Risk prediction of hepatitis B or C or HIV among newly diagnosed cancer patients

Abstract

Background: Screening for viral infection in cancer patients is inconsistent. A mechanism to readily identify cancer patients at increased risk of existing or prior viral infection could enhance screening efforts while reducing costs.

Methods: We identified factors associated with increased risk of past or chronic hepatitis virus B, hepatitis virus C, or HIV infection before initiation of systemic cancer therapy. Data were from a multicenter prospective cohort study of 3051 patients with newly diagnosed cancer (SWOG-S1204) enrolled between 2013 and 2017. Patients completed a survey with questions pertaining to personal history and behavioral, socioeconomic, and demographic risk factors for viral hepatitis or HIV. We derived a risk model to predict the presence of viral infection in a random set of 60% of participants using best subset selection. The derived model was validated in the remaining 40% of participants. Logistic regression was used.

Results: A model with 7 risk factors was identified, and a risk score with 4 levels was constructed. In the validation cohort, each increase in risk level was associated with a nearly threefold increased risk of viral positivity (odds ratio = 2.85, 95% confidence interval = 2.26 to 3.60, P < .001). Consistent findings were observed for individual viruses. Participants in the highest risk group (with >3 risk factors), comprised of 13.4% of participants, were 18 times more likely to be viral positive compared with participants with no risk factors (odds ratio = 18.18, 95% confidence interval = 8.00 to 41.3, P < .001).

Conclusions: A risk-stratified screening approach using a limited set of questions could serve as an effective strategy to streamline screening for individuals at increased risk of viral infection.