Clinical Decision Support Systems to Guide Healthcare Providers on HIV Testing

Mikaela Smit; Carlijn C.E. Jordans; Jitte M. Reinhard; Wichor M. Bramer; Annelies Verbon; Casper Rokx; Alexandra Calmy

Disclosures

AIDS. 2022;36(8):1083-1093. 

In This Article

Methods

Search Strategy and Selection Criteria

An original systematic review was carried out and reported according to PRISMA guidelines[16] (see Supplement for PRISMA checklist), to identify studies reporting on any type of automated system triggering an alert to perform an HIV test. A CDSS was defined as any clinical or electronic system that generated an alert encouraging healthcare workers to perform an HIV test.

Medline ALL via Ovid, Embase via Embase.com, Cochrane CENTRAL Register of Trial via Wiley and CINAHL via EBSCOhost were searched from inception to 17 November 2020 (date last searched). Searches were not restricted by language or quality of study. The searches were designed by clinicians and an experienced medical information specialist (M.S./C.C.E.J./W.M.B.). Details of the search terms and strategy can be found in Supplement 1, https://links.lww.com/QAD/C467.

After duplicates were removed, two reviewers (M.S. and C.C.E.J.) screened titles and abstracts independently according to prespecified inclusion and exclusion criteria (Table 1). Selected studies underwent full text reviewing subsequently. Where two studies reported on the same CDSS, the most recent or the one reporting on the impact of the CDSS was chosen. Conflicts were resolved by the two reviewers on a case-by-case basis. All data was extracted into Microsoft Excel by M.S., C.C.E.J. and J.M.R. for qualitative syntheses. Data extracted included first author, publication year, study type, country of study, CDSS type and details, population targeted, time period, sample size and details on the outcome of the CDSS on HIV test rate and diagnosis. We reported harmonized sample size and outcome measures wherever possible (patients and percentage tested or diagnosed). Some studies reported other sample size measures, such as number of visits and other outcome measures, such as rates without providing additional information to generate comparable sample sizes or outcomes. Wherever relevant data was not available, corresponding authors were contacted. Reference lists of included full-text articles were screened to identify additional studies based on the same inclusion and exclusion criteria.

The design and conduct of this review were specified in advance and documented in a protocol. Details of the protocol were registered on PROSPERO (CRD42020175576) and can be accessed at https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42020175576.

Data Syntheses

A qualitative analysis of included studies was performed, describing the geographical distribution of studies, the types of CDSS, which were tested and criteria for targeted testing. Furthermore, meta-analysis was performed, to quantify the impact of CDSS on HIV test rate. The meta-analyses were restricted to studies reporting comparable outcomes (proportion of eligible patients tested with and without the CDSS) and having comparable designs. Studies reporting on other outcome measures, for example, tests per months or over a given period without providing information on the overall patient load could not be included in the meta-analyses. Due to insufficient data on HIV diagnosis, it was not possible to perform a meta-analysis on this outcome.

Crude risk ratios (cRRs) and 95% confidence intervals (CIs) with CDSSs compared with without were obtained from studies. Where studies did not report risk ratios, they were calculated by dividing proportion tested, with all studies included in the meta-analyses reporting proportion tested and total patient flow. 95% CIs were calculated in Microsoft Excel.

The cRRs were pooled in R version 4.0.4 using a random-effect model to adjust for variance within and between studies. Heterogeneity was quantified using the I2 statistic.[17] Risk of bias was assessed for all published (peer-reviewed and preprint) studies using the Version 2 of the Cochrane risk-of-bias tool for randomized control studies (RCTs) (RoB 2) and the Risk of Bias in Nonrandomized Studies of Interventions (ROBINS-I) tool.[18,19]

processing....