Advertisement

Risk of complications and mortality following recurrent and non-recurrent Clostridioides difficile infection: a retrospective observational database study in England

Open AccessPublished:September 25, 2020DOI:https://doi.org/10.1016/j.jhin.2020.09.025

      Summary

      Background

      Clostridioides difficile infection (CDI) increases the risk of complications and mortality. We assessed the magnitude of these outcomes in a large cohort of English patients with initial and recurrent CDI.

      Aim

      To compare the risk of complications and all-cause mortality, within 12 months, among hospitalized patients ≥18 years old with hospital-associated- (HA-) CDI and recurrent CDI.

      Methods

      Patients with HA-CDI during 2002–2013 were identified using inpatient hospital data linked to primary care and death data. Each HA-CDI case was frequency matched to two hospitalized patients without CDI on age group, sex, calendar year of admission, admission method and number of hospital care episodes. A second CDI episode starting on days 13–56 was defined as recurrence. Risks of mortality and complications at 12 months were analysed using Cox proportional hazard models.

      Findings

      We included 6862 patients with HA-CDI and 13,724 without CDI. Median age was 81.0 years (IQR 71.0–87.0). Patients with HA-CDI had more comorbidities than those without CDI, and significantly higher risks of mortality (adjusted hazard ratio (95% confidence interval) 1.77 (1.67–1.87)) and complications (1.66 (1.46–1.88)) within 12 months from hospital admission. Of those with HA-CDI, 1140 (16.6%) experienced CDI recurrence. Patients with recurrent versus non-recurrent CDI also had significantly increased risk of mortality (1.32 (1.20–1.45)) and complications (1.37 (1.01–1.84)) in the 12 months from the initial CDI.

      Conclusions

      HA-CDI (versus no CDI) and recurrent CDI are both associated with significantly higher risks of complications or death within 12 months of the initial CDI episode.

      Keywords

      Introduction

      Clostridioides difficile is a Gram-positive, spore-forming, anaerobic bacillus that is predominantly a nosocomial pathogen [
      • Zhu D.
      • Sorg J.A.
      • Sun X.
      Clostridioides difficile biology: sporulation, germination, and corresponding therapies for C. difficile infection.
      ]. Traditional risk factors for infection include older age, hospital duration, prolonged use of antimicrobials, and use of immunosuppressive or gastric-suppressive medications [
      • Schäffler H.
      • Breitruck A.
      Clostridium difficile – from colonization to infection.
      ,
      • Cohen S.H.
      • Gerding D.N.
      • Johnson S.
      • Kelly C.P.
      • Loo V.G.
      • McDonald L.C.
      • et al.
      Clinical practice guidelines for Clostridium difficile infection in adults: 2010 update by the Society for Healthcare Epidemiology of America (SHEA) and the Infectious Diseases Society of America (IDSA).
      ,
      • Haines C.F.
      • Moore R.D.
      • Bartlett J.G.
      • Sears C.L.
      • Cosgrove S.E.
      • Carroll K.
      • et al.
      Clostridium difficile in a HIV-infected cohort: incidence, risk factors, and clinical outcomes.
      ,
      • Dial S.
      • Delaney J.A.C.
      • Barkun A.N.
      • Suissa S.
      Use of gastric acid-suppressive agents and the risk of community-acquired Clostridium difficile-associated disease.
      ,
      • Asha N.J.
      • Tompkins D.
      • Wilcox M.H.
      Comparative analysis of prevalence, risk factors, and molecular epidemiology of antibiotic-associated diarrhea due to Clostridium difficile, Clostridium perfringens, and Staphylococcus aureus.
      ,
      • Howell M.D.
      • Novack V.
      • Grgurich P.
      • Soulliard D.
      • Novack L.
      • Pencina M.
      • et al.
      Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile infection.
      ]. Other risk factors include a history of inflammatory bowel disease (IBD), cystic fibrosis and diabetes [
      • Lo Vecchio A.
      • Zacur G.M.
      Clostridium difficile infection: an update on epidemiology, risk factors, and therapeutic options.
      ].
      C. difficile infection (CDI) is a major cause of morbidity and mortality in hospitalized patients [
      • Bouza E.
      Consequences of Clostridium difficile infection: understanding the healthcare burden.
      ,
      • Wiegand P.N.
      • Nathwani D.
      • Wilcox M.H.
      • Stephens J.
      • Shelbaya A.
      • Haider S.
      Clinical and economic burden of Clostridium difficile infection in Europe: a systematic review of healthcare-facility-acquired infection.
      ]. Several studies have reported increased toxic megacolon, perforation and sepsis among patients with CDI [
      • Cohen S.H.
      • Gerding D.N.
      • Johnson S.
      • Kelly C.P.
      • Loo V.G.
      • McDonald L.C.
      • et al.
      Clinical practice guidelines for Clostridium difficile infection in adults: 2010 update by the Society for Healthcare Epidemiology of America (SHEA) and the Infectious Diseases Society of America (IDSA).
      ,
      • Bartlett J.G.
      • Gerding D.N.
      Clinical recognition and diagnosis of Clostridium difficile infection.
      ,
      • Chakra C.N.A.
      • Pepin J.
      • Sirard S.
      • Valiquette L.
      Risk factors for recurrence, complications and mortality in Clostridium difficile infection: a systematic review.
      ,
      • McDonald L.C.
      • Gerding D.N.
      • Johnson S.
      • Bakken J.S.
      • Carroll K.C.
      • Coffin S.E.
      • et al.
      Clinical practice guidelines for Clostridium difficile infection in adults and children: 2017 update by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA).
      ]. Significantly higher risks of all-cause mortality have also been observed in the USA [
      • Olsen M.A.
      • Stwalley D.
      • Demont C.
      • Dubberke E.R.
      Clostridium difficile infection increases acute and chronic morbidity and mortality.
      ] and Europe [
      • Bauer M.P.
      • Notermans D.W.
      • van Benthem B.H.
      • Brazier J.S.
      • Wilcox M.H.
      • Rupnik D.L.
      • et al.
      Clostridium difficile infection in Europe: a hospital-based survey.
      ,
      • Barbut F.
      • Bouée S.
      • Longepierre L.
      • Goldberg M.
      • Bensoussan C.
      • Levy-Bachelot L.
      Excess mortality between 2007 and 2014 among patients with Clostridium difficile infection: a French health insurance database analysis.
      ]. In the UK, one study using surveillance data from Scotland estimated an almost three-fold increase in 30-day mortality [
      • Banks A.
      • Moore E.K.
      • Bishop J.
      • Coia J.E.
      • Brown D.
      • Mather H.
      • et al.
      Trends in mortality following Clostridium difficile infection in Scotland, 2010-2016: a retrospective cohort and case-control study.
      ], and another reported a 50% increased risk of death up to 5–8 years after hospital admission among patients with CDI at an English teaching hospital [
      • Reacher M.
      • Verlander N.Q.
      • Roddick I.
      • Trundle C.
      • Brown N.
      • Farrington M.
      • et al.
      Excess mortality attributable to Clostridium difficile and risk factors for infection in an historic cohort of hospitalised patients followed up in the United Kingdom death register.
      ]. The incidence of CDI recurrence ranges from 11% to 33% and recurrence is also associated with higher patient mortality and complication rates [
      • Kuntz J.L.
      • Baker J.M.
      • Kipnis P.
      • Li S.X.
      • Liu V.
      • Xie Y.
      • et al.
      Utilization of health services among adults with recurrent Clostridium difficile infection: a 12-year population-based study.
      ,
      • Shah D.N.
      • Aitken S.L.
      • Barragan L.F.
      • Bozorgui S.
      • Goddu S.
      • Navarro M.E.
      • et al.
      Economic burden of primary compared with recurrent Clostridium difficile infection in hospitalized patients: a prospective cohort study.
      ,
      • Olsen M.A.
      • Yan Y.
      • Reske K.A.
      • Zilberberg M.D.
      • Dubberke E.R.
      Recurrent Clostridium difficile infection is associated with increased mortality.
      ], although lower mortality rates have been reported among patients with recurrent CDI compared with first episodes [
      • Reveles K.R.
      • Lawson K.A.
      • Mortensen E.M.
      • Pugh M.J.V.
      • Koeller J.M.
      • Argamany J.R.
      • et al.
      National epidemiology of initial and recurrent Clostridium difficile infection in the Veterans Health Administration from 2003 to 2014.
      ].
      Although several studies have examined complications and mortality rates among patients with CDI, they were sometimes limited by small sample sizes [
      • Barbut F.
      • Bouée S.
      • Longepierre L.
      • Goldberg M.
      • Bensoussan C.
      • Levy-Bachelot L.
      Excess mortality between 2007 and 2014 among patients with Clostridium difficile infection: a French health insurance database analysis.
      ,
      • Reacher M.
      • Verlander N.Q.
      • Roddick I.
      • Trundle C.
      • Brown N.
      • Farrington M.
      • et al.
      Excess mortality attributable to Clostridium difficile and risk factors for infection in an historic cohort of hospitalised patients followed up in the United Kingdom death register.
      ], focused on either in-hospital [
      • Khanna S.
      • Gupta A.
      • Baddour L.M.
      • Pardi D.S.
      Epidemiology, outcomes, and predictors of mortality in hospitalized adults with Clostridium difficile infection.
      ,
      • Esteban-Vasallo M.D.
      • de Miguel-Díez J.
      • López-de-Andrés A.
      • Hernández-Barrera V.
      • Jiménez-García R.
      Clostridium difficile-related hospitalizations and risk factors for in-hospital mortality in Spain between 2001 and 2015.
      ] or post-discharge mortality [
      • Olsen M.A.
      • Yan Y.
      • Reske K.A.
      • Zilberberg M.D.
      • Dubberke E.R.
      Recurrent Clostridium difficile infection is associated with increased mortality.
      ], or did not include clinical information from outside the hospital setting [
      • Barbut F.
      • Bouée S.
      • Longepierre L.
      • Goldberg M.
      • Bensoussan C.
      • Levy-Bachelot L.
      Excess mortality between 2007 and 2014 among patients with Clostridium difficile infection: a French health insurance database analysis.
      ,
      • Banks A.
      • Moore E.K.
      • Bishop J.
      • Coia J.E.
      • Brown D.
      • Mather H.
      • et al.
      Trends in mortality following Clostridium difficile infection in Scotland, 2010-2016: a retrospective cohort and case-control study.
      ,
      • Reacher M.
      • Verlander N.Q.
      • Roddick I.
      • Trundle C.
      • Brown N.
      • Farrington M.
      • et al.
      Excess mortality attributable to Clostridium difficile and risk factors for infection in an historic cohort of hospitalised patients followed up in the United Kingdom death register.
      ,
      • Sammons J.S.
      • Localio R.
      • Xiao R.
      • Coffin S.E.
      • Zaoutis T.
      Clostridium difficile infection is associated with increased risk of death and prolonged hospitalization in children.
      ]. Some studies were also restricted to populations from specialist centres or other clinical settings that may not be representative of wider groups of patients with CDI [
      • Shah D.N.
      • Aitken S.L.
      • Barragan L.F.
      • Bozorgui S.
      • Goddu S.
      • Navarro M.E.
      • et al.
      Economic burden of primary compared with recurrent Clostridium difficile infection in hospitalized patients: a prospective cohort study.
      ,
      • Olsen M.A.
      • Yan Y.
      • Reske K.A.
      • Zilberberg M.D.
      • Dubberke E.R.
      Recurrent Clostridium difficile infection is associated with increased mortality.
      ].
      To overcome these issues, we conducted a frequency-matched, longitudinal, retrospective cohort study using data from hospitalized patients linked to a primary care database and a national death registry, to assess the risk of mortality and complications among patients with hospital-associated (HA)-CDI versus no CDI, between 1st January 2002 and 31st December 2013. We separately assessed the risk of all-cause mortality and complications in patients with recurrent CDI versus patients with non-recurrent CDI.

      Methods

      Data sources

      Data were extracted from the UK Clinical Practice Research Datalink General Practitioner Online Data (CPRD GOLD), a longitudinal primary care electronic database of anonymized patient records, collected from a sample of general practices in the UK. Analyses were restricted to 380 practices in England with individual record linkage to Hospital Episode Statistics (HES) Admitted Patient Care (APC) data, which comprise admission and discharge dates plus information on clinical diagnoses and procedures collected from English hospitals []; the Office for National Statistics (ONS) database, which holds data on the date and cause of death [

      UK Statistics Authority. Office for national Statistics n.d. https://www.ons.gov.uk/(last accessed March 2020).

      ]; and/or the Index of Multiple Deprivation (IMD), which combines a number of economic and social indicators into a single deprivation score based on area of residence []. In the case of the IMD, which ranks areas relative to one another from 1 to 32, 844, the raw scores were categorized into quintiles, with 1 being the least deprived and 5 the most deprived. Analyses of primary outcomes were restricted to patients in CPRD GOLD eligible for linkage to both HES APC and ONS death data. For secondary outcomes where death was not evaluated, patients were required to be eligible for linkage to HES and IMD data only.

      Study population

      The risk of mortality and complications among hospitalized patients with a first episode of HA-CDI was compared with frequency-matched (1:2) hospitalized patients without CDI. Recurrence was evaluated among the sub-cohort of patients with a first episode of HA-CDI. Only patients with ≥6 months of follow up during the study period were eligible for inclusion.

      CDI case selection

      Cases were patients with a diagnosis of HA-CDI recorded between 1st January 2002 and 31st December 2013. Eligible patients were aged ≥18 years at index hospital admission with ≥1 diagnosis of HA-CDI documented during their first or second episode of hospital care (periods during which a patient was under the care of a particular consultant []). Pregnancy-related admissions, between-hospital transfers and patients with a first CDI occurrence recorded in hospital care episode ≥3 were excluded.
      Cases were ascertained based on ICD-10 (International Classification of Diseases 10th revision) A04.7 ‘Enterocolitis due to Clostridium difficile’ or related codes, recorded in HES APC data. HA-CDI was defined using an algorithm developed from previous work by Jen et al. [
      • Jen M.-H.
      • Saxena S.
      • Bottle A.
      • Pollok R.
      • Holmes A.
      • Aylin P.
      Assessment of administrative data for evaluating the shifting acquisition of Clostridium difficile infection in England.
      ], and was specified as a primary diagnosis of CDI (CDI recorded in the first hospital care episode and in the first diagnosis order) and a history of hospitalization within the preceding 4 weeks, or CDI recorded as a secondary diagnosis during the first or second episode of a hospital stay of >2 days (Supplementary Figure S1). As the exact dates of clinical onset are not recorded in HES, the index date of the CDI episode was defined as the start of the care episode in which the CDI was recorded.

      Control group selection

      Cases were frequency matched to two hospitalized patients without CDI (unexposed patients) through random selection. Potential controls with CDI or diarrhoeal illness in HES or primary care up to 90 days prior to their index admission were excluded. Unexposed patients were selected to reflect the distribution of the case pool on the following parameters: calendar year of diagnosis, sex, age group, index admission method (emergency or non-emergency) and the number of hospital care episodes (one or more). The matched hospital care episode for each unexposed patient was randomly sampled to reflect the distribution of episodes in which CDI was recorded. The index date was defined as the start of the randomly selected care episode.
      Recurrence of CDI was defined as a second CDI episode documented on days 13–56 inclusive, after the index episode. The date of CDI recurrence was defined as the start of the care episode in which the diagnosis was made. Otherwise, patients with a single record of CDI documented during the study period, with all CDI events recorded within 12 days of their first CDI episode or with their second CDI event recorded after day 56 were included in the non-recurrent CDI cohort. Follow-up time for these patients was censored at the date of their second CDI diagnosis.

      Outcomes

      The primary outcome was time to all-cause mortality, up to 12 months, as evidenced by a record of death in the ONS mortality record. Only mortality events recorded during the patient's registration period at the practice were included. Secondary outcomes were the time to first complication and time to the combined endpoint of complications or death, up to 12 months. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis, and were identified using ICD-10 codes in HES or ONS mortality data, Read Codes in primary care, and/or Office of Population Censuses and Surveys procedure codes in hospital (Supplementary Table S1). Admission to augmented or critical care for any reason during the follow-up period was also classed as a complication. Events occurring prior to the index date were excluded.

      Study size

      Assuming a cumulative risk of 21.5% for 1-year mortality among patients without diarrhoea and unexposed to CDI [
      • Hensgens M.P.M.
      • Goorhuis A.
      • Dekkers O.M.
      • Van Benthem B.H.B.
      • Kuijper E.J.
      All-cause and disease-specific mortality in hospitalized patients with Clostridium difficile infection: a multicenter cohort study.
      ], 1571 patients with and without CDI would provide 90% power to detect a statistically significant difference between groups at a two-sided significance level of 0.05, if the real difference in 1-year mortality was ≥5%. Using a 1-year mortality rate of 15.3% among patients with non-recurrent CDI [
      • Escobar G.J.
      • Baker J.M.
      • Li S.X.
      • Xie Y.
      • Kipnis P.
      Clinical and economic burden of recurrent Clostridium difficile infections – a 10-year retrospective large database analysis.
      ], 1228 patients with and without recurrent CDI would provide 90% power to detect a statistically significant difference between groups at a two-sided significance level of 0.05, if the real difference in mortality was ≥5%. We elected to use a ratio of 1:2 of cases to controls for any incremental power that this might add.

      Statistical analysis

      Baseline characteristics were summarized using Chi-squared tests, t-tests and Mann–Whitney U-tests, where appropriate. Kaplan–Meier analysis was initially used to estimate the cumulative incidence function of time to mortality or complications up to 12 months. Cox proportional hazard regression models, with backward stepwise selection of candidate variables, were fitted to compare hazards of endpoints during the outcome periods. To assess the impact of immortal time bias [
      • Shintani A.K.
      • Girard T.D.
      • Eden S.K.
      • Arbogast P.G.
      • Moons K.G.M.
      • Ely E.W.
      Immortal time bias in critical care research: application of time-varying Cox regression for observational cohort studies.
      ], Cox regression models were alternatively specified with time-fixed or time-dependent covariates. For the comparison of HA-CDI versus no CDI, two models were used: CDI was specified as either a time-dependent or time-fixed covariate, the index date of CDI was set as the start of the CDI episode, and follow up was from either hospital admission or from the start of the CDI episode. The comparison of recurrent versus non-recurrent CDI used four different models, with CDI specified as either a time-dependent or time-fixed covariate; the index date of recurrent CDI specified as either the start of the recurrent episode or final CDI status; and follow up from either the episode start of the index CDI, day 13 (the start of the recurrence risk window) or day 57 (the end of the recurrence risk window). Tests for proportional hazards were undertaken visually using log–log plots and statistically by assessing Schoenfeld's residuals. Hazard ratios (HRs) with 95% confidence intervals (CIs) were reported and two-sided P-values were calculated using the Wald test.
      Adjustment factors measured at cohort entry using primary care and/or HES data included age in years, sex, body mass index, smoking history, alcohol use, history of comorbidities, drug therapies and socioeconomic deprivation category (Supplementary data). To further assess the robustness of our findings, we conducted sensitivity analyses taking the midpoint of the hospital care episode as the CDI index date (Supplementary Tables S2 and S3). The end of follow up was defined as the earliest of the following: practice last collection date, patient deregistration date, date of complication diagnosis or death, end of HES data collection, end of ONS mortality data collection, or study end.
      All analyses were performed using STATA version 14.

      Ethics

      CPRD has ethical approval for the collection and use of pseudonymized primary care records and linked data from the Health Research Authority Research Ethics Committee (reference number: 05/MRE04/87). The study was approved by the Independent Scientific Advisory Committee of the Medicines and Healthcare products Regulatory Agency, under protocol number 14_187A3.

      Results

      Patient populations

      Between 1st January 2002 and 31st December 2013, 5,552,368 patients registered in CPRD GOLD were eligible for linkage to HES and ONS data (Figure 1). Following the implementation of inclusion and exclusion criteria, 6862 patients met our definition of HA-CDI (Figure 1) and were matched to 13,724 hospitalized patients without CDI.
      Figure 1
      Figure 1Selection of cases with hospital-associated- (HA-) Clostridioides difficile infection (CDI) recorded in the first or second episode of hospital care. a Patients are considered acceptable by the Clinical Practice Research Datalink (CPRD) based on 15-point criteria that excludes patients with non-continuous follow up, or patients with poor data recording that raises suspicion as to the validity of those patients' records. GOLD, General Practitioner Online Data; HES, Hospital Episode Statistics; UTS, up-to-standard. Grey boxes indicate patients excluded from the dataset.

      Demographic and clinical characteristics

      Patients with HA-CDI versus no CDI

      The median age (interquartile range) of all participants was 81.0 (71.0–87.0) years; females comprised 59.0% (12,141/20,586) of patients. Compared with patients with no CDI, patients with HA-CDI had higher healthcare service use in the 12 months prior to CDI diagnosis, more recorded comorbidities (including a history of IBD and diabetes mellitus), and more antibiotics prescribed in the 90 days prior to admission (Table I). Higher prior use of proton-pump inhibitors and more admissions to augmented or intensive care were also observed (Table I).
      Table IBaseline characteristics of patients with hospital-associated- (HA-) Clostridioides difficile infection (CDI) and no CDI, and patients with recurrent and non-recurrent CDI
      HA-CDI and no CDI cohort
      Characteristics were assessed for the specified duration prior to index hospital admission.
      Recurrent CDI and non-recurrent CDI cohort
      Characteristics were assessed for the specified period prior to the index HA-CDI episode.
      HA-CDI (N=6862)No CDI (N=13,724)Total (N=20,586)PRecurrent CDI (N=1140)Non-recurrent CDI (N=5722)Total (N=6862)P
      Baseline characteristics
      Age in years, median (IQR)81.0 (72.0–87.0)81.0 (71.0–87.0)81.0 (71.0–87.0)0.09281.0 (72.0–87.0)81.0 (72.0–87.0)81.0 (72.0–87.0)0.688
      Female4047 (59.0)8094 (59.0)12,141 (59.0)1.000
      Patients with no CDI were matched to those with HA-CDI for this parameter.
      678 (59.5)3369 (58.9)4047 (59.0)0.709
      BMI
       Median (IQR)25.4 (22.2–29.1)25.4 (22.5–28.9)25.4 (22.4–28.9)0.38125.5 (22.1–29.1)25.4 (22.2–29.2)25.4 (22.2–29.1)0.901
       Unknown1382 (20.1)2761 (20.1)4143 (20.1)222 (19.5)1160 (20.3)1382 (20.1)
      Smoking status<0.0010.513
       Non-smoker2659 (38.7)5772 (42.1)8431 (41.0)420 (36.8)2239 (39.1)2659 (38.7)
       Ex-smoker2679 (39.0)4917 (35.8)7596 (36.9)468 (41.1)2211 (38.6)2679 (39.0)
       Smoker1136 (16.6)2173 (15.8)3309 (16.1)205 (18.0)931 (16.3)1136 (16.6)
       Unknown388 (5.7)862 (6.3)1250 (6.1)222 (19.5)1160 (20.3)1382 (20.1)
      Drinking status<0.0010.513
       Non-drinker1010 (14.7)1852 (13.5)2862 (13.9)163 (14.3)847 (14.8)1010 (14.7)
       Ex-drinker769 (11.2)1321 (9.6)2090 (10.2)122 (10.7)647 (11.3)769 (11.2)
       Drinker3867 (56.4)8005 (58.3)11 872 (57.7)667 (58.5)3200 (55.9)3867 (56.4)
       Unknown1216 (17.7)2546 (18.6)3762 (18.3)188 (16.5)1028 (18.0)1216 (17.7)
      Socioeconomic deprivation level
      Grouping based on Index of Multiple Deprivation scores, with 1 being the least deprived and 5 the most deprived.
      <0.0010.704
       11200 (17.5)2787 (20.3)3987 (19.4)210 (18.4)990 (17.3)1200 (17.5)
       21590 (23.2)3284 (23.9)4874 (23.7)262 (23.0)1328 (23.2)1590 (23.2)
       31396 (20.3)2892 (21.1)4288 (20.8)215 (18.9)1181 (20.6)1396 (20.3)
       41416 (20.6)2586 (18.8)4002 (19.4)237 (20.8)1179 (20.6)1416 (20.6)
       51249 (18.2)2151 (15.7)3400 (16.5)215 (18.9)1034 (18.1)1249 (18.2)
       Unknown11 (0.2)24 (0.2)35 (0.2)1 (<0.1)10 (0.2)11 (0.2)
      Current or prior comorbidities
      Diabetes mellitus
      At any time previously.
      1369 (20.0)1979 (14.4)3348 (16.3)<0.001247 (21.7)1130 (19.7)1377 (20.1)0.140
      Hypertension
      At any time previously.
      5036 (73.4)9172 (66.8)14 208 (69.0)<0.001864 (75.8)4172 (72.9)5036 (73.4)0.045
      Inflammatory bowel disease
      At any time previously.
      214 (3.1)81 (0.6)295 (1.4)<0.00144 (3.9)176 (3.1)220 (3.2)0.170
      Renal disease
      At any time previously.
      2386 (34.8)3163 (23.0)5549 (27.0)<0.001440 (38.6)1993 (34.8)2433 (35.5)0.015
      Cerebrovascular accident
      At any time previously.
      1910 (27.8)3249 (23.7)5159 (25.1)<0.001349 (30.6)1625 (28.4)1974 (28.8)0.131
      Moderate-to-severe liver disease
      At any time previously.
      158 (2.3)88 (0.6)246 (1.2)<0.00140 (3.5)132 (2.3)172 (2.5)0.018
      Peripheral vascular disease
      At any time previously.
      1076 (15.7)1418 (10.3)2494 (12.1)<0.001191 (16.8)899 (15.7)1090 (15.9)0.379
      Dementia
      At any time previously.
      850 (12.4)1694 (12.3)2544 (12.4)0.928125 (11.0)758 (13.2)883 (12.9)0.036
      Malignant cancer ≤5 years previously1485 (21.6)1781 (13.0)3266 (15.9)<0.001213 (18.7)1299 (22.7)1512 (22.0)0.003
      Transplantation ≤24 months previously87 (1.3)15 (0.1)102 (0.5)<0.00114 (1.2)75 (1.3)89 (1.3)0.822
      Medication prescriptions ≤90 days previously
      Any antibiotic2749 (40.1)3961 (28.9)6710 (32.6)<0.001478 (41.9)2254 (39.4)2732 (39.8)0.110
      Lipid-lowering drugs1904 (27.7)3492 (25.4)5396 (26.2)<0.001344 (30.2)1547 (27.0)1891 (27.6)0.030
      Immunosuppressants excluding hydrocortisone and prednisolone59 (0.9)27 (0.2)86 (0.4)<0.0019 (0.8)50 (0.9)59 (0.9)0.778
      Hydrocortisone and/or prednisolone788 (11.5)1023 (7.5)1811 (8.8)<0.001137 (12.0)649 (11.3)786 (11.5)0.513
      Proton-pump inhibitors2285 (33.3)2806 (20.4)5091 (24.7)<0.001382 (33.5)1888 (33.0)2270 (33.1)0.737
      Glucocorticoids804 (11.7)1041 (7.6)1845 (9.0)<0.001139 (12.2)663 (11.6)802 (11.7)0.561
      Health service use ≤12 months previously
      GP consultations, mean (SD)17.2 (15.4)12.3 (11.7)13.9 (13.2)<0.00117.8 (15.8)17.1 (15.4)17.3 (15.4)0.537
      ≥2 prior hospitalizations2926 (42.6)2124 (15.5)5050 (24.5)<0.001574 (50.4%)2809 (49.1%)3383 (49.3%)0.032
      Index hospitalization
      Duration of hospital stay, median (IQR)28.0 (15.0–50.0)6.0 (1.0–15.0)11.0 (3.0–28.0)<0.00139.0 (24.0–62.0)26.0 (14.0–47.0)28.0 (15.0–50.0)<0.001
      Emergency admission5832 (85.0)11,664 (85.0)17,496 (85.0)1.000
      Patients with no CDI were matched to those with HA-CDI for this parameter.
      997 (87.5)4835 (84.5)5832 (85.0)0.011
      Augmented/intensive care
      Assessed from admission to index date.
      234 (3.4)119 (0.9)353 (1.7)<0.00144 (3.9)229 (4.0)273 (4.0)0.822
      BMI, body mass index; GP, general practitioners; IQR, interquartile range; SD, standard deviation. Data are N (%) unless otherwise given.
      a Characteristics were assessed for the specified duration prior to index hospital admission.
      b Characteristics were assessed for the specified period prior to the index HA-CDI episode.
      c Patients with no CDI were matched to those with HA-CDI for this parameter.
      d Grouping based on Index of Multiple Deprivation scores, with 1 being the least deprived and 5 the most deprived.
      e At any time previously.
      f Assessed from admission to index date.

      Patients with recurrent CDI versus non-recurrent CDI

      Within the HA-CDI sub-cohort, 16.6% (1140/6862) of patients had a recurrent CDI episode during days 13–56. Notably, patients with recurrent CDI had a longer index hospital stay and were more likely to have had an emergency index hospital admission (Table I). A history of hypertension, renal disease, moderate/severe liver disease and use of lipid-lowering therapy were also more commonly documented among those with recurrent CDI (Table I). A history of malignant cancer (up to 5 years prior to the index CDI diagnosis) and dementia were more prevalent in the non-recurrent CDI group (Table I).

      Twelve-month risk of death or complications

      Patients with HA-CDI versus no CDI

      There were 3427/6862 (49.9%) and 4146/13,724 (30.2%) deaths among patients with and without CDI, respectively, at 12 months from the initial hospital care episode (P<0.001) (Table II). In the time-dependent and time-fixed analyses, the adjusted HRs for death among patients with HA-CDI versus no CDI were 1.77 (95% CI: 1.67, 1.87) and 1.55 (95% CI: 1.47, 1.64), respectively (Table III).
      Table IIIncidence of death or complications in patients with Clostridioides difficile infection (CDI)
      HA-CDI versus no CDIRecurrent CDI versus non-recurrent CDI
      OutcomeTime to outcomeHA-CDI (N=6862)No CDI (N=13,724)PRecurrent CDI (N=1,140)Non-recurrent CDI (N=5,722)P
      Death
      Ascertained using death registration recorded in Office of National Statistics mortality data.
      , N (%)
      ≤30 days1117 (16.3)2041 (14.9)0.00887 (7.6)1102 (19.3)<0.001
      ≤90 days2535 (36.9)3094 (22.5)<0.001381 (33.4)2164 (37.8)0.005
      ≤180 days3076 (44.8)3613 (26.3)<0.001492 (43.2)2494 (43.6)0.790
      ≤365 days3427 (49.9)4146 (30.2)<0.001564 (49.5)2737 (47.8)0.311
      Complications
      Evaluated using primary care (CPRD GOLD), hospital (HES) and Office of National Statistics mortality data. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis (Supplementary Table S1).
      , N (%)
      ≤30 days1227 (17.9)1009 (7.4)<0.001183 (16.1)868 (15.2)0.450
      ≤90 days1353 (19.7)1096 (8.0)<0.001226 (19.8)940 (16.4)0.005
      ≤180 days1436 (20.9)1164 (8.5)<0.001241 (21.1)996 (17.4)0.003
      ≤365 days1486 (21.7)1240 (9.0)<0.001249 (21.8)1036 (18.1)0.003
      Complications
      Evaluated using primary care (CPRD GOLD), hospital (HES) and Office of National Statistics mortality data. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis (Supplementary Table S1).
      or death
      Ascertained using death registration recorded in Office of National Statistics mortality data.
      , N (%)
      ≤30 days2032 (29.6)2639 (19.2)<0.001247 (21.7)1690 (29.5)<0.001
      ≤90 days3251 (47.4)3620 (26.4)<0.001520 (45.6)2628 (45.9)0.846
      ≤180 days3742 (54.5)4137 (30.1)<0.001615 (53.9)2943 (51.4)0.121
      ≤365 days4055 (59.1)4668 (34.0)<0.001682 (59.8)3173 (55.5)0.007
      Follow up was from the start of the initial hospital care episode. CDI, Clostridioides difficile infection; CPRD, Clinical Practice Research Datalink; GOLD, General Practitioner Online Data; HA, hospital-associated; HES, Hospital Episode Statistics.
      a Ascertained using death registration recorded in Office of National Statistics mortality data.
      b Evaluated using primary care (CPRD GOLD), hospital (HES) and Office of National Statistics mortality data. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis (Supplementary Table S1).
      Table IIITwelve-month risk of death or complications for patients with hospital-associated- (HA-) Clostridioides difficile infection (CDI) versus no CDI and for recurrent CDI versus non-recurrent CDI
      Death
      Ascertained using death registration recorded in Office of National Statistics mortality data.
      Complications
      Evaluated using primary care (CPRD GOLD), hospital (HES) and Office of National Statistics mortality data. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis (Supplementary Table S1).
      Complications
      Evaluated using primary care (CPRD GOLD), hospital (HES) and Office of National Statistics mortality data. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis (Supplementary Table S1).
      or death
      Ascertained using death registration recorded in Office of National Statistics mortality data.
      ModelCDI covariate specificationIndex date of CDIStart of follow upUnadjusted HR (95% CI)Adjusted HR (95% CI)Unadjusted HR (95% CI)Adjusted HR (95% CI)Unadjusted HR (95% CI)Adjusted HR (95% CI)
      HA-CDI versus no CDI
      1Time-dependentStart of CDI episodeHospital admission1.96 (1.87–2.06)∗1.77 (1.67–1.87)∗1.80 (1.58–2.03)∗1.66 (1.46–1.88)∗1.99 (1.89–2.08)∗1.75 (1.65–1.86)∗
      2Time-fixedStart of CDI episodeStart of CDI/selected hospital care episode1.77 (1.69–1.85)∗1.55 (1.47–1.64)∗3.27 (3.00–3.56)∗3.36 (3.06–3.68)∗2.05 (1.96–2.14)∗1.88 (1.78–1.98)∗
      Recurrent CDI versus non-recurrent CDI
      3Time-dependentStart of recurrent episodeEpisode start of the index CDI1.37 (1.25–1.50)∗1.32 (1.20–1.45)∗1.37 (1.02–1.85)∗1.37 (1.01–1.84)∗1.40 (1.26–1.54)∗1.36 (1.22–1.50)∗
      4Time-dependentStart of recurrent episodeDay 13 (start of recurrence risk window)
      As evaluated from the date of index CDI.
      1.37 (1.25–1.50)∗1.34 (1.22–1.47)∗1.37 (1.02–1.85)∗1.35 (1.00–1.83)∗1.40 (1.26–1.54)∗1.36 (1.23–1.50)∗
      5Time-fixedFinal CDI status
      Defined as ever having a diagnosis of CDI (yes or no) during the study period.
      Day 57 (end of the recurrence risk window)
      As evaluated from the date of index CDI.
      1.45 (1.27–1.64)∗1.40 (1.23–1.59)∗1.39 (0.95–2.02)1.34 (0.91–1.95)1.50 (1.31–1.71)∗1.46 (1.27–1.67)∗
      6Time-fixedFinal CDI status
      Defined as ever having a diagnosis of CDI (yes or no) during the study period.
      Episode start of the index CDI0.91 (0.83–1.00)∗0.88 (0.80–0.96)∗1.18 (1.03–1.36)∗1.17 (1.02–1.34)∗1.01 (0.93–1.10)0.99 (0.91–1.08)
      CDI, Clostridioides difficile infection; CPRD, Clinical Practice Research Datalink; GOLD, General Practitioner Online Data; HES, Hospital Episode Statistics; HR, hazard ratio. Recurrent CDI was defined as a second CDI episode diagnosed on days 13–56.
      P<0.05 for comparison.
      a Ascertained using death registration recorded in Office of National Statistics mortality data.
      b Evaluated using primary care (CPRD GOLD), hospital (HES) and Office of National Statistics mortality data. Complications included ulcerative pancolitis, megacolon, intestinal perforation, toxic gastroenteritis/colitis, colectomy, renal failure and sepsis (Supplementary Table S1).
      c As evaluated from the date of index CDI.
      d Defined as ever having a diagnosis of CDI (yes or no) during the study period.
      Similarly, significantly more patients with HA-CDI experienced complications up to 12 months following index hospital admission, compared with patients with no CDI (P<0.001) (Table II). The adjusted HR from time-dependent analysis for complications in the group of patients with HA-CDI versus no CDI was 1.66 (95% CI: 1.46, 1.88) (Table III).

      Patients with recurrent versus non-recurrent CDI

      There were 564 (49.5%) and 2737 (47.8%) deaths among patients with recurrent and non-recurrent CDI, respectively, at 12 months from the initial hospital care episode (Table II). When evaluating the risk of death from the start of the initial CDI episode, the incidence of mortality at ≤30 days and ≤90 days was significantly lower in patients with recurrent CDI and the difference in mortality at ≤365 days was not significant (Table II). Adjusted time-fixed analyses showed a seemingly implausible protective effect of recurrence on 12-month mortality (HR: 0.88, 95% CI: 0.80, 0.96) (Table III). When immortal time was correctly classified (Figure 2), 12-month survival was higher in those with non-recurrent CDI versus recurrent CDI (HR: 1.32, 95% CI: 1.20, 1.45) (Table III).
      Figure 2
      Figure 2Twelve-month mortality, assessed from the start of the Clostridioides difficile infection (CDI) recurrence window, among patients with recurrent and non-recurrent CDI. Recurrent CDI was defined as a second CDI episode documented on days 13–56 inclusive. Time to death was assessed from the start of the CDI recurrence window at day 13. The date of recurrence was defined as the start of the hospital episode in which the recurrent CDI episode was identified.
      The incidence of complications (excluding death) up to 90–365 days from the start of the initial care episode was significantly higher (P≤0.005) in patients with recurrent CDI (Table II). The adjusted HR for complications excluding death, up to 12 months from the start of the initial care episode, was 1.37 (95% CI: 1.01, 1.84) (Table III).
      Results of further sensitivity analyses, with outcomes assessed from the midpoint of the initial hospital care episode, are included in the Supplementary data).

      Discussion

      We conducted a retrospective study of patients hospitalized in England with a first episode of CDI and observed that the risk of 12-month mortality and complications was significantly higher compared with matched hospitalized patients with no evidence of CDI, a finding supported by previous research [
      • Olsen M.A.
      • Stwalley D.
      • Demont C.
      • Dubberke E.R.
      Clostridium difficile infection increases acute and chronic morbidity and mortality.
      ,
      • Banks A.
      • Moore E.K.
      • Bishop J.
      • Coia J.E.
      • Brown D.
      • Mather H.
      • et al.
      Trends in mortality following Clostridium difficile infection in Scotland, 2010-2016: a retrospective cohort and case-control study.
      ,
      • Khanna S.
      • Gupta A.
      • Baddour L.M.
      • Pardi D.S.
      Epidemiology, outcomes, and predictors of mortality in hospitalized adults with Clostridium difficile infection.
      ,
      • Delgado A.
      • Reveles I.A.
      • Cabello F.T.
      • Reveles K.R.
      Poorer outcomes among cancer patients diagnosed with Clostridium difficile infections in United States community hospitals.
      ,
      • Forster A.J.
      • Taljaard M.
      • Oake N.
      • Wilson K.
      • Roth V.
      • van Walraven C.
      The effect of hospital-acquired infection with Clostridium difficile on length of stay in hospital.
      ,
      • Yasunaga H.
      • Horiguchi H.
      • Hashimoto H.
      • Matsuda S.
      • Fushimi K.
      The burden of Clostridium difficile-associated disease following digestive tract surgery in Japan.
      ]. Interestingly, in a study of CDI in Scotland [
      • Banks A.
      • Moore E.K.
      • Bishop J.
      • Coia J.E.
      • Brown D.
      • Mather H.
      • et al.
      Trends in mortality following Clostridium difficile infection in Scotland, 2010-2016: a retrospective cohort and case-control study.
      ], the 30-day all-cause mortality rate of 17.5% was similar to the rate we observed in England (16.3%), despite the use of differing time periods and database methodology. Additionally, mortality and complications were increased significantly in patients with recurrent CDI compared with non-recurrent CDI, findings which align with earlier research in the USA [
      • Kuntz J.L.
      • Baker J.M.
      • Kipnis P.
      • Li S.X.
      • Liu V.
      • Xie Y.
      • et al.
      Utilization of health services among adults with recurrent Clostridium difficile infection: a 12-year population-based study.
      ,
      • Olsen M.A.
      • Yan Y.
      • Reske K.A.
      • Zilberberg M.D.
      • Dubberke E.R.
      Recurrent Clostridium difficile infection is associated with increased mortality.
      ]. It is, however, important to note that our study population was older and had more comorbidities than some populations studied in North America [
      • Kuntz J.L.
      • Baker J.M.
      • Kipnis P.
      • Li S.X.
      • Liu V.
      • Xie Y.
      • et al.
      Utilization of health services among adults with recurrent Clostridium difficile infection: a 12-year population-based study.
      ,
      • Shah D.N.
      • Aitken S.L.
      • Barragan L.F.
      • Bozorgui S.
      • Goddu S.
      • Navarro M.E.
      • et al.
      Economic burden of primary compared with recurrent Clostridium difficile infection in hospitalized patients: a prospective cohort study.
      ,
      • Forster A.J.
      • Taljaard M.
      • Oake N.
      • Wilson K.
      • Roth V.
      • van Walraven C.
      The effect of hospital-acquired infection with Clostridium difficile on length of stay in hospital.
      ] and Europe [
      • Hensgens M.P.M.
      • Goorhuis A.
      • Dekkers O.M.
      • Van Benthem B.H.B.
      • Kuijper E.J.
      All-cause and disease-specific mortality in hospitalized patients with Clostridium difficile infection: a multicenter cohort study.
      ]. This age difference is similar to findings from studies of fidaxomicin for the treatment of CDI, in which the median age in the UK was 70–81 years [
      • Goldenberg S.D.
      • Brown S.
      • Edwards L.
      • Gnanarajah D.
      • Howard P.
      • Jenkins D.
      • et al.
      The impact of the introduction of fidaxomicin on the management of Clostridium difficile infection in seven NHS secondary care hospitals in England: a series of local service evaluations.
      ] whilst in North America and Europe, the mean age was 61–63 years [
      • Louie T.J.
      • Miller M.A.
      • Mullane K.M.
      • Weiss K.
      • Lentnek A.
      • Golan Y.
      • et al.
      Fidaxomicin versus vancomycin for Clostridium difficile infection.
      ,
      • Cornely O.A.
      • Crook D.W.
      • Esposito R.
      • Poirier A.
      • Somero M.S.
      • Weiss K.
      • et al.
      Fidaxomicin versus vancomycin for infection with Clostridium difficile in Europe, Canada, and the USA: a double-blind, non-inferiority, randomised controlled trial.
      ].
      We also found a higher frequency of risk factors recorded among CDI cases: more comorbidities, more patients with prior prescriptions of antibiotics and/or proton-pump inhibitors, a greater incidence of hospitalization within the previous 12 months, and significantly higher use of augmented or intensive care; these results are also consistent with previous studies [
      • Stevens V.
      • Dumyati G.
      • Fine L.S.
      • Fisher S.G.
      • van Wijngaarden E.
      Cumulative antibiotic exposures over time and the risk of Clostridium difficile infection.
      ,
      • Stevens V.
      • Concannon C.
      • van Wijngaarden E.
      • McGregor J.
      Validation of the chronic disease score-infectious disease (CDS-ID) for the prediction of hospital-associated Clostridium difficile infection (CDI) within a retrospective cohort.
      ,
      • Khanafer N.
      • Vanhems P.
      • Barbut F.
      • Luxemburger C.
      Factors associated with Clostridium difficile infection: a nested case-control study in a three year prospective cohort.
      ,
      • Ingle M.
      • Deshmukh A.
      • Desai D.
      • Abraham P.
      • Joshi A.
      • Rodrigues C.
      • et al.
      Prevalence and clinical course of Clostridium difficile infection in a tertiary-care hospital: a retrospective analysis.
      ,
      • Šuljagić V.
      • Miljković I.
      • Starčević S.
      • Stepić N.
      • Kostić Z.
      • Jovanović D.
      • et al.
      Risk factors for Clostridium difficile infection in surgical patients hospitalized in a tertiary hospital in Belgrade, Serbia: a case-control study.
      ]. The finding that lipid-lowering drugs had a statistically significant risk for CDI is novel and requires further study, although the difference is small. Among patients with CDI, we also found higher rates of hypertension, renal disease, moderate-to-severe liver disease, IBD and malignant cancer, compared with patients with no CDI. Conversely, the incidence of IBD was not significantly higher among patients with recurrent versus non-recurrent CDI, and the rate of malignant cancer was significantly higher in the non-recurrent CDI group. Hypertension, renal disease, liver disease and malignant cancer have been found to be associated with recurrence in previous studies [
      • Chung M.S.
      • Kim J.
      • Kang J.O.
      • Pai H.
      Impact of malignancy on Clostridium difficile infection.
      ,
      • Phatharacharukul P.
      • Thongprayoon C.
      • Cheungpasitporn W.
      • Edmonds P.J.
      • Mahaparn P.
      • Bruminhent J.
      The risks of incident and recurrent Clostridium difficile-associated diarrhea in chronic kidney disease and end-stage kidney disease patients: a systematic review and meta-analysis.
      ,
      • Abdelfatah M.
      • Nayfe R.
      • Nijim A.
      • Enriquez K.
      • Ali E.
      • Watkins R.R.
      • et al.
      Factors predicting recurrence of Clostridium difficile infection (CDI) in hospitalized patients: retrospective study of more than 2000 patients.
      ,
      • Mathews S.N.
      • Lamm R.
      • Yang J.
      • Park J.
      • Tzimas D.
      • Buscaglia J.M.
      • et al.
      Factors associated with health care utilization of recurrent Clostridium difficile infection in New York State.
      ,
      • Gómez S.
      • Chaves F.
      • Orellana M.A.
      Clinical, epidemiological and microbiological characteristics of relapse and re-infection in Clostridium difficile infection.
      ]. The reason for the higher rate of malignant cancer among patients with non-recurrent CDI in our study needs to be further explored and may be due to the methodology used in the definition of the recurrence window, which is a particular challenge in this type of large database study, or the immortal time bias already described.
      Strengths of this study include its use of objective death data from the ONS mortality registry to confirm vital status and its inclusion of both primary and secondary care data in risk factor and covariate definitions. This is likely to have increased outcome ascertainment in our study, as well as the accuracy and completeness of data on baseline characteristics and covariates. Our study also provides insights on the risk of outcomes in both the early and later periods following CDI diagnosis by taking account of all deaths recorded up to 12 months.
      Sensitivity analyses to assess the robustness of our findings offered insights into some of the methodological issues involved in the use of real-world data for infectious disease epidemiology. For example, comparison of the mortality incidence between recurrent and non-recurrent CDI groups showed that patients with recurrent CDI had significantly lower mortality up to 90 days from the start of initial hospitalization, than those with non-recurrent CDI. This can be explained by the effects of immortal time bias: by definition, patients with recurrent CDI would need to survive long enough to experience a second CDI episode 13–56 days after the first. Similarly, our time-fixed analyses showed that recurrent CDI was protective regarding the 12-month risk of death. However, this implausible benefit was invalidated once immortal time was correctly classified by following patients from the start of the recurrence window in a time-dependent analysis or from the end of the recurrence risk window in a time-fixed analysis. In contrast with mortality, the occurrence of a non-fatal complication does not preclude a recurrent CDI episode, and the 12-month risk of complications were thus more frequently observed in the group with recurrent CDI versus non-recurrent CDI, regardless of the analyses used.
      One limitation of this study was that some patients in our primary and secondary analyses may have been incorrectly classified as CDI-positive due to the lack of laboratory confirmation of infection. Secondly, a precise onset or CDI diagnosis date was not available for either the primary or recurrent episode of CDI, as clinical data in the HES database are grouped into episodes of care rather than presented as separate encounters []. This limitation may have led to misclassification of initial CDI or recurrence status, and/or an inaccurate time to event in the survival analysis. Given this important limitation, throughout our study we selected analysis parameters that would provide more conservative estimates of the effect of CDI exposure on outcomes. Thirdly, the CPRD GOLD and HES databases lack information on some important factors, such as disease severity, infection strain, metabolic changes associated with CDI, severity of comorbidities and prescribing in the secondary care setting, which may introduce residual confounding. It is, however, reasonable to assume that some factors may be non-differentially distributed between exposure and comparison groups. Additionally, location and organizational information such as hospital provider codes or hospital ward are not routinely available in the HES data collected by CPRD, to minimize the potential risk of patient re-identification. Hospital and ward information was therefore not used as a matching parameter for cases and controls. Instead, we opted to take account of whether patients were admitted for emergency or elective procedures as a proxy for the severity of the patient condition at the time of admission. There is also no linkage from HES to microbiology data or outbreak surveillance systems, so we were unable to account for impact of local outbreaks. For the CDI versus no CDI analysis, we matched on year of admission, and for the recurrent CDI versus CDI analysis, we adjusted for year of admission, in an attempt to mitigate the impact of large national outbreaks or important temporal trends. Lastly, in the absence of objective laboratory or pathology evidence to link CDI to cause of death, it was not possible to estimate the proportion of deaths attributable to CDI; given this, we assessed all-cause mortality up to 12 months to capture both short and longer term, and direct and indirect effects of CDI.
      In summary, this study draws on a substantial body of data from routine clinical practice to demonstrate that patients who have CDI during their hospital stay have higher mortality rates than patients who do not. Among patients who have HA-CDI, those with recurrent CDI are at greater risk of complications or death than those with non-recurrent CDI. Our data provide a comprehensive picture of the impact of CDI in England, and highlight the importance of preventing and managing initial and recurrent episodes of CDI.

      Acknowledgements

      This study was funded by Astellas Pharma Europe Ltd. We wish to thank Reiner Tretter, statistician at Astellas Pharma Inc. at the time of this study, for his contribution to the analysis and Iona Easthope, DPhil, of Cello Health MedErgy for medical writing services. Previous presentations: oral presentation at the 25th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID), 25th−28th April 2015, Copenhagen, Denmark; e-poster at the 27th ECCMID, 22nd–25th April 2017, Vienna, Austria.

      Author contributions

      D.A.E., T.M-T., N.A., D.D., N.F. and A.K. were involved in the concept and design of the study. T.M-T., N.A., D.D., A.G. and A.K. were involved in the acquisition, analysis and/or interpretation of the data. All authors participated in drafting the manuscript and its critical revisions for important intellectual content. All authors approved the final submitted article.

      Conflict of interest statement

      D.A.E. has received fees for conference attendance from Astellas Pharma, Eumedica, Pfizer, Gilead and MSD, and for consultancy work from Cardiome. T.M-T. and D.D. are full-time employees of CPRD, which provides contract research services and which received payment from Astellas Pharma for work on the study. N.A. and A.G. were full-time employees of Astellas Pharma at the time of the study. A.K. was a full-time employee of Astellas Pharma at the time of the study, and has patents WO2015169451 A1 and EP17167541.6 pending, which are licenced to Astellas Pharma.

      Funding sources

      This research study and medical writing services were funded by Astellas Pharma Europe Ltd.
      Astellas Pharma Europe Ltd. was involved in the study design; the collection, analysis and interpretation of data; the writing of the report; and the decision to submit the article for publication.

      Appendix A. Supplementary data

      The following is the Supplementary data to this article:

      References

        • Zhu D.
        • Sorg J.A.
        • Sun X.
        Clostridioides difficile biology: sporulation, germination, and corresponding therapies for C. difficile infection.
        Front Cell Infect Microbiol. 2018; 8: 1-29
        • Schäffler H.
        • Breitruck A.
        Clostridium difficile – from colonization to infection.
        Front Microbiol. 2018; 9: 646
        • Cohen S.H.
        • Gerding D.N.
        • Johnson S.
        • Kelly C.P.
        • Loo V.G.
        • McDonald L.C.
        • et al.
        Clinical practice guidelines for Clostridium difficile infection in adults: 2010 update by the Society for Healthcare Epidemiology of America (SHEA) and the Infectious Diseases Society of America (IDSA).
        Infect Control Hosp Epidemiol. 2010; 31: 431-455
        • Haines C.F.
        • Moore R.D.
        • Bartlett J.G.
        • Sears C.L.
        • Cosgrove S.E.
        • Carroll K.
        • et al.
        Clostridium difficile in a HIV-infected cohort: incidence, risk factors, and clinical outcomes.
        AIDS. 2013; 27: 2799-2807
        • Dial S.
        • Delaney J.A.C.
        • Barkun A.N.
        • Suissa S.
        Use of gastric acid-suppressive agents and the risk of community-acquired Clostridium difficile-associated disease.
        J Am Med Assoc. 2005; 294: 2989-2995
        • Asha N.J.
        • Tompkins D.
        • Wilcox M.H.
        Comparative analysis of prevalence, risk factors, and molecular epidemiology of antibiotic-associated diarrhea due to Clostridium difficile, Clostridium perfringens, and Staphylococcus aureus.
        J Clin Microbiol. 2006; 44: 2785-2791
        • Howell M.D.
        • Novack V.
        • Grgurich P.
        • Soulliard D.
        • Novack L.
        • Pencina M.
        • et al.
        Iatrogenic gastric acid suppression and the risk of nosocomial Clostridium difficile infection.
        Arch Intern Med. 2010; 170: 784-790
        • Lo Vecchio A.
        • Zacur G.M.
        Clostridium difficile infection: an update on epidemiology, risk factors, and therapeutic options.
        Curr Opin Gastroenterol. 2012; 28: 1-9
        • Bouza E.
        Consequences of Clostridium difficile infection: understanding the healthcare burden.
        Clin Microbiol Infect. 2012; 18: 5-12
        • Wiegand P.N.
        • Nathwani D.
        • Wilcox M.H.
        • Stephens J.
        • Shelbaya A.
        • Haider S.
        Clinical and economic burden of Clostridium difficile infection in Europe: a systematic review of healthcare-facility-acquired infection.
        J Hosp Infect. 2012; 81: 1-14
        • Bartlett J.G.
        • Gerding D.N.
        Clinical recognition and diagnosis of Clostridium difficile infection.
        Clin Infect Dis. 2008; 46: S12-S18
        • Chakra C.N.A.
        • Pepin J.
        • Sirard S.
        • Valiquette L.
        Risk factors for recurrence, complications and mortality in Clostridium difficile infection: a systematic review.
        PLoS One. 2014; 9 (e98400)
        • McDonald L.C.
        • Gerding D.N.
        • Johnson S.
        • Bakken J.S.
        • Carroll K.C.
        • Coffin S.E.
        • et al.
        Clinical practice guidelines for Clostridium difficile infection in adults and children: 2017 update by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA).
        Clin Infect Dis. 2018; 66: e1-48
        • Olsen M.A.
        • Stwalley D.
        • Demont C.
        • Dubberke E.R.
        Clostridium difficile infection increases acute and chronic morbidity and mortality.
        Infect Control Hosp Epidemiol. 2019; 40: 65-71
        • Bauer M.P.
        • Notermans D.W.
        • van Benthem B.H.
        • Brazier J.S.
        • Wilcox M.H.
        • Rupnik D.L.
        • et al.
        Clostridium difficile infection in Europe: a hospital-based survey.
        Lancet. 2011; 377: 63-73
        • Barbut F.
        • Bouée S.
        • Longepierre L.
        • Goldberg M.
        • Bensoussan C.
        • Levy-Bachelot L.
        Excess mortality between 2007 and 2014 among patients with Clostridium difficile infection: a French health insurance database analysis.
        J Hosp Infect. 2018; 98: 21-28
        • Banks A.
        • Moore E.K.
        • Bishop J.
        • Coia J.E.
        • Brown D.
        • Mather H.
        • et al.
        Trends in mortality following Clostridium difficile infection in Scotland, 2010-2016: a retrospective cohort and case-control study.
        J Hosp Infect. 2018; 100: 133-141
        • Reacher M.
        • Verlander N.Q.
        • Roddick I.
        • Trundle C.
        • Brown N.
        • Farrington M.
        • et al.
        Excess mortality attributable to Clostridium difficile and risk factors for infection in an historic cohort of hospitalised patients followed up in the United Kingdom death register.
        PLoS One. 2016; 11: 1-23
        • Kuntz J.L.
        • Baker J.M.
        • Kipnis P.
        • Li S.X.
        • Liu V.
        • Xie Y.
        • et al.
        Utilization of health services among adults with recurrent Clostridium difficile infection: a 12-year population-based study.
        Infect Control Hosp Epidemiol. 2017; 38: 45-52
        • Shah D.N.
        • Aitken S.L.
        • Barragan L.F.
        • Bozorgui S.
        • Goddu S.
        • Navarro M.E.
        • et al.
        Economic burden of primary compared with recurrent Clostridium difficile infection in hospitalized patients: a prospective cohort study.
        J Hosp Infect. 2016; 93: 286-289
        • Olsen M.A.
        • Yan Y.
        • Reske K.A.
        • Zilberberg M.D.
        • Dubberke E.R.
        Recurrent Clostridium difficile infection is associated with increased mortality.
        Clin Microbiol Infect. 2015; 21: 164-170
        • Reveles K.R.
        • Lawson K.A.
        • Mortensen E.M.
        • Pugh M.J.V.
        • Koeller J.M.
        • Argamany J.R.
        • et al.
        National epidemiology of initial and recurrent Clostridium difficile infection in the Veterans Health Administration from 2003 to 2014.
        PLoS One. 2017; 12: 1-14
        • Khanna S.
        • Gupta A.
        • Baddour L.M.
        • Pardi D.S.
        Epidemiology, outcomes, and predictors of mortality in hospitalized adults with Clostridium difficile infection.
        Intern Emerg Med. 2016; 11: 657-665
        • Esteban-Vasallo M.D.
        • de Miguel-Díez J.
        • López-de-Andrés A.
        • Hernández-Barrera V.
        • Jiménez-García R.
        Clostridium difficile-related hospitalizations and risk factors for in-hospital mortality in Spain between 2001 and 2015.
        J Hosp Infect. 2019; 102: 148-156
        • Sammons J.S.
        • Localio R.
        • Xiao R.
        • Coffin S.E.
        • Zaoutis T.
        Clostridium difficile infection is associated with increased risk of death and prolonged hospitalization in children.
        Clin Infect Dis. 2013; 57: 1-8
        • UK National Health Service
        Hospital episode Statistics.
        2019 (last accessed March 2020)
      1. UK Statistics Authority. Office for national Statistics n.d. https://www.ons.gov.uk/(last accessed March 2020).

        • UK Government
        English indices of deprivation.
        2019 (last accessed March 2020)
        • UK National Health Service
        Hospital Episode Statistics (HES) Analysis Guide.
        2019 (last accessed March 2020)
        • Jen M.-H.
        • Saxena S.
        • Bottle A.
        • Pollok R.
        • Holmes A.
        • Aylin P.
        Assessment of administrative data for evaluating the shifting acquisition of Clostridium difficile infection in England.
        J Hosp Infect. 2012; 80: 229-237
        • Hensgens M.P.M.
        • Goorhuis A.
        • Dekkers O.M.
        • Van Benthem B.H.B.
        • Kuijper E.J.
        All-cause and disease-specific mortality in hospitalized patients with Clostridium difficile infection: a multicenter cohort study.
        Clin Infect Dis. 2013; 56: 1108-1116
        • Escobar G.J.
        • Baker J.M.
        • Li S.X.
        • Xie Y.
        • Kipnis P.
        Clinical and economic burden of recurrent Clostridium difficile infections – a 10-year retrospective large database analysis.
        54th Intersci. Conf. Antimicrob. Agents Chemother. 2014;
        • Shintani A.K.
        • Girard T.D.
        • Eden S.K.
        • Arbogast P.G.
        • Moons K.G.M.
        • Ely E.W.
        Immortal time bias in critical care research: application of time-varying Cox regression for observational cohort studies.
        Crit Care Med. 2009; 37: 2939-2945
        • Delgado A.
        • Reveles I.A.
        • Cabello F.T.
        • Reveles K.R.
        Poorer outcomes among cancer patients diagnosed with Clostridium difficile infections in United States community hospitals.
        BMC Infect Dis. 2017; 17: 448
        • Forster A.J.
        • Taljaard M.
        • Oake N.
        • Wilson K.
        • Roth V.
        • van Walraven C.
        The effect of hospital-acquired infection with Clostridium difficile on length of stay in hospital.
        CMAJ. 2012; 184: 37-42
        • Yasunaga H.
        • Horiguchi H.
        • Hashimoto H.
        • Matsuda S.
        • Fushimi K.
        The burden of Clostridium difficile-associated disease following digestive tract surgery in Japan.
        J Hosp Infect. 2012; 82: 175-180
        • Goldenberg S.D.
        • Brown S.
        • Edwards L.
        • Gnanarajah D.
        • Howard P.
        • Jenkins D.
        • et al.
        The impact of the introduction of fidaxomicin on the management of Clostridium difficile infection in seven NHS secondary care hospitals in England: a series of local service evaluations.
        Eur J Clin Microbiol Infect Dis. 2016; 35: 251-259
        • Louie T.J.
        • Miller M.A.
        • Mullane K.M.
        • Weiss K.
        • Lentnek A.
        • Golan Y.
        • et al.
        Fidaxomicin versus vancomycin for Clostridium difficile infection.
        N Engl J Med. 2011; 364: 422-431
        • Cornely O.A.
        • Crook D.W.
        • Esposito R.
        • Poirier A.
        • Somero M.S.
        • Weiss K.
        • et al.
        Fidaxomicin versus vancomycin for infection with Clostridium difficile in Europe, Canada, and the USA: a double-blind, non-inferiority, randomised controlled trial.
        Lancet Infect Dis. 2012; 12: 281-289
        • Stevens V.
        • Dumyati G.
        • Fine L.S.
        • Fisher S.G.
        • van Wijngaarden E.
        Cumulative antibiotic exposures over time and the risk of Clostridium difficile infection.
        Clin Infect Dis. 2011; 53: 42-48
        • Stevens V.
        • Concannon C.
        • van Wijngaarden E.
        • McGregor J.
        Validation of the chronic disease score-infectious disease (CDS-ID) for the prediction of hospital-associated Clostridium difficile infection (CDI) within a retrospective cohort.
        BMC Infect Dis. 2013; 13: 1-8
        • Khanafer N.
        • Vanhems P.
        • Barbut F.
        • Luxemburger C.
        Factors associated with Clostridium difficile infection: a nested case-control study in a three year prospective cohort.
        Anaerobe. 2017; 44: 117-123
        • Ingle M.
        • Deshmukh A.
        • Desai D.
        • Abraham P.
        • Joshi A.
        • Rodrigues C.
        • et al.
        Prevalence and clinical course of Clostridium difficile infection in a tertiary-care hospital: a retrospective analysis.
        Indian J Gastroenterol. 2011; 30: 89-93
        • Šuljagić V.
        • Miljković I.
        • Starčević S.
        • Stepić N.
        • Kostić Z.
        • Jovanović D.
        • et al.
        Risk factors for Clostridium difficile infection in surgical patients hospitalized in a tertiary hospital in Belgrade, Serbia: a case-control study.
        Antimicrob Resist Infect Control. 2017; 6: 31
        • Chung M.S.
        • Kim J.
        • Kang J.O.
        • Pai H.
        Impact of malignancy on Clostridium difficile infection.
        Eur J Clin Microbiol Infect Dis. 2016; 35: 1771-1776
        • Phatharacharukul P.
        • Thongprayoon C.
        • Cheungpasitporn W.
        • Edmonds P.J.
        • Mahaparn P.
        • Bruminhent J.
        The risks of incident and recurrent Clostridium difficile-associated diarrhea in chronic kidney disease and end-stage kidney disease patients: a systematic review and meta-analysis.
        Dig Dis Sci. 2015; 60: 2913-2922
        • Abdelfatah M.
        • Nayfe R.
        • Nijim A.
        • Enriquez K.
        • Ali E.
        • Watkins R.R.
        • et al.
        Factors predicting recurrence of Clostridium difficile infection (CDI) in hospitalized patients: retrospective study of more than 2000 patients.
        J Investig Med. 2015; 63: 747-751
        • Mathews S.N.
        • Lamm R.
        • Yang J.
        • Park J.
        • Tzimas D.
        • Buscaglia J.M.
        • et al.
        Factors associated with health care utilization of recurrent Clostridium difficile infection in New York State.
        J Clin Gastroenterol. 2019; 53: 298-303
        • Gómez S.
        • Chaves F.
        • Orellana M.A.
        Clinical, epidemiological and microbiological characteristics of relapse and re-infection in Clostridium difficile infection.
        Anaerobe. 2017; 48: 147-151