The Effect of Antidepressants on the Course of Inflammatory …downloads.hindawi.com/journals/cjgh/2018/2047242.pdf · 2019-07-30 · ResearchArticle The Effect of Antidepressants

Research ArticleThe Effect of Antidepressants on the Course ofInflammatory Bowel Disease

Barry J. Hall ,1 P. John Hamlin,1,2 David J. Gracie,1,2 and Alexander C. Ford1,2

1Leeds Gastroenterology Institute, St. James’s University Hospital, Leeds, UK2Leeds Institute of Biomedical and Clinical Sciences, University of Leeds, Leeds, UK

Correspondence should be addressed to Barry J. Hall; [email protected]

Received 15 May 2018; Accepted 17 August 2018; Published 9 September 2018

Academic Editor: Kevork M. Peltekian

Copyright © 2018 Barry J. Hall et al. This is an open access article distributed under the Creative Commons Attribution License,which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Background and Aims.Moodmay have an important role in the natural history of inflammatory bowel disease (IBD). However, theimpact of antidepressant use on prognosis is unknown. We aimed to address this in a longitudinal study in a referral population.Methods.We collected demographic data, clinical disease activity andmood using validated questionnaires, and antidepressant useat baseline. Longitudinal disease activity was defined by disease flare or need for glucocorticosteroids, escalation ofmedical therapy,hospitalisation, or intestinal resection. We compared rates of these over a minimum period of 2 years according to antidepressantuse at baseline. Results. In total, 331 patients provided complete data, of whom 54 (15.8%) were taking an antidepressant at studyentry.Older age, female gender, and abnormalmood scoreswere associatedwith antidepressant use.During longitudinal follow-up,there was a trend towards lower rates of any of the four endpoints of IBD activity of interest in patients with abnormal anxiety scoresat baseline and who were receiving an antidepressant (42.3% versus 64.6%, P = 0.05). Based on univariate Cox regression analysis,there was a trend towards lower rates of escalation of medical therapy among patients receiving antidepressants at baseline (hazardratio (HR) = 0.59; 95% confidence interval (CI) 0.35-1.00, P = 0.05). None of the differences observed persisted after multivariateCox regression.Conclusions. Antidepressants may have some beneficial effects on the natural history of IBD, but larger studies withlonger follow-up are required. Whether these effects are limited to patients with abnormal mood remains uncertain.

1. Introduction

Crohn’s disease (CD) and ulcerative colitis (UC) are chronicinflammatory organic disorders of the gastrointestinal (GI)tract of unknown aetiology, which may arise from dysreg-ulation of both the innate and adaptive immune systems,leading to persistent and damaging inflammation of the intes-tine. Conventional therapies for inflammatory bowel dis-ease (IBD) include glucocorticosteroids, 5-aminosalicylates,immunosuppressants, and biological therapies [1–7], with theaim of treatment being to induce remission of GI inflamma-tion and prevent future disease relapse.

Patientswith IBDappear to be at increased risk of exhibit-ing mood disorders, such as anxiety or depression, comparedwith healthy individuals [8–12]. However, whether this is aconsequence of, or an etiological factor in, GI inflammationremains uncertain. In people with functional GI disorders,such as irritable bowel syndrome, longitudinal studies suggestthat there is an increased likelihood of asymptomatic people

who demonstrate anxiety or depression at baseline develop-ing GI symptoms de novo [13]. This raises the possibility thatcoexistent anxiety or depression in patients with IBD, partic-ularly if unrecognised or untreated, may have a deleteriouseffect on the natural history of IBD.

Evidence to support a role of abnormal mood in sub-sequent IBD activity comes mainly from animal models.Studies have demonstrated that inmousemodels of quiescentcolitis, inducing depression leads to reactivation ofGI inflam-mation [14] but that this can be prevented by preadminis-tration of antidepressants [15]. There is also limited evidencein humans to suggest that acute psychological stress inducesproinflammatory cytokine production in both the serum andthe mucosa of IBD patients [16]. Studies that have examinedthe effect of psychological comorbidity on the natural historyof IBD suggest that patients with coexistent mood disordershave higher rates of disease relapse, escalation of therapy,and IBD-related surgery than patients with IBD withoutpsychological disorders [17–20]. Despite this, the benefit of

HindawiCanadian Journal of Gastroenterology and HepatologyVolume 2018, Article ID 2047242, 11 pageshttps://doi.org/10.1155/2018/2047242

2 Canadian Journal of Gastroenterology and Hepatology

treatments, such as psychological therapies or antidepres-sants, which target mood abnormalities in patients with IBDis uncertain. A systematic review and meta-analysis of 14randomised controlled trials (RCTs) of cognitive behaviouraltherapy, hypnotherapy, and other psychological interventionsfailed to demonstrate any effect on subsequent disease activity[21]. However, none of the included trials recruited onlypatients with psychological comorbidity. Another systematicreview examining the effect of antidepressants on the courseof IBD identified 15 studies [22], but the majority of thesewere case series or case reports. There were few longitudinalstudies and only one small pilot RCT of fluoxetine in CD [23],which appeared to be of no benefit in terms of maintenanceof disease remission.

This is an important issue, as patients with quiescent IBDmay be at a lower risk of subsequent relapse of disease activityif coexisting mood disorders are identified and treated. Thiscould lead to a more benign disease course, with fewerepisodes of flares of disease activity and reductions in treat-ment escalation, hospitalisation, or surgery. There is, there-fore, a need for epidemiological studies that examine theeffects of antidepressants on the course of IBD in a settingthat is generalisable to the type of patients with IBD seen insecondary care. We have therefore conducted a longitudinalfollow-up study examining these issues in a large numberof patients with IBD. Our a priori hypothesis was that therewould be lower rates of, and longer time of onset to, objectivemarkers of subsequent disease activity among those patientsprescribed antidepressants at baseline.

2. Materials and Methods

2.1. Participants and Setting. Individuals recruited into aprevious cross-sectional survey were included in this longitu-dinal follow-up study [24, 25]. All patients had an establishedradiological, histological, or endoscopic diagnosis of CD orUC and were aged ≥16 years at the time of baseline recruit-ment. Exclusion criteria were an inability to understandwritten English, a diagnosis of IBD-unclassified, and anyonewith an end ileostomy or colostomy, due to the difficultiesin assessing disease activity indices in these patients. Partici-pants were followed up after aminimumperiod of 2 years hadelapsed from initial recruitment. The longitudinal follow-upstudy was approved by the local research ethics committee inSeptember 2014 (REC ref: 12/YH/0443), and data collectioncontinued until June 2017. Study findings were reported inaccordance with the STROBE guidelines for observationalstudies [26].

2.2. Data Collection and Synthesis. Date of recruitment intothe original cross-sectional survey, demographic data, type ofIBD, medication use for both IBD and non-IBD-related dis-ease, and data concerning anxiety, depression, somatisation,and quality of life data were recorded at baseline, as describedin the original cross-sectional survey [24, 25].

2.2.1. Definition of Disease Activity, Anxiety or Depression,and Somatisation. At baseline assessment, a combination of

disease activity indices (the Harvey-Bradshaw index (HBI)for CD [27] and the simple clinical colitis activity index(SCCAI) for UC [28]) were utilised. Clinical remission wasdefined as a score <5 for both CD and UC, as has beenrecommended by previous investigators [29, 30].

Anxiety and depression data were collected using thehospital anxiety and depression scale (HADS) [31]. This 14-item questionnaire consists of seven questions screening forthe presence of anxiety symptoms and seven for depressionsymptoms, with a 4-point response for each item, rangingfrom 0 to 3. The total HADS score ranges from a minimumof 0 to a maximum of 21 for both anxiety and depression.Severity was categorised, according to total HADS score,into normal (total HADS depression or anxiety score 0-7),borderline normal (8-10), and abnormal (≥11) as previouslyrecommended [31].

Somatisation data were collected using the patient healthquestionnaire-15 (PHQ-15), which is derived from the val-idated full PHQ [32]. Individuals were asked to rate theseverity of 15 predefined symptoms as “not bothered at all”(scored as 0), “bothered a little” (scored as 1), or “bothered alot” (scored as 2). Therefore the total PHQ-15 score rangesfrom a minimum of 0 to a maximum of 30. Somatisationseverity was categorised, using the total PHQ-15 score, intohigh (total PHQ-15≥15), medium (10-14), low (5-9), andmin-imal (≤4) levels of somatisation severity, again as previouslyrecommended [32].

2.2.2. Antidepressant Use. Use of antidepressants was cap-tured at the baseline visit. We recorded the regular prescrip-tion of any of the following drugs at study entry: selec-tive serotonin reuptake inhibitors (SSRIs) including fluoxe-tine, fluvoxamine, paroxetine, sertraline, citalopram, escitalo-pram; tricyclic antidepressants (TCAs) including amitripty-line, nortriptyline, imipramine, clomipramine, desipramine,doxepin, or dosulepin; serotonin and noradrenaline reuptakeinhibitors (SNRIs) including venlafaxine and duloxetine;tetracyclic antidepressants including mirtazapine; and atyp-ical antidepressants, such as trazodone.

2.2.3. Longitudinal Objective Assessment of IBD Activity.Objective assessment of disease activity during longitudinalfollow-up was made by detailed case note review by a soleinvestigator (DJG), blinded to the baseline questionnairedata. The case notes of each of the patients included at base-line were assessed for the following four clinical endpoints,with the date of each endpoint recorded, where applicable:documented glucocorticosteroid prescription or a flare ofdisease activity identified by physician’s global assessment;escalation of medical therapy due to uncontrolled diseaseactivity, as judged by a physician’s global assessment; hospi-talisation secondary to objectively confirmed IBD activity;and intestinal resection. All these decisions were made as perusual care, as this was a noninterventional study conductedin routine clinical practice. Escalation of medical therapy inresponse to therapeutic drug monitoring, but in the absenceof inflammatory activity, was not included as an endpoint,nor was surgical intervention for isolated perianal Crohn’sdisease.

Canadian Journal of Gastroenterology and Hepatology 3

2.3. Statistical Analysis. Baseline demographic, disease-relat-ed, and psychological data for all participants were comparedbetween those taking and not taking antidepressants at studyentry. Due to multiple comparisons we considered a 2-tailed P value of <0.01 to be statistically significant for theseanalyses. Proportions of patients experiencing the occurrenceof one, or any, of the four objective markers of IBD activityduring longitudinal follow-up were then compared betweenthose patients taking an antidepressant at baseline and thosewho were not. We planned to perform subgroups analysesaccording to type of antidepressant used (SSRI, TCA, SNRI,or tetracyclic) as well as presence or absence of anxietyand/or depression at baseline, to assess whether these hadany impact on the association between antidepressant use andIBD activity during longitudinal follow-up. We comparedproportions in all these analyses using a𝜒2 test for categoricalvariables and an independent samples t-test for continuousdata.

We then performed univariate Cox regression analysis toexamine the effect of antidepressant use on IBD activity dur-ing longitudinal follow-up for all patients, and according topresence or absence of anxiety and/or depression at base-line. Independent predictors of the occurrence of any ofthe objective disease activity outcomes of interest werethen determined by performing multivariate Cox regressionanalysis to control for antidepressant use, as well as all otherbaseline data. The results of these analyses were expressed ashazard ratios (HR) with 95% confidence intervals (CI). Asour hypothesis that there would be lower rates of objectivemarkers of IBD activity among those patients prescribedantidepressants at baseline was defined a priori, for theseanalyses we considered a 2-tailed P value of <0.05 to be sta-tistically significant. All statistical analyses were performedusing SPSS forWindows version 22.0 (SPSS Inc., Chicago, IL,USA).

3. Results

In total, 54 (15.8%) of 342 patients were taking an antide-pressant at study entry. Of these, 32 (59.3%) were taking anSSRI, 17 (31.5%) a TCA, two (3.7%) an SNRI, two (3.7%) bothan SSRI and a TCA, and one (1.9%) trazodone. Older ageand female gender were both associated with antidepressantuse, and there was a trend towards a higher body mass index(BMI) among those using antidepressants (Table 1). Patientswith abnormal HADS anxiety or depression scores and thosewith high somatisation levels were more likely to be taking anantidepressant at baseline (P < 0.001).

3.1. Effect of Antidepressant Use on Subsequent IBD Activity.Of these 342 patients, 331 (96.8%) provided complete follow-up data for longitudinal IBD activity after case note reviewand were included in subsequent analyses. In all, 142 (42.9%)of 331 patients had a flare of disease activity or required glu-cocorticosteroid therapy during the 2-year follow-up period.Furthermore, 137 (41.4%) patients underwent escalation ofmedical therapy in response to uncontrolled disease activ-ity. In terms of hospitalisation and intestinal resection, 43(13.0%) and 19 (0.6%) patients experienced these endpoints

Hazard Function for patterns 1 -2

Cum

Haz

ard

Total follow-up for escalation

Antidepressantuse yes or no

noyes

1.2

1.0

0.8

0.6

0.4

0.2

0.0

0 200 400 600 800 1000 1200

Figure 1: Cumulative hazard for escalation of medical therapy inresponse to uncontrolled IBD between patients receiving or notreceiving antidepressants.

of interest, respectively. There were no differences in pro-portions of patients experiencing a flare of disease activityor glucocorticosteroid use, need for escalation of therapy,hospitalisation, intestinal resection, or any of these endpointsaccording to use of antidepressants at baseline (Table 2).Limiting our analyses for flare of disease activity or requiringglucocorticosteroid therapy or escalation of medical therapyto only those who were in clinical remission at baseline didnot affect our findings—9 (39.1%) of 23 patients receivingantidepressants experienced a flare of disease activity orneeded a prescription for glucocorticosteroids versus 49(32.0%) of 153 not receiving (P = 0.50); 7 (30.4%) of 23patients receiving antidepressants required escalation of ther-apy versus 60 (39.2%) of 153 nor receiving (P = 0.42).

Based on univariate Cox regression analysis, there was atrend towards lower rates of escalation of medical therapyamong patients receiving antidepressants at baseline, com-pared with those who were not (HR = 0.59; 95% CI 0.35 to1.00, P = 0.05) (Table 3, Figure 1), but this was no longer thecase based on multivariate Cox regression controlling for allbaseline data.

We performed subgroup analyses according to pres-ence or absence of anxiety and/or depression at baseline(Table 2). There was a trend towards lower rates of escalationof therapy among those with either abnormal anxiety ordepression scores at baseline and who were receiving anantidepressant, compared with those with either abnormalanxiety or depression scores at baseline but who were notreceiving an antidepressant (25.5% versus 46.6%, P = 0.06).There were significantly lower rates of escalation of therapyamong patients with abnormal anxiety scores at baseline andwho were receiving an antidepressant, compared with thosewith abnormal anxiety scores at baseline but who were notreceiving an antidepressant (23.1% versus 47.7%, P = 0.03).Finally there was a trend towards lower rates of any of the

4 Canadian Journal of Gastroenterology and Hepatology

Table1:Ba

selin

edem

ograph

ic,dise

ase-related,andpsycho

logicalcharacteristicso

fpatientsw

ithIBDaccordingto

useo

fantidepressants.

Prescribed

Antidepressan

tsat

Baselin

e(n

=54

)

Not

prescribed

Antidepressan

tsat

Baselin

e(n

=28

8)P

value∗

Meanagein

years(

SD)

47.1(13.8)

39.6(16.2)

0.00

1Meandu

ratio

nof

follo

w-upin

days

(SD

)for

each

endp

oint

Glucocorticosteroidprescriptio

nor

flare

ofdiseasea

ctivity

648.3(409)

502.9(361.8)

0.02

Escalatio

nof

medicaltherapyinrespon

seto

uncontrolleddiseasea

ctivity

664.0(399.1)

531.0

(345.3)

0.03

Hospitalisationdu

etodiseasea

ctivity

(%)

841.4

(284.1)

733.2(252.1)

0.01

Intestinalresectio

n(%

)861.4

(265.6)

770.7(218.1)

0.02

Femaleg

ender(

%)

47(87.0

)158(54.9)

<0.00

1Cau

casian

ethn

icity

(%)

51(96.2)

267(93.7)

0.47

Married

orco-habiting

(%)

27(50.9)

165(58.3)

0.32

Uni

versity

/postgradu

ate(

%)

16(29.6

)102(36.0)

0.37

Mean

BMI(

SD)

28.0(6.2)

25.6(5.3)

0.01

Toba

ccouser

(%)

14(25.9)

48(16.8)

0.11

Alcoh

oluser

(%)

31(57.4

)200(69.9

)0.07

Cro

hn’sdisease(

%)

20(37.0

)118

(41.0

)0.59

5-A

SAuse(

%)

27(50.0)

124(43.1)

0.35

Immun

osup

pressant

use(

%)

16(29.6

)117

(40.6)

0.13

Biolog

icuse(

%)

6(11.1)

63(21.9

)0.07

Glucocorticosteroid

use(

%)

5(9.3)

26(9.0)

0.96

HBI

/SCC

AI<

5(%

)23

(45.1)

159(58.7)

0.07

HA

DSan

xiet

yscores

atba

selin

e(%

)Normal

16(29.6

)167(58.6)

Borderlin

eabn

ormal

10(18.5)

10(18.5)

Abno

rmal

28(51.9

)69

(24.2)

<0.00

1H

AD

Sdepression

scores

atba

selin

e(%

)Normal

29(53.7)

230(80.1)

Borderlin

eabn

ormal

11(20.4)

35(12.2)

Abno

rmal

14(25.9)

22(7.7)

<0.00

1PH

Q-1

5som

atisationscores

atba

selin

e(%

)Mild

0(0)

50(18.8)

Low

8(15.7)

83(31.2

)Medium

20(39.2

)78

(29.3

)High

23(45.1)

55(20.7)

<0.00

1∗Independ

entsam

ples

t-testfor

continuo

usdataand𝜒2forc

omparis

onof

categoric

aldata.

Canadian Journal of Gastroenterology and Hepatology 5

Table2:Eff

ecto

fantidepressantu

seon

subsequent

developm

ento

fIBD

activ

ityandaccordingto

anxietyanddepressio

nscores

atbaselin

e.

Antidepressan

tUse

:All

Patie

nts

Antidepressan

tUse

:Normal

Anx

iety

and

Depression

Scores

atBa

selin

e

Antidepressan

tUse

:Abn

ormal

Anx

iety

orDepression

Scores

atBa

selin

e

Antidepressan

tUse

:Abn

ormal

Anx

iety

Scores

atBa

selin

e

Antidepressan

tUse

:Abn

ormal

Depression

Scores

atBa

selin

e

No

Yes

Pvalue

No

Yes

Pvalue

No

Yes

Pvalue

No

Yes

Pvalue

No

Yes

Pvalue

Glucocorticosteroid

prescriptio

nor

flare

ofdiseasea

ctivity

(%)

121/2

79(43.4)

21/52

(40.4)

0.69

82/203

(40.4)

10/25

(40.0)

0.97

38/73

(52.1)

11/27

(40.1)

0.32

35/65

(53.8)

10/26

(38.5)

0.19

9/21

(42.9)

6/14

(42.9)

1.00

Escalatio

nof

med

icaltherap

yin

respon

seto

uncontrolle

ddisease

activ

ity(%

)

120/279

(43.0)

17/52

(32.7)

0.17

85/203

(41.9

)10/25

(40.0)

0.86

34/73

(46.6)

7/27

(25.9)

0.06

31/65

(47.7

)6/26

(23.1)

0.03

8/21

(38.1)

5/14

(35.7)

0.89

Hospitalisationdu

eto

diseasea

ctivity

(%)

38/279

(13.6)

5/52

(9.6)

0.43

26/203

(12.8)

4/25

(16.0)

0.66

12/73

(16.4)

1/27

(3.7)

0.09

11/65

(16.9)

1/26

(3.8)

0.10

2/21

(9.5)

0/14 (0)

0.23

Intestinalresection

(%)

15/279

(5.4)

4/52

(7.7)

0.51

8/203

(3.9)

3/25

(12.0)

0.08

7/73

(9.6)

1/27

(3.7)

0.34

6/65

(9.2)

1/26

(3.8)

0.38

2/21

(9.5)

0/14 (0)

0.23

Any

adverseo

utcome

(%)

148/279

(53.0)

24/52

(46.2)

0.36

102/203

(50.2)

12/25

(48.0)

0.83

45/73

(61.6

)12/27

(44.4)

0.12

42/65

(64.6)

11/26

(42.3)

0.05

10/21

(47.6

)6/14

(42.9)

0.78

6 Canadian Journal of Gastroenterology and Hepatology

Table3:Eff

ecto

fantidepressanto

rSSR

Iuse

onsubsequent

IBDactiv

itybasedon

univariatecoxregressio

nanalysis.

Glucocorticosteroid

prescriptio

nor

flare

ofdiseasea

ctivity

Escalatio

nof

Med

ical

�erap

yin

Respon

seto

Uncon

trolled

Disease

Activity

Hospitalisationdu

eto

Disease

Activity

Intestinalresection

Hazardratio

(95%

CI)

Pvalue

Hazardratio

(95%

CI)

Pvalue

Hazardratio

(95%

CI)

Pvalue

Hazardratio

(95%

CI)

Pvalue

Antidepressan

tuse:

allp

atients

0.76

(0.47-1.2

2)0.25

0.59

(0.35-1.0

0)0.05

0.59

(0.231.50)

0.26

1.15

(0.37

3.53)

0.81

Antidepressan

tuse:no

rmal

anxiet

yan

ddepression

scores

atba

selin

e

0.77

(0.38-1.5

4)0.77

0.72

(0.36-1.4

4)0.35

1.01

(0.352.93)

0.98

2.15

(0.558.50)

0.27

Antidepressan

tuse:abno

rmal

anxiet

yor

depression

scores

atba

selin

e

0.63

(0.32

-1.23)

0.17

0.47

(0.21-1.05)

0.07

0.20

(0.031.57)

0.13

0.37

(0.053.04)

0.36

SSRI

use:all

patie

nts

0.74

(0.41-1.31

)0.30

0.47

(0.24-0.93)

0.03

0.35

(0.081.46)

0.15

0.86

(0.19

3.85)

0.84

SSRI

use:no

rmal

anxiet

yan

ddepression

scores

atba

selin

e

0.76

(0.31-1.91)

0.56

0.49

(0.18

-1.37

)0.18

0.87

(0.203.75)

0.86

2.39

(0.4712.2)

0.29

SSRI

use:

abno

rmalan

xiet

yor

depression

scores

atba

selin

e

0.58

(0.27-1.2

6)0.17

0.43

(0.17-1.11)

0.08

0.03

(0.0010.1)

0.24

0.03

(0.0073.9)

0.39

Canadian Journal of Gastroenterology and Hepatology 7

Hazard Function for patterns 1 -2

Cum

Haz

ard

Total follow-up for escalation

Antidepressantuse yes or no

noyes

1.0

0.8

0.6

0.4

0.2

0.0

0 200 400 600 800 1000

Figure 2: Cumulative hazard for escalation of medical therapyin response to uncontrolled IBD between patients with abnormalanxiety or depression scores at baseline receiving or not receivingantidepressants.

four endpoints of IBD activity of interest in patients withabnormal anxiety scores at baseline and who were receiv-ing an antidepressant, compared with those with abnormalanxiety scores at baseline but who were not receiving anantidepressant (42.3% versus 64.6%, P = 0.05). Based onunivariate Cox regression analysis, again there was a trendtowards lower rates of escalation of medical therapy amongpatients with either abnormal anxiety or depression scores atbaseline who were receiving antidepressants, compared withthose with either abnormal anxiety or depression scores atbaseline but who were not (HR = 0.47; 95% CI 0.21 to 1.05,P = 0.07) (Table 3) (Figure 2). However, this was no longerthe case based on multivariate Cox regression.

3.2. Effect of SSRI Use on Subsequent IBD Activity. In total,33 (61.1%) of all 54 patients taking an antidepressant atbaseline were receiving an SSRI. There were no significantdifferences in the proportions of patients prescribed anSSRI experiencing one, or any, of the endpoints of interest,compared with those who were not (Table 4).

Based on univariate Cox regression analysis, there wasa significantly lower rate of escalation of medical therapyamong patients receiving SSRIs at baseline, compared withthose whowere not (HR= 0.47; 95%CI 0.24 to 0.93, P = 0.03)(Table 3) (Figure 3). This was no longer observed based onmultivariate Cox regression.

Due to a reduced number of individuals taking SSRIs wewere unable to perform subgroup analyses according to thepresence of only abnormal anxiety scores, or only abnormaldepression scores, at baseline. In the remaining analyses,there were trends towards higher rates of intestinal resectionamong patients with normal anxiety and depression scoresat baseline who were taking SSRIs (15.4% versus 3.9%, P= 0.06) and lower rates of hospitalisation among patientswith either abnormal anxiety or depression scores at baseline

Hazard Function for patterns 1 -2

Cum

Haz

ard

Total follow-up for escalation

SSRIuse

noyes

1.2

1.0

0.8

0.6

0.4

0.2

0.0

0 200 400 600 800 1000 1200

Figure 3: Cumulative hazard for escalation of medical therapy inresponse to uncontrolled IBD between patients receiving or notreceiving SSRIs.

receiving SSRIs (0% versus 16.4%, P = 0.05) (Table 4). Basedon both univariate (Table 3) and multivariate Cox regressionanalysis, there were no significant differences in rates offlare or glucocorticoid prescription, escalation of medicaltherapy, hospitalisation, or intestinal resection among SSRIusers according to presence or absence of abnormal anxietyor depression scores at baseline.

4. Discussion

Results from univariate Cox regression analysis in thislongitudinal study, adjusted for total duration of follow-upamong all individuals, suggest a trend towards lower ratesof subsequent escalation of medical therapy among patientswith IBD receiving antidepressants at baseline, and thispersisted when only those patients with abnormal anxietyor depression scores at study entry were considered in theanalysis. Similar findings were observed when only thoseusing SSRIs at baseline were considered in the analysis.However, none of these associations remained statisticallysignificant based on multivariate Cox regression analysis.Our findings, although interesting, should be consideredas preliminary only, and larger studies with longer follow-up, or rigorous RCTs, which are powered to examine thisissue specifically, will be required before the true effect ofantidepressant use on objective disease endpoints in patientswith IBD is known.

We obtained longitudinal disease activity outcome datafor a cohort of 331 patients with IBD. The observationaldesign and the use of a secondary care population mean thatour findings are likely to be generalisable to the majority ofpatients with IBD seen outside specialist centres. Our use ofvalidated questionnaires to assess for the presence of anxiety[31], depression [31], or somatisation [32] at baseline is afurther strength. We controlled for differences in duration

8 Canadian Journal of Gastroenterology and Hepatology

Table4:Eff

ecto

fSSR

Iuse

onsubsequent

developm

ento

fIBD

activ

ityandaccordingto

anxietyanddepressio

nscores

atbaselin

e.

SSRI

Use

:All

Patie

nts

SSRI

Use

:Normal

Anx

iety

and

Depression

Scores

atBa

selin

e

SSRI

use:

Abn

ormal

Anx

iety

orDepression

Scores

atBa

selin

eNo

Yes

PVa

lue

No

Yes

Pvalue

No

Yes

Pvalue

Glucocorticosteroid

prescriptio

nor

flare

ofdiseasea

ctivity

(%)

121/2

79(43.4)

14/33

(42.4)

0.92

82/203

(40.4)

6/13

(46.2)

0.68

38/73

(52.1)

8/20

(40.0)

0.34

Escalatio

nof

med

icaltherap

yinrespon

seto

uncontrolle

ddisease

activ

ity(%

)120/279

(43.0)

10/33

(30.3)

0.16

85/203

(41.9

)5/13

(38.5)

0.81

34/73

(46.6)

5/20

(25.0)

0.08

Hospitalisationdu

etodiseasea

ctivity

(%)

38/279

(13.6)

2/33

(6.1)

0.22

26/203

(12.8)

2/13

(15.4)

0.79

12/73

(16.4)

0/20 (0)

0.05

Intestinalresection

(%)

15/279

(5.4)

2/33

(6.1)

0.87

8/203

(3.9)

2/13

(15.4)

0.06

7/73

(9.6)

0/20 (0)

0.15

Any

adverseo

utcome(

%)

148/279

(53.0)

14/33

(42.4)

0.25

102/203

(50.2)

6/13

(46.2)

0.78

45/73

(61.6

)8/20

(40.0)

0.08

Canadian Journal of Gastroenterology and Hepatology 9

of follow-up between those receiving, or not receiving,antidepressants at baseline and all other baseline data, usingmultivariate Cox regression. Finally, objective longitudinaldisease activity assessment data were collected blinded tobaseline disease activity and psychological data, thereforeeliminating researcher confirmation bias.

Weaknesses include the sample size and length of follow-up. Based on multivariate Cox regression analysis, there wereno statistically significant differences observed in rates ofany of the objective markers of disease activity accordingto antidepressant use at baseline, despite the fact that theabsolute proportion of patients reaching these endpointsweregenerally lower among those taking antidepressants at base-line, particularly among those with abnormal anxiety and/ordepression scores at baseline. This suggests that if our samplesize had been larger, or the duration of follow-up longer,some of these comparisons may have become statisticallysignificant. Alternatively, there may be a true beneficial effectthat was lost when other factors were controlled for in ourmultivariate analysis.This could be due to either confoundingor a loss of time contributed to the analysis from individualswith missing data for these demographic characteristics.Our assessment of psychological well-being was based onquestionnaire responses at a single point in time, and ouruse of HADS scores as a marker of anxiety or depressioncould be criticised as, although these are widely used, theirpsychometric properties have been challenged [33]. Althoughwe collected disease activity endpoints based on objectivelydefined criteria, there is the possibility that some of theseendpoints, such as escalation of therapy, were reached basedon symptom reporting alone, rather than true inflammatoryactivity. Due to the low event rate for some of the longitudinaldisease activity outcomes, we chose not to present data onIBD patients dichotomised into those with UC and CD, oraccording to the use of antidepressant subclasses, other thanSSRIs. Finally, use of antidepressants at baseline was classifiedaccording to a documented current prescription for the drugsof interest, but whether patients were actually adherent totheir antidepressant therapy is uncertain. In addition, wedid not record the discontinuation or commencement ofantidepressant drugs during follow-up and therefore cannotexclude some misclassification of patients.

Due to the chronicity of their disease and the requiredshifts in coping and self-management skills over time, it is notsurprising that patients with IBD are at increased risk ofmen-tal health disorders, yet previous studies have demonstratedthat these coexistent psychological problems are often under-treated [34]. Undertreatment of mental health disorders canlead to long-term disability and a stigma against seekingtreatment [35, 36]. Most IBD specialists have little formaltraining in the detection andmanagement of such psycholog-ical comorbidities. This lack of training, combined with thepotential hesitancy of many patients to report psychologicalsymptoms, may influence therapy options. Although ourstudy demonstrated that patients with abnormal anxiety ordepression scores and those with high somatisation levelswere more likely to be taking an antidepressant at baseline,the prevalence of abnormal anxiety scores, abnormal depres-sion scores, and high somatisation scores in patients not

receiving antidepressant therapy was also relatively high at24.2%, 7.7%, and 20.7%, respectively.

Decreasing potential long-term sequelae by minimisingthe inflammatory burden in IBD patients is an importanttreatment strategy. There is some evidence, from studies inboth animals and humans with IBD, to suggest that psycho-logical comorbidity can trigger inflammatory activity [14, 16].Furthermore, patients with IBD are at increased risk of psy-chological comorbidity, particularly depression and anxiety[37].Thus, pharmacological therapy for mood disorders mayhave the potential to alter the natural history of CD and UC.Limited data, to date, have suggested improvements inmood,inflammatory activity, and quality of life with antidepressanttherapy [23, 38]. Our own longitudinal follow-up data didnot show any statistically significant differences in the pro-portion of patients experiencing endpoints consistent withinflammatory activity, according to use of antidepressants atbaseline. However, the absolute proportions reaching theseendpoints were generally lower among those prescribedantidepressants, particularly when only those with abnormalanxiety or depression scores were considered in the analysis.

Mechanisms of any beneficial effect of antidepressantdrugs on the course of IBD are uncertain. Some antidepres-sants, like amitriptyline, appear to have both pain modify-ing properties [39] and direct effects on proinflammatorycytokines that may arise via actions on the nuclear factor-𝜅B and nitric oxide pathways, which are implicated in thepathogenesis of IBD [40]. The former may explain the trendtowards a reduction in escalation of therapy we observed,although given that this was also observed when only patientstaking SSRIs, which have less in the way of analgesiceffects [39], were considered, this seems unlikely. Therefore,antidepressant therapy may indeed have a beneficial effecton inflammatory activity. However, differentiating the trueinflammatory burden in a symptomatic patient with IBD andcoexistent mood disorder can be difficult, as the availablesymptom scoring systems cannot confirm the true origin ofsymptoms, and patients with IBD with psychological comor-bidity may be more likely to report GI symptoms in general[24, 25]. As a result patients with mood disorders and whowere on antidepressants in our study may potentially havebeen less likely to be offered, or indeed accept, a glucocor-ticosteroid prescription, a step-up in therapy, hospitalisation,or surgery.

In summary, our longitudinal follow-up study demon-strated lower rates of some objective markers of disease activ-ity among patients with IBD receiving antidepressants andSSRIs, at baseline, particularly among those with abnormalanxiety or depression scores at baseline. However, none werestatistically significant based on multivariate Cox regression.Further observational studies or RCTs with longer follow-up, which are powered to examine this issue specifically, willbe required before the true effect of antidepressant use ondisease activity in patients with CD and UC is known.

Abbreviations

BMI: Body mass indexCD: Crohn’s disease

10 Canadian Journal of Gastroenterology and Hepatology

CI: Confidence intervalGI: GastrointestinalHADS: Hospital anxiety and depression scaleHBI: Harvey-Bradshaw indexHR: Hazard ratioIBD: Inflammatory bowel diseasePHQ-15: Patient health questionnaire-15RCT: Randomised controlled trialSCCAI: Simple clinical colitis activity indexSNRI: Serotonin and noradrenaline reuptake

inhibitorSSRI: Selective serotonin reuptake inhibitorTCA: Tricyclic antidepressantUC: Ulcerative colitis.

Data Availability

The data used to support the findings of this study are avail-able from the corresponding author upon request.

Disclosure

David J. Gracie and Alexander C. Ford are joint last authors,and Alexander C. Ford is the guarantor of the article. Thestudy sponsor had no input into the concept, design, analysis,or reporting of the study.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Authors’ Contributions

David J. Gracie, Alexander C. Ford, Barry J. Hall, and P.John Hamlin devised the study. David J. Gracie conductedthe data collection. Alexander C. Ford, David J. Gracie, andBarry J. Hall analysed and interpreted the data. Barry J. Halland Alexander C. Ford drafted the manuscript. All authorscontributed to and approved the final draft of the manuscript.

Acknowledgments

The authors are thankful for the support provided by theLeeds Teaching Hospitals Charitable Foundation (9R11/14-02) and the unrestricted research funds provided by TillottsPharma UK Ltd.

References

[1] S. J. Bickston, B. W. Behm, D. J. Tsoulis et al., “Vedolizumab forinduction and maintenance of remission in ulcerative colitis,”Cochrane Database of Systematic Reviews, vol. 8, 2014.

[2] J. K. MacDonald and J. W. McDonald, “Natalizumab forinduction of remission in Crohn’s disease,” Cochrane Databaseof Systematic Reviews, vol. 1, 2007.

[3] K. J. Khan,M.C.Dubinsky, A. C. Ford, T. A.Ullman,N. J. Talley,and P. Moayyedi, “Efficacy of immunosuppressive therapy forinflammatory bowel disease: a systematic review and meta-analysis,” American Journal of Gastroenterology, vol. 106, no. 4,pp. 630–642, 2011.

[4] A. C. Ford, J.-P. Achkar, K. J. Khan et al., “Efficacy of 5-aminosalicylates in ulcerative colitis: systematic review andmeta-analysis,” American Journal of Gastroenterology, vol. 106,no. 4, pp. 601–616, 2011.

[5] A. C. Ford, C. N. Bernstein, K. J. Khan et al., “Glucocorti-costeroid therapy in inflammatory bowel disease: Systematicreview and meta-analysis,” American Journal of Gastroenterol-ogy, vol. 106, no. 4, pp. 590–599, 2011.

[6] A. C. Ford, S. V. Kane, K. J. Khan et al., “Efficacy of 5-aminosali-cylates in crohn’s disease: systematic review andmeta-analysis,”American Journal of Gastroenterology, vol. 106, no. 4, pp. 617–629, 2011.

[7] A. C. Ford, W. J. Sandborn, K. J. Khan, S. B. Hanauer, N. J.Talley, and P. Moayyedi, “Efficacy of biological therapies ininflammatory bowel disease: systematic review and meta-analysis,” American Journal of Gastroenterology, vol. 106, no. 4,pp. 644–659, 2011.

[8] A. J. Panara, A. J. Yarur, B. Rieders et al., “The incidence andrisk factors for developing depression after being diagnosedwith inflammatory bowel disease: A cohort study,” AlimentaryPharmacology &Therapeutics, vol. 39, no. 8, pp. 802–810, 2014.

[9] L. A. Graff, J. R. Walker, and C. N. Bernstein, “Depression andanxiety in iflammatory bowel disease: a review of comorbidityand management,” Inflammatory Bowel Diseases, vol. 15, no. 7,pp. 1105–1118, 2009.

[10] E. Fuller-Thomson and J. Sulman, “Depression and inflamma-tory bowel disease: Findings from two nationally representativeCanadian surveys,” Inflammatory Bowel Diseases, vol. 12, no. 8,pp. 697–707, 2006.

[11] C.Mittermaier, C.Dejaco, T.Waldhoer et al., “Impact of depres-sive mood on relapse in patients with inflammatory boweldisease: a prospective 18-month follow-up study,”PsychosomaticMedicine, vol. 66, no. 1, pp. 79–84, 2004.

[12] J. R. Goodhand, M. Wahed, J. E. Mawdsley, A. D. Farmer, Q.Aziz, and D. S. Rampton, “Mood disorders in inflammatorybowel disease: relation to diagnosis, disease activity, perceivedstress, and other factors,” Inflammatory Bowel Diseases, vol. 18,no. 12, pp. 2301–2309, 2012.

[13] N. A. Koloski, M. Jones, J. Kalantar, M. Weltman, J. Zaguirre,and N. J. Talley, “The brain—gut pathway in functional gas-trointestinal disorders is bidirectional: a 12-year prospectivepopulation-based study,”Gut, vol. 61, no. 9, pp. 1284–1290, 2012.

[14] J.-E. Ghia, P. Blennerhassett, Y. Deng, E. F. Verdu, W. I. Khan,and S. M. Collins, “Reactivation of Inflammatory bowel diseasein a mouse model of depression,” Gastroenterology, vol. 136, no.7, pp. 2280–e4, 2009.

[15] J.-E. Ghia, P. Blennerhassett, and S. M. Collins, “Impairedparasympathetic function increases susceptibility to inflamma-tory bowel disease in amousemodel of depression,”The Journalof Clinical Investigation, vol. 118, no. 6, pp. 2209–2218, 2008.

[16] J. E. Mawdsley, M. G. Macey, R. M. Feakins, L. Langmead,and D. S. Rampton, “The effect of acute psychologic stress onsystemic and rectal mucosal measures of inflammation in ulcer-ative colitis,” Gastroenterology, vol. 131, no. 2, pp. 410–419, 2006.

[17] A. N. Ananthakrishnan, V. S. Gainer, R. G. Perez et al., “Psychi-atric co-morbidity is associatedwith increased risk of surgery inCrohn’s disease,” Alimentary Pharmacology &Therapeutics, vol.37, no. 4, pp. 445–454, 2013.

[18] L. S. Gaines, J. C. Slaughter, S. N. Horst et al., “Associationbetween affective-cognitive symptoms of depression and exac-erbation of Crohn’s disease,”American Journal of Gastroenterol-ogy, vol. 111, no. 6, pp. 864–870, 2016.

Canadian Journal of Gastroenterology and Hepatology 11

[19] B. Kochar, E. L. Barnes, M. D. Long et al., “Depression isassociated with more aggressive inflammatory bowel disease,”American Journal of Gastroenterology, vol. 113, no. 1, pp. 80–85,2017.

[20] D. J. Gracie, E. A. Guthrie, P. J. Hamlin, and A. C. Ford,“Bi-directionality of brain–gut interactions in patients withinflammatory bowel disease,” Gastroenterology, vol. 154, no. 6,pp. 1635–1646.e3, 2018.

[21] D. J. Gracie, A. J. Irvine, R. Sood, A. Mikocka-Walus, P. J.Hamlin, and A. C. Ford, “Effect of psychological therapy ondisease activity, psychological comorbidity, and quality of lifein inflammatory bowel disease: a systematic review and meta-analysis,” The Lancet Gastroenterology and Hepatology, vol. 2,no. 3, pp. 189–199, 2017.

[22] B. J. D. Macer, S. L. Prady, and A. Mikocka-Walus, “Antide-pressants in inflammatory bowel disease: a systematic review,”Inflammatory Bowel Diseases, vol. 23, no. 4, pp. 534–550, 2017.

[23] A. Mikocka-Walus, P. A. Hughes, P. Bampton et al., “Fluoxetinefor maintenance of remission and to improve quality of lifein patients with Crohn’s disease: A pilot randomized placebo-controlled trial,” Journal of Crohn’s and Colitis, vol. 11, no. 4, pp.509–514, 2017.

[24] D. J. Gracie, C. J. M. Williams, R. Sood et al., “Negative effectson psychological health and quality of life of genuine irritablebowel syndrome–type symptoms in patientswith inflammatorybowel disease,”Clinical Gastroenterology andHepatology, vol. 15,no. 3, pp. 376–384.e5, 2017.

[25] D. J. Gracie, C. J. M. Williams, R. Sood et al., “Poor correlationbetween clinical disease activity and mucosal inflammation,and the role of psychological comorbidity, in inflammatorybowel disease,” American Journal of Gastroenterology, vol. 111,no. 4, pp. 541–551, 2016.

[26] E. von Elm, D. G. Altman,M. Egger, S. J. Pocock, P. C. Gøtzsche,and J. P. Vandenbroucke, “The strengthening the reporting ofobservational studies in epidemiology (STROBE) statement:guidelines for reporting observational studies,” Journal of Clin-ical Epidemiology, vol. 61, no. 4, pp. 344–349, 2008.

[27] R. F. Harvey and J. M. Bradshaw, “A simple index of Crohn’s-disease activity.,”The Lancet, vol. 1, no. 8167, p. 514, 1980.

[28] R. S. Walmsley, R. C. S. Ayres, R. E. Pounder, and R. N. Allan,“A simple clinical colitis activity index,” Gut, vol. 43, no. 1, pp.29–32, 1998.

[29] S. Vermeire, S. Schreiber, W. J. Sandborn, C. Dubois, and P.Rutgeerts, “Correlation between theCrohn’s disease activity andHarvey-Bradshaw indices in assessing Crohn’s disease severity,”Clinical Gastroenterology and Hepatology, vol. 8, no. 4, pp. 357–363, 2010.

[30] S. L. Jowett, C. J. Seal, E. Phillips, W. Gregory, J. R. Barton, andM. R. Welfare, “Defining relapse of ulcerative colitis using asymptom-based activity index,” Scandinavian Journal of Gas-troenterology, vol. 38, no. 2, pp. 164–171, 2003.

[31] A. S. Zigmond and R. P. Snaith, “The hospital anxiety and de-pression scale,”Acta Psychiatrica Scandinavica, vol. 67, no. 6, pp.361–370, 1983.

[32] K. Kroenke, R. L. Spitzer, and J. B. W. Williams, “The PHQ-15:Validity of a newmeasure for evaluating the severity of somaticsymptoms,” Psychosomatic Medicine, vol. 64, no. 2, pp. 258–266,2002.

[33] J. C. Coyne and E. van Sonderen, “No further research needed:abandoning the Hospital and Anxiety Depression Scale(HADS),” Journal of Psychosomatic Research, vol. 72, no. 3, pp.173-174, 2012.

[34] I. Marın-Jimenez, M. G. Montoya, A. Panadero et al., “Man-agement of the psychological impact of inflammatory boweldisease: perspective of doctors and patients - The ENMENTEproject,” Inflammatory Bowel Diseases, vol. 23, no. 9, pp. 1492–1498, 2017.

[35] K. A. Collins, H. A. Westra, D. J. A. Dozois, and D. D. Burns,“Gaps in accessing treatment for anxiety and depression: Chal-lenges for the delivery of care,” Clinical Psychology Review, vol.24, no. 5, pp. 583–616, 2004.

[36] P. S.Wang,M. Lane,M.Olfson,H. A. Pincus, K. B.Wells, and R.C. Kessler, “Twelve-month use of mental health services in theUnited States: Results from the National Comorbidity SurveyReplication,” Archives of General Psychiatry, vol. 62, no. 6, pp.629–640, 2005.

[37] A. Mikocka-Walus, S. R. Knowles, L. Keefer, and L. Graff,“Controversies revisited: a systematic review of the comorbidityof depression and anxiety with inflammatory bowel diseases,”Inflammatory Bowel Diseases, vol. 22, no. 3, pp. 752–762, 2015.

[38] O. Yanartas, H. T. Kani, E. Bicakci et al., “The effects of psychi-atric treatment on depression, anxiety, quality of life, and sexualdysfunction in patients with inflammatory bowel disease,”Neuropsychiatric Disease and Treatment, vol. 12, pp. 673–683,2016.

[39] D. A. Drossman, J. Tack, A. C. Ford, E. Szigethy, H. Tornblom,and L. Van Oudenhove, “Neuromodulators for functionalgastrointestinal disorders (disorders of gut−brain interaction):a rome foundation working team report,”Gastroenterology, vol.154, no. 4, pp. 1140–1171.e1, 2018.

[40] H. R. Rahimi,M. Shiri, andA. Razmi, “Antidepressants can treatinflammatory bowel disease through regulation of the nuclearfactor-𝜅B/nitric oxide pathway and inhibition of cytokineproduction: A hypothesis,” World Journal of GastrointestinalPharmacology andTherapeutics, vol. 3, no. 6, p. 83, 2012.

Stem Cells International

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

MEDIATORSINFLAMMATION

of

EndocrinologyInternational Journal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Disease Markers

Hindawiwww.hindawi.com Volume 2018

BioMed Research International

OncologyJournal of

Hindawiwww.hindawi.com Volume 2013

Hindawiwww.hindawi.com Volume 2018

Oxidative Medicine and Cellular Longevity

Hindawiwww.hindawi.com Volume 2018

PPAR Research

Hindawi Publishing Corporation http://www.hindawi.com Volume 2013Hindawiwww.hindawi.com

The Scientific World Journal

Volume 2018

Immunology ResearchHindawiwww.hindawi.com Volume 2018

Journal of

ObesityJournal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Computational and Mathematical Methods in Medicine

Hindawiwww.hindawi.com Volume 2018

Behavioural Neurology

OphthalmologyJournal of

Hindawiwww.hindawi.com Volume 2018

Diabetes ResearchJournal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Research and TreatmentAIDS

Hindawiwww.hindawi.com Volume 2018

Gastroenterology Research and Practice

Hindawiwww.hindawi.com Volume 2018

Parkinson’s Disease

Evidence-Based Complementary andAlternative Medicine

Volume 2018Hindawiwww.hindawi.com

Submit your manuscripts atwww.hindawi.com