Patterns of Anti-Seizure Medication Non-adherence in Post-Stroke Prophylaxis Among Older Adults

Purpose: Anti-seizure medications (ASMs) are commonly prescribed for post-stroke seizures. Yet long-term use and non-adherence lack clear clinical guidance. We examined patterns of medication non-adherence among older adults who initiated Levetiracetam–the most frequently prescribed ASM–within 30 days after an acute ischemic stroke (AIS) discharge. Methods : We analyzed a national 20% random sample of U.S. Medicare beneficiaries aged 65 and over, who were hospitalized for a first AIS between 2009-2021 and initiated Levetiracetam within 30 days of discharge. We used an adjusted proportion of days covered (PDC) measure, accounting for prescription overlap, hospital readmissions, and non-persistency, to analyze medication non-adherence patterns during the first year after initiation. Latent class mixed models on PDC trajectories were applied to characterize patterns of non-adherence and factors associated with non-adherence. Results : In our sample of 1,697 Levetiracetam initiators, the mean age was 77.4 years, with 58% female. The latent class mixed model on PDC trajectories identified three distinct non-adherence pattern groups: (i) 907 patients (53%) were non-adherent after two months, (ii) 99 patients (6%) were non-adherent after ten months, and (iii) 692 patients (41%) remained adherent. Non-white patients and males were more likely to be non-adherent to the medication. Conclusions: The adjusted PDC method and latent class model incorporate hospital readmission and adjust for relevant covariates. Approximately 60% older adults were non-adherent to Levetiracetam within a year after outpatient initiation, with race and gender significantly associated with non-adherence patterns. This study offers both methodological innovation and clinical meaningful insights into ASM non-adherence following AIS. Patterns of Anti-Seizure Medication Non-adherence in Post-Stroke Prophylaxis Among Older Adults Shuo Sun, PhD a ; Rafaella Cazé de Medeiros, MD b ; Julianne D. Brooks, MPH b ; Joseph P. Newhouse, PhD c,d,e,f ; Lee H. Schwamm, MD g ; Sebastien Haneuse, PhD a *; Lidia M. V. R. Moura, MD PhD MPH b * *co-senior authors. a Department of Biostatistics, Harvard T.H. Chan School of Public Health, Boston, Massachusetts. b Department of Neurology, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts. c Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts d Department of Health Policy and Management, Harvard T.H. Chan School of Public Health, Boston, Massachusetts. e Harvard Kennedy School, Cambridge, Massachusetts. f National Bureau of Economic Research, Cambridge, Massachusetts. g Department of Neurology, Yale School of Medicine, New Haven, Connecticut. not-yet-known not-yet-known not-yet-known unknown Corresponding Author: Lidia MVR Moura [email protected] 617.726.3311 55 Fruit Street, Wang 739D Boston, MA 02114 not-yet-known not-yet-known not-yet-known unknown ABSTRACT Purpose: Anti-seizure medications (ASMs) are commonly prescribed for post-stroke seizures. Yet long-term use and non-adherence lack clear clinical guidance. We examined patterns of medication non-adherence among older adults who initiated Levetiracetam–the most frequently prescribed ASM–within 30 days after an acute ischemic stroke (AIS) discharge. Methods : We analyzed a national 20% random sample of U.S. Medicare beneficiaries aged 65 and over, who were hospitalized for a first AIS between 2009-2021 and initiated Levetiracetam within 30 days of discharge. We used an adjusted proportion of days covered (PDC) measure, accounting for prescription overlap, hospital readmissions, and non-persistency, to analyze medication non-adherence patterns during the first year after initiation. Latent class mixed models on PDC trajectories were applied to characterize patterns of non-adherence and factors associated with non-adherence. Results : In our sample of 1,697 Levetiracetam initiators, the mean age was 77.4 years, with 58% female. The latent class mixed model on PDC trajectories identified three distinct non-adherence pattern groups: (i) 907 patients (53%) were non-adherent after two months, (ii) 99 patients (6%) were non-adherent after ten months, and (iii) 692 patients (41%) remained adherent. Non-white patients and males were more likely to be non-adherent to the medication. Conclusions: The adjusted PDC method and latent class model incorporate hospital readmission and adjust for relevant covariates. Approximately 60% older adults were non-adherent to Levetiracetam within a year after outpatient initiation, with race and gender significantly associated with non-adherence patterns. This study offers both methodological innovation and clinical meaningful insights into ASM non-adherence following AIS. Keywords: Acute Ischemic Stroke; Discontinuation; Epilepsy; Latent Class Mixed Model; Medicare; Seizure; Trajectory Model. Key points: • Three non-adherence pattern groups were identified, with early and later non-adherence to Levetiracetam. Approximately 53% of older adults became non-adherent after two months, and 6% became non-adherent after ten months. • Non-white and males were more likely to be non-adherent to Levetiracetam. • An adjusted proportion of days covered (PDC) approach and latent class mixed models, accounting for prescription overlap, hospital readmissions, and clinical factors, were applied to characterize adherence patterns. not-yet-known not-yet-known not-yet-known unknown INTRODUCTION Stroke affects over 700,000 affected individuals per year.1 Acute ischemic stroke (AIS) patients can present with post-stroke seizures, with an incidence that can reach up to 40%.2,3 Considering the high risk of seizure recurrence in post-stroke patients and its negative impact on clinical prognosis, therapy with antiseizure medications (ASMs) is common in older adults.4–6 Among the different available ASMs, Levetiracetam is typically chosen as a first-line agent.7 The long-term use versus non-adherence of antiseizure treatment is a subject of considerable debate.8 A significant challenge in managing ASM use in post-stroke patients is the lack of clear guidelines regarding its use and non-adherence.8–10 Some patients discontinue ASMs due to adverse effects5,9 while others continue indefinitely, influenced by what is known as ”therapeutic inertia” or ”physician inertia,” where the adherence of treatment occurs without a clear clinical indication because stopping the medication is not actively considered.11,12 Different objective methodologies can be used to identify and measure these non-adherence patterns, including PDC (Proportion of Days Covered), MPR (Medication Possession Ratio), and others.13,14 Previously used in several studies, PDC calculates the proportion of days within a defined period that a patient can access their medication, considering the number of days the medication refills.15,16 To accurately describe the analysis we conducted, from this point forward, we will reference Levetiracetam “adherence” as it refers to how closely patients follow prescribed dosing regimens, measured by metrics like PDC and MPR. This study hypothesizes the existence of considerable variation in the patterns of Levetiracetam non-adherence among older adults following an acute ischemic stroke discharge. We examined patients initiating Levetiracetam therapy within the first 30 days after discharge to define patterns of non-adherence within 12 months.

This study was approved by the Mass General Brigham Institutional Review Board and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. 17 The data supporting this study’s findings are collected routinely by The Centers for Medicare & Medicaid Services (CMS) for billing purposes and were made available by CMS with no direct identifiers. 18 All results were aggregated following CMS Cell Suppression Policies. Restrictions apply to the availability of these data, which were used under license for this study. Medicare data are available through CMS with their permission. The code that produced the findings is available in the Supplemental Materials. Data Sources Demographic variables, including age, sex, race, ethnicity, and insurance status, were obtained for all study patients from the Medicare’s Master Beneficiary Summary (MBSF) File. Hospitalization and mortality information was extracted from the Medicare Provider Analysis and Review (MedPAR) database based on principal diagnosis codes for AIS. 19 We used a validated strategy to capture acute stroke in administrative databases using the International Classification of Diseases, 9th Revision (ICD-9) codes 433, 434, 436, and ICD-10 codes I63. 20 Study Population We conducted a retrospective analysis of a 20% random sample of U.S. Medicare beneficiaries discharged after acute ischemic stroke episodes between April 1, 2009, and September 30, 2021, who initiated Levetiracetam within the 30-day period post-discharge in an outpatient setting. For the patients with more than one stroke episode during the study period, the first stroke hospitalization was considered as their AIS admission index date. We excluded patients who initiated Levetiracetam during any readmission. Eligible patients were 65 years or older and had continuous enrollment in traditional Medicare Part D (drug prescription coverage) for one year before the index stroke hospitalization and at least one year plus 30 days following their index stroke hospitalization or died before one year after their first Levetiracetam initiation post-discharge. For each patient, time 0 is defined as the date of the first Levetiracetam prescription filled in Medicare Part D post discharge from their index stroke hospitalization. Measures of Medication Adherence The standard PDC calculates the proportion of days within a defined period that a patient has access to their medication, considering the number of days available refills cover the medication. 21,22 It is defined as: \(PDC=\frac{\text{Number\ of\ days\ covered\ by\ medication}}{\text{Length\ of\ time\ interval}}=1-\frac{\text{Number\ of\ gap\ days}}{\text{Length\ of\ time\ interval}}\). To acknowledge readmission, prescription overlap, non-persistency, and death, we used a series of pragmatic strategies to construct an adjusted PDC. Death was treated as a censoring factor. While Medicare Part D claims data contains records of outpatient prescriptions, individuals may also receive medication if they are readmitted to an inpatient setting such as a long-term care facility or skilled nursing facility (SNF). Typically, patients receive medication through the hospital rather than from home, filled through an outpatient pharmacy. We established rules to account for the possibility of patients receiving additional medication doses during readmissions. For patients readmitted for 14 days or less, we assumed they received medication from the inpatient provider for their readmission stay. This situation could result in patients having extra tablets at home. If not accounted for, it could appear that they picked a prescription up late and had poor medication adherence. Upon discharge from the readmission stay, we assumed these patients would have extra tablets equivalent to the length of the readmission stay. This rule helped account for outpatient prescriptions and inpatient medication administration during short readmissions. For patients with prolonged hospital readmission of 15 days or more, if they refilled their Levetiracetam prescriptions within 90 days after discharge from the readmission, we assumed the extra tablets available were for 14 days. Otherwise, readmission was considered a disruption, and these patients were censored at the date of readmission. When calculating the PDC, “prescription overlap” was where the patient refilled a prescription prior to the end date of the previous (current) supply. For example, if a patient picked up a 30-day supply of medication five days early, they would have 35 tablets on hand, causing the PDC to exceed 100%. To acknowledge overlap, prescription refills were shifted forward by the number of days of overlap. The same method was applied to extra medication doses during short readmissions (deemed non-disruptive). If leftover tablets overlapped with subsequent refills, the medication supply was adjusted ensure accurate adherence calculations. Patients who did not refill their Levetiracetam prescriptions within 90 days after the previous prescription’s end date were considered non-persistent with the medication. Non-persistent periods were incorporated into the adjusted PDC calculation by treating days without medication as non-adherent days. This adjustment ensures a comprehensive assessment of the patient’s behavior of adherence, encompassing both active medication use and gaps due to non-persistence. An example of the adjustments to account for overlaps and readmissions is illustrated in Figure 1. Using a sequence of time intervals, a patient is defined as non-adherent (i.e., having discontinued the medication) during an interval if the PDC within that interval is less than 0.8; otherwise, the patient is defined as adherent. Statistical Analysis We utilized a latent class mixed model 23 on adjusted PDC trajectories to identify and characterize patients’ non-adherence patterns over the first 12 months after Levetiracetam initiation. Specifically, we used a 60-day slicing window, with adjusted PDC estimates calculated for each patient at every 60-day interval. This resulted in estimated adjusted PDC trajectories with up to seven distinct time points per patient. A threshold of 80%, commonly used in in health research and supported by the Pharmacy Quality Alliance (PQA) 22 and the Centers for Disease Control and Prevention (CDC), 24 was applied to dichotomize adjusted PDCs at each time point into two categories: non-adherence (<80%) and adherence (\(\geq\) 80%). The binary outcome was coded as 1 for non-adherence and 0 otherwise. The latent class mixed model conceptualizes the patient population in questions as a mixture of \(G\) latent classes, each characterized by a unique mean trajectory. The \(G\) mean trajectory profiles are defined as functions of time and covariates through latent class-specific mixed effects models. The difference with a standard linear mixed model is that both fixed effects and random effects can be class specific. Throughout we used a logistc model for the binary outcome (non-adherence/adherence) and selected the number of classes and covariates for our optimal model based on BIC. The analysis was performed using ‘lcmm’ package in R version 4.4.0. Cumulative adjusted PDCs over the first year after Levetiracetam initiation were also calculated and summarized in the Supplementary Materials.

We presented sample characteristics in Supplemental Table S1 and sample inclusion and exclusion criteria in Supplemental Figure S1. The 1,697 included patients presented with a mean age of 77.5 years (median: 77; IQR: 71–83); 57.5% were female, and 79% were white. Among these patients, 1,016 (60%) were readmitted within one year plus 30 days after their initial discharge following an AIS admission, and 473 (27.9%) died within the same period. Within the 1,016 patients with at least one readmission episode during the study period, 81.4% had a length of stay of 14 days or fewer (median: 5; IQR: 3–11). The average number of readmissions was 2.6 (median: 2; IQR: 1–3), and the average time from initial discharge to first readmission was 75.8 days (median: 31; IQR: 9–105). For computation and interpretation purposes, age was categorized in 5-year intervals, starting at 65. We considered class numbers from 1 to 4 under various model specifications (see Tables S2 and S3). The optimal model had three classes with BIC equal to 8,218 (Table S3), with the posterior classification given in Table 1. The estimated parameters, the estimated standard errors, the Wald test statistics, and the corresponding \(p\)-values of the optimal latent class mixed model were summarized in Table 2. The class 1 intercept in the longitudinal model was not estimated due to the location constraint. 23 We plotted the comparison between white males and females at age 75, and between White Females and Non-white Females at age 75 in Figures 2 and 3, respectively. The y-axis is the predicted probability of non-adherence, i.e., \(\ Pr(PDC<0.8).\) We categorized the three classes for interpretability as follows: (i) non-adherence after two-month (class 1) included 907 patients (53.45%), with median time to first non-adherence (e.g., time to first PDC less than 0.8) being 60 days; (ii) non-adherence after ten-month (class 2) included 99 patients (5.83%), with median time to first non-adherence being 300 days; and (iii) adherence (class 3) included 691 patients (40.72%), of whom 65% remained adherent to their medication until the end of the study. The Non-white patients and males were more likely to discontinue the medication within a year (\(p\)-value\(\approx\)0). The descriptive results of the adjusted PDC trajectories can be found in the Supplementary Material Section S2. Also, the descriptive results of cumulative adjusted PDCs over the first year after Levetiracetam initiation are provided in Supplementary Materials Section S3. \papertype Original Article DISCUSSION Our study assessed a national U.S. Medicare sample to identify patterns of use and adherence of Levetiracetam in acute ischemic stroke survivors over 65 years old. By examining these patterns, we aimed to explore the utility of PDC as a tool for evaluation and insights into factors that influence medication adherence. We used PDC to measure medication adherence, as it considers medication overlap and excess in its methodology, to gain insights into how long vulnerable patients continue to take these medications. 2122 This tool and its importance have been explored in previous studies for adherence of ASMs and other drug classes with cumulative PDC, including the use of claims-based data. 25–28 In our study, we successfully delineated ASM adherence by adding time-varying PDC and identifying different factors related to drug non-adherence. Additionally, our analysis was built on assumptions on prescription overlap and readmission (short/long stay, extra tablets) and non-persistence assumptions. We introduced a novel framework by making the assumptions explicit in the methodology, providing a transparent replication pathway in future studies. Moreover, there remains a questionable indication of ASM prophylaxis in elderly post-stroke patients presenting with seizures, with a significant gap existing in clinical guidelines regarding the optimal treatment duration and the ideal timing for medication cessation. 4,29 Several factors complicate ASM use in older adults, including an increased susceptibility to adverse drug reactions and the potential for harmful drug interactions, particularly when polypharmacy is involved. 30,31,32 In this context, we can use classes of non-adherence in future studies to understand outcomes and optimal treatment duration. In our study, non-White populations presented with decreased Levetiracetam adherence rate. Increased non-adherence in non-White populations, such as African Americans, Hispanics, and others, has been documented in the literature, configuring a significant social barrier to antiseizure treatment. 33,34 Our results also highlighted a decreased adherence rate in male patients, as seen in previous studies. 35 This finding is pointed out as existent due to barriers to medication adherence, such as forgetfulness, medication side effects, and others. 36,37,38 Delineating medication adherence trajectories can help identify gaps in follow-up and alert providers to re-examine patients’ medications, thus resulting in an essential tool for surveillance to prevent adverse outcomes and improve health care. 39,40,41 This study is critical to understanding variations and addressing inconsistencies in the anti-seizure prophylaxis of discharged AIS patients. Our findings are consistent with previous findings in the literature, further confirming the differences in ASM non-adherence across different racial and sex groups. The existing divergent periods of non-adherence highlight again the need for guidelines for post-stroke seizure treatment choice and indication in older populations.

This study presents with limitations. Our population findings may not be generalizable to others, considering the use of data from Medicare and its availability. Moreover, the study findings may not be generalizable to patients in other parts of Medicare not included in this study, such as Part C (health plan associated with Medicare Advantage). Lastly, our study did not reflect patients discharged to SNFs or inpatient rehabilitation units, as our study focused on outpatient patterns of medication use and adherence. The use of claims-based data, despite its reliability, can present with common limitations. Entry errors and missing data, such as baseline variables, may exist that would add to our analysis. We also did not consider daily dosages of treatment or drug switches in the analysis. This may overestimate adherence without considering the start of a new therapy with a different ASM. Lastly, in our methodology, the adjusted PDC trajectories were continuous but asymmetric (heavily concentrated at 0), therefore not allowing a Gaussian assumption (using dichotomized outcomes might present a loss of information). PLAIN LANGUAGE SUMMARY Antiseizure medications (ASMs) like Levetiracetam are commonly prescribed after acute ischemic stroke (AIS). However, the duration of ASM therapy and patterns of non-adherence remain unclear, particularly in the absence of established guidelines. We conducted a retrospective cohort study using a 20% national sample of Medicare beneficiaries aged 65 and older who were hospitalized for a first AIS between 2009 and 2021 and initiated Levetiracetam within 30 days of discharge. An adjusted proportion of days covered (PDC) measure was used to assess non-adherence, accounting for prescription overlap, hospital readmissions, and treatment non-persistency. Latent class mixed models identified non-adherence trajectories over one year and examined associated factors. Among 1,697 patients, approximately 60% were non-adherent within the first year. We identified three non-adherence patterns: early non-adherence (53%), late non-adherence (6%), and sustained adherence (41%). Non-white race and male sex were significantly associated with higher a likelihood of non-adherence. This study reveals high rates of early non-adherence and sociodemographic disparities in post-stroke ASM use, highlighting the need for improved guidance and follow-up strategies in clinical practice. ETHICS STATEMENT This study was approved by the Mass General Brigham Institutional Review Board and followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. The requirement for informed consent was waived in our study as we performed a secondary analysis of data routinely collected for billing. FUNDING This study was supported by the NIH (1R01AG073410-01). not-yet-known not-yet-known not-yet-known unknown DISCLOSURES OF CONFLICT OF INTEREST S.S., R.C.M., J.D.B., S.H., and L.H.S. have no conflict of interest to disclose. J.P.N. is the National Committee for Quality Assurance director and reports no conflict of interest. L.M.V.R.M. receives support from the Epilepsy Foundation of America and reports no conflict of interest. DATA AVAILABILITY STATEMENT The data supporting this study’s findings were collected by The Centers for Medicare & Medicaid Services (CMS) and were made available by CMS with no direct identifiers. All results were aggregated following CMS Cell Suppression Policies. Restrictions apply to the availability of these data, which were used under license for this study. Medicare data are available through CMS with their permission. We included the code which produced the findings in the supplemental materials (Supplementary materials – analytical code). \papertype Original Article REFERENCES 1. Tsao CW, Aday AW, Almarzooq ZI, Anderson CAM, Arora P, Avery CL, Baker-Smith CM, Beaton AZ, Boehme AK, Buxton AE, Commodore-Mensah Y, Elkind MSV, Evenson KR, Eze-Nliam C, Fugar S, Generoso G, Heard DG, Hiremath S, Ho JE, Kalani R, Kazi DS, Ko D, Levine DA, Liu J, Ma J, Magnani JW, Michos ED, Mussolino ME, Navaneethan SD, Parikh NI, Poudel R, Rezk-Hanna M, Roth GA, Shah NS, St-Onge M-P, Thacker EL, Virani SS, Voeks JH, Wang N-Y, Wong ND, Wong SS, Yaffe K, Martin SS, on behalf of the American Heart Association Council on Epidemiology and Prevention Statistics Committee and Stroke Statistics Subcommittee. Heart Disease and Stroke Statistics—2023 Update: A Report From the American Heart Association. Circulation . 2023;147. doi:10.1161/CIR.0000000000001123.2. Camilo O, Goldstein LB. Seizures and epilepsy after ischemic stroke. Stroke . 2004;35:1769–1775.3. Menon B, Shorvon SD. Ischaemic stroke in adults and epilepsy. Epilepsy Research . 2009;87:1–11.4. Moura LMVR, Donahue MA, Yan Z, Smith LH, Hsu J, Newhouse JP, Schwamm LH, Haneuse S, Hernandez-Diaz S, Blacker D. Comparative Effectiveness and Safety of Seizure Prophylaxis Among Adults After Acute Ischemic Stroke. Stroke . 2023;54:527–536.5. Brigo F, Lattanzi S, Zelano J, Bragazzi NL, Belcastro V, Nardone R, Trinka E. Randomized controlled trials of antiepileptic drugs for the treatment of post-stroke seizures: A systematic review with network meta-analysis. Seizure . 2018;61:57–62.6. Fisher RS, Acevedo C, Arzimanoglou A, Bogacz A, Cross JH, Elger CE, Engel J, Forsgren L, French JA, Glynn M, Hesdorffer DC, Lee BI, Mathern GW, Moshé SL, Perucca E, Scheffer IE, Tomson T, Watanabe M, Wiebe S. ILAE Official Report: A practical clinical definition of epilepsy. Epilepsia . 2014;55:475–482.7. Terman SW, Youngerman BE, Choi H, Burke JF. Antiseizure medication treatment pathways for US Medicare beneficiaries with newly treated epilepsy. Epilepsia . 2022;63:1571–1579.8. Moura LM, Westover MB, Kwasnik D, Cole AJ, Hsu J. Causal inference as an emerging statistical approach in neurology: an example for epilepsy in the elderly. Clin Epidemiol . 2017;9:9–18.9. Moura LMVR, Smith JR, Yan Z, Blacker D, Schwamm LH, Newhouse JP, Hernandez-Diaz S, Hsu J. Patterns of anticonvulsant use and adverse drug events in older adults. Pharmacoepidemiol Drug Saf . 2021;30:28–36.10. Lezaic N, Gore G, Josephson CB, Wiebe S, Jetté N, Keezer MR. The medical treatment of epilepsy in the elderly: A systematic review and meta‐analysis. Epilepsia . 2019;60:1325–1340.11. Phillips LS, Branch WT, Cook CB, Doyle JP, El-Kebbi IM, Gallina DL, Miller CD, Ziemer DC, Barnes CS. Clinical Inertia. Ann Intern Med . 2001;135:825.12. Aujoulat I, Jacquemin P, Darras E, Rietzschel E, Wens J, Hermans M, Scheen A, Trefois P. Factors associated with clinical inertia: an integrative review. AMEP . 2014;:141.13. Malek N, Heath CA, Greene J. A review of medication adherence in people with epilepsy. Acta Neurol Scand . 2017;135:507–515.14. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiology and Drug . 2006;15:565–574.15. Hess LM, Raebel MA, Conner DA, Malone DC. Measurement of Adherence in Pharmacy Administrative Databases: A Proposal for Standard Definitions and Preferred Measures. Ann Pharmacother . 2006;40:1280–1288.16. Benner JS. Long-term Persistence in Use of Statin Therapy in Elderly Patients. JAMA . 2002;288:455.17. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, STROBE Initiative. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ . 2007;335:806–808.18. Centers for Medicare and Medicaid Services. Centers for Medicare and Medicaid Services. Available at https://www.cms.gov. Accessed January 7, 2025.19. Medicare Provider Analysis and Review (MedPAR). ResDAC (Research Data Assistance Center). Available at https://resdac.org/cms-data/files/medpar. Accessed January 7, 2025.20. McCormick N, Bhole V, Lacaille D, Avina-Zubieta JA. Validity of Diagnostic Codes for Acute Stroke in Administrative Databases: A Systematic Review. PLOS ONE . 2015;10:e0135834.21. Loucks J, Zuckerman AD, Berni A, Saulles A, Thomas G, Alonzo A. Proportion of days covered as a measure of medication adherence. American Journal of Health-System Pharmacy . 2022;79:492–496.22. PQA Adherence Measures. Available at https://www.pqaalliance.org/adherence-measures. Accessed January 13, 2025.23. Proust-Lima C, Philipps V, Liquet B. Estimation of Extended Mixed Models Using Latent Classes and Latent Processes: The R Package lcmm . J Stat Soft . 2017;78. doi:10.18637/jss.v078.i02.24. Pharmacy-Based Interventions to Improve Medication Adherence. Centers for Disease Control and Prevention (CDC) Cardiovascular Disease Data, Tools, and Evaluation Resources. Available at https://www.cdc.gov/cardiovascular-resources/php/medication-adherence/index.html#. Accessed January 13, 2025.25. Bansilal S, Castellano JM, Garrido E, Wei HG, Freeman A, Spettell C, Garcia-Alonso F, Lizano I, Arnold RJG, Rajda J, Steinberg G, Fuster V. Assessing the Impact of Medication Adherence on Long-Term Cardiovascular Outcomes. Journal of the American College of Cardiology . 2016;68:789–801.26. Ho PM, Bryson CL, Rumsfeld JS. Medication Adherence: Its Importance in Cardiovascular Outcomes. Circulation . 2009;119:3028–3035.27. Terman SW, Kerr WT, Marcum ZA, Wang L, Burke JF. Antiseizure medication adherence trajectories in Medicare beneficiaries with newly treated epilepsy. Epilepsia . 2021;62:2778–2789.28. Terman SW, Kerr WT, Aubert CE, Hill CE, Marcum ZA, Burke JF. Adherence to Antiseizure vs Other Medications Among US Medicare Beneficiaries With and Without Epilepsy. Neurology . 2022;98:e427–e436.29. Zaccara G, Lattanzi S, Brigo F. Acute symptomatic seizures after stroke: A scoping review on primary prevention, treatment with antiseizure medications and drug discontinuation. Epilepsy & Behavior . 2023;149:109499.30. Loikas D, Linnér L, Sundström A, Wettermark B, von Euler M. Post-stroke epilepsy and antiepileptic drug use in men and women. Basic Clin Pharmacol Toxicol . 2021;129:148–157.31. Dovjak P. Polypharmacy in elderly people. Wien Med Wochenschr . 2022;172:109–113.32. Terman SW, Aubert CE, Hill CE, Maust DT, Betjemann JP, Boyd CM, Burke JF. Polypharmacy in patients with epilepsy: A nationally representative cross-sectional study. Epilepsy & Behavior . 2020;111:107261.33. Von Gaudecker JR, Buelow JM, Miller WR, Tanner AL, Austin JK. Social determinants of health associated with epilepsy treatment adherence in the United States: A scoping review. Epilepsy & Behavior . 2021;124:108328.34. Bautista RED, Graham C, Mukardamwala S. Health disparities in medication adherence between African-Americans and Caucasians with epilepsy. Epilepsy & Behavior . 2011;22:495–498.35. Byrnes M, Chandan P, Newey C, Hantus S, Punia V. Acute Symptomatic Seizure Associated With Chronic Antiseizure Medication Use After Stroke. Neurol Clin Pract . 2022;12:e154–e161.36. Hovinga CA, Asato MR, Manjunath R, Wheless JW, Phelps SJ, Sheth RD, Pina-Garza JE, Zingaro WM, Haskins LS. Association of non-adherence to antiepileptic drugs and seizures, quality of life, and productivity: survey of patients with epilepsy and physicians. Epilepsy Behav . 2008;13:316–322.37. Nakhutina L, Gonzalez JS, Margolis SA, Spada A, Grant A. Adherence to antiepileptic drugs and beliefs about medication among predominantly ethnic minority patients with epilepsy. Epilepsy Behav . 2011;22:584–586.38. Jones RM, Butler JA, Thomas VA, Peveler RC, Prevett M. Adherence to treatment in patients with epilepsy: associations with seizure control and illness beliefs. Seizure . 2006;15:504–508.39. Bensken WP, Fernandez Baca Vaca G, Alberti PM, Khan OI, Ciesielski TH, Jobst BC, Williams SM, Stange KC, Sajatovic M, Koroukian SM. Racial and Ethnic Differences in Antiseizure Medications Among People With Epilepsy on Medicaid: A Case of Potential Inequities. Neurol Clin Pract . 2023;13:e200101.40. Gonzalez A, Pandey D, Digiannantonio N, Serafini A. Adherence to antiseizure medications in an underserved population with epilepsy. Epilepsy & Behavior . 2023;149:109484.41. Bensken WP, Alberti PM, Khan OI, Williams SM, Stange KC, Vaca GF-B, Jobst BC, Sajatovic M, Koroukian SM. A framework for health equity in people living with epilepsy. Epilepsy Research . 2022;188:107038. Table 1. Posterior classification of the optimal latent class mixed model. | Non-adherence after two-month (Class 1) | Non-adherence after ten-month (Class 2) | Adherence (Class 3) | | | Number of patients (%) | 907 (53.5) | 99 (5.8) | 691 (40.7) | | Age, mean (SD) | 2.62 (1.63) | 2.37 (1.51) | 2.27 (1.54) | | Male, N (%) | 397 (43.8) | 283 (41.0) | 42 (42.4) | | Non-white, N (%) | 190 (20.9) | 145 (21.0) | 20 (20.2) | Table 2. Latent class mixed model results. | Coefficient estimate | SE | Wald test statistics | \(p-\)value | | | Intercept class 1 | 1.62 | 0.31 | 5.16 | 0.00 | | Intercept class 3 | 2.01 | 0.30 | 6.72 | 0.00 | | Male class 1 | 0.06 | 0.28 | 0.23 | 0.82 | | Male class 3 | -0.08 | 0.27 | -0.30 | 0.77 | | Age class 1 | 0.21 | 0.09 | 2.25 | 0.02 | | Age class 3 | 0.09 | 0.09 | 1.02 | 0.31 | | Fixed effects in the longitudinal model | |||| | Coefficient estimate | SE | Wald test statistics | \(p-\)value | | | Intercept class 1 (not estimated) | - | - | - | - | | Intercept class 2 | -24.00 | 7.25 | -3.31 | 0.00 | | Intercept class 3 | 1.07 | 0.30 | 3.58 | 0.00 | | Time class 1 | 2.40 | 0.22 | 10.69 | 0.00 | | Time class 2 | 5.65 | 1.44 | 3.92 | 0.00 | | Time class 3 | -0.13 | 0.05 | -2.59 | 0.01 | | Male | 0.35 | 0.12 | 2.87 | 0.00 | | Non-white | 0.42 | 0.13 | 3.32 | 0.00 | Legend : SE: standard error. Figure 1 : Illustration of the re-admission and overlap adjusted PDC (Proportion of Days Covered). *For non-disruption readmissions. not-yet-known not-yet-known not-yet-known unknown Figure 2: Comparison between white males and females at age 75. Figure 3: Comparison Between White Females and Non-white Females at age 75. Information & Authors Information Version history Copyright This work is licensed under a Non Exclusive No Reuse License.

Authors Metrics & Citations Metrics Article Usage 469views 125downloads Citations Download citation Shuo Sun, Rafaella Cazé de Medeiros, Julianne D. Brooks, et al. Patterns of Anti-Seizure Medication Non-adherence in Post-Stroke Prophylaxis Among Older Adults. Authorea. 29 April 2025. DOI: https://doi.org/10.22541/au.174589943.38621886/v1 DOI: https://doi.org/10.22541/au.174589943.38621886/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.