The quality of the reporting and methodology of the included trials was generally poor (Table 5), with only two studies86,88judged not to be at high risk of bias on at least one domain, and all RCTs including at least one domain judged to be at unclear risk of bias, with few instances of low risk of bias. Five trials74,76,85–87 reported adequate random sequence generation, but only one of the seven included trials clearly reported how allocation was concealed and, therefore, the remaining six were judged to be at an unclear risk of bias. Most trials were judged to be at an unclear risk of performance and detection bias, because they did not report if participants, study personnel or outcome assessors were blinded to treatment allocation. Balato and colleagues85and Staab and colleagues87were judged to be at high risk of performance bias; in Balato and colleagues,85although it was noted that physicians were blinded to group assignment until the end of the study, no details were provided about the blinding of participants, and in Staab and colleagues87it was stated that participants and trainers were not blinded to treatment allocation. The trial by Staab and colleagues87was also judged to be at high risk of detection bias (on self-reported measures) because participants were not blinded, and it was unclear if the investigators who rated eczema severity were blinded to treatment allocation. However, it may be difficult to blind participants to the fact they are taking part in an educational intervention rather than just receiving standard medical care, but this criterion is still appropriate for demonstrating potential risks of bias.
Incomplete outcome data were adequately addressed in three trials85,86,88either because all participants completed the trial and were analysed according to their randomised groups,85,88or because attrition was balanced across arms with similar reasons for drop-out and was therefore considered unlikely to bias the results.86One trial74was rated as being at an unclear risk of bias on this criterion, because, although attrition rates were small and balanced across groups, clear reasons were not reported for the attrition, and, in addition, one participant was randomised and excluded from the analysis because technical difficulties resulted in baseline data not being available for this participant. The other three trials76,87,89 were considered to be at high risk of attrition bias, either because drop-outs were unbalanced between groups76,87or because there was a high overall rate of attrition89(with no exact reasons provided other than that participants did not return study measures at particular time points and that four participants in the control group received the intervention and therefore were excluded) and no intention-to-treat (ITT) analyses were used. In addition, one participant in the educational intervention group in Bostoen and colleagues76was excluded from the analysis owing to experiencing extreme stress at work during the study.
Two trials were assessed as being at low risk of selective reporting, because the authors reported results for all the outcome measurements specified in the methods section of the paper.85,86Four were considered to be at an unclear risk of bias for various reasons, including narratively reporting results for many of the outcomes without providing supporting numerical data;76converting a continuous measure to a dichotomous measure for a severity of itching and scratching outcome;89reporting within-group p-values for two HRQoL outcomes, with between-group p-values reported for only one of the measures;88and measuring outcomes at 6 and 12 months’ follow-up, but not reporting the 6-month outcomes.87The RCT by Santer and colleagues was considered to be at high risk of selective reporting bias, because results for the IDQoL and CDLQI measures of HRQoL were not reported.74Results were also not reported for a number of other measures [specifically, self-report measure of emollient use; Problematic Experiences of Therapy Scale (PETS); attitudes measure; and Patient Enablement Instrument].
Other potential sources of bias were identified in three trials where this was rated as‘unclear’. In Bostoen and colleagues76most results were presented as subgroup analyses for atopic dermatitis and psoriasis patients and it was unclear if these were adequately powered or pre-specified. In Ersser and colleagues86 a statistically significant higher proportion of participants in the intervention group compared with the control group were women. In Santer and colleagues,74there were baseline imbalances in disease severity across groups, with both website groups having slightly more severe disease at baseline than the usual care group, and it was unclear if this had been adjusted for in the data analysis. In Santer and colleagues,74
TABLE 5 Quality assessment of the included RCTs (Cochrane risk of bias criteria)
Study, skin condition (population) Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias Balato et al., 2013,85 plaque psoriasis (adults)
Low risk Unclear risk High risk Unclear risk Low risk Low risk Low risk
Ersser et al., 2011,86psoriasis
(adults)
Low risk Unclear risk Unclear risk Unclear risk Low risk Low risk Unclear risk Bostoen et al., 2012,76 psoriasis or atopic dermatitis (adults)
Low risk Unclear risk Unclear risk Unclear risk High risk Unclear risk Unclear risk Santer et al., 2014,74eczema (children and parents/carers)
Low risk Low risk Unclear risk Unclear risk Unclear risk High risk Unclear risk Staab et al., 2006,87 atopic dermatitis (children, adolescents and parents)
Low risk Unclear risk High risk High risk High risk Unclear risk
Low risk
Matsuoka et al., 2006,88acne
vulgaris (adults)
Unclear risk Unclear risk Unclear risk Unclear risk Low risk Unclear risk
Low risk
van Os-Medendorp et al., 2007,89
chronic pruritic skin disease (adults)
Unclear risk Unclear risk Unclear risk Unclear risk High risk Unclear risk
statistical analyses were also reported inconsistently; specifically, mean change in the POEM score was presented for the combined website groups versus usual care, not individual website groups versus usual care (therefore this result was not data extracted or presented in this review).