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


AIDS. 2022;36(8):1083-1093. 

In This Article


This study indicates that implementation of CDSSs across a variety of healthcare settings improves HIV testing rates inside and outside the hospital. All identified studies reported an absolute increase in HIV testing rate following the implementation of CDSSs. All but two out of the 14 studies that performed statistical analyses on these outcomes reported a statistically significant impact on HIV testing. Our pooled analyses of the data on universal screening in hospital and nonhospital setting showed an over two-fold increase in HIV testing rates. Although no study outlined a priori definitions of clinical significance regarding the impact on HIV testing, we suggest these pooled findings could be considered a clinically relevant improvement. However, few studies investigated whether CDSSs improved HIV diagnosis (overall and early), and none reported on the long-term impact of CDSSs on increasing HIV testing or factors affecting their successful implementation. As a result, the data regarding these outcomes are mixed and limited conclusions can be drawn in relation to them. They do, however, highlight the need for additional data on the impact of CDSSs in these areas as this is necessary to estimate the impact on UNAIDS 90–90–90 goals.

This systematic review highlighted the relatively small number of studies reporting on the impact of CDSSs to improve HIV testing, and especially, diagnosis, and their limited geographical distribution. This and the resulting risk of bias hinders making too firm conclusions on their impact to curb the HIV pandemic. Although the majority of studies did report an absolute or significant increase in HIV testing, it is important to note that the two RCTs were unable to establish a significant impact of CDSSs on this outcome. Possible reasons for this include that the one study delegated the prompts to healthcare workers unfamiliar with HIV testing whereas the other found that providers did not have sufficient time to perform HIV tests and did not believe risk assessments were appropriate for the veteran patients, both resulting in low uptake.[34,37] Overall, our results highlight that steps need be made to generate appropriately powered RCTs or interventional studies to assess the clinical impact on HIV diagnosis with either universal or targeted CDSSs and to identify the optimal CDSS method and optimal selection of patients per setting. Finally, there seems to be an underuse of CDSSs to adequately capture HIV diagnoses based on current and recognized clinical indicator conditions[5] (lymphoma, cervical cancer, HHV8-related diagnosis, mononucleosis-like illness, oral candidiasis, etc.); such CDSSs have the potential of significant individual benefits.

However, the wider potential of CDSSs are clear as they have been used across disease areas to improve adherence to best practice for patient care, including timely testing.[8–10] Coronavirus disease 2019 (COVD-19) in particular, has demonstrated the feasibility of rapid implementation of CDSS for disease testing and diagnosis.[42] The importance of targeted testing for HIV has been highlighted in a number of studies, which identified the missed opportunity for timely testing, particularly amongst patients presenting for cancer care[43,44] and certain key populations.[45] The missed opportunities of recognizing HIV indicator conditions or risk factors as triggers for physicians to test for HIV suggests that gaps in timely diagnosis of HIV remain to be fully leveraged and CDSSs could play a role not only in reaching public health objectives but also to improve individual health condition in affected patients.[46] Comprehensive evaluations of setting-specific missed opportunities for timely diagnosis and key patient characteristics will help inform targeted CDSSs and likely result in considerable efficient improvements to the care cascade of low HIV endemic settings.

Future studies on this topic should also allow us to better understand the design of such systems and the way in which they fit into daily clinical care, adjusted to the setting, and their clinical impact on particularly the first pillar of the HIV cascade of care to decrease the proportion of people currently living without being aware of their HIV infection. From that point of view, we agree with the authors of the included articles that practical and economic barriers in CDSSs implementation should be evaluated and solved to increase chances of success, including healthcare ICT (information and communication technology) infrastructure, limited resources, limited time, lack of priority, and acceptability by, and fatigability of, physicians to be guided by electronic alerts. Although the latter aspect was rarely reported on and the long-term effects of CDSSs on HIV testing and diagnoses, indicating its durability, are therefore, unknown.[47,48]

Unfortunately, the majority of CDSSs were used in hospital healthcare settings in wealthy, and non-HIV endemic countries. Implementation of universal CDSSs are likely to increase timely HIV diagnosis in HIV endemic countries, with poor testing infrastructure and with evidence of missed opportunities for timely diagnosis during previous contact with healthcare. Whereas targeted testing is likely to be most applicable in non-HIV endemic settings with strong healthcare infrastructure including comprehensive patient history and wider diagnostic protocols to determine indicator conditions. Furthermore, CDSSs may be more feasible in hospital settings, where longer patient interaction and on-site laboratory facilities simplifies rapid testing. Yet, questions remain on the feasibility and acceptability of CDSSs in HIV endemic settings, which often suffer from a lack of ICT infrastructure and high out-of-pocket costs for patients, making implementation of such systems unpractical. Here, HIV indicator conditions or risk factors commonly remain undiagnosed or undocumented in patients' medical history, making targeted screening more difficult. Population-wide testing campaigns may be an alternative to increase diagnosis rate in these settings, although crucial to these efforts and their wider public health impact will be the affordable care provision for individuals who are diagnosed positive for HIV.

We would promote that aside from the CDSSs, a number of strategies should be used complementary to improve HIV testing. These include the recommendation for routine screening during pregnancy to reduce mother-to-child transmission,[49] partner notification,[50] the use of community health workers or lay counsellors to improve HIV testing in vulnerable, marginalized or rural populations,[51,52] the use of questionnaires for risk factor assessment,[53,54] and the increased use of self-testing,[55] testing reminder strategies including those using text messaging[56–59] and ring-strategies to offer (self)-testing to all contacts of newly diagnosed HIV patients as suggested in new WHO recommendations.[60] Importantly, implementing new strategies, for example, by CDSSs aimed at improving HIV screening need to be accompanied by the possibility to facilitate linkage to care and treatment in order to have lasting beneficial impact on both individual and population-level health.[61] Ensuring this is essential to optimize the use of CDSSs to reduce HIV infections and deaths.[62]

This is the first global systematic review of the use of CDSSs to improve both HIV testing and diagnosis, providing insights into their impact and done by using standard methodologies for systematic review to optimally collate available data in this field. Yet, the study also has limitations. First, the study was unable to quantify the overall effect of CDSSs especially on HIV or AIDS diagnosis, let alone to delineate their effect per healthcare setting or HIV prevalence area. This was in part because of the large heterogeneity across studies; with studies reporting on the use of CDSSs across a number of healthcare setting and reporting the impact on HIV testing and diagnosis in differing ways (e.g. test per months, proportion of patients). This heterogeneity also means that interpretation of absolute or pooled increases in testing and diagnosis should be interpreted with care. Studies had varying baseline testing rates (most were low testing rates), meaning that, for example, a two-fold increase in testing should be interpreted in the context of the original studies. The low number of studies that were sufficiently homogenous in their design and the fact that these were non-RCTs limits the interpretation of the pooled results. As future appropriately powered and designed studies become available, these estimates can be updated. Second, we were unable to report on the long-term impact of CDSSs on HIV testing, diagnosis and ultimately the HIV cascade of care, as studies reported short-term outcomes or did not assess these outcomes in their settings. Third, studies were only reporting CDSSs in the context of healthcare settings, whereas communities do play a large role in the overall testing strategies. Fourth, overall the data came from nonrandomized studies with a significant risk of bias, which is also likely reflected in the variation in estimated risk ratios between individual studies. This can only be overcome by a randomized study that has sufficient power to delineate the impact of CDSSs in multiple settings. Last, outstanding questions remain regarding cost-effectiveness, compliance and acceptability by patients and physicians. The identified studies reported insufficient to make recommendations and conclusions.

The results of our systematic review indicate an improvement in HIV testing by CDSS, which can also potentially benefit the number of HIV diagnoses. The broader application of electronic reminders within healthcare settings may contribute positively to the timely detection of HIV-infections. However, data on CDSSs, and type (universal vs. targeted) in the field of HIV screening accounting for the wide variety of HIV prevalence remains scarce. More studies are needed, particularly studies on the clinical impact on 90–90–90 goals, in paediatric populations and across different HIV prevalence settings.