NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Chou R, Selph S, Dana T, et al. Screening for HIV: Systematic Review to Update the U.S. Preventive Services Task Force Recommendation [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2012 Nov. (Evidence Syntheses, No. 95.)

Cover of Screening for HIV: Systematic Review to Update the U.S. Preventive Services Task Force Recommendation

Screening for HIV: Systematic Review to Update the U.S. Preventive Services Task Force Recommendation [Internet].

Show details

3RESULTS

Key Question 1. What Are the Benefits of Universal or Targeted HIV Screening Versus No Screening or Each Other in Asymptomatic, Nonpregnant Adolescents and Adults on Disease Transmission, Morbidity, Mortality, and Quality of Life?

No randomized trial or observational study compared clinical outcomes between adults and adolescents screened and not screened for HIV infection.

Key Question 2a. What Is the Yield (Number of New Diagnoses) of HIV Screening at Different Intervals in Nonpregnant Adolescents and Adults?

No randomized trial or observational study evaluated the yield of repeat HIV screening compared with one-time screening, or compared the yield of different strategies for repeat screening (e.g., risk-based repeat screening vs. a routine repeat test). The yield of repeated screening would depend in part on the frequency of new HIV infections. Some modeling studies have estimated the cost-effectiveness of strategies involving repeat screening (see the contextual question below).

Key Question 2b. What Are the Effects of Universal Versus Targeted HIV Screening on Testing Acceptability and Uptake in Nonpregnant Adolescents and Adults?

Summary

No study directly evaluated the acceptability of universal versus targeted HIV screening. One fair-quality, nonrandomized study of emergency department (ED) patients found universal, opt-out rapid screening associated with higher likelihood of testing compared with physician-directed, targeted rapid screening (25% vs. 0.8%; relative risk [RR], 30 [95% CI, 26 to 34]), but testing uptake (the proportion of patients offered testing who accepted) was not reported. In two uncontrolled implementation studies of universal HIV screening conducted in primary care settings, 35 percent (standard test) and 60 percent (rapid test) of those offered screening underwent it.

Evidence

The prior USPSTF review found no studies that directly compared acceptance of universal versus targeted HIV screening.1-3 It found that general acceptance of voluntary HIV testing in the United States varied from 11 to 91 percent, with greater uptake in higher prevalence settings, in patients with perceived or acknowledged HIV risk factors, when confidentiality protections were present, and when providers believed testing was beneficial.1-3 Other factors that appeared to increase HIV testing uptake were use of opt-out testing, anonymous testing, and for adolescents, removal of parental consent.

One nonrandomized study published since the prior USPSTF review compared testing rates during periods of universal opt-out rapid HIV screening versus physician-directed, targeted rapid screening in sequential 4-month intervals over 2 years in an ED.68 Universal screening was associated with a much higher likelihood of testing (25% [6,933/28,043] vs. 0.8% [243/29,925]; RR, 30 [95% CI, 26 to 34]), but testing uptake (the proportion of patients offered testing who accepted) was not reported. One uncontrolled implementation study of universal testing in a primary care setting reported 60 percent (574/954) of patients were offered and accepted rapid HIV testing,69 and another reported that 35 percent (105/300) of patients accepted standard HIV testing70 (Appendixes B1 and B2).

Key Question 2c. What Is the Effect of Opt-Out Versus Opt-In Testing or Different Pre- or Post-Test HIV Counseling Methods on Screening Uptake or Rates of Followup and Linkage to Care in Nonpregnant Adolescents and Adults?

Summary

One observational study of computerized, kiosk-based screening found an opt-out approach associated with higher likelihood of testing compared with an opt-in approach (13% vs. 7%; RR, 2.1 [95% CI, 1.9 to 2.4]), but patients who underwent opt-out testing were more likely to report that they had not been informed of HIV testing. Only two patients had newly diagnosed HIV infection, precluding conclusions regarding rates of followup or linkage to care. One other study found opt-out testing associated with lower testing uptake compared with opt-in testing, but results may have been confounded by differences in who offered the testing.

No study compared effects of different pre- or post-test HIV counseling methods on screening uptake or rates of followup and linkage to care.

Evidence

The prior USPSTF review included an uncontrolled implementation study that found that 35 percent (26/74) of HIV-infected persons identified through a routine voluntary screening program in an urgent care center had entered care within 4 months.71 No study was found on effects of opt-out versus opt-in testing on screening uptake in nonpregnant persons, or on effects of different pre- or post-test HIV counseling methods on screening uptake or rates of followup and linkage to care.

One fair-quality, prospective observational study (n=12,827) published since the prior USPSTF review of computerized, kiosk-based screening in the ED found opt-out screening associated with a higher likelihood of testing compared with opt-in testing (13% vs. 7%; RR, 2.1 [95% CI, 1.9 to 2.4])72 (Appendixes B3 and B4). However, patients who underwent opt-out testing were also more likely to report that they had not been informed of HIV testing compared with those who underwent opt-in testing (54% vs. 2.5%; RR, 21 [95% CI, 5.4 to 85]). Only two patients in the study were diagnosed with HIV infection (both during the opt-in period); both were successfully linked to ongoing HIV care.

One other observational study (n=8,732) in an ED setting reported lower testing uptake with opt-out screening offered by ED front desk registration staff compared with opt-in screening offered by ED triage nurses and providers (31% vs. 63%; p<0.01), but results may have been confounded by differences in who offered the testing.73

No study compared effects of different pre- or post-test HIV counseling methods on screening uptake or rates of followup and linkage to care. One randomized trial compared streamlined versus traditional pretest counseling but was excluded because of other differences between arms other than the counseling intervention, including referral for possible testing versus on-site counseling and testing, physician- versus nurse-directed testing, and use of rapid versus standard HIV testing.74

Key Question 2d. What Are the Adverse Effects (Including False-Positive Results and Anxiety) of Rapid Versus Standard HIV Testing in Nonpregnant Adolescents and Adults Not Known to Be at Higher Risk?

Summary

The prior USPSTF review found standard and rapid HIV testing with confirmatory Western blot associated with high sensitivities and specificities, though the positive predictive value associated with rapid testing prior to confirmatory testing decreased as the prevalence of HIV infection decreased. One randomized trial published since the prior USPSTF review directly compared rapid versus standard HIV testing but only identified one confirmed HIV infection. In large studies of rapid testing (without a comparison to standard testing), the positive predictive value was 95 percent in one study of a higher-prevalence (1.1%) setting, and varied widely (16% to 83%) in four studies of lower-prevalence (0.2% to 0.4%) settings. No study compared psychological or other harms associated with rapid versus standard HIV testing.

Evidence

No study in the prior USPSTF review directly compared harms associated with rapid versus standard HIV testing. The prior USPSTF review found standard HIV testing followed by confirmatory Western blot associated with sensitivity ≥99.7 percent and specificity ≥98.5 percent, with a false-alarm rate (1 – positive predictive value) in low-prevalence settings of about 1 in 250,000 (95% CI, 1 in 173,000 to 1 in 379,000).1-3 It found rapid testing prior to confirmatory testing associated with a sensitivity ≥94 percent and specificity ≥99 percent, with positive predictive values of 25 to 50 percent (in settings with a prevalence of 0.3%), and 85 to 95 percent (in settings with a prevalence of 5%). The prior USPSTF review also identified anecdotal reports of other harms of screening, including stigmatization (including verbal and physical abuse) and anxiety, but found insufficient evidence to estimate their magnitude.1-3

One trial published since the prior USPSTF review randomized patients in a Department of Veterans Affairs primary care setting to universal HIV screening based on one of three strategies: nurse-initiated rapid testing (n=84), nurse-initiated standard testing (n=84), or physician-initiated standard testing (n=83), but only identified one patient with a preliminary positive result.74 No other study directly compared rapid versus standard testing and reported positive predictive values.

Five large (sample sizes, 2,002 to 23,900) uncontrolled observational studies published since the 2005 USPSTF review reported positive predictive values associated with rapid testing68, 73, 75-77 (Table 1). In lower-prevalence (0.2% to 0.4%) settings, positive predictive values varied dramatically, from 16 to 83 percent.68, 73, 76, 77 One study appeared to be an outlier,77 reporting a positive predictive value of 16 percent compared with 77 to 83 percent in the other studies. Stratification of the low-prevalence studies according to whether they evaluated a rapid test using an oral fluid (16% and 78%)77 versus finger-stick (77%)76 or whole-blood specimen (83%)68 did not explain the variability in positive predictive values.

Table 1. Positive Predictive Values and False-Positive Results Associated With HIV Rapid Testing.

Table 1

Positive Predictive Values and False-Positive Results Associated With HIV Rapid Testing.

One large study (n=23,900) in a higher-prevalence (1.1%) setting reported a positive predictive value following a positive rapid HIV test (oral fluid or finger-stick specimen) of 94 percent.75

No study evaluated psychological or other adverse effects associated with rapid versus standard HIV testing.

Key Question 2e. What Are the Effects of Universal Versus Targeted HIV Screening on CD4 Counts at the Time of Diagnosis?

Summary

One fair-quality study found universal testing associated with a higher median CD4 count and lower likelihood of CD4 count <0.200 × 109 cells/L at the time of diagnosis compared with targeted HIV screening, but these differences were not statistically significant. No other studies directly compared effects of universal versus targeted HIV screening, though epidemiologic data indicate temporal trends suggesting earlier diagnosis since the 2006 CDC recommendation on routine HIV screening was issued.

Evidence

A high proportion of HIV-infected patients are diagnosed at late stages of disease. In 2008, about one third of patients received an AIDS diagnosis within 1 year of testing HIV-positive.13 The prior USPSTF review1-3 identified no studies on the effects of universal screening on the proportion of patients with HIV infection identified shortly before being diagnosed with AIDS or concurrently with their AIDS diagnosis.

One fair-quality cohort study published since the prior USPSTF review of patients in a large urban ED compared universal opt-out rapid HIV testing (n=6,702) with targeted HIV testing (n=243) (Appendixes B5 and B6).68 The median CD4 count at the time of HIV diagnosis was 0.069 × 109 cells/L (interquartile range, 0.017 to 0.430 × 109 cells/L) for 16 confirmed infections identified during opt-out testing (prevalence, 0.24%) versus 0.013 × 109 cells/L (interquartile range, 0.011 to 0.015 × 109 cells/L) for five confirmed infections (prevalence, 2.1%) identified during diagnostic testing phases (p=0.02 for difference). Nine of 15 patients with HIV infections identified during universal opt-out testing had an initial CD4 count <0.200 × 109 cells/L compared with all four confirmed HIV infections identified during targeted testing (60% vs. 100%; RR, 0.66 [95% CI, 0.40 to 1.1]).

One other observational study (n=8,732) reported a mean CD4 count of 0.415 × 109 cells/L (standard deviation [SD], 0.237 × 109 cells/L) in eight new, confirmed HIV infections (prevalence, 0.2%) identified during universal opt-in screening offered by ED triage nurses and providers versus 0.307 × 109 cells/L (SD, 0.274 × 109 cells/L) in 21 infections (prevalence, 0.4%) identified during universal opt-out screening offered by ED front desk registration staff (p=0.84).73 Twenty-five percent of patients diagnosed during opt-in screening had a CD4 count <0.200 × 109 cells/L versus 48 percent diagnosed during opt-out screening (RR, 0.52 [95% CI, 0.15 to 1.9]). Results may have been confounded by differential HIV testing acceptance rates in the two groups (31% for opt-out testing and 63% for opt-in testing), perhaps due in part to differences in who offered the testing.

No other study directly evaluated effects of universal versus targeted screening on CD4 counts at the time of diagnosis, though epidemiologic data may provide some indirect evidence. The CDC reported that the proportion of newly diagnosed patients in the United States with a late diagnosis (defined as CD4 cell count <0.200 × 109 cells/L or AIDS-defining illness within 12 months of HIV diagnosis) decreased from 37 percent between 2001 and 2004 to 32 percent in 2007.60 Similarly, a large cohort study (n=44,491) reported a decrease over time in the proportion of HIV-positive patients initially presenting to care with a CD4 cell count <0.350 × 109 cells/L from 1997 to 2007 (from 62% to 54%), with an increase in median CD4 count at presentation of 0.061 × 109 cells/L.78 One study (n=4,478) in Washington, D.C., found that the median CD4 count at the time of HIV diagnosis increased from 0.266 × 109 cells/L in 2005 to 0.361 × 109 cells/L in 2009, though the statistical significance of the difference was not reported.79 Another, smaller study (n=1,203) also reported a temporal trend for lower likelihood of late diagnosis (39% in 2000–2001 and 35% in 2008–2009), though the difference was not statistically significant.80 Although these trends appear to temporally coincide with the CDC recommendations for universal opt-out HIV screening released in 2006,6 it is not possible to determine causality between increased testing and earlier diagnosis based on these data.

Key Question 2f. What Are the Effects of Universal Versus Targeted HIV Screening on Rates of Followup and Linkage to Care in Nonpregnant Adolescents and Adults Who Screen Positive?

Summary

Three observational studies published since the prior USPSTF review reported rates of followup or linkage to care following a new HIV diagnosis found during universal testing, ranging from 75 to 100 percent. The only study that directly compared universal with targeted testing reported very high rates of followup (defined as attending at least one HIV clinic visit) with either strategy (97% to 100%). All studies were limited by small numbers of patients with newly diagnosed HIV infection.

Evidence

In order to realize the potential clinical benefits from HIV screening, patients must be successfully linked to HIV care following diagnosis. The prior USPSTF review identified little evidence on the effect of universal versus targeted HIV screening on linkage to care following HIV diagnosis.1-3 It included one uncontrolled study that found that 35 percent (26/74) of HIV-infected persons identified through a universal voluntary screening program in an urgent care center had entered care within 4 months.71 Another uncontrolled study, also performed in an urgent care center, found that at least 70 percent (42/60) of newly diagnosed HIV-infected persons had one or more documented followup visits following identification through routine screening.81

Three studies68, 73, 76 published since the prior USPSTF review reported linkage to care following universal HIV testing (Appendixes B7 and B8). One study compared universal with targeted screening and two reported rates of linkage to care after universal testing (one study73 evaluated two strategies of universal testing). All studies were limited by small numbers of newly diagnosed HIV infections (17 to 36 cases).

The study that directly compared universal with targeted screening (36 new HIV cases) was a fair-quality, nonrandomized study conducted in a large urban ED that found a very high likelihood of attending at least one HIV clinic appointment in patients diagnosed with either universal or targeted testing (97% vs. 100%; RR, 1.0 [95% CI, 0.81 to 1.3]).68 An uncontrolled study of universal rapid HIV testing in Federally Qualified Health Centers found 14 of 17 (82%) patients with confirmed HIV infections were linked to HIV care following diagnosis.76 A pre-post evaluation of universal opt-in or opt-out rapid oral HIV screening implementation (29 new HIV cases identified) in an ED reported similar rates of linkage to care within 90 days following HIV diagnosis with either strategy (75% [6/8] vs. 77% [16/21]; RR, 0.98 [95% CI, 0.62 to 1.6]).73

Key Question 3a. To What Extent Does Knowledge of HIV-Positive Status Affect Behaviors Associated With Increased Risk of HIV Transmission in Nonpregnant Adolescents and Adults?

Summary

Four before-after studies not included in the prior USPSTF review addressed effects of knowledge of HIV-positive status on risk behaviors. As in the prior USPSTF review, the studies found knowledge of HIV-positive status associated with reduced self-reported risky behaviors in all populations studied.

Evidence

The prior USPSTF evidence review1-3 included two systematic reviews on the association between HIV-positive status and high-risk behaviors.82, 83 Both reviews found greater self-reported reductions in unprotected intercourse in persons testing HIV positive and in serodiscordant couples compared with those testing negative or those who were untested or unaware of their status. Interpretation of these findings was difficult because the primary studies in the reviews evaluated diverse populations and frequently had methodological shortcomings, such as retrospective design, low participation rates, or high loss to followup. Although these studies relied on self-reported behavior, with its attendant shortcomings, there is no practical alternative for assessing these outcomes. Reasons for HIV testing were typically not reported in the primary studies, so the applicability of results to asymptomatic patients undergoing screening was unclear.

Four before-after studies not considered in the prior USPSTF review evaluated the association between knowledge of HIV-positive status and behaviors associated with increased risk of HIV transmission84-87 (Table 2, Appendix B9). Sample sizes ranged from 73 to 560 and behaviors were evaluated from 1 month to 2 or more years following diagnosis. One study was rated good-quality87 and three studies fair-quality84-86 (Appendix B10). All studies relied on self-reported risky behaviors and one86 relied on retrospective recall of pre-HIV diagnosis behaviors. Two studies focused on high-risk groups (MSM or injection drug users), potentially limiting applicability to individuals without these risk factors.85, 87

Table 2. Effect of Knowledge of HIV-Positive Status on Risky Behaviors.

Table 2

Effect of Knowledge of HIV-Positive Status on Risky Behaviors.

One retrospective before-after study of a mixed population (n=487) of HIV-positive persons (injection drug users, noninjection drug-using heterosexual individuals, and MSM) found a significantly lower likelihood of self-reported injection drug use 2 or more years following HIV diagnosis compared with prior to diagnosis (32% vs. 54%).86 The study also found increased condom use after compared with before HIV diagnosis during vaginal (40% vs. 5.5%), anal (32% vs. 4.1%), or oral-genital sex (9.0% vs. 0.9%) with stable partners (p<0.0005 for all differences). Although patients were also less likely to have stable partners following an HIV diagnosis compared with before diagnosis (77% vs. 89%; p<0.0005), likelihood of condom use during intercourse with occasional partners also increased after HIV diagnosis. Patients were also less likely to report engaging in sex for money or drugs following HIV diagnosis (6.8% vs. 13%; p<0.0005) or engaging in sex with sex workers (7.2% vs. 16%; p<0.0005). One other small (n=16) before-after study of heterosexual individuals also found reduced risky sexual behaviors 3 months following a diagnosis of HIV infection compared with before diagnosis.84

Two studies of high-risk populations also found decreases in high-risk behaviors following HIV diagnosis. A prospective before-after study of MSM with primary HIV infection (n=98) found greater self-reported condom use (proportion always using during insertive anal intercourse, 61% vs. 31%; p<0.01) and fewer sexual partners (66% reported fewer sex partners, 27% no change, and 7.1% more partners; p<0.001) 3 months following diagnosis compared with at the time of testing.87 Seventy-six percent reported no high-risk behaviors at all following HIV diagnosis (high-risk behaviors defined as unprotected anal intercourse with a regular partner of unknown or HIV-negative status, unprotected anal intercourse with a casual male partner, or incident STI). Another prospective before-after study found that 26 percent (11/42) of HIV-positive injection drug users reported cessation of injection drug use 1 to 6 months following diagnosis, 73 percent (19/26) stopped lending needles, 62 percent (23/37) stopped borrowing needles, and 38 percent (27/72) increased use of needle exchange programs.85 Among males, 50 percent (9/18) had ceased sexual relations over the past 3 months, and all five men previously engaged in sex work had stopped this activity.

Key Question 3b. To What Extent Does Use of Antiretroviral Therapy Affect Behaviors Associated With Increased Risk of HIV Transmission in Nonpregnant Adolescents and Adults?

Summary

Seven observational studies not included in the prior USPSTF review addressed the effect of ART use on HIV risk behaviors. The studies primarily used a cross-sectional design and had methodological shortcomings, including failure to report baseline differences or to adjust for potential confounders. They found no clear association between ART use and increase in self-reported risky behaviors, with some studies showing decreased risky behaviors.

Evidence

The prior USPSTF review identified one good-quality meta-analysis that found no association between ART use in HIV-infected persons and increased likelihood of unprotected sex.89 However, some individual studies included in the prior USPSTF review reported associations between ART use and increased risk of high-risk sexual behaviors and in MSM,90 as well as associations between ART use and increased likelihood of developing an STD91 and higher risk for pregnancy.92

Five cross-sectional studies,93-97 one prospective cohort study,98 and one before-after study99 not included in the prior USPSTF review evaluated the association between ART use and high-risk behaviors (Table 3, Appendix B11). Sample sizes ranged from 67 to 4,016. In the prospective cohort study, duration of followup averaged 8 years.98 All studies were rated fair-quality93-99 (Appendix B12). Methodological shortcomings included group differences between those taking and not taking ART93 or insufficient information to compare groups by ART use at baseline.94-97 Three studies did not adjust for or did not clearly describe statistical adjustments for potential confounders,93, 96, 98 and one study did not adjust for sex.95 Risky behaviors were self-reported in all studies.93-99 Three studies included only high-risk groups (MSM or injection drug users).96, 98, 99

Table 3. Effect of Use of Antiretroviral Therapy on Risky Behaviors.

Table 3

Effect of Use of Antiretroviral Therapy on Risky Behaviors.

Three observational studies of women or mixed (male or female) populations of heterosexual patients found no association between ART use and increased risky sexual behaviors, with two studies showing decreased risk.93, 95, 97 A cross-sectional Spanish study of 625 HIV-serodiscordant heterosexual couples found lower likelihood of self-reported unprotected sexual intercourse in the preceding 6 months in couples in which the index partner was taking ART compared with couples in which the index partner was not taking ART (46% vs. 57%; p=0.02).93 A cross-sectional U.S. study also found trends toward reduced likelihood of engaging in risky behaviors in women (n=1,104) or heterosexual men (n=803) taking ART compared with those not taking ART, though differences were not statistically significant.97 A cross-sectional United Kingdom study found no association between ART use and unprotected intercourse in women (n=480) or heterosexual men (n=224).95

Six observational studies of high-risk populations (MSM or injection drug users) also found no clear increases in risky behaviors (high-risk sexual behaviors or injection drug use) after initiation of ART compared with before initiation of therapy, or in HIV-infected patients taking ART compared with those not on therapy.94-99 Two of these studies found ART use associated with reduced likelihood of high-risk behaviors.94, 97 One (n=4,016) found ART use associated with decreased risk of engaging in risky sexual behaviors over the past 6 months in MSM (adjusted odds ratio [OR], 0.73 [95% CI, 0.54 to 1.0]).97 The other (n=874) found ART use associated with decreased risk of unprotected anal or vaginal intercourse (adjusted OR, 0.70 [95% CI, 0.50 to 1.0]) in a population primarily consisting of gay men and injection drug users.94

Key Question 4a. How Effective Is Antiretroviral Therapy in Reducing Transmission of HIV in Nonpregnant Adolescents and Adults With Chronic HIV Infection?

Summary

A good-quality systematic review found consistent evidence from one randomized, controlled trial and seven observational studies that ART use is associated with decreased risk of HIV transmission from HIV-positive persons to uninfected sexual partners. In the randomized trial, the risk of HIV seroconversion in uninfected sexual partners of patients with baseline CD4 counts of 0.350 to 0.550 × 109 cells/L was much lower in those randomized to immediate versus delayed ART after 1.7 years of followup (HR, 0.04 [95% CI, 0.01 to 0.27] for genomically linked seroconversion), consistent with the pooled risk estimate from observational studies (HR, 0.16 [95% CI, 0.07 to 0.35]).

Evidence

The prior USPSTF review1-3 identified no studies that directly evaluated the association between ART use and risk of transmission. However, ART could decrease risk of HIV transmission from infected persons by decreasing viral load.19, 104-106 One pre-HAART era cohort study found zidovudine associated with lower risk of heterosexual transmission compared with no treatment in monogamous men (RR, 0.5 [95% CI, 0.1 to 0.9]).107

A recent, good-quality systematic review evaluated the association between ART use and risk of HIV transmission from HIV-positive persons to uninfected sexual partners108 (Appendixes B13 and B14). It included one randomized, controlled trial109 and seven observational studies.93, 107, 110-114

The good-quality randomized, controlled trial (HIV Prevention Trials Network [HPTN] 052) compared early ART (started at enrollment) versus delayed therapy (after a decline in CD4 count to <0.250 × 109 cells/L or onset of symptoms) in HIV-infected patients with baseline CD4 counts of 0.350 to 0.550 × 109 cells/L and an HIV-negative partner109 (Appendixes B15 and B16). Fifty-four percent of the 1,763 couples were from Africa, with the remainder from Brazil, India, Thailand, and the United States. Ninety-seven percent of couples were heterosexual and 94 percent were married. All couples received condoms and counseling along with quarterly HIV testing of uninfected partners. The trial was designed to follow patients for 5 years, but was terminated early after meeting prespecified criteria for efficacy in interim analyses. At a median followup of 1.7 years, there were 39 seroconversions among all participants in the trial (1.2 per 100 person-years). Risk of seroconversion in HIV-negative partners was much lower in the early compared with the delayed therapy group (0.3 vs. 2.2 per 100 person-years; HR, 0.11 [95% CI, 0.04 to 0.32]). When restricted to the 28 cases that were genomically linked to the HIV-infected patient enrolled in the trial (one transmission in the early-therapy group and 27 transmissions in the delayed-therapy group), the HR was 0.04 (95% CI, 0.01 to 0.27). All cases of linked transmission in the delayed-therapy group occurred prior to initiation of ART in the HIV-infected partner.

Results of seven observational studies93, 107, 110-114 (Appendixes B15 and B17) included in the systematic review108 were consistent with the randomized trial.109 Sample sizes ranged from 93 to 3,408 couples, with typical followup between 1 and 3 years (range, 3 months to 9 years). All seven observational studies were cohort studies of HIV-serodiscordant, heterosexual couples from Africa, Italy, Spain, Brazil, or China. Six cohort studies were rated fair-quality93, 107, 110-112, 114 and the seventh113 was reported as a conference abstract only and could not be quality rated. Three studies adjusted for possible confounding variables such as age, sex, condom use, or frequency of sexual intercourse.107, 110, 114 Four studies reported low loss to followup.107, 110, 111, 114

Six of the seven observational studies reported decreased risk of HIV transmission from persons taking ART compared with those who were untreated.93, 107, 110-113 Of the 436 total HIV transmissions in the seven observational studies, 71 were in couples in which the HIV-infected individual was receiving ART and 365 transmissions were in couples in which the HIV-infected individual was not receiving ART (pooled HR, 0.34 [95% CI, 0.13 to 0.92]).108 However, there was substantial statistical heterogeneity (I2=73%). Excluding one study with inadequate person-time data114 and one older study that included persons treated with monotherapy only107 resulted in a pooled HR of 0.16 (95% CI, 0.07 to 0.35) and eliminated the statistical heterogeneity (I2=0%). The treatment effect was also more pronounced when the analysis was restricted to couples in which the HIV-infected individual had a CD4 count <0.200 × 109 cells/L (pooled HR, 0.06 [95% CI, 0.01 to 0.54]),93, 110-112 couples in which the index case was male (pooled HR, 0.02 [95% CI, 0.00 to 0.89]),93, 113 or couples residing in low/middle income countries (pooled HR, 0.24 [95% CI, 0.06 to 1.03]).110-114

Key Question 4b. How Effective Is Behavioral Counseling in Reducing Transmission of HIV in Nonpregnant Adolescents and Adults With Chronic HIV Infection?

Summary

Two studies of counseling interventions identified too few cases of new HIV infection to reliably estimate effects of counseling on risk of transmission.

Evidence

The prior USPSTF review1-3 found no randomized trials or controlled observational studies on the effects of counseling HIV-positive persons regarding risky behaviors on HIV transmission risk. One uncontrolled prospective U.S. study of 144 serodiscordant heterosexual couples reported reduced risky behaviors and no HIV transmission following counseling after 193 couple-years of followup.115

There remains little direct evidence on effects of testing and counseling regarding risky behaviors on HIV transmission (Appendixes B18, B19, and B20). Trials of counseling have generally not been designed to assess the effect of counseling on HIV transmission rates and have been underpowered. A cluster-randomized, controlled trial of African American, HIV-serodiscordant, heterosexual couples (n=536 couples) from four U.S. cities who had recently engaged in unprotected sexual intercourse found an Afrocentric HIV-STD risk-reduction counseling intervention116 associated with increased likelihood of condom use compared with an attention-matched, individual-focused health promotion comparison group (63% vs. 48%; RR, 1.4 [95% CI, 1.2 to 1.7]), but after 12 months, there were only two HIV transmissions out of 260 couples in the counseling group and only three HIV transmissions out of 275 couples in the comparison group.117 Similarly, a before-after study of 564 serodiscordant couples who participated in couples counseling and testing in Madrid from 1989 to 2007 found an increased likelihood of 100 percent condom use following counseling compared with before counseling (69% vs. 49%; p<0.001), but there were only five seroconversions during 1,279 couple-years of followup.118

No study estimated the effects of testing and counseling HIV-positive persons on injection drug use behaviors and transmission rates.

Key Question 4c. In Asymptomatic, Nonpregnant Adolescents and Adults With Chronic HIV Infection, What Are the Effects of Initiating Antiretroviral Therapy at Different CD4 Counts or Viral Load Thresholds on Morbidity, Mortality, and Quality of Life?

Summary

The prior USPSTF review found good-quality evidence that ART is associated with decreased risk of AIDS events and mortality compared with placebo or less-intensive regimens in patients with CD4 counts <0.200 × 109 cells/L. Two randomized, controlled trials (including one subgroup analysis) published after the prior USPSTF review found initiation of ART at CD4 counts <0.250 × 109 cells/L associated with substantially increased risk of death or AIDS events compared with initiation at CD4 counts >0.350 × 109 cells/L. Recent large, observational studies incorporating data from 12 to 23 cohorts also consistently found initiation of ART at CD4 counts between 0.350 and 0.500 × 109 cells/L associated with decreased risk of mortality, or a trend toward decreased risk, compared with deferred or no ART. Four studies evaluating initiation of ART at CD4 counts >0.500 × 109 cells/L were inconsistent, with one study showing beneficial effects on clinical outcomes and three studies finding no clear benefit.

Two studies reported inconsistent results for the association between viral load at the time of initiation of ART and subsequent mortality.

Evidence

CD4 count. The prior USPSTF review included good-quality randomized, controlled trials119-121 and observational studies36, 122-128 that consistently found ART associated with decreased risk of AIDS events and mortality compared with placebo or less-intensive regimens in patients with CD4 counts <0.200 × 109 cells/L. Evidence showing benefits of starting ART at higher CD4 counts was limited. Although a Swiss cohort study found starting ART at CD4 counts >0.350 × 109 cells/L associated with reduced risk of mortality and progression to AIDS compared with starting at counts <0.350 × 109 cells/L,129 three U.S. cohort studies found no difference in risk between starting ART at CD4 counts between 0.350 and 0.500 × 109 cells/L versus delaying until CD4 counts were between 0.200 and 0.350 × 109 cells/L.126-128

Two good-quality randomized trials109, 130 published since the prior USPSTF and one subgroup analysis131 from another good-quality randomized trial evaluated effects of initiating ART at different CD4 count thresholds (Appendixes B21 and B22). Five observational studies (reported in six publications) that each combined data from 12 to 23 U.S., European, and Australian cohorts (∼9,000 to >60,000 participants; duration of followup, 1 to 5 years, with substantial overlap in the cohorts included in the studies) also evaluated effects of starting ART at different CD4 count thresholds132-137 (Appendix B23). All studies were rated fair-quality (Appendix B24). None reported blinding of outcome assessors or those analyzing data, and attrition rates were often not reported or unclear. Although all studies adjusted for important confounders in their analyses, most provided insufficient information to determine baseline comparability of patients started and not started on ART in different CD4 count strata.

Three randomized, controlled trials found delayed initiation of ART associated with increased risk of the combined outcome of death or AIDS-related events (Table 4). A retrospective subgroup analysis of patients (n=477) in the Strategies for Management of Antiretroviral Therapy (SMART) randomized trial who were treatment-naive or had been off therapy for at least 6 months found initiation of ART at CD4 counts <0.250 × 109 cells/L associated with increased risk of death or AIDS events compared with initiation at CD4 counts >0.350 × 109 cells/L after a mean of 18 months (HR, 5.3 [95% CI, 1.3 to 9.6]).131 The SMART trial was conducted in 33 primarily nonresource-poor countries. HPTN 052, conducted in 1,763 serodiscordant partners from primarily resource-poor countries, found initiation of ART at CD4 counts <0.250 × 109 cells/L associated with increased risk for the combined endpoint of death or AIDS events compared with initiation at CD4 counts between 0.350 and 0.550 × 109 cells/L (adjusted HR, 1.7 [95% CI, 1.1 to 2.5]), though these results were strongly influenced by the incidence of extrapulmonary tuberculosis (RR, 5.6 [95% CI, 1.7 to 20]).109 Results for mortality or pulmonary tuberculosis were not significant when these outcomes were considered individually. The third randomized trial (n=816) found initiation of ART at CD4 counts <0.200 × 109 cells/L associated with increased risk of mortality compared with initiation at CD4 counts of 0.201 to 0.350 × 109 cells/L (HR, 4.0 [95% CI, 1.6 to 9.8]; p=0.001), but is less directly applicable to the U.S. population, as it was conducted in Haiti and CD4 count thresholds for treatment in both groups were lower than those typically used in the United States.130

Table 4. Effect of Initiating Antiretroviral Therapy at Different CD4 Counts or Viral Load Thresholds on Progression to AIDS or Mortality.

Table 4

Effect of Initiating Antiretroviral Therapy at Different CD4 Counts or Viral Load Thresholds on Progression to AIDS or Mortality.

Four observational studies consistently found initiation of ART at CD4 counts between 0.350 and 0.500 × 109 cells/L associated with decreased risk of mortality compared with deferred or no ART (Table 4).132, 134-136 One other study found a reduction in risk that was not statistically significant.137 The largest study, the HIV Cohorts Analyzed Using Structural Approaches to Longitudinal Data (HIV-CAUSAL) (n=62,760 from 12 cohorts), found initiation of ART at CD4 counts of 0.350 to 0.500 × 109 cells/L associated with decreased risk of mortality compared with noninitiation within this CD4 count range after an average of 3.3 years of followup (adjusted HR, 0.55 [95% CI, 0.41 to 0.74]).134 Similarly, the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) (n=17,517 from 22 cohorts) found initiation of ART at CD4 counts of 0.350 to 0.500 × 109 cells/L associated with decreased risk of death compared with deferred treatment within these thresholds after an average of 3 years of followup (adjusted RR, 0.61 [95% CI, 0.46 to 0.83]).135 Initiation of ART at CD4 counts >0.350 × 109 cells/L was also associated with decreased risk of the combined outcome of AIDS-defining events or death compared with deferred or no initiation of ART in two studies (Table 4).132, 136 One other study found a reduction in risk that was not statistically significant.137

Studies on initiation of ART at CD4 counts >0.500 × 109 cells/L were less consistent. NA-ACCORD found initiation of ART at CD4 counts >0.500 × 109 cells/L associated with decreased mortality compared with deferred therapy (adjusted RR, 0.54 [95% CI, 0.35 to 0.83]),135 and HIV-CAUSAL found decreased mortality risk that was not statistically significant after 3 years (adjusted HR, 0.77 [95% CI, 0.58 to 1.0]).134 Another analysis from HIV-CAUSAL that directly compared initiation of ART at CD4 counts >0.500 × 109 cells/L versus initiation at counts >0.350 × 109 cells/L found no difference in mortality (HR, 0.99 [95% CI, 0.73 to 1.4]).133 Two other large cohort studies found initiation of ART at CD4 counts >0.500 × 109 cells/L associated with no difference in risk of mortality when compared with noninitiation after 5 years (adjusted HR, 1.0 [95% CI, 0.49 to 2.1])132 or when compared with slightly delayed initiation after 3 years (adjusted HR, 0.93 [95% CI, 0.60 to 1.4] for starting at CD4 counts of 0.451 to 0.550 versus 0.351 to 0.450 × 109 cells/L).137 In all four studies, absolute mortality rates were low in patients with CD4 counts >0.500 × 109 cells/L (range, 2% to 5%).

Results were also mixed for the combined outcome of mortality plus AIDS-defining events, which was not reported in NA-ACCORD.135 HIV-CAUSAL found initiation above a threshold of 0.500 × 109 cells/L associated with decreased risk of AIDS-defining events or death compared with initiation above 0.350 × 109 cells/L (HR, 0.72 [95% CI, 0.59 to 0.88]).133 Two other studies found no clear association between starting versus not starting ART at CD4 counts >0.500 × 109 cells/L and risk of AIDS-defining events or death (Table 4).132, 137

Viral load. Two studies reported inconsistent results for the association between viral load at time of initiation of ART and subsequent mortality (Appendix B23).134, 136 HIV-CAUSAL (n=62,760 from 12 cohorts) found initiation of ART at higher viral loads associated with greater reduction in mortality risk (adjusted HR, 0.82 [95% CI, 0.64 to 1.0] for initiation at viral load <10,000 copies/mL vs. noninitiation; adjusted HR, 0.46 [95% CI, 0.36 to 0.60] for viral load of 10,000 to 100,000 copies/mL; and adjusted HR, 0.36 [95% CI, 0.28 to 0.45] for viral load >100,000 copies/mL).134 Another study, the Antiretroviral Therapy Cohort Collaboration (n=20,379 from 12 cohorts), found initiation of ART at viral loads of 10,000 to <100,000 copies/mL and 1,000 to <10,000 copies/mL each associated with decreased risk of mortality or progression to AIDS compared with initiation at a viral load ≥100,000 copies/mL (adjusted HRs, 0.80 [95% CI, 0.73 to 0.88] and 0.80 [95% CI, 0.68 to 0.95], respectively).136

Key Question 5. What Are the Longer-Term Harms Associated With Antiretroviral Therapy in Nonpregnant Adolescents and Adults With Chronic HIV Infection?

Summary

The 2005 USPSTF review included results from the Data Collection on Adverse Events of Anti-HIV Drugs (DAD) study, which found longer duration of exposure to ART associated with increased risk of myocardial infarction (adjusted RR per year of exposure, 1.3 [95% CI, 1.1 to 1.4]). More recent analyses from DAD with up to 6 years of followup were consistent with earlier results in finding slightly increased risk of myocardial infarction with use of some protease inhibitors. Two studies (DAD and one other cohort study) found abacavir associated with increased risk of cardiovascular events, but two other studies found no such association. There was no clear association between use of other nucleoside reverse transcriptase inhibitors or nonnucleoside reverse transcriptase inhibitors and increased risk of adverse cardiovascular events.

Evidence

The prior USPSTF review included results from the large (n=23,468), ongoing DAD study, which found increased risk of myocardial infarction associated with longer exposure to ART (adjusted RR per year of exposure, 1.3 [95% CI, 1.1 to 1.4]), though absolute event rates were low (3.5/1,000 person-years).141

Subsequent analyses from DAD142-144 and three other cohort studies reported cardiovascular harms associated with ART through up to 4 to 6 years of followup (Appendix B25).145-147 Sample sizes ranged from 2,952 to >30,000. All of the studies were rated good-quality except for one,145 which was rated fair-quality due to lack of detail about baseline patient characteristics and blinding of study personnel (Appendix B26).

Like the earlier DAD results, the most recent DAD analysis found longer exposure to indinavir alone (adjusted RR per year of exposure, 1.1 [95% CI, 1.1 to 1.2]), ritonavir-boosted indinavir (adjusted RR per year of exposure, 1.2 [95% CI, 1.1 to 1.3]), and ritonavir-boosted lopinavir (adjusted RR per year of exposure, 1.1 [95% CI, 1.0 to 1.2]) each associated with slightly increased risk of myocardial infarction compared with nonuse, after adjustment for age, sex, HIV infection risk group, ethnicity, calendar year, family history of cardiovascular disease, prior cardiovascular disease, smoking status, body mass index, and other factors (Table 5).144 No other protease inhibitor was associated with increased myocardial risk.

Table 5. Cardiovascular Events and Antiretroviral Therapy Use.

Table 5

Cardiovascular Events and Antiretroviral Therapy Use.

Evidence on the association between the nucleoside reverse transcriptase inhibitor abacavir and risk of myocardial infarction is mixed. Although two studies found abacavir use associated with increased risk (adjusted RRs, 1.7 and 2.0),144, 146 two others found no association (adjusted HRs, 0.6 and 1.2)145, 147 after 4 to 6 years of followup (Table 5).

The DAD study also found recent didanosine use associated with increased myocardial infarction risk (adjusted RR, 1.4 [95% CI, 1.1 to 1.8]), but no association when analyses were based on cumulative didanosine exposure.144 There was no association between use of other nucleoside reverse transcriptase inhibitors or the nonnucleoside reverse transcriptase inhibitors nevirapine or efavirenz and increased risk of cardiovascular events.144

Key Question 6a. To What Extent Are Improvements in Viremia Associated With Reductions in HIV Transmission Rates in Nonpregnant Adolescents and Adults?

Summary

The prior USPSTF review included seven observational studies that consistently found a strong association between lower individual viral load and decreased risk of heterosexual transmission of HIV infection. Three observational studies not included in the prior USPSTF review reported results consistent with these findings. Three other observational studies (two with overlapping populations) found lower community viral load (defined as the average viral load in a defined population) associated with decreased risk of HIV transmission.

Evidence

The prior USPSTF report3 included seven observational studies19, 106, 150-154 that consistently found an association between lower individual viral load and lower risk of heterosexual transmission of HIV infection (Table 6). The strongest evidence was from a good-quality prospective cohort study of 415 serodiscordant couples in rural Uganda (a setting in which ART was not available), which found viral load to be the strongest predictor for heterosexual HIV transmission (male to female or female to male).19 The rate of transmission in patients with HIV-1 viral load <1,500 copies/mL was zero out of 51, and increased in a dose-response fashion to 23 per 100 person-years at a viral load ≥50,000 copies/mL. The adjusted RR for transmission was 12 (95% CI, 5.0 to 35) for a viral load >50,000 copies/mL compared with <3,500 copies/mL. Another analysis of the same Ugandan cohort reported an adjusted RR of transmission per coital act of 16 (95% CI, 3.1 to 296) for a viral load of 1,700 to 12,499 copies/mL versus <1,700 copies/mL, 18 (95% CI, 3.4 to 329) for viral loads of 12,500 to 38,499 copies/mL, and 28 (95% CI, 5.4 to 507) for viral loads ≥38,500 copies/mL.20

Table 6. HIV Transmission by Viral Load.

Table 6

HIV Transmission by Viral Load.

Three observational studies not included in the prior USPSTF review evaluated the association between viral load in individual patients and risk of HIV transmission110, 155, 156 (Table 6, Appendixes B27 and B28). Two evaluated heterosexual couples in Africa.110, 155 One cohort study evaluated 3,408 HIV-discordant couples in seven African countries in which the index case was infected with both HIV and herpes simplex virus-2 and had CD4 counts >0.250 × 109 cells/L.110 The group at greatest risk for HIV transmission were individuals with CD4 counts of 0.200 to 0.349 × 109 cells/L and a viral load of ≥50,000 copies/mL (incidence per 100 person-years, 4.7 [95% CI, 3.2 to 6.6]). A case-control study of heterosexual couples in Zambia (109 cases of HIV transmission to the uninfected partner and 208 control couples with no transmission) found a dose-dependent association between higher viral load and risk of transmission from females to males (RR per log viral load, 2.5 [95% CI, 1.5 to 4.0]) as well as from males to females (RR per log viral load, 1.8 [95% CI, 1.2 to 2.8]).155 HIV RNA viral load was also a predictor of transmission risk in a cohort study (1,144 men, 41 cases) of MSM in the United Kingdom (RR per log viral load, 1.6 [95% CI, 1.2 to 2.3]).156

Studies that evaluated community viral load (the average viral load in a defined population) also found an association between higher viral load and increased risk of transmission157-159 (Appendixes B27 and B28). One study found that for every 10-fold decrease in the median viral load of all HIV-infected individuals in a specific year in British Columbia, the number of new HIV cases decreased by a factor of 0.86 (95% CI, 0.75 to 0.98) after adjusting for year and number of individuals taking ART, despite increased rates of other STDs in this population.158 An analysis of a subgroup of the above population, consisting of injection drug users in inner-city Vancouver, also found community viral load independently associated with time to HIV seroconversion (HR per log10 increase, 3.3 [95% CI, 1.8 to 6.1]) after adjusting for other markers of risk.159 Similarly, a study based on San Francisco's HIV/AIDS surveillance system found both higher mean community viral load and sum total viral load associated with increased risk of HIV incidence (unadjusted, p=0.003 and p=0.002, respectively).157

Key Question 6b. To What Extent Are Improvements in Risky Behaviors Associated With Reductions in HIV Transmission Rates in Nonpregnant Adolescents and Adults?

Summary

The prior USPSTF review included two systematic reviews that found consistent condom use associated with substantially reduced risk of sexual transmission of HIV infection. Two observational studies not included in the prior USPSTF review were consistent with these findings.

No study evaluated effects of safer injection drug use behaviors by HIV-positive patients on risk of HIV transmission.

Evidence

The prior USPSTF review included a systematic review1-3 (11 prospective studies, two retrospective studies, and one case report) of primarily HIV-discordant heterosexual couples from the United States, Europe, Africa, and Haiti that found consistent use of condoms (defined as use of a condom for all acts of penetrative vaginal intercourse) associated with an 80 percent reduction in heterosexual transmission of HIV.160 Another pooled analysis found consistent condom users were 10 to 20 times less likely to become infected when exposed to the virus than inconsistent or nonusers.161

The 2007 USPSTF update1 also included a study that used a mathematical formula to estimate that transmission risk was 3.5 times higher in HIV-positive patients unaware of their status (6.9%) compared with those aware (2.0%) of their HIV infection, resulting in a projected 31 percent decline in new sexual infections per year (from 32,000 to 22,150) if all HIV-positive patients unaware of their status became aware.51 However, these results were based on estimates for reduced risky behaviors from studies with methodological shortcomings, and may not have adequately accounted for other important factors that might affect transmission risk (such as type of risky behaviors, number of risky behavior episodes, number of sexual partners, viral load, use of ART, presence of other STDs, CD4 count, and time since diagnosis).162

Two observational studies published since the prior USPSTF review reported results consistent with previous findings93, 114 (Appendixes B29 and B30). One prospective cohort study of 476 heterosexual Spanish individuals (1,355 couple-years of followup) found self-reported condom use associated with decreased risk of HIV transmission per act of intercourse compared with intercourse without a condom (unadjusted RR, 0.07 [95% CI, 0.01 to 0.58]).93 A Chinese study of 1,927 serodiscordant couples found not always using condoms associated with increased risk of seroconversion (RR, 8.4 [95% CI, 4.8 to 15]) in multivariate analysis when compared with always using a condom, after adjusting for frequency of sexual intercourse, switching of ART regimen, and physical and psychological quality-of-life scores.114

No study evaluated effects of safer injection drug use behaviors by HIV-positive patients on risk of HIV transmission.

Contextual Question. What Is the Cost-Effectiveness of Universal Versus Targeted HIV Screening in Low- or Average-Prevalence Populations?

The 2005 USPSTF review included two good-quality studies8, 9 that estimated cost-effectiveness of HIV screening in low- or average-prevalence populations. One study by Sanders et al estimated <$50,000 (2004 U.S. dollars) per quality-adjusted life-year (QALY) for one-time screening versus no screening at an HIV prevalence of 0.5 percent, excluding potential transmission benefits.9 After incorporating potential transmission benefits, cost-effectiveness remained <$50,000 per QALY at an HIV prevalence of 0.05 percent, or substantially lower than seen in the general population. Another study by Paltiel et al, which did not directly incorporate secondary transmission benefits, estimated incremental cost-effectiveness of one-time screening in the general population (prevalence of undiagnosed HIV infection, 0.1%; corresponding to an overall HIV prevalence of about 0.4%) of $113,000 (2001 U.S. dollars) per QALY compared with no screening.8 Neither study evaluated the incremental cost-effectiveness of a strategy of universal versus targeted screening in low-prevalence populations,163 though one of the studies included assumptions about background testing rates in the no screening arm.8 Long-term cardiovascular harms were not accounted for in either model. In the study that included secondary transmission benefits, cost-effectiveness in low-prevalence settings was sensitive to estimates of beneficial effects of screening on transmission.9 The other cost-effectiveness analysis did not directly incorporate secondary transmission benefits when estimating cost-effectiveness,8 though a subsequent analysis found that increasing rates of test notification and entry into care had a greater impact on cost-effectiveness than similar increases in rates of testing.164

The cost-effectiveness analyses included in the prior USPSTF review also evaluated screening strategies involving repeat testing.8, 9 They found screening every 5 years in a population with 1 percent prevalence associated with a cost-effectiveness ratio <$50,000 per QALY when secondary transmission benefits were included and annual incidence was at least 0.09 percent. In low-prevalence (0.1% undiagnosed HIV infection) settings, one of these studies found that repeat screening at any interval cost >$100,000 per QALY at all plausible incidences.8 This study also found that in a high-risk setting (incidence, 1.2%; prevalence, 3.0%), screening every 5 years cost $50,000 per QALY compared with one-time screening, screening every 3 years cost $63,000 per QALY compared with screening every 5 years, and screening annually cost $100,000 per QALY compared with screening every 3 years.

Subsequent cost-effectiveness analyses based on the models used in the above studies have been published.165-167 Paltiel et al estimated cost-effectiveness ratios of <$50,000 (2004 U.S. dollars) per QALY for one-time rapid screening compared with no screening in settings with HIV prevalence as low as 0.20 percent, when assuming moderately favorable effects of ART on transmission (decrease in the basic reproductive number [R0], 1.44 to 1.27).167 Cost-effectiveness ratios remained <$50,000 per QALY for screening every 5 years compared with no screening at prevalences as low as 0.45 percent and annual incidences as low as 0.0075 percent. Sanders et al estimated cost-effectiveness ratios of <$60,000 (2007 U.S. dollars) per QALY for one-time screening with streamlined counseling compared with no screening in persons ages 55 to 75 years with a sexual partner at risk at an HIV prevalence as low as 0.1 percent, assuming favorable effects on transmission.165 Cost-effectiveness ratios were also <$60,000 per QALY for one-time screening with streamlined counseling compared with no screening in persons ages 55 to 65 years without a sexual partner at risk at an HIV prevalence of 0.5 percent. With traditional counseling, cost-effectiveness ratios of screening compared with no screening were >$100,000 per QALY for screening persons ages 75 years or older with a sexual partner at risk or persons ages 65 years or older without a sexual partner at risk. In a separate study, Sanders et al estimated a cost-effectiveness ratio of $10,660 per QALY for nurse-initiated routine screening with rapid HIV testing and streamlined counseling compared with traditional HIV counseling and testing.166

One other study published since the 2005 USPSTF review estimated a cost-effectiveness ratio of $22,382 (2009 U.S. dollars) per QALY for one-time screening of low-risk persons (HIV prevalence, 0.10% in men and 0.22% in women) plus annual screening of high-risk persons compared with current practice (annual rate of screening, 23% in high-risk persons and 10% in low-risk persons), assuming a 20 percent reduction in sexual activity after screening, with an associated reduction in risk of HIV transmission.168 Assuming the same screening strategy plus an increase in ART utilization in 75 percent of infected persons resulted in a similar cost-effectiveness ratio, though more infections would be prevented. Screening low-risk persons every 3 years or more frequently was associated with cost-effectiveness ratios of >$100,000 per QALY compared with one-time screening of low-risk persons, with annual screening of high-risk persons included as part of both strategies.

No study directly compared cost-effectiveness of universal versus targeted screening in low-prevalence populations, or explicitly included potential long-term cardiovascular harms of combination ART in models.

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.8M)

In this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...