Summary of findings 1. Summary of findings table ‐ Stem Cell compared to Placebo or Control for Induction of Remission in Medically Refractory Crohn?s Disease.

Stem Cell compared to Placebo or Control for Induction of Remission in Medically Refractory Crohn's Disease
Patient or population: Induction of Remission in Medically Refractory Crohn's Disease Setting: Specialised centres Intervention: Stem Cell Comparison: Placebo or Control
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with Placebo or Control	Risk with Stem Cell
Clinical remission	315 per 1000	592 per 1000 (252 to 1000)	RR 1.88 (0.80 to 4.41)	301 (3 RCTs)	⊕⊕⊝⊝ Lowa,b	The three studies that included data about clinical remission were Hawkey 2015, Melmet 2015, and Panes 2016. Each had a different definition of clinical remission.
CDAI <150 at 24 weeks	506 per 1000	516 per 1000 (339 to 789)	RR 1.02 (0.67 to 1.56)	352 (4 RCTs)	⊕⊝⊝⊝ Very lowc,d,e	The data regarding the CDAI at 24 weeks was not the primary outcome of any of the included studies. They also used different cut‐offs and different ranges. Here we included local and systemic stem cell therapy. Zhang et al 2018 had zero weight, as the number of events was 0 in both intervention and control arms (the authors stated that no patients achieved CDAI <150 at 24 weeks). The baseline CDAI was already low in the studies examining fistula management.
Fistula Closure short‐term assessed with: Clinically or MRI	349 per 1000	516 per 1000 (391 to 684)	RR 1.48 (1.12 to 1.96)	269 (4 RCTs)	⊕⊕⊝⊝ Lowf,g	Fistula closure was assessed in four studies. Garcia‐Olmo et al 2009 assessed the outcome at 8 weeks, while Molendijk et al 2015, Panes et al 2016, and Zhou et al 2020 assessed the outcome at 24 weeks. It was assessed both clinically and with MRI
Fistula closure in long‐term Follow up of original studies assessed with: Clinicaaly or MRI	390 per 1000	554 per 1000 (425 to 729)	RR 1.42 (1.09 to 1.87)	250 (4 RCTs)	⊕⊕⊝⊝ Lowh,i	The data on the long‐term effects were gathered from published papers after a long‐term follow‐up of the original studies. Except Zhou 2020, which reported their long‐term 1‐year follow‐up.
Total Adverse Events assessed with: Clinically	730 per 1000	723 per 1000 (643 to 825)	RR 0.99 (0.88 to 1.13)	293 (4 RCTs)	⊕⊝⊝⊝ Very lowj,k,l	The range of total adverse effects stated was very wide across studies, from minimal abdominal pain or low‐grade fever to sepsis and the need for surgical operation. We collected the data on total adverse events without stating the level, or severity of the adverse events.
Serious Adverse Events assessed with: Clinically	112 per 1000	137 per 1000 (99 to 187)	RR 1.22 (0.88 to 1.67)	433 (7 RCTs)	⊕⊕⊝⊝ Lowm,n	All 7 studies stated the number of patients suffering from serious adverse effects, mostly because these trials are addressing the safety issues of stem cell administration. The different definitions of serious adverse events among the studies make the outcome assessment heterogeneous and inconsistent, some studies did not define clearly what a serious adverse event stands for.
Withdrawal due to adverse events	74 per 1000	58 per 1000 (24 to 140)	RR 0.78 (0.32 to 1.89)	272 (3 RCTs)	⊕⊝⊝⊝ Very lowo,p	Both Panes 2016 and Zhou 2020 stated clearly that the withdrawal was due to adverse events. But in Hawkey 2015, it was stated that one patient from the control group withdrew directly after randomization, and one patient in the active group withdrew after 26 weeks of mobilization for accelerated transplantation. In Zhou 2020, the patients withdrew due to adverse effects to receive a subsequent reoperation.
*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). CI: confidence interval; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_423896885310176737.

^a The risk of bias is downgraded to serious, as two of the three studies are open‐label, thus the participants are not blinded including Panes 2016 which contains most of the studied population, and this outcome is a subjective outcome.
^b The inconsistency was downgraded by one level as the heterogeneity in the included studies was high in the form of the variability of the results due to different definitions of the outcome.
^c Inconsistency downgraded by one level to serious as the level of CDAI at the start of the studies varied according to the route of intervention, CDAI was low or normal at baseline in local intervention, while high at baseline in systemic intervention.
^d The imprecision was downgraded by one level to serious as the RR is 1.08 with CI 0.94‐1.24, which makes the intervention equally capable of doing harm or benefit.
^e Publication bias strongly suspected and downgraded by one level as there were two abstracts of unpublished complete data (Arturo 2017 and Lichtiger 2012).
^f The outcome was downgraded by one point to serious as the risk of allocation concealment was unclear in three studies (Garcia‐Olmo 2009 and Zhou 2020), and randomization was unclear in (Zhou 2020). Although the risk of bias for blinding of participants is high in (Zou 2020, Panes 2016, Garcia‐Olmo 2009), and unclear in (Molendijk 2015); three of these studies had a low risk for detection bias, while only Zhou 2020 had high risk. Also, the outcome is an objective outcome that doesn't change by the participant or the personnel being unblinded.
^g Imprecision was downgraded by one level to serious as the number of the studied population was small.
^h The risk of bias is downgraded to serious: as there are a lot of patients lost to long‐term follow‐up (missing data), thus high attrition.
ⁱ The imprecision was downgraded by one level to serious as the number of patients and number of events were low. Also, the confidence interval was wide.
^j The risk of Bias was downgraded by one level to serious as three of the four studies had a high risk of performance bias and one unclear risk. Considering that the outcome is reporting about the adverse events, which are mostly subjective in the case of mild and moderate adverse events, we downgraded by one level.
^k Inconsistency is downgraded by one level to serious as the studies reported the category (All adverse events differently)
^l Impression was downgraded by one level as the number of participants was low
^m The risk of bias was downgraded by one level, as 5 trials had high‐performance bias (open‐label trials) but it is not an objective outcome, and allocation concealment is unclear in 4 trials and randomization is unclear in 2 trials.
ⁿ Imperceision is downgraded by one level to serious because the CI was very wide (0.89‐1.93), so we are not certain if the intervention causes benefit or harm. 
^o Indirectness was downgraded by one level to seriuos as the causes of withdrawal in (Hawkey 2015) were not stated as due to adverse events.
^p Imprecision was downgraded by two levels to very serious due to the low number of participants and wide CI (0.33‐1.91), so we are not confident if the intervention causes benefit or harm.