This systematic review and meta-analysis was conducted and reported according to the established guideline for meta-analysis [
All published and unpublished randomized controlled trials and observational studies including prospective cohort, retrospective cohort, case-control, and cross-sectional studies, involving patients infected with
The primary outcome was the comparison of the number of participants with a history of NSAID exposure and CDAD versus those without CDAD.
Anawin Sanguankeo and Sikarin Upala independently searched published studies indexed in the PubMed/MEDLINE and EMBASE from database inception to October 2014. References of selected retrieved articles were also examined. The search terms used were
Anawin Sanguankeo and Sikarin Upala independently reviewed the titles and abstracts of all citations that were identified. After all the studies were abstracted, face-to-face data comparisons were conducted between investigators to ensure completeness and reliability. The inclusion criteria were independently applied to all identified studies. Differing decisions were resolved by consensus.
Full-text versions of potentially relevant papers identified in the initial screening were retrieved. If multiple articles from the same study were found, only the article with the longest follow-up period was included. Data concerning study design (cross-sectional, case-control, prospective cohort, and retrospective cohort), participant characteristics (age, sex, and settings), NSAID use (previous or current use, overall, and specific NSAID use), and outcome measures (definition of CDAD, number of participants, odds ratio, or risk ratio) were independently extracted. We contacted the authors of the primary reports to request any unpublished data. If the authors did not reply, we used the available data for our analyses.
The quality of observational studies (OBS) was evaluated by each investigator using the Newcastle-Ottawa quality assessment scale [
Data analysis was performed using the Comprehensive Meta-Analysis 3.3 software from the Biostat, Inc. We reported the estimated pooled odds ratio (OR) of NSAID use using a random effects model because of the high likelihood of heterogeneity among the studies. Subgroup analyses were performed based on types of NSAID, duration of NSAID (60 days or less and more than 60 days), age (50 years or less and more than 50 years), and risk of bias (high and low risk of bias). The heterogeneity of the effect size estimates across these studies was quantified using the
The initial search yielded 987 articles (Figure
Results of information search.
A total of 16 articles underwent full-length review. Finally, data were extracted from nine observational studies [
Characteristics of included studies.
Study (year) | Design | Participants characteristics | CDAD criteria | Controls | Participants ( |
---|---|---|---|---|---|
Dial et al. [ |
Population-based case-control studies | Participants aged ≥18 years |
Presence of a |
Age matched controls were not hospitalized in the |
CDAD, 1,233 |
|
|||||
El Feghaly et al. [ |
Case-control study | Children with diarrhea from inpatient, outpatient, and emergency department visits | Diarrhea with positive |
Symptomatic |
CDAD, 65 |
|
|||||
Loo et al. [ |
Prospective cohort study | Participants aged ≥18 years admitted to units with a historically |
Presence of diarrhea and a positive |
Asymptomatic |
CDAD, 117 |
|
|||||
Manges et al. [ |
Nested case-control study | Subjects enrolled in a large cohort study supported by |
( |
Matched controls in a 1 : 2 ratio according to |
CDAD, 25 |
|
|||||
Naggie et al. [ |
Case-control study | Participants aged ≥18 years | Diarrhea (increased stool output and unformed |
Matched controls by geographic |
CDAD, 66 |
|
|||||
Pépin et al. [ |
Retrospective cohort study | Adult patients hospitalized at |
( |
— | CDAD, 293 |
|
|||||
Regnault et al. [ |
Retrospective study | All patients hospitalized for IBD flares in the Gastroenterology Department of the Saint-Antoine IBD Center | Positive stool toxigenic culture and a positive stool cytotoxicity assay or, in cases of negative stool cytotoxicity assays, a positive toxigenic culture | — |
|
|
|||||
Soes et al. 2014 [ |
Prospective matched case-control study | All patients aged ≥2 years that had a fecal sample submitted because of diarrhea |
Patients with diarrhea or other gastrointestinal |
Matched controls by age, sex, and site for laboratory analyses of |
CDAD, 177 |
|
|||||
Suissa et al. [ |
Case-control study | Participants aged ≥18 years and have at least 2 years of records in the GPRD | First clinical diagnosis |
Matched controls by age, medical practice | CDAD, 1360 |
CDAD:
Quality assessment scores using the Newcastle-Ottawa Scale tool for observational studies are summarized in Table
Summary of quality assessment.
Author | Selection (max. 4) | Comparability (max. 2) | Exposure (max. 3) |
---|---|---|---|
Dial et al. [ |
4 | 2 | 2 |
El Feghaly et al. [ |
3 | 0 | 2 |
Loo et al. [ |
3 | 2 | 3 |
Manges et al. [ |
4 | 2 | 2 |
Naggie et al. [ |
4 | 2 | 2 |
Pépin et al. [ |
2 | 2 | 2 |
Regnault et al. [ |
2 | 0 | 2 |
Soes et al. [ |
4 | 2 | 2 |
Suissa et al. [ |
4 | 2 | 2 |
The meta-analysis was performed using the random effects model (Figure
Forest plot of the included studies comparing odds ratio of CDAD in patients who used NSAID and those who did not.
Forest plot of subgroup analysis in (a) nonselective NSAID and (b) each type of NSAID.
Forest plot of subgroup analysis by risk of bias.
Forest plot of subgroup analysis by age group.
Forest plot of subgroup analysis by duration of NSAID use.
Sensitivity analysis was performed using a fixed effects model rather than a random effects model. The result of the point estimate and its statistical significance were not different from the main result.
To investigate potential publication bias, we examined the contour-enhanced funnel plot of the included studies. Vertical axis represents study size (standard error) while horizontal axis represents effect size (log odds ratio). From this plot, bias is not present because there is symmetrical distribution of studies on both sides of the mean. Egger’s regression test for bias was −0.45 (95% CI, −3.50 to 2.60) (Supplemental Figure
To the best of our knowledge, this is the first systematic review and meta-analysis of published observational studies to demonstrate the association between the use of NSAIDs and CDAD. The results showed that odds of CDAD among patients with NSAID exposure were significantly increased. This association between NSAID and CDAD was found in nonselective NSAID and patients who were 50 years or older, regardless of duration of NSAID use.
It should be noted that the case definitions of CDAD in the individual studies were not exactly the same. The first prescription of oral vancomycin was added into the case definition, in addition to clinical diarrhea, laboratory diagnosis of CDAD, and the presence of pseudomembranous colitis, to increase statistical power and reduce exposure misclassification of the study by Suissa et al. [
This review has several limitations, and, thus, our results should be interpreted with caution. First, the major limitation of our review is the small number of studies that met our inclusion criteria; only eight studies were included in the meta-analysis. Pépin et al. did not provide the number of participants with CDAD or control data [
In conclusion, we found a significant association between the use of NSAIDs and having CDAD. The results of this systematic review and meta-analysis have important implications: CDAD should be considered part of the differential diagnosis when faced with patients that present with acute diarrhea and a history of recent NSAID exposure, in addition to the use of antimicrobial or acid-suppressive agents. However, it should be noted that the results of this meta-analysis of observational studies can only demonstrate the association, not the causal relationship. Further studies are necessary to evaluate whether any specific types of NSAIDs can increase the risk of CDAD or if the course of CDAD treatment should be prolonged if NSAIDs are being used concurrently.
The authors have no competing interests to declare.