|Home | About | Journals | Submit | Contact Us | Français|
Gillian D. Sanders PhD, Duke Clinical Research Institute, PO Box 17969, Duke University, Durham, NC. 27715.
Ahmed M Bayoumi MD, MSc, Centre for Research on Inner City Health, St. Michael's Hospital, 30 Bond St., Toronto ON, Canada M5B 1W8
Mark Holodniy MD, VA Palo Alto Health Care System, 3801 Miranda Avenue (132), Palo Alto, CA 94304
Douglas K. Owens MD, MS, Center for Primary Care and Outcomes Research/Center for Health Policy, Stanford University, 117 Encina Commons, Stanford, CA 94305-6019.
Although HIV infection is more prevalent in people <45, a significant number of infections occur in older individuals. Recent guidelines recommend HIV screening of patients aged 13–64. The cost-effectiveness of HIV screening patients aged 55–75 years is uncertain.
To examine the costs and benefits of HIV screening in patients aged 55–75 years.
Derived from the literature.
Patients aged 55–75 with unknown HIV status.
HIV screening program for patients aged 55–75 years compared to current practice.
Life-years, quality-adjusted life-years (QALYs), costs, incremental cost-effectiveness.
For a 65-year-old patient, HIV screening using traditional counseling costs $55,440/QALY compared with current practice when the prevalence of HIV was 0.5% and they did not have a sexual partner-at-risk. In sexually-active patients, the incremental cost-effectiveness ratio was $30,020/QALY. At a prevalence of 0.1%, HIV screening cost <$60,000/QALY for patients <75 years old with a partner-at-risk if less costly streamlined counseling is used.
Cost-effectiveness of HIV screening depended on HIV prevalence, age of the patient, counseling costs, and whether the patient was sexually active. Sensitivity analyses with other variables did not change our results substantially.
Uncertainty regarding effects of age on HAART toxicity, efficacy, and AIDS mortality. Sensitivity analyses exploring these variables did not qualitatively affect our results.
If the tested population has a prevalence of ≥0.1%, HIV screening in individuals between 55–75 years old reaches conventional levels of cost-effectiveness when counseling is streamlined and if the screened patient has a partner at risk. HIV screening of patients with advanced age is economically attractive in many circumstances.
The Centers for Disease Control and Prevention’s (CDC) recently revised human immunodeficiency virus (HIV) screening guidelines recommend that all patients age 13 to 64 in health care settings be tested (1). Previous work by our group (2) and others (3–5) has demonstrated that screening for HIV in younger patients is cost effective even in relatively low-prevalence settings. For example, we found screening 40-year old individuals to cost less than $50,000 per quality-adjusted life year (QALY) when the prevalence of undiagnosed HIV infection is 0.05% (1 in 2000) or higher (2). Although screening is cost effective in younger age groups, approximately 19% of all people with AIDS in the U.S. were aged 50 or older at the time of their diagnosis (6). The cost effectiveness of screening older patients has not been evaluated and is uncertain.
There are several reasons that screening in patients over 55 years of age may be less cost effective than screening younger individuals. First, the prevalence of HIV, a major determinant of cost effectiveness, is lower in older compared to younger age groups (7). Second, treatment of HIV will be less cost effective in older individuals because the competing risk of death from other causes will diminish the absolute benefits of therapy. Third, if older patients have fewer sex partners, screening will be less economically attractive because reduction of HIV transmission is an important benefit of screening and hence, a determinant of cost effectiveness (2). Finally, neither the effectiveness nor toxicity of highly active antiretroviral therapy (HAART) in patients aged 55–75 years old have been well studied; both could be less favorable than in younger patients (8). To assess the importance of these factors, we modified our previous cost-effectiveness model to assess the cost effectiveness of HIV screening in individuals over 55 years of age.
We estimated the health and economic effects of a program of voluntary HIV screening in patients between the ages of 55 and 75 using a decision model, following the recommendations of the Panel on Cost Effectiveness in Health and Medicine. We performed the analysis using a societal perspective on health benefits and costs and applying a 3% annual discount rate (9).
We adapted a Markov model (10, 11) developed to assess the cost effectiveness of voluntary HIV screening in healthcare settings (2) using Decision Maker software (version 2007.0.1 Pratt Medical Group). The model tracked a cohort of older patients over their lifetime. Patients were either screened through a one-time voluntary HIV screening program or were followed without routine screening. Each month, the model assessed the patients’ HIV status, whether their HIV status was identified, the clinical course of HIV disease, the costs and consequences of HIV transmission, and the costs and consequences of HAART for patients identified with HIV and eligible for treatment (Figure 1). Individuals were also at risk for age- and gender-specific non-HIV related mortality. Probability estimates were based on high-quality published literature and expert clinical judgment (Table 1) (2, 9, 12–156).
We assessed the cost-effectiveness of screening patients between 55 and 75 years of age whose HIV status was unknown in populations with a prevalence of unidentified HIV infection that varied from 0.1% to 1% (157). We estimated the incidence of HIV in age- and sex-specific strata based on previous models (18, 19)
Patients who were not tested through an HIV screening program or who tested negative through screening could be selected for future testing through symptom-based case finding. We assumed that such case finding occurred only in patients with a CD4 count of 0.350 109 cells/L or lower and that the probability of identifying and testing for HIV increased as the CD4 count fell. We incorporated standard HIV testing using a serum enzyme-linked immunosorbent assay followed by confirmatory Western blotting.
In our base-case analysis, patients received comprehensive pre- and post-test counseling with HIV testing. We explored the effects of streamlined counseling (abbreviated pre-test counseling requirements) as recommended by the CDC (1) on HIV screening in sensitivity analyses. The benefits of testing and counseling accrued only to those patients who received their test results and entered care.
We modeled HIV disease progression using HIV viral load copy number and CD4 count and drawing on natural-history data (Table 1) (52, 77, 84). We assumed treatment of HIV infection would adhere to published treatment guidelines, including appropriate prophylaxis and immunization (41, 42). HAART was assumed to be initiated when an identified HIV-infected patient’s CD4 count fell to or below 0.35 109 cells/L. Once HAART was initiated, we modeled the possible rise in CD4 count, virologic suppression, treatment-related toxicity, and virologic rebound. We assumed that patients with drug-related adverse effects switched to a new regimen and had the same probability of achieving suppression on this new regimen as they did on the initial regimen. In contrast, we assumed that patients who experienced virologic failure and were identified through viral load testing were switched to a new regimen with a reduced likelihood of achieving virologic suppression (Table1). We assumed that after being treated with three successive antiretroviral regimens, a patient would only be able to achieve partial virologic suppression due to multi-drug resistance. We assessed varying assumptions about the efficacy of treatment of such patients in sensitivity analysis.
Substantial uncertainty exists regarding the effectiveness of HAART and the course of HIV infection in the older aged population. Although patients over 55 may have higher rates of virologic suppression (perhaps due to better adherence or altered pharmacodynamics), the evidence regarding immunologic effects in patients over 55 is inconsistent. While several studies have suggested older patients have a longer time to CD4 reconstitution after starting HAART (89–91, 94, 98), other studies have found no differences, or differences in the first year only (8, 92, 93, 95–97). Studies have similarly differed as to whether older age is an independent risk factor for disease progression (90, 158). The evidence regarding toxicity of antiretrovirals in older patients is even more limited. Although there are many reasons to suspect that older patients will tolerate antiretrovirals less well than younger patients (such as cardiovascular and hepatic events), the evidence for this assertion is still weak (8, 91, 94, 96).
We approached the modeling of HAART effectiveness and toxicity in patients over 55 recognizing that the evidence for age effects are generally based on observational studies that are often small and with considerable potential for residual confounding . To address the uncertainty around these age effects and explore their potential impact, our base-case model assumed no difference in antiretroviral efficacy for older patients, but we included specific scenarios in which we modeled three effects that may be related to older age: greater failure rates with HAART, greater treatment-limiting toxicities, and greater non-AIDS age-specific mortality.
We included in our analysis the potential costs and health benefits associated with transmission of HIV from an infected individual to their sexual partners. Transmission depended on the infected patient’s gender, type of sexual activity, number of sexual partners, knowledge of HIV status, and viral load (Table 1). We assumed that a sexually active patient aged 55–75 years old would have on average 1 partner at risk for infection and also tested the situation in which patients had no sexual partners or multiple partners (see Technical Appendix).
The model incorporated adjustments for the quality of life associated with age-specific current health, HIV disease (asymptomatic HIV, symptomatic HIV, and AIDS), and HAART (Table 1). Utilities measure the patient’s quality-of-life and were rated on a scale from 0 to 1, where 0 represents death and 1 ideal health. We multiplied utilities based on HIV-related health status and knowledge by age-specific utility weights derived from the Beaver Dam Health Outcomes Study (151).
Our analysis included the direct costs of medical care associated with HIV testing and counseling, follow-up, and treatment of patients identified with HIV infection through a screening program or through symptom-based case finding (Table 1). We included age-specific medical expenditures unrelated to HIV care (http://www.bls.gov/cex/csxann05.pdf) and updated all costs to 2007 dollars using the gross domestic product (GDP) deflator (159). While the GDP deflator is considered a more robust estimate of inflation than the consumer price index by some economists, it generally yields similar or slightly more conservative estimates of current costs.
We performed one-way and multi-way sensitivity analyses to account for important model assumptions and uncertainties. For clinical variables, our ranges for sensitivity analyses represent our judgment of the variation likely to be encountered in clinical practice on the basis of both the literature and discussion with experts.
To estimate the uncertainty in our findings we performed a probabilistic sensitivity analysis. We assigned distributions to eighteen variables which described the natural history of HIV and effectiveness of HAART in older patients (see Technical Appendix). We selected distributions appropriate to the type of variable (for example, we modeled probabilities using a beta distribution) and selected corresponding parameters using the method of moments. For variables which represented alternative scenarios rather than uncertain estimates, we sampled from discrete distributions reflecting our belief in the likelihood of each scenario being true. We performed a second-order Monte Carlo simulation with 1,000 iterations over the sampled distributions.
Calibration and validation of the decision model and its outputs occurred as an iterative process throughout its development. To assess the output of the model, we compared our predicted time from infection until death for untreated patients (10.3 to 13.4 years) to published estimates (12.1 years) (103). To assess our long-term estimates of treatment efficacy, we compared the estimated proportion of patients with multi-drug resistant virus at 10 years (7.2%) to published estimates (9.2%, 95% CI 5.0% to 13.4%) (66). Finally, we compared our model's estimate of life expectancy of identified HIV infected patients on HAART (varying between 14.9 to 18.5 years) to other modeled estimates by King et al (160) and Paltiel et al (4).
This research is supported in part by the Department of Veterans Affairs, the National Institute on Drug Abuse (R01 DA15612-01), the National Institute of Aging through the Stanford Center on the Demography and Economics of Health and Aging (CDEHA) grant P30-AG017253 and support of the Ontario HIV Treatment Network. The funding sources had no role in the design or implementation of the study or in the decision to seek publication. Drs. Sanders, Bayoumi, Holodniy, and Owens have no potential conflicts of interest for this work.
We estimated the increase in survival that resulted from early identification of asymptomatic HIV and initiation of HAART through a screening program compared with identification on the basis of case finding. The benefit varied depending on age and decreased for older patients. For a 65-year old HIV-infected patient, early identification and treatment increased life expectancy by 0.58 years or 0.41 QALYs. For a 75-year old HIV-infected patient, the benefit was 0.30 years or 0.20 QALYs (Figure 2). In the Technical Appendix we demonstrate how such increases in life expectancy from a one-time HIV screening program in the current US population aged 55–64 could save over 120,000 life years attributed to almost 170,000 individuals. In the 65–74 year cohort, an additional 40,000 life years would be gained by almost 92,000 individuals.
The costs and benefits of voluntary one-time HIV screening for patients aged 55–75 years old depended on the prevalence of unidentified HIV, the age of the patient, and whether the patient has a sexual partner at risk of infection.
If a patient has an uninfected partner at risk for infection and the prevalence of unidentified HIV was 0.5%, a voluntary one-time screening program improved life expectancy in 65-year old patients by 1.56 days or 1.31 quality-adjusted days at a cost of $107 relative to current practice (incremental cost-effectiveness of $30,020/QALY) (Figure 3a). For a 75-year old patient, the incremental cost was $90, while incremental life expectancy was 0.95 days or 0.79 quality-adjusted days (incremental cost-effectiveness of $41,520/QALY). At a lower HIV prevalence of 0.1%, the incremental cost-effectiveness of screening a 65-year old with a sexual partner at risk was $91,410/QALY. At this prevalence, HIV screening exceeds $100,000/QALY for patients 70 and older (Table 2 and Figure 3a) if traditional pre-test (estimated cost of $35) and post-test counseling (estimated cost of $10 for negative tests and $35 for positive tests) is used.
If a patient does not have a partner at risk for infection, HIV screening costs less than $100,000 per quality-adjusted life year for all patients (≤ 80 years) if the prevalence of unidentified HIV is greater than 0.5% (Figure 3b). For a 65-year-old patient, HIV screening using traditional counseling costs $55,440/QALY compared with current practice when the prevalence of HIV was 0.5%. At an HIV prevalence of 0.1% however, screening a patient (≥ 50 years) without a sexual partner increases costs more than $100,000/QALY (Table 2 and Figure 3b) with traditional counseling.
Current CDC guidelines for HIV screening recommend streamlined counseling, which reduces the time and costs associated with pre-test counseling and, for those who test negative, post-test counseling (1). Based on a recent time-costing analysis by the Veterans Affairs (VA) Quality Enhancement Research Initiative for HIV and Hepatitis (QUERI-HIV/HEP), in this analysis we assumed the pre-test counseling costs would be reduced, relative to comprehensive counseling, from $35 to $5, post-test counseling of patients who tested negative would be reduced from $10 to $2, and post-test counseling of patients who tested positive would remain unchanged ($35). We further assumed streamlined counseling did not affect either adherence with screening or modification of risk behaviors. Under these assumptions, the incremental cost effectiveness of HIV screening was more economically attractive at all ages and prevalences of HIV (Figure 4). For patients with a sexual partner at risk, HIV screening incorporating streamlined counseling would cost less than $60,000/QALY compared with current practice for patients up to 75 years of age, even if the prevalence of unidentified HIV was as low as 0.1% (Figure 4). For patients without a partner at risk, HIV screening incorporating streamlined counseling costs less than $100,000/QALY for patients from populations where the HIV prevalence is 0.1% if patients were 75 or younger (Figure 4), and cost less than $50,000/QALY at a prevalence of 0.5%.
We evaluated scenarios in which HAART was less effective or more toxic in older patients. After assuming that the failure rate of HAART was increased by 20% in patients over 55, HIV screening with traditional counseling cost less than $60,000 per QALY for patients of all ages if they had a sexual partner at risk and the unidentified HIV prevalence in the population was greater than 0.5%. If the unidentified prevalence of HIV was 0.1%, HIV screening cost less than $65,000/QALY for patients aged 75 years and younger with streamlined counseling.
Assuming that HAART-related toxicity requiring treatment change increased by 25% in patients over 55 had minimal effects on the analysis and did not qualitatively change our results. It is also uncertain whether the non-AIDS age-specific mortality of HIV-positive patients is increased compared with uninfected patients. We therefore evaluated the effect of HIV screening if we assumed that HIV-positive patients had an age-specific mortality 25% greater than that of an uninfected cohort. Again, these assumptions did not qualitatively affect the results of our analysis. Finally, when we varied all three of these assumptions simultaneously for patients with a partner at risk, HIV screening with traditional counseling cost less than $65,000 per QALY if the unidentified prevalence of HIV was 0.5% or greater. If the unidentified prevalence of HIV was 0.1%, HIV screening cost less than $70,000/QALY for patients 75 years and younger with streamlined counseling.
We performed a probabilistic sensitivity analysis in which we simultaneously varied the values of variables defining the natural history of HIV and effectiveness of HAART in older patients. Assuming traditional counseling costs, for 65–74 year old patients with a sexual partner at risk from a population with the prevalence of unidentified HIV which ranged from 0.2 to 1.2%, the probabilistic sensitivity analysis estimated a median and mean incremental cost-effectiveness ratio of $36,980/QALY and $44,530/QALY respectively. A cost-effectiveness acceptability curve (Technical Appendix) demonstrated that, at a willingness-to-pay threshold of $50,000/QALY, the probability that HIV screening in this population was cost-effective was 75%. The probability that HIV screening with traditional counseling costs was cost-effective with a willingness-to-pay threshold of $100,000/QALY was 96%.
With streamlined counseling, the probabilistic sensitivity analysis predicted a median and mean incremental cost-effectiveness ratio of $22,460/QALY and $24,420/QALY respectively. The cost-effectiveness acceptability curve (Technical Appendix) with streamlined counseling demonstrates that, at a willingness-to-pay threshold of $50,000/QALY, the probability that HIV screening in this population with streamlined counseling is cost effective was 97%.
We found that the cost effectiveness of screening in patients between 55–75 years compares favorably to that of other interventions that are accepted as good uses of resources, particularly if providers implement screening with streamlined counseling and if the person being screened has a sexual partner at risk. Under these circumstances, screening cost less than $60,000 per QALY gained, even with an unidentified prevalence as low as 0.1%. Screening at age 64 years, as recommended by the CDC, cost about $41,000 per QALY gained with streamlined counseling and a partner at risk. Screening is more expensive in patients who do not have a partner at risk, or if traditional counseling is used.
Although evidence about the prevalence of HIV in the older age groups is sparse, the limited available data suggest that the prevalence is sufficiently high for screening to be cost effective. In a blinded serologic survey we performed of 8627 inpatients and outpatients at six Department of Veterans Affairs Health Care Systems, we found the prevalence of undocumented HIV infection was 0.7% (95% confidence interval [CI] 0.2% to 1.7%) in outpatients aged 55 to 64 years, 0.5% (95% CI, 0.2% to 1.2%) in outpatients aged 65 to 74, and 0.1% (0.0% to 0.06%) in outpatients aged 75 and older (157). Because the VA population differs from that in other health care settings, evidence about the prevalence of undocumented HIV in the over 55 age groups from other populations would be useful to help guide screening decisions. We note however, that among outpatients aged 65 to 74, the prevalence in other settings could be only one-fifth as high as we found and screening with streamlined counseling would still be cost effective for patients who are sexually active.
As noted, an important determinant of the cost effectiveness of screening is whether the screened person has a sexual partner or partners at risk. A recent study used a probability sample of 3005 U.S. adults and found that 73% of people who were 57 to 64 years of age were sexually active, as were 53% of people 65 to 74 years of age, and 26% of people 75 to 85 years of age (161). The National Health and Social Science Survey in 1992 studied 3,492 members of the US general population and found that 84% of individuals between age 50–54 and 69% of those aged 55–59 had at least one sexual partner in the past year. The National AIDS Behavioral Surveys administered in 1991–1992 estimated that the prevalence of having at least one risk factor for HIV infection was approximately 10% among individuals aged 50 years or older. In addition, a small percentage of those individuals with a known risk for HIV infection used condoms during sex, or had undergone testing and were much less likely to have adopted these prevention strategies than younger individuals who engaged in the same behavioral risks (162). These studies suggest that a significant number of people over age 50 have risk factors for HIV and that majority of individuals up to age 75 have a partner at risk for infection.
Based on the results of our cost-effectiveness analyses, the data available on prevalence and the relatively high rates of sexual activity in people over age 55, we recommend one-time voluntary HIV screening with streamlined counseling on a routine basis for all persons aged 55–64 and one-time screening on a targeted basis to sexually active persons aged 65–74, if the HIV prevalence is > 0.1%. For people aged 65 to 74 who do not have a partner at risk, screening costs between $50,000/QALY and $100,000/QALY gained with prevalence between 0.1% and 0.5%. Thus, screening is more expensive if the person is not sexually active, but is still a reasonable option, particularly if prevalence approaches 0.5%.
One approach that providers can use to estimate the prevalence of HIV is to begin a screening program, and assess the number of positive tests in the screened population. If approximately 4000 individuals are screened without a positive test, a prevalence of 0.1% can be excluded with 95% confidence, and the population would fall outside the prevalence threshold (0.1%) recommended for screening by the CDC (1).
The cost effectiveness of screening remained favorable if HAART were modestly less effective, or led to modestly higher rates of adverse events than in younger patients. An increase in age-specific mortality of 25% in infected patients did not substantively change the cost effectiveness of screening. This finding would not hold for patients whose life expectancy was substantially shortened by diseases with high mortality such as cancer or congestive heart failure. In general, however, our findings support the usefulness of screening older individuals who do not have high-mortality comorbidities if the prevalence of HIV is above 0.1% to 0.5%. Our analysis included three suppressive antiviral regimens before the start of nonsuppressive therapy. With the approval of new antiretrovirals, a fourth suppressive regimen may be feasible, thus we may have underestimated the benefit from antiretroviral therapy in the elderly.
In conclusion, we found that routine HIV screening is cost effective in the age range (up to age 64) and prevalence (greater than 0.1%) recommended by the CDC. If the screened population has an unidentified prevalence of 0.1% or greater, HIV screening in older individuals aged 65 to 75 can also reach conventional levels of cost effectiveness if screening can be done inexpensively, such as by using streamlined counseling. Our analyses suggest that screening decisions in patients over 64 years of age should consider whether there are partners at risk, the expense of screening, and whether there are life-threatening comorbid conditions. Advanced age alone should not preclude screening for HIV. Rather, for many people in this age group, the cost effectiveness of screening falls within the range of other accepted interventions.
Drs. Sanders, Bayoumi, Holodniy, and Owens each made substantial contributions to the conception and design, acquisition of data, and analysis and interpretation of data; Dr. Sanders drafted the article and Drs. Bayoumi, Holodniy, and Owens each revised it critically for important intellectual content; all authors provided final approval of the version to be published. Dr. Sanders affirms that everyone who has contributed significantly to this work has been listed.
GRANT SUPPORT: This research is supported in part by Department of Veterans Affairs, the National Institute on Drug Abuse (R01 DA15612-01), and funding from the National Institute of Aging through the Stanford Center on the Demography and Economics of Health and Aging (CDEHA) grant P30-AG017253. Dr. Bayoumi was supported by a career scientist award from the Ontario HIV Treatment Network. The authors gratefully acknowledge the support of the Ontario Ministry of Health and Long-Term Care.
To estimate the public health benefit of HIV screening within the current U.S. population, we used the results of a blinded serologic survey of HIV within the VA population (157), the effect of HIV screening and early identification on life expectancy (shown graphically in Figure 2), and the current population of the United States. These data were used to estimate the life expectancy gain in the current US population from early HIV identification through a one-time voluntary HIV screening program. As the table below demonstrates, under these assumptions, a one-time HIV screening program in the current US population aged 55–64 would save over 120,000 life years attributed to almost 170,000 individuals. In the 65–74 year cohort, an additional 40,000 life years would be gained by almost 92,000 individuals.
|Variable||Age Cohort, years|
|U.S. population, N*||24,274,684||18,390,986|
|Prevalence of unidentified HIV, % (95% confidence interval)†||0.7|
|Estimated number of HIV-infected individuals unaware of their HIV Status, N (range)||169,923|
|Estimated life expectancy gain per individual from early HIV identification, years||0.7227||0.4391|
|Estimated life expectancy gain in current US population from early HIV identification, years (range)||122,805|
In our main analyses we focused on the incremental cost-effectiveness of a one-time voluntary HIV screening program as compared with current practice. In sensitivity analyses however, we wanted to explore the incremental costs and benefits of a repeated HIV screening program – especially in populations where the incidence of HIV may be increased. The figure below demonstrates the effect of screening frequency on the incremental cost-effectiveness of screening at various HIV incidence rates assuming patients are 55-years old and the prevalence of HIV is 1% in the cohort. The solid line marked with diamonds represents the baseline incidence, the solid line marked with circles represents the cost-effectiveness of recurrent screening when the incidence of HIV is twice the baseline rate, and the dashed line represents the cost-effectiveness of recurrent screening when the incidence of HIV is three times the baseline rate. The incremental cost-effectiveness ratio compares screening every A years with screening every B years, where B refers to the screening frequency directly to the left of A on the x-axis (i.e. comparing screening every 3 years with screening every 5 years). Recurrent screening becomes more economically favorable as the incidence increased – yet one time screening appears to be the best use of limited healthcare resources in these older cohorts. Similar findings were found in older cohorts and at varying prevalences.
To explore the effect of additional regimens or increased efficacy of non-suppressive therapy, we varied the decrement in viral load which patients received while on non-suppressive therapy. Under our base-case assumptions (1.0 log10 copies/mL decrement with non-suppressive therapy) screening a 55-year old individual with a sexual partner-at-risk for HIV cost $71,060/QALY compared with current practice if the prevalence of unidentified HIV was 0.1%. Under these same assumptions, screening a 75-year old individual for HIV cost $143,060/QALY compared with current practice. If the decrement in viral load realized through non-suppressive therapy was increased to 2 log10 copies/mL, the incremental cost-effectiveness of HIV screening became more favorable at $56,250/QALY and $118,480/QALY in the 55- and 75-year old individuals with a partner-at-risk. Similarly, if the decrement in viral load was increased to 4 log10 copies/mL, the incremental cost-effectiveness ratios of HIV screening was further reduced to $43,780/QALY and $110,980/QALY in the 55- and 75-year old cohorts.
Because the cost and feasibility of streamlined counseling in elderly patients who have competing comorbid conditions and may not have been tested previously is uncertain, we explored the effect of a more expensive streamlined counseling scenario in sensitivity analyses. In this analysis we assumed that a pre-test counseling session cost $15 dollars (base-case value = $5 in our streamlined analysis), and that post-test counseling for a negative test result cost $5 (base-case value = $2 in our original streamlined analysis). In our original analysis, if the prevalence of unidentified HIV was 0.1%, then screening patients aged 75 who had a partner-at-risk cost $83,680/QALY compared with current practice. Under our assumptions of a more costly streamlined counseling scenario, this incremental cost-effectiveness ratio increased marginally to $88,120/QALY. Although the incremental cost-effectiveness ratio increased for all age and prevalence combinations, these increases did not change our qualitative results or our clinical recommendations.
We also explored a scenario where the efficacy of streamlined counseling was reduced. It is important to note that under streamlined counseling, patients who test positive for HIV infection receive the same counseling (both in terms of costs and efficacy) as those who receive traditional counseling and therefore the true reduction of streamlined counseling effects on behavior modification is small. If we assumed that streamlined counseling reduced the efficacy of counseling from our base-case value of 20% reduction in transmission to a 10% reduction, then the incremental cost-effectiveness of HIV screening increases to $19,800/QALY, $23,090/QALY, and $45,550/QALY in a cohort of 65-year olds with prevalence of unidentified HIV of 1%, 0.5%, and 0.1% respectively. If we assumed that streamlined counseling completely eradicated any benefit from counseling on transmission then the incremental cost-effectiveness of screening increased although the effects again are minimal ($23,050/QALY, $26,700/QALY, and $50,730/QALY in 65-year old cohort with 1%, 0.5%, and 0.1% prevalence of HIV respectively).
Finally, in our exploration of the effects of streamlined counseling, we assume that individuals who are sexually active have one partner-at-risk for transmitting their infection. In a sensitivity analysis, we varied this assumption to explore the effects on our results if individuals who were sexually active had 2 partners at risk. As expected, if individuals had a greater number of sexual partners-at-risk, this increased the benefits of HIV screening and early identification. As an example, if the prevalence of unidentified HIV in a population was 0.05%, HIV screening individuals aged 75 years who had one partner-at-risk would cost $21,560/QALY if streamlined counseling was used. If however these individuals had two partners-at-risk, such HIV screening would become more favorable with an incremental cost-effectiveness ratio of $10,100/QALY compared with current practice.
To estimate the uncertainty in our findings we performed a probabilistic sensitivity analysis. We assigned distributions to twenty variables which described the natural history of HIV and effectiveness of HAART in older patients. For 15 variables, we used parametric distributions as described below:
|Prevalence of HIV, %*||0.5||0.27||BETA|
|Decrease in viral load during virologic rebound||0.5||0.5||GAMMA|
|Decrease in viral load with nonsuppressive therapy||1||0.5||GAMMA|
|Increase in viral load after failed therapy (with AIDS)||1||0.5||GAMMA|
|Increase in viral load after failed therapy||0.8||0.4||GAMMA|
|Viral load during virologic suppression||1.3||0.6||GAMMA|
|Probability of not achieving virologic suppression during 1st regimen||0.20||0.20||BETA|
|Probability of not achieving virologic suppression during 2nd regimen||0.35||0.21||BETA|
|Probability of not achieving virologic suppression during 3rd regimen||0.7||0.15||BETA|
|Probability of intolerance to HAART requiring discontinuation||0.25||0.13||BETA|
|Relative risk for discontinuation of HAART during 2nd regimen||1.4||0.84||LOG NORMAL|
|Relative hazard of AIDS per increase in CD4 count||0.0154||0.12||LOG NORMAL|
|Relative hazard of AIDS per increase in viral load||0.43||1.25||LOG NORMAL|
|Relative hazard of death from AIDS per increase in CD4 count||0.118||1.43||LOG NORMAL|
|Relative hazard of death from AIDS per increase in viral load||0.64||0.92||LOG NORMAL|
For 5 variables, we sampled from discrete values, as defined below:
|Variable||Discrete Values Sampled*|
|Age, years||65, 66, 67, 68, 69, 70, 71, 72, 73, 74†|
|Multiplier on age-specific mortality for HIV infected patients over 55||1.0, 1.5|
|Relative risk of achieving suppression in patients over 55||0.5, 1.0|
|Relative risk of HAART discontinuation in patients over 55||1.0, 1.5|
|Relative risk for discontinuation of HAART during 3rd regimen||1,2,3,4‡|
We evaluated the incremental cost-effectiveness of HIV screening a 65–74 year old patient with a sexual partner in a population where the unidentified prevalence of HIV ranged from 0.2–1.2 %.
The following two figures show the results of our Monte Carlo simulation under assumptions of either traditional or streamlined counseling as an acceptability curve. Cost effectiveness acceptability curves allow decision makers to determine the probability that HIV screening in the given population is cost-effective at various willingness-to-pay thresholds (the highest incremental cost-effectiveness ratio that people would be willing to accept as reasonable value for their health care dollar).
The views expressed in this publication are the views of the authors and do not necessarily reflect the views of the Ontario Ministry of Health and Long-Term Care.
Drs. Sanders, Bayoumi, Holodniy, and Owens have no potential conflicts of interest for this work.