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

Results

The database search identified 1424 records. After removing duplicates, 929 records were screened, and 82 records were assessed in full text (Figure 1). Of these, 22 met the inclusion criteria.

Figure 1.

Systematic search flow diagram and search terms. *Database search record excluded with reasons: 28 did not report on alert systems; 17 were duplicates of other studies; 13 did not report any outcomes; one news briefing; one could not find full-text.

Included studies were published between 2008 and 2020. The geographical distribution of the included studies was limited in scope and from HIV low endemic settings predominantly (Figure 2); the majority of the studies were conducted in the United States of America (n = 17),[20–36] with the remaining studies from the United Kingdom (n = 2),[37,38] Kenya (n = 1),[39] and Argentina (n = 1),[40] with one not stating a country.[41] Most studies were implementation studies (n = 20) comparing HIV testing preimplementation and postimplementation of CDSSs, with only two randomized control trials.[34,39] Study settings were hospitals (n = 12),[21,22,26,29–31,33,36,38–41] nonhospital (n = 9)[20,23–25,27,28,32,34,35] or combined (n = 1).[37] The setting was defined based on the clinical medical setting and the electronic data tools that tend to be available to these settings. Hospital settings included emergency departments, outpatient primary care departments, ICUs and paediatric HIV clinics. Nonhospital settings included primary care practices, veteran facilities and community health centres. 'Combined' setting referred to studies looking at CDSSs within both hospital and nonhospital-based settings. Studies mainly targeted general adult populations aged 13 and over (n = 16),[20–22,24,26,29–33,35–38,40,41] five focused on veterans[23,25,27,28,34] and one study from Kenya focused solely on a paediatric population.[39]

Figure 2.

Global distribution of studies.

Overview of Clinical Decision Support System Types

All CDSSs were developed for physician or nurse use and relied on electronic medical records or laboratory ordering systems displaying information, which could prompt offering an HIV test. Amongst the 22 included studies reporting on CDSSs, two types of alert systems were identified. The majority were electronic clinical reminders (n = 18),[20,22–28,30–36,39–41] with the remaining studies reporting on electronic laboratory reminders (n = 3)[21,37,38] or a combination of both (n = 1).[29] CDSSs that acted as electronic clinical reminders mainly focused on universal screening (alerting healthcare providers to perform a routine HIV test for all patients without known HIV infection and without a documented recent HIV test). Most CDSSs of this type relied on displaying patients' prior HIV testing history alone. Some electronic clinical reminders combined the display of HIV testing history with more targeted alerts, such as additional alerts for patients with risk factors (e.g. hepatitis, use of postexposure prophylaxis, recreational drug use, sexually transmitted infections or other HIV indicator conditions). Electronic laboratory reminders on the other hand focused on providing 'targeted' HIV tests (based on generating alerts when other laboratory tests were ordered). These included prompting providers to offer an HIV test when ordering tests for sexually transmitted infections (STIs), hepatitis or pneumonia and a subset of AIDS-defining conditions (wasting syndrome, herpes simplex >1 month and recurrent pneumonia). The risk of bias across studies was relatively homogenous and moderate on average (Supplement 1, https://links.lww.com/QAD/C467), as is common when evaluating largely non-RCTs. A full summary of the studies is presented in Table 2.

Summary of Impact on HIV Testing and Diagnosis

All studies reported the impact of the CDSSs on the number of HIV tests and 11 studies reported the impact of CDSSs on new HIV diagnosis.[20,23–27,29–32,35] Almost all studies reporting the impact of CDSSs on HIV testing and the majority reporting on HIV diagnosis reported a crude increase in HIV testing with CDSSs. However, not all studies established whether the increase was statistically significant. Amongst the 14 studies that evaluated the statistical significance of the impact of CDSSs on the increase in HIV testing, all but two found that CDSSs significantly increased HIV testing.[20,26–32,35,39] Only 11 of 22 studies reported the impact of CDSSs on HIV diagnosis with increases reported in five. Two out of these five studies also established statistically significant increases in HIV diagnosis, one found no statistical significance.[29,30,32]

Out of the 22 studies, 15 used comparable outcome measures regarding the impact of HIV testing (proportion of patients tested with and without the CDSS).[20,21,24,26–35,38] The remainder used other outcome measures including tests per months or crude numbers without providing information on total patient load.[22,23,25,37,39,40] Their design was heterogenous. Amongst the 15 studies, however, some had comparable designs. Specifically, three reported on universal screening in hospital settings and four on universal screening in nonhospital settings and were pooled in two sub-group meta-analyses. The meta-analyses showed that the use of CDSSs could be more than double HIV testing (cRR = 2.57, 95% CI 1.53–4.33 in hospital settings and cRR = 2.13, 95% CI 1.78–4.14 in nonhospital settings, random effects model) (Figure 3a and b), with a higher impact in hospital settings.

Figure 3.

Forest plot of the risk ratio for HIV testing.Estimates were pooled using a random effects model. CI, confidence interval; RR, risk ratio. (a) Universal screening in nonhospital settings and (b) Universal screening in hospital settings.

processing....