Type D Personality Is Associated With Depressive Symptoms and Clinical Activity in Inflammatory Bowel Disease

Sebastian Bruno Ulrich Jordi; Federica Botte; Brian Matthew Lang; Thomas Greuter; Niklas Krupka; Bianca Auschra; Philipp Schreiner; Michael Christian Sulz; Luc Biedermann; Roland von Känel; Gerhard Rogler; Stefan Begré; Benjamin Misselwitz


Aliment Pharmacol Ther. 2021;54(1):53-67. 

In This Article


Design of the Swiss IBD Cohort

We analysed the data of the Swiss IBD cohort study which recruits IBD patients nationwide in Switzerland. The Swiss IBD cohort study was started in 2006 as a prospective cohort study with yearly follow-ups.[26,27] Disease-related clinical data are assessed by physicians (physician questionnaire) at enrolment and follow-ups. Moreover, clinical, sociodemographic and psychosocial data are collected via patient questionnaires at enrolment and follow-ups. The detailed aims and methodology of the Swiss IBD cohort study are described elsewhere.[26,27]

Patient Characteristics

We exported sociodemographic and clinical variables acquired at enrolment and regular follow-ups (of up to 4250 days): Disease activity was measured using the Crohn's disease activity index (CDAI) or the modified Truelove and Witts activity index (MTWAI) for patients with CD and UC/IBDu, respectively. For intercomparable measures, CDAI or MTWAI score were Z-score normalised and combined
or cut-off values defining active disease were used (see below).

A set of binary descriptor variables for the IBD course was defined and reached if one of their defining criteria was observed: Complications included colon dysplasia, colorectal cancer, intestinal lymphoma, anaemia (not as an adverse event of medical therapies), deep vein thrombosis, pulmonary embolism, massive haemorrhage, nephrolithiasis, gallstones,[28] malabsorption syndrome, osteopenia/osteoporosis,[29] perforation or peritonitis. Extraintestinal manifestations comprised peripheral arthritis/arthralgia, uveitis/iritis, aphthous oral ulcers/stomatitis, erythema nodosum, pyoderma gangrenosum, ankylosing spondylitis/sacroiliitis and primary sclerosing cholangitis (PSC). The presence of any stenosis localised in the oesophagus, duodenum/jejunum, ileum, large bowel, rectum, or anus was noted as stenosis. The fistula variable consisted of IBD-related fistula and abscesses and anal fissure, whereas the variable surgery included any abdominal or fistula and abscess-related surgery. Systemic steroids comprised therapies with the following oral steroids with systemic activity: prednisone, prednisolone, deflazacort. TNF inhibitors referred to treatments with infliximab, certolizumab pegol or adalimumab (no patient with golimumab treatment was reported at the time of data export). The variable number of current therapies was the sum of all administered therapies during the respective time. Such therapies could be any of the following: any steroids, any antibiotics, 5-aminosalicylic acid, sulfasalazine, azathioprine, 6-mercaptopurine, methotrexate, cyclosporine, tacrolimus, infliximab, adalimumab, certolizumab pegol, cholestyramine, Escherichia coli Nissle (Mutaflor®), ursodeoxycholic acid, bisphosphonates, the probiotic VSL#3®, or any other medication that was noted by physicians. Smoking status was assessed binarily independent of the smoking quantity.

Outcome Measures

Type D was assessed once at enrolment via the validated Type D Scale-14 questionnaire for patients with age ≥16 years.[8,30] No paediatric patient was considered for this study. Subjects were classified as Type D when exceeding the cut-off of 10 for both subscales (negative affectivity ≥10 and social inhibition ≥10).[8] Only patients with known Type D status at enrolment were included in this study.

Depressive symptoms were measured at enrolment and each follow-up with the Hospital Anxiety and Depression Scale's (HADS) subscale for depression (HADS-D). The HADS is a valid instrument to measure depressive symptoms in the presence of somatic symptoms.[31] Following well-established criteria, depression was defined as HADS-D ≥11, indicating probable moderate or severe depression.[32–34] We used the terms depression and depressive symptoms with reference to this definition henceforward.

Active disease was defined by CDAI ≥150[35] or MTWAI ≥10,[36] respectively. Clinical IBD recurrence was measured according to published composite endpoints.[5,6] The first clinical composite endpoint[6] physician-reported f lare, n on-response to any administered therapy with consequent change in medication, new complication or new e xtraintestinal manifestation (FNCE) was defined as the occurrence of any one of these events after enrolment.

The second clinical composite endpoint[5,6] a ctive disease, physician-reported f lare, f istula, stenosis, s urgery or new s ystemic t herapy (AFFSST) was reached upon the presence of active disease (see definition above), a physician-reported flare, new fistula, new stenosis, new surgery, intake of systemic steroids and/or start of therapy with a new TNF inhibitor (this also includes a change to another TNF inhibitor) after enrolment.

For either composite endpoint, we counted the first occurrence of at least one criterion after enrolment as clinical recurrence.

For time-to-event analyses, all endpoints were coded with right censoring. We defined time intervals tn−1 − tn where tn corresponds to the nth follow-up visit when an event was documented. Thereby, we described the time intervals during which events had occurred [tn−1, tn] and then assumed an event's occurrence at tn−1 + (tn − tn−1)/2 on average.

Association Between Type D and Depressive Symptoms at Enrolment

We tested for associations between Type D and depressive symptoms by implementing regression models. Based on clinical expertise we chose the following control variables (set 1) for an adjusted model: sex, diagnosis, time since diagnosis, age, body mass index, smoking status, daily alcohol consumption, disease activity (standardised CDAI/MTWAI), disease-related surgery prior to enrolment, systemic steroids, TNF inhibitors, number of current therapies. To obtain a more parsimonious secondary model we applied automated variable selection. In this selection process, the optimal model was calculated based on the Akaike information criterion using the glmulti R package[37] version

Time-to-event Analyses

We distinguished between occurrence and new occurrence of events. Occurrence refers to the presence of a certain variable value independent of its antecedent value, while new occurrence refers only to the presence of a certain variable value following at least one measurement with the negative state of this variable.

To confirm cross-sectional associations between Type D and depressive symptoms, we performed time-to-event analyses using the Survival R package version 2.38.[38] We implemented Cox proportional hazards models with the first occurrence of HADS ≥11 in a follow-up examination as an endpoint. A multivariable model included the variables of set 1 (described above).

Deterioration of the IBD course was defined by the first occurrence of active disease (CDAI ≥150/MTWAI ≥10), the first new occurrence of any extraintestinal manifestation or the first adverse event according to a composite endpoint definition (see above) at follow-up. We obtained adjusted P values (q values or simply q) by applying the Bonferroni correction method for multiple testing (q = P × n); for these analyses, we corrected for the testing of four endpoints (q = P × 4). To assess the impact of Type D and the co-occurrence of Type D and depressive symptoms on these endpoints, we implemented Cox proportional hazards models. For these models, we coded depressive symptoms as a time-varying variable that reflects a patient's development over time.[39] We chose a conservative coding approach to ensure an unambiguous time sequence: only if a recording of depressive symptoms preceded (at the previous yearly follow-up assessment) observed disease recurrence, an association was assumed. For these models, q values were calculated by correction for 12 tests (q = P × 12).

For all multivariable time-to-event analyses with clinical endpoints, we adapted the set of control variables by removing potential proxy-variables for clinical endpoints; set 2: sex, IBD diagnosis, time since IBD diagnosis, age, body mass index, disease-related surgery prior to enrolment, smoking status, daily alcohol consumption. All calculations were performed using R[40] version 3.6.1.