The Need for Systematic Evaluations of Diagnostic Tests

The need to perform cost-efficiency analysis of health care interventions is increasingly brought to the attention of policy makers in both developed and developing countries. This is not only because of the fact that all settings, even the richer ones, have limited resources allocated to the health care sector but also because of the constantly changing dynamics of health determinants and disease epidemics. Of note, the development of new health care technologies almost continuously brings new elements for consideration by policy makers. Few resource-poor countries have institutionalized a formal system for the comprehensive evaluation of laboratory assays and for the translation of the scientific evidence generated by these evaluations into clinical guidelines. Generally speaking, a diagnostic test or strategy is recommended when its good performance is documented by high-quality evidence and when its clinical benefits outweigh the unwanted effects [1]. The most reliable approach to evaluate a diagnostic test or strategy is a randomized, controlled trial, in which patients are randomly assigned to experimental or control arms and in which patient-important outcomes, such as morbidity, mortality, and quality of life, are meticulously investigated [1, 2]. Unfortunately, however, because of the scarcity of such trials, recommendations for use are often based on observational studies designed to measure the analytical accuracy of a test or diagnostic strategy. The translation of international clinical recommendations into national public health care policies is particularly challenging. For example, decision making in public health care must also take into account elements from areas of knowledge outside clinical medicine, including ethical, cultural, political, infrastructural, and economical considerations [3, 4]. In fact, it has been suggested that clinical guidelines should also incorporate considerations regarding resource use [5]. The inclusion of antiretroviral therapy (ART) in high-value intervention packages for control of human immunodeficiency virus (HIV) and AIDS in resourcepoor settings has been supported by cost-effectiveness data. Moreover, it has been further argued that some of the indirect benefits of ART are difficult to capture in cost-effectiveness models [6]. In richer countries, ARTmonitoring includes routine visits to specialized clinics, as well as laboratory testing to check efficacy and to rule out toxicity. In resource-poor settings, especially those with a high burden of HIV/AIDS, routine laboratory monitoring has remained a challenge. Recently, the Development of Anti- Retroviral Therapy in Africa (DART) study, an open, randomized, noninferiority trial, investigated the long-term effects of routine laboratory testing on clinical outcomes in 3 centers in Uganda and 1 in Zimbabwe [7]. After performing baselinelaboratory tests for all patients, the authors enrolled 3321 symptomatic, ART-naive, HIV-infected adults with absolute CD4 T cell counts <200 cells/mm3 into 2 study arms—namely, laboratory and clinical monitoring and clinical driven monitoring. Patients were clinically assessed and tested for hematological characteristics, clinical chemistry, and CD4+ T cell counts every 3 months, but in the arm with clinical driven monitoring, laboratory results weremade available onlywhen clinically indicated. After amedian followup time of 4.9 years, the authors did not find any difference in the rate of loss to follow-up and 5-year survival between the 2 study arms. Moreover, there was no difference in the rate at which new serious adverse events occurred between the both arms. However, the arm with laboratory and clinical monitoring registered fewer new World Health Organization stage 4 events and deaths (P = .001), with higher rates of switches to second-line ART from the second year of follow-up. The authors concluded that routine laboratory testing for toxicity was not required for ART, but CD4 T cell counts should be used from the second year of follow-up to guide the switch to second-line ART. In this issue of Clinical Infectious Diseases, Koenig et al [8] describe a retrospective analysis of a cohort of 1800 patients who started ART at a clinic in Haiti. The authors investigated whether routine laboratory testing was better than clinical driven testing at detecting abnormalities and at saving costs per disability-adjusted life year (DALY) averted by early detection of toxic events. Koenig et al conclude that, in their setting, monitoring for asymptomatic anemia is the only routine laboratory test that is cost-effective, with a savings of US$317 per DALY averted in follow-up. The study found that anemia was most cost-effectively monitored using hematocrit testing, because the other abnormalities identified by complete blood counts did not have significant clinical impact. Similarly, the routine laboratory monitoring for hepatitis and renal insufficiency was not cost-efficient in Haiti. Both the DART study [7] and the article by Koenig et al [8] describe country-level evaluations of accepted diagnostic strategies to gather information about cost-effectiveness. Both studies suggest that routine laboratory follow-up of ART using hematological analysis and clinical chemistry panels is not cost-effective, whereas baseline testing may be of use in identifying specific comorbidities (anemia being the most important in both studies). Although it may be tempting to generalize these results to other resource-poor settings, clinicians, researchers, and policy makers should carefully consider geographical differences in the epidemiology of coinfections (hepatitis B and C, tuberculosis, intestinal parasites, and malaria), the frequency of other morbidities (hepatotoxicity, anemia, thrombocytopenia, neutropenia, lipid metabolism disturbances, and neuropathy), and the ART protocols in use. Moreover, genetic differences in drug metabolism may also significantly influence the frequency of some toxic effects [9]. One other important limitation of the generalization of these results is that clinical research is performed under conditions that are much more carefully controlled, compared with routine medical services. For example, it is conceivable that the high quality of medical care in the arms with clinical drivenmonitoring identified pathological conditions that would have been missed by unsupervised staff working under less-than-ideal conditions in everyday clinical settings. The process to select laboratory tests for public health care programs is a complex decision, and cost-efficiency is undoubtedly amajor consideration to be taken into account. However, it is critical that other criteria, such as the effect of the introduction of a technology on the inequity of health care services or on the expectations of health care workers and patients, are also taken into account. Additionally, clinical protocols may need to be adapted as newer assays are introduced into the market. In the specific case of ART, the arrival of affordable, point-of-care type assays for hematological characteristics, clinical chemistry, opportunistic infections, CD4+ T cell counts, and HIV load may completely alter the current clinical and public health care scenario [10, 11]. The systematic evaluation of diagnostic tests and strategies for adoption in public health care policies is a goal that should be pursued by every resource-poor nation. This goal can be achieved only if appropriate resources are allocated for establishing sound regulatory frameworks and for strengthening national health research institutions. Potential conflicts of interest. I.V.J.: no conflicts.