Temporal relationship between multiple drugs and multiple events in patient reports on adverse drug reactions: findings in a pilot study in Japan

Patient reporting of suspected adverse drug reactions (ADRs) has been accepted as an essential component of pharmacovigilance.1,2 Important signals may be detected when combined with reports from health care professionals (HCPs)3 but also from “patient-only” dataset.4 In January 2011, the Health and Labour Sciences Research Grant study (“HLSRG study”) supported by the Ministry of Health, Labour andWelfare (MHLW) started a pilot study to determine whether the web-based patient reporting works in Japan. The online patient report form was linked from the main page of Pharmaceuticals and Medical Devices Agency (PMDA). To call for patient reports, a message was shown on the PMDAwebsite and in 955 (1.7%) of about 55000 community pharmacies, 1000 posters were displayed and 100000 flyers were handed to patients. The study was approved by the Research Ethics Committee of Faculty of Pharmacy, Keio University. In an interim analysis of the data collected during the first 2 months of the study, we found patients sometimes reported multiple drugs and multiple ADRs occurring at different occasions. It was often very difficult to specify the temporal relationship between the exposure to drugs and occurrence of ADRs. To improve the clarity of the temporal relationship, a revised form with multiple entry boxes for dates of ADR occurrence was used in the latter half of the study (see online Figure 1). At the end of the study, we conducted an ad hoc analysis to know whether the revised form worked as intended. The drugs were classified by the anatomical therapeutic chemical (ATC) classification. ADRs were coded by Japanese translation of lowest level terms of the medical dictionary for regulatory activities (MedDRA) version 14.0J and converted to preferred terms (PTs) by two of authors (AH and KK) and then further checked independently by each of two psychiatric specialists (AI and TK) because many of reported ADRs were classified under the primary systemic organ class (SOC) “nervous” or “psychiatry”. The temporal relationship was classified into “clear” (clear for all drug–event combinations), “clear/unclear” (clear for some but unclear for the other combinations) and “unclear” (unclear for all combinations) also by these two specialists. Any discrepancy between two specialists was resolved by discussion. We also analyzed the spontaneous reports submitted by HCPs directly or through the drug company to the PMDA (“HCP reports”) and reports submitted by the “consumer or other non-health professional” through the drug company (“indirect patient reports”) during the same period using Japanese Adverse Drug Event Report database (http://www.info.pmda.go.jp/) and compared with “direct patient reports” in the HLSRG study. Table 1 shows standardized differences5 for demographic and other characteristics between a total of 214 direct patient reports with at least one identifiable drug and one identifiable ADR and 657 indirect patient reports as well as 34843 HCP reports submitted to the PMDA between January and December 2011. In direct patient reports, median age of patients was 40 years old and much younger than “60’s” in other two report types. There was a difference of ADRs and drugs frequently reported between three reporter types. The 214 reports were submitted by 166 (78%) patients and 48 (22%) family members, and 16 (7%) reports had one or more over-the-counter drugs. As shown in Table 2, during the latter half of the study period where the revised form was used, two or more dates for ADR occurrence were often reported. The ratio of the odds of the clarity (“clear” vs. others (“clear/ unclear” or “unclear”)) in the latter half of the study period (51/3) to the odds in the first half (110/50) was 7.7 (95% CI:2.3–26.0). The logistic regression revealed that the “odds ratio” was 9.8 (95%CI:2.5–38.1) when adjusted for age and sex of reporters and other potentially relevant factors indicating that the revised form improved the clarity of the temporal relationship. The number of reports in the first to fourth quarter during the one-year study period was 108, 52, 30 and 24, respectively. This decrease was potentially because patients who used the internet regularly to avidly collect the information of ADRs tended to discover that the pilot study was in place and report ADRs they experienced early in the study period.