In this retrospective multicentre cohort study, we described a total of 23 new drug–ADR associations that were not mentioned in the SmPC. In addition, a total of seven ADRs were reported with a higher cumulative incidence in the IBDREAM registry compared with the corresponding SmPC. Finally, there was one ADR that was mentioned with an unknown incidence in the SmPC, for which we calculated a cumulative incidence in IBDREAM.
Most new drug–ADR associations that were found in this study were related to azathioprine, such as pyrexia, malaise, arthralgia, nausea and diarrhoea. Although these symptoms are mentioned in the SmPC as part of a hypersensitivity reaction , our study suggests that they may also occur independently and may not always point towards a hypersensitivity reaction. Similarly, arthralgia, pyrexia and rash associated with the use of mercaptopurine are only reported in the SmPC as part of a hypersensitivity reaction , and not separately. These findings were supported by recent studies [21,22,23,24] in which pyrexia, malaise, nausea and diarrhoea associated with the use of azathioprine, and arthralgia, pyrexia and rash associated with the use of mercaptopurine, were reported separately. Symptoms such as arthralgia may also be considered underlying symptoms of IBD. Similarly, digestive intolerance such as diarrhoea, nausea, vomiting and abdominal pain may be due to other causes such as irritable bowel syndrome or underlying IBD and may therefore be potentially misclassified as an ADR despite careful assessment by HCPs and assessors during analysis. As we did not correct for the background incidence of these symptoms, the calculated cumulative incidences in our study might be an overestimation of true occurrence.
Skin rash associated with the use of mercaptopurine was observed with a ‘common’ (≥ 1% to < 10%) cumulative incidence in IBDREAM (1.7%) while the ADR is documented as ‘rare’ (≥ 0.01% to < 0.1%) in the SmPC. A retrospective study  reported a cumulative incidence of 3.0% for skin rash associated with the use of mercaptopurine in azathioprine-intolerant patients, suggesting a ‘common’ frequency, which is in line with the findings in our study.
There were four new ADRs associated with the use of thioguanine, including arthralgia. Arthralgia is a common idiosyncratic ADR during thioguanine use  and may occur at any time during the use of thioguanine. Arthralgia associated with the use of thioguanine was reported with a ‘common’ (≥ 1% to < 10%) cumulative incidence in IBDREAM (1.9%; n = 11). This finding is in line with the cumulative incidence of 5.8% reported in a Dutch retrospective study (n = 16) , while a Dutch prospective study reported an even higher cumulative incidence of ‘very common’, with 10.6% (n = 10) . From the latter 10 cases, seven were possibly related to arthralgia and three cases resulted in discontinuation. As stated previously, arthralgia is also associated with the use of azathioprine (cumulative incidence 1.9%) and mercaptopurine (cumulative incidence 2.5%). Patients who experience arthralgia during the use of azathioprine may benefit from a switch to mercaptopurine [28,29,30]. This might be due to nitro-imidazole, which is released from azathioprine when converted to mercaptopurine. In general, this would suggest a higher cumulative incidence of arthralgia during azathioprine use compared with mercaptopurine, which is in line with the findings in our study. However, other factors might also play a role in developing arthralgia, such as comorbidities and comedication. In addition, arthralgia is the most common extraintestinal manifestation of IBD, which occurs in one-third of IBD patients . One notable finding in the current study is that the drug was withdrawn as a result of arthralgia in 66.7% (n = 18) of the azathioprine cases, 75% (n = 21) of the mercaptopurine cases and 100% (n = 11) of the thioguanine cases. In all cases, patients reported resolution of symptoms, which points to a causal relationship.
Nausea and abdominal pain associated with the use of thioguanine were reported with a ‘common’ cumulative incidence (4.8% and 3.2%, respectively) in IBDREAM, but were not mentioned in the SmPC of thioguanine.
These findings were confirmed by a recent retrospective cohort study  that observed a ‘common’ cumulative incidence for both nausea and abdominal pain, with 2.2% (n = 6). In that study, patients experienced these ADRs as mild and the ADRs did not lead to treatment discontinuation.
The presence of antidrug antibodies directed at infliximab was reported with a higher cumulative incidence compared with the SmPC. However, antidrug antibodies themselves are usually not considered an ADR, and consequently the incidence rates of antidrug antibodies are underreported and underestimated in IBDREAM compared with the SmPC. Furthermore, ADRs that included investigations rather than symptoms (e.g. anaemia, leukocytosis) were only reported if this resulted in discontinuation of the treatment. In addition, only symptoms as a result of these type of ADRs were reported (e.g. fatigue in the context of anaemia). Therefore, the calculated cumulative incidences of these objective ADRs should be interpreted with caution as these are likely to be underestimations of cumulative incidences compared with the corresponding SmPCs.
This study shows that in the current system for ADR collection, as reflected by the SmPCs, there is room for improvement to provide a representative overview of the ADRs occurring in daily clinical practice. A previous study showed that HCPs significantly underreport ADRs . HCPs may want to avoid the administrative burden of ADR reporting. In order to complement the information in the SmPC, other sources for ADR reporting, such as patient registries, should be considered. The advantages of these registries comprise the long follow-up per patient, the large number of patients included and the inclusion of patients who are normally not included in clinical trials due to specific selection criteria. Digital support for reporting and assessing ADRs is crucial to further facilitate this process.
A strength of this study is the use of real-world data from daily practice, including 2993 IBD patients. Because of the large number of patients treated in different hospitals across the country, the study population is a representative sample of IBD patients treated in The Netherlands. Therefore, the external validity of this study is high. Even though the ADRs were not assessed for causality, the HCP verified and registered all ADRs. Therefore, we believe that all drug–ADR combinations included in this study, especially the 23 new combinations, are strong drug–ADR associations that require further research.
A limitation of this retrospective observational study is that patients may not always report every complaint to their HCP. Additionally, HCPs will not always register every ADR reported by patients. These may both cause an underestimation of cumulative incidences reported in the IBDREAM registry compared with reality. Although we did not adjust for the background incidence and because some ADRs may be misclassified, the calculated cumulative incidences could be interpreted as the lower limit of the range of the related frequency category since the ADRs in this study were most often reported in case the drug was withdrawn (75%) or the dosage was altered (2%). These ADRs are therefore considered clinically relevant since they resulted in adjustment (dose reduction or discontinuation) of therapy. Second, the reported ADRs may also be a result of the underlying disease, co-medication or co-morbidities. Although careful assessment by the HCP and the pharmacovigilance assessor limits this risk, it cannot be excluded in all cases. The clinical impact (e.g. improvement of symptoms) of stopping therapy has facilitated this analysis to correctly assign ADRs to reported symptoms.