Factors driving increases in Clostridioides difficile infection rates in England: A national case-control study, 2019 to 2024

Factors driving increases in Clostridioides difficile infection rates in England: A national case-control study, 2019 to 2024 | 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 Factors driving increases in Clostridioides difficile infection rates in England: A national case-control study, 2019 to 2024 Timothy Whiteley, Andre Charlett, Dimple Chudasama, Zahin Amin-Chowdhury, and 9 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8165881/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Numbers of Clostridioides difficile infections (CDI) in England have increased since 2021. The factors driving the increase are unknown. We aimed to understand the factors associated with CDI and what proportion of this recent increase could be explained by changes to these risk factors. Methods: Case-control study, with CDI case data linked to hospitalisation episodes. Control cases were non-CDI related hospitalisation episodes. A logistic regression model estimated odds ratios (ORs) for individual and regional demographic factors, comorbidities, testing rates and primary care antibiotic consumption. To assess temporal changes, the model was fitted to data from financial year 2019/20 and used to predict cases in 2020/21 onwards, compared to observed data. Results: Factors associated with testing positive for CDI were age, white ethnicity, comorbidities and primary care prescribing. The regions with the highest ORs were the North West and Yorkshire and Humber. Predicting cases using 2019/20 coefficients showed that demographic changes and antibiotic use explained 12.5% of the difference between observed cases and our baseline model in 2023/24. Including comorbidities and testing rates explained 63% of this increase. Conclusion: The strongest risk factors for CDI were age and antibiotic consumption. Increased hospitalisations, comorbidities and testing explained a large proportion of the rise in CDI cases whereas an ageing population, increases in antibiotic prescribing and demographic changes predicted smaller increases in CDI cases. However, the factors considered were not enough to fully explain the rise in cases and further investigation is needed into other factors that may be contributing. Statistical Epidemiology CDI Clostridioides difficile logistic regression Figures Figure 1 Figure 2 Figure 3 Introduction Clostridioides difficile infection (CDI) is a major healthcare-associated infection worldwide associated with increased risk of morbidity and mortality [1]. CDI is associated with a considerable economic healthcare burden, with hospitalisation for CDI requiring an average length of stay of 17 days for an initial episode, extending to 33 days for recurrent CDI [1]. Average total costs per patient to an NHS hospital range from £12,710 to £31,121 [1]. C. difficile is transmitted via the faecal-oral route by spores which may contaminate the environment within healthcare settings [2]. CDI has a devastating impact on patients’ health-related quality of life with an emotional impact that continues after infection, resulting in isolation and fear of recurrence [3]. Risk factors for CDI include older age (≥ 65 years), recent antibiotic use, the number of antibiotics prescribed and prior hospitalisation [4]. Recurrent CDI infections are common, with around a quarter of patients experiencing a recurrence, and as many as 40–60% multiple recurrences [5]. In 2018/19, CDI rates in England were at their lowest level since surveillance began, at 21.9 cases per 100,000 – declining from 107.6 cases per 100,000 at start of surveillance in 2007/08 – and remained broadly stable between 2013/14 until 2020/21 (ranging from 21.9 to 26.1). From 2020/21, the rates of CDI were increasing annually with a 33% increase in cases between 2020/21 and 2023/24 seen nationally. Increases occurred across all age groups and sexes, and amongst both hospital- and community-onset cases [6]. A descriptive analysis of this increase in cases identified changes in multiple factors which were known to be associated with CDI but was not able to identify a clear reason for this change [6]. A comprehensive understanding of this epidemiological trend and risk factors driving this increase in CDI is essential for informing targeted interventions and policy decisions. The aim of this study was to understand the risk factors associated with CDI and what proportion of this recent increase could be explained by changes to these risk factors. Methods Study design We undertook a population-level case-control study using a logistic regression to estimate the likelihood of an individual acquiring CDI during their hospitalisation. Cases were defined as laboratory-confirmed CDI positive stool samples reported to UKHSA’s mandatory surveillance system [7] between 01 April 2019 and 31 March 2024, with a matching record in the Hospital Episode Statistics database (HES). The HES record was used to source information on patient characteristics, including comorbidities. CDI positive stool samples with no matching HES record were excluded from the study. The HES records matched with CDI positive stool samples were not necessarily for CDI-related hospital admissions, and so the study cases included Community onset CDI cases (patients that tested positive and were treated for CDI in the community and patients that were already infected on admission to hospital) and Hospital onset CDI cases (patients who were infected during their hospital stay). The control population was derived from the population that were admitted to an acute English NHS hospital between 01 April 2019 and 31 March 2024, but excluding HES spells matched to a CDI case. To minimise data size, a random sample of 1% of patients in the control population were used as controls in this study. Data We used data from the following sources. HES We matched CDI cases to their hospital admission using patient NHS number, first name and surname. We considered continuous in-patient spells (CIPs) for each admission. Where a CDI case had multiple admissions to hospital, we chose the spell that is closest to their CDI diagnosis which may have been before or after. Hospital onset (HO) cases were those that had positive samples taken two days or more into their hospital spell where admission is counted as day 0, but before discharge [7]. Everything else was community onset (CO) [7]. The majority of cases, 84%, had a hospitalisation spell less than 30 days before or after a positive CDI result (Appendix A). Sensitivity analysis was conducted excluding positive CDI results > 84 days before or after a hospitalisation spell (Appendix B) and the results were broadly similar. To maximise the data availability, we included these cases. Frailty and comorbidities : These were estimated from the ICD-10 codes listed against the hospital spell. Frailty was calculated as the Dr Foster Fragility metric from Soong et al [8] for which there are seven groups of ICD-10 codes as follows: cognition, mobility, falls, malnutrition, incontinence, dependencies and mood. For comorbidities, we derived a modified Charlson Comorbidity index [9] by removing age, cancer and renal components. Age was considered separately. Cancer (lymphoma, leukaemia and solid tumour) and renal disease (moderate to severe chronic kidney disease) were deemed epidemiologically important and therefore picked out separately from this index. We then summed the remaining factors in the index without previous weighting to give a measure of other comorbidities. We adapted the python package comorbidipy [10] to extract the ICD-10 codes and implement the metric. Previous hospitalisation For each previous hospitalisation spell in the preceding 12 months to the matched HES record, we summed the number of overnight bed-days in hospital (Total CIPS duration). We found that cancer and renal disease were sensitive to the inclusion of previous hospitalisation, and we therefore consider the second order interaction of duration and cancer and renal disease. Testing Trust-level reports of total stool samples tested for CDI were taken from quarterly mandatory laboratory returns data (QMLR). This data was normalised by the number of admissions to the hospital [11] giving a metric of tests per 1000 admissions. Prescribing : Primary care antibiotics and Proton pump inhibitor (PPI) prescribing data comes from the ‘Medicines Dispensed in Primary Care’ (MDPC) dataset [12]. Prescribing that was chosen were: Amoxicillin, antibiotics deemed high risk of CDI (Co-Amoxiclav, Clindamycin, Cephalosporins and Fluoroquinolones), and PPIs. This data gives the month that the prescription was submitted to NHS Business Services Authority for processing. For each antibiotic class, we calculated the difference between the month of prescription and the month of CDI onset (for cases) or hospital admission (for controls). If this difference was 6 months or less, then this antibiotic variable is set to True. Details of the BNF codes used as well as sensitivity analysis on the choice of six months can be seen in appendix C. Deprivation We used the 2019 index of multiple deprivation (IMD)[13] which is the deprivation score of the lower super output area (LSOA) of the postcode registered for the patient. This gives a score is from 1–10 where IMD 10 is the least deprived. The IMD is treated as a factor with the default IMD level as 10. Missing, unmatched, or non-England postcodes were recorded in a category ‘missing’. This metric is static over time. Age & Ethnicity data : We removed any cases older than 130 as this indicated an error in the data. We consolidated ethnicity into the following groups: ‘White’, ‘Black’, ‘Asian’, ‘Mixed’, ‘Other/Unknown’. The reference group for ethnicity was White. Ethnicity data was enhanced using the CHIME methodology as detailed in the UKHSA quality and methodology information for health care acquired infection reports [14]. Region The UKHSA region of the trust. The default region is East Midlands. This is first alphabetically and incidence are mid-range nationally. Financial year and quarter For cases, this is the date of their first positive CDI result. For controls, the date of their admission. This is a factor in the model i.e. each time point is independent. Analysis: Initial investigations into missingness within the data showed that several hospital trusts had significant biased missingness for QMLR testing rate data and sex. We therefore excluded data from 36 trusts with an incomplete CDI testing rate history for the reporting period and excluded two trusts for missingness in sex data. Details can be seen in appendix A. The analysis was performed in R 4.4.0 using the glm function. The output metrics of interest were odds ratios (ORs) and total predicted case numbers. Control cases used weights = 100 to rebalance given we used a 1% sample of our total controls. Sensitivity to this is shown in appendix D. Univariable and multivariable analysis were performed using the same methodology. In the first part of the analysis, we fitted this model to all data from 2019/20 to 2023/24 and identified CDI risk factors in the dataset. To focus on hospital onset we excluded the community onset case data before fitting the model and estimating ORs (and vice versa to get community onset cases). This gives the ORs showing the factors that impact on CDI. In the second part, we sought to understand what has changed between 2019 and 2024 that might cause an increase in cases. We used financial year 2019/20 data only to fit the model coefficients and then predicted cases in future years and compared them to the number of observed cases. To understand the contribution of each factor to the model prediction, we broke this model into sub-models of increasing complexity. We started by fitting a baseline model of hospitalisations only ( \(\:cdi\sim1\) ) to 2019/20 and predicting forward. Then we added a group of variables to the equation, fitted the new model to 2019/20 and predicted forwards. We continued adding groups of variables, beginning with those perceived to be more distal and then those that are more proximal until we reached the original model. This gave us the following: Baseline ( \(\:cdi\sim1\) ): This accounts for total admission numbers only Model 1: Demography : Age, sex, ethnicity, IMD and region Model 2: As above plus prescribing Model 3: As above plus comorbidities and previous hospitalisation The full model: As above plus testing . In no scenario in part 2 were financial year and quarter included in the model. Ethical statement: All data were collected within statutory approvals granted to the UKHSA for infectious disease surveillance and control. Information was held securely and in accordance with the Data Protection Act 2018 and Caldicott guidelines. Results Risk factors for CDI Table 1 shows the breakdown for cases and controls for each variable. The CDI population was significantly older (mean 71.6 years compared to 53.4), predominantly White ethnicity (91.6% compared to 84.5%), frailer (mean 1.9 compared to 0.51), more likely to have renal disease (26.4% compared to 12.4%) and more likely to have had antibiotics (e.g. 11.3% compared to 3.5% recorded as having had Co-Amoxiclav). This is also shown in the univariable analysis column in Table 2. Odds ratios from the full logistic regression model can be seen in Table 2 and Figure 1 with community onset, hospital onset and univariable analysis in Table 2. In the multivariable analysis, almost all factors showed statistical significance except ‘other’ ethnicity, various IMD deciles, Q4 of 2021/22 (compared to Q1 2019), the North East and South West regions (compared to East Midlands) and amoxicillin use (Table 2). Positive associations were seen for age (OR 1.03 per additional year of age), and all comorbidities included especially renal disease (OR of 1.38). Non white ethnicities (compared to white) and male sex (compared to female) all had negative associations with CDI. For aggregated variables, the effect was weaker. Testing rates had an odds ratio of 1.005 per additional stool sample tested per 1000 admissions. As deprivation increased, the odds ratio slightly increased. The regions with the highest odds ratios (reference East Midlands) were the North West (OR 1.17) and Yorkshire and Humber (OR 1.16). The lowest regions were London (OR 0.78) and the South East (OR 0.93). All antibiotics modelled except for amoxicillin showed a strong positive association with the odds of CDI. Fluoroquinolones, cephalosporins and co-amoxiclav had ORs of 1.56, 1.58 and 2.41 respectively. Clindamycin showed the largest association with an odds ratio of 5.22. The odds ratios for financial quarters (Figure 1d) show all quarters have higher odds than 2019/20 Q1, varying from 1.06 to 1.33. We do not however see a clear increasing linear trend. There are some significant differences in community and hospital onset data. For antibiotics, the OR is always substantially higher for community onset than hospital onset. For example, CO clindamycin has an OR of 7.11 compared to 2.70 for HO. For all considered comorbidities, the ORs are increased for HO than CO. For example, the OR for cancer goes from 0.93 (negative association) for CO to 1.36 (positive association) and HO. Fitting 2019 and predicting forwards The model predicts an overall increase in CDI cases based on underlying changes to the characteristics of the hospitalised population, but it does not fully explain the variation in observed case numbers (Figure 2). Total predicted cases rose from 9273 in 2019/20 to 11074 in 2023/24. However, observed cases rose more steeply to 11752 cases. Therefore 94% of the cases are explained using this model (Appendix E). There is a significant step change around 2020/21 where the model underpredicts cases (88% explained, Appendix E) and that underprediction continues for future years. This trend is also true for both community and hospital onset cases with 94.7% and 93.8% of cases explained respectively. The hospital onset model fits better over the COVID pandemic period than the community onset, with 94% explained for HO compared to 84% for CO. The HO model fits less well in 2022/23, explaining only 86% of cases compared to 99% explained by the CO model. Figure 3 shows testing and comorbidities as the key drivers of the rise in the model. Corresponding data points are in Appendix E. The baseline model gives the overall shape of our prediction and shows a rise in cases since a low in 2020/21. In 2023/24 there were 9797 baseline cases compared to 11752 observed. Adding demography (Model 1) explains 6% of the difference between baseline and observed. Adding prescribing (Model 2) increases predicted cases in 2023/24 to 10043, explaining 12.5%. Adding comorbidities (Model 3) increases the predicted number of cases to 10370. This is 29% of the difference between baseline and observed. Adding testing rates into the model further increases the prediction to 11032 cases in 2023/34, explaining 63% of the difference between baseline and observed. Discussion The odds ratios that we see in the model are in line with risk factors observed in previous studies [4]. Increased age, frailty and comorbidities are all known risk factors for CDI, the scale of which is seen in the ORs. We also see that there are lower odds of infection for males and non-White ethnicities. Regional differences also tie in with previous work [6] and we encourage further research into unmodelled drivers of regional difference such as demographic factors, individual-level deprivation or local hospital conditions. Antibiotic consumption provided the highest discrete risk factor for the individual, especially Clindamycin (OR of 5.22) confirming known associations with this antibiotic [15]. All other antibiotics except Amoxicillin increased the risk of CDI significantly. Therefore, extra care must be taken to protect individuals who have had previous exposure to any of these antibiotics, aligning with current research emphasising the impact of antibiotics on the gut microbiome [16]. The difference between the ORs for community and hospital onset must be explained by context as well as epidemiology. For example, a cancer patient is likely to be in hospital regularly and therefore, they are more likely to pick up their CDI in the hospital rather than the community versus someone with average hospital exposure. Conversely, antibiotics usage data is only for community prescribing and excludes antibiotic exposure information for anyone who took antibiotics in hospital or was prescribed via outpatient services. Whilst there are also different risk factors in hospital such as transmission, it is difficult to disentangle these without hospital prescribing data. Temporal analysis modelling showed that changes in the frequency of exposures known to be associated with CDI risk would predict an increase in CDI cases, but not to the extent that we observed, and sources of variation exist which were not captured in this model. Using coefficients derived from 2019, cases were predicted to increase over time at a similar rate to observed, but the total number of predicted cases were lower overall. The model significantly underpredicted cases in 2020/21 and this carried forward to later years. This implies that there might be a lasting effect of the COVID-19 pandemic that is not captured by our model but which is ongoing, such as changes in hospital pressures, treatment delays or increased transmission. This model does not support changing demography, increased aging and increased antibiotic use as substantial drivers of the modelled increase. The case increase that is explained by the model is predicted by increases in: overall hospitalisation rates; comorbidities especially renal disease, increased frailty, and testing rate increases. Limitations The comparison population for this study are all hospitalised patients. This has two limitations. Firstly, there may be hospitalised patients whose risk of CDI is unrelated to their hospitalisation but might impact the odds ratios. For example, we do not exclude under 2s or maternity cases from our control cases which might modestly increase the odds for age and male sex respectively. Secondly, the hospitalised population will have changed significantly over time. If, for example, control cases were reduced, then the comparative chance of contracting CDI in the logistic regression would increase. The association between testing rates and risk of CDI should be interpreted with care. Although an increase in testing might explain a rise in cases, it may also be a response to a rise in cases. It is worth noting that although testing rates have increased at a national level, this may not be the case in all trusts with admitted patients that were included in the study. At a national level, the test positivity rate is relatively unchanged over the past five years and shows consistent seasonal variation, with peaks in Q3 (October – December) and troughs in Q1 (April – June) [6]. Because the positivity rates have not declined as testing rates have increased, the association between testing rates and risk of CDI observed in this study indicates that testing has increased in response to an increase in the underlying risk of CDI amongst people presenting with diarrhoea, and clinicians’ testing thresholds have remained constant. The increase in testing might be also caused by an increase in overall diarrhoea causing illnesses that may also explain why both renal disease and frailty (which includes incontinence) have increased over time. This model shows correlation, not causation and care must be taken when interpreting the results. For example, we might think antibiotics cause an infection by disrupting the gut microbiome; but they may also correlate with other comorbidities not captured that make the individual more susceptible to infection. In this model we do not consider recurrent infections; each infection or hospitalisation is considered as a separate data point. This was the most parsimonious assumption we could make. Data limitations: Deprivation as a risk factor is not captured at the individual level, nor does it vary over time, this study may therefore underestimate its impact. Hospital level risk factors such as care in unconventional places (‘corridor care’), staff pressures, or IPC compliance do not have sufficient data to include in the model and they may be explanatory factors. Antibiotic data has two key limitations. Firstly, we are missing any individual data on hospital prescriptions; meaning we may be missing key causal links. Secondly, the date of claiming a prescription for primary care data means it is possible that a prescription occurs after the CDI incident or hospitalisation episode. However, to exclude this possibility we would lose cases where antibiotic prescribing occurs close to infection which would underestimate the effect of prescribing on the odds of infection. This underscores the need for individual level prescribing data, with dates that reflect exposure to improve the accuracy of our modelling. Lastly, we exclude any individuals without a hospitalisation episode (less than 1.9%, Appendix A). This is unavoidable as the data on their comorbidities are not available in the data sources used in this study. We include the small number of CDI cases that has a HES match which is significantly further away from their CDI episode (E.g. 9% of cases had their CDI episode more than 84 days before or after their hospitalisation; Appendix A & B). This may mean that their comorbidities and regional data may have changed in this time. However, this allows us to include as much information on true community cases as we can. Conclusion The study shows that a substantial proportion of the increase of cases of CDI could be explained by changes in hospitalisations, comorbidities and testing rates. Conversely, the model allows us to say that an aging population and recent antibiotic use in primary care only explain a small proportion of the rises we see. There is a significant underestimation of the number of cases predicted, beginning in the COVID-19 pandemic (2020/21) and continuing onwards. We hope that further work can explore what might be underpinning this unexplained rise in cases, and that further data on hospital pressures and hospital antibiotic prescribing might soon be available to improve the capabilities of this model and better understand the key drivers associated with the recent increase in CDI incidence. Declarations Funding statement: This work was primarily supported by the UKHSA. J.R. was supported by the National Institute for Health Research (NIHR) Health Protection Research Unit in Healthcare Associated Infections and Antimicrobial Resistance at the University of Oxford (NIHR200915) in partnership with the UKHSA. References Tresman R, Goldenberg SD. Healthcare resource use and attributable cost of Clostridium difficile infection: a micro-costing analysis comparing first and recurrent episodes. Journal of Antimicrobial Chemotherapy. 2018 Oct 1;73(10):2851–5. Nicholson WL, Munakata N, Horneck G, Melosh HJ, Setlow P. Resistance of Bacillus Endospores to Extreme Terrestrial and Extraterrestrial Environments. Microbiol Mol Biol Rev. 2000 Sept;64(3):548–72. Armstrong EP, Malone DC, Franic DM, Pham SV, Gratie D, Amin A. Patient Experiences with Clostridioides difficile Infection and Its Treatment: A Systematic Literature Review. Infect Dis Ther. 2023 July;12(7):1775–95. Davies K, Lawrence J, Berry C, Davis G, Yu H, Cai B, et al. Risk Factors for Primary Clostridium difficile Infection; Results From the Observational Study of Risk Factors for Clostridium difficile Infection in Hospitalized Patients With Infective Diarrhea (ORCHID). Front Public Health. 2020 July 17;8:293. Ghosh S, Antunes A, Rinta-Kokko H, Chaparova E, Lay-Flurrie S, Tricotel A, et al. Clostridioides difficile infections, recurrences, and clinical outcomes in real-world settings from 2015 to 2019: The RECUR England study. International Journal of Infectious Diseases. 2024 Mar 1;140:31–8. UK Health Security Agency. Increase in Clostridioides difficile infections (CDI): current epidemiology, data and investigations – Technical report [Internet]. GOV.UK. [cited 2025 Oct 17]. Available from: https://www.gov.uk/government/publications/increase-in-clostridioides-difficile-infections-technical-report/increase-in-clostridioides-difficile-infections-cdi-current-epidemiology-data-and-investigations-technical-report UK Health Security Agency. Mandatory enhanced MRSA, MSSA and Gram-negative bacteraemia, and Clostridioides difficile infection surveillance - Protocol version 4.4 [Internet]. GOV.UK. [cited 2025 Nov 12]. Available from: https://hcaidcs.ukhsa.gov.uk/ContentManagement/LinksAndAnnouncements/HCAIDCS_Mandatory_Surveillance_Protocol_v4.4.pdf Soong JTY, Gammall J, Liew D, Peden CJ, Bottle A, Bell D, et al. Dr Foster global frailty score: an international retrospective observational study developing and validating a risk prediction model for hospitalised older persons from administrative data sets. BMJ Open. 2019 June 1;9(6):e026759. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al. Coding Algorithms for Defining Comorbidities in ICD-9-CM and ICD-10 Administrative Data. Medical Care. 2005 Nov;43(11):1130. vvcb. comorbidipy: Python package to calculate comorbidity scores and other clinical risk scores. [Internet]. [cited 2025 Oct 23]. Available from: https://github.com/vvcb/comorbidipy Fingertips | Department of Health and Social Care [Internet]. [cited 2025 Oct 24]. Available from: https://fingertips.phe.org.uk/ NHS England. Medicines dispensed in Primary Care NHS Business Services Authority data product [Internet]. Medicines dispensed in Primary Care NHS Business Services Authority data product. [cited 2025 Oct 17]. Available from: https://digital.nhs.uk/data-and-information/data-tools-and-services/data-services/medicines-dispensed-in-primary-care-nhsbsa-data English indices of deprivation 2019 [Internet]. GOV.UK. [cited 2025 Nov 18]. Available from: https://www.gov.uk/government/statistics/english-indices-of-deprivation-2019 Quality and methodology information (QMI) for healthcare-associated infections (HCAI) reports [Internet]. GOV.UK. [cited 2025 Oct 17]. Available from: https://www.gov.uk/government/statistics/hcai-qmi-report/quality-and-methodology-information-qmi-for-healthcare-associated-infections-hcai-reports Clindamycin | Drugs | BNF content published by NICE [Internet]. [cited 2025 Oct 23]. Available from: https://bnf.nice.org.uk/drugs/clindamycin/ Cusumano G, Flores GA, Venanzoni R, Angelini P. The Impact of Antibiotic Therapy on Intestinal Microbiota: Dysbiosis, Antibiotic Resistance, and Restoration Strategies. Antibiotics. 2025 Apr 3;14(4):371. Tables Tables 1 and 2 are available in the Supplementary Files section Additional Declarations The authors declare no competing interests. Supplementary Files Table1.xlsx Table 1: Characteristics of the cases and controls used in the model. We present total numbers and percentage breakdowns for categorical variables and means and standard deviation for continuous variables. Table2.xlsx Table 2: Odds ratios and p-values for each modelled factors, broken down into all cases, community onset, hospital onset and univariate. We also show the p-value that community onset and hospital onset values are significantly different. AppendixAmissingness.docx Appendix A: Data missingness AppendixBCompaingwithcloseCOcases.docx Appendix B: Comparing CO cases including and excluding those >84 days before or after hospitalisation AppendixCAntibiotics.docx Appendix C: Antibiotics used and sensitivity analysis to choice of 6 months AppendixD0pt1pc.docx Appendix D: Sensitivity to the choice of a 1% random sample AppendixETablescorrespondingtofigures.docx Appendix E: Tables corresponding to figures. Cite Share Download PDF Status: Posted Version 1 posted 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-8165881","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":548245846,"identity":"79040495-19f9-44c1-ab86-e98570477ac5","order_by":0,"name":"Timothy Whiteley","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABCUlEQVRIiWNgGAWjYBAC+8MMDMy8DRIMDOxA3geGhAS4lAQOLWzMEC0SDEAG4wyitDCAtTCAtTDzEKWFnTuBmXeHRR0/M/Mxadu2tDz59gPMLz62MSTObMDlMN4NzLlnJCQkm9nSpHPbcooNziSwWc4EapmN0y8gLW0SEgaHecyAWioSNzAksBnznGFInEdIi/1h/m/SlkAt8/sfEKHlL8gWZh42aca2nMSGGwnMj3kq8Dvs8Nw2CckZh9mMLXvOpSVuuPGwjXFGhYQxTu/zn934mLetjp+/vfnhjR9lyUCHJR/+8MHARnbGARzWAAFMigUaE4xtErhjBRUwf0BnjIJRMApGwSgAAQBqQFKCD6F2HgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0003-4990-4644","institution":"UK Health Security Agency","correspondingAuthor":true,"prefix":"","firstName":"Timothy","middleName":"","lastName":"Whiteley","suffix":""},{"id":548560308,"identity":"5bdef221-6d0c-44a9-b740-29816da0ee51","order_by":1,"name":"Andre Charlett","email":"","orcid":"https://orcid.org/0000-0001-7154-0432","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Andre","middleName":"","lastName":"Charlett","suffix":""},{"id":548560309,"identity":"47bdb500-b243-4393-a70d-6c29fe09dbe6","order_by":2,"name":"Dimple Chudasama","email":"","orcid":"https://orcid.org/0000-0002-2214-5900","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Dimple","middleName":"","lastName":"Chudasama","suffix":""},{"id":548560310,"identity":"824d7a26-96c9-4dce-b82e-14c380725c60","order_by":3,"name":"Zahin Amin-Chowdhury","email":"","orcid":"https://orcid.org/0000-0002-3106-992X","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Zahin","middleName":"","lastName":"Amin-Chowdhury","suffix":""},{"id":548560311,"identity":"df4750f8-4610-4a78-9d74-d0b9ddebf1fd","order_by":4,"name":"Russell Hope","email":"","orcid":"https://orcid.org/0000-0003-2838-6833","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Russell","middleName":"","lastName":"Hope","suffix":""},{"id":548560312,"identity":"e0cf2c1e-52d0-4a2c-b8cc-0e760d89879b","order_by":5,"name":"Laura Fellowes","email":"","orcid":"https://orcid.org/0000-0002-0167-1872","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Laura","middleName":"","lastName":"Fellowes","suffix":""},{"id":548560313,"identity":"c4b557b8-42b4-4711-b34b-ad0c4778f96b","order_by":6,"name":"Dakshika Jeyaratnam","email":"","orcid":"https://orcid.org/0000-0001-6671-1073","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Dakshika","middleName":"","lastName":"Jeyaratnam","suffix":""},{"id":548560314,"identity":"4ddb9a13-878c-4bea-a848-adaab84c74ac","order_by":7,"name":"Rebecca Oettle","email":"","orcid":"","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Rebecca","middleName":"","lastName":"Oettle","suffix":""},{"id":548560315,"identity":"d04fe8cb-dbe8-499b-93f2-6391286ce68c","order_by":8,"name":"Julie Robotham","email":"","orcid":"https://orcid.org/0000-0003-2515-4084","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Julie","middleName":"","lastName":"Robotham","suffix":""},{"id":548560316,"identity":"a05d5272-6fb8-476f-85e9-51f44e74de96","order_by":9,"name":"Charlotte Stevens","email":"","orcid":"","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Charlotte","middleName":"","lastName":"Stevens","suffix":""},{"id":548560317,"identity":"e36501d1-c8cc-498c-b1ca-5925f1a56985","order_by":10,"name":"James Stimson","email":"","orcid":"https://orcid.org/0000-0003-1855-6229","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"James","middleName":"","lastName":"Stimson","suffix":""},{"id":548560318,"identity":"3445e915-b94a-43d8-ab13-f6b9c777c8b6","order_by":11,"name":"Thomas Inns","email":"","orcid":"https://orcid.org/0000-0002-1218-958X","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Inns","suffix":""},{"id":548560319,"identity":"68daad2a-9361-4e5d-9215-771b6d8c11b4","order_by":12,"name":"Stephanie Evans","email":"","orcid":"https://orcid.org/0000-0002-2345-5249","institution":"UK Health Security Agency","correspondingAuthor":false,"prefix":"","firstName":"Stephanie","middleName":"","lastName":"Evans","suffix":""}],"badges":[],"createdAt":"2025-11-20 15:04:03","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-8165881/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8165881/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":96805826,"identity":"2c285fe8-b4d4-4d97-9d78-8ffe6b85a446","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":83236,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/20abf879404027e10f6ef515.docx"},{"id":96916125,"identity":"0f0ff516-2c4b-4d8e-ab81-45bf3da964d8","added_by":"auto","created_at":"2025-11-27 14:08:04","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":342,"visible":true,"origin":"","legend":"","description":"","filename":"rs8165881.json","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/cca7eab289b12b919da46a7e.json"},{"id":96805828,"identity":"c2b654b7-f022-49f8-9fea-c7762575f7d6","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":43046,"visible":true,"origin":"","legend":"","description":"","filename":"rs81658810enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/f8c5db30406a23b1dd77fedd.xml"},{"id":96917374,"identity":"7d2eedc9-4cea-4b91-a223-ba1672c8f985","added_by":"auto","created_at":"2025-11-27 14:09:38","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40543,"visible":true,"origin":"","legend":"","description":"","filename":"rs81658810structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/fc29b5bcfe66faf324c65fe1.xml"},{"id":96917294,"identity":"b9c97d18-5243-4d16-9b9f-80e020bb5492","added_by":"auto","created_at":"2025-11-27 14:09:31","extension":"html","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":47153,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/a428d96ccc40ba9234576579.html"},{"id":96805823,"identity":"7c475903-458b-4d78-bd76-2473dfea10a9","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1093259,"visible":true,"origin":"","legend":"\u003cp\u003eOdds ratios from the logistic regression model for (a) discrete variables (b) continuous variables (c) Prescribing data (d) Financial quarters.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/8c37344e3a50f6aad4bb86f1.png"},{"id":96805830,"identity":"28c25eef-7cc2-4b3e-a32a-d9b8dbd4929d","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":772253,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted and observed cases using the logistic regression model trained on 2019/20 data and predicting 2020/21 onwards. (a) All cases (b) Community onset (c) Hospital onset.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/de857f35974f9a445d627a44.png"},{"id":96805834,"identity":"97379ed3-1e50-4857-af3f-a198ef7599a2","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":985834,"visible":true,"origin":"","legend":"\u003cp\u003eBreakdown of the model into submodels, fitted to 2019/20 and predicting 2020/21 onwards. (a) Predicted and observed cases. (b) The percentage difference between baseline and observed cases.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/705a2618da37aed1c32a5732.png"},{"id":96923235,"identity":"edd7a7a8-3dca-4032-9485-17fb67c9cbf7","added_by":"auto","created_at":"2025-11-27 14:21:19","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2950925,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/69b79efd-8f35-4b34-93bf-a02ec71d13e1.pdf"},{"id":96805822,"identity":"4f61a3b4-55a4-4c6d-afc6-00c612bce69d","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":19431,"visible":true,"origin":"","legend":"\u003cp\u003eTable 1: Characteristics of the cases and controls used in the model. We present total numbers and percentage breakdowns for categorical variables and means and standard deviation for continuous variables.\u003c/p\u003e","description":"","filename":"Table1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/d267a709d5d7ecca376f5bcd.xlsx"},{"id":96805825,"identity":"52683c40-9543-478e-8f67-8d0c9bde9e97","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":20647,"visible":true,"origin":"","legend":"\u003cp\u003eTable 2: Odds ratios and p-values for each modelled factors, broken down into all cases, community onset, hospital onset and univariate. We also show the p-value that community onset and hospital onset values are significantly different.\u003c/p\u003e","description":"","filename":"Table2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/a4e79f9d9eba44dacfb20550.xlsx"},{"id":96917195,"identity":"16dbe1bb-1aaf-4b20-94cf-43853c1c3cdd","added_by":"auto","created_at":"2025-11-27 14:09:21","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":26003,"visible":true,"origin":"","legend":"\u003cp\u003eAppendix A: Data missingness\u003c/p\u003e","description":"","filename":"AppendixAmissingness.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/fd2103fa8624feb2bdebf326.docx"},{"id":96805832,"identity":"d7827f66-4e4a-4b6b-9168-8812845b03d9","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"docx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":31151,"visible":true,"origin":"","legend":"\u003cp\u003eAppendix B: Comparing CO cases including and excluding those \u0026gt;84 days before or after hospitalisation\u003c/p\u003e","description":"","filename":"AppendixBCompaingwithcloseCOcases.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/1a1fa977852ae16700456bfa.docx"},{"id":96805836,"identity":"a96a92e7-ab6b-42c7-8dcb-32bd54827776","added_by":"auto","created_at":"2025-11-26 09:11:55","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":98432,"visible":true,"origin":"","legend":"\u003cp\u003eAppendix C: Antibiotics used and sensitivity analysis to choice of 6 months\u003c/p\u003e","description":"","filename":"AppendixCAntibiotics.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/18ceff02769ce1d144c51511.docx"},{"id":96916327,"identity":"79f7aa7f-694e-4c07-a9b9-1e641f79e25d","added_by":"auto","created_at":"2025-11-27 14:08:28","extension":"docx","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":27710,"visible":true,"origin":"","legend":"\u003cp\u003eAppendix D: Sensitivity to the choice of a 1% random sample\u003c/p\u003e","description":"","filename":"AppendixD0pt1pc.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/d693c09cca5bbd394e33926d.docx"},{"id":96918242,"identity":"d6d41c58-95fb-4067-ba89-0577de26c526","added_by":"auto","created_at":"2025-11-27 14:11:28","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":27587,"visible":true,"origin":"","legend":"\u003cp\u003eAppendix E: Tables corresponding to figures.\u003c/p\u003e","description":"","filename":"AppendixETablescorrespondingtofigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-8165881/v1/5483f23fac71e5541fa5d5ea.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eFactors driving increases in \u003cem\u003eClostridioides difficile\u003c/em\u003e infection rates in England: A national case-control study, 2019 to 2024\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003e\u003cem\u003eClostridioides difficile\u003c/em\u003e infection (CDI) is a major healthcare-associated infection worldwide associated with increased risk of morbidity and mortality [1]. CDI is associated with a considerable economic healthcare burden, with hospitalisation for CDI requiring an average length of stay of 17 days for an initial episode, extending to 33 days for recurrent CDI [1]. Average total costs per patient to an NHS hospital range from \u0026pound;12,710 to \u0026pound;31,121 [1]. \u003cem\u003eC. difficile\u003c/em\u003e is transmitted via the faecal-oral route by spores which may contaminate the environment within healthcare settings [2]. CDI has a devastating impact on patients\u0026rsquo; health-related quality of life with an emotional impact that continues after infection, resulting in isolation and fear of recurrence [3].\u003c/p\u003e\u003cp\u003eRisk factors for CDI include older age (\u0026ge;\u0026thinsp;65 years), recent antibiotic use, the number of antibiotics prescribed and prior hospitalisation [4]. Recurrent CDI infections are common, with around a quarter of patients experiencing a recurrence, and as many as 40\u0026ndash;60% multiple recurrences [5].\u003c/p\u003e\u003cp\u003eIn 2018/19, CDI rates in England were at their lowest level since surveillance began, at 21.9 cases per 100,000 \u0026ndash; declining from 107.6 cases per 100,000 at start of surveillance in 2007/08 \u0026ndash; and remained broadly stable between 2013/14 until 2020/21 (ranging from 21.9 to 26.1). From 2020/21, the rates of CDI were increasing annually with a 33% increase in cases between 2020/21 and 2023/24 seen nationally. Increases occurred across all age groups and sexes, and amongst both hospital- and community-onset cases [6]. A descriptive analysis of this increase in cases identified changes in multiple factors which were known to be associated with CDI but was not able to identify a clear reason for this change [6].\u003c/p\u003e\u003cp\u003eA comprehensive understanding of this epidemiological trend and risk factors driving this increase in CDI is essential for informing targeted interventions and policy decisions. The aim of this study was to understand the risk factors associated with CDI and what proportion of this recent increase could be explained by changes to these risk factors.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eStudy design\u003c/h2\u003e\n \u003cp\u003eWe undertook a population-level case-control study using a logistic regression to estimate the likelihood of an individual acquiring CDI during their hospitalisation. Cases were defined as laboratory-confirmed CDI positive stool samples reported to UKHSA\u0026rsquo;s mandatory surveillance system [7] between 01 April 2019 and 31 March 2024, with a matching record in the Hospital Episode Statistics database (HES). The HES record was used to source information on patient characteristics, including comorbidities. CDI positive stool samples with no matching HES record were excluded from the study. The HES records matched with CDI positive stool samples were not necessarily for CDI-related hospital admissions, and so the study cases included Community onset CDI cases (patients that tested positive and were treated for CDI in the community and patients that were already infected on admission to hospital) and Hospital onset CDI cases (patients who were infected during their hospital stay). The control population was derived from the population that were admitted to an acute English NHS hospital between 01 April 2019 and 31 March 2024, but excluding HES spells matched to a CDI case. To minimise data size, a random sample of 1% of patients in the control population were used as controls in this study.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eData\u003c/h3\u003e\n\u003cp\u003eWe used data from the following sources.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHES\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe matched CDI cases to their hospital admission using patient NHS number, first name and surname. We considered continuous in-patient spells (CIPs) for each admission. Where a CDI case had multiple admissions to hospital, we chose the spell that is closest to their CDI diagnosis which may have been before or after. Hospital onset (HO) cases were those that had positive samples taken two days or more into their hospital spell where admission is counted as day 0, but before discharge [7]. Everything else was community onset (CO) [7]. The majority of cases, 84%, had a hospitalisation spell less than 30 days before or after a positive CDI result (Appendix A). Sensitivity analysis was conducted excluding positive CDI results\u0026thinsp;\u0026gt;\u0026thinsp;84 days before or after a hospitalisation spell (Appendix B) and the results were broadly similar. To maximise the data availability, we included these cases.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eFrailty and comorbidities\u003c/em\u003e: These were estimated from the ICD-10 codes listed against the hospital spell. Frailty was calculated as the Dr Foster Fragility metric from Soong et al [8] for which there are seven groups of ICD-10 codes as follows: cognition, mobility, falls, malnutrition, incontinence, dependencies and mood. For comorbidities, we derived a modified Charlson Comorbidity index [9] by removing age, cancer and renal components. Age was considered separately. Cancer (lymphoma, leukaemia and solid tumour) and renal disease (moderate to severe chronic kidney disease) were deemed epidemiologically important and therefore picked out separately from this index. We then summed the remaining factors in the index without previous weighting to give a measure of other comorbidities. We adapted the python package \u003cem\u003ecomorbidipy\u003c/em\u003e [10] to extract the ICD-10 codes and implement the metric.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePrevious hospitalisation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor each previous hospitalisation spell in the preceding 12 months to the matched HES record, we summed the number of overnight bed-days in hospital (Total CIPS duration). We found that cancer and renal disease were sensitive to the inclusion of previous hospitalisation, and we therefore consider the second order interaction of duration and cancer and renal disease.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTesting\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTrust-level reports of total stool samples tested for CDI were taken from quarterly mandatory laboratory returns data (QMLR). This data was normalised by the number of admissions to the hospital [11] giving a metric of tests per 1000 admissions.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ePrescribing\u003c/em\u003e: Primary care antibiotics and Proton pump inhibitor (PPI) prescribing data comes from the \u0026lsquo;Medicines Dispensed in Primary Care\u0026rsquo; (MDPC) dataset [12]. Prescribing that was chosen were: Amoxicillin, antibiotics deemed high risk of CDI (Co-Amoxiclav, Clindamycin, Cephalosporins and Fluoroquinolones), and PPIs. This data gives the month that the prescription was submitted to NHS Business Services Authority for processing. For each antibiotic class, we calculated the difference between the month of prescription and the month of CDI onset (for cases) or hospital admission (for controls). If this difference was 6 months or less, then this antibiotic variable is set to True. Details of the BNF codes used as well as sensitivity analysis on the choice of six months can be seen in appendix C.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeprivation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe used the 2019 index of multiple deprivation (IMD)[13] which is the deprivation score of the lower super output area (LSOA) of the postcode registered for the patient. This gives a score is from 1\u0026ndash;10 where IMD 10 is the least deprived. The IMD is treated as a factor with the default IMD level as 10. Missing, unmatched, or non-England postcodes were recorded in a category \u0026lsquo;missing\u0026rsquo;. This metric is static over time.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eAge \u0026amp; Ethnicity data\u003c/em\u003e: We removed any cases older than 130 as this indicated an error in the data. We consolidated ethnicity into the following groups: \u0026lsquo;White\u0026rsquo;, \u0026lsquo;Black\u0026rsquo;, \u0026lsquo;Asian\u0026rsquo;, \u0026lsquo;Mixed\u0026rsquo;, \u0026lsquo;Other/Unknown\u0026rsquo;. The reference group for ethnicity was White. Ethnicity data was enhanced using the CHIME methodology as detailed in the UKHSA quality and methodology information for health care acquired infection reports [14].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRegion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe UKHSA region of the trust. The default region is East Midlands. This is first alphabetically and incidence are mid-range nationally.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFinancial year and quarter\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFor cases, this is the date of their first positive CDI result. For controls, the date of their admission. This is a factor in the model i.e. each time point is independent.\u003c/p\u003e\n\u003cp\u003eAnalysis:\u003c/p\u003e\n\u003cp\u003eInitial investigations into missingness within the data showed that several hospital trusts had significant biased missingness for QMLR testing rate data and sex. We therefore excluded data from 36 trusts with an incomplete CDI testing rate history for the reporting period and excluded two trusts for missingness in sex data. Details can be seen in appendix A.\u003c/p\u003e\n\u003cp\u003eThe analysis was performed in R 4.4.0 using the glm function. The output metrics of interest were odds ratios (ORs) and total predicted case numbers. Control cases used \u003cem\u003eweights\u0026thinsp;=\u0026thinsp;100\u003c/em\u003e to rebalance given we used a 1% sample of our total controls. Sensitivity to this is shown in appendix D. Univariable and multivariable analysis were performed using the same methodology.\u003c/p\u003e\n\u003cp\u003eIn the first part of the analysis, we fitted this model to all data from 2019/20 to 2023/24 and identified CDI risk factors in the dataset. To focus on hospital onset we excluded the community onset case data before fitting the model and estimating ORs (and vice versa to get community onset cases). This gives the ORs showing the factors that impact on CDI.\u003c/p\u003e\n\u003cp\u003eIn the second part, we sought to understand what has changed between 2019 and 2024 that might cause an increase in cases. We used financial year 2019/20 data only to fit the model coefficients and then predicted cases in future years and compared them to the number of observed cases.\u003c/p\u003e\n\u003cp\u003eTo understand the contribution of each factor to the model prediction, we broke this model into sub-models of increasing complexity. We started by fitting a baseline model of hospitalisations only (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:cdi\\sim1\\)\u003c/span\u003e\u003c/span\u003e) to 2019/20 and predicting forward. Then we added a group of variables to the equation, fitted the new model to 2019/20 and predicted forwards. We continued adding groups of variables, beginning with those perceived to be more distal and then those that are more proximal until we reached the original model. This gave us the following:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eBaseline (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:cdi\\sim1\\)\u003c/span\u003e\u003c/span\u003e): This accounts for \u003cstrong\u003etotal admission numbers\u003c/strong\u003e only\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eModel 1: \u003cstrong\u003eDemography\u003c/strong\u003e: Age, sex, ethnicity, IMD and region\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eModel 2: As above plus \u003cstrong\u003eprescribing\u003c/strong\u003e\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eModel 3: As above plus \u003cstrong\u003ecomorbidities\u003c/strong\u003e and previous hospitalisation\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eThe full model: As above plus \u003cstrong\u003etesting\u003c/strong\u003e.\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eIn no scenario in part 2 were financial year and quarter included in the model.\u003c/p\u003e\n\u003cp\u003eEthical statement: All data were collected within statutory approvals granted to the UKHSA for infectious disease surveillance and control. Information was held securely and in accordance with the Data Protection Act 2018 and Caldicott guidelines.\u0026nbsp;\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eRisk factors for CDI\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTable 1 shows the breakdown for cases and controls for each variable. The CDI population was significantly older (mean 71.6 years compared to 53.4), predominantly White ethnicity (91.6% compared to 84.5%), frailer (mean 1.9 compared to 0.51), more likely to have renal disease (26.4% compared to 12.4%) and more likely to have had antibiotics (e.g. 11.3% compared to 3.5% recorded as having had Co-Amoxiclav). This is also shown in the univariable analysis column in Table 2.\u003c/p\u003e\n\u003cp\u003eOdds ratios from the full logistic regression model can be seen in Table 2 and Figure 1 with community onset, hospital onset and univariable analysis in Table 2. In the multivariable analysis, almost all factors showed statistical significance except \u0026lsquo;other\u0026rsquo; ethnicity, various IMD deciles, Q4 of 2021/22 (compared to Q1 2019), the North East and South West regions (compared to East Midlands) and amoxicillin use (Table 2).\u003c/p\u003e\n\u003cp\u003ePositive associations were seen for age (OR 1.03 per additional year of age), and all comorbidities included especially renal disease (OR of 1.38). Non white ethnicities (compared to white) and male sex (compared to female) all had negative associations with CDI.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor aggregated variables, the effect was weaker. Testing rates had an odds ratio of 1.005 per additional stool sample tested per 1000 admissions. As deprivation increased, the odds ratio slightly increased. The regions with the highest odds ratios (reference East Midlands) were the North West (OR 1.17) and Yorkshire and Humber (OR 1.16). The lowest regions were London (OR 0.78) and the South East (OR 0.93).\u003c/p\u003e\n\u003cp\u003eAll antibiotics modelled except for amoxicillin showed a strong positive association with the odds of CDI. Fluoroquinolones, cephalosporins and co-amoxiclav had ORs of 1.56, 1.58 and 2.41 respectively. Clindamycin showed the largest association with an odds ratio of 5.22. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe odds ratios for financial quarters (Figure 1d) show all quarters have higher odds than 2019/20 Q1, varying from 1.06 to 1.33. We do not however see a clear increasing linear trend.\u003c/p\u003e\n\u003cp\u003eThere are some significant differences in community and hospital onset data. For antibiotics, the OR is always substantially higher for community onset than hospital onset. For example, CO clindamycin has an OR of 7.11 compared to 2.70 for HO. For all considered comorbidities, the ORs are increased for HO than CO. For example, the OR for cancer goes from 0.93 (negative association) for CO to 1.36 (positive association) and HO.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFitting 2019 and predicting forwards\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe model predicts an overall increase in CDI cases based on underlying changes to the characteristics of the hospitalised population, but it does not fully explain the variation in observed case numbers (Figure 2). Total predicted cases rose from 9273 in 2019/20 to 11074 in 2023/24. However, observed cases rose more steeply to 11752 cases. Therefore 94% of the cases are explained using this model (Appendix E). There is a significant step change around 2020/21 where the model underpredicts cases (88% explained, Appendix E) and that underprediction continues for future years.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis trend is also true for both community and hospital onset cases with 94.7% and 93.8% of cases explained respectively. The hospital onset model fits better over the COVID pandemic period than the community onset, with 94% explained for HO compared to 84% for CO. The HO model fits less well in 2022/23, explaining only 86% of cases compared to 99% explained by the CO model.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFigure 3 shows testing and comorbidities as the key drivers of the rise in the model. Corresponding data points are in Appendix E. The baseline model gives the overall shape of our prediction and shows a rise in cases since a low in 2020/21. In 2023/24 there were 9797 baseline cases compared to 11752 observed. Adding demography (Model 1) explains 6% of the difference between baseline and observed. Adding prescribing (Model 2) increases predicted cases in 2023/24 to 10043, explaining 12.5%. Adding comorbidities (Model 3) increases the predicted number of cases to 10370. This is 29% of the difference between baseline and observed. Adding testing rates into the model further increases the prediction to 11032 cases in 2023/34, explaining 63% of the difference between baseline and observed.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe odds ratios that we see in the model are in line with risk factors observed in previous studies [4]. Increased age, frailty and comorbidities are all known risk factors for CDI, the scale of which is seen in the ORs. We also see that there are lower odds of infection for males and non-White ethnicities. Regional differences also tie in with previous work [6] and we encourage further research into unmodelled drivers of regional difference such as demographic factors, individual-level deprivation or local hospital conditions.\u003c/p\u003e\n\u003cp\u003eAntibiotic consumption provided the highest discrete risk factor for the individual, especially Clindamycin (OR of 5.22) confirming known associations with this antibiotic [15]. All other antibiotics except Amoxicillin increased the risk of CDI significantly. Therefore, extra care must be taken to protect individuals who have had previous exposure to any of these antibiotics, aligning with current research emphasising the impact of antibiotics on the gut microbiome [16].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe difference between the ORs for community and hospital onset must be explained by context as well as epidemiology. For example, a cancer patient is likely to be in hospital regularly and therefore, they are more likely to pick up their CDI in the hospital rather than the community versus someone with average hospital exposure. Conversely, antibiotics usage data is only for community prescribing and excludes antibiotic exposure information for anyone who took antibiotics in hospital or was prescribed via outpatient services. Whilst there are also different risk factors in hospital such as transmission, it is difficult to disentangle these without hospital prescribing data.\u003c/p\u003e\n\u003cp\u003eTemporal analysis modelling showed that changes in the frequency of exposures known to be associated with CDI risk would predict an increase in CDI cases, but not to the extent that we observed, and sources of variation exist which were not captured in this model. Using coefficients derived from 2019, cases were predicted to increase over time at a similar rate to observed, but the total number of predicted cases were lower overall. The model significantly underpredicted cases in 2020/21 and this carried forward to later years. This implies that there might be a lasting effect of the COVID-19 pandemic that is not captured by our model but which is ongoing, such as changes in hospital pressures, treatment delays or increased transmission.\u003c/p\u003e\n\u003cp\u003eThis model does not support changing demography, increased aging and increased antibiotic use as substantial drivers of the modelled increase. The case increase that is explained by the model is predicted by increases in: overall hospitalisation rates; comorbidities especially renal disease, increased frailty, and testing rate increases.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eLimitations\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eThe comparison population for this study are all hospitalised patients. This has two limitations. Firstly, there may be hospitalised patients whose risk of CDI is unrelated to their hospitalisation but might impact the odds ratios. For example, we do not exclude under 2s or maternity cases from our control cases which might modestly increase the odds for age and male sex respectively. Secondly, the hospitalised population will have changed significantly over time. If, for example, control cases were reduced, then the comparative chance of contracting CDI in the logistic regression would increase.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe association between testing rates and risk of CDI should be interpreted with care. \u0026nbsp; Although an increase in testing might explain a rise in cases, it may also be a response to a rise in cases. It is worth noting that although testing rates have increased at a national level, this may not be the case in all trusts with admitted patients that were included in the study. At a national level, the test positivity rate is relatively unchanged over the past five years and shows consistent seasonal variation, with peaks in Q3 (October \u0026ndash; December) and troughs in Q1 (April \u0026ndash; June) [6]. Because the positivity rates have not declined as testing rates have increased, the association between testing rates and risk of CDI observed in this study indicates that testing has increased in response to an increase in the underlying risk of CDI amongst people presenting with diarrhoea, and clinicians\u0026rsquo; testing thresholds have remained constant. \u0026nbsp;The increase in testing might be also caused by an increase in overall diarrhoea causing illnesses that may also explain why both renal disease and frailty (which includes incontinence) have increased over time.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis model shows correlation, not causation and care must be taken when interpreting the results. For example, we might think antibiotics \u003cem\u003ecause\u003c/em\u003e an infection by disrupting the gut microbiome; but they may also \u003cem\u003ecorrelate\u003c/em\u003e with other comorbidities not captured that make the individual more susceptible to infection.\u003c/p\u003e\n\u003cp\u003eIn this model we do not consider recurrent infections; each infection or hospitalisation is considered as a separate data point. This was the most parsimonious assumption we could make.\u003c/p\u003e\n\u003cp\u003eData limitations: Deprivation as a risk factor is not captured at the individual level, nor does it vary over time, this study may therefore underestimate its impact. Hospital level risk factors such as care in unconventional places (\u0026lsquo;corridor care\u0026rsquo;), staff pressures, or IPC compliance do not have sufficient data to include in the model and they may be explanatory factors. Antibiotic data has two key limitations. Firstly, we are missing any individual data on hospital prescriptions; meaning we may be missing key causal links. Secondly, the date of claiming a prescription for primary care data means it is possible that a prescription occurs after the CDI incident or hospitalisation episode. However, to exclude this possibility we would lose cases where antibiotic prescribing occurs close to infection which would underestimate the effect of prescribing on the odds of infection. This underscores the need for individual level prescribing data, with dates that reflect exposure to improve the accuracy of our modelling.\u003c/p\u003e\n\u003cp\u003eLastly, we exclude any individuals without a hospitalisation episode (less than 1.9%, Appendix A). This is unavoidable as the data on their comorbidities are not available in the data sources used in this study. We include the small number of CDI cases that has a HES match which is significantly further away from their CDI episode (E.g. 9% of cases had their CDI episode more than 84 days before or after their hospitalisation; Appendix A \u0026amp; B). This may mean that their comorbidities and regional data may have changed in this time. However, this allows us to include as much information on true community cases as we can.\u0026nbsp;\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe study shows that a substantial proportion of the increase of cases of CDI could be explained by changes in hospitalisations, comorbidities and testing rates. Conversely, the model allows us to say that an aging population and recent antibiotic use in primary care only explain a small proportion of the rises we see.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThere is a significant underestimation of the number of cases predicted, beginning in the COVID-19 pandemic (2020/21) and continuing onwards. We hope that further work can explore what might be underpinning this unexplained rise in cases, and that further data on hospital pressures and hospital antibiotic prescribing might soon be available to improve the capabilities of this model and better understand the key drivers associated with the recent increase in CDI incidence.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding statement:\u003c/h2\u003e\u003cp\u003eThis work was primarily supported by the UKHSA. J.R. was supported by the National Institute for Health Research (NIHR) Health Protection Research Unit in Healthcare Associated Infections and Antimicrobial Resistance at the University of Oxford (NIHR200915) in partnership with the UKHSA.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eTresman R, Goldenberg SD. Healthcare resource use and attributable cost of Clostridium difficile infection: a micro-costing analysis comparing first and recurrent episodes. Journal of Antimicrobial Chemotherapy. 2018 Oct 1;73(10):2851\u0026ndash;5. \u003c/li\u003e\n\u003cli\u003eNicholson WL, Munakata N, Horneck G, Melosh HJ, Setlow P. Resistance of \u003cem\u003eBacillus\u003c/em\u003e Endospores to Extreme Terrestrial and Extraterrestrial Environments. Microbiol Mol Biol Rev. 2000 Sept;64(3):548\u0026ndash;72. \u003c/li\u003e\n\u003cli\u003eArmstrong EP, Malone DC, Franic DM, Pham SV, Gratie D, Amin A. Patient Experiences with Clostridioides difficile Infection and Its Treatment: A Systematic Literature Review. Infect Dis Ther. 2023 July;12(7):1775\u0026ndash;95. \u003c/li\u003e\n\u003cli\u003eDavies K, Lawrence J, Berry C, Davis G, Yu H, Cai B, et al. Risk Factors for Primary Clostridium difficile Infection; Results From the Observational Study of Risk Factors for Clostridium difficile Infection in Hospitalized Patients With Infective Diarrhea (ORCHID). Front Public Health. 2020 July 17;8:293. \u003c/li\u003e\n\u003cli\u003eGhosh S, Antunes A, Rinta-Kokko H, Chaparova E, Lay-Flurrie S, Tricotel A, et al. Clostridioides difficile infections, recurrences, and clinical outcomes in real-world settings from 2015 to 2019: The RECUR England study. International Journal of Infectious Diseases. 2024 Mar 1;140:31\u0026ndash;8. \u003c/li\u003e\n\u003cli\u003eUK Health Security Agency. Increase in Clostridioides difficile infections (CDI): current epidemiology, data and investigations \u0026ndash; Technical report [Internet]. GOV.UK. [cited 2025 Oct 17]. Available from: https://www.gov.uk/government/publications/increase-in-clostridioides-difficile-infections-technical-report/increase-in-clostridioides-difficile-infections-cdi-current-epidemiology-data-and-investigations-technical-report\u003c/li\u003e\n\u003cli\u003eUK Health Security Agency. Mandatory enhanced MRSA, MSSA and Gram-negative bacteraemia, and Clostridioides difficile infection surveillance - Protocol version 4.4 [Internet]. GOV.UK. [cited 2025 Nov 12]. Available from: https://hcaidcs.ukhsa.gov.uk/ContentManagement/LinksAndAnnouncements/HCAIDCS_Mandatory_Surveillance_Protocol_v4.4.pdf\u003c/li\u003e\n\u003cli\u003eSoong JTY, Gammall J, Liew D, Peden CJ, Bottle A, Bell D, et al. Dr Foster global frailty score: an international retrospective observational study developing and validating a risk prediction model for hospitalised older persons from administrative data sets. BMJ Open. 2019 June 1;9(6):e026759. \u003c/li\u003e\n\u003cli\u003eQuan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al. Coding Algorithms for Defining Comorbidities in ICD-9-CM and ICD-10 Administrative Data. Medical Care. 2005 Nov;43(11):1130. \u003c/li\u003e\n\u003cli\u003evvcb. comorbidipy: Python package to calculate comorbidity scores and other clinical risk scores. [Internet]. [cited 2025 Oct 23]. Available from: https://github.com/vvcb/comorbidipy\u003c/li\u003e\n\u003cli\u003eFingertips | Department of Health and Social Care [Internet]. [cited 2025 Oct 24]. Available from: https://fingertips.phe.org.uk/\u003c/li\u003e\n\u003cli\u003eNHS England. Medicines dispensed in Primary Care NHS Business Services Authority data product [Internet]. Medicines dispensed in Primary Care NHS Business Services Authority data product. [cited 2025 Oct 17]. Available from: https://digital.nhs.uk/data-and-information/data-tools-and-services/data-services/medicines-dispensed-in-primary-care-nhsbsa-data\u003c/li\u003e\n\u003cli\u003eEnglish indices of deprivation 2019 [Internet]. GOV.UK. [cited 2025 Nov 18]. Available from: https://www.gov.uk/government/statistics/english-indices-of-deprivation-2019\u003c/li\u003e\n\u003cli\u003eQuality and methodology information (QMI) for healthcare-associated infections (HCAI) reports [Internet]. GOV.UK. [cited 2025 Oct 17]. Available from: https://www.gov.uk/government/statistics/hcai-qmi-report/quality-and-methodology-information-qmi-for-healthcare-associated-infections-hcai-reports\u003c/li\u003e\n\u003cli\u003eClindamycin | Drugs | BNF content published by NICE [Internet]. [cited 2025 Oct 23]. Available from: https://bnf.nice.org.uk/drugs/clindamycin/\u003c/li\u003e\n\u003cli\u003eCusumano G, Flores GA, Venanzoni R, Angelini P. The Impact of Antibiotic Therapy on Intestinal Microbiota: Dysbiosis, Antibiotic Resistance, and Restoration Strategies. Antibiotics. 2025 Apr 3;14(4):371. \u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTables 1 and 2 are available in the Supplementary Files section\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"UK Health Security Agency","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"CDI, Clostridioides difficile, logistic regression","lastPublishedDoi":"10.21203/rs.3.rs-8165881/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8165881/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Numbers of \u003cem\u003eClostridioides difficile\u003c/em\u003e infections (CDI) in England have increased since 2021. The factors driving the increase are unknown. We aimed to understand the factors associated with CDI and what proportion of this recent increase could be explained by changes to these risk factors.\u003c/p\u003e\n\u003cp\u003eMethods: Case-control study, with CDI case data linked to hospitalisation episodes. Control cases were non-CDI related hospitalisation episodes. A logistic regression model estimated odds ratios (ORs) for individual and regional demographic factors, comorbidities, testing rates and primary care antibiotic consumption. To assess temporal changes, the model was fitted to data from financial year 2019/20 and used to predict cases in 2020/21 onwards, compared to observed data.\u003c/p\u003e\n\u003cp\u003eResults: Factors associated with testing positive for CDI were age, white ethnicity, comorbidities and primary care prescribing. The regions with the highest ORs were the North West and Yorkshire and Humber. Predicting cases using 2019/20 coefficients showed that demographic changes and antibiotic use explained 12.5% of the difference between observed cases and our baseline model in 2023/24. Including comorbidities and testing rates explained 63% of this increase.\u003c/p\u003e\n\u003cp\u003eConclusion: The strongest risk factors for CDI were age and antibiotic consumption. Increased hospitalisations, comorbidities and testing explained a large proportion of the rise in CDI cases whereas an ageing population, increases in antibiotic prescribing and demographic changes predicted smaller increases in CDI cases. However, the factors considered were not enough to fully explain the rise in cases and further investigation is needed into other factors that may be contributing.\u003c/p\u003e","manuscriptTitle":"Factors driving increases in Clostridioides difficile infection rates in England: A national case-control study, 2019 to 2024","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-26 09:11:50","doi":"10.21203/rs.3.rs-8165881/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e2cd5bcb-8ad3-4417-a2e8-6d1eef1f1cc4","owner":[],"postedDate":"November 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":58367955,"name":"Statistical Epidemiology"}],"tags":[],"updatedAt":"2025-11-26T09:11:51+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-26 09:11:50","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8165881","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8165881","identity":"rs-8165881","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}