Study selection, assessment of methodological
quality and data extraction
All abstracts identified from the literature search were
screened by two reviewers. There was 100% concordance
between the two reviewers. The relevant full text articles/
reports were obtained and assessed for eligibility against the
inclusion/exclusion criteria independently by two reviewers.Studies that met the inclusion criteria were selected for
potential inclusion. Studies that were reported in more
than one publication were included only once. Any disagreements
were resolved by discussion with a third reviewer.
The methodological quality of the eligible studies was
assessed independently by two reviewers using the Joanna
Briggs quality assessment tool for experimental and nonexperimental
studies (Table 1). There was 100% concordance
between the two reviewers. Methodological quality
assessment was according to the following criteria: (i)
detailed description of inclusion and exclusion criteria used
to obtain the sample; (ii) evidence of allocation concealment
at randomisation; (iii) the validity of methods of outcome
assessment; (iv) description of withdrawals and dropouts;
and (v) the potential for bias in outcome assessment. The
minimum score obtainable for methodological quality using
these tools was 10 and maximum of 30. For this review, the
mean quality score minus one SD was adopted as the threshold
for defining studies of adequate quality.20 Studies that
obtained a quality score equal to or above this threshold
were included in the analysis, while those that did not meet
the threshold quality score were excluded. Data extraction
was undertaken by one reviewer using a data extraction tool
that was developed for the review. All data extracted were
checked by a second reviewer.
Data analysis
Data were analysed using the RevMan 5.1 software (Nordic
Cochrane Centre, Copenhagen, Denmark). Odds ratios and
95% confidence intervals were calculated for dichotomous
outcomes. Analysis of continuous outcomes involved calculation
of the mean and SD to derive standardised mean
differences and 95% confidence intervals. As the studies
were heterogeneous in terms of the various models of care
and outcomes assessed, meta-analysis could not be undertaken.
Therefore, the results are summarised as a narrative
report with forest plots presented where relevant.