2023 Volume 46 Issue 5 Pages 655-660
Appendicitis is one of the most common abdominal surgical emergencies worldwide; however, its causes remain poorly understood. The Japanese Adverse Drug Event Report (JADER) database is a spontaneous reporting system (SRS) that can be utilized to analyze the safety signals of adverse events. In this study, we investigated the association between drug use and the onset of appendicitis using the JADER database. We first used the reporting odds ratio (ROR) as the signal and found signals for appendicitis, perforated appendicitis, and complicated appendicitis for 23, 9, and 1 drug, respectively. To investigate the level of hazard over time in drug-associated appendicitis, the Weibull shape parameter β was calculated using a Weibull plot, which revealed drug-dependent patterns for changes in the risk of appendicitis over time for the eight drugs. Furthermore, logistic regression analysis was performed to account for the influence of age, sex, and primary disease, and a significant association was detected between two drugs and appendicitis. Several types of drugs, such as antitumor, antirheumatic, and anti-inflammatory drugs, were included in our analyses; however, only clozapine, which is used for patients with schizophrenia, was commonly identified in these analyses. The resulting data suggest that certain drugs may be associated with appendicitis and may require adequate attention.
Appendicitis is characterized by inflammation of the vermiform appendix and is one of the most common surgical emergencies. Appendicitis is reported to have a 7–8% lifetime risk. It occurs at a rate of approximately 100 patients per 100000 inhabitants per year in developed countries and is slightly more common in men.1–3) The peak incidence is usually in the second or third decade of life, and the disease is less common in children and elderly people.2) Despite the fact that appendicitis is a very common disease, its causes remain poorly understood.
Appendicitis is classified as simple or complicated. Simple appendicitis is defined as acute appendicitis without perforation, whereas complicated appendicitis is defined as rupture of the appendix with subsequent abscess or phlegmon formation. Although obstruction of the appendiceal outlet caused by appendicoliths, calculi, lymphoid hyperplasia, impacted stool, infection, or tumors have generally been proposed as the primary factors for appendicitis that often cause perforation, these factors tend to be exceptions rather than regular occurrences.2–4) In contrast, a population-based study found no significant negative relationship between negative appendectomy and perforation rate, suggesting that perforated and non-perforated appendicitis may have different pathologies.5)
The Japanese Adverse Drug Event Report (JADER) database of the Pharmaceuticals and Medical Devices Agency (PMDA) is a spontaneous reporting system (SRS), similar to the U.S. Food and Drug Administration’s Adverse Event Reporting System (FAERS). As an SRS is used for pharmacovigilance assessments that reflect the clinical practice, the JADER database could be used to analyze the safety signal detection of adverse events. Indexes for signal detection in SRS, represented by the reporting odds ratio (ROR) and proportional reporting ratios, have been developed to identify drug-associated adverse events using disproportionality analyses. However, limitations of SRS, such as underestimation, errors, and inaccuracies, are well known. Therefore, statistically detected signals do not necessarily suggest a causal relationship between the drug and adverse events but indicate a sufficient likelihood for validation.6)
In this study, we investigated the reported frequency and time to onset of appendicitis to detect associations between drugs and appendicitis using the JADER database. This study aimed to identify drugs that may cause appendicitis as an adverse event.
Data from the JADER database for April 2004–December 2021 were obtained from the PMDA website (www.pmda.go.jp). The structure of the JADER database complies with international safety reporting guidelines (ICH E2B). The database consisted of four data tables: 1) patient demographic information (demo), 2) drug information (drug), 3) adverse events (reac), and 4) primary illness (hist). The JADER database does not contain codes for identifying case reports. The “drug” file contains the role code assigned to each drug: suspected, interacting, and concomitant. In this study, we analyzed the records of suspected drugs.
Definition of Adverse Events and Drug NamesAccording to the terminology preferred by the Medical Dictionary for Regulatory Activities (MedDRA) version 25.0, which was coded in the “reac” table, we extracted the case reports of appendicitis, perforated appendicitis, complicated appendicitis, noninfective appendicitis, and stump appendicitis. We excluded records that did not include the starting or termination dates of administration while constructing the Weibull plot and those without age, sex, or primary disease information were excluded from logistic regression analysis. The drug name was referred to by the International Non-proprietary Name (INN) or the United States Adopted Name (USAN).
Signal DetectionFor signal detection, we calculated the ROR, which is the ratio of the odds of a certain adverse event occurring with the drug of interest to the odds of the same event occurring with all other drugs in the database. Specifically, we selected drugs that were reported in at least three cases as the suspected drug for each adverse event in the JADER database and confirmed the statistical significance between the two categorical variables using Fisher’s exact test. We then calculated the ROR and 95% confidence intervals (CIs) for drugs that showed statistical significance using Fisher’s exact test. To be defined as a signal, the lower limit of the 95% CI of the ROR had to be greater than 1.
Time-to-Onset AnalysisIn the time-to-onset analysis, we excluded the records without information pertaining to the date of starting the administration or the date of occurrence of the adverse events. Using the time-to-onset data, which was calculated as the time between a patient’s first prescription and the occurrence of an adverse event, the Weibull shape parameter β, which determines the shape of the distribution function, was calculated and used to indicate the level of hazard over time without a reference population.7–9) Specifically, we applied the Weibull plot and estimated the shape parameter β. The Weibull plot has a special scale designed to make the points straight or nearly straight if the data follows a Weibull distribution. A least-squares fit of this straight line yielded an estimate of the Weibull shape parameter β, which is the slope of the fitted line. In the time-to-onset analysis, the cumulative distribution function was obtained using the mean rank if the sample size was greater than 20 or the median rank if the sample size was ≤20. Additionally, the statistical significance of the straight line fitted to the Weibull plot was confirmed. When β is equal to 1, the hazard is estimated to be constant over time (random failure). If β is greater than 1 and the 95% CI of β excluded the value 1, the hazard increases over time (wear-out failure). Finally, if β is less than 1 and the 95% CI of β excluded the value 1, the hazard is considered to decrease over time (early failure).
Logistic RegressionThe data were stratified by age (less than or greater than 50 years). To construct the logistic model, sex, stratified age groups, and the primary disease for which the drug was most frequently used were coded. In this study, we calculated the adjusted ROR and 95% CI for drugs with more than five events per variable (EPV) and confirmed that multicollinearity was absent because the variance inflation factor for the independent variables was less than 4.0.10,11)
Statistical AnalysisAll statistical data analyses were performed using R version 4.2.1. and Microsoft Excel for Mac version 16.63.1.
In total, the JADER database comprised 25157119 records. Of these, 23605854 records were removed due to reasons such as role code being interacting and concomitant in the drug table, duplicates, or missing values. The total number of adverse events recorded in this study was 1551265, of which 591, 129, and 21 were records of appendicitis, perforated appendicitis, and complicated appendicitis, respectively. No reports of non-infective appendicitis or stump appendicitis were found. The ROR signals for appendicitis were detected for 23 drugs, including barium sulfate, clozapine, vedolizumab, liraglutide, ustekinumab, lanthanum carbonate, adalimumab, tocilizumab, nintedanib esylate, dapagliflozin propanediol, tofacitinib citrate, cytarabine, olaparib, esomeprazole magnesium, mafgnesium oxide, infliximab, calcium levofolinate, docetaxel, fluorouracil, temozolomide, bevacizumab, methotrexate, and prednisolone (Table 1). Clozapine, tocilizumab, bevacizumab, calcium levofolinate, cytarabine, fluorouracil, irinotecan, gemcitabine, and carboplatin showed ROR signals for perforated appendicitis, whereas only clozapine showed ROR signal for complicated appendicitis (Table 1).
| Disease | Drug | Cases | Non-cases | Total | Ratio (%) | ROR (95% CI) |
|---|---|---|---|---|---|---|
| Appendicitis | Total | 591 | 1550674 | 1551265 | 0.038 | |
| Barium sulfate | 7 | 199 | 206 | 3.40 | 93.39 (43.8–199) | |
| Clozapine | 88 | 5069 | 5157 | 1.71 | 53.3 (42.5–67.0) | |
| Vedolizumab | 4 | 613 | 617 | 0.65 | 17.2 (6.42–46.2) | |
| Liraglutide | 5 | 802 | 807 | 0.62 | 16.5 (6.82–39.9) | |
| Ustekinumab | 6 | 1253 | 1259 | 0.48 | 12.7 (5.66–28.4) | |
| Lanthanum carbonate | 3 | 818 | 821 | 0.37 | 9.67 (3.10–30.1) | |
| Adalimumab | 6 | 1786 | 1792 | 0.33 | 8.89 (3.97–19.9) | |
| Tocilizumab | 31 | 12260 | 12291 | 0.25 | 6.95 (4.84–9.98) | |
| Nintedanib esylate | 3 | 1231 | 1234 | 0.24 | 6.42 (2.06–20.0) | |
| Dapagliflozin propanediol | 3 | 1233 | 1236 | 0.24 | 6.41 (2.06–20.0) | |
| Tofacitinib citrate | 7 | 3289 | 3296 | 0.21 | 5.64 (2.67–11.9) | |
| Cytarabine | 9 | 4329 | 4338 | 0.21 | 5.52 (2.86–10.7) | |
| Olaparib | 4 | 2245 | 2249 | 0.18 | 4.70 (1.76–12.6) | |
| Esomeprazole magnesium | 3 | 1998 | 2001 | 0.15 | 3.95 (1.27–12.3) | |
| Magnesium oxide | 3 | 2140 | 2143 | 0.14 | 3.69 (1.19–11.5) | |
| Infliximab | 8 | 5851 | 5859 | 0.14 | 3.62 (1.80–7.28) | |
| Calcium levofolinate | 7 | 5123 | 5130 | 0.14 | 3.62 (1.72–7.62) | |
| Docetaxel | 5 | 4072 | 4077 | 0.12 | 3.24 (1.34–7.82) | |
| Fluorouracil | 16 | 14843 | 14859 | 0.11 | 2.88 (1.75–4.73) | |
| Temozolomide | 8 | 8001 | 8009 | 0.10 | 2.65 (1.32–5.32) | |
| Bevacizumab | 16 | 21173 | 21189 | 0.08 | 2.01 (1.75–4.73) | |
| Methotrexate | 17 | 23230 | 23247 | 0.07 | 1.95 (1.32–5.32) | |
| Prednisolone | 20 | 32299 | 32319 | 0.06 | 1.65 (1.05–2.57) | |
| Perforated appendicitis | Total | 129 | 1551136 | 1551265 | 0.0083 | |
| Clozapine | 16 | 5141 | 5157 | 0.31 | 42.58 (25.2–71.9) | |
| Tocilizumab | 14 | 12277 | 12291 | 0.11 | 15.3 (8.76–26.6) | |
| Bevacizumab | 17 | 21172 | 21189 | 0.080 | 11.0 (6.58–18.3) | |
| Calcium levofolinate | 4 | 5126 | 5130 | 0.078 | 9.65 (3.56–26.1) | |
| Cytarabine | 3 | 4335 | 4338 | 0.069 | 8.50 (2.70–26.7) | |
| Fluorouracil | 8 | 14851 | 14859 | 0.054 | 6.84 (3.34–14.0) | |
| Irinotecan | 4 | 9107 | 9111 | 0.044 | 5.42 (2.00–14.7) | |
| Gemcitabine | 3 | 6953 | 6956 | 0.043 | 5.29 (1.68–16.6) | |
| Carboplatin | 5 | 13081 | 13086 | 0.038 | 4.74 (1.94–11.6) | |
| Complicated appendicitis | Total | 21 | 1551244 | 1551265 | 0.0014 | |
| Clozapine | 7 | 5150 | 5157 | 0.14 | 150 (60.6–372) |
CI, confidence interval; ROR, reported odds ratio.
Several drugs that indicated the ROR signal of appendicitis showed statistically significant regression coefficients in the Weibull plot and estimated a shape parameter β that indicated a failure pattern (Table 2). Box charts of the time of onset of appendicitis with these drugs are shown in Fig. 1. The hazards associated with the use of tocilizumab, tofacitinib citrate, fluorouracil, and infliximab were estimated to be early failure patterns; thus, the hazards of these drugs decreased over time. The hazards for appendicitis with bevacizumab, methotrexate, and prednisolone were estimated to be random failure patterns; thus, the hazards of these drugs were considered to be constant over time. Only clozapine was estimated to have a wear-out failure pattern, indicating an increasing hazard over time. In perforated appendicitis, only bevacizumab showed a statistically significant regression coefficient in the Weibull plot (Supplementary Table S1, Fig. S1).
| Drug | Case | Shape parameter (β) | Pattern | |
|---|---|---|---|---|
| β | 95% CI | |||
| Clozapine | 40 | 1.56 | (1.50–1.62) | Wear out failure |
| Methotrexate | 6 | 1.33 | (0.81–1.85) | Random failure |
| Bevacizumab | 9 | 1.15 | (0.83–1.46) | Random failure |
| Prednisolone | 5 | 0.78 | (0.46–1.10) | Random failure |
| Tofacitinib citrate | 7 | 0.64 | (0.41–0.86) | Early failure |
| Tocilizumab | 23 | 0.53 | (0.50–0.57) | Early failure |
| Fluorouracil | 6 | 0.55 | (0.20–0.91) | Early failure |
| Infliximab | 6 | 0.37 | (0.31–0.43) | Early failure |
CI, confidence interval.
Logistic regression was performed to remove the effects of confounding factors such as age, sex, and primary disease. Clozapine and tocilizumab with more than five EPV in appendicitis were applied to logistic regression and showed significant correlations with appendicitis regardless of age, sex, or primary disease (Table 3). However, no drugs showed more than five EPV in perforated appendicitis and complicated appendicitis. In performing logistic regression, schizophrenia and rheumatoid arthritis were coded as the primary diseases for clozapine and tocilizumab, respectively.
| Drug | Case | Total | Crude ROR (95% CI) | Adjusted ROR (95% CI) |
|---|---|---|---|---|
| Clozapine | 84 | 4960 | 55.6 (44.0–70.3) | 40.2 (29.6–54.6) |
| Tocilizumab | 30 | 12291 | 7.34 (5.08–10.6) | 4.58 (2.95–7.13) |
CI, confidence interval; ROR, reported odds ratio.
Although appendicitis is one of the most common surgical emergencies worldwide, its cause remains unclear. The SRS, represented by JADER and FAERS, plays an important role in pharmacovigilance by providing information from clinical practice in the post-marketing environment.12–14) Nevertheless, the limitations of the data should be understood in interpreting the results obtained from the SRS, since the SRS is a passive reporting system that is liable to over-reporting, under-reporting, missing data on patient characteristics, exclusion of healthy subjects, missing denominators, and the presence of confounding factors.15,16) In this study, we applied the ROR as a signal to find unknown correlations between adverse events and drugs. Notably, the ROR is an indicator of increased risk of adverse events and does not represent the absolute risk of occurrence of adverse events.17) Hence, we additionally applied the Weibull plot for the analysis of onset time and failure pattern of appendicitis and logistic regression for removal of the influences of confounding factors. As a result of these analyses, we have drawn the following inferences in the present study: (1) ROR signals suggesting an association between clinical drugs and appendicitis, perforated appendicitis, and complicated appendicitis were detected in several drugs; (2) several drugs with an ROR signal showed characteristic onset time profiles, and (3) several drugs were significantly associated with appendicitis regardless of age, sex, and primary disease.
In signal analysis, 23 drugs showed signals for appendicitis (Table 1). Notably, appendicitis is not mentioned in the Japanese package inserts for any of these drugs with the exception of barium sulfate, which was found to be associated with appendicitis. The concomitant use of barium sulfate was examined for all drugs (Table 1); however, barium sulfate was not used concomitantly. Thus, none of the signals found in Table 1, except for barium sulfate, could be attributed to the concomitant use of barium sulfate. The drugs that showed ROR signals in appendicitis are prescribed for cancer, rheumatoid arthritis, ulcerative colitis, Crohn’s disease, diabetes mellitus, and other ailments. Considering that obstruction of the appendiceal outlet due to appendicoliths, calculi, lymphoid hyperplasia, impacted stool, infection, or tumors is the commonly proposed cause of appendicitis, the relationships between drugs and appendicitis suggested in Table 1 do not appear to be consistent with the proposed causes. In addition, logistic regression analysis in Table 3 shows that both clozapine and tocilizumab had statistically significant correlations with appendicitis, but their underlying mechanisms were quite different. Future pharmacological experiments, randomized clinical trials, case-control studies, and cohort analyses are needed to confirm these drugs as causative agents of appendicitis. This may contribute to identifying new aspects of the pharmacological mechanisms underlying the action of these drugs.
Of the 23 drugs with suggested association with appendicitis, only clozapine, tocilizumab, cytarabine, calcium levofolinate, fluorouracil, and bevacizumab showed ROR signals in perforated appendicitis (Table 1). In contrast, irinotecan, gemcitabine, and carboplatin, which showed signals in perforated appendicitis, did not show any signals in appendicitis. Given that the pharmacological mechanisms underlying the action of the six drugs with signals in perforated appendicitis were different, these results seem consistent with the previous suggestion that perforated and non-perforated appendicitis may have different pathologies.5)
Recently, the Weibull shape parameter β was used to detect the time-to-onset profile for adverse events.7–9) Therefore, we used the Weibull plot and estimated the Weibull shape parameter β (Table 2; Supplementary Table S1). As shown in Table 2, the pattern of failure varied between the drugs and did not appear to depend on the pharmacological mechanisms. Since the risk associated with drugs showing the early failure pattern is expected to decrease with time, appendicitis should be considered immediately after treatment with these drugs. Only the failure pattern of clozapine was estimated to be wear-out failure, predicting an increasing hazard with time. Some adverse events of clozapine have been reported to occur long after administration.18,19) Considering our results and previous reports about the onset time of adverse events, attention should be paid for an extended duration to the onset of appendicitis in patients taking clozapine.
In this study, clozapine, which is an antipsychotic, showed signals for both cases of appendicitis: perforated and complicated; additionally, significant correlations with appendicitis were detected in the logistic regression analysis (Tables 1, 3). Clozapine is used worldwide for the treatment of schizophrenia, and many adverse effects on the central nervous system, gastrointestinal tract, heart, and others have been reported.20–22) Among the adverse effects of clozapine, constipation, which is likely to have been caused by the anticholinergic and antiserotonergic effects of clozapine, is a frequent adverse effect that often causes fecal impaction related to appendicitis.23–25) Previous studies have revealed that clozapine is associated with significant gastrointestinal hypomotility, which is related to constipation26,27); however, luminal obstruction resulting from fecal impaction is not always the primary cause of appendicitis.2) Although drugs cannot be simplistically compared due to the limitations of SRS, the drugs suggested to be associated with appendicitis in Table 1, such as those for schizophrenia, are not the only ones that have been shown to exert anticholinergic or antiserotonin effects. These facts suggest that further in vitro, in vivo, and clinical studies are needed to clarify the mechanism underlying the action of clozapine leading to appendicitis.
Of the several studies were that were returned as the result of PubMed and Google Scholar searches using the terms “clozapine” and “appendicitis,” one was a case series with reported six cases of appendicitis within 14 years. All cases were of perforated appendicitis and were observed in a German forensic psychiatric hospital that treated 220 patients annually using clozapine.28) Another study is a case report of a 46-year-old Japanese man with treatment-resistant schizophrenia who suffered from constipation and slight abdominal discomfort while taking clozapine and developed appendicitis, leading to perforation of the large intestine.29) Another study found in a Google Scholar search that was limited to second-generation antipsychotics shows an association between clozapine and appendicitis.30) A systematic review has been published regarding the disproportionality analysis studies focusing on safety of antipsychotic drugs.31) Most recently, a retrospective cohort study investigated and reported a relationship between clozapine exposure and the onset of appendicitis in schizophrenia patients.32) As these previous studies limited the range of the confounding factors such as antipsychotic drugs or presence of schizophrenia in the patients, the inference presented in the current study may be more general and significantly different from the previous studies. Despite the difference in the limitation of confounding factors, both the present and previous studies using SRS suggest a significant relationship between clozapine and appendicitis, indicating sufficient likelihood for validation in further studies.
The present study suggests that tocilizumab, an antirheumatic drug against interleukin-6 (IL-6), might also be involved in the development of appendicitis (Tables 1, 3), but only a few other reports have shown a relationship between tocilizumab and appendicitis.33,34) IL-6 is well known as a cytokine with various physiological activities and plays an important role in inflammation and immune reactions. Tocilizumab binds to the IL-6 receptor and suppresses IL-6 activity, potentially causing appendicitis; however, the detailed mechanism remains unclear. The present study suggests that this relationship may be validated with future studies as in the case of clozapine.
A recently published FAERS-based study, which was not limited to antipsychotic and schizophrenic patients and focused on perforated appendicitis, reported that perforated appendicitis occurred in patients treated for rheumatoid arthritis or receiving drugs including clozapine and bevacizumab.35) Although the adverse events of interest in this older study using FAERS and the present study (i.e., perforated appendicitis and appendicitis) are different, some trends are similar with respect to the primary disease and drugs. Taking into account the possible differences in pathologies between perforated and non-perforated appendicitis and the differences in characteristics between JADER and FAERS, further pharmacovigilance studies are needed to confirm the relationship between drugs and appendicitis.5,36)
This study suggests that several drugs may be associated with the development of appendicitis. Therefore, paying focusing sufficient attention on appendicitis during intake of these drugs may be necessary.
The authors declare no conflict of interest.
This article contains supplementary materials.
