14 Feb, 14 | by Leslie Goode, Blogmaster
A recent paper (Krölov&Langel) describes a technique for the diagnosis of Chlamydia trachomatis which, if developed into a point-of-care test (POCT), could be performed in just twenty minutes and would achieve a considerably greater sensitivity (83%) than any of the POCT alternatives to the current laboratory testing process using Polymerase Chain Reaction (PCR). The technique – Recombinase Polymerase Amplification – is a Nuclear Acid Amplification (NAAT) test (like PCR), but one that allows a simplification of the complex stage of sample preparation such that it can be accomplished swiftly outside a laboratory setting.
The problem with PCR is that because it depends on a laboratory process, patients have to make a second visit to their health care provider in order to receive their diagnosis and treatment – and, some may fail to make that visit. Existing POCT alternatives, however, seriously under-perform on the score of sensitivity (generally under 60% – and in many cases much worse).
Some researchers – including contributors to STIs (STIs/Skidmore; STIs/Gaydos; STIs/Dommelen&Hoebe) – have viewed the availability of such poorly performing tests as entirely detrimental. On the other hand, the scale of the problem of patients lost to follow-up no doubt varies with place and time – and what is required of POCT may not be the same in all contexts (STIs/Rompalo&Gaydos). Others, therefore, have taken a more positive view of the impact of less than perfect POCTs – to extent of acknowledging their potential role in certain contexts, or in combination with other interventions (Swain&Singh). A recent modelling study (STIs/Huang&Barnes) has sought to set the conditions of cost-effectiveness for such a POCT in terms of thresholds for sensitivity, cost of POCT, and “willingness to wait” for the result.
Above all, however, the potential of the less than perfect POCT depends on the size of the “lost to follow up” problem. The Swain and Singh (S&S) model assumes a 93.5% rate of return for treatment achieved for their sample by the Baltimore STD clinics engaged in their study in 2010. This rate does not seem higher that than would be expected in STD clinics in developed countries nowadays (e.g. the UK). It is certainly a good deal better than the 80% given by Schwebke & Hook for Alabama in 1994 – which has since been cited by Swain & Singh (2004) and Krölov & Langel (2014). Assuming the rate of 93.5% lost to follow-up, Swain and Singh estimate degrees of sensitivity that would be needed at a given cost per test for the POCT to achieve a cost-effectiveness that would equal that of the current PCR tests. Thus for a cost per test of $30 a POCT with approx. 85% sensitivity would be required; for a cost per test of $35, a sensitivity of 90%. All of these predictions assume a POCT that would require the patient to wait 40’, and fairly low stated levels of “readiness to wait” (e.g. only 48% on the basis of a patient questionnaire). On the basis of such admittedly rather rough and ready assumptions, the sensitivity of the proposed POCT seems to come at least range of parity with PCR as regards cost-effectiveness.
But these estimates all assume levels of “lost to follow up” such as we find in an STD clinic in developed countries. This situation could, of course, be very different – and more favourable to POCTs – amongst specific populations and less developed countries where rates of return for follow-up could be much lower. A more interesting and relevant modelling study than that of Swain and Singh would have shown the relation between proportion of “lost to follow up” and sensitivity for a given cost per test, rather than treating proportion of “lost to follow up” as the fixed parameter.