Exposure to the Four First-Line Anti-Tuberculosis Drugs and Treatment Outcomes: A Target Trial Emulation Study in Ghana

Exposure to the Four First-Line Anti-Tuberculosis Drugs and Treatment Outcomes: A Target Trial Emulation Study in Ghana | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Exposure to the Four First-Line Anti-Tuberculosis Drugs and Treatment Outcomes: A Target Trial Emulation Study in Ghana Michael Opoku-Mireku, Awewura Kwara, Margaret Y. M. Lartey, Charles A. Peloquin, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9255541/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 12 You are reading this latest preprint version Abstract Background Tuberculosis (TB) treatment outcomes in sub-Saharan Africa remain suboptimal despite high adherence to first-line therapy. Variability in drug pharmacokinetics, resulting in subtherapeutic plasma concentrations, may contribute to treatment failure and the development of resistance. This study estimated the causal effect of subtherapeutic plasma concentrations of first-line anti-TB drugs on treatment failure or death among individuals with drug-susceptible pulmonary TB in Ghana. Methods We conducted a prospective cohort study of 164 adults receiving standard WHO weight-band dosing at five Ghanaian hospitals. Peak plasma concentrations (C max ) of rifampicin, isoniazid, pyrazinamide, and ethambutol were measured at months 1–2 using validated LC-MS/MS. We emulated a target trial comparing two static strategies: ( 1 ) therapeutic C max of at least one first-line drug versus ( 2 ) subtherapeutic C max of all four drugs. Using the clone-censor-weight approach, we estimated the per-protocol analogue risk difference (RD) and risk ratio (RR) for treatment failure (smear positive at months 5 or 6) or death by month 6. Models were adjusted for baseline covariates using inverse probability of censoring weighting. Sensitivity analyses included inverse probability weighting with regression adjustment, plain inverse probability weighting, and E-values. Results Of 164 participants, 120 had complete pharmacokinetic and outcome data; 20.0% (24/120) had subtherapeutic concentrations of all four drugs. The 6-month risk of treatment failure or death was 33.3% under the low-exposure strategy versus 4.2% under the adequate-exposure strategy (crude RD: 29.1 percentage points). In weighted analyses, low drug exposure was associated with a 25.6 percentage-point increase in absolute risk of treatment failure or death (95% CI: 5.7–45.6; p = 0.012) and an 8.6-fold higher relative risk (95% CI: 2.34–31.92; p = 0.001), corresponding to approximately one additional poor outcome for every four patients with subtherapeutic levels. Sensitivity analyses were consistent (ATE: 19.9%, 95% CI: 2.3–37.5). The E-value was 15.5 (lower bound 5.0). Conclusions Subtherapeutic exposure to all four first-line drugs was strongly associated with increased risk of treatment failure or death. Preventing multidrug subtherapeutic exposure through therapeutic drug monitoring or optimized dosing warrants randomized evaluation in high-burden settings. Trial registration Clinical trial number: not applicable. Tuberculosis Pharmacokinetics Treatment failure Target trial emulation Subtherapeutic concentrations Ghana Figures Figure 1 INTRODUCTION Tuberculosis remains a major public health challenge, with control efforts falling behind global targets. The WHO End TB strategy targets a 50% reduction in the 2015 incidence rate by 2025, however, by 2023 only an 8.3% reduction had been achieved. The mortality rate reduction achieved in 2023 was 23% of the 2015 rate, although the target for 2025 was 75% ( 1 ). Treatment success rates have improved, reaching 88% for drug-susceptible TB. However, the success rate for drug-resistant TB remains substantially lower at 68%. This disparity underscores the urgent need to prevent the emergence and spread of drug resistance as a key strategy in the global fight against tuberculosis. In sub-Saharan Africa, which bears the highest burden of disease, there has not been much improvement in the treatment success rate in recent years ( 1 ). For example, Ghana’s TB treatment success rate has stagnated between 84% and 87% since 2012 ( 2 ). Although this represents a substantial improvement from the 50% reported in 2000, the lack of further progress in recent years may be due to a complex interplay of factors that warrant urgent investigation. A plausible consequence of this observation is the increased emergence of resistant strains of Mycobacterium tuberculosis in recent years. Whereas globally, the emergence of resistance among previously treated patients is six times higher compared to treatment-naïve patients, it is estimated to be 19 times higher in Ghana ( 1 , 3 ). The magnitude of the difference in the rate between previously treated and treatment-naïve patients suggests a treatment-related process that facilitates acquired resistance. Although nonadherence is often thought to be primarily responsible, studies by Pasipanodya and colleagues point to the person-to-person variability in the pharmacokinetics of the drugs rather than nonadherence, unless beyond a low threshold (> 60%) ( 4 , 5 ). A systematic review and meta-analysis has established that there is a high prevalence of low plasma concentrations of the first-line anti-TB medications at standard doses, thereby substantiating the role of pharmacokinetic variability in promoting resistance ( 6 ). These observations underscore the importance of pharmacokinetic variability as a potential determinant of treatment outcomes, providing a key rationale for the present study. Investigations into the relationship between plasma concentrations of the anti-TB medication at standard doses and treatment outcomes are not conclusive. Despite high adherence to the regimen, poor outcomes may still result ( 4 ). Despite high rates of low plasma concentrations successful TB treatment outcome are common ( 7 , 8 ). The challenge to addressing the dilemma emanates from methodological limitations of the available studies. The evidence is derived from a limited number of studies conducted in settings that may not fully represent the broader population, often involving small sample sizes. Published TB pharmacokinetic studies from Ghana and other West African countries have important limitations. Many were not designed to evaluate the relationship between drug exposure and treatment outcomes or were conducted only among selected clinical subgroups ( 9 – 11 ). Several others relied on retrospective clinical data or did not follow the routine programmatic treatment duration ( 12 – 14 ), reducing their generalizability. These limitations are compounded by the fact that potential confounders were rarely measured or adjusted for, as observed by Perumal and team ( 7 ). Even in studies that adjusted for measured confounders, the resulting estimates are primarily associative rather than causal, reflecting biological signals that may not be robust for causal inference. Furthermore, most existing studies focus on individual drug concentrations, even though tuberculosis treatment outcomes arise from the combined synergistic or antagonistic effects of the full multi-drug regimen. This further constrains the interpretability of prior evidence. A valid estimation of the effect of the pharmacokinetics of the drugs on treatment outcomes is essential. Understanding whether, and how much, low drug exposure despite treatment adherence influences poor outcomes could explain much of the remaining gap in cure rates and open a practical new path to improvement. This would ensure patients receive doses optimized to achieve a cure, thereby raising success rates, shortening infectious periods, and reducing the emergence of resistance. As has been suggested, optimization of the first-line treatment regimen holds the greatest potential for reducing TB incidence and mortality( 15 , 16 ). We therefore analyzed data from a prospective cohort study in Ghana to answer the following causal question: Among patients with drug-susceptible pulmonary tuberculosis treated with standard WHO-recommended weight-band dosing, what is the 6-month risk of treatment failure or death under a hypothetical dosing strategy that guarantees therapeutic exposure to at least one first-line drug from treatment initiation, compared with a hypothetical dosing strategy that results in subtherapeutic exposure to all four first-line drugs? To estimate this effect, we used target trial emulation with the clone-censor-weight approach. METHODS Target Trial Specification and Emulation We emulated a two-arm randomized controlled trial comparing treatment strategies among patients with drug-susceptible pulmonary tuberculosis who survived and remained on treatment until pharmacokinetic (PK) assessment (month 1–2): ( 1 ) adequate-exposure strategy: therapeutic peak concentrations (C max ) of at least one of the four first-line drugs. ( 2 ) low-exposure strategy: subtherapeutic C max of all four drugs. Time zero was defined as the date of pharmacokinetic measurement. Follow-up extended from pharmacokinetic assessment to the end of standard therapy. We sought to estimate the population-averaged risk difference (RD) in poor treatment outcomes (failure or death), comparing the adequate exposure strategy with the low exposure strategy, under the assumptions of exchangeability, positivity, and consistency. The main components of the target trial are summarized in Table 1 . Exposure and Outcome Definitions The C max of the drugs were categorized as “subtherapeutic” if they were below the published threshold and “therapeutic” if equal to or above this threshold ( 17 ). The thresholds were 8 µg/ml for rifampicin, 3 µg/ml for isoniazid, 20 µg/ml for pyrazinamide, and 2µg/ml for ethambutol. Because a threshold of 35 µg/ml has been used in clinical practice for pyrazinamide ( 18 , 19 ), proportions based on both thresholds are presented for completeness. However, all primary analyses were conducted using the 20 µg/ml threshold. The primary outcome was treatment response, categorized as “success” for clinical treatment outcomes of cured and treatment completed, or “poor” for treatment failure and death. Clinical outcome categories were defined using the World Health Organization’s standard criteria ( 20 ). A patient was considered cured if they had bacteriologically confirmed pulmonary TB at baseline and became smear negative in the final month of treatment, as well as at least one earlier time point. Treatment completion was assigned to patients who finished the full course of therapy without evidence of treatment failure, but for whom bacteriological results in the final month were unavailable or not performed. Treatment failure referred to patients who remained smear positive at month five or later. Patients who died from any cause during treatment were categorized under mortality. Those who interrupted treatment for two or more consecutive months were classified as lost to follow-up. Adherence was defined as the percentage of prescribed doses missed in the month preceding pharmacokinetic sampling. High adherence was categorized as missing fewer than 10% of doses, while low adherence was defined as missing 10% or more. Data source, Eligibility, and Follow-up Data was obtained from a prospective cohort study conducted at five hospitals (Komfo Anokye Teaching Hospital, Kumasi South Hospital, Holy Family Hospital - Techiman, Suntreso Government Hospital, and Tafo Government Hospital) to investigate the risk factors of poor drug-susceptible tuberculosis treatment outcomes. Patients were enrolled consecutively from 1st August 2022 until the pre-determined sample size was reached on 23rd August 2023. Eligible participants were patients aged ≥ 15 years, weighing at least 30 kg, and had rifampicin-susceptible pulmonary tuberculosis confirmed by GeneXpert MTB/RIF. Patients with multidrug-resistant TB, exclusively extrapulmonary TB, or unknown prior treatment history were excluded. Previously treated patients were included only if their last episode had ended in cure or treatment completion (n = 1 out of the analytic cohort of 120 (0.8%). Peak plasma concentrations of rifampicin, isoniazid, pyrazinamide, and ethambutol were measured once between months 1 and 2 of treatment. Participants were followed monthly until the end of the standard 6-month regimen, with treatment outcomes recorded according to national guidelines. Sample size and Recruitment A sample of size 164 was enrolled. This number was based on an assumed 5% rate of treatment failure or death among adequately exposed patients ( 21 ), a design effect of 2.0 to account for clustering by facility, 80% power, and a two-sided α of 0.05, with an additional 20% added for missing pharmacokinetic data or loss to follow-up. Recruitment continued until the target was reached: 92 patients from Komfo Anokye Teaching Hospital, 37 from Holy Family Hospital - Techiman, 21 from Kumasi South Hospital, and 7 each from Suntreso and Tafo Government Hospitals. Drug Dosing and Administration All participants received daily fixed-dose combinations of the four first-line anti-TB drugs during the intensive phase (first 2 months) and rifampicin plus isoniazid during the continuation phase (next 4 months), administered by DOTS in line with WHO recommendations ( 22 ). Dosing followed standard weight bands: 30–39 kg (300 mg rifampicin, 150 mg isoniazid, 550 mg ethambutol, 800 mg pyrazinamide), 40–54 kg (450 mg rifampicin, 225 mg isoniazid, 825 mg ethambutol, 1,200 mg pyrazinamide), and ≥ 55 kg (600 mg rifampicin, 300 mg isoniazid, 1,100 mg ethambutol, 1,600 mg pyrazinamide) ( 23 ). Pharmacokinetic Data Collection and Analysis At month 1 or 2, 4 ml of blood was collected two and four hours after dosing. Participants fasted overnight (≥ 8 hours), took their observed dose, and received a light meal 30 minutes post-dose. Blood samples were transferred into heparinized tubes, centrifuged at 3,000 g for 10 minutes at 4°C, and the resulting plasma stored at − 80°C until analysis. All specimens were subsequently shipped on dry ice to the Infectious Disease Pharmacokinetic Laboratory at the University of Florida for quantification of drug concentrations. Quantification was done using a validated high-performance liquid chromatography–tandem mass spectrometry (LC-MS/MS) assay. The method demonstrated precision with percentage relative standard deviations between 0.2% and 6.6% across high, medium, and low quality-control levels, and accuracy ranging from 99.03% to 99.67% for all four drugs. Measured concentrations ranged from 0.012–14.382 µg/mL for rifampicin, 0.052–5.162 µg/mL for isoniazid, 0.122–54.142 µg/mL for pyrazinamide, and 0.0052–5.892 µg/mL for ethambutol. Maximum plasma concentrations were estimated by maximum likelihood using the 2-hour and 4-hour post-dose samples modeled in Monolix 2023 ( 24 ). Concentrations below the lower limits of quantification (0.25 µg/mL for rifampicin, 0.15 µg/mL for isoniazid, 0.50 µg/mL for pyrazinamide, and 0.05µg/ml ethambutol) were excluded from the C max calculation to prevent unreliable model inputs that would bias C max estimates downward. Table 1 Specification of the target trial and its emulation using observational pharmacokinetic data Component Target Trial (Ideal but infeasible) Emulation using observational data Eligibility Patients aged 15 years or older with a minimum body weight of 30kg, diagnosed with rifampicin-susceptible TB using the GeneXpert MTB/Rif. Same, but must have survived and remained on treatment until pharmacokinetic sampling (complete-case restriction) Treatment strategies Strategy A: Dosing regimen that guarantees a therapeutic C max of at least one of the four first-line drugs from day 0 onward Strategy A: Observed month 1–2 C max shows therapeutic level for at least one drug (adequate-exposure phenotype) B: Dosing regimen that results in subtherapeutic C max of all four first-line drugs from day 0 onward B: Observed month 1–2 C max subtherapeutic for all four drugs (low-exposure phenotype) Assignment Randomization at baseline (day 0) Clone-censor-weight: each participant cloned and assigned to both strategies; clones censored when their observed exposure violates the assigned strategy Follow-up start From day 0 of treatment until month 6, death, treatment failure, or loss to follow-up Same, with time zero set at pharmacokinetic sampling (month 1–2) and follow-up to month 6 Outcome Treatment failure or death by month 6 Same Causal contrast Intention-to-treat effect Per-protocol analogue via clone-censor-weighting Statistical Analysis Risk difference/ risk ratio Clone-censor-weight estimation for per-protocol analogue effect of static regimes. Sensitivity analyses included IPWRA, plain IPW, and E-value. Key assumptions Randomization ensures exchangeability Conditional exchangeability, positivity, and consistency given measured covariates (age, sex, weight, HIV, diabetes, smoking, adherence, alcohol, diet, smear grade). IPW, Inverse Probability Weighting; IPWRA, Inverse Probability Weighted Regression Adjustment Statistical Analysis and Causal Inference The pharmacokinetic phenotype reflects a final common pathway of diverse upstream factors, including body weight, malabsorption, drug-drug interactions, genetic variability, adherence, among others, many of which cannot be fully measured in routine settings ( 25 , 26 ). We estimated the per-protocol effect of sustained adequate versus low pharmacokinetic exposure using standardized static-regime estimation with baseline cloning and artificial censoring. Each participant was duplicated at baseline and assigned to one of two static strategies: therapeutic C max of at least one of the four first-line drugs versus subtherapeutic C max of all four drugs. Clones whose assigned strategy was incompatible with the observed month 1–2 pharmacokinetic phenotype were artificially censored at baseline and excluded from follow-up. Compatible clones were retained in the analysis and weighted by the inverse of the estimated probability of compatibility, conditional on baseline covariates (age, sex, weight, HIV co-infection, diabetes comorbidity, current smoking, adherence, alcohol use, dietary diversity, and baseline smear grade). To reduce variance, weights were stabilized by dividing by the marginal probability of compatibility and winsorized at the 1st and 99th percentiles. This procedure standardizes outcomes to the baseline covariate distribution and estimates the population-averaged risk difference and risk ratio, and adheres to each exposure phenotype, independent of upstream determinants of exposure. Risk differences and risk ratios were obtained from weighted binomial generalised linear models with identity and log links, respectively, with robust variance estimation clustered by participant. The number needed to treat (NNT) was calculated as the reciprocal of the primary risk difference (RD = 25.6%), yielding an NNT of 4 (95% CI: 2 to 18), indicating that shifting 4 patients from the low-exposure phenotype to adequate exposure would prevent one additional case of treatment failure or death. Sensitivity Analysis To assess the robustness of the primary findings, we conducted three sensitivity analyses. First, we used augmented inverse probability weighted regression adjustment (IPWRA) to estimate the average treatment effect (ATE) of the low-exposure phenotype versus adequate exposure, adjusting for baseline covariates (age, sex, weight, HIV co-infection, diabetes comorbidity, current smoking, adherence, alcohol use, dietary diversity, and smear grade) with robust standard errors. This doubly robust approach remains consistent if either the treatment or outcome model is correctly specified. We also examined covariate balance using standardized mean differences before and after weighting. Raw standardized mean differences (SMD) ranged from − 0.90 to + 0.21 (largest imbalance for weight). After weighting, all SMDs were reduced, with most falling within |<0.30| (maximum absolute SMD = 0.41 for alcohol; range − 0.30 to + 0.41). Variance ratios remained acceptable (0.45–1.62), supporting effective reweighting. The treated group's effective sample size increased to 54.3 (from raw n = 21), while the control group's effective sample size decreased to 53.7 (from raw n = 87), reflecting the reweighting process to achieve balance. Second, we applied plain inverse probability weighting (IPW) to estimate the ATE, using the same covariates and robust standard errors, with an oversampling strategy to handle potential positivity violations. Weighted balance was again assessed via standardized mean differences. Third, we calculated E-values to quantify the strength of unmeasured confounding required to nullify the primary risk difference estimate (RD = 25.6%). The E-value assesses how strongly an unmeasured confounder would need to be associated with both the exposure and outcome (on the risk ratio scale) to explain away the observed association, assuming no true effect ( 27 ). All sensitivity analyses were restricted to patients with complete follow-up (i.e., uncensored patients) and utilized Stata’s “teffects” suite with robust variance estimation. Exploratory Firth-penalized Poisson regression models were fitted to examine individual-drug associations under sparse data and potential separation; these were regarded as hypothesis-generating only and were not used for primary causal inference. Results are shown for completeness only in Additional file 3. All analyses were performed in Stata 17. Limitations of the estimand The clone-censor-weight approach identifies population-averaged effects attributable to the observed pharmacokinetic phenotype, independent of its upstream causes. It does not estimate the effects of hypothetical interventions like therapeutic drug monitoring-guided dosing. Key assumptions include no unmeasured confounding of phenotype-outcome and correct trial specification. RESULTS Of the 164 patients enrolled, 17 (10.4%) were lost to follow-up, and 27 (16.5%) did not have complete four-drug pharmacokinetic measurements at months 1–2, leaving 120 patients (73.2%) for the emulated target trial and all primary analyses. Patients excluded because of loss to follow-up or incomplete pharmacokinetic data (n = 44) were broadly similar to the analytic sample (n = 120). The largest standardized mean differences were 0.47 for baseline weight, 0.46 for treatment facility, and 0.39 for dietary diversity score; all other characteristics had standardized differences < 0.25 (Additional file 1). These same variables also had statistically significant p-values. These modest imbalances indicate that selection into the final analytic cohort introduced limited bias. Table 2 summarizes the baseline characteristics of participants by their pharmacokinetic phenotype. Of the 120 participants, 85 (70.8%) were male, 81 (67.5%) had a high school education, 77 (64.2%) were employed, 65 (54.2%) were unmarried, and 65 (54.2%) sought care at the Komfo Anokye Teaching Hospital. Mean (SD) age was 41.7 years (13.5), median weight was 58.5 kg (IQR: 52.4, 64.0), and median dietary diversity score was 7.0 (IQR: 5.4, 9.6) out of a maximum of 10. While 20.0% (24/120) were HIV co-infected, 7.5% (9/120) had diabetes, and 38.3% (46/120) were on non-TB medications. Thirteen out of 120 (10.8%) missed 10% or more doses within the first two months of treatment initiation. Table 2 Baseline characteristics by Pharmacokinetic Phenotype Characteristics Pharmacokinetic Phenotype Total, N = 120 Adequate exposure n = 96 Low exposure n = 24 Age in years, mean (SD) 41.7 (13.5) 42.0 (13.9) 40.5 (12.0) Weight in kilograms, median (IQR) 58.5 (52.4, 64.0) 59.0 (53.0, 65.8) 56.0 (49.0, 62.1) Dietary diversity Score, median (IQR) 7.0 (5.4, 9.6) 6.9 (5.4, 9.4) 7.6 (4.9, 9.8) Household size, median (IQR) 4.0 (3.0, 7.0) 4.0 (3.0, 6.5) 4.0 (2.5, 7.5) Male sex, n (%) 85 (70.8) 68 (70.8) 17 (70.8) Highest Educational Level, n (%) No formal Education 23 (19.2) 20 (20.8) 3 (12.5) Primary School 9 (7.5) 9 (9.4) 0 (0.0) High School 81 (67.5) 62 (64.6) 19 (79.2) Tertiary 7 (5.8) 5 (5.2) 2 (8.3) Employment status, n (%) Unemployed 32 (26.7) 24 (25.0) 8 (33.3) Employed 77 (64.2) 63 (65.6) 14 (58.3) Student 8 (6.7) 6 (6.3) 2 (8.3) Other# 3 (2.5) 3 (3.1) 0 (0.0) Not married, n (%) 65 (54.2) 51 (53.1) 14 (58.3) Facility, n (%) Komfo Anokye Teaching Hospital 65 (54.2) 53 (55.2) 12 (50.0) Suntreso Government Hospital 4 (3.3) 3 (3.1) 1 (4.2) Kumasi South Hospital 11 (9.2) 9 (9.4) 2 (8.3) Tafo Government Hospital 4 (3.30 3 (3.1) 1 (4.2) Holy Family Hospital, Techiman 36 (30.0) 28 (29.2) 8 (33.3) Current alcohol use, n (%) 21 (17.6) 17 (17.9) 4 (16.7) Current smoker, n (%) 8 (6.9) 7 (7.4) 1 (4.5) Non-adherent (≥ 10% of doses missed), n (%) 13 (10.8) 10 (10.4) 3 (12.5) Concomitant non-TB drugs, n (%) 46 (38.3) 38 (39.6) 8 (33.3) HIV Co-infected n (%) 24 (20.0) 21 (21.9) 3 (12.5) Diabetes Co-morbidity, n (%) 9 (7.5) 8 (8.3) 1 (4.2) Severe Smear Grade (3+), n (%) 27 (23.7) 20 (22.2) 7 (29.2) SD, Standard Deviation; IQR, Interquartile Range # Includes retirees and apprentices *Adequate exposure = therapeutic C max of at least one drug †Low exposure = subtherapeutic C max of all four drugs Table 3 summarizes the distribution of C max for the four first-line anti-TB drugs. Subtherapeutic C max levels were observed in 93.3% of participants for rifampicin, 86.7% for isoniazid, and 53.3% for ethambutol. For pyrazinamide, using the widely reported threshold of 20 µg/ml, 20.0% of participants had subtherapeutic C max . However, when the higher clinical practice threshold of 35 µg/ml is applied, this proportion increases markedly to 80.0%. Overall, 3.3% (4/120) of participants had therapeutic concentrations for all drugs, whereas 20.0% (24/120) exhibited subtherapeutic concentrations for all four medicines. Table 3 Distribution of peak plasma concentration of anti-TB drugs for participants Drug Peak Concentration, µg/ml median (IQR) Subtherapeutic C max n (%) Therapeutic C max n (%) Rifampicin 4.1 (2.0, 5.6) 112 (93.3) 8 (6.7) Isoniazid 1.6 (0.7, 2.4) 104 (86.7) 16 (13.3) Pyrazinamide (< 20 µg/ml) 24.9 (12.8, 32.5) 24 (20.0) 96 (80.0) Pyrazinamide (< 35 µg/ml) 24.9 (12.8, 32.5) 96 (80.0) 24 (20.0) Ethambutol 1.9 (1.2, 2.5) 64 (53.3) 56 (46.7) All four drugs - 24 (20.0) 96 (80.0) C max, Peak Plasma Concentration; IQR, Interquartile Range; IQR, Interquartile Range Figure 1: Bar chart showing clinical treatment outcomes by pharmacokinetic phenotype Clinical outcomes differed between the adequate exposure phenotype and the low exposure phenotype (Fig. 1). Among patients with adequate exposure (n = 96), 95.8% achieved treatment success (70.8% cured; 25.0% completed treatment), and only 4.1% experienced treatment failure or death. In contrast, among those with subtherapeutic concentrations of all four drugs (n = 24), treatment success declined to 66.6% (58.3% cured; 8.3% completed), while poor outcomes increased to 33.4% (29.2% failure; 4.2% death) (p < 0.001). The distribution of baseline covariates by treatment response is presented in Additional file 2. The results in Table 4 show a strong and consistent link between subtherapeutic C max of all four first-line anti-TB drugs and a much higher risk of treatment failure or death. In the emulated target trial, the 6-month risk of poor treatment response was 33.3% (8 of 24 patients) under the low-exposure strategy compared to 4.2% (4 of 96 patients) under the adequate-exposure strategy. This corresponds to a large and clinically significant causal effect: a crude risk difference of 29.1 percentage points. In the primary analysis (weighted binomial regression with identity link, accounting for censoring via inverse probability of censoring weights), patients with subtherapeutic levels of all four drugs had an absolute risk increase of 25.6 percentage points (95% CI: 5.7% to 45.6%; p = 0.012) compared to those with adequate levels of at least one drug. This translates to an estimated 3–4 poor outcomes per 100 patients with therapeutic drug levels, rising to approximately 29–50 per 100 among those with low levels of all four drugs. This difference is clearly meaningful. Using a log-linked binomial model (risk ratio scale), the same exposure was associated with an 8.6-fold higher risk of failure or death (RR = 8.64, 95% CI: 2.34–31.92; p = 0.001), meaning patients with low exposure to all four drugs were more than eight times as likely to experience a poor outcome compared to those with better drug levels. Regarding sensitivity analyses, both IPWRA and plain IPW yielded identical average treatment effects of 19.9 percentage points higher risk (95% CI: 2.3%–37.5%; p = 0.027), after adjustment for age, sex, weight, HIV status, diabetes, smoking, alcohol use, adherence, dietary diversity, and baseline smear grade. The consistency across the two approaches strengthens confidence that the association is not explained by measured confounding or model misspecification ( 28 ). The third sensitivity analysis was the E-value. The point estimate was 15.48, meaning that any unmeasured confounder(s) would need to be associated with both exposure and outcome by a risk ratio of at least 15.5 in both directions to nullify the observed effect. The corresponding E-value for the lower bound of the confidence interval was 4.96, indicating that an unmeasured confounder would still require a substantial risk ratio of approximately 5.0. These high E-values suggest that the association is unlikely to be explained entirely by unmeasured confounding, providing additional support for a causal interpretation of the relationship between low all-four drug exposure and increased risk of poor treatment outcomes. Overall, these complementary analyses showed that low exposure to all four first-line anti-TB drugs is associated with an elevated risk of treatment failure or death, approximately 20–26 percentage points higher in absolute terms, or more than eight times higher in relative terms. Table 4: Risk of Treatment Failure or Death Associated with All Four Subtherapeutic Anti-TB Drugs Exposure Model Type Estimand Effect [95% CI] P-value Adjustment Primary Analyses (n = 108) Weighted Binomial regression (identity link) Risk Difference 25.6% [5.7%, 45.6%] 0.012 IPCW-weighted analysis restricted to uncensored cases, with robust standard errors (clustered by participant ID) Weighted Binomial regression (log link) Risk Ratio 8.64 [2.34, 31.92] 0.001 IPCW-weighted analysis restricted to uncensored cases, with cluster-robust standard errors by participant ID Sensitivity Analyses (n = 108) Inverse Probability Weighting Regression Adjustment Average Treatment Effect (%) 19.9% [2.3%, 37.5%] 0.027 Robust SE, adjusted for baseline covariates Inverse Probability Weighting Average Treatment Effect (%) 19.9% [2.3%, 37.5%] 0.027 Robust SE, oversample strategy, adjusted for baseline covariates E-value (for primary RD = 25.6%) E-value 15.48 [Lower bound: 4.96] - An unmeasured confounder would need RR ≥15.5 with both exposure & outcome to nullify RD Abbreviation : CI, Confidence Interval;IPCW, Inverse Probability of Censoring Weight; IPW, Inverse Probability Weighting; IPWRA, Inverse Probability Weighting Regression Adjustment; SE, Standard Error Notes : Log-binomial provides a multiplicative RR scale. Covariates were age, sex, weight, HIV, diabetes, smoking, alcohol use, adherence, dietary diversity, and smear grade. All models restricted to uncensored cases (n=108); robust standard errors throughout. No poor outcomes were observed among patients with therapeutic rifampicin C max . However, small numbers precluded formal multivariable estimation (Additional file 2). Low C max was associated with a 7.3-fold (95% CI: 1.89, 28.33) higher risk of treatment failure or death than adequate C max . Subtherapeutic C max for pyrazinamide (IRR = 16.7; 95% CI: 4.0–69.4, p < 0.001) and ethambutol (IRR = 4.1; 95% CI: 1.2–13.9, p = 0.024) were each strongly associated with poor treatment response. There wasn't enough evidence of an association between isoniazid C max and treatment response (Additional file 3). DISCUSSION In this prospective cohort study of adults with drug-susceptible pulmonary tuberculosis in Ghana, subtherapeutic concentrations of all four first-line anti-TB drugs were associated with an increased risk of treatment failure or death. This contrasted with a hypothetical strategy in which therapeutic concentrations of at least one drug were achieved at treatment initiation. The findings were consistent across primary and sensitivity analyses. Moreover, the E-value indicates that an unmeasured confounder would need to be associated with both the exposure and the outcome by risk ratios greater than 15.5 (lower bound 5.0) to nullify the observed effect. Unmeasured confounding greater than 15.5 is implausible since confounders typically exert far smaller effects, but cannot be ruled out (29–32). Even the lower confidence limit (5.7%) represents a significant harm. Our findings suggest that averting the low-exposure phenotype could prevent one poor outcome per 4 patients (NNT=3.9), supporting research into pharmacokinetics-guided interventions. Our findings are biologically plausible, underscored by the synergistic effect of the drugs in the standard regimen. The clinical efficacy is largely preserved if at least one companion drug achieves adequate exposure, even if the others have low exposure. This may explain the common paradoxical relationship between high overall treatment success rates and low drug concentrations (4,8). However, low exposure to all four drugs disrupts the synergistic bactericidal and sterilizing activity of the regimen, rendering these patients at an elevated risk of failure, death, relapse, and selection of drug-resistant strains (33,34). Ghana’s disproportionately high (19-fold) risk of resistance among previously treated patients compared to treatment-naïve ones likely reflects this phenomenon (1,3). The high prevalence of sub-therapeutic C max levels for the four drugs, despite high adherence, is consistent with previous studies in sub-Saharan African populations, where malnutrition, drug-drug interactions, and genetic variations affecting drug metabolism are common (7,35). Clinically, this highlights the fact that patients may fail therapy even when they take their medication correctly. This is because standard doses based on weight bands may not achieve the desired therapeutic exposure in all patients. Another potential contributor to subtherapeutic exposure across all four drugs is inter-individual variability in drug metabolism. Genetic polymorphisms affecting drug-metabolizing enzymes and transporters, such as NAT2 polymorphisms governing isoniazid acetylation, transporter variants such as ABCB1 affecting rifampicin disposition, and hepatic enzymatic pathways involved in pyrazinamide metabolism, are prevalent in African populations and can lead to low plasma concentrations in some individuals (26,36,37). While we did not assess pharmacogenetic markers or metabolite levels, such differences could partly explain the observed phenotype and warrant further investigation in larger cohorts with pharmacokinetic and genetic data. Our results underscore the need to integrate pharmacokinetic considerations in routine TB care. Implementing therapeutic drug monitoring (TDM) in resource-limited settings like Ghana poses challenges, including the need for advanced laboratory infrastructure and trained personnel. Our study’s use of LC-MS/MS for plasma concentration quantification, conducted at a specialized facility, underscores these logistical barriers. Pragmatic clinical proxies such as early sputum non-conversion, persistent symptoms despite adherence, and markers of severe disease (cavitation, low body weight) can prompt dose intensification (38,39). Alternatively, population pharmacokinetic models that incorporate readily available covariates (weight, HIV status, diabetes, age) can be used to guide empiric dose adjustments (10). Although LC–MS/MS remains the gold standard for drug quantification due to its high sensitivity and specificity, high-performance liquid chromatography with ultraviolet detection (HPLC-UV) may represent a more affordable alternative in resource-constrained settings (40,41). In addition, dried blood spot (DBS) sampling offers a practical option for therapeutic drug monitoring in remote or decentralized settings, despite being technically more demanding than plasma-based assays (42). These provide opportunities worth exploring to enable the introduction of TDM into routine TB care. At the population level, the stagnation of treatment success rates in Ghana and similar settings may reflect prevalent low drug exposure rather than programmatic failure alone. If a critical mass of patients systematically achieves subtherapeutic levels of the drugs while on the recommended regimen, even an effective Directly Observed Treatment, Short-course (DOTS) programme cannot fully achieve its expected outcomes. Improving treatment outcomes will require moving beyond adherence-based interventions alone. Public health programmes need to consider differentiated dosing for at-risk subtherapeutic populations, such as those with comorbidities, or a regimen with optimized doses of one or two of the drugs. In the case of the latter, our study suggested that pyrazinamide and ethambutol concentrations signaled an effect on poor response. It is worth noting that these proposals remain exploratory. This estimand demonstrates that preventing the low-exposure phenotype would substantially reduce poor treatment outcomes. However, it does not establish whether TDM, higher-dose regimens, or any other specific intervention can reliably achieve this phenotype shift in routine clinical practice. Definitive evidence of effectiveness and feasibility will require further randomized studies explicitly designed to test these implementation strategies locally. To comprehensively address this under-recognized pharmacokinetic barrier to TB elimination, we recommend piloting TDM in high-risk patients at teaching hospitals, accelerating research into higher-dose regimens, and building sustainable local capacity for pharmacokinetic research. The strengths of this study include its rigorous target trial emulation framework, clearly defining eligibility, time zero, exposure strategies, follow-up, and estimand, reducing common biases like immortal time bias and improving transparency over standard observational analyses. Drug exposure was measured directly via observed C max (not proxies like dose or weight), providing strong biological grounding. Confounding was addressed with stabilized inverse probability weighting and formal balance diagnostics, yielding population-averaged estimates relevant to clinical and policy decisions. Both absolute (risk difference, NNT) and relative (risk ratio) effect measures were reported, enhancing interpretability and relevance. The exposure contrast focused on a severe, biologically extreme phenotype (all four drugs subtherapeutic), making the large observed effect more plausible as a pharmacologic signal. Together, these features strengthen internal validity and support cautious causal interpretation within the study’s assumptions and constraints. Some limitations existed despite these strengths. Although inverse probability weighting approximated exchangeability, residual confounding remains possible. Unmeasured markers of disease severity (such as cavitary disease, radiographic extent, serum albumin, malabsorption, or inflammatory burden) may influence both exposure and outcomes. While measured covariates achieved acceptable balance post-weighting, causal interpretation depends on the assumption that all important confounders were captured. Exposure was defined at months 1–2, with time zero at pharmacokinetic sampling. The effect, therefore, applies only to patients who survived until sampling. Early deaths or failures before measurement were excluded, limiting generalizability and introducing potential survivor selection bias. A single C max measurement may not fully capture longitudinal exposure. Within-person variability, assay error, or absorption fluctuations could cause misclassification, likely biasing estimates toward the null if nondifferential, though differential misclassification cannot be excluded. Despite stabilized weights and improved balance, the small number of patients with the low-exposure phenotype raises concerns about positivity and weight instability. Extreme weights were truncated, but finite-sample variability may affect precision. Approximately one-quarter of eligible participants were excluded due to missing pharmacokinetic or outcome data. Baseline differences between included and excluded patients suggest potential selection bias, limiting external validity. Finally, the exposure contrast reflects a severe composite phenotype (all four drugs subtherapeutic) rather than drug-specific effects. Findings should not be interpreted as the isolated causal effect of any single agent, but rather as the estimated impact of avoiding profound multidrug subtherapeutic exposure. Despite these limitations, the clearly defined time zero, explicit causal estimand, and robust weighting framework strengthen internal validity relative to conventional regression-based analyses. CONCLUSION Ghana and most sub-Saharan African countries have not been able to acheive TB treatment success above 85–90% despite generally high reported adherence and well-functioning programmes. Our findings suggest that some individuals do not achieve therapeutic concentrations of any of the four first-line drugs on currently recommended doses. Recognizing and correcting this hidden pharmacokinetic shortfall would directly increase the chance of cure for those individuals, shorten infectiousness, and reduce the emergence of drug-resistant strains. These phenotype effects suggest that interventions averting subtherapeutic exposure, such as TDM-guided dosing or empiric high-dose regimens, merit randomized evaluation in high-burden settings like Ghana. Declarations Ethics approval and consent to participate The study was conducted in accordance with the Declaration of Helsinki. Ethical clearance for the study was received from the Ghana Health Service Ethics Review Committee (GHS-ERC 002/02/21) and the KATH Ethics Committee (KATH IRB/AP/023/21 and KATH IRB/CR01/023/22). Written informed consent was obtained from all adult participants aged 18 years and above. For participants under the age of 18 (minors), written informed consent was obtained from their parents or legal guardians, and assent was also obtained from the minors themselves. Consent for publication Not applicable Availability of data and materials The dataset contains sensitive personal health information from human participants, including identifiable details such as dates of treatment, clinical measurements, and demographic data. Public release would violate participant confidentiality and contravene ethical approval conditions and data protection regulations. The data are therefore available upon reasonable request from the corresponding author, MOM. Competing interests The authors declare that they have no competing interests. Funding The study was funded by the Fogarty International Centre of the National Institutes of Health, US, under the UG-Florida Academic Partnership project (D43 TW010055). The Infectious Disease Pharmacokinetic Laboratory at the University of Florida performed the drug assays, as part of MOM’s PhD research. Authors’ Contributions The study was conceptualized by M.O.M., M.Y.M.L., A.K., C.A.P., and K.A.K. Data collection was supervised by M.O.M. Data analysis was conducted by M.O.M., with methodological support and guidance from A.A.M. and K.A.K. Data interpretation was performed by D.A.Y.A., M.O.M., P.A.N., K.A.T and KAK. K.A.T interpreted radiographic images where available. Pharmacokinetic analysis was supervised by C.A.P., with the involvement of M.O.M. The first draft of the manuscript was written by M.O.M. All authors reviewed and edited the manuscript, with high-level critical review provided by M.Y.M.L., A.K., C.A.P., A.A.M., P.A.N., and K.A.K. All authors read and approved the final version of the manuscript. Acknowledgments The authors would like to acknowledge the study participants for their participation in the study, as well as the field team for their role in data collection. Trial registration Clinical trial number: not applicable. References WHO. Global Tuberculosis Report 2024 [Internet]. 2024 [cited 2025 Jan 23]. Available from: https://www.who.int/teams/global-tuberculosis-programme/tb-reports/global-tuberculosis-report-2024 World Bank Group. World Bank Open Data [Internet]. 2025 [cited 2025 Mar 10]. World Bank Open Data. Available from: https://data.worldbank.org Sylverken AA, Kwarteng A, Twumasi-Ankrah S, Owusu M, Arthur RA, Dumevi RM, et al. The burden of drug resistance tuberculosis in Ghana; results of the First National Survey. PLoS ONE. 2021;16(6):e0252819. 10.1371/journal.pone.0252819 . PubMed PMID: 34111159; PubMed Central PMCID: PMC8191906. Pasipanodya JG, Srivastava S, Gumbo T. Meta-Analysis of Clinical Studies Supports the Pharmacokinetic Variability Hypothesis for Acquired Drug Resistance and Failure of Antituberculosis Therapy. Clin Infect Dis. 2012;55(2):169–77. 10.1093/cid/cis353 . Srivastava S, Pasipanodya JG, Meek C, Leff R, Gumbo T. Multidrug-Resistant Tuberculosis Not Due to Noncompliance but to Between-Patient Pharmacokinetic Variability. J Infect Dis. 2011;204(12):1951–9. 10.1093/infdis/jir658 . Mota L, Al-Efraij K, Campbell JR, Cook VJ, Marra F, Johnston J. Therapeutic drug monitoring in anti-tuberculosis treatment: a systematic review and meta-analysis. Int J Tuberc Lung Dis. 2016;20(6):819–26. 10.5588/ijtld.15.0803 . Perumal R, Naidoo K, Naidoo A, Ramachandran G, Requena-Mendez A, Sekaggya-Wiltshire C, et al. A systematic review and meta-analysis of first-line tuberculosis drug concentrations and treatment outcomes. Int J Tuberc Lung Dis. 2020;24(1):48–64. Perumal R, Naidoo K, Naidoo A, Padayatchi N. Clinical impact of plasma concentrations of first-line antituberculosis drugs. S Afr Med J. 2023;113(3):148–53. 10.7196/SAMJ.2023.v113i3.16761 . Antwi S, Yang H, Enimil A, Sarfo AM, Gillani FS, Ansong D et al. Pharmacokinetics of the First-line Antituberculosis Drugs in Ghanaian Children with Tuberculosis with and without HIV Coinfection. Antimicrob Agents Chemother. 2016;AAC.01701-16. 10.1128/AAC.01701-16 Kwara A, Enimil A, Gillani FS, Yang H, Sarfo AM, Dompreh A, et al. Pharmacokinetics of First-Line Antituberculosis Drugs Using WHO Revised Dosage in Children With Tuberculosis With and Without HIV Coinfection. J Pediatr Infect Dis Soc. 2016;5(4):356–65. 10.1093/jpids/piv035 . Saleri N, Dembélé SM, Villani P, Carvalho ACC, Cusato M, Bonkoungou V, et al. Systemic exposure to rifampicin in patients with tuberculosis and advanced HIV disease during highly active antiretroviral therapy in Burkina Faso. J Antimicrob Chemother. 2012;67(2):469–72. 10.1093/jac/dkr445 . PubMed PMID: 22028201. Maze MJ, Paynter J, Chiu W, Hu R, Nisbet M, Lewis C. Therapeutic drug monitoring of isoniazid and rifampicin during anti-tuberculosis treatment in Auckland, New Zealand. Int J Tuberc Lung Dis Off J Int Union Tuberc Lung Dis. 2016;20(7):955–60. 10.5588/ijtld.15 . .0792 PubMed PMID: 27287650. Meloni M, Corti N, Müller D, Henning L, Gutteck U, von Braun A, et al. Cure of tuberculosis despite serum concentrations of antituberculosis drugs below published reference ranges. Swiss Med Wkly. 2015;145(5152):5152. 10.4414/smw.2015.14223 . Park JS, Lee JY, Lee YJ, Kim SJ, Cho YJ, Yoon HI, et al. Serum Levels of Antituberculosis Drugs and Their Effect on Tuberculosis Treatment Outcome. Antimicrob Agents Chemother. 2015;60(1):92–8. 10.1128/AAC.00693-15 . PubMed PMID: 26459901; PubMed Central PMCID: PMC4704228. Egelund EF, Alsultan A, Peloquin CA. Optimizing the clinical pharmacology of tuberculosis medications. Clin Pharmacol Ther. 2015;9(4). https://doi.org/10.1002/cpt.180 . Kendall EA, Shrestha S, Cohen T, Nuermberger E, Dooley KE, Gonzalez-Angulo L, et al. Priority-Setting for Novel Drug Regimens to Treat Tuberculosis: An Epidemiologic Model. PLOS Med. 2017;14(1):e1002202. 10.1371/journal.pmed.1002202 . Peloquin CA. Therapeutic Drug Monitoring in the Treatment of Tuberculosis. Drugs. 2002;62(15):2169–83. 10.2165/00003495-200262150-00001 . Chideya S, Winston CA, Peloquin CA, Bradford WZ, Hopewell PC, Wells CD, et al. Isoniazid, Rifampin, Ethambutol, and Pyrazinamide Pharmacokinetics and Treatment Outcomes among a Predominantly HIV-Infected Cohort of Adults with Tuberculosis from Botswana. Clin Infect Dis. 2009;48(12):1685–94. 10.1086/599040 . Tappero JW, Bradford WZ, Agerton TB, Hopewell P, Reingold AL, Lockman S, et al. Serum Concentrations of Antimycobacterial Drugs in Patients with Pulmonary Tuberculosis in Botswana. Clin Infect Dis. 2005;41(4):461–9. 10.1086/431984 . WHO. Definitions and reporting framework for tuberculosis – 2013 revision (updated December 2014 and January 2020). Report No; 2013. Pasipanodya JG, McIlleron H, Burger A, Wash PA, Smith P, Gumbo T. Serum drug concentrations predictive of pulmonary tuberculosis outcomes. J Infect Dis. 2013;208(9):1464–73. 10.1093/infdis/jit352 . PubMed PMID: 23901086; PubMed Central PMCID: PMC3789573. World Health Organization, editor. Community involvement in tuberculosis care and prevention: towards partnerships for health: guiding principles and recommendations based on a WHO review. Geneva, Switzerland: World Health Organization; 2008. p. 1. Ministry of Health, Ghana, Standard Treatment, Guidelines. 2017. 7th ed. Ghana National Drugs Programme; 2017. Lixoft SAS. a Simulations Plus company. Monolix 2023R1. 2023. Devaleenal Daniel B, Ramachandran G, Swaminathan S. The challenges of pharmacokinetic variability of first-line anti-TB drugs. Expert Rev Clin Pharmacol. 2017;10(1):47–58. 10.1080/17512433.2017.1246179 . PubMed PMID: 27724114. Sileshi T, Makonnen E, Telele NF, Barclay V, Zumla A, Aklillu E. Variability in plasma rifampicin concentrations and role of SLCO1B1, ABCB1, AADAC2 and CES2 genotypes in Ethiopian patients with tuberculosis. Infect Dis Lond Engl. 2024;56(4):308–19. PubMed PMID: 38315168; PubMed Central PMCID: PMC11134291. VanderWeele TJ, Ding P. Sensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med. 2017;167(4):268–74. 10.7326/M16-2607 . PubMed PMID: 28693043. Bang H, Robins JM. Biometrics. 2005;61(4):962–73. 10.1111/j.1541-0420.2005.00377.x . Doubly Robust Estimation in Missing Data and Causal Inference Models. Houck K, Chakhaia T, Gorvetzian S, Critchley JA, Schechter MC, Magee MJ. Diabetes Mellitus and Tuberculosis Treatment Outcomes: Interaction Assessment Between Hyperglycemia and Human Immunodeficiency Virus in the State of Georgia, 2015–2020. Open Forum Infect Dis. 2023;10(6):ofad255. 10.1093/ofid/ofad255 . Huangfu P, Ugarte-Gil C, Golub J, Pearson F, Critchley J. The effects of diabetes on tuberculosis treatment outcomes: an updated systematic review and meta-analysis. Int J Tuberc Lung Dis Off J Int Union Tuberc Lung Dis. 2019;23(7):783–96. 10.5588/ijtld.18.0433 . PubMed PMID: 31439109. Mekonen H, Negesse A, Dessie G, Desta M, Mihiret GT, Tarik YD, et al. Impact of HIV coinfection on tuberculosis treatment outcomes in Ethiopia: a systematic review and meta-analysis. BMJ Open. 2024;14(7):e087218. 10.1136/bmjopen-2024-087218 . PubMed PMID: 38969385; PubMed Central PMCID: PMC11228389. Wagnew F, Alene KA, Kelly M, Gray D. Undernutrition increases the risk of unsuccessful treatment outcomes of patients with tuberculosis in Ethiopia: A multicenter retrospective cohort study. J Infect. 2024;89(1):106175. 10.1016/j.jinf.2024.106175 . PubMed PMID: 38729526. Clewe O, Wicha SG, De Vogel CP, De Steenwinkel JEM, Simonsson USH. A model-informed preclinical approach for prediction of clinical pharmacodynamic interactions of anti-TB drug combinations. J Antimicrob Chemother. 2018;73(2):437–47. 10.1093/jac/dkx380 . Zhu C, Liu Y, Hu L, Yang M, He ZG. Molecular mechanism of the synergistic activity of ethambutol and isoniazid against Mycobacterium tuberculosis . J Biol Chem. 2018;293(43):16741–50. 10.1074/jbc.RA118.002693 . Sekaggya-Wiltshire C, von Braun A, Lamorde M, Ledergerber B, Buzibye A, Henning L, et al. Delayed Sputum Culture Conversion in Tuberculosis–Human Immunodeficiency Virus–Coinfected Patients With Low Isoniazid and Rifampicin Concentrations. Clin Infect Dis Off Publ Infect Dis Soc Am. 2018;67(5):708–16. 10.1093/cid/ciy . 179 PubMed PMID: 29514175; PubMed Central PMCID: PMC6094003. Hussain Z, Zhu J, Ma X. Metabolism and Hepatotoxicity of Pyrazinamide, an Antituberculosis Drug. Drug Metab Dispos. 2021;49(8):679–82. 10.1124/dmd.121 . .000389 PubMed PMID: 34074731; PubMed Central PMCID: PMC8407665. Malinga TVB, Othman H, Paximadis M, Tiemessen CT, Ramsay M, Hazelhurst S, et al. Characterization of NAT, GST, and CYP2E1 Genetic Variation in Sub-Saharan African Populations: Implications for Treatment of Tuberculosis and Other Diseases. Clin Pharmacol Ther. 2025;117(5):1338–57. 10.1002/cpt.3557 . PubMed PMID: 39829327; PubMed Central PMCID: PMC11993289. Imperial MZ, Nahid P, Phillips PPJ, Davies GR, Fielding K, Hanna D, et al. A patient-level pooled analysis of treatment-shortening regimens for drug-susceptible pulmonary tuberculosis. Nat Med. 2018;24(11):11. 10.1038/s41591-018-0224-2 . Strydom N, Gupta SV, Fox WS, Via LE, Bang H, Lee M, et al. Tuberculosis drugs’ distribution and emergence of resistance in patient’s lung lesions: A mechanistic model and tool for regimen and dose optimization. PLOS Med. 2019;16(4):e1002773. 10.1371/journal.pmed.1002773 . Paal M, Habler K, Northoff B, Vogeser M. Comparative Routine Therapeutic Drug Monitoring of Mycophenolic Acid in human Plasma with HPLC-UV and Isotope Dilution LC-MS/MS. Clin Lab. 2020;66(4). 10.7754/Clin . .Lab.2019.190820 PubMed PMID: 32255280. Pršo K, Žideková N, Porvazník I, Solovič I, Mokrý J, Kertys M. A high-throughput LC–MS/MS method for simultaneous determination of isoniazid, ethambutol and pyrazinamide in human plasma. Rapid Commun Mass Spectrom. 2023;37(2):e9425. 10.1002/rcm.9425 . PubMed PMID: 36329637; PubMed Central PMCID: PMC9787364. Capiau S, Veenhof H, Koster RA, Bergqvist Y, Boettcher M, Halmingh O, et al. Official International Association for Therapeutic Drug Monitoring and Clinical Toxicology Guideline: Development and Validation of Dried Blood Spot–Based Methods for Therapeutic Drug Monitoring. Ther Drug Monit. 2019;41(4):409. 10.1097/FTD.0000000000000643 . Additional Declarations No competing interests reported. Supplementary Files Additionalfile1.docx Additionalfile2.docx Additionalfile3.docx Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 13 May, 2026 Reviews received at journal 12 May, 2026 Reviewers agreed at journal 29 Apr, 2026 Reviews received at journal 20 Apr, 2026 Reviewers agreed at journal 07 Apr, 2026 Reviewers agreed at journal 06 Apr, 2026 Reviewers agreed at journal 05 Apr, 2026 Reviewers invited by journal 03 Apr, 2026 Editor invited by journal 02 Apr, 2026 Editor assigned by journal 02 Apr, 2026 Submission checks completed at journal 01 Apr, 2026 First submitted to journal 01 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9255541","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":618821953,"identity":"d22af857-4c69-41c1-b1b8-d624734c78e9","order_by":0,"name":"Michael Opoku-Mireku","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2klEQVRIiWNgGAWjYFACxgYGhgMgBhsDwwcQxU6KFsYZIIqZKJugWph5QDQhLbrTDrd9+HDmsJw5e1uatM2vbfJ8zAyMHz7m4NZidjuxeeaMG4eNLXuOHZPO7btt2MbMwCw5cxt+Lcw8Hw4nbriR3iad23ObEaiFjZmXkJY/MC2WPbftidPCcAOkJe2YNMOP24lEaWHsOZMO8kuyZW/D7eQ2ZsZmAn5Jf8zw45g1KMQMb/z4c9t2fnvzwQ8f8WiBAwMQwdgGJhuIUA/TwvCHOMWjYBSMglEwsgAAvrVWEgv39/kAAAAASUVORK5CYII=","orcid":"","institution":"University of Ghana","correspondingAuthor":true,"prefix":"","firstName":"Michael","middleName":"","lastName":"Opoku-Mireku","suffix":""},{"id":618821954,"identity":"9dc309e6-cc3f-4ba8-92e6-fed89c2fff65","order_by":1,"name":"Awewura Kwara","email":"","orcid":"","institution":"University of Florida College of Medicine","correspondingAuthor":false,"prefix":"","firstName":"Awewura","middleName":"","lastName":"Kwara","suffix":""},{"id":618821955,"identity":"57e96d7c-d130-47c3-8f5b-b75f81e10968","order_by":2,"name":"Margaret Y. M. Lartey","email":"","orcid":"","institution":"Korle Bu Teaching Hospital","correspondingAuthor":false,"prefix":"","firstName":"Margaret","middleName":"Y. M.","lastName":"Lartey","suffix":""},{"id":618821956,"identity":"154454a1-6fde-4e12-a3b7-918404e367a0","order_by":3,"name":"Charles A. Peloquin","email":"","orcid":"","institution":"University of Florida","correspondingAuthor":false,"prefix":"","firstName":"Charles","middleName":"A.","lastName":"Peloquin","suffix":""},{"id":618821957,"identity":"689cd242-aa28-4f67-93b6-2c6856180bed","order_by":4,"name":"Divine Yao Aseye Amenuke","email":"","orcid":"","institution":"Komfo Anokye Teaching Hospital","correspondingAuthor":false,"prefix":"","firstName":"Divine","middleName":"Yao Aseye","lastName":"Amenuke","suffix":""},{"id":618821958,"identity":"b6298258-cb97-417f-b665-ab4dc01fd836","order_by":5,"name":"Kwasi Adjepong Twum","email":"","orcid":"","institution":"Komfo Anokye Teaching Hospital","correspondingAuthor":false,"prefix":"","firstName":"Kwasi","middleName":"Adjepong","lastName":"Twum","suffix":""},{"id":618821959,"identity":"9feff2b1-4cee-477a-a7c8-420e1460c806","order_by":6,"name":"Priscilla A. Nortey","email":"","orcid":"","institution":"University of Ghana","correspondingAuthor":false,"prefix":"","firstName":"Priscilla","middleName":"A.","lastName":"Nortey","suffix":""},{"id":618821960,"identity":"0a3bc4cb-8ca4-4b7e-8e56-ca2a0df41382","order_by":7,"name":"Alexander Ansah Manu","email":"","orcid":"","institution":"University of Ghana","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"Ansah","lastName":"Manu","suffix":""},{"id":618821961,"identity":"d5da815d-41dd-4e15-b4fd-3eb2cfc06212","order_by":8,"name":"Kwadwo Ansah Koram","email":"","orcid":"","institution":"University of Ghana","correspondingAuthor":false,"prefix":"","firstName":"Kwadwo","middleName":"Ansah","lastName":"Koram","suffix":""}],"badges":[],"createdAt":"2026-03-28 23:23:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9255541/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9255541/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106579736,"identity":"5589bff1-7fe2-430b-92ad-8f8307e9a3d8","added_by":"auto","created_at":"2026-04-10 06:34:02","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":205942,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBar chart showing clinical treatment outcomes by pharmacokinetic phenotype\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Fig.1.png","url":"https://assets-eu.researchsquare.com/files/rs-9255541/v1/494c7cb09f884805d641a5a5.png"},{"id":106727435,"identity":"a8cdb6b8-98ce-40d6-acd1-1d8b4c571083","added_by":"auto","created_at":"2026-04-12 18:39:07","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1545780,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9255541/v1/f4c29058-55c7-40db-9aec-c8f2a8368821.pdf"},{"id":106579733,"identity":"4fea4b5d-94b3-4a15-b8c9-0211a84c5494","added_by":"auto","created_at":"2026-04-10 06:34:02","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":20080,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfile1.docx","url":"https://assets-eu.researchsquare.com/files/rs-9255541/v1/d85151d68a58d53c2690c60b.docx"},{"id":106725940,"identity":"31b8558f-f1b4-4e57-bf25-e0796af939a7","added_by":"auto","created_at":"2026-04-12 18:34:34","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":18497,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfile2.docx","url":"https://assets-eu.researchsquare.com/files/rs-9255541/v1/5415ae90190882a49f649e6f.docx"},{"id":106579734,"identity":"21364776-a8fc-49dd-95c3-58375c340a39","added_by":"auto","created_at":"2026-04-10 06:34:02","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":19583,"visible":true,"origin":"","legend":"","description":"","filename":"Additionalfile3.docx","url":"https://assets-eu.researchsquare.com/files/rs-9255541/v1/8b5768bdd375f7dfefc785e7.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Exposure to the Four First-Line Anti-Tuberculosis Drugs and Treatment Outcomes: A Target Trial Emulation Study in Ghana","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eTuberculosis remains a major public health challenge, with control efforts falling behind global targets. The WHO End TB strategy targets a 50% reduction in the 2015 incidence rate by 2025, however, by 2023 only an 8.3% reduction had been achieved. The mortality rate reduction achieved in 2023 was 23% of the 2015 rate, although the target for 2025 was 75% (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTreatment success rates have improved, reaching 88% for drug-susceptible TB. However, the success rate for drug-resistant TB remains substantially lower at 68%. This disparity underscores the urgent need to prevent the emergence and spread of drug resistance as a key strategy in the global fight against tuberculosis. In sub-Saharan Africa, which bears the highest burden of disease, there has not been much improvement in the treatment success rate in recent years (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e). For example, Ghana\u0026rsquo;s TB treatment success rate has stagnated between 84% and 87% since 2012 (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Although this represents a substantial improvement from the 50% reported in 2000, the lack of further progress in recent years may be due to a complex interplay of factors that warrant urgent investigation.\u003c/p\u003e \u003cp\u003eA plausible consequence of this observation is the increased emergence of resistant strains of \u003cem\u003eMycobacterium tuberculosis\u003c/em\u003e in recent years. Whereas globally, the emergence of resistance among previously treated patients is six times higher compared to treatment-na\u0026iuml;ve patients, it is estimated to be 19 times higher in Ghana (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). The magnitude of the difference in the rate between previously treated and treatment-na\u0026iuml;ve patients suggests a treatment-related process that facilitates acquired resistance. Although nonadherence is often thought to be primarily responsible, studies by Pasipanodya and colleagues point to the person-to-person variability in the pharmacokinetics of the drugs rather than nonadherence, unless beyond a low threshold (\u0026gt;\u0026thinsp;60%) (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e). A systematic review and meta-analysis has established that there is a high prevalence of low plasma concentrations of the first-line anti-TB medications at standard doses, thereby substantiating the role of pharmacokinetic variability in promoting resistance (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). These observations underscore the importance of pharmacokinetic variability as a potential determinant of treatment outcomes, providing a key rationale for the present study.\u003c/p\u003e \u003cp\u003eInvestigations into the relationship between plasma concentrations of the anti-TB medication at standard doses and treatment outcomes are not conclusive. Despite high adherence to the regimen, poor outcomes may still result (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Despite high rates of low plasma concentrations successful TB treatment outcome are common (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). The challenge to addressing the dilemma emanates from methodological limitations of the available studies. The evidence is derived from a limited number of studies conducted in settings that may not fully represent the broader population, often involving small sample sizes. Published TB pharmacokinetic studies from Ghana and other West African countries have important limitations. Many were not designed to evaluate the relationship between drug exposure and treatment outcomes or were conducted only among selected clinical subgroups (\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Several others relied on retrospective clinical data or did not follow the routine programmatic treatment duration (\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e), reducing their generalizability. These limitations are compounded by the fact that potential confounders were rarely measured or adjusted for, as observed by Perumal and team (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Even in studies that adjusted for measured confounders, the resulting estimates are primarily associative rather than causal, reflecting biological signals that may not be robust for causal inference. Furthermore, most existing studies focus on individual drug concentrations, even though tuberculosis treatment outcomes arise from the combined synergistic or antagonistic effects of the full multi-drug regimen. This further constrains the interpretability of prior evidence.\u003c/p\u003e \u003cp\u003eA valid estimation of the effect of the pharmacokinetics of the drugs on treatment outcomes is essential. Understanding whether, and how much, low drug exposure despite treatment adherence influences poor outcomes could explain much of the remaining gap in cure rates and open a practical new path to improvement. This would ensure patients receive doses optimized to achieve a cure, thereby raising success rates, shortening infectious periods, and reducing the emergence of resistance. As has been suggested, optimization of the first-line treatment regimen holds the greatest potential for reducing TB incidence and mortality(\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWe therefore analyzed data from a prospective cohort study in Ghana to answer the following causal question: Among patients with drug-susceptible pulmonary tuberculosis treated with standard WHO-recommended weight-band dosing, what is the 6-month risk of treatment failure or death under a hypothetical dosing strategy that guarantees therapeutic exposure to at least one first-line drug from treatment initiation, compared with a hypothetical dosing strategy that results in subtherapeutic exposure to all four first-line drugs? To estimate this effect, we used target trial emulation with the clone-censor-weight approach.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eTarget Trial Specification and Emulation\u003c/h2\u003e \u003cp\u003eWe emulated a two-arm randomized controlled trial comparing treatment strategies among patients with drug-susceptible pulmonary tuberculosis who survived and remained on treatment until pharmacokinetic (PK) assessment (month 1\u0026ndash;2):\u003c/p\u003e \u003cp\u003e(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) adequate-exposure strategy: therapeutic peak concentrations (C\u003csub\u003emax\u003c/sub\u003e) of at least one of the four first-line drugs. (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) low-exposure strategy: subtherapeutic C\u003csub\u003emax\u003c/sub\u003e of all four drugs.\u003c/p\u003e \u003cp\u003eTime zero was defined as the date of pharmacokinetic measurement. Follow-up extended from pharmacokinetic assessment to the end of standard therapy. We sought to estimate the population-averaged risk difference (RD) in poor treatment outcomes (failure or death), comparing the adequate exposure strategy with the low exposure strategy, under the assumptions of exchangeability, positivity, and consistency.\u003c/p\u003e \u003cp\u003eThe main components of the target trial are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eExposure and Outcome Definitions\u003c/h3\u003e\n\u003cp\u003eThe C\u003csub\u003emax\u003c/sub\u003e of the drugs were categorized as \u0026ldquo;subtherapeutic\u0026rdquo; if they were below the published threshold and \u0026ldquo;therapeutic\u0026rdquo; if equal to or above this threshold (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). The thresholds were 8 \u0026micro;g/ml for rifampicin, 3 \u0026micro;g/ml for isoniazid, 20 \u0026micro;g/ml for pyrazinamide, and 2\u0026micro;g/ml for ethambutol. Because a threshold of 35 \u0026micro;g/ml has been used in clinical practice for pyrazinamide (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e), proportions based on both thresholds are presented for completeness. However, all primary analyses were conducted using the 20 \u0026micro;g/ml threshold. The primary outcome was treatment response, categorized as \u0026ldquo;success\u0026rdquo; for clinical treatment outcomes of cured and treatment completed, or \u0026ldquo;poor\u0026rdquo; for treatment failure and death. Clinical outcome categories were defined using the World Health Organization\u0026rsquo;s standard criteria (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). A patient was considered cured if they had bacteriologically confirmed pulmonary TB at baseline and became smear negative in the final month of treatment, as well as at least one earlier time point. Treatment completion was assigned to patients who finished the full course of therapy without evidence of treatment failure, but for whom bacteriological results in the final month were unavailable or not performed. Treatment failure referred to patients who remained smear positive at month five or later. Patients who died from any cause during treatment were categorized under mortality. Those who interrupted treatment for two or more consecutive months were classified as lost to follow-up. Adherence was defined as the percentage of prescribed doses missed in the month preceding pharmacokinetic sampling. High adherence was categorized as missing fewer than 10% of doses, while low adherence was defined as missing 10% or more.\u003c/p\u003e\n\u003ch3\u003eData source, Eligibility, and Follow-up\u003c/h3\u003e\n\u003cp\u003e Data was obtained from a prospective cohort study conducted at five hospitals (Komfo Anokye Teaching Hospital, Kumasi South Hospital, Holy Family Hospital - Techiman, Suntreso Government Hospital, and Tafo Government Hospital) to investigate the risk factors of poor drug-susceptible tuberculosis treatment outcomes. Patients were enrolled consecutively from 1st August 2022 until the pre-determined sample size was reached on 23rd August 2023.\u003c/p\u003e \u003cp\u003eEligible participants were patients aged\u0026thinsp;\u0026ge;\u0026thinsp;15 years, weighing at least 30 kg, and had rifampicin-susceptible pulmonary tuberculosis confirmed by GeneXpert MTB/RIF. Patients with multidrug-resistant TB, exclusively extrapulmonary TB, or unknown prior treatment history were excluded. Previously treated patients were included only if their last episode had ended in cure or treatment completion (n\u0026thinsp;=\u0026thinsp;1 out of the analytic cohort of 120 (0.8%). Peak plasma concentrations of rifampicin, isoniazid, pyrazinamide, and ethambutol were measured once between months 1 and 2 of treatment. Participants were followed monthly until the end of the standard 6-month regimen, with treatment outcomes recorded according to national guidelines.\u003c/p\u003e\n\u003ch3\u003eSample size and Recruitment\u003c/h3\u003e\n\u003cp\u003eA sample of size 164 was enrolled. This number was based on an assumed 5% rate of treatment failure or death among adequately exposed patients (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e), a design effect of 2.0 to account for clustering by facility, 80% power, and a two-sided α of 0.05, with an additional 20% added for missing pharmacokinetic data or loss to follow-up. Recruitment continued until the target was reached: 92 patients from Komfo Anokye Teaching Hospital, 37 from Holy Family Hospital - Techiman, 21 from Kumasi South Hospital, and 7 each from Suntreso and Tafo Government Hospitals.\u003c/p\u003e\n\u003ch3\u003eDrug Dosing and Administration\u003c/h3\u003e\n\u003cp\u003eAll participants received daily fixed-dose combinations of the four first-line anti-TB drugs during the intensive phase (first 2 months) and rifampicin plus isoniazid during the continuation phase (next 4 months), administered by DOTS in line with WHO recommendations (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Dosing followed standard weight bands: 30\u0026ndash;39 kg (300 mg rifampicin, 150 mg isoniazid, 550 mg ethambutol, 800 mg pyrazinamide), 40\u0026ndash;54 kg (450 mg rifampicin, 225 mg isoniazid, 825 mg ethambutol, 1,200 mg pyrazinamide), and \u0026ge;\u0026thinsp;55 kg (600 mg rifampicin, 300 mg isoniazid, 1,100 mg ethambutol, 1,600 mg pyrazinamide) (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003ePharmacokinetic Data Collection and Analysis\u003c/h2\u003e \u003cp\u003eAt month 1 or 2, 4 ml of blood was collected two and four hours after dosing. Participants fasted overnight (\u0026ge;\u0026thinsp;8 hours), took their observed dose, and received a light meal 30 minutes post-dose. Blood samples were transferred into heparinized tubes, centrifuged at 3,000 g for 10 minutes at 4\u0026deg;C, and the resulting plasma stored at \u0026minus;\u0026thinsp;80\u0026deg;C until analysis. All specimens were subsequently shipped on dry ice to the Infectious Disease Pharmacokinetic Laboratory at the University of Florida for quantification of drug concentrations. Quantification was done using a validated high-performance liquid chromatography\u0026ndash;tandem mass spectrometry (LC-MS/MS) assay. The method demonstrated precision with percentage relative standard deviations between 0.2% and 6.6% across high, medium, and low quality-control levels, and accuracy ranging from 99.03% to 99.67% for all four drugs. Measured concentrations ranged from 0.012\u0026ndash;14.382 \u0026micro;g/mL for rifampicin, 0.052\u0026ndash;5.162 \u0026micro;g/mL for isoniazid, 0.122\u0026ndash;54.142 \u0026micro;g/mL for pyrazinamide, and 0.0052\u0026ndash;5.892 \u0026micro;g/mL for ethambutol. Maximum plasma concentrations were estimated by maximum likelihood using the 2-hour and 4-hour post-dose samples modeled in Monolix 2023 (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). Concentrations below the lower limits of quantification (0.25 \u0026micro;g/mL for rifampicin, 0.15 \u0026micro;g/mL for isoniazid, 0.50 \u0026micro;g/mL for pyrazinamide, and 0.05\u0026micro;g/ml ethambutol) were excluded from the C\u003csub\u003emax\u003c/sub\u003e calculation to prevent unreliable model inputs that would bias C\u003csub\u003emax\u003c/sub\u003e estimates downward.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSpecification of the target trial and its emulation using observational pharmacokinetic data\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eComponent\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTarget Trial (Ideal but infeasible)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEmulation using observational data\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEligibility\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePatients aged 15 years or older with a minimum body weight of 30kg, diagnosed with rifampicin-susceptible TB using the GeneXpert MTB/Rif.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSame, but must have survived and remained on treatment until pharmacokinetic sampling (complete-case restriction)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eTreatment strategies\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStrategy A: Dosing regimen that guarantees a therapeutic C\u003csub\u003emax\u003c/sub\u003e of at least one of the four first-line drugs from day 0 onward\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStrategy A: Observed month 1\u0026ndash;2 C\u003csub\u003emax\u003c/sub\u003e shows therapeutic level for at least one drug (adequate-exposure phenotype)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eB: Dosing regimen that results in subtherapeutic C\u003csub\u003emax\u003c/sub\u003e of all four first-line drugs from day 0 onward\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eB: Observed month 1\u0026ndash;2 C\u003csub\u003emax\u003c/sub\u003e subtherapeutic for all four drugs (low-exposure phenotype)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAssignment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRandomization at baseline (day 0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClone-censor-weight: each participant cloned and assigned to both strategies; clones censored when their observed exposure violates the assigned strategy\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFollow-up start\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFrom day 0 of treatment until month 6, death, treatment failure, or loss to follow-up\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSame, with time zero set at pharmacokinetic sampling (month 1\u0026ndash;2) and follow-up to month 6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOutcome\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTreatment failure or death by month 6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSame\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCausal contrast\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIntention-to-treat effect\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePer-protocol analogue via clone-censor-weighting\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStatistical Analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRisk difference/ risk ratio\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eClone-censor-weight estimation for per-protocol analogue effect of static regimes. Sensitivity analyses included IPWRA, plain IPW, and E-value.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKey assumptions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRandomization ensures exchangeability\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eConditional exchangeability, positivity, and consistency given measured covariates (age, sex, weight, HIV, diabetes, smoking, adherence, alcohol, diet, smear grade).\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIPW, Inverse Probability Weighting; IPWRA, Inverse Probability Weighted Regression Adjustment\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStatistical Analysis and Causal Inference\u003c/h3\u003e\n\u003cp\u003eThe pharmacokinetic phenotype reflects a final common pathway of diverse upstream factors, including body weight, malabsorption, drug-drug interactions, genetic variability, adherence, among others, many of which cannot be fully measured in routine settings (\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e). We estimated the per-protocol effect of sustained adequate versus low pharmacokinetic exposure using standardized static-regime estimation with baseline cloning and artificial censoring. Each participant was duplicated at baseline and assigned to one of two static strategies: therapeutic C\u003csub\u003emax\u003c/sub\u003e of at least one of the four first-line drugs versus subtherapeutic C\u003csub\u003emax\u003c/sub\u003e of all four drugs. Clones whose assigned strategy was incompatible with the observed month 1\u0026ndash;2 pharmacokinetic phenotype were artificially censored at baseline and excluded from follow-up. Compatible clones were retained in the analysis and weighted by the inverse of the estimated probability of compatibility, conditional on baseline covariates (age, sex, weight, HIV co-infection, diabetes comorbidity, current smoking, adherence, alcohol use, dietary diversity, and baseline smear grade). To reduce variance, weights were stabilized by dividing by the marginal probability of compatibility and winsorized at the 1st and 99th percentiles.\u003c/p\u003e \u003cp\u003eThis procedure standardizes outcomes to the baseline covariate distribution and estimates the population-averaged risk difference and risk ratio, and adheres to each exposure phenotype, independent of upstream determinants of exposure. Risk differences and risk ratios were obtained from weighted binomial generalised linear models with identity and log links, respectively, with robust variance estimation clustered by participant. The number needed to treat (NNT) was calculated as the reciprocal of the primary risk difference (RD\u0026thinsp;=\u0026thinsp;25.6%), yielding an NNT of 4 (95% CI: 2 to 18), indicating that shifting 4 patients from the low-exposure phenotype to adequate exposure would prevent one additional case of treatment failure or death.\u003c/p\u003e\n\u003ch3\u003eSensitivity Analysis\u003c/h3\u003e\n\u003cp\u003eTo assess the robustness of the primary findings, we conducted three sensitivity analyses. First, we used augmented inverse probability weighted regression adjustment (IPWRA) to estimate the average treatment effect (ATE) of the low-exposure phenotype versus adequate exposure, adjusting for baseline covariates (age, sex, weight, HIV co-infection, diabetes comorbidity, current smoking, adherence, alcohol use, dietary diversity, and smear grade) with robust standard errors. This doubly robust approach remains consistent if either the treatment or outcome model is correctly specified. We also examined covariate balance using standardized mean differences before and after weighting. Raw standardized mean differences (SMD) ranged from \u0026minus;\u0026thinsp;0.90 to +\u0026thinsp;0.21 (largest imbalance for weight). After weighting, all SMDs were reduced, with most falling within |\u0026lt;0.30| (maximum absolute SMD\u0026thinsp;=\u0026thinsp;0.41 for alcohol; range\u0026thinsp;\u0026minus;\u0026thinsp;0.30 to +\u0026thinsp;0.41). Variance ratios remained acceptable (0.45\u0026ndash;1.62), supporting effective reweighting. The treated group's effective sample size increased to 54.3 (from raw n\u0026thinsp;=\u0026thinsp;21), while the control group's effective sample size decreased to 53.7 (from raw n\u0026thinsp;=\u0026thinsp;87), reflecting the reweighting process to achieve balance.\u003c/p\u003e \u003cp\u003eSecond, we applied plain inverse probability weighting (IPW) to estimate the ATE, using the same covariates and robust standard errors, with an oversampling strategy to handle potential positivity violations. Weighted balance was again assessed via standardized mean differences.\u003c/p\u003e \u003cp\u003eThird, we calculated E-values to quantify the strength of unmeasured confounding required to nullify the primary risk difference estimate (RD\u0026thinsp;=\u0026thinsp;25.6%). The E-value assesses how strongly an unmeasured confounder would need to be associated with both the exposure and outcome (on the risk ratio scale) to explain away the observed association, assuming no true effect (\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAll sensitivity analyses were restricted to patients with complete follow-up (i.e., uncensored patients) and utilized Stata\u0026rsquo;s \u0026ldquo;teffects\u0026rdquo; suite with robust variance estimation. Exploratory Firth-penalized Poisson regression models were fitted to examine individual-drug associations under sparse data and potential separation; these were regarded as hypothesis-generating only and were not used for primary causal inference. Results are shown for completeness only in Additional file 3. All analyses were performed in Stata 17.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eLimitations of the estimand\u003c/h2\u003e \u003cp\u003eThe clone-censor-weight approach identifies population-averaged effects attributable to the observed pharmacokinetic phenotype, independent of its upstream causes. It does not estimate the effects of hypothetical interventions like therapeutic drug monitoring-guided dosing. Key assumptions include no unmeasured confounding of phenotype-outcome and correct trial specification.\u003c/p\u003e \u003c/div\u003e"},{"header":"RESULTS","content":"\u003cp\u003eOf the 164 patients enrolled, 17 (10.4%) were lost to follow-up, and 27 (16.5%) did not have complete four-drug pharmacokinetic measurements at months 1\u0026ndash;2, leaving 120 patients (73.2%) for the emulated target trial and all primary analyses. Patients excluded because of loss to follow-up or incomplete pharmacokinetic data (n\u0026thinsp;=\u0026thinsp;44) were broadly similar to the analytic sample (n\u0026thinsp;=\u0026thinsp;120). The largest standardized mean differences were 0.47 for baseline weight, 0.46 for treatment facility, and 0.39 for dietary diversity score; all other characteristics had standardized differences\u0026thinsp;\u0026lt;\u0026thinsp;0.25 (Additional file 1). These same variables also had statistically significant p-values. These modest imbalances indicate that selection into the final analytic cohort introduced limited bias.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e summarizes the baseline characteristics of participants by their pharmacokinetic phenotype. Of the 120 participants, 85 (70.8%) were male, 81 (67.5%) had a high school education, 77 (64.2%) were employed, 65 (54.2%) were unmarried, and 65 (54.2%) sought care at the Komfo Anokye Teaching Hospital. Mean (SD) age was 41.7 years (13.5), median weight was 58.5 kg (IQR: 52.4, 64.0), and median dietary diversity score was 7.0 (IQR: 5.4, 9.6) out of a maximum of 10. While 20.0% (24/120) were HIV co-infected, 7.5% (9/120) had diabetes, and 38.3% (46/120) were on non-TB medications. Thirteen out of 120 (10.8%) missed 10% or more doses within the first two months of treatment initiation.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics by Pharmacokinetic Phenotype\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eCharacteristics\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003ePharmacokinetic Phenotype\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTotal, N\u0026thinsp;=\u0026thinsp;120\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAdequate exposure n\u0026thinsp;=\u0026thinsp;96\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLow exposure n\u0026thinsp;=\u0026thinsp;24\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAge in years, mean (SD)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e41.7 (13.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e42.0 (13.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e40.5 (12.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eWeight in kilograms, median (IQR)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e58.5 (52.4, 64.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e59.0 (53.0, 65.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e56.0 (49.0, 62.1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eDietary diversity Score, median (IQR)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7.0 (5.4, 9.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6.9 (5.4, 9.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e7.6 (4.9, 9.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHousehold size, median (IQR)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.0 (3.0, 7.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.0 (3.0, 6.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.0 (2.5, 7.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eMale sex, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e85 (70.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e68 (70.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e17 (70.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHighest Educational Level, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo formal Education\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e23 (19.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20 (20.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3 (12.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrimary School\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e9 (7.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9 (9.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh School\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e81 (67.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e62 (64.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e19 (79.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTertiary\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e7 (5.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5 (5.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2 (8.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEmployment status, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnemployed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e32 (26.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e24 (25.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8 (33.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEmployed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e77 (64.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e63 (65.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e14 (58.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8 (6.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6 (6.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2 (8.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther#\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3 (2.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3 (3.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0 (0.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNot married, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65 (54.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e51 (53.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e14 (58.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eFacility, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKomfo Anokye Teaching Hospital\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e65 (54.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e53 (55.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e12 (50.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuntreso Government Hospital\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4 (3.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3 (3.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1 (4.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKumasi South Hospital\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11 (9.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e9 (9.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2 (8.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTafo Government Hospital\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4 (3.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3 (3.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1 (4.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHoly Family Hospital, Techiman\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e36 (30.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e28 (29.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8 (33.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCurrent alcohol use, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e21 (17.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e17 (17.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4 (16.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCurrent smoker, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e8 (6.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7 (7.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1 (4.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eNon-adherent (\u0026ge;\u0026thinsp;10% of doses missed), n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13 (10.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10 (10.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3 (12.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eConcomitant non-TB drugs, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e46 (38.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e38 (39.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e8 (33.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eHIV Co-infected n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e24 (20.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e21 (21.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3 (12.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eDiabetes Co-morbidity, n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e9 (7.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e8 (8.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1 (4.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSevere Smear Grade (3+), n (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e27 (23.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20 (22.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e7 (29.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eSD, Standard Deviation; IQR, Interquartile Range\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003csup\u003e#\u003c/sup\u003eIncludes retirees and apprentices\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e*Adequate exposure\u0026thinsp;=\u0026thinsp;therapeutic C\u003csub\u003emax\u003c/sub\u003e of at least one drug\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u0026dagger;Low exposure\u0026thinsp;=\u0026thinsp;subtherapeutic C\u003csub\u003emax\u003c/sub\u003e of all four drugs\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e summarizes the distribution of C\u003csub\u003emax\u003c/sub\u003e for the four first-line anti-TB drugs. Subtherapeutic C\u003csub\u003emax\u003c/sub\u003e levels were observed in 93.3% of participants for rifampicin, 86.7% for isoniazid, and 53.3% for ethambutol. For pyrazinamide, using the widely reported threshold of 20 \u0026micro;g/ml, 20.0% of participants had subtherapeutic C\u003csub\u003emax\u003c/sub\u003e. However, when the higher clinical practice threshold of 35 \u0026micro;g/ml is applied, this proportion increases markedly to 80.0%. Overall, 3.3% (4/120) of participants had therapeutic concentrations for all drugs, whereas 20.0% (24/120) exhibited subtherapeutic concentrations for all four medicines.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDistribution of peak plasma concentration of anti-TB drugs for participants\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrug\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePeak Concentration, \u0026micro;g/ml median (IQR)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003eSubtherapeutic C\u003csub\u003emax\u003c/sub\u003e n (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eTherapeutic C\u003csub\u003emax\u003c/sub\u003e n (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eRifampicin\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.1 (2.0, 5.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e112 (93.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8 (6.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eIsoniazid\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.6 (0.7, 2.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e104 (86.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e16 (13.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePyrazinamide (\u0026lt;\u0026thinsp;20 \u0026micro;g/ml)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24.9 (12.8, 32.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e24 (20.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96 (80.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003ePyrazinamide (\u0026lt;\u0026thinsp;35 \u0026micro;g/ml)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24.9 (12.8, 32.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96 (80.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e24 (20.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eEthambutol\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.9 (1.2, 2.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e64 (53.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e56 (46.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eAll four drugs\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c4\" namest=\"c3\"\u003e \u003cp\u003e24 (20.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e96 (80.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eC\u003csub\u003emax,\u003c/sub\u003e Peak Plasma Concentration; IQR, Interquartile Range; IQR, Interquartile Range\u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure\u0026nbsp;1: Bar chart showing clinical treatment outcomes by pharmacokinetic phenotype\u003c/b\u003e \u003c/p\u003e \u003cp\u003eClinical outcomes differed between the adequate exposure phenotype and the low exposure phenotype (Fig.\u0026nbsp;1). Among patients with adequate exposure (n\u0026thinsp;=\u0026thinsp;96), 95.8% achieved treatment success (70.8% cured; 25.0% completed treatment), and only 4.1% experienced treatment failure or death. In contrast, among those with subtherapeutic concentrations of all four drugs (n\u0026thinsp;=\u0026thinsp;24), treatment success declined to 66.6% (58.3% cured; 8.3% completed), while poor outcomes increased to 33.4% (29.2% failure; 4.2% death) (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). The distribution of baseline covariates by treatment response is presented in Additional file 2.\u003c/p\u003e \u003cp\u003eThe results in Table\u0026nbsp;4 show a strong and consistent link between subtherapeutic C\u003csub\u003emax\u003c/sub\u003e of all four first-line anti-TB drugs and a much higher risk of treatment failure or death. In the emulated target trial, the 6-month risk of poor treatment response was 33.3% (8 of 24 patients) under the low-exposure strategy compared to 4.2% (4 of 96 patients) under the adequate-exposure strategy. This corresponds to a large and clinically significant causal effect: a crude risk difference of 29.1 percentage points.\u003c/p\u003e \u003cp\u003eIn the primary analysis (weighted binomial regression with identity link, accounting for censoring via inverse probability of censoring weights), patients with subtherapeutic levels of all four drugs had an absolute risk increase of 25.6 percentage points (95% CI: 5.7% to 45.6%; p\u0026thinsp;=\u0026thinsp;0.012) compared to those with adequate levels of at least one drug. This translates to an estimated 3\u0026ndash;4 poor outcomes per 100 patients with therapeutic drug levels, rising to approximately 29\u0026ndash;50 per 100 among those with low levels of all four drugs. This difference is clearly meaningful. Using a log-linked binomial model (risk ratio scale), the same exposure was associated with an 8.6-fold higher risk of failure or death (RR\u0026thinsp;=\u0026thinsp;8.64, 95% CI: 2.34\u0026ndash;31.92; p\u0026thinsp;=\u0026thinsp;0.001), meaning patients with low exposure to all four drugs were more than eight times as likely to experience a poor outcome compared to those with better drug levels.\u003c/p\u003e \u003cp\u003eRegarding sensitivity analyses, both IPWRA and plain IPW yielded identical average treatment effects of 19.9 percentage points higher risk (95% CI: 2.3%\u0026ndash;37.5%; p\u0026thinsp;=\u0026thinsp;0.027), after adjustment for age, sex, weight, HIV status, diabetes, smoking, alcohol use, adherence, dietary diversity, and baseline smear grade. The consistency across the two approaches strengthens confidence that the association is not explained by measured confounding or model misspecification (\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe third sensitivity analysis was the E-value. The point estimate was 15.48, meaning that any unmeasured confounder(s) would need to be associated with both exposure and outcome by a risk ratio of at least 15.5 in both directions to nullify the observed effect. The corresponding E-value for the lower bound of the confidence interval was 4.96, indicating that an unmeasured confounder would still require a substantial risk ratio of approximately 5.0. These high E-values suggest that the association is unlikely to be explained entirely by unmeasured confounding, providing additional support for a causal interpretation of the relationship between low all-four drug exposure and increased risk of poor treatment outcomes.\u003c/p\u003e \u003cp\u003eOverall, these complementary analyses showed that low exposure to all four first-line anti-TB drugs is associated with an elevated risk of treatment failure or death, approximately 20\u0026ndash;26 percentage points higher in absolute terms, or more than eight times higher in relative terms.\u003c/p\u003e \u003cp\u003e \u003cb\u003eTable\u0026nbsp;4: Risk of Treatment Failure or Death Associated with All Four Subtherapeutic Anti-TB Drugs Exposure\u003c/b\u003e \u003c/p\u003e\u003ctable border=\"0\" cellspacing=\"3\" cellpadding=\"0\" width=\"605\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eModel Type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eEstimand\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eEffect [95% CI]\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eP-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAdjustment\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrimary Analyses (n = 108)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eWeighted Binomial regression (identity link)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRisk Difference \u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e25.6% [5.7%, 45.6%]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.012\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eIPCW-weighted analysis restricted to uncensored cases, with robust standard errors (clustered by participant ID)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eWeighted Binomial regression (log link)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRisk Ratio\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e8.64 [2.34, 31.92]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eIPCW-weighted analysis restricted to uncensored cases, with cluster-robust standard errors by participant ID\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eSensitivity Analyses (n = 108)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eInverse Probability Weighting Regression Adjustment\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAverage Treatment Effect (%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e19.9% [2.3%, 37.5%]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRobust SE, adjusted for baseline covariates\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eInverse Probability Weighting\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAverage Treatment Effect (%)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e19.9% [2.3%, 37.5%]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.027\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRobust SE, oversample strategy, adjusted for baseline covariates\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003eE-value (for primary RD = 25.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eE-value\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e15.48 [Lower bound: 4.96]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAn unmeasured confounder would need RR ≥15.5 with both exposure \u0026amp; outcome to nullify RD\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eAbbreviation\u003c/strong\u003e: CI, Confidence Interval;IPCW, Inverse Probability of Censoring Weight; IPW, Inverse Probability Weighting; IPWRA, Inverse Probability Weighting Regression Adjustment; SE, Standard Error\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eNotes\u003c/strong\u003e: Log-binomial provides a multiplicative RR scale. Covariates were\u0026nbsp;age, sex, weight, HIV, diabetes, smoking, alcohol use, adherence, dietary diversity, and smear grade. All\u0026nbsp;models restricted to uncensored cases (n=108); robust standard errors throughout.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNo poor outcomes were observed among patients with therapeutic rifampicin C\u003csub\u003emax\u003c/sub\u003e. However, small numbers precluded formal multivariable estimation (Additional file 2). Low C\u003csub\u003emax\u003c/sub\u003e was associated with a\u0026nbsp;7.3-fold (95% CI: 1.89, 28.33) higher risk of treatment failure or death than adequate\u0026nbsp;C\u003csub\u003emax\u003c/sub\u003e.\u0026nbsp;Subtherapeutic C\u003csub\u003emax\u003c/sub\u003e for pyrazinamide (IRR = 16.7; 95% CI: 4.0–69.4, p \u0026lt; 0.001) and ethambutol (IRR = 4.1; 95% CI: 1.2–13.9, p = 0.024) were each strongly associated with poor treatment response. There wasn't enough evidence of an association between isoniazid C\u003csub\u003emax\u003c/sub\u003e and treatment response (Additional file 3). \u0026nbsp;\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this prospective cohort study of adults with drug-susceptible pulmonary tuberculosis in Ghana, subtherapeutic concentrations of all four first-line anti-TB drugs were associated with an increased risk of treatment failure or death. This contrasted with a hypothetical strategy in which therapeutic concentrations of at least one drug were achieved at treatment initiation. The findings were consistent across primary and sensitivity analyses. Moreover, the E-value indicates that an unmeasured confounder would need to be associated with both the exposure and the outcome by risk ratios greater than 15.5 (lower bound 5.0) to nullify the observed effect. Unmeasured confounding greater than 15.5 is implausible since confounders typically exert far smaller effects, but cannot be ruled out (29–32). Even the lower confidence limit (5.7%) represents a significant harm. Our findings suggest that averting the low-exposure phenotype could prevent one poor outcome per 4 patients (NNT=3.9), supporting research into pharmacokinetics-guided interventions.\u003c/p\u003e\n\u003cp\u003eOur findings are biologically plausible, underscored by the synergistic effect of the drugs in the standard regimen. The clinical efficacy is largely preserved if at least one companion drug achieves adequate exposure, even if the others have low exposure. This may explain the common paradoxical relationship between high overall treatment success rates and low drug concentrations\u0026nbsp;(4,8). However, low exposure to all four drugs disrupts the synergistic bactericidal and sterilizing activity of the regimen, rendering these patients at an elevated risk of failure, death, relapse, and selection of drug-resistant strains\u0026nbsp;(33,34). Ghana’s disproportionately high (19-fold) risk of resistance among previously treated patients compared to treatment-naïve ones likely reflects this phenomenon\u0026nbsp;(1,3).\u003c/p\u003e\n\u003cp\u003eThe high prevalence of sub-therapeutic C\u003csub\u003emax\u003c/sub\u003e levels for the four drugs, despite high adherence, is consistent with previous studies in sub-Saharan African populations, where malnutrition, drug-drug interactions, and genetic variations affecting drug metabolism are common\u0026nbsp;(7,35). Clinically, this highlights the fact that patients may fail therapy even when they take their medication correctly. This is because standard doses based on weight bands may not achieve the desired therapeutic exposure in all patients. Another potential contributor to subtherapeutic exposure across all four drugs is inter-individual variability in drug metabolism. Genetic polymorphisms affecting drug-metabolizing enzymes and transporters, such as \u003cem\u003eNAT2\u003c/em\u003e polymorphisms governing isoniazid acetylation, transporter variants such as \u003cem\u003eABCB1\u003c/em\u003e affecting rifampicin disposition, and hepatic enzymatic pathways involved in pyrazinamide metabolism, are prevalent in African populations and can lead to low plasma concentrations in some individuals\u0026nbsp;(26,36,37). While we did not assess pharmacogenetic markers or metabolite levels, such differences could partly explain the observed phenotype and warrant further investigation in larger cohorts with pharmacokinetic and genetic data.\u003c/p\u003e\n\u003cp\u003eOur results underscore the need to integrate pharmacokinetic considerations in routine TB care. Implementing therapeutic drug monitoring (TDM) in resource-limited settings like Ghana poses challenges, including the need for advanced laboratory infrastructure and trained personnel. Our study’s use of LC-MS/MS for plasma concentration quantification, conducted at a specialized facility, underscores these logistical barriers. Pragmatic clinical proxies such as early sputum non-conversion, persistent symptoms despite adherence, and markers of severe disease (cavitation, low body weight) can prompt dose intensification\u0026nbsp;(38,39). Alternatively, population pharmacokinetic models that incorporate readily available covariates (weight, HIV status, diabetes, age) can be used to guide empiric dose adjustments\u0026nbsp;(10).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough LC–MS/MS remains the gold standard for drug quantification due to its high sensitivity and specificity, high-performance liquid chromatography with ultraviolet detection (HPLC-UV) may represent a more affordable alternative in resource-constrained settings\u0026nbsp;(40,41). In addition, dried blood spot (DBS) sampling offers a practical option for therapeutic drug monitoring in remote or decentralized settings, despite being technically more demanding than plasma-based assays\u0026nbsp;(42). These provide opportunities worth exploring to enable the introduction of TDM into routine TB care.\u003c/p\u003e\n\u003cp\u003eAt the population level, the stagnation of treatment success rates in Ghana and similar settings may reflect prevalent low drug exposure rather than programmatic failure alone. If a critical mass of patients systematically achieves subtherapeutic levels of the drugs while on the recommended regimen, even an effective Directly Observed Treatment, Short-course (DOTS) programme cannot fully achieve its expected outcomes. Improving treatment outcomes will require moving beyond adherence-based interventions alone.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePublic health programmes need to consider differentiated dosing for at-risk subtherapeutic populations, such as those with comorbidities, or a regimen with optimized doses of one or two of the drugs. In the case of the latter, our study suggested that pyrazinamide and ethambutol concentrations signaled an effect on poor response. It is worth noting that these proposals remain exploratory. This estimand demonstrates that preventing the low-exposure phenotype would substantially reduce poor treatment outcomes. However, it does not establish whether TDM, higher-dose regimens, or any other specific intervention can reliably achieve this phenotype shift in routine clinical practice. Definitive evidence of effectiveness and feasibility will require further randomized studies explicitly designed to test these implementation strategies locally.\u003c/p\u003e\n\u003cp\u003eTo comprehensively address this under-recognized pharmacokinetic barrier to TB elimination, we recommend piloting TDM in high-risk patients at teaching hospitals, accelerating research into higher-dose regimens, and building sustainable local capacity for pharmacokinetic research.\u003c/p\u003e\n\u003cp\u003eThe strengths of this study include its rigorous target trial emulation framework, clearly defining eligibility, time zero, exposure strategies, follow-up, and estimand, reducing common biases like immortal time bias and improving transparency over standard observational analyses. Drug exposure was measured directly via observed C\u003csub\u003emax\u003c/sub\u003e (not proxies like dose or weight), providing strong biological grounding. Confounding was addressed with stabilized inverse probability weighting and formal balance diagnostics, yielding population-averaged estimates relevant to clinical and policy decisions. Both absolute (risk difference, NNT) and relative (risk ratio) effect measures were reported, enhancing interpretability and relevance. The exposure contrast focused on a severe, biologically extreme phenotype (all four drugs subtherapeutic), making the large observed effect more plausible as a pharmacologic signal. Together, these features strengthen internal validity and support cautious causal interpretation within the study’s assumptions and constraints. Some limitations existed despite these strengths.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough inverse probability weighting approximated exchangeability, residual confounding remains possible. Unmeasured markers of disease severity (such as cavitary disease, radiographic extent, serum albumin, malabsorption, or inflammatory burden) may influence both exposure and outcomes. While measured covariates achieved acceptable balance post-weighting, causal interpretation depends on the assumption that all important confounders were captured. Exposure was defined at months 1–2, with time zero at pharmacokinetic sampling. The effect, therefore, applies only to patients who survived until sampling. Early deaths or failures before measurement were excluded, limiting generalizability and introducing potential survivor selection bias. A single C\u003csub\u003emax\u003c/sub\u003e measurement may not fully capture longitudinal exposure. Within-person variability, assay error, or absorption fluctuations could cause misclassification, likely biasing estimates toward the null if nondifferential, though differential misclassification cannot be excluded. Despite stabilized weights and improved balance, the small number of patients with the low-exposure phenotype raises concerns about positivity and weight instability. Extreme weights were truncated, but finite-sample variability may affect precision. Approximately one-quarter of eligible participants were excluded due to missing pharmacokinetic or outcome data. Baseline differences between included and excluded patients suggest potential selection bias, limiting external validity. Finally, the exposure contrast reflects a severe composite phenotype (all four drugs subtherapeutic) rather than drug-specific effects. Findings should not be interpreted as the isolated causal effect of any single agent, but rather as the estimated impact of avoiding profound multidrug subtherapeutic exposure. Despite these limitations, the clearly defined time zero, explicit causal estimand, and robust weighting framework strengthen internal validity relative to conventional regression-based analyses.\u003c/p\u003e"},{"header":"CONCLUSION","content":"\u003cp\u003eGhana and most sub-Saharan African countries have not been able to acheive TB treatment success above 85\u0026ndash;90% despite generally high reported adherence and well-functioning programmes. Our findings suggest that some individuals do not achieve therapeutic concentrations of any of the four first-line drugs on currently recommended doses. Recognizing and correcting this hidden pharmacokinetic shortfall would directly increase the chance of cure for those individuals, shorten infectiousness, and reduce the emergence of drug-resistant strains. These phenotype effects suggest that interventions averting subtherapeutic exposure, such as TDM-guided dosing or empiric high-dose regimens, merit randomized evaluation in high-burden settings like Ghana.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was conducted in accordance with the Declaration of Helsinki. Ethical clearance for the study was received from the\u0026nbsp;Ghana Health Service Ethics Review Committee (GHS-ERC 002/02/21) and the KATH Ethics Committee (KATH IRB/AP/023/21 and KATH IRB/CR01/023/22).\u0026nbsp;Written informed consent was obtained from all adult participants aged 18 years and above.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor participants under the age of 18 (minors), written informed consent was obtained from their parents or legal guardians, and assent was also obtained from the minors themselves.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset contains sensitive personal health information from human participants, including identifiable details such as dates of treatment, clinical measurements, and demographic data. Public release would violate participant confidentiality and contravene ethical approval conditions and data protection regulations. The data are therefore available upon reasonable request from the corresponding author, MOM.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp id=\"_Toc60734491\"\u003eThe study was funded by the Fogarty International Centre of the National Institutes of Health, US, under the UG-Florida Academic Partnership project (D43 TW010055). The Infectious Disease Pharmacokinetic Laboratory at the University of Florida performed the drug assays, as part of MOM’s PhD research.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors’ Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was conceptualized by M.O.M., M.Y.M.L., A.K., C.A.P., and K.A.K. Data collection was supervised by M.O.M. Data analysis was conducted by M.O.M., with methodological support and guidance from A.A.M. and K.A.K. Data interpretation was performed by D.A.Y.A., M.O.M., P.A.N., K.A.T and KAK. K.A.T interpreted radiographic images where available. Pharmacokinetic analysis was supervised by C.A.P., with the involvement of M.O.M. The first draft of the manuscript was written by M.O.M. All authors reviewed and edited the manuscript, with high-level critical review provided by M.Y.M.L., A.K., C.A.P., A.A.M., P.A.N., and K.A.K. All authors read and approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors would like to acknowledge the study participants for their participation in the study, as well as the field team for their role in data collection.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTrial registration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClinical trial number: not applicable.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eWHO. Global Tuberculosis Report 2024 [Internet]. 2024 [cited 2025 Jan 23]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.who.int/teams/global-tuberculosis-programme/tb-reports/global-tuberculosis-report-2024\u003c/span\u003e\u003cspan address=\"https://www.who.int/teams/global-tuberculosis-programme/tb-reports/global-tuberculosis-report-2024\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Bank Group. World Bank Open Data [Internet]. 2025 [cited 2025 Mar 10]. World Bank Open Data. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://data.worldbank.org\u003c/span\u003e\u003cspan address=\"https://data.worldbank.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSylverken AA, Kwarteng A, Twumasi-Ankrah S, Owusu M, Arthur RA, Dumevi RM, et al. The burden of drug resistance tuberculosis in Ghana; results of the First National Survey. PLoS ONE. 2021;16(6):e0252819. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0252819\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0252819\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 34111159; PubMed Central PMCID: PMC8191906.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePasipanodya JG, Srivastava S, Gumbo T. Meta-Analysis of Clinical Studies Supports the Pharmacokinetic Variability Hypothesis for Acquired Drug Resistance and Failure of Antituberculosis Therapy. Clin Infect Dis. 2012;55(2):169\u0026ndash;77. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/cid/cis353\u003c/span\u003e\u003cspan address=\"10.1093/cid/cis353\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSrivastava S, Pasipanodya JG, Meek C, Leff R, Gumbo T. Multidrug-Resistant Tuberculosis Not Due to Noncompliance but to Between-Patient Pharmacokinetic Variability. J Infect Dis. 2011;204(12):1951\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/infdis/jir658\u003c/span\u003e\u003cspan address=\"10.1093/infdis/jir658\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMota L, Al-Efraij K, Campbell JR, Cook VJ, Marra F, Johnston J. Therapeutic drug monitoring in anti-tuberculosis treatment: a systematic review and meta-analysis. Int J Tuberc Lung Dis. 2016;20(6):819\u0026ndash;26. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5588/ijtld.15.0803\u003c/span\u003e\u003cspan address=\"10.5588/ijtld.15.0803\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerumal R, Naidoo K, Naidoo A, Ramachandran G, Requena-Mendez A, Sekaggya-Wiltshire C, et al. A systematic review and meta-analysis of first-line tuberculosis drug concentrations and treatment outcomes. Int J Tuberc Lung Dis. 2020;24(1):48\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePerumal R, Naidoo K, Naidoo A, Padayatchi N. Clinical impact of plasma concentrations of first-line antituberculosis drugs. S Afr Med J. 2023;113(3):148\u0026ndash;53. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7196/SAMJ.2023.v113i3.16761\u003c/span\u003e\u003cspan address=\"10.7196/SAMJ.2023.v113i3.16761\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAntwi S, Yang H, Enimil A, Sarfo AM, Gillani FS, Ansong D et al. Pharmacokinetics of the First-line Antituberculosis Drugs in Ghanaian Children with Tuberculosis with and without HIV Coinfection. Antimicrob Agents Chemother. 2016;AAC.01701-16. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1128/AAC.01701-16\u003c/span\u003e\u003cspan address=\"10.1128/AAC.01701-16\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKwara A, Enimil A, Gillani FS, Yang H, Sarfo AM, Dompreh A, et al. Pharmacokinetics of First-Line Antituberculosis Drugs Using WHO Revised Dosage in Children With Tuberculosis With and Without HIV Coinfection. J Pediatr Infect Dis Soc. 2016;5(4):356\u0026ndash;65. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jpids/piv035\u003c/span\u003e\u003cspan address=\"10.1093/jpids/piv035\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaleri N, Demb\u0026eacute;l\u0026eacute; SM, Villani P, Carvalho ACC, Cusato M, Bonkoungou V, et al. Systemic exposure to rifampicin in patients with tuberculosis and advanced HIV disease during highly active antiretroviral therapy in Burkina Faso. J Antimicrob Chemother. 2012;67(2):469\u0026ndash;72. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jac/dkr445\u003c/span\u003e\u003cspan address=\"10.1093/jac/dkr445\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 22028201.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaze MJ, Paynter J, Chiu W, Hu R, Nisbet M, Lewis C. Therapeutic drug monitoring of isoniazid and rifampicin during anti-tuberculosis treatment in Auckland, New Zealand. Int J Tuberc Lung Dis Off J Int Union Tuberc Lung Dis. 2016;20(7):955\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5588/ijtld.15\u003c/span\u003e\u003cspan address=\"10.5588/ijtld.15\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .0792 PubMed PMID: 27287650.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeloni M, Corti N, M\u0026uuml;ller D, Henning L, Gutteck U, von Braun A, et al. Cure of tuberculosis despite serum concentrations of antituberculosis drugs below published reference ranges. Swiss Med Wkly. 2015;145(5152):5152. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.4414/smw.2015.14223\u003c/span\u003e\u003cspan address=\"10.4414/smw.2015.14223\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePark JS, Lee JY, Lee YJ, Kim SJ, Cho YJ, Yoon HI, et al. Serum Levels of Antituberculosis Drugs and Their Effect on Tuberculosis Treatment Outcome. Antimicrob Agents Chemother. 2015;60(1):92\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1128/AAC.00693-15\u003c/span\u003e\u003cspan address=\"10.1128/AAC.00693-15\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 26459901; PubMed Central PMCID: PMC4704228.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEgelund EF, Alsultan A, Peloquin CA. Optimizing the clinical pharmacology of tuberculosis medications. Clin Pharmacol Ther. 2015;9(4). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/cpt.180\u003c/span\u003e\u003cspan address=\"10.1002/cpt.180\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKendall EA, Shrestha S, Cohen T, Nuermberger E, Dooley KE, Gonzalez-Angulo L, et al. Priority-Setting for Novel Drug Regimens to Treat Tuberculosis: An Epidemiologic Model. PLOS Med. 2017;14(1):e1002202. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pmed.1002202\u003c/span\u003e\u003cspan address=\"10.1371/journal.pmed.1002202\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeloquin CA. Therapeutic Drug Monitoring in the Treatment of Tuberculosis. Drugs. 2002;62(15):2169\u0026ndash;83. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2165/00003495-200262150-00001\u003c/span\u003e\u003cspan address=\"10.2165/00003495-200262150-00001\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChideya S, Winston CA, Peloquin CA, Bradford WZ, Hopewell PC, Wells CD, et al. Isoniazid, Rifampin, Ethambutol, and Pyrazinamide Pharmacokinetics and Treatment Outcomes among a Predominantly HIV-Infected Cohort of Adults with Tuberculosis from Botswana. Clin Infect Dis. 2009;48(12):1685\u0026ndash;94. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1086/599040\u003c/span\u003e\u003cspan address=\"10.1086/599040\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTappero JW, Bradford WZ, Agerton TB, Hopewell P, Reingold AL, Lockman S, et al. Serum Concentrations of Antimycobacterial Drugs in Patients with Pulmonary Tuberculosis in Botswana. Clin Infect Dis. 2005;41(4):461\u0026ndash;9. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1086/431984\u003c/span\u003e\u003cspan address=\"10.1086/431984\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWHO. Definitions and reporting framework for tuberculosis \u0026ndash; 2013 revision (updated December 2014 and January 2020). Report No; 2013.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePasipanodya JG, McIlleron H, Burger A, Wash PA, Smith P, Gumbo T. Serum drug concentrations predictive of pulmonary tuberculosis outcomes. J Infect Dis. 2013;208(9):1464\u0026ndash;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/infdis/jit352\u003c/span\u003e\u003cspan address=\"10.1093/infdis/jit352\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 23901086; PubMed Central PMCID: PMC3789573.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Health Organization, editor. Community involvement in tuberculosis care and prevention: towards partnerships for health: guiding principles and recommendations based on a WHO review. Geneva, Switzerland: World Health Organization; 2008. p. 1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Health, Ghana, Standard Treatment, Guidelines. 2017. 7th ed. Ghana National Drugs Programme; 2017.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLixoft SAS. a Simulations Plus company. Monolix 2023R1. 2023.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDevaleenal Daniel B, Ramachandran G, Swaminathan S. The challenges of pharmacokinetic variability of first-line anti-TB drugs. Expert Rev Clin Pharmacol. 2017;10(1):47\u0026ndash;58. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/17512433.2017.1246179\u003c/span\u003e\u003cspan address=\"10.1080/17512433.2017.1246179\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 27724114.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSileshi T, Makonnen E, Telele NF, Barclay V, Zumla A, Aklillu E. Variability in plasma rifampicin concentrations and role of SLCO1B1, ABCB1, AADAC2 and CES2 genotypes in Ethiopian patients with tuberculosis. Infect Dis Lond Engl. 2024;56(4):308\u0026ndash;19. PubMed PMID: 38315168; PubMed Central PMCID: PMC11134291.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVanderWeele TJ, Ding P. Sensitivity Analysis in Observational Research: Introducing the E-Value. Ann Intern Med. 2017;167(4):268\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7326/M16-2607\u003c/span\u003e\u003cspan address=\"10.7326/M16-2607\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 28693043.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBang H, Robins JM. Biometrics. 2005;61(4):962\u0026ndash;73. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.1541-0420.2005.00377.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1541-0420.2005.00377.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Doubly Robust Estimation in Missing Data and Causal Inference Models.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHouck K, Chakhaia T, Gorvetzian S, Critchley JA, Schechter MC, Magee MJ. Diabetes Mellitus and Tuberculosis Treatment Outcomes: Interaction Assessment Between Hyperglycemia and Human Immunodeficiency Virus in the State of Georgia, 2015\u0026ndash;2020. Open Forum Infect Dis. 2023;10(6):ofad255. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/ofid/ofad255\u003c/span\u003e\u003cspan address=\"10.1093/ofid/ofad255\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuangfu P, Ugarte-Gil C, Golub J, Pearson F, Critchley J. The effects of diabetes on tuberculosis treatment outcomes: an updated systematic review and meta-analysis. Int J Tuberc Lung Dis Off J Int Union Tuberc Lung Dis. 2019;23(7):783\u0026ndash;96. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5588/ijtld.18.0433\u003c/span\u003e\u003cspan address=\"10.5588/ijtld.18.0433\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 31439109.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMekonen H, Negesse A, Dessie G, Desta M, Mihiret GT, Tarik YD, et al. Impact of HIV coinfection on tuberculosis treatment outcomes in Ethiopia: a systematic review and meta-analysis. BMJ Open. 2024;14(7):e087218. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmjopen-2024-087218\u003c/span\u003e\u003cspan address=\"10.1136/bmjopen-2024-087218\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 38969385; PubMed Central PMCID: PMC11228389.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWagnew F, Alene KA, Kelly M, Gray D. Undernutrition increases the risk of unsuccessful treatment outcomes of patients with tuberculosis in Ethiopia: A multicenter retrospective cohort study. J Infect. 2024;89(1):106175. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jinf.2024.106175\u003c/span\u003e\u003cspan address=\"10.1016/j.jinf.2024.106175\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 38729526.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eClewe O, Wicha SG, De Vogel CP, De Steenwinkel JEM, Simonsson USH. A model-informed preclinical approach for prediction of clinical pharmacodynamic interactions of anti-TB drug combinations. J Antimicrob Chemother. 2018;73(2):437\u0026ndash;47. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jac/dkx380\u003c/span\u003e\u003cspan address=\"10.1093/jac/dkx380\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu C, Liu Y, Hu L, Yang M, He ZG. Molecular mechanism of the synergistic activity of ethambutol and isoniazid against \u003cem\u003eMycobacterium tuberculosis\u003c/em\u003e. J Biol Chem. 2018;293(43):16741\u0026ndash;50. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1074/jbc.RA118.002693\u003c/span\u003e\u003cspan address=\"10.1074/jbc.RA118.002693\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSekaggya-Wiltshire C, von Braun A, Lamorde M, Ledergerber B, Buzibye A, Henning L, et al. Delayed Sputum Culture Conversion in Tuberculosis\u0026ndash;Human Immunodeficiency Virus\u0026ndash;Coinfected Patients With Low Isoniazid and Rifampicin Concentrations. Clin Infect Dis Off Publ Infect Dis Soc Am. 2018;67(5):708\u0026ndash;16. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/cid/ciy\u003c/span\u003e\u003cspan address=\"10.1093/cid/ciy\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. 179 PubMed PMID: 29514175; PubMed Central PMCID: PMC6094003.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHussain Z, Zhu J, Ma X. Metabolism and Hepatotoxicity of Pyrazinamide, an Antituberculosis Drug. Drug Metab Dispos. 2021;49(8):679\u0026ndash;82. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1124/dmd.121\u003c/span\u003e\u003cspan address=\"10.1124/dmd.121\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .000389 PubMed PMID: 34074731; PubMed Central PMCID: PMC8407665.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMalinga TVB, Othman H, Paximadis M, Tiemessen CT, Ramsay M, Hazelhurst S, et al. Characterization of NAT, GST, and CYP2E1 Genetic Variation in Sub-Saharan African Populations: Implications for Treatment of Tuberculosis and Other Diseases. Clin Pharmacol Ther. 2025;117(5):1338\u0026ndash;57. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/cpt.3557\u003c/span\u003e\u003cspan address=\"10.1002/cpt.3557\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 39829327; PubMed Central PMCID: PMC11993289.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImperial MZ, Nahid P, Phillips PPJ, Davies GR, Fielding K, Hanna D, et al. A patient-level pooled analysis of treatment-shortening regimens for drug-susceptible pulmonary tuberculosis. Nat Med. 2018;24(11):11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41591-018-0224-2\u003c/span\u003e\u003cspan address=\"10.1038/s41591-018-0224-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStrydom N, Gupta SV, Fox WS, Via LE, Bang H, Lee M, et al. Tuberculosis drugs\u0026rsquo; distribution and emergence of resistance in patient\u0026rsquo;s lung lesions: A mechanistic model and tool for regimen and dose optimization. PLOS Med. 2019;16(4):e1002773. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pmed.1002773\u003c/span\u003e\u003cspan address=\"10.1371/journal.pmed.1002773\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaal M, Habler K, Northoff B, Vogeser M. Comparative Routine Therapeutic Drug Monitoring of Mycophenolic Acid in human Plasma with HPLC-UV and Isotope Dilution LC-MS/MS. Clin Lab. 2020;66(4). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7754/Clin\u003c/span\u003e\u003cspan address=\"10.7754/Clin\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. .Lab.2019.190820 PubMed PMID: 32255280.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePršo K, Židekov\u0026aacute; N, Porvazn\u0026iacute;k I, Solovič I, Mokr\u0026yacute; J, Kertys M. A high-throughput LC\u0026ndash;MS/MS method for simultaneous determination of isoniazid, ethambutol and pyrazinamide in human plasma. Rapid Commun Mass Spectrom. 2023;37(2):e9425. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/rcm.9425\u003c/span\u003e\u003cspan address=\"10.1002/rcm.9425\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. PubMed PMID: 36329637; PubMed Central PMCID: PMC9787364.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCapiau S, Veenhof H, Koster RA, Bergqvist Y, Boettcher M, Halmingh O, et al. Official International Association for Therapeutic Drug Monitoring and Clinical Toxicology Guideline: Development and Validation of Dried Blood Spot\u0026ndash;Based Methods for Therapeutic Drug Monitoring. Ther Drug Monit. 2019;41(4):409. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/FTD.0000000000000643\u003c/span\u003e\u003cspan address=\"10.1097/FTD.0000000000000643\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-infectious-diseases","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"infd","sideBox":"Learn more about [BMC Infectious Diseases](http://bmcinfectdis.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/infd","title":"BMC Infectious Diseases","twitterHandle":"#bmcinfectdis","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Tuberculosis, Pharmacokinetics, Treatment failure, Target trial emulation, Subtherapeutic concentrations, Ghana","lastPublishedDoi":"10.21203/rs.3.rs-9255541/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9255541/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eTuberculosis (TB) treatment outcomes in sub-Saharan Africa remain suboptimal despite high adherence to first-line therapy. Variability in drug pharmacokinetics, resulting in subtherapeutic plasma concentrations, may contribute to treatment failure and the development of resistance. This study estimated the causal effect of subtherapeutic plasma concentrations of first-line anti-TB drugs on treatment failure or death among individuals with drug-susceptible pulmonary TB in Ghana.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe conducted a prospective cohort study of 164 adults receiving standard WHO weight-band dosing at five Ghanaian hospitals. Peak plasma concentrations (C\u003csub\u003emax\u003c/sub\u003e) of rifampicin, isoniazid, pyrazinamide, and ethambutol were measured at months 1\u0026ndash;2 using validated LC-MS/MS. We emulated a target trial comparing two static strategies: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) therapeutic C\u003csub\u003emax\u003c/sub\u003e of at least one first-line drug versus (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) subtherapeutic C\u003csub\u003emax\u003c/sub\u003e of all four drugs. Using the clone-censor-weight approach, we estimated the per-protocol analogue risk difference (RD) and risk ratio (RR) for treatment failure (smear positive at months 5 or 6) or death by month 6. Models were adjusted for baseline covariates using inverse probability of censoring weighting. Sensitivity analyses included inverse probability weighting with regression adjustment, plain inverse probability weighting, and E-values.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eOf 164 participants, 120 had complete pharmacokinetic and outcome data; 20.0% (24/120) had subtherapeutic concentrations of all four drugs. The 6-month risk of treatment failure or death was 33.3% under the low-exposure strategy versus 4.2% under the adequate-exposure strategy (crude RD: 29.1 percentage points). In weighted analyses, low drug exposure was associated with a 25.6 percentage-point increase in absolute risk of treatment failure or death (95% CI: 5.7\u0026ndash;45.6; p\u0026thinsp;=\u0026thinsp;0.012) and an 8.6-fold higher relative risk (95% CI: 2.34\u0026ndash;31.92; p\u0026thinsp;=\u0026thinsp;0.001), corresponding to approximately one additional poor outcome for every four patients with subtherapeutic levels. Sensitivity analyses were consistent (ATE: 19.9%, 95% CI: 2.3\u0026ndash;37.5). The E-value was 15.5 (lower bound 5.0).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eSubtherapeutic exposure to all four first-line drugs was strongly associated with increased risk of treatment failure or death. Preventing multidrug subtherapeutic exposure through therapeutic drug monitoring or optimized dosing warrants randomized evaluation in high-burden settings.\u003c/p\u003e\u003ch2\u003eTrial registration\u003c/h2\u003e \u003cp\u003eClinical trial number: not applicable.\u003c/p\u003e","manuscriptTitle":"Exposure to the Four First-Line Anti-Tuberculosis Drugs and Treatment Outcomes: A Target Trial Emulation Study in Ghana","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-10 06:33:58","doi":"10.21203/rs.3.rs-9255541/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-05-13T07:36:21+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-12T19:29:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"30805188178563834414537122109116927148","date":"2026-04-29T16:01:26+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-21T02:09:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"86139869584241102362384259801586363940","date":"2026-04-07T09:44:48+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"72715290039602896011327622508832277985","date":"2026-04-07T00:43:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"277119297070446952219567154883830250407","date":"2026-04-06T01:05:54+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-03T05:45:06+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-02T05:01:07+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-02T04:58:02+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-01T07:37:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Infectious Diseases","date":"2026-04-01T07:21:57+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-infectious-diseases","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"infd","sideBox":"Learn more about [BMC Infectious Diseases](http://bmcinfectdis.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/infd","title":"BMC Infectious Diseases","twitterHandle":"#bmcinfectdis","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"a3962781-770a-43d5-8268-2892c61f6be6","owner":[],"postedDate":"April 10th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Revision requested","date":"2026-05-13T07:36:21+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-12T19:29:50+00:00","index":41,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[],"tags":[],"updatedAt":"2026-05-13T07:42:19+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-10 06:33:58","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9255541","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9255541","identity":"rs-9255541","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}