Rawatan adjuvan dengan antidepresan untuk pengurusan penyakit radang usus
Abstract
Background
Symptoms of anxiety and depression are common in inflammatory bowel disease (IBD). Antidepressants are taken by approximately 30% of people with IBD. However, there are no current guidelines on treating co‐morbid anxiety and depression in people with IBD with antidepressants, nor are there clear data on the role of antidepressants in managing physical symptoms of IBD.
Objectives
The objectives were to assess the efficacy and safety of antidepressants for treating anxiety and depression in IBD, and to assess the effects of antidepressants on quality of life (QoL) and managing disease activity in IBD.
Search methods
We searched MEDLINE; Embase, CINAHL, PsycINFO, CENTRAL, and the Cochrane IBD Group Specialized Register from inception to 23 August 2018. Reference lists, trials registers, conference proceedings and grey literature were also searched.
Selection criteria
Randomised controlled trials (RCTs) and observational studies comparing any type of antidepressant to placebo, no treatment or an active therapy for IBD were included.
Data collection and analysis
Two authors independently screened search results, extracted data and assessed bias using the Cochrane risk of bias tool. We used the Newcastle‐Ottawa Scale to assess quality of observational studies. GRADE was used to evaluate the certainty of the evidence supporting the outcomes. Primary outcomes included anxiety and depression. Anxiety was assessed using the Hospital Anxiety and Depression Scale (HADS) or the Hamilton Anxiety Rating Scale (HARS). Depression was assessed using HADS or the Beck Depression Inventory. Secondary outcomes included adverse events (AEs), serious AEs, withdrawal due to AEs, quality of life (QoL), clinical remission, relapse, pain, hospital admissions, surgery, and need for steroid treatment. QoL was assessed using the WHO‐QOL‐BREF questionnaire. We calculated the risk ratio (RR) and corresponding 95% confidence intervals (CI) for dichotomous outcomes. For continuous outcomes, we calculated the mean difference (MD) with 95% CI. A fixed‐effect model was used for analysis.
Main results
We included four studies (188 participants). Two studies were double‐blind RCTs, one was a non‐randomised controlled trial, and one was an observational retrospective case‐matched study. The age of participants ranged from 27 to 37.8 years. In three studies participants had quiescent IBD and in one study participants had active or quiescent IBD. Participants in one study had co‐morbid anxiety or depression. One study used duloxetine (60 mg daily), one study used fluoxetine (20 mg daily), one study used tianeptine (36 mg daily), and one study used various antidepressants in clinical ranges. Three studies had placebo controls and one study had a no treatment control group. One RCT was rated as low risk of bias and the other was rated as high risk of bias (incomplete outcome data). The non‐randomised controlled trial was rated as high risk of bias (random sequence generation, allocation concealment, blinding). The observational study was rated as high methodological quality, but is still considered to be at high risk of bias given its observational design.
The effect of antidepressants on anxiety and depression is uncertain. At 12 weeks, the mean anxiety score in antidepressant participants was 6.11 + 3 compared to 8.5 + 3.45 in placebo participants (MD ‐2.39, 95% ‐4.30 to ‐0.48, 44 participants, low certainty evidence). At 12 months, the mean anxiety score in antidepressant participants was 3.8 + 2.5 compared to 4.2 + 4.9 in placebo participants (MD ‐0.40, 95% ‐3.47 to 2.67, 26 participants; low certainty evidence). At 12 weeks, the mean depression score in antidepressant participants was 7.47 + 2.42 compared to 10.5 + 3.57 in placebo participants (MD ‐3.03, 95% CI ‐4.83 to ‐1.23, 44 participants; low certainty evidence). At 12 months, the mean depression score in antidepressant participants was 2.9 + 2.8 compared to 3.1 + 3.4 in placebo participants (MD ‐0.20, 95% ‐2.62 to 2.22, 26 participants; low certainty evidence).
The effect of antidepressants on AEs is uncertain. Fifty‐seven per cent (8/14) of antidepressant participants group reported AEs versus 25% (3/12) of placebo participants (RR 2.29, 95% CI 0.78 to 6.73, low certainty evidence). Commonly reported AEs include nausea, headache, dizziness, drowsiness, sexual problems, insomnia, fatigue, low mood/anxiety, dry mouth, muscle spasms and hot flushes. None of the included studies reported any serious AEs. None of the included studies reported on pain.
One study (44 participants) reported on QoL at 12 weeks and another study (26 participants) reported on QoL at 12 months. Physical, Psychological, Social and Environmental QoL were improved at 12 weeks compared to placebo (all low certainty evidence). There were no group differences in QoL at 12 months (all low certainty evidence). The effect of antidepressants on maintenance of clinical remission and endoscopic relapse is uncertain. At 12 months, 64% (9/14) of participants in the antidepressant group maintained clinical remission compared to 67% (8/12) of placebo participants (RR 0.96, 95% CI 0.55 to 1.69; low certainty evidence). At 12 months, none (0/30) of participants in the antidepressant group had endoscopic relapse compared to 10% (3/30) of placebo participants (RR 0.14, 95% CI 0.01 to 2.65; very low certainty evidence).
Authors' conclusions
The results for the outcomes assessed in this review are uncertain and no firm conclusions regarding the efficacy and safety of antidepressants in IBD can be drawn. Future studies should employ RCT designs, with a longer follow‐up and develop solutions to address attrition. Inclusion of objective markers of disease activity is strongly recommended as is testing antidepressants from different classes, as at present it is unclear if any antidepressant (or class thereof) has differential efficacy.
PICOs
Ringkasan bahasa mudah
Antidepresan untuk penyakit radang usus
Apakah penyakit radang usus?
Penyakit radang usus (IBD) adalah penyakit radang kronik yang melibatkan salur gastrousus (kolon atau usus kecil atau kedua‐duanya). IBD kebanyakannya merangkumi penyakit Crohn’s dan kolitis ulseratif. Simptom IBD termasuklah cirit‐birit, tergesa‐gesa membuang air besar (termasuk kehilangan kawalan membuang air besar), sakit perut, pendarahan rektum, kelesuan dan penurunan berat badan. Apabila seseorang itu mengalami simptom‐simptom IBD, mereka dikategorikan mempunyai keaktifan penyakit yang tinggi. Apabila simptom‐simptom IBD berhenti, penyakit tersebut berada dalam tempoh kesembuhan sementara. IBD dikaitkan dengan beban psikososial dan kadar kemurungan pada orang yang mempunyai IBD adalah dua kali ganda lebih tinggi berbanding dengan populasi umum. Keresahan dan kemurungan yang mengiringi IBD mungkin dikaitkan dengan kualiti hidup yang teruk, aktiviti IBD yang semakin buruk, kadar kemasukan hospital yang lebih tinggi dan kepatuhan rawatan yang lebih rendah. Seramai 30% orang yang menghidapi IBD mengambil antidepresan yang dipreskripsi sama ada untuk kesihatan mental atau simptom‐simptom usus atau kedua‐duanya.
Apakah itu antidepresan?
Antidepresan adalah ubat yang digunakan untuk merawat kemurungan dan kecelaruan mental yang lain seperti keresahan. Pada masa ini, tiada antidepresan yang telah diluluskan oleh agensi penguatkuasaan khususnya untuk merawat keresahan dan kemurungan, untuk menguruskan simptom‐simptom fizikal atau untuk mengurangkan radang usus pada orang yang menghidapi IBD. Walau bagaimanapun, sesetengah antidepresan mempunyai indikasi untuk merawat kesakitan dalam keadaan kronik dan telah biasa digunakan untuk menguruskan simptom‐simptom usus dalam keadaan seperti sindrom usus rengsa.
Apakah yang disiasat oleh penyelidik?
Kajian lepas tentang terapi antidepresan dalam IBD telah disemak. Data daripada beberapa kajian tersebut telah digabungkan menggunakan kaedah yang dipanggil analisis meta. Semasa analisis, orang yang mengambil antidepresan telah dibandingkan dengan mereka yang tidak mengambil antidepresan dari segi kadar keresahan dan kemurungan serta ukuran‐ukuran lain seperti kualiti hidup, kesan sampingan dan aktiviti penyakit IBD.
Apakah yang ditemui oleh para penyelidik?
Para penyelidik telah mencari sastera perubatan sehingga 23 Ogos 2018. Empat kajian yang diterbitkan, yang merangkumi seramai 188 orang, telah menyelidik terapi antidepresan pada orang yang menghidapi IBD. Usia peserta adalah dari 27 sehingga 37.8. Dalam tiga kajian, peserta‐peserta mempunyai IBD dalam tempoh kesembuhan sementara dan dalam satu kajian, peserta‐peserta mempunyai sama ada IBD aktif atau IBD dalam tempoh kesembuhan sementara. Peserta‐peserta dalam satu kajian mempunyai keresahan atau kemurungan yang wujud bersama. Satu kajian menggunakan duloxetine (60 mg sehari), satu kajian menggunakan fluoxetine (20 mg sehari), satu kajian menggunakan tianeptin (36 mg sehari), dan satu kajian menggunakan pelbagai antidepresan. Tiga kajian mempunyai kumpulan kawalan plasebo (contohnya pil gula) dan satu kajian mempunyai kumpulan kawalan tanpa rawatan.
Analisis menunjukkan bahawa simptom‐simptom keresahan dan kemurungan bertambah baik dalam kalangan orang yang mengambil antidepresan berbanding dengan plasebo. Peserta yang menerima antidepresan mengalami lebih banyak kesan sampingan daripada mereka yang menerima plasebo. Kesan‐kesan sampingan yang dilaporkan oleh mereka yang mengambil antidepresan termasuklah; rasa loya, sakit kepala, pening, mengantuk, masalah seksual, insomnia, kelesuan, muram/gelisah, mulut kering, tidur yang kurang lelap, kaki gelisah dan hangat memerah. Sebahagian daripada aspek kualiti hidup serta aktiviti IBD telah bertambah baik dalam kumpulan antidepresan. Kualiti keseluruhan kajian yang dimasukkan ke dalam ulasan ini adalah rendah kerana kajian‐kajian melibatkan bilangan peserta yang kecil, dan juga melibatkan populasi IBD yang berbeza antara satu sama lain dalam ciri‐ciri utama. Tambahan pula, pelbagai jenis antidepresan telah dinilai, maka bukti untuk mana‐mana satu antidepresan adalah tidak jelas. Oleh itu, kajian lanjutan adalah diperlukan untuk mengesahkan pemerhatian‐pemerhatian ini.
Kesimpulan
Keputusan untuk hasil yang dinilai dalam kajian ini adalah tidak jelas dan tiada kesimpulan yang kukuh boleh dibuat mengenai manfaat dan kemudaratan antidepresan dalam IBD. Lebih banyak kajian diperlukan untuk membuat kesimpulan kukuh mengenai manfaat dan kemudaratan penggunaan antidepresan pada orang yang menghidapi IBD.
Authors' conclusions
Summary of findings
| Antidepressants compared to placebo for inflammatory bowel disease | ||||||
| Patient or population: participants with active and inactive inflammatory bowel disease | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
| Risk with Placebo | Risk with Antidepressants | |||||
| Anxiety at 12 weeks | The mean anxiety was 8.5 (SD = 3.45) | The mean anxiety was 6.11 (SD = 3) MD 2.39 lower | ‐ | 44 | ⊕⊕⊝⊝ | Anxiety was assessed using the HADS |
| Anxiety at 12 months | The mean anxiety was 4.2 (SD = 4.9) | The mean anxiety was 3.8 (SD = 2.5) MD ‐0.40 lower | ‐ | 26 | ⊕⊕⊝⊝ | Anxiety was assessed using the HADS A second non‐randomised study using the HARS reported a mean score of 12.65 + 3.76 in the antidepressant group (n = 30) compared to 17.85 + 3.33 in the placebo group (n = 30) (MD ‐5.20, 95% CI ‐7 to ‐3.40; very low certainty evidence) |
| Depression at 12 weeks | The mean depression was 10.5 (SD = 3.57) | The mean depression was 7.47 (SD = 2.42) MD ‐3.03 lower | ‐ | 44 | ⊕⊕⊝⊝ | Depression was assessed using the HADS |
| Depression at 12 months | The mean depression was 3.1 (SD = 3.4) | The mean depression was 2.9 (SD = 2.8) MD ‐0.20 lower | ‐ | 26 | ⊕⊕⊝⊝ | Depression was assessed using the HADS A second non‐randomised study using the Beck Depression Inventory reported a mean score of 9.6 + 2.76 in the antidepressant group (n = 30) compared to 16.35 + 5.41 in the placebo group (n = 30) (MD ‐6.75, 95% CI ‐8.92 to ‐4.58; very low certainty evidence) |
| Adverse events at 12 months | 250 per 1,000 | 573 per 1,000 | RR 2.29 | 26 | ⊕⊕⊝⊝ | Commonly reported adverse events include nausea, headache, dizziness, drowsiness, sexual problems, insomnia, fatigue, low mood/anxiety, dry mouth muscle spasms and hot flushes None of the included studies reported any serious adverse events |
| Quality of life | ‐ | ‐ | See comment | 70 | ⊕⊕⊝⊝ | Quality of life was assessed using the WHO‐QOL‐BREF We were unable to pool data as the outcome was reported at 12 weeks in 1 trial (44 participants) and 12 months in 1 trial (26 participants). Physical, Psychological, Social and Environmental QoL were improved only at 12 weeks with no group difference at 12 months |
| Pain | Not reported | No studies reported this outcome | ||||
| Maintenance of remission at 12 months | 667 per 1,000 | 640 per 1,000 | RR 0.96 | 26 | ⊕⊕⊝⊝ | Maintenance of remission was measured by the CDAI (< 150) and fecal calprotectin levels |
| Endoscopic relapse at 12 months | 0 per 1,000 | 0 per 1,000 | RR 0.14 | 60 | ⊕⊝⊝⊝ | Non‐randomised study. We were unable to calculate absolute effects. Endoscopic relapse occurred in 0% (0/30) of participants in the antidepressants group compared to 10% (3/30) in the placebo group |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 Downgraded one level due to serious imprecision (44 participants) 2 Downgraded one level due to high risk of bias (incomplete outcome data) 3 Downgraded two levels due very serious imprecision (26 participants) 4 Downgraded two levels due very serious imprecision (11 events) 5 Downgraded two levels due very serious imprecision (17 events). 6 Downgraded one level due to very serious imprecision (3 events) and a large CI around the point estimate | ||||||
Background
Description of the condition
Inflammatory bowel disease (IBD) is a chronic, inflammatory disease affecting the gastrointestinal tract. The aetiology of IBD is thought to involve an inappropriate immune response to intestinal microbiota, triggered by environmental factors, in genetically susceptible people. The typical symptoms of IBD include diarrhoea, urgency of defecation, abdominal pain and cramping, fatigue, and weight loss. IBD affects 2.2 million people in Europe (Loftus 2004), 1.4 million people in the USA (CCFA 2012), 233,000 people in Canada (Rocchi 2012), and over 75,000 people in Australia (CCA 2015).
IBD is associated with a psychosocial burden. People with IBD have a higher life‐time prevalence of depression compared to the general community, with estimated rates of 27% in persons with IBD compared to 12% in the general population (Walker 2008). During IBD remission, over 20% of people report symptoms of anxiety or depression but this number rises to 60% when IBD is active (Mikocka‐Walus 2016a). Psychological stress has been found to predict symptomatic disease course (Bernstein 2011), and is also linked to increased inflammation (Maunder 2008). Associations between symptoms of depression and clinical recurrence over time (Mikocka‐Walus 2016d), higher hospitalisation rates (Van Langenberg 2010), and lower adherence to treatment (Nigro 2001), have also been suggested.
Despite the high prevalence of mental co‐morbidities with IBD and the effect on disease course, mental disorders are not routinely treated in this population. In fact, fewer than 40% of those with IBD reporting mental symptoms receive psychotherapy (Bennebroek Evertsz 2012). Poor access to psychologists may contribute to this finding. In the UK, for example, only a fraction of IBD services (12%) have access to clinical psychology (RCP 2014). However, psychotherapy is not a universal treatment for mental and physical symptoms associated with IBD (Timmer 2011). While the most recent meta‐analysis demonstrated that psychological therapies, and cognitive behavioural therapy in particular, might have small short‐term beneficial effects on depression scores and quality of life (QoL) in IBD (Gracie 2017), there is no evidence that psychotherapies are effective for IBD activity. The limitations of the current studies on psychotherapy in IBD are discussed elsewhere (Knowles 2013).
Depending on the population, 10% to 30% of IBD patients take antidepressants (Fuller‐Thomson 2006; Haapamaki 2013; Mikocka‐Walus 2012). However, studies have shown that those IBD patients who receive antidepressants do not necessarily suffer from depression but often are treated for pain, insomnia or functional bowel symptoms which overlap with IBD (Mikocka‐Walus 2007; Mikocka‐Walus 2012). This resembles treatment for functional gut disorders such as irritable bowel syndrome, where there is good evidence of antidepressants’ efficacy for physical symptoms (Ford 2009; Ford 2014). However, while antidepressants are used in IBD, the efficacy of this intervention in this population has not been established to date.
Description of the intervention
Antidepressants are drugs used to treat depression and other mental disorders such as anxiety. While lithium was known in the 19th century, it wasn't introduced to common psychiatry practice until the 1950s (Shorter 2009). Other antidepressants ‐ monoamine oxidase inhibitors and tricyclics were also introduced in the 1950s while tetracyclics were introduced in the 1970s. Presently, the most commonly used antidepressants are selective serotonin reuptake inhibitors (SSRIs) which were introduced in the 1980s. Serotonin‐norepinephrine reuptake inhibitors (SNRIs) became available in the 1990s. Other less commonly known groups of antidepressants include: heterocyclics, norepinephrine reuptake inhibitors (NARIs), norepinephrine–dopamine reuptake inhibitors (NDRIs), noradrenergic and specific serotonergic antidepressants (NaASSAs), and serotonin antagonist and reuptake inhibitors (SARIs).
Dosage regimens differ between the different classes and individual antidepressants, and depend on the severity of symptoms. Antidepressants are usually taken daily (either morning or night) and the treatment ranges from several months to several years or even lifetime use. The efficacy of older antidepressants (e.g. tricyclics) and newer, second‐generation antidepressants (e.g. SSRI) is similar (Williams 2000). However, the use of first generation antidepressants is associated with more serious adverse events, with increased lethality with overdose (Gartlehner 2007; Gartlehner 2011), and thus these agents are no longer first line pharmacotherapy treatment for depression or anxiety. Among the new generation antidepressants, escitalopram and sertraline are considered to be superior to other commonly used antidepressants in terms of efficacy and acceptability (Cipriani 2009).
No antidepressants are currently approved by regulatory agencies for specifically treating anxiety and depression comorbid with IBD, to manage physical symptoms of IBD or to reduce bowel inflammation. However, some antidepressants have indications for treatment of pain in chronic conditions. For example, duloxetine has an indication for diabetic peripheral neuropathy (AMH 2012).
How the intervention might work
Antidepressants are thought to work through compensating for transmitter deficits in the brain, which are considered to be the underlying cause of depression (Ritter 2015). Antidepressants can either inhibit the reuptake of neurotransmitters from the synaptic cleft or inhibit the metabolism of neurotransmitters. Thus, for example, tricyclics inhibit the uptake of noradrenaline or serotonin or both. SSRIs inhibit serotonin uptake, while SNRIs inhibit both noradrenaline and serotonin uptake, and monoamine oxidase inhibitors inhibit the metabolism of mono‐amine neurotransmitters such as serotonin. However, it is also hypothesized that antidepressants may help treat depression due to immunoregulatory effects (Maes 2001). A significant drop in serum C‐reactive protein concentrations (independent of depressive symptoms being resolved) has been observed following four weeks of treatment with SSRIs in people with a major depressive disorder (O'Brien 2006). Even in healthy volunteers, antidepressants have been shown to improve immunoregulatory activity (Szuster‐Ciesielska 2003); and in sufferers of chronic inflammatory conditions such as asthma, antidepressants are reported to reduce the need for steroids (Brown 2005), and improve overall immune function (Krommydas 2005).
Given the immunoregulatory effect of antidepressants, it is possible that when given to patients with inflammatory conditions such as IBD, antidepressants may exert an effect on inflammation outside the brain and thus improve not only mood but also bowel symptoms, by extending or inducing remission. Animal studies examining models of colitis can serve as a proof of concept (Mikocka‐Walus 2009). For example, mice receiving desipramine (a tricyclic antidepressant) have significantly reduced microscopic damage (P < 0.05) and attenuation of colonic myeloperoxidase activity (P < 0.05) when compared to placebo (Varghese 2006). Furthermore, serum Il‐1β concentrations were significantly lower in rats receiving 10 mg fluoxetine (an SSRI), 20 mg fluoxetine, 20 mg desipramine or 10 mg desipramine compared to controls (all P < 0.001) (Guemei 2008). Similarly, reductions in serum tumour necrosis factor‐alpha were observed in rats receiving either desipramine or fluoxetine (10 or 20 mg) compared to controls (all P < 0.001). Thus, antidepressants can induce an anti‐inflammatory response which is not related to antidepressive effects.
Further, treatments which improve inflammation in IBD, such as biologics, are known to also improve QoL (Feagan 2007). Thus, it is hypothesised that antidepressants can reduce symptoms of anxiety and depression and improve QoL in IBD. It is further hypothesised that, similarly to what occurs in animal models where antidepressants have been shown to have anti‐inflammatory properties, antidepressants may induce remission of IBD and reduce the number of flares in humans.
Why it is important to do this review
There is a growing interest in mental health and antidepressant use in chronic illness, to manage comorbid depression as well as physical symptoms, with recent Cochrane systematic reviews conducted on diabetes (Baumeister 2014), coronary artery disease (CAD) (Baumeister 2011), and functional gut disorders (Ruepert 2011). These reviews have shown improved glycaemic control after the use of SSRIs versus placebo in patients with diabetes (Baumeister 2014); improvements in depression, reduction in hospitalisations and emergency room visits (though no beneficial effects on mortality, cardiac events or QoL) after SSRI use compared to placebo in CAD (Baumeister 2011); and improvements in abdominal pain and symptoms (after tricyclics as compared to placebo) and in global assessment (after SSRIs as compared to placebo) in irritable bowel syndrome (Ruepert 2011). However, there is currently no Cochrane systematic review exploring the role of antidepressants in IBD.
The first systematic review on the use of antidepressants in IBD was conducted in 2005 and identified 12 uncontrolled studies (Mikocka‐Walus 2006). While the review observed a beneficial effect of antidepressants on mental and physical status of IBD patients, the available research was of low quality, making it impossible to provide definitive conclusions on the efficacy of antidepressants for improving outcomes in patients with IBD. A more recent systematic review (Macer 2017), included 15 studies including 1 randomised controlled trial, 2 cohort studies, 1 case‐control study, 1 cross‐sectional survey, 1 qualitative study, 2 audits, 1 case series, and 6 case reports. Twelve studies suggested that antidepressants have a positive impact on IBD course. Nine studies reported on anxiety and depression as outcomes. Eight of these studies reported beneficial effects of antidepressants. Most of the studies were deemed to be at low risk of bias, apart from the case reports, which were at high risk of bias. While this review confirmed the beneficial effect of antidepressants on IBD course, it concluded that it was not possible to determine efficacy of antidepressants for certain due to the lack of randomised controlled trials (RCTs). Since the publication of the latest review another trial of antidepressant use in IBD has been published (Mikocka‐Walus 2017). It is now time to review current evidence on the effectiveness and safety of antidepressants for mood and disease activity in IBD patients. It is also critical to conduct the first meta‐analysis of the effects of antidepressants in IBD management.
Given the widespread use of antidepressants in IBD (Fuller‐Thomson 2006; Haapamaki 2013; Mikocka‐Walus 2012), and the potential for not only addressing poor mental health but also immunoregulatory activity (Krommydas 2005), it is important to assess the efficacy and safety of antidepressants in IBD. This review explores the adjuvant role of antidepressants in IBD.
Objectives
Primary objectives
-
To assess the efficacy and safety of antidepressants for treating anxiety and depression in IBD.
Secondary objectives
-
To assess the efficacy and safety of antidepressants for improving QoL in IBD.
-
To assess the efficacy and safety of antidepressants for managing IBD disease activity.
Methods
Criteria for considering studies for this review
Types of studies
All published and unpublished quantitative studies including: RCTs, and non‐randomised controlled studies, prospective and retrospective studies including cohort, case control, cross‐sectional and audit studies, were eligible for inclusion. Studies without a comparison group were excluded.
Types of participants
Humans, clinically diagnosed with IBD of any type (i.e. Crohn's disease, ulcerative colitis or indeterminate colitis) – according to standard practice (i.e. a combination of clinical, radiologic, endoscopic and histologic grounds), were considered for inclusion.
Types of interventions
All types of antidepressants (in any dose) were included:
-
SSRIs (citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, sertraline);
-
Tricyclics (amitriptyline, clomipramine, desipramine, dothiepin, doxepin, imipramine, lofepramine, nortriptyline, protriptyline, trimipramine);
-
Heterocyclics (mianserin);
-
MAO inhibitors (isocarboxazid, phenelzine, tranylcypromine, brofaromine, moclobemide, tyrima);
-
NARIs (reboxetine);
-
NDRIs (amineptine, buproprion);
-
SNRIs (duloxetine, milnacipran, venlafaxine);
-
NASSAs (mirtazapine);
-
SARIs (trazodone); and
-
Other unclassified antidepressants (agomelatine, vilazodone).
Any comparator including any of the following was considered for inclusion:
-
No intervention;
-
Placebo;
-
Standard care/treatment as usual;
-
Surgery;
-
Alternative interventions used to treat depression and anxiety, e.g. anxiolytics, psychotherapy;
-
Another antidepressant; and
-
Any other active comparators.
Types of outcome measures
Primary outcomes
Efficacy
-
Anxiety and depression as measured by any well‐established anxiety or depression scale
Secondary outcomes
Safety
-
Adverse events;
-
Serious adverse events;
-
Study withdrawal due to adverse events.
Efficacy
-
QoL as measured by any well‐established QoL scale;
-
IBD clinical remission or relapse;
-
Pain severity as established using any well‐established pain scale; and
-
Hospital admissions, surgery, need for steroid treatment.
Search methods for identification of studies
Electronic searches
The following sources were searched from inception to 23rd August 2018 and without language restrictions:
-
MEDLINE via PubMed (Appendix 1);
-
Embase (Appendix 2);
-
CINAHL (Appendix 3);
-
PsycINFO (Appendix 4);
-
CENTRAL; and
-
The Cochrane IBD Group Specialized Register.
Trial registries were searched to identify any unpublished or ongoing studies. These registries included:
-
The WHO Trials portal (ICTRP);
-
ClinicalTrials.gov; and
-
The EU clinical trials register.
Conference proceedings were searched to identify studies published in abstract form. These conferences included:
-
Digestive Disease Week;
-
United European Gastroenterology Week;
-
European Crohn’s and Colitis Organisation; and
-
Advances in IBD.
The grey literature database Open Grey was searched to identify studies not indexed in the major databases.
Searching other resources
We searched the reference lists of included studies and applicable systematic reviews to identify studies missed by the database searches.
Data collection and analysis
Selection of studies
Two authors (AMW, JP) independently screened titles and abstracts identified by the search and excluded those studies not meeting the selection criteria. Full text reports were obtained for all the studies deemed eligible and were read independently by two review authors (AMW, JP). For the studies co‐authored by AMW, eligibility was assessed by other researchers (JP, SLP). If information pertaining to eligibility was missing, we contacted the authors of the studies for further information. In cases where the two authors could not reach consensus on study eligibility, a third investigator (SLP or SK) was consulted.
Data extraction and management
Data were independently extracted by two authors (AMW, JP or JP and SLP in the case of the trial co‐authored by AMW). Any disagreements were resolved by consensus and, if this could not be reached, a third author (SLP or SK) was asked to arbitrate.
The following information was extracted:
-
General study information: authors, year, country;
-
Method: design (including details such as: randomisation, allocation concealment, duration, follow‐up), setting, recruitment, intervention (type of antidepressant, dose, frequency, type of controls, adherence), clinical measures (e.g. disease activity measure, measures of anxiety/depression), sample size calculation;
-
Participants: number of participants, age, sex, IBD type, per cent in remission; and
-
Outcomes: descriptives (mean/SD or median/inter‐quartile range (or range), frequency (%) plus accompanying statistics, e.g. OR, P value) for primary and secondary outcome measures at time points, adverse events, and loss to follow‐up.
We contacted the authors of one study about missing or unclear information and the study authors provided the requested data (Chojnacki 2011).
Assessment of risk of bias in included studies
Two authors independently assessed the risk of bias. The variety of study designs included in this review necessitated the use of several different quality assessment tools. For RCTs, the Cochrane risk of bias tool was used (Higgins 2011). The following types of bias were examined: selection bias (sequence generation and allocation sequence concealment, two items), performance bias (blinding of participants and personnel, two items), detection bias (blinding of outcome assessment, one item), attrition bias (incomplete outcome data at short‐term (two to six weeks) and at long‐term (greater than six weeks, two items), reporting bias (selective outcome reporting, one item). Each item was rated as either ‘Low risk’, ‘High risk’ or 'Unclear risk'. For observational studies (case‐control), we used the Newcastle‐Ottawa Scale (Wells 2000), for which a study could score a possible of nine points, with a higher score consistent with better methodological quality.
In addition, the GRADE approach was used to evaluate the overall quality of the evidence supporting the primary outcomes and selected secondary outcomes (Guyatt 2008). Following the GRADE, evidence from randomised trials starts as high quality but may be downgraded due to within‐study risk of bias (methodological quality), indirect evidence, unexplained heterogeneity, imprecision of effect estimates and risk of publication bias. Evidence from non‐randomised studies starts as low quality. Each outcome was assigned one of the following scores: high quality (future research unlikely to change confidence in the estimate); moderate quality (future research likely to impact confidence in the estimate); low quality (future research very likely to impact confidence in the estimate); very low quality (the estimate is uncertain).
Summary of findings tables were prepared for the following outcomes post‐treatment:
-
Anxiety symptoms;
-
Depression symptoms;
-
Adverse events;
-
Quality of life;
-
Pain;
-
Clinical remission; and
-
Relapse.
Measures of treatment effect
We used the RevMan software for data analysis. For dichotomous outcomes, we calculated the risk ratio (RR) and corresponding 95% confidence interval (CI). The number needed to treat (NNT) and risk difference (RD) were calculated where appropriate. For continuous outcomes, the mean difference (MD) or standardised mean difference (SMD) and corresponding 95% CI were calculated.
Unit of analysis issues
Where the efficacy of multiple antidepressants (on IBD activity) was meant to be compared, it was planned to split the shared comparison group (e.g. standard care or psychotherapy) equally between the antidepressants arms as comparison groups. However, such a study was not identified. Cross‐over trials were to be included only when antidepressant and comparator data were extracted from the first treatment period or when the sufficient wash‐out period occurred between treatment periods (e.g. two weeks for all antidepressants except for fluoxetine where four weeks are required in light of the long plasma half‐life). However, no cross‐over trial was identified. SE was converted into SD using the following formula: SD = SE / √1/NE+1/NC.
Dealing with missing data
Where possible, the intention‐to‐treat principle was adhered to. In the case of dichotomous data when treatment response was compared, the total number of participants in each pre‐treatment comparison group (as the denominator) was included. In the analyses of treatment response, only the data from studies reporting a group size prior to drop‐outs were included. For continuous outcome measures, we included summary statistics derived from (in order of preference) mixed‐effects models, observed cases summary statistics, and last observation carried forward where possible. This was dictated by the notion that mixed‐effects models are considered less biased than the analyses of the last observation carried forward (Verbeke 2000).
Assessment of heterogeneity
We planned to assess clinical homogeneity using the forest plot of the risk ratio. We also planned to review the results of the Chi2 test. A P value of less than 0.10 was to be considered evidence of statistically significant heterogeneity (assuming the low power of the Chi2 statistic when few trials are available) (Deeks 2011). This proved impractical due to the very small number of studies identified.
The I2 statistic was used to assess heterogeneity across trials (Higgins 2003). An I2 statistic greater than 30% was considered moderate heterogeneity and greater than 50% was considered severe heterogeneity.
Subgroup differences in continuous measures of antidepressant efficacy were to be investigated using Deeks' stratified test of heterogeneity (Deeks 2001). Herein the sum of the Chi2 statistics for each of the subgroups included in the study is subtracted from the Chi2 statistic for all the studies, to provide a measure (Qb) of heterogeneity between groups. As different antidepressants may exert different effects, we planned to stratify all of the outcome comparisons by the individual antidepressant used (excluding subgroup and sensitivity analyses). This however proved impossible due to each study using a different type of antidepressant.
Assessment of reporting biases
Small‐sample effects were to be investigated by visual inspection of a funnel plot of treatment response (Sterne 2011). This was however deemed inappropriate as we identified fewer than 10 studies and the method is not robust in such cases (Egger 1997).
Data synthesis
The pooled RR and corresponding 95% CI was calculated for dichotomous outcomes. For continuous outcomes, the pooled MD or SMD with 95% CI was calculated as appropriate. It was planned to combine dichotomous and continuous variables using the standard Cochrane procedure (InOR = SMD X π / √ 3) (Deeks 2011), but this proved unnecessary. We obtained categorical and continuous treatment effects using a fixed‐effect model. The outcomes were expressed as an average effect size for each subgroup and 95% CIs. In some models, heterogeneity was present and in such cases random‐effects models are usually preferred. However, the Cochrane Handbook does warn that if the effect size is associated with sample size, then using a random‐effects model will award relatively more weight to the smaller studies, and will exacerbate bias (Deeks 2011). This is further confirmed by a recent evidence synthesis (Bender 2018). As this review includes a small number of studies, a fixed‐effect model was applied for the analyses.
Subgroup analysis and investigation of heterogeneity
Subgroup analysis was to be conducted for the following subgroups:
IBD subtype: Crohn's disease versus ulcerative colitis or indeterminate colitis;
Sex: Male versus female; and
Types of antidepressants: SSRI versus tricyclics.
This was not deemed practical due to the small number of studies.
Sensitivity analysis
Sensitivity analysis was to be performed to check the robustness of our conclusions for the meta‐analysis of the primary outcome. We planned to follow the same procedure as was applied in our previous protocol on a similar topic (Gordon 2013):
We planned to assess whether treatment response varies as a function of the use of treatment response versus non‐response as outcomes. Treatment response may produce less consistent outcome statistics than non‐response in cases when the control group event rate is greater than 50% (Deeks 2002). This analysis was only to be conducted if the majority of studies reported a control group event rate greater than 50%. This was not the case for the analysis.
Conducting a 'worst case/best case' analysis was considered to examine the impact of the exclusion of those lost to follow‐up on treatment efficacy effect estimates (Deeks 2011). Herein, for the worst case scenario, all the missing data for the treatment group were to be recorded as non‐responders. For the best case scenario, all missing data in the control group were to be considered non‐responders. Where the effect estimates of treatment efficacy would not differ between these two comparisons, it would be concluded that missing data in the studies did not have a marked impact on outcomes. This analysis was to be done in case we had access to full data sets for the included studies. This was the case for one study only (Mikocka‐Walus 2016c).
Results
Description of studies
Results of the search
The search was conducted on 23 August 2018 and identified 3920 records. After duplicates were removed, 3144 records were screened for inclusion. Of the studies that were screened, 16 were selected for full text review. Overall, 4 studies met the inclusion criteria (Figure 1). No additional studies were identified through other sources.
Study flow diagram.
Included studies
Country
Included studies came from four countries: one each conducted in Australia (Mikocka‐Walus 2016c), Iran (Daghaghzadeh 2015), Poland (Chojnacki 2011) and the United Kingdom (Goodhand 2012).
Study design
The search identified two double blind RCTs (Daghaghzadeh 2015; Mikocka‐Walus 2016c), one non‐randomised controlled trial (Chojnacki 2011), and one observational retrospective case‐matched study (Goodhand 2012).
Participant characteristics
The studies included a total of 188 patients with IBD (96 assigned to intervention and 93 assigned to controls). The age of participants ranged from 27 (Goodhand 2012) to 37.8 years (Daghaghzadeh 2015). The proportion of female participants ranged from 46% (Daghaghzadeh 2015; Mikocka‐Walus 2016c) to 65% (Chojnacki 2011). In three studies participants were in IBD remission (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c) and in one study participants had mixed IBD activity (Goodhand 2012). In one study participants had to have co‐morbid symptoms of anxiety and/or depression (Chojnacki 2011) while in the remaining studies this was not part of the inclusion criteria. Two studies included both participants with Crohn's disease and ulcerative colitis (Daghaghzadeh 2015; Goodhand 2012), one study included only participants with Crohn's disease (Mikocka‐Walus 2016c) and another only those with ulcerative colitis (Chojnacki 2011).
Treatment
In the experimental groups, one study used duloxetine 60 mg daily (an SNRI antidepressant) (Daghaghzadeh 2015), one study used fluoxetine 20 mg daily (an SSRI antidepressant) (Mikocka‐Walus 2016c), one study used tianeptine 12 mg three times a day (an atypical antidepressant) (Chojnacki 2011), and one study used various antidepressants in clinical ranges (Goodhand 2012). Three studies used a placebo control (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c), and one study used a no treatment control group matched for various clinical and demographic characteristics (Goodhand 2012).
Follow‐up
The follow‐up periods ranged from 12 weeks to 12 months, with one study (Goodhand 2012) observing participants 12 months before and 12 months after being prescribed an antidepressant. In two studies no attrition was recorded (Chojnacki 2011; Goodhand 2012), while in one study 79% of participants remained in the study at 12‐weeks of follow‐up (Daghaghzadeh 2015), and in another study 69% of participants remained in the study at 12‐months of follow‐up (Mikocka‐Walus 2016c).
Outcome measures
In terms of the primary outcome measures, three studies measured symptoms of anxiety and depression (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c). Symptoms of anxiety and depression were measured using the Hospital Anxiety and Depression Scale (HADS) in two studies (Daghaghzadeh 2015; Mikocka‐Walus 2016c), and the Hamilton Anxiety Rating Scale (HARS) and the Beck Depression Inventory (BDI) in one study (Chojnacki 2011).
Regarding the secondary outcome measures, three studies measured adverse events (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c), two studies measured study withdrawal due to adverse events and QoL (Mikocka‐Walus 2016c; Daghaghzadeh 2015). All studies measured IBD activity. Three studies used an IBD activity index (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c), two studies used blood tests ( Chojnacki 2011; Mikocka‐Walus 2016c), one study used faecal calprotectin (Mikocka‐Walus 2016c), and one study used endoscopy (Chojnacki 2011).
QoL was measured using the World Health Organization Quality of Life (WHOQOL‐BREF) questionnaire (Daghaghzadeh 2015; Mikocka‐Walus 2016c). The WHOQOL‐BREF is a short version of the World Health Organization Quality of Life (WHOQOL‐100) questionnaire and is a tool which can be used cross‐culturally to evaluate quality of life (WHOQOL‐BREF). WHOQOL‐BREF measures four major domains of QoL: physical (corresponding with physical health, e.g. fatigue, pain, sleep), psychological (corresponding with psychological well‐being, e.g. self‐esteem, body image, positive or negative feelings), social relationships (corresponding with personal relationships, social support and sexual functioning) and environment (corresponding with people's relationship to their environment, e.g. safety, financial resources, transport, physical environment).
Disease activity indices included the Crohn's Disease Activity Index (CDAI) (Mikocka‐Walus 2016c), the Lichtiger Colitis Activity Index (Daghaghzadeh 2015), and the Mayo Clinic Disease Activity Index (Chojnacki 2011). The blood tests included C‐reactive protein (CRP) (Chojnacki 2011), and cytokines/chemokines (Mikocka‐Walus 2016c).
One study measured hospital admissions and need for steroid treatment (Goodhand 2012). None of the studies measured pain or surgery.
For details of studies see Characteristics of included studies.
Excluded studies
Studies were excluded for not meeting the inclusion criteria of study design (Drossmann 2014; Eirund 1998), presenting data overlapping with another paper (Iskandar 2012; Iskandar 2011), lack of information regarding the efficacy of antidepressants (Loftus 2011; Virta 2014), including combination therapy without separate data on antidepressant efficacy (Xie 2014; NCT02162862), no validated measure of outcomes (Mikocka‐Walus 2016b), no control group (Yanartas 2016), a control group not comprised of IBD patients (Iskandar 2014), and a trial registration without published results (NCT00126373). See Characteristics of excluded studies.
Risk of bias in included studies
The results of the risk of bias analysis for the three controlled trials are summarized in Figure 2 and Figure 3 (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c). Table 1 reports the Newcastle‐Ottawa Scale results for the observational study (Goodhand 2012). Mikocka‐Walus 2016c was rated as low risk of bias. Daghaghzadeh 2015 was rated as high risk of bias for incomplete outcome data. The non‐randomised controlled trial was associated with low risk of bias on two items, unclear risk on two items (blinding of outcome assessment and selective reporting) and high risk on three items (random sequence generation, allocation concealment, blinding of participants and personnel) (Chojnacki 2011).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
| Case Control Study | Is the case definition adequate? (/1) | Representativeness of the cases (/1) | Selection of controls (/1) | Definition of controls (/1) | Comparability of cases and controls on the basis of the design or analysis (/2) | Assessment of exposure (/1) | Same method of ascertainment for cases and controls (/1) | Non‐Response Rate (/1) | |
| Goodhand (2012) | 1 | ‐ | 1 | ‐ | 2 | 1 | 1 | 1 | 7 |
The observational study (Goodhand 2012), was considered to be of reasonable methodological quality and was given a score of seven stars. However, there were concerns with two items: representativeness of the cases and definition of controls. Cases were not completely representative as some of them were excluded based on the lack of data (e.g. when the date of commencement of the antidepressant was missing). The definition of controls was considered incomplete as it did not mention the history of outcome in this group. Importantly, this study was non‐randomised and as such is likely to be associated with a higher risk of bias than RCTs.
Allocation
Random sequence generation was rated as low risk of bias in two studies (Daghaghzadeh 2015; Mikocka‐Walus 2016c), and as high risk in the non‐randomised controlled trial (Chojnacki 2011). Allocation concealment was rated as high risk of bias in one study (Chojnacki 2011), as low risk in two studies (Daghaghzadeh 2015; Mikocka‐Walus 2016c).
Blinding
Blinding of participants and personnel was judged to be adequate in two studies (Daghaghzadeh 2015; Mikocka‐Walus 2016c), and high risk of bias in one study (Chojnacki 2011). Blinding of outcome assessment was judged to be adequate in two studies (Daghaghzadeh 2015; Mikocka‐Walus 2016c), and unclear risk of bias in one study (Chojnacki 2011).
Incomplete outcome data
One study was judged to be at high risk of bias for Incomplete outcome data (Daghaghzadeh 2015), while two studies were judged to be at low risk of bias for this item (Chojnacki 2011; Mikocka‐Walus 2016c).
Selective reporting
Selective reporting was considered at low risk of bias in two studies (Daghaghzadeh 2015; Mikocka‐Walus 2016c), and unclear risk of bias in one study (Chojnacki 2011).
Other potential sources of bias
The three trials were considered at low risk of bias for other types of bias (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c) .
Effects of interventions
Comparison 1: Antidepressants versus placebo
Overall, four studies have contributed to this comparison (Chojnacki 2011; Daghaghzadeh 2015; Goodhand 2012; Mikocka‐Walus 2016c).
Primary outcome measures
Anxiety
Three studies examined the effect of antidepressants on symptoms of anxiety (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c).
At 12 weeks, using the HADS, Daghaghzadeh 2015 reported a mean score of 6.11 + 3 in the antidepressant group (n = 22) compared to 8.5 + 3.45 in the placebo group (n = 22) (MD ‐2.39, 95% ‐4.30 to ‐0.48; low certainty evidence, see Analysis 1.1 and summary of findings Table for the main comparison).
At 12 months, two trials reported that symptoms of anxiety were improved in participants receiving antidepressants compared to placebo (Chojnacki 2011; Mikocka‐Walus 2016c). We initially attempted to pool these studies using the SMD but a very high degree of heterogeneity was detected (I2= 87%). Thus we report the results for each trial separately. Using the HARS, Chojnacki 2011 reported a mean score of 12.65 + 3.76 in the antidepressant group (n = 30) compared to 17.85 + 3.33 in the placebo group (n = 30) (MD ‐5.20, 95% CI ‐7 to ‐3.40; very low certainty evidence). Using the HADS, Mikocka‐Walus 2016c reported a mean score of 3.8 + 2.5 in the antidepressant group (n = 14) compared to 4.2 + 4.9 in the placebo group (n = 12) (MD ‐0.40, 95% ‐3.47 to 2.67, low certainty evidence; See Analysis 1.2 and summary of findings Table for the main comparison).
Depression
Three studies examined the effect of antidepressants on symptoms of depression (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c).
At 12 weeks, using the HADS, Daghaghzadeh 2015 reported a mean score of 7.47 + 2.42 in the antidepressant group (n = 22) compared to 10.5 + 3.57 in the placebo group (n = 22) (MD ‐3.03, 95% CI ‐4.83 to ‐1.23, low certainty evidence; see Analysis 1.3 and summary of findings Table for the main comparison).
At 12 months, two trials reported that symptoms of depression were improved in participants receiving antidepressants compared to placebo (Chojnacki 2011; Mikocka‐Walus 2016c). We initially attempted to pool these studies using the SMD but a very high degree of heterogeneity was detected (I2= 89%). Thus we report the results for each trial separately. Using the BDI, Chojnacki 2011 reported a mean score of 9.6 + 2.76 in the antidepressant group (n = 30) compared to 16.35 + 5.41 in the placebo group (n = 30) (MD ‐6.75, 95% CI ‐8.92 to ‐4.58; very low certainty evidence). Using the HADS, Mikocka‐Walus 2016c reported a mean score of 2.9 + 2.8 in the antidepressant group (n = 14) compared to 3.1 + 3.4 in the placebo group (n = 12) (MD ‐0.20, 95% ‐2.62 to 2.22, low certainty evidence; See Analysis 1.4 and summary of findings Table for the main comparison).
Secondary outcome measures
Adverse events
Adverse events were reported in three studies, with nausea being an adverse event common to all three studies (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c).
At 12 weeks, higher rates of nausea were reported in the antidepressant group compared to placebo (Daghaghzadeh 2015). Thirty‐two per cent (7/22) of participants in the antidepressant group reported nausea compared to nine per cent (2/22) of placebo participants (RR 3.50, 95% CI 0.82 to 15.01; very low certainty evidence). The very low GRADE rating was due to a small sample size and incomplete outcome data.
At 12 months, two trials showed no group difference in nausea between those taking antidepressants and placebo (Chojnacki 2011; Mikocka‐Walus 2016c). Thirteen per cent (6/44) of those taking antidepressants reported nausea compared to two per cent (1/42) of placebo participants (RR 4.02, 95% CI 0.74 to 22.03).
Adverse events in the group who received antidepressants included nausea, headache, dizziness, drowsiness, sexual problems, insomnia, fatigue, low mood/anxiety, dry mouth, poor sleep, restless legs and hot flushes. Adverse events in the control group included dizziness, insomnia and muscle spasms (Daghaghzadeh 2015; Mikocka‐Walus 2016c).
Mikocka‐Walus 2016c reported on the number of participants who had an adverse event. Fifty‐seven per cent (8/14) of those in the antidepressant group reported adverse events compared to 25% (3/12) of the placebo group (RR 2.29, 95% CI 0.78 to 6.73, low certainty evidence; See Analysis 1.5 and summary of findings Table for the main comparison).
Serious adverse events
Serious adverse events were not reported by the included studies.
Study withdrawal due to adverse events
One RCT examined the effect of antidepressants on study withdrawal due to adverse events at 12 weeks (Daghaghzadeh 2015), and one RCT examined the effect of antidepressants on study withdrawal due to adverse events at 12 months (Mikocka‐Walus 2016c).
At 12 weeks, no group difference in study withdrawal due to adverse events was observed, with 4% (1/22) of participants taking antidepressants withdrawing from the study due to adverse events (adverse event type not reported) compared to 0% (0/22) of placebo group participants (RR 3, 95% CI 0.13 to 69.9; see Analysis 1.8).
At 12 months, no group difference in study withdrawal due to adverse events was observed, with 7% (1/14) of participant taking antidepressants withdrawing from the study due to adverse events (including poor sleep, anxiety, restless legs) compared to 0% (0/12) of placebo group participants (RR 2.6, 95% CI 0.12 to 58.5; see Analysis 1.9).
Quality of life
One RCT examined the effect of antidepressants on QoL at 12 weeks (Daghaghzadeh 2015) and one RCT examined the effect of antidepressants on QoL at 12 months (Mikocka‐Walus 2016c). Both studies used the WHOQOL‐BREF questionnaire.
Physical QoL
At 12 weeks, Daghaghzadeh 2015 reported a mean score of 60.24 + 12.94 in the antidepressant group (n = 22) compared to 49.52 + 10.12 in the placebo group (n = 22) (MD 10.72, 95% CI 3.86 to 17.58, low certainty evidence; See Analysis 1.10 and summary of findings Table for the main comparison).
At 12 months, Mikocka‐Walus 2016c reported a mean score of 68.83 + 13.34 in the antidepressant group (n = 14) compared to 66.66 + 21.72 in the placebo group (n = 12) (MD 2.17, 95% CI ‐11.97 to 16.31, low certainty evidence; See Analysis 1.11 and summary of findings Table for the main comparison).
Psychological QoL
At 12 weeks, Daghaghzadeh 2015 reported a mean score of 51.81 + 13.6 in the antidepressant group (n = 22) compared to 43.5 + 11.94 in the placebo group (n = 22) (MD 8.31, 95% CI 0.75 to 15.87, low certainty evidence; See Analysis 1.12 and summary of findings Table for the main comparison).
At 12 months, Mikocka‐Walus 2016c reported a mean score of 75.37 + 14.84 in the antidepressant group (n = 14) compared to 72.22 + 16.79 in the placebo group (n = 12) (MD 3.15, 95% CI ‐9.12 to 15.42, low certainty evidence; See Analysis 1.13 and summary of findings Table for the main comparison).
Social QoL
At 12 weeks, Daghaghzadeh 2015 reported a mean score of 51.2 + 15.1 in the antidepressant group (n = 22) compared to 38.88 + 12.12 in the placebo group (n = 22) (MD 12.32, 95% CI 4.23 to 20.41, low certainty evidence; See Analysis 1.14 and summary of findings Table for the main comparison).
At 12 months, Mikocka‐Walus 2016c reported a mean score of 73.48 + 18.56 in the antidepressant group (n = 14) compared to 75 + 23.19 in the placebo group (n = 12) (MD ‐1.52, 95% CI ‐17.85 to 14.81, low certainty evidence; See Analysis 1.15 and summary of findings Table for the main comparison).
Environmental QoL
At 12 weeks, Daghaghzadeh 2015 reported a mean score of 51.79 + 10.24 in the antidepressant group (n = 22) compared to 44.13 + 12.27 in the placebo group (n = 22) (MD 7.66, 95% CI 0.98 to 14.34, low certainty evidence; See Analysis 1.16 and summary of findings Table for the main comparison).
At 12 months, Mikocka‐Walus 2016c reported a mean score of 73.86 + 14.41 in the antidepressant group (n = 14) compared to 75.69 + 9.85 in the placebo group (n = 12) (MD ‐1.83, 95% CI ‐11.21 to 7.55; low certainty evidence; See Analysis 1.17 and summary of findings Table for the main comparison).
The low GRADE rating at both 12 weeks and 12 months was due to very serious imprecision (26 participants) in one study (Mikocka‐Walus 2016c), and incomplete outcome data and imprecision (44 participants) in the other study (Daghaghzadeh 2015).
Clinical remission
One trial reported on remission rates at 12 months post treatment (Mikocka‐Walus 2016c). In the group receiving an antidepressant, 64% (9/14) of participants remained in remission (based on CDAI and faecal calprotectin) compared to 66% (8/12) in the placebo group (RR 0.96, 95% CI 0.55 to 1.69, low certainty evidence; see Analysis 1.18 and summary of findings Table for the main comparison).
Three studies used a disease activity index to measure disease activity (Chojnacki 2011; Daghaghzadeh 2015; Mikocka‐Walus 2016c). This post hoc outcome was not pre‐specified in our protocol.
At 12 weeks, using the Lichtiger Colitis Activity Index, Daghaghzadeh 2015 reported a mean score of 4.52 + 11.63 in the antidepressant group (n = 22) compared to 6.83 + 2.09 in the placebo group (n = 22) (MD ‐2.31, 95% CI ‐3.42 to ‐1.20; See Analysis 1.19).
At 12 months, two trials reported that disease activity was improved in the group taking antidepressants as compared to placebo (Chojnacki 2011; Mikocka‐Walus 2016c). We initially attempted to pool these studies using the SMD but a very high degree of heterogeneity was detected (I2= 87%). Thus we report the results for each trial separately. Using the Mayo Clinic Disease Activity Index, Chojnacki 2011 reported a mean score of 3.05 + 1.36 in the antidepressant group (n = 30) compared to 4.65 + 1.69 in the placebo group (n = 30) (MD ‐1.60, 95% CI ‐2.38 to ‐0.82; See Analysis 1.20). Using the CDAI, Mikocka‐Walus 2016c reported a mean score of 84.4 + 82.5 in the antidepressant group compared (n = 14) to 60.63 + 46.5 in the placebo group (n = 12) (MD 23.77, 95% CI ‐26.82 to 74.36; See Analysis 1.20).
No study collected data on clinical remission at longitudinal follow‐up beyond trial completion.
Relapse
One non‐randomised trial reported on endoscopic relapse up to 12 months post treatment (Chojnacki 2011). At 12 months, 0% (0/30) of participants in the antidepressant group had endoscopic relapse compared to 10% (3/30) of placebo group participants (RR 0.14, 95% CI 0.01 to 2.65, very low certainty evidence; see Analysis 1.21 and summary of findings Table for the main comparison).
The following relevant post hoc outcomes were not pre‐specified in our review protocol: relapse using clinician's assessment, faecal calprotectin and blood tests (CRP, cytokines/chemokines).
Goodhand 2012 reported on the number of relapses (clinician assessed based on symptoms/blood tests) in the year preceding treatment with antidepressants and in the year after the treatment commenced. In the year after starting an antidepressant, patients treated with an antidepressant had fewer relapses than controls (median[range] = 0 [0–4) versus 1 [0–3]).
In one trial (Mikocka‐Walus 2016c) there was no group difference in the relapse rate as measured using faecal calprotectin at 12 months (post treatment), with 7% (1/14) of the participants in the antidepressant group relapsing (faecal calprotectin > 200) compared to 0% (0/12) of the placebo group (MD 2.60, 85% CI 0.12 to 58.48; See Analysis 1.22).
Regarding blood tests, one trial reported data on CRP (Chojnacki 2011), while another trial reported data for cytokines and chemokines (Mikocka‐Walus 2016c). Chojnacki 2011 reported a mean CRP of 6.99 + 5.65 in the antidepressant group (n = 30) compared to 9.40 + 6.78 in the placebo group (n = 30) (MD ‐2.41, 95% CI ‐5.57 to 0.75; See Analysis 1.23).
Mikocka‐Walus 2016c reported a mean proportion of TH Effector Memory RA cells of 45.8 + 4.5 in the antidepressant group (n = 14) compared to 39.7 + 3.1 in the placebo group (n = 12) (MD 6.10, 95% CI 3.16 to 9.04; See Analysis 1.24).
Mikocka‐Walus 2016c reported a mean proportion of TC Effector Memory RA cells of 3.5 + 0.48 in the antidepressant group (n = 14) compared to 4.75 + 0.9 in the placebo group (n = 12) (MD ‐1.25, 95% CI ‐1.82 to ‐0.68; See Analysis 1.25).
Regarding interleukin‐10 (IL‐10) secretion, Mikocka‐Walus 2016c reported a mean of CD3/CD28 stimulated cytokine concentrations in peripheral blood mononuclear cells supernatants of 525.3 + 93.2 in the antidepressant group (n = 14) compared to 222.9 + 63.2 in the placebo group (n = 12) (MD 302.4, 95% CI 241.89 to 362.91; See Analysis 1.26).
No study collected data on relapse at longitudinal follow‐up beyond trial completion.
Pain severity
None of the included studies examined the impact of antidepressants on pain.
Hospital admissions
Hospital admissions were included as an outcome in only one study (Goodhand 2012), and thus a meta‐analysis was not conducted. At 12‐month follow‐up, no participants in either group had hospital admissions due to IBD.
Surgery
The included studies did not examine the impact of antidepressants on the need for surgery.
Need for steroid treatment
The need for steroid treatment was included as an outcome in only one study (Goodhand 2012). At 1‐year follow‐up, no participants in the antidepressant group (0/29) required steroids compared to 3% (1/29) of those in the control group (MD 0.33, 95% CI 0.01 to 7.86; See Analysis 1.27).
Discussion
Up to 30% of people with IBD take antidepressants (Fuller‐Thomson 2006; Haapamaki 2013; Mikocka‐Walus 2012). Despite the clinical relevance of the present topic, there were only four studies (examining 188 people in total) of sufficient quality to include in this systematic review.
The review cautiously suggests that antidepressants improved the symptoms of anxiety and depression. There was no group difference in nausea or study withdrawal due to adverse events. Antidepressants were associated with some benefits for QoL and disease activity. However, the GRADE analysis indicated that the overall certainty of the evidence was very low, due to a small sample size, incomplete outcome data, and heterogeneity in population and antidepressant treatment type, thus more well‐designed studies are needed. Future trials examining the role of antidepressants in IBD are therefore needed to clarify whether the present findings are consistent.
Summary of main results
Up to 30% of people with IBD take antidepressants (Fuller‐Thomson 2006; Haapamaki 2013; Mikocka‐Walus 2012). Despite the clinical relevance of the present topic, there were only four studies (including 188 participants) meeting the inclusion criteria (Chojnacki 2011; Daghaghzadeh 2015; Goodhand 2012; Mikocka‐Walus 2016c). Two studies were double‐blind RCTs (Daghaghzadeh 2015; Mikocka‐Walus 2016c). One study was non‐randomised controlled trial (Chojnacki 2011), and the final study was an observational retrospective case‐matched study (Goodhand 2012).
Symptoms of anxiety and depression were improved at 12 weeks and 12 months in antidepressant participants compared to placebo. There were no group differences in adverse events at 12 months or study withdrawal due to adverse events at 12 weeks or 12 months. Physical, Psychological, Social and Environmental QoL were improved at 12 weeks with no group differences at 12 months. Disease activity as measured by disease activity indices was also improved in the group receiving antidepressants. However, there was no group difference in clinical remission at 12 months (based on the CDAI and faecal calprotectin), or relapse rate at 12 months (based on endoscopy or faecal calprotectin). There were no group differences in hospital admissions or need for steroid treatment. Pain severity or surgery were not reported in the included studies.
Overall completeness and applicability of evidence
The results of this review are applicable to adults with Crohn's disease and ulcerative colitis, though at this point it is unclear if patients with either IBD subtype may benefit more from antidepressant treatment. The studies included in this review assessed different IBD populations. For example, one study limited the intervention to the participants reporting symptoms of anxiety and depression (Chojnacki 2011), while the other studies did not. One study included participants with mixed disease activity (Goodhand 2012), while the other studies included participants who were in remission. Two studies included participants with Crohn's disease and ulcerative colitis (Daghaghzadeh 2015; Goodhand 2012), while the other studies examined just one IBD subtype. The overall evidence base is not complete. All of the included studies had small sample sizes and we were unable to collect data for some of our pre‐specified outcomes (e.g. pain severity and surgery). Several outcomes were only reported by one study (e.g. CRP, cytokines, faecal calprotectin, endoscopic relapse, hospital admissions, need for steroids). The four studies assessed different classes of antidepressants, thus the evidence supporting the use of any particular type of antidepressant is sparse. The certainty of this evidence was very low and further studies are needed before firm conclusions can be drawn.
Quality of the evidence
One RCT was rated as low risk of bias (Mikocka‐Walus 2016c). The other RCT was rated as high risk of bias for incomplete outcome data. The non‐randomised controlled trial was rated as high risk of bias for random sequence generation, allocation concealment, and blinding of participants and personnel. Although the observational study scored well on the Newcastle‐Ottawa scale, it is still considered to be at high risk of bias in comparison to RCTs given its observational design (Goodhand 2012).
The GRADE analysis indicated that the overall certainty of the evidence supporting the outcomes of anxiety, depression, QoL, adverse events, and disease activity was low or very low due to very serious imprecision and high risk of bias (incomplete outcome data) in one study.
Potential biases in the review process
Measures were taken to ensure the reviewers who co‐authored one of the included trials would not extract data or assess study quality (Mikocka‐Walus 2016c). Authors not involved in the previous trial (SLP, SK, JP) undertook this task. All studies were assessed for inclusion by two independent authors and any disagreements were resolved by a third author. All data were extracted independently by two authors. Further, to reduce any language bias, language restrictions were not imposed on the current review and the included Polish study was translated (Chojnacki 2011), as well as the two non‐English excluded studies (Eirund 1998; Xie 2014).
The limitations of the present review include the deviation from an RCT design usually used in effectiveness reviews. We decided to broaden our inclusion criteria to include non‐randomised studies in order to increase the number of included studies in the review. We decided against conducting a subgroup analysis based on type of IBD (i.e. Crohn's disease or ulcerative colitis) due to the very small sample size per comparison group. We also decided to use a fixed‐effect model for our analysis even when heterogeneity was considerable. We realise this decision may be controversial, but it was dictated by the desire to reduce bias inherent in reviews including studies with small sample sizes. While we attempted a meta‐analysis, only the data for nausea at 12 months could be combined as the heterogeneity was low. All other data where two studies are presented on the forest plot could not be combined due to high levels of heterogeneity.
Agreements and disagreements with other studies or reviews
The results of this review agree with the two previous reviews which relied on data synthesis only (Macer 2017; Mikocka‐Walus 2006). The present review is the first attempt at a meta‐analysis in the area.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Comparison 1 Antidepressants versus placebo, Outcome 1 Anxiety at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 2 Anxiety at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 3 Depression at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 4 Depression at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 5 Adverse events at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 6 Adverse events: nausea at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 7 Adverse events: nausea at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 8 Study withdrawal due to adverse events at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 9 Study withdrawal due to adverse events at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 10 Physical QoL at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 11 Physical QoL at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 12 Psychological QoL at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 13 Psychological QoL at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 14 Social QoL at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 15 Social QoL at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 16 Environmental QoL at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 17 Environmental QoL at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 18 Clinical remission at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 19 Disease activity at 12 weeks.
Comparison 1 Antidepressants versus placebo, Outcome 20 Disease activity at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 21 Endoscopic relapse at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 22 Relapse using faecal calprotectin at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 23 CRP at 12 months.
Comparison 1 Antidepressants versus placebo, Outcome 24 Cytokines TH Effector Memory RA at 6 months.
Comparison 1 Antidepressants versus placebo, Outcome 25 Cytokines TC Effector Memory RA at 6 months.
Comparison 1 Antidepressants versus placebo, Outcome 26 Interleukin‐10 at 6 months.
Comparison 1 Antidepressants versus placebo, Outcome 27 Need for steroids at 12 months.
| Antidepressants compared to placebo for inflammatory bowel disease | ||||||
| Patient or population: participants with active and inactive inflammatory bowel disease | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | № of participants | Certainty of the evidence | Comments | |
| Risk with Placebo | Risk with Antidepressants | |||||
| Anxiety at 12 weeks | The mean anxiety was 8.5 (SD = 3.45) | The mean anxiety was 6.11 (SD = 3) MD 2.39 lower | ‐ | 44 | ⊕⊕⊝⊝ | Anxiety was assessed using the HADS |
| Anxiety at 12 months | The mean anxiety was 4.2 (SD = 4.9) | The mean anxiety was 3.8 (SD = 2.5) MD ‐0.40 lower | ‐ | 26 | ⊕⊕⊝⊝ | Anxiety was assessed using the HADS A second non‐randomised study using the HARS reported a mean score of 12.65 + 3.76 in the antidepressant group (n = 30) compared to 17.85 + 3.33 in the placebo group (n = 30) (MD ‐5.20, 95% CI ‐7 to ‐3.40; very low certainty evidence) |
| Depression at 12 weeks | The mean depression was 10.5 (SD = 3.57) | The mean depression was 7.47 (SD = 2.42) MD ‐3.03 lower | ‐ | 44 | ⊕⊕⊝⊝ | Depression was assessed using the HADS |
| Depression at 12 months | The mean depression was 3.1 (SD = 3.4) | The mean depression was 2.9 (SD = 2.8) MD ‐0.20 lower | ‐ | 26 | ⊕⊕⊝⊝ | Depression was assessed using the HADS A second non‐randomised study using the Beck Depression Inventory reported a mean score of 9.6 + 2.76 in the antidepressant group (n = 30) compared to 16.35 + 5.41 in the placebo group (n = 30) (MD ‐6.75, 95% CI ‐8.92 to ‐4.58; very low certainty evidence) |
| Adverse events at 12 months | 250 per 1,000 | 573 per 1,000 | RR 2.29 | 26 | ⊕⊕⊝⊝ | Commonly reported adverse events include nausea, headache, dizziness, drowsiness, sexual problems, insomnia, fatigue, low mood/anxiety, dry mouth muscle spasms and hot flushes None of the included studies reported any serious adverse events |
| Quality of life | ‐ | ‐ | See comment | 70 | ⊕⊕⊝⊝ | Quality of life was assessed using the WHO‐QOL‐BREF We were unable to pool data as the outcome was reported at 12 weeks in 1 trial (44 participants) and 12 months in 1 trial (26 participants). Physical, Psychological, Social and Environmental QoL were improved only at 12 weeks with no group difference at 12 months |
| Pain | Not reported | No studies reported this outcome | ||||
| Maintenance of remission at 12 months | 667 per 1,000 | 640 per 1,000 | RR 0.96 | 26 | ⊕⊕⊝⊝ | Maintenance of remission was measured by the CDAI (< 150) and fecal calprotectin levels |
| Endoscopic relapse at 12 months | 0 per 1,000 | 0 per 1,000 | RR 0.14 | 60 | ⊕⊝⊝⊝ | Non‐randomised study. We were unable to calculate absolute effects. Endoscopic relapse occurred in 0% (0/30) of participants in the antidepressants group compared to 10% (3/30) in the placebo group |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 Downgraded one level due to serious imprecision (44 participants) 2 Downgraded one level due to high risk of bias (incomplete outcome data) 3 Downgraded two levels due very serious imprecision (26 participants) 4 Downgraded two levels due very serious imprecision (11 events) 5 Downgraded two levels due very serious imprecision (17 events). 6 Downgraded one level due to very serious imprecision (3 events) and a large CI around the point estimate | ||||||
| Case Control Study | Is the case definition adequate? (/1) | Representativeness of the cases (/1) | Selection of controls (/1) | Definition of controls (/1) | Comparability of cases and controls on the basis of the design or analysis (/2) | Assessment of exposure (/1) | Same method of ascertainment for cases and controls (/1) | Non‐Response Rate (/1) | |
| Goodhand (2012) | 1 | ‐ | 1 | ‐ | 2 | 1 | 1 | 1 | 7 |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Anxiety at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 2 Anxiety at 12 months Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 3 Depression at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 4 Depression at 12 months Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 5 Adverse events at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 6 Adverse events: nausea at 12 weeks Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 7 Adverse events: nausea at 12 months Show forest plot | 2 | 86 | Risk Ratio (M‐H, Fixed, 95% CI) | 4.02 [0.74, 22.03] |
| 8 Study withdrawal due to adverse events at 12 weeks Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 9 Study withdrawal due to adverse events at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 10 Physical QoL at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 11 Physical QoL at 12 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 12 Psychological QoL at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 13 Psychological QoL at 12 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 14 Social QoL at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 15 Social QoL at 12 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 16 Environmental QoL at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 17 Environmental QoL at 12 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 18 Clinical remission at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 19 Disease activity at 12 weeks Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 20 Disease activity at 12 months Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 21 Endoscopic relapse at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 22 Relapse using faecal calprotectin at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 23 CRP at 12 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 24 Cytokines TH Effector Memory RA at 6 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 25 Cytokines TC Effector Memory RA at 6 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 26 Interleukin‐10 at 6 months Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| 27 Need for steroids at 12 months Show forest plot | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
