The US Centres for Disease Control and Prevention (CDC) recommends universal “opt-out” HIV screening in all health care settings, including emergency departments and primary care settings (CDC 2006 HIV testing guidelines); the UK Guidelines for HIV Testing (BHIVA, BASHH, BIS) have, since 2008, recommended universal “opt-out-screening” in communities where HIV prevalence ≥ 2 per 1000 (HPA 2011 Time to test for HIV). “Universal” screening is, of course, easier to recommend, than to implement. However, a recent US study, Lyons & Fichtenbaum (L&F), raises the question whether, given adequate resources for implementation of universal screening, it is even the right goal to aim for. Isn’t some degree of targeting always bound to have a beneficial influence on the yield of screening?
This randomized control study conducted over two years in a US emergency department within a low HIV prevalence area and in the context of an ongoing HIV screening program, compared the yield of universal and targeted HIV screening. Sufficient resources were provided to ensure that within the targeted arm, eligibility criteria could be targeted as broadly as possible, and would be implemented effectively. The researchers were thus able to investigate the possibility that some degree of targeting would increase the yield of screening even in the ideal situation where it was not being forced on them by inadequate resourcing.
The answer of the study is that, even in the context of this low-prevalence area (0.23% in the surrounding county), targeting conferred no detectable clinical advantage. Of the 4,692 eligible visits in the universal arm 40.8% consented with 6 new diagnoses resulting (0.31%); of the 4880 eligible visits in the targeted arm 47.4% consented with 3 new diagnoses (0.22%). The lower level of consent that might have been expected in the universal arm was not accompanied by any decrease in yield. So even in this optimal situation for targeting, it seems to have brought no advantage.
What is the relevance of this finding, we might ask – especially outside the US, where emergency departments may not have the same role within the health system as the first and last resort of the uninsured? A large trial of universal HIV screening in Paris emergency departments, for example, generated an entirely negative result (STI Blogs 28/11/11); whereas the 2009-10 UK screening pilots of the HINT study picked up rather more cases (4/3433), though not as many as L&F (HPA 2011 Time to test for HIV). A more generalizable message of this study may be the difficulty of using targeting effectively to reach the undiagnosed. Palfreeman & McNally, in a short report on a pilot screening study at a UK acute medical unit (Leicester), undertaken concurrently with the HINT study, comments that 3/10 diagnosed did not have an indicator condition, and that many would have been missed but for the pilot. This might be a point in favour of universal screening.
The findings of L&F have decreasing relevance to the extent that screening operates in situations where resources are constrained. The authors themselves comment, in connection with the clinical screening program into which their study was inserted, that its high positivity rate (1%) before and after the study reflects the degree to which health workers were forced by limited resources to operate more selective targeting than the study itself. Of course, L&F do not mean to imply that, in the more straitened context of the program prior to the study intervention, health workers should not have been applying more restrictive criteria. Their point is simply that, where such constraints do not apply, targeted screening should not be assumed to offer any greater benefit than universal screening. The relevance of such a finding, supposing it to be well-founded, will depend on how far, in the particular context considered, real-life conditions approximate those of the study itself, rather than those of the screening program as it operated prior to and after the study!