Skip to main content
. 2023 Apr 25;2023(4):CD013871. doi: 10.1002/14651858.CD013871.pub2

Risk of bias for analysis 1.1 Resolution of rCDI: intention‐to‐treat analysis.

Study Bias
Randomisation process Deviations from intended interventions Missing outcome data Measurement of the outcome Selection of the reported results Overall
Authors' judgement Support for judgement Authors' judgement Support for judgement Authors' judgement Support for judgement Authors' judgement Support for judgement Authors' judgement Support for judgement Authors' judgement Support for judgement
Cammarota 2015 Low risk of bias Quote from article: "Blocked randomization of subjects was performed by an external person not involved in the study. An online random number generator software was used to provide random permuted blocks with a block size of six and an equal allocation ratio; the sequence was concealed until the interventions were assigned."
Comment: We think the allocation sequence was random and appropriately concealed in this study and the baseline characteristics of the two groups were comparable.
Low risk of bias The study was not blinded however we do not believe there was a deviation from intended intervention due to non‐blinding nor did this affect the outcome measured as we performed analysis on an intention‐to‐treat basis. Low risk of bias There was no missing data at the end of the study period. Low risk of bias This was an open‐label study. We did not think that the measurement of the resolution of rCDI was at high risk of bias because the definition required two negative stool tests for C. difficile toxin. Low risk of bias The clinical trial was registered at clinicaltrial.gov (NCT02148601). The authors followed the pre‐specified plan of analysis. The study was conducted in accordance with the Consolidated Standards of Reporting Trials (CONSORT) Statement. There were no concerns about selective reporting for this outcome. Low risk of bias No major concerns were noted in any of the domains of risk of bias for the outcomes of resolution of rCDI from this study. Even though the study was open‐label, we did not have any major concerns about bias in the measurement of this outcome due to objective measurement with C. difficile stool testing.
Hota 2017 Some concerns No information was available on sequence generation and concealment of allocation sequence. However, the baseline characteristics of both groups were similar. We contacted the authors for further information on the randomization process but did not get a reply. Low risk of bias The study was not blinded however we do not believe there was a significant deviation from intended intervention due to non‐blinding nor did this affect the outcome measured. The authors of the study performed a per‐protocol analysis and we created an intention to treat analysis by considering all the participants who were randomized to FMT and vancomycin group and assumed that the participants excluded by the study authors in the analysis did not have the outcome. Low risk of bias Two participants were excluded from the vancomycin group, one withdrew and one was withdrawn by the investigator due to non‐compliance to the protocol. We included all participants in the intention‐to‐treat analysis and assumed that those excluded were failures (had an episode of rCDI). We did a sensitivity analysis to assess our assumption and the results were similar. So the missing data is less likely to create a bias for this outcome. Low risk of bias The study was open‐label but we did not consider it as high risk of bias for this domain as the outcome definition included an objective measure with stool C. difficle testing. Low risk of bias The study was registered at clinicaltrial.gov (NCT01226992). The authors followed the pre‐specified plan of analysis. We did not have any major concerns about selective reporting for this outcome. Low risk of bias Even though we had some concerns about the lack of description of methods of the randomization process, we did not think that this study was at high risk of bias. The two groups were balanced at the start of the trial. Even though this was an open‐label study the outcome measured was objective hence low risk of bias from non‐blinding.
Hvas 2019 Some concerns No information was available on allocation sequence and concealment of allocation sequence. However, the baseline characteristics of both groups were similar. We contacted the authors for further information on the randomization process but did not get a reply. Low risk of bias Quote from article: ''All 64 randomized patients received the allocated treatment.''
The study was not blinded however we do not believe there were significant deviations from intended intervention due to non‐blinding as all participants received allocated intervention and authors of the study performed intention to treat analysis.
Low risk of bias There was no attrition and data were available for all the participants for this outcome. Low risk of bias The study was open‐label but the outcome measured was defined objectively. We did not think that outcome of the resolution of rCDI was at high risk of bias because the definition of the outcome included laboratory‐confirmed CDI. Low risk of bias Comment: The trial was registered at clinicaltrial.gov (NCT02743234). The authors followed the pre‐specified plan of analysis. We did not have any concerns about selective reporting for this outcome. Low risk of bias Even though we had some concerns about the lack of description of methods of randomization process and that study was open‐label, we did not consider this study was at high risk of bias for this outcome. All the study groups were similar at the start of the trial and the outcome measured was objective in nature.
Kelly 2016 Low risk of bias Quote from article: "Patients were equally allocated to the donor and autologous FMT groups via block randomization by C difficile positivity at baseline, with stratification by study site."
Comment: The allocation sequence was random and appropriately concealed in this study and the baseline characteristics of the two groups were comparable.
Low risk of bias This was a double‐blinded study and the authors performed an intention‐to‐treat analysis hence low risk of bias due to deviation from intended intervention. Low risk of bias Two patients were lost to follow‐up in the control group and one patient in the intervention group. All the participants were considered in the intention to treat analysis in our review assuming that patients who were lost to follow up were failures. We did a sensitivity analysis by considering those participants for whom the data were available and the results were similar. So we think that missing data is less likely to bias this outcome. Low risk of bias The study was double‐blind, hence we do not have any concerns about the risk of bias in the measurement of this outcome. Low risk of bias The trial was registered at clinicaltrial.gov (NCT01703494). The authors followed the pre‐specified plan of analysis. We did not have any concerns about selective reporting for this outcome. Low risk of bias This was a very well conducted double‐blinded study with no major concerns about risk of bias for any of the domains considered.
Rode 2021 Low risk of bias Quote from article: "Computer‐generated stratified randomization in blocks of six was used with allocation concealment in sealed opaque envelopes with sequential numbers for each stratum. "
Comment: The allocation sequence was random and appropriately concealed in this study and the baseline characteristics of the two groups were comparable.
Low risk of bias The study was not blinded however we do not believe there was a deviation from intended intervention due to non‐blinding nor did this affect the outcome measured as we included the analyzed data on an intention to treat basis by assuming that the participants who did not complete the study did not have the outcome. Low risk of bias The majority of missing data was due to mortality (2 in the FMT group and 11 in the comparison group). We created the intention‐to‐treat analysis basis by analyzing the participants in which they were randomized irrespective if they received the intervention and were loss to follow‐up. For participants with loss to follow up, we considered them to have failures meaning they did not have resolution of rCDI. In order to investigate our assumption, we did a sensitivity analysis just on available cases and the summary estimate has a slight reduction in effect (from RR 1.63 to 1.58) but the direction of effect was same and width of confidence did not change much. We, therefore, think that this outcome was not at high risk of bias due to missing data. Low risk of bias The study was open‐label but the outcome measured was defined objectively. We did not think that measurement of resolution of rCDI was at high risk of bias because the definition of the outcome included the absence of laboratory‐confirmed CDI in case diarrhea was present. Low risk of bias The trial was registered at ClinicalTrials.gov (NCT02774382). The authors followed the pre‐specified plan of analysis. We did not have any major concerns about selective reporting for this outcome. Low risk of bias Even though the study was open‐label, we did not have any major concerns about bias in the measurement of this outcome due to the objective definition of the outcome and intention to treat analysis.
van Nood 2013 Low risk of bias Quote from protocol: "Patients will be randomized by a computer according to their first, second or >2 relapses, and hospitalization status."
Quote from Supplementary Appendix: ''To achieve adequate allocation concealment, each patient was randomized by applying automated biased coin minimization in ALEA with stratification for hospitalization status (clinical or outpatient) and the number of previous recurrences (1, 2, >2). The coin bias factor was set at 3, the bias coin lower threshold at 2. Study physicians at the coordinating center in charge of randomization were unaware of the model specifications used.''
Comment: The allocation sequence was random and appropriately concealed in this study and the baseline characteristics of the two groups were comparable.
Low risk of bias The study was not blinded however we do not believe there was a deviation from intended intervention due to non‐blinding nor did this affect the outcome measured as we included the analyzed data on an intention‐to‐treat basis by assuming that participants who did not complete the study did not have the outcome. Low risk of bias Only two participants did not complete the study and one of them was mortality. We created the intention to treat analysis assuming the two participants with loss to follow up were failures, i.e. did not have resolution of rCDI. In order to investigate our assumptions, we did a sensitivity analysis and the effect size remained the same. We, therefore, do not think that this outcome was at high risk of bias due to missing data from this study. Low risk of bias The study was open‐label. Resolution of rCDI was defined based on the absence of diarrhea or persistent diarrhea that could be explained by other causes with three consecutive negative stool tests for C. difficile toxin. We did not have any major concerns about bias in the measurement of this outcome due to the objective definition of the outcome. Low risk of bias The trial was registered in the Netherlands Trial Register number, NTR1177. We did not have any concerns about selective outcome or analysis reporting for this outcome. The authors followed the pre‐specified plan of analysis. Low risk of bias Even though the study was open‐label, we did not have any major concerns about bias in the measurement of this outcome due to objective measurement with C. difficile stool testing and intention to treat analysis.