Clinical and Psychosocial Characteristics of Adult Primary Care IBS Patients: A Post hoc Analysis of the DOMINO Study.

The DOMINO study was approved by the Ethical Committee Research UZ/KU Leuven, Belgium ( S59482 ).

Written informed consent was obtained from each participant in the DOMINO study.

This manuscript did not receive grants from funding agencies. The DOMINO study was funded by the Belgian Health Care Knowledge Center (KCE) Trials Program (study ID KCE16001).

A total of 483 primary care IBS patients were enrolled [ 13 ]. Baseline characteristics of the DOMINO study participants are shown in Table  1 . The Rome IV criteria were fulfilled by 70% of the participants (Rome+). Of the patients not fulfilling Rome IV criteria (Rome−), 64% did not meet the criteria based on frequency of abdominal pain and 56% based on stool requirements. Absence of abdominal pain was uncommon in this cohort (6%) (Table  2 ). Baseline characteristics of DOMINO study participants. Abbreviations: BMI, body mass index; GAD, generalised anxiety disorder; GI, gastrointestinal; GP, general practitioner; IBS, irritable bowel syndrome; IBS‐C, constipation‐predominant IBS; IBS‐D, diarrhoea‐predominant IBS; IBS‐M, IBS with mixed bowel habits; IBS‐U, IBS unclassified; PHQ, patient health questionnaire; QoL, quality of life; SD, standard deviation; SSS, symptom severity score. Cause of IBS Rome IV negativity ( n  = 133). Abbreviation: IBS, irritable bowel syndrome. Patients were subtyped by the Rome IV questionnaire as follows: 20% IBS‐C, 33% IBS‐D, 31% IBS‐M and 16% IBS‐U. The average score on the IBS‐SSS was 268 ± 98, with 46% and 36% of the cases reporting moderate and severe symptoms, respectively. The most bothersome symptom in the past 3 months was bloating (37%), followed by abdominal pain (28%) and diarrhea (19%). On average, the somatic symptom disorder level was medium and the depression and anxiety level moderate. Differences between the Rome+ and Rome− groups are shown in Table  3 . The Rome + group was significantly younger compared to Rome− and displayed a higher female proportion. The IBS Rome IV subtypes differed significantly according to Rome IV positivity. The prevalence of IBS‐D and IBS‐M was significantly higher in the Rome+ group, while this group consisted proportionally of less IBS‐U patients. Differences according to IBS Rome IV positivity. Abbreviations: GAD, generalised anxiety disorder; IBS, irritable bowel syndrome; IBS‐C, constipation‐predominant IBS; IBS‐D, diarrhoea‐predominant IBS; IBS‐M, IBS with mixed bowel habits; IBS‐U, IBS unclassified; PHQ, patient health questionnaire; QoL, quality of life; SD, standard deviation; SSS, symptom severity score. Rome + participants scored significantly higher on the IBS‐SSS compared to Rome−. Likewise, they had a significantly lower quality of life. In addition, Rome + patients had significantly higher somatic symptom disorder and depression and anxiety levels, indicating a more substantial psychological burden. According to the Rome IV questionnaire, patients were subdivided into 20% IBS‐C, 33% IBS‐D, 31% IBS‐M and 16% IBS‐U, while the GP subdivided patients into 21% IBS‐C, 28% IBS‐D, 39% IBS‐M and 12% IBS‐U. Significant differences were shown between both, also after excluding IBS‐U cases ( p  < 0.001) (Supporting Information  S1 : Table 1). In the IBS‐U group, more than half of the patients were categorized differently by GPs compared with the Rome IV questionnaire. A reasonable concordance was found for the other subtypes. Differences according to IBS Rome IV subtype are shown in Table  4 . IBS‐M and IBS‐D patients were significantly younger than IBS‐U patients. Furthermore, IBS‐M participants had a significantly lower age compared with IBS‐C participants. In the IBS‐C and IBS‐M group, the female proportion was significantly higher than in the IBS‐U group. Differences according to IBS Rome IV subtype. Overall: < 0.001 IBS‐D versus IBS‐U: 0.001 IBS‐M versus IBS‐U: < 0.001 IBS‐C versus IBS‐M: 0.01 Overall: < 0.001 IBS‐C/IBS‐M versus IBS‐U: < 0.05 Overall: 0.003 IBS‐C versus IBS‐D: 0.004 IBS‐C versus IBS‐U: 0.02 Overall: < 0.001 IBS‐C/IBS‐M versus IBS‐U: < 0.001 IBS‐D versus IBS‐U: 0.001 IBS‐D versus IBS‐M: 0.03 Overall: < 0.001 IBS‐C/IBS‐D/IBS‐M versus IBS‐U: < 0.05 IBS‐C/IBS‐M versus IBS‐U: < 0.05 Overall: < 0.001 IBS‐C versus IBS‐U: 0.02 IBS‐D versus IBS‐U: 0.003 IBS‐M versus IBS‐U: < 0.001 Overall: < 0.001 IBS‐C versus IBS‐U: 0.02 IBS‐D versus IBS‐U: 0.008 IBS‐M versus IBS‐U: < 0.001 Overall: < 0.001 IBS‐D/IBS‐M versus IBS‐U: < 0.05 IBS‐M versus IBS‐U: < 0.05 IBS‐M versus IBS‐U: < 0.05 Overall: 0.001 IBS‐M versus IBS‐U: 0.001 IBS‐D versus IBS‐M: 0.04 Overall: 0.03 IBS‐M versus IBS‐U: < 0.05 Overall: 0.008 IBS‐M versus IBS‐U: 0.005 Overall: 0.09 IBS‐M versus IBS‐U: < 0.05 Overall: 0.008 IBS‐M versus IBS‐U: 0.007 Overall: 0.30 IBS‐M versus IBS‐U: < 0.05 Abbreviations: GAD, generalised anxiety disorder; IBS, irritable bowel syndrome; IBS‐C, constipation‐predominant IBS; IBS‐D, diarrhoea‐predominant IBS; IBS‐M, IBS with mixed bowel habits; IBS‐U, IBS unclassified; PHQ, patient health questionnaire; QoL, quality of life; SD, standard deviation; SSS, symptom severity score. Only p ‐values of statistically significant one‐to‐one post hoc comparisons were reported. IBS‐M, IBS‐D and IBS‐C participants had a significantly higher IBS‐SSS score compared with IBS‐U patients. Also, the IBS‐M group scored significantly higher on the IBS‐SSS than the IBS‐D group. IBS‐M, IBS‐D and IBS‐C patients reported a significantly lower quality of life compared with IBS‐U. PHQ‐15 scores were significantly higher in IBS‐M, IBS‐D and IBS‐C than in IBS‐U. IBS‐M patients scored significantly higher on the PHQ‐12 than IBS‐U and IBS‐D patients. In addition, IBS‐M participants had significantly higher depression and anxiety levels compared with IBS‐U participants. We identified lower age, higher somatic symptom disorder levels and having a defined IBS Rome IV subtype as statistically significant predictors of higher IBS‐SSS (Supporting Information  S1 : Table 2). Furthermore, higher depression levels, higher anxiety levels and higher IBS‐SSS were significant predictors of higher IBS‐QoL (Supporting Information  S1 : Table 3).

This cross‐sectional post hoc analysis investigated the clinical and psychosocial characteristics of adult primary care IBS patients from the DOMINO trial. Recruitment in the DOMINO study (trial registration number NCT04270487 , approved by the Ethical Committee Research UZ/KU Leuven ( S59482 )) ran from July 2018 to December 2019. The trial design was published previously [ 13 ]. All authors had access to the study data and reviewed and approved the final manuscript. Newly diagnosed or to be treated IBS patients were included in the DOMINO trial by GPs. Patients with a concurrent gastrointestinal disease, a history of major abdominal surgery, diabetes, uncontrolled coexisting diseases such as thyroid dysfunction, active malignancy, symptomatic endometriosis and a major psychiatric disorder or dosage alteration of antidepressants in the last 3 months were excluded. Women of childbearing potential not using contraception could not participate. Patients with a history of treatment with OB for more than 3 consecutive weeks and/or any intake of OB in the last 3 months were not eligible. Furthermore, patients with a history of a FODMAP‐lowering diet or an ongoing elimination diet were excluded. Use of medication for IBS during the last 3 weeks prior to the trial intervention was forbidden [ 13 ]. Baseline characteristics, including age, sex, weight, body mass index (BMI), predominant stool type and lifestyle, were collected by the GP. The latter encompassed smoking, alcohol consumption and sportiness level. GPs determined the predominant stool type during routine clinical history‐taking. At baseline, patients filled out the Rome IV IBS diagnostic questionnaire to determine fulfilment of the Rome IV IBS criteria (Rome+) and the IBS subtype [ 14 ]. Additionally, the IBS Symptom Severity Scale (IBS‐SSS) [ 15 ] and questionnaires assessing quality of life (IBS‐Quality of Life (IBS‐QoL)) [ 16 ], somatic symptom disorder whether or not taking gastrointestinal symptoms into account (Patient Health Questionnaire (PHQ)‐15 and PHQ‐12, respectively) [ 17 , 18 ], depression (PHQ‐9) [ 19 ] and anxiety (Generalised Anxiety Disorder (GAD)‐7) [ 20 ] were completed. The questionnaires are described in more detail in Supporting Information  S1 . A descriptive characterisation of the primary care IBS cohort was performed in terms of demographics, lifestyle, Rome IV positivity, IBS subtype, symptom severity, quality of life and psychological burden. The IBS subtype derived from the Rome IV IBS diagnostic questionnaire was compared with the IBS subtype defined by the GP. Furthermore, differences in baseline data and questionnaire results according to IBS Rome IV positivity and IBS subtype were analyzed. Finally, we identified statistically significant predictors of IBS‐SSS and IBS‐QoL. Continuous variables are reported as mean ± standard deviation and categorical variables as percentage and number. Between‐group differences were studied using the Mann‐Whitney, Kruskal‐Wallis, Fisher's Exact and Chi‐square tests. One‐to‐one post hoc comparisons were performed when analysing more than 2 groups, with a Bonferroni correction for multiple tests. To identify statistically significant predictors of IBS‐SSS and IBS‐QoL, multiple linear regressions and general linear models were used. The assumptions of all statistics were checked and fulfilled. Statistical significance was defined as a two‐sided p ‐value < 0.05. Participants with missing data important for a certain analysis were excluded from this analysis. All statistical analyses were performed using GraphPad Prism version 9.4.1 (GraphPad Software Inc.) and SPSS Statistics version 28.0.1.1 (IBM Corporation).

Although most IBS patients are diagnosed and managed in primary care, this clinical setting is underinvestigated to date. The present study was a large‐scale cross‐sectional analysis which aimed to explore the clinical and psychosocial characteristics of newly diagnosed adult primary care IBS patients. We found that the majority of patients fulfilled the Rome IV criteria and were subtyped as IBS‐D or IBS‐M. Most subjects were characterised by moderate or severe IBS‐SSS and on average, the level of psychological burden was rated as moderate. Rome+ and IBS‐M participants reported higher symptom severity, lower quality of life and more pronounced psychosocial comorbidity. Participant demographics were in line with previous epidemiological research and clinical trials in primary care IBS, that is, female predominance and a mean age of around 40 years [ 2 , 3 , 10 , 21 ]. In this study, as in clinical reality, not all IBS patients fulfilled the Rome IV criteria, which have been shown to have a diagnostic sensitivity of 63% and specificity of 97% [ 14 ]. The most common reason for patients not meeting the Rome IV IBS criteria was insufficient abdominal pain frequency. Notably, the Rome IV framework does not consider pain severity, even though infrequent but intense pain may affect quality of life as much as more frequent and milder pain. In our cohort, the mean abdominal pain intensity over the preceding 10 days (0–100 scale) was 54.4 among patients who met the frequency criterion ( n  = 365) and 30.6 among those who did not ( n  = 85; p  < 0.0001). Importantly, the pain‐intensity range in the latter group spanned 0 to 100, indicating that patients with low pain frequency may still experience episodes of substantial severity. Previous research in primary care and the overall IBS population found total absence of abdominal pain as the most frequent reason for patients not being characterised as IBS according to the Rome IV criteria [ 22 , 23 ]. However, these studies evaluated abdominal discomfort and pain using self‐designed questionnaires based on the Rome III criteria, whereas we only assessed abdominal pain using the Rome IV IBS diagnostic questionnaire. Perhaps, patients reporting a low frequency of abdominal pain in our cohort would be more likely to describe this as discomfort if given the opportunity, and therefore might fit into the category of total absence of abdominal pain. The distribution of IBS stool pattern subtypes is not consistent across different studies. A worldwide prevalence analysis found IBS‐C and IBS‐M as the most frequent subtypes, in contrast to IBS‐D and IBS‐M in our cohort [ 2 ]. Similarly, in the Atlantis trial, which investigated the effect of low‐dose amitriptyline in Rome + primary care patients, IBS‐D and IBS‐M were the most prevalent subtypes, encompassing > 80% of the participants [ 21 ]. The stool consistency and frequency tend to fluctuate over time within an individual, which might make a single assessment of these parameters less accurate [ 24 ]. Furthermore, the distribution is known to depend on the population investigated, the geographical location and the criteria used to determine the subtypes [ 25 ]. The latter could explain the difference between the subtype determined by the GP (via history‐taking) and the subtype determined by the Rome IV IBS diagnostic questionnaire (completed by the patient) in this study. Because of the well‐defined and systematic way of questioning bowel habits, the Rome IV questionnaire seems more accurate in identifying the subtype. The notably high proportion of IBS‐U in our cohort warrants attention. To remain as close as possible to routine primary care practice, the DOMINO trial did not apply the Rome IV recommendation of using at least 2 weeks of daily diary data for subtype classification [ 1 ]. Instead, subgrouping was based on a questionnaire, which may have made it more difficult for patients to accurately recall stool consistency over recent bowel movements. However, this factor alone is unlikely to fully explain the finding, as the Atlantis trial, also relying on a questionnaire rather than a diary‐based approach, reported a substantially lower proportion of IBS‐U, namely 3% [ 21 ]. The vast majority of participants reported moderate or severe symptoms, in contrast to the commonly accepted estimated prevalence of IBS‐SSS gradations, that is, mild 40%, moderate 35% and severe 25% [ 26 ]. The present study included treatment‐naïve IBS patients, which could at least partly explain this finding. Also, previous research suggested that IBS patients only managed in primary care are affected as much as those presenting to secondary care [ 27 ]. The Atlantis trial, which included patients with a history of prior treatments, reported similar results, with 85% of participants scoring in the moderate‐to‐severe range on the IBS‐SSS [ 21 ]. These findings underline the important role of GPs in the diagnosis and treatment of IBS. The mean IBS‐QoL in this cohort was in accordance with previous reports across different care settings [ 16 , 28 , 29 ]. The on average moderate level of psychosocial comorbidity in the present study was compatible with the established association between DGBIs and mental health [ 8 , 9 ]. The participants scored numerically higher on the psychological burden questionnaires compared with the general population [ 30 , 31 , 32 , 33 ]. In general, Rome + patients reported higher symptom severity scores compared to Rome−. However, the symptom severity, somatic symptom disorder and depression level remained moderate in Rome− cases. Furthermore, a population‐based survey found that self‐reported IBS patients were characterised by a similar quality of life impact and a more pronounced psychosocial comorbidity compared with Rome + patients [ 23 ]. Our findings of the comparisons between Rome+ and Rome− participants were largely in line with earlier research in the overall IBS population [ 34 , 35 ]. However, a tertiary care analysis did not demonstrate a more severe depression and anxiety level in Rome + subjects, in contrast to our findings [ 36 ]. Psychosocial comorbidity has been shown to predict attending tertiary care, which could explain a higher prevalence of depression and anxiety in this setting regardless of Rome positivity [ 37 ]. The Rome + group had a higher representation of younger and female participants in this study. Consequently, apart from being more prevalent in the overall IBS population, these age and sex categories seem to exhibit a more severe clinical presentation of the disorder. When comparing the IBS subtypes, our findings suggested that IBS‐U patients represent a less severe subgroup of the disorder. Previously, differences between IBS subtypes were shown to mostly focus on psychological burden, with IBS‐M having the highest level of somatic symptom disorder, depression and anxiety [ 38 ]. Our results were in line with these findings, but we identified additional domains of differentiation. Interestingly, IBS‐D, but not IBS‐C, had a lower symptom severity and milder somatic symptom disorder compared to IBS‐M. Perhaps, the different impact of IBS‐D and IBS‐C on daily life, as shown in a cross‐sectional analysis, could account for this discrepancy. IBS‐D patients avoided leaving the house or traveling because of concerns about bathroom availability, whereas IBS‐C patients developed a reluctance to sexual intercourse, reported difficulty concentrating and felt more self‐conscious about their body image [ 39 ]. These affected areas could influence the assessment of symptom severity and somatic symptom disorder, especially as problems during sexual intercourse are part of the PHQ‐12 questionnaire [ 18 ]. Similarly, previous research showed that the frequency of a high level of somatic symptom disorder was lower in IBS‐D, but not in IBS‐C compared with IBS‐M [ 8 ]. In our cohort, IBS‐C patients were older and more likely to be female compared with other subtypes. This finding is in accordance with other studies showing that women report more constipation than men and that the prevalence of constipation increases with age [ 40 ]. Furthermore, IBS‐U consisted of older subjects and less women, suggesting once again an association between age and sex on the one hand and severity of the clinical presentation on the other hand. We identified age, Rome subtype and somatic symptom disorder, but not depression and anxiety, as statistically significant predictors of higher IBS‐SSS scores. By contrast, Spiller et al. demonstrated that PHQ‐12 and depression were both independent predictors of IBS‐SSS, despite the fact that PHQ‐12 was more strongly correlated with IBS‐SSS than anxiety and depression [ 18 ]. Furthermore, we found that IBS‐SSS, depression and anxiety, but not somatic symptom disorder, were predictors of IBS‐QoL. Knowles et al. identified somatic symptom disorder as an additional significant predictor. However, they used a different tool to assess the quality of life, with physical and mental subscales. Somatic symptom disorder was the strongest predictor of decreased physical quality of life, whereas depression, followed by anxiety, was the strongest predictor of decreased mental quality of life [ 6 ]. IBS‐SSS has already been identified as a predictor of IBS‐QoL previously [ 16 ]. In the present study, Rome+ and IBS‐M participants had a higher symptom severity, a lower quality of life and a more pronounced psychosocial comorbidity. However, little attention is paid to Rome IV positivity and stool pattern subtype in primary care, and the Rome criteria are not included in current primary care IBS guidelines [ 41 ]. Only a minority of GPs are familiar with and use the Rome criteria for diagnosing IBS, and 70% still believe that IBS is a diagnosis of exclusion [ 42 , 43 , 44 ]. The Rome IV criteria could be useful for clinical stratification in a Rome+ and a Rome− group, as well as into different stool pattern subtype groups, each of which is associated with a distinct severity profile and, particularly for the stool‐pattern subtypes, a specific therapeutic approach [ 45 ]. In that way, assessing Rome IV positivity and stool pattern subtype in newly diagnosed IBS patients might immediately provide GPs insight into disease severity, and help them to establish an individualised therapeutic strategy (Figure  1 ). for example, a Rome + patient was more likely to have a higher psychosocial comorbidity in this study and therefore could hypothetically benefit more from the initiation of a neuromodulator as part of the management of IBS. In addition, the Rome criteria may support establishing a positive diagnosis of IBS, which in turn has been associated with lower healthcare costs and utilisation, earlier initiation of treatment, reduced anxiety, and improved patient satisfaction [ 46 , 47 ]. Of course, the Rome diagnostic criteria should not be applied uncritically, as they are restrictive and exclude a substantial proportion of IBS patients, approximately 30% in our cohort. The recently published Rome V criteria represent an important update in this regard, as less frequent pain or discomfort is now sufficient for an IBS diagnosis, improving diagnostic performance [ 48 ]. Furthermore, the Rome Foundation stated in 2022, and reaffirmed on the occasion of the Rome V update, that the clinical setting does not require the strict duration and frequency criteria when diagnosing IBS [ 49 , 50 ]. In that perspective, the patients who do not fulfill the Rome IV criteria in the current study should be considered representing a true clinical IBS phenotype. Clinical stratification of IBS patients according to Rome IV positivity and stool pattern subtype in primary care after making a positive diagnosis of IBS might help GPs to establish an individualised therapeutic strategy and to optimise long‐term outcome. GP, general practitioner; IBS, irritable bowel syndrome; IBS‐C, constipation‐predominant IBS; IBS‐D, diarrhoea‐predominant IBS; IBS‐M, IBS with mixed bowel habits; IBS‐U, IBS unclassified. This study provides novel comprehensive information on important characteristics in a large number of primary care IBS patients in a pragmatic setting. However, the study has a number of limitations. Because of the cross‐sectional design, we could not determine causal relationships based on the results. Also, a randomised controlled trial population is inevitably subject to some degree of selection and participation bias. Nonetheless, we are confident that this post hoc analysis provides an accurate representation of the real‐world primary care IBS population, because: (1) treatment‐naïve patients were identified by trained GPs, (2) participants were informed that they could switch to the alternative intervention after trial completion, potentially reducing the impact of treatment preferences, (3) this post hoc analysis included all screened individuals, and (4) the DOMINO trial used only limited exclusion criteria, essential for preventing distortion of trial outcomes, while minimally compromising generalisability. A final limitation is that the baseline data collected were closely tied to the DOMINO trial endpoints, which meant that some general descriptive variables typically used to characterise a population cohort were not captured. To conclude, in a large primary care IBS cohort, the majority of patients fulfilled the Rome IV criteria, were subtyped as IBS‐D or IBS‐M and were characterised by moderate or severe IBS‐SSS. Rome+ and IBS‐M participants had a higher symptom severity, a lower quality of life and a more pronounced psychosocial comorbidity. Assessing these variables in newly diagnosed IBS patients might help GPs to establish an individualised therapeutic strategy and to optimise long‐term outcomes.

Irritable bowel syndrome (IBS) is a disorder of gut‐brain interaction (DGBI) defined by the Rome IV criteria as recurrent abdominal pain related to defecation and/or a change in stool form or frequency [ 1 ]. Worldwide, 4.1% of adults meet the Rome IV criteria for IBS, and older studies have shown that approximately 30% of patients presenting to general practitioners (GPs) with gastrointestinal complaints ultimately receive a diagnosis of IBS [ 2 , 3 ]. Based on predominant stool pattern, four IBS subtypes are distinguished: constipation‐predominant (IBS‐C), diarrhoea‐predominant (IBS‐D), mixed bowel habits (IBS‐M), and unclassified (IBS‐U) [ 1 ]. Clear risk factors for developing IBS include female sex, age under 50 years, previous acute enteric infection, and being affected by functional somatic syndromes, the latter partly explaining the overlap with other DGBIs [ 4 , 5 ]. IBS is associated with decreased quality of life, in such a way that patients would give up 10–15 years of their residual life expectancy for a permanent and immediate remedy [ 6 , 7 ]. Also, IBS is associated with a considerable psychological burden, including somatic symptom disorder, depression and anxiety, even in primary care [ 8 , 9 , 10 ]. Guidelines state that physicians should establish a positive diagnosis of IBS based on symptoms, provided there are no alarm signs or abnormalities on simple blood and stool tests. Hence, the majority of IBS patients are diagnosed and managed in a primary care setting, whereas only a minority of GPs, estimated at 7%–32% in Europe, would refer patients to a specialist before establishing an IBS diagnosis [ 11 , 12 ]. Despite this, IBS in primary care remains understudied. Most research has focused on the general population or on secondary and tertiary care cohorts, and the available primary‐care evidence is largely outdated, leaving important gaps in understanding how IBS presents in the clinical context where the majority of patients are actually seen. The DOMINO study, a pragmatic randomised open‐label parallel group trial, investigated the effect of otilonium bromide (OB) versus a simplified fermentable oligosaccharides, disaccharides, monosaccharides and polyols (FODMAP) lowering diet, provided by a mobile application, on IBS symptoms. In this study, 69 Belgian GPs recruited 483 newly diagnosed adult primary care IBS patients [ 13 ]. In the present manuscript, we describe a cross‐sectional post hoc analysis of this trial, evaluating the clinical and psychosocial characteristics of the primary care IBS cohort. The aim of this study was to improve our understanding of IBS characteristics in primary care.

J.T. has given scientific advice to Adare, AlfaWassermann, Arena, Bayer, Christian Hansen, Clasado, Danone, Devintec, Falk, FitForMe, Grünenthal, Ironwood, Janssen, Kiowa Kirin, Menarini, Mylan, Neurogastrx, Neutec, Novartis, Nutricia, Reckitt Benckiser, Ricordati, Shionogi, Takeda, Truvion, Tsumura, Zealand and Zeria pharmaceuticals, has received research support from Biohit, Shire, Sofar and Takeda, and has served on the Speaker Bureau for Abbott, Allergan, AstraZeneca, FitForMe, Janssen, Kyowa Kirin, Mayoly, Menarini, Mylan, Novartis, Schwabe Parmaceuticals, Takeda, Wellspect and Zeria. K.V.d.H. is supported by a junior postdoctoral fellowship of the Flanders Research Foundation (FWO Vlaanderen). F.C., B.B. and R.W. does not have conflicts of interest to declare.

Supporting Information S1