| Popis: |
Background: Shortly after the introduction of the first licensed vaccine against dengue fever (Dengvaxia), a serious outcome was attributed to the vaccine: vaccinated individuals without a previous dengue infection were at increased risk of developing severe dengue if subsequently infected by a heterologous serotype. In response, the World Health Organization recommended vaccination in regions where the seroprevalence of dengue is at least 50% and, ideally, greater than 70%. Hence, accurate estimates of regional seroprevalence are crucial for both population vaccination strategies and test-then-vaccinate decisions at the individual level. Currently, estimates of seroprevalence are based on surveys, using screening tests for previous dengue exposure. These estimates must consider the sensitivity and specificity of the tests, which depend on identification of those who have been exposed, ostensibly through a test, regarded as the gold standard. There is, however, no easily accessible gold standard test for dengue. Methods: We propose an approach to estimate the seroprevalence of dengue that does not require a gold standard test by modeling: (i) the uncertainty in the sensitivity and specificity, and (ii) the uncertainty in the “true” disease prevalence. Results: Through simulations, we demonstrate the effect of these extra sources of uncertainty on post-test estimates of dengue seroprevalence. Our simulations show, for example, that in a population of 1 million it is possible to overestimate or underestimate the number who are truly seropositive by as much as 76,000. Conclusions: Current estimates can substantially overestimate or underestimate the true probability of previous exposure when these extra sources of variability are not accounted for. |
