2019 HSR&D/QUERI National Conference

1164 — Analysis of Sequences by Finite State Machine Finds Systemic Anomalies in HIV Diagnostic Testing

Lead/Presenter: Ronald Hauser,  COIN - West Haven
All Authors: Hauser RG (VA Connecticut Health Care System), Chartier M (Deputy Director for the HIV, Hepatitis, and Related Conditions Program in the VHA Office of Specialty Care Services), Maier MM (VA Portland Health Care System)

The misuse of HIV diagnostic tests can delay the diagnosis of HIV, have significant cost implications, pose legal risks to a healthcare system and its staff, and contribute to the spread of HIV. Prompted by case reports of misused HIV diagnostic tests, we sought to develop a technique to systematically evaluate the adherence of the VHA to HIV diagnostic testing guidelines proposed by the Centers for Disease Control and Prevention (CDC).

Using all HIV diagnostic tests present in the Corporate Data Warehouse (CDW) at the time of the analysis (1999-2018), we recreated in silico the chronological sequence of HIV tests performed on each Veteran in the VHA. The CDC's HIV diagnostic testing guidelines were mathematically represented as a finite state machine (FSM). To compare actual HIV testing to the recommendations, each Veteran's HIV testing sequence was compared to the permitted state transitions in the FSM. Anomalous transitions were recorded and subjected to clinician review.

The diagnostic test sequences reviewed consisted of 8.790 million HIV diagnostic tests from 3.870 million Veterans. Deviation from recommended practice was found at nearly all facilities (n = 112/130) including multiple facilities that for periods of time had (1) screened for HIV with an HIV genotype, an expensive and insensitive test for HIV diagnosis, (2) screened for HIV with an HIV viral load, also an expensive test, and (3) confirmed positive HIV screens with Western blot after it became outdated by newer guidelines. Veterans potentially affected by these analyses were contacted through their facilities. Nonadherence to HIV diagnostic testing guidelines likely contributed to a delayed diagnosis of HIV in at least nine cases.

We have developed a novel technique to assess the sequential ordering of laboratory tests, which detected costly and high-risk clinical events related to HIV diagnostic screening across the VHA. This technique is broadly applicable to the analysis of any sequence of events, such as the diagnostic workup of Hepatitis C or syphilis.

Our detection of improper HIV screening led to reviews of policy and health care delivery to Veterans at 112 VHA facilities.