Table 6. Summary of Findings of Systematic Evidence Reviewa

Question Number Key Question Level and Type of Evidence Overall Evidence for the Link Findings
1 Does screening for HIV in asymptomatic adolescents and adults reduce premature death and disability and spread of disease? None Not applicable No controlled studies or observational studies link screening directly to health outcomes.
2 Can clinical or demographic characteristics (including specific settings) identify a subgroup of asymptomatic adolescents and adults at increased risk for HIV compared to the general population? II-2. Cohort and cross-sectional studies Good The strongest risk factors for HIV infection from multiple large observational studies are intravenous drug use, male to male sex, and high-risk sexual behaviors. The largest U.S. study found that in federally funded testing sites, 20%-26% of HIV-positive patients reported no risk factors.28 In high-risk settings, several observational studies found that targeted screening based on broad criteria could increase the yield of screening but would still miss 7%-13% of positive patients while testing a much higher proportion.37-39
3 What are the test characteristics of HIV antibody test strategies? Studies of diagnostic test accuracy Good for standard and OraQuick rapid testb; fair for other testing and collection methods Standard testing is associated with a sensitivity and specificity >99%.45-47 Initial studies indicate that FDA-approved rapid tests are associated with similar diagnostic test accuracy, but data from clinical settings are limited for rapid tests other than OraQuick on blood specimens.50-55 Home sampling and oral specimen sampling appear to have diagnostic accuracy similar to that of standard testing,58,59,64 but urine specimens may be associated with lower accuracy.60-63
4 What are the harms (including labeling and anxiety) associated with screening? Is screening acceptable to patients? Studies of diagnostic test accuracy II-2; cohort and cross-sectional studies for harms of screening and acceptability Good for false-positive rates and false-negative rates; fair to good for harms from screening and acceptability of testing False-positive results appear rare with standard testing, even in low-prevalence settings (1 of 250,000 blood donors).48 False-positive results from rapid tests could occur if results are given before confirmatory testing. False-negative results could occur during the window period before seroconversion and provide false reassurance. True-negative results could also provide false reassurance in patients practicing high-risk behaviors. True-positive results are associated with social consequences, anxiety, and labeling, but these harms are difficult to measure. Violence is very frequent in HIV-infected persons, but the impact of screening is not clear. Larger or more recent observational studies have not clearly shown that disclosure increases partner dissolution,87-89 intimate partner violence,84-86 or suicide risk.79 Acceptance rates vary widely even in similar settings (10%-97%) and may be improved by the availability of newer screening methods (rapid tests, noninvasive samples, home-based collection, on-site testing).94 An opt-out testing policy increased testing rates in 1 study.95
5 How many newly diagnosed HIV-positive patients meet criteria for antiretroviral treatment or prophylaxis against opportunistic infections? How many patients who meet criteria for interventions receive them? II-2. Cohort and cross-sectional studies Fair for proportion of patients qualifying for intervention at treatment (little information on initial viral load); good for proportion receiving interventions Seven U.S. studies found that 12%-43% of patients are diagnosed with CD4 cell counts below 0.200 × 109 cells/L, and 46%-80% with CD4 cell counts below 0.500 × 109 cells/L.26,41,112-116 No studies reported initial CD4 cell counts and viral loads in asymptomatic patients. No studies estimated the effects of screening on the proportion of patients qualifying for interventions or the effects of HAART on the proportion of patients qualifying for prophylaxis. A substantial proportion of HIV-positive patients do not receive or decline care. An estimated 36%-63% of infected patients were receiving care at least once every 6 months in 1996;124 38%-58% with positive test results do not return for initial post-test counseling (although about 90% are eventually located),30,117-120 and 53%-85% of infected patients who met guidelines for antiretroviral treatment were receiving them.129-132 Patients with lower CD4 cell counts and higher viral loads appear to have poorer response to antiretroviral therapy, but data on long-term outcomes are lacking.
6 What are the harms associated with the work-up for HIV infection? None Not applicable No evidence.
7a 1. How effective is antiretroviral treatment in improving clinical outcomes (mortality, functional status, quality of life, symptoms, opportunistic infections, or transmission rates)? I, II-2. Randomized, controlled trials; large cohort studies Good for clinical progression and death; fair for quality of life and spread of disease HAART is associated with improved clinical outcomes (clinical progression and death) compared to 2-drug therapy (OR, 0.62 [95% CI, 0.51-0.70]) and other less intense regimens.133 Quality-of-life outcomes from HAART have not been well studied. Beneficial effects of HAART on reducing horizontal transmission by reducing viral load may be offset by increases in risky behaviors,154,159 but there was insufficient evidence with which to estimate the effects of HAART on transmission rates.
2. How effective is counseling on risky behaviors in reducing transmission rates? II. Cohort studies Fair Few data address the effects of counseling and testing on HIV transmission rates in the United States. In Africa, uninfected women's knowledge of the HIV-positive status of their male partner was associated with a reduction in transmission by about 50%.184 Several observational studies indicate that sexually transmitted disease rates decline after an HIV diagnosis but may increase in persons testing negative.185,186 Interactive HIV counseling and testing was more effective than standard didactic counseling and testing in reducing sexually transmitted disease rates in 1 large, good-quality randomized trial, although there were too few cases to determine whether it was more effective at reducing new HIV infections.187 There is insufficient evidence with which to estimate effects of counseling on drug behaviors and transmission rates.
3. How effective are immunizations in improving clinical outcomes (mortality, functional status, quality of life, symptoms, opportunistic infections)? I, II-2. Randomized, controlled trials; large cohort studies Fair for pneumococcal, influenza, and hepatitis B vaccinations; poor for others In 1 randomized trial from Uganda, pneumococcal vaccination was associated with an increased risk for all-cause pneumonia (HR, 1.89 [95% CI, 1.1-3.2]),253 although long-term followup found an unexpected survival advantage (HR, 0.84 [CI, 0.7-1.0])254 (Observational studies mostly found a benefit from vaccination, particularly in patients with higher CD4 cell counts).255-259 Influenza vaccination was associated with a lower risk for symptomatic respiratory illness (49% vs. 29%; P = 0.04) in a clinical trial of HIV-infected patients in a military clinic.260 Hepatitis B vaccination was associated with a lower risk for acute hepatitis B in 1 observational study of HIV-infected persons.261 No studies had clinical outcomes of other immunizations in HIV-positive patients.
4. How effective is prophylaxis against opportunistic infections in improving clinical outcomes (mortality, functional status, quality of life, symptoms, opportunistic infections, or transmission rates)? I, II-2. Randomized, controlled trials; large cohort studies Good overall Good-quality systematic reviews found that preventive medication against PCP reduced the risk for PCP (RR, 0.39 [95% CI, 0.27-0.55]) and was associated with a nonsignificant mortality benefit (RR, 0.87 [CI, 0.60-1.25]).194,195 Some medications effective for PCP prophylaxis were also effective for toxoplasmosis prophylaxis. Two good-quality systematic reviews found that prophylaxis was effective at preventing active tuberculosis (risk reduced by 60%-86%) and death (risk reduced by 21%-23%) in patients with a positive skin test result.199,200 Multiple randomized, controlled trials found that preventive medication was effective for preventing disseminated Mycobacterium avium intracellulare infection, and may be associated with a mortality benefit (HR, ~0.75).201-206 In 2 randomized trials of ganciclovir, prophylaxis against CMV in patients who are positive for CMV antibody may prevent invasive CMV disease but does not appear associated with a significant mortality benefit.206,207
7b In asymptomatic patients with HIV infection, does immediate antiretroviral treatment result in reduced rates of premature death or disability compared to delayed treatment until symptomatic? II-2. Cohort studies Fair Large observational studies that controlled for lead-time bias consistently found that starting HAART at CD4 cell counts > 0.350 ×109 cells/L is associated with better clinical outcomes than starting at a count < 0.200 ×109 cells.210-213 The optimal CD4 cell count at which to start HAART in patients with counts between 0.200 and 0.350 ×109 cells is unclear. Observational studies that have controlled for lead-time bias did not control for other potentially important confounders (such as level of adherence or physician experience).
7c How well do interventions reduce the rate of viremia, improve CD4 cell counts, or reduce risky behaviors? I, II-2. Randomized, controlled trials; large cohort studies Good A fair-quality systematic review of HAART regimens found a rate of viral load suppression to <50 copies/mL at 48 wk of 47% overall (95% CI, 43%-51%).262 Observational studies found that 40%-50% of patients reached and maintained CD4 cell counts > 0.500 ×109 cells during HAART after 4-5 y,263,264 and 47% had a viral load < 50 copies/mL after 6 y.265 Two good-quality systematic reviews found that HIV counseling and testing are associated with decreases in risky sexually behaviors in persons testing positive, but the strength of the association varied according to the group studied.168-169 The strongest association was in heterosexual couples and in those testing positive. More intense or targeted counseling was more effective than standard counseling in several randomized trials.174-178
8 What are the harms associated with antiretroviral therapy? I, II-2. Randomized, controlled trials; large cohort studies Good In numerous clinical trials and observational studies, HAART regimens were associated with clinically significant short-term adverse events. Many patients can be switched to effective alternative regimens. Specific antiretroviral drugs and combinations are associated with specific adverse event profiles. A large, good-quality prospective cohort study found that the incidence of myocardial infarction and cardiac or cerebrovascular events increased with longer exposure to HAART (adjusted RR per year, 1.26 [95% CI, 1.12-1.41] and 1.26 [CI, 1.14-1.38], respectively) for the first 4 y, but the overall rate was low at 3.5 and 5.7 events, respectively, per 1000 person-years.228,229
9 Have improvements in intermediate outcomes (CD4 cell counts, viremia, risky behaviors) been shown to reduce premature death and disability or spread of disease? I, II-2. Randomized, controlled trials; large cohort studies Good for CD4 cell count or viral load and clinical progression and transmission risk; fair for behavior changes and transmission risk A large collaborative analysis of 13 cohort studies found that 6-mo CD4 cell count and viral load were strongly, independently associated with clinical outcomes in patients starting HAART.266 Observational studies found that low viral load was strongly correlated with decreased risk for HIV transmission in heterosexual couples,267 but data from patients treated with HAART are lacking. Condoms have been shown to be associated with decreased risk for transmission from HIV-infected persons.165,166 In mixed populations of infected and uninfected drug users, lower rates of HIV infection were associated with decreased risky drug use behaviors, participation in needle exchange programs, and participation in drug treatment programs.268-270
10 What is the cost-effectiveness of screening for HIV infection? Cost-effectiveness analyses Good Two good-quality cost-effectiveness analyses found that the cost-effectiveness of screening for HIV infections compared to no screening in settings with 1% prevalence was $38 000 to $42 000 per quality-adjusted life-year.243,244 One study found that when transmission benefits were incorporated into estimates, cost-effectiveness remained less than $50 000 per quality-adjusted life-year in settings with prevalences lower than that in the general population.243 Neither study evaluated the incremental cost-effectiveness of universal screening compared to targeted screening strategies in different populations.

aNotes: CMV = cytomegalovirus; FDA = U.S. Food and Drug Administration; HAART = highly active antiretroviral therapy; HR = hazard ratio; OR = odds ratio; PCP = Pneumocystis carinii pneumonia; RR = relative risk.
bOraSure Technologies, Bethlehem, Pennsylvania.

Return to Document

 

AHRQ Advancing Excellence in Health Care