Table 3

Assessment of bias
Domain From the study Review authors’ judgement
Giltin and Corcoran 1993[26]
Selection bias. “randomly assigned to either attention control group who received home-making services or a treatment group” [26 p14] No data presented on the control group characteristics Method of allocation not specified. No comment
Performance bias (blinding of participants and personnel) Participants aware of receiving OT or home making service. OT aware they were providing the intervention. Risk of bias
Detection bias (blinding of outcome assessment). OT providing the intervention also provided the assessment of reported outcomes (care giver acceptance of solutions and elimination of ineffective care giver approaches). Risk of bias
Attrition bias (incomplete outcome data) No attrition from the study reported Data only provided for the intervention group on Risk of bias
Reporting bias. Reporting only on the intervention group. Risk of bias
Jirovec and Templin 2001[27]
Selection bias. “Using a table of random numbers, volunteers were randomly assigned to either intervention or control group [27 p2]. Low risk
Performance bias (blinding of participants and personnel) Participants aware of receiving intervention or in control group. Personnel aware of those in the intervention or control group Risk of bias
Detection bias (blinding of outcome assessment). “The same person collected data from the intervention participants and the control group data collectors were not involved in the intervention”p5 Risk of bias
Attrition bias (incomplete outcome data). Attrition rate, 37%, and reasons reported. “The loss of participants between the groups was not significantly different” [27 p5]. Three of four study measures reported. The implementation of IST by carers was not reported. Low risk Risk of bias
Reporting bias. The intervention arm was assigned into two groups, those that received bi-monthly visits and those that received a visit at 6 months. The data from these two arms were combined as the 6-month outcomes for percentage of time incontinent were “not significantly different” [27 p2] but not presented. Risk of bias
Other points The paper reports that this is a significant decrease in the experimental group using the non-parametric sign test (Z= −1.83, p<.05) [27 p 5]. As these figures appeared inconsistent we re-ran the sign test using the reported data which gave Z=−1.81, p=0.07 which is borderline but not significant. We re-ran the data on another version of the sign test, the exact binomial which gave a value of p=0.09 i.e. still not statistically significant between the groups.
Endberg et al. 2002[28]
Selection bias. “Randomly assigned with use of a computerised minimisation algorithm” ( 28 p255) “Despite randomisation, the control group tended to have more severe incontinence than the treatment group” [28p259] Low risk of bias
Performance bias (blinding of participants and personnel) Participants aware of receiving intervention or in control group. Personnel aware of those in the intervention or control group Risk of bias
Detection bias (blinding of outcome assessment). “The 2 study NPs collected a comprehensive continence and medical history for the subject and caregiver” [28 p254] and provided the intervention and attention control.. “at the end of the treatment the subjects were reassessed” [28 p256] Risk of bias
Attrition bias (incomplete outcome data). Three of 9 subjects randomly assigned to the treatment group dropped out or were excluded”[28 p260] All outcomes and measures reported Low risk of bias
Reporting bias. All outcomes and measures reported Low risk

Drennan et al.

Drennan et al. BMC Geriatrics 2012 12:77   doi:10.1186/1471-2318-12-77

Open Data