Algorithm Prompts HIV Testing Along with Routine Blood Tests

An intelligent risk stratification algorithm can trigger a prompt for clinicians to offer an HIV test when they are ordering routine blood tests, potentially leading to earlier HIV diagnoses.

Late diagnosis of HIV infection is common. Studies show that in the year prior to an HIV diagnosis, more than half of all patients who presented to primary care or outpatient clinics had symptoms that suggested HIV, but they were not tested. Targeted HIV testing is “likely to lead to earlier diagnoses in those with undiagnosed HIV infection, even in lower prevalence areas, while remaining cost-effective,” stated researchers led by David Chadwick, MD, of James Cook University Hospital in Middlesbrough, United Kingdom.

The researchers published their results in June 2017 HIV Medicine.

Computerized physician order entry (CPOE) systems offer the opportunity to incorporate HIV tests into groups of tests, such as for pneumonia or lymphadenopathy. For example, if the clinician orders tests, such as syphilis serology or an infectious mononucleosis test, or the patient has been diagnosed with candidiasis or has had thrombocytopenia on a recent blood test, the patient is more likely to have undiagnosed HIV infection.

Chadwick and colleagues set out to evaluate a risk-based prototype application to prompt HIV testing when patients undergo routine blood tests. Two CPOE systems were modified to prompt healthcare workers, including general practitioners, to add an HIV test when other tests suggested that the patient was at higher risk of HIV infection.

The application was applied for a 3-month period in two areas, in a large London hospital and in general practices in more rural Teesside/North Yorkshire. At the end of the evaluation period, healthcare workers were asked to assess the usability and acceptability of the system’s prompt. Numbers of HIV tests ordered in the general practice areas were also compared before and after the prompt's introduction.

The system was found to be both useable and generally acceptable to primary physicians, nurse practitioners, and hospital doctors with little evidence of prompt/alert fatigue. “The issue of the prompt appearing late in the patient consultation did lead to some difficulties, particularly around discussion of the test and consent,” the researchers stated.

In the general practices, about 1 in 10 prompts were accepted and there was a 6% increase in testing rates over the 3-month study period.

The researchers concluded that the “clinical decision support (CDS) application to prompt HIV testing appears both feasible and acceptable.”

One benefit of the application was that healthcare workers, particularly general practitioners, in the low-prevalence areas outside London found it “a useful reminder for testing, which may well have contributed to their general satisfaction with the application,” the researchers stated. “In the large areas of the UK where HIV prevalence is not high, a CDS prompt may well have significant potential to increase HIV testing rates.”

The application needs refining to provide more accurate risk stratification and to make it more effective, they stated. Further development should include testing in a larger range of clinical settings to determine its wider applicability within health care systems.