Results of the search

We completed the literature search on the 28 April March 2021 (Appendix 1), identifying a total of 2969 records through database searching. We found 17 additional records from alternative sources. After removal of duplicates 2576 unique records remained. Examination of the titles and abstracts found 67 records for full‐text screening. After assessing all 68 records, we identified six records of five studies that met the inclusion criteria and included them in the review. There were also eight records of eight ongoing studies and 16 records of 13 studies awaiting classification. We excluded 37 records of 32 studies for various reasons. We present the results of the search in the PRISMA flow diagram (Figure 1).

Figure 1

---

---

Study flow diagram.

Included studies

Excluded studies

We excluded 33 studies for various reasons. The reasons for exclusion of each study are presented in the excluded studies table and are summarised below.

• 8 studies were excluded as having ineligible outcomes (ACTRN12617000876392; ACTRN12619000150145; Chen 2015b; Dai 2017; Engel 2016; Gibson 2013; Pullan 1994; Tripp 2017)

• 6 studies were excluded for mixed IBD population without separate UC data. We contacted the authors for separate data and they either did not respond or were unable to provide it (Berrill 2014; Mizrahi 2012; Ozgursoy Uran 2019; Tapete 2018; Tapete 2019; Volz 2016)

• 5 studies were excluded for ineligible interventions (Chen 2015a; Collawn 1992; Danese 2019 ; Faghfoori 2014; Ghosh 2018)

• 2 studies were excluded for not being RCTs (McCormick 2010; Spagnuolo 2017)

• 5 studies were excluded for ineligible indication (Hallert 2003; Hanauer 1993; Huang 2013; ISRCTN98226923; Johari 2016)

• 3 studies were excluded for ineligible participant population (Cohen 1999; NCT02763293; Zai 2018)

• 3 studies were excluded for ineligible study design (Bae 2016; Forbes 2019; Gearry 2009)

Allocation

Randomisation was described clearly in three of the five studies (Cox 2020; Sharma 2015; Yilmaz 2019), which we rated low for risk of bias, and was not sufficiently described in two studies (Shaw 1987; Zhao 2017) which we rated unclear for risk of bias.

We rated two studies at low risk of selection bias (Cox 2020; Sharma 2015), as the method of random allocation of participants to intervention and control groups and allocation concealment was described and we judged it to be adequate. We rated the other three studies at unclear risk of selection bias and allocation concealment (Shaw 1987; Yilmaz 2019; Zhao 2017), as they did not provide enough information (or none at all) about their selection and allocation concealment process.

Blinding

All studies were rated as high in performance bias, as the interventions they studied could not be blinded for both participants and personnel or they were open‐label studies. However, Cox 2020 used a sham diet to keep their participants blind to the intervention, which is not typical in diet RCTs due to the difficulties entailed.

Detection bias was rated as low in one study (Yilmaz 2019) which provided details about it, and unclear in the other studies, as they did not provide enough information for a judgement (Cox 2020; Sharma 2015; Shaw 1987; Zhao 2017).

Incomplete outcome data

Attrition bias was judged as low in four studies that provided enough information for judgement (Cox 2020; Sharma 2015; Shaw 1987; Yilmaz 2019). Zhao 2017 was rated as unclear.

Selective reporting

Reporting bias was rated as low in four studies that reported all outcomes they had set out to report (Cox 2020; Shaw 1987; Yilmaz 2019; Zhao 2017). Sharma 2015 lacked enough information in their report to judge whether they had reported all outcomes, and was rated as unclear.

Other potential sources of bias

We rated four studies as low in other potential sources of bias (Cox 2020; Shaw 1987; Sharma 2015; Yilmaz 2019). We rated Zhao 2017 as having high potential for other sources of bias because of significant differences in their participants' baseline characteristics that were highly likely to affect the results (Zhao 2017).

Funding source and conflict of interest

Three studies reported their sources of funding (Cox 2020; Sharma 2015; Yilmaz 2019). Sharma 2015 was funded via a government grant, Cox 2020 was funded by private foundations, and Yilmaz 2019 reported that they received no funding.

Four studies made declarations on conflicts of interest (Cox 2020; Sharma 2015; Yilmaz 2019; Zhao 2017). Three declared no conflicts of interest (Sharma 2015; Yilmaz 2019; Zhao 2017) and Cox 2020 declared industry connections and ownership of an invention connected to their intervention.

1. Low FODMAPs diet versus sham diet

Cox 2020 (n = 52) compared a diet low in FODMAPs to a sham diet. It included participants with either CD or UC who were at an inactive stage of their disease, which was defined by all of the following: physician global assessment, stable medications, no IBD flare in the previous six months, faecal calprotectin less than 250 mg/g, and serum C‐reactive protein (CRP) less than 10 mg/L. The intervention lasted for four weeks.

Primary outcomes

Treatment success was not reported.

Pain was measured using the pain subscale of the IBS‐SSS that rates pain on a scale of 0 to 100 and the GSRS pain rating scale of 0 to 3.

Pain frequency

Pain frequency was measured.

At end of study, for the 13 UC participants in the 'low FODMAPs' group, the mean (SD) IBS‐SSS pain frequency in days was 31 (21.6) days and for the 13 participants in the sham‐diet group 35 (21.6). There was no clear difference in days of pain for CD participants when a low FODMAPs diet was compared to a sham diet in Cox 2020 (MD −4.00, 95% CI −20.61 to 12.61). The certainty of evidence was very low, due to risk of bias and imprecision (Analysis 1.1; summary of findings Table 1).

Separate UC data for the GSRS scale was requested but not provided by the authors.

Pain intensity

Pain intensity was measured.

At end of study, for the 13 UC participants in the 'low FODMAPs' group the mean (SD) IBS‐SSS pain intensity was 20 (14.4) and for the 13 participants in the sham‐diet group 29 (14.4). There was no clear difference in intensity of pain for CD participants when a low FODMAPs diet was compared to a sham diet in Cox 2020 (MD −9.00, 95% CI −20.07 to 2.07). The certainty of evidence was very low, due to risk of bias and imprecision (Analysis 1.2; summary of findings Table 1).

Separate UC data for the GSRS scale was requested but not provided by the authors.

Withdrawals due to adverse events

There were two withdrawals due to adverse events in the 'low FODMAPs' group (one IBD relapse, one beginning antibiotics) and one in the sham‐diet group (IBD relapse). There was no clear difference in withdrawals due to adverse effects between the 'low FODMAPs' and sham‐diet groups (RR 1.85, 95% CI 0.18 to 19.19; 52 participants). The certainty of evidence was very low due to risk of bias and imprecision (Analysis 1.3; summary of findings Table 1).

Secondary outcomes

Anxiety/depression were not reported.

No serious adverse events were reported. Adverse events were reported in five participants. One participant in the IG group and one in the CG group reported flu‐like symptoms and sinusitis, and one reported worsening of abdominal pain in the IG group. Two participants, one in the IG and one in the CG group, reported IBD relapse.

2. Relaxation training versus wait‐list

Shaw 1987 compared a relaxation training to a wait‐list (n = 40). It included only people with UC at an unclear stage of the disease. The intervention lasted six weeks, with participants followed up for six weeks after the end of the intervention.

Primary outcomes

Treatment success was not reported.

Pain frequency was measured in hours between pain episodes, and pain intensity on an unidentified 0 to 10 scale. Withdrawals due to adverse events were not reported.

Pain frequency

At end of study the mean (SD) score for the 20 UC participants in the relaxation‐training group was 5.6 (2.64) and for the 20 UC participants in the wait‐list group 3.0 (2.05). There was a difference in hours between pain episodes at end of study in the relaxation training when compared to a wait‐list (MD 2.60, 95% CI 1.14 to 4.06). The certainty of evidence was very low, due to imprecision and risk of bias (Analysis 2.1; summary of findings Table 2).

Six weeks after end of study the mean (SD) score for the 20 UC participants in the relaxation training group was 6.7 (3.07) and for the 20 UC participants in the wait‐list group 3.4 (2.23). There was a small difference in hours between pain episodes six weeks after end of study in the relaxation‐training when compared to a wait‐list (MD 3.30, 95% CI 1.64 to 4.96). The certainty of evidence was very low, due to imprecision and risk of bias (Analysis 2.2; summary of findings Table 2).

Pain intensity

In Shaw 1987 at end of study the mean (SD) score for the 20 UC participants in the relaxation‐training group was 5.4 (1.84) and for the 20 UC participants in the wait‐list group 7.1 (2.08). There was a difference in pain intensity at end of study in the relaxation‐training when compared to a wait‐list (MD −1.70, 95% CI −2.92 to −0.48). The certainty of evidence was very low due to imprecision and risk of bias (Analysis 2.3; summary of findings Table 2).

Six weeks after end of study the mean (SD) score for the 20 UC participants in the relaxation‐training group was 4.5 (2.01) and for the 20 UC participants in the wait‐list group 6.8 (2.48). There was a small difference in hours between pain episodes at end of study in the relaxation‐training when compared to a wait‐list (MD −2.30, 95% CI ‐3.70 to ‐0.90). The certainty of evidence was very low due to imprecision and risk of bias (Analysis 2.4; summary of findings Table 2).

Withdrawals due to adverse events were not reported.

Secondary outcomes

Anxiety/depression and adverse events or serious adverse events were not reported.

3. Yoga intervention versus no intervention

Sharma 2015 (n = 60) compared a yoga intervention to no intervention (both groups received standard medical therapy). The study included only participants with CD or UC at an inactive stage of the disease with a CDAI score under 150, while UC activity was measured on the Truelove and Witts index (Truelove 1954). The intervention lasted eight weeks. The study reported data for the 30 UC participants in the yoga group and 30 UC participants in the control group.

Primary outcomes

Treatment success in this study was measured as a dichotomous outcome of presence or absence of pain. The types of pain reported were tenesmus, intestinal colic pain, peri‐anal pain and arthralgia, of which intestinal colic pain was the most relevant to the topic of this review.

The authors report at baseline that three participants in the yoga group and four in the control group reported the presence of pain, and 23 in the yoga group and 22 in the control group reported absence of pain.

At end of study, in the yoga group five reported presence of pain and in the control group 14 reported presence of pain, while 20 in the yoga group and 12 in the control group reported absence of pain.

After we contacted the authors, they explained that these results are for the 25 UC participants in the yoga group and the 26 participants in the control group who completed the study. However, if the baseline and end‐of‐study numbers for presence of pain are summed for the yoga group the total is 26. The authors did not provide further clarification on this discrepancy, nor on how presence and absence of pain were defined, and whether this was a yes/no question for participants or a pain scale with a cut‐off score.

Due to the reasons outlined above and the fact that we could not determine whether presence and absence of pain were measured for all 60 UC participants randomised to the study at baseline we did not perform an analysis for the results provided for this outcome.

Pain frequency and pain intensity were not reported in this study.

Withdrawals due to adverse events

There were three cases of increased disease activity that led participants to drop out from the study. One occurred in the yoga group and two in the standard medical‐therapy group. There was no clear difference in withdrawals due to adverse events between the two groups (RR 0.50, 95% CI 0.05 to 5.22). The certainty of this result is very low due to imprecision and risk of bias (Analysis 3.1; summary of findings Table 3).

Secondary outcomes

The authors reported that there were no serious adverse events in either group. Total adverse events were also zero in both groups.

Anxiety level results were reported using the State and Trait Anxiety Inventory, consisting of two 20‐item subscales. Participants were asked to mark: not at all, somewhat, moderately, very much, on a scale of 1 to 4. Scoring was done as the sum of these individual scores. The range of anxiety score was 20 to 80.

At end of study the mean (SD) state anxiety score for the 30 UC participants in the yoga group was 32.8 (8.21) and for the 30 UC participants in the standard medical‐therapy group 39 (9.05). There was a difference in state anxiety at end of study in the yoga group when compared to standard medical therapy (MD −6.20, 95% CI −10.57 to −1.83). The certainty of evidence was very low due to imprecision and risk of bias (Analysis 3.2; summary of findings Table 3).

At end of study the mean (SD) trait anxiety score for the 30 UC participants in the yoga group was 41.24 (8.22) and for the 30 UC participants in the standard medical‐therapy group 42.26 (8.29). There was no difference in trait anxiety at end of study in the yoga group when compared to standard medical therapy (MD −1.02, 95% CI −5.25 to 3.21). The certainty of evidence was very low, due to imprecision and risk of bias (Analysis 3.3; summary of findings Table 3).

Depression was not measured.

4. Kefir diet (Lactobacillus bacteria) versus no intervention

Yilmaz 2019 (n = 48) compared a Kefir diet with Lactobacillus bacteria versus no intervention. The study included participants with CD or UC whose disease activity ranged from inactive to moderate. The intervention lasted four weeks.

Primary outcomes

Treatment success in Yilmaz 2019 was not reported.

Pain was measured on a four‐point rating scale from 0 to 3. Pain frequency was not reported.

Pain intensity

At end of study mean(SD) the score for pain intensity for the 15 participants in the UC Kefir group was 0.33 (0.61) and for the 10 participants in the no‐intervention group 0.5 (1.08). No clear difference was detected in pain intensity scores when Kefir diet was compared to no intervention (MD −0.17, 95% CI −0.91 to 0.57). The certainty of the evidence is low due to risk of bias and imprecision (Analysis 4.1; summary of findings Table 4)

Withdrawals due to adverse events

There were no withdrawals due to adverse events in either group, so there was no estimable relative effect of the intervention on withdrawals due to adverse events (summary of findings Table 4).

Secondary outcomes

Anxiety/depression and adverse events or serious adverse events were not reported.

5. Stellate ganglion block versus sulphasalazine

Zhao 2017 (n = 120) compared stellate ganglion block treatment once a day to four doses of sulphasalazine twice a day orally. The study included UC participants at an unclear stage of the disease. The intervention lasted 30 days. Ninety UC participants were randomised to the intervention group and 30 participants to the control group.

Primary outcomes

Treatment success in Zhao 2017 was not reported. Pain was measured as "stomachache", with no further details.

At baseline, stomachache was experienced by 17 participants in the stellate ganglion block group and by 50 participants in the sulphasalazine group. At end of study, stomachache was experienced by eight participants in the stellate ganglion block group and by 13 participants in the sulphasalazine group.

We contacted the study authors to determine if the end‐of‐study results were from the same subgroup of participants who reported stomachache at baseline. We received no response and we could not determine if these results were for the participants experiencing stomachache at baseline or if they refer to the whole study cohort. We therefore decided not to perform any meta‐analysis for these results.

Pain frequency and intensity were not reported.

Withdrawals due to adverse events were not reported.

Secondary outcomes

Anxiety/depression and adverse events or serious adverse events were not reported.