Background: Structural interventions are theoretically promising for populations with a low prevalence of HIV, because they can reach large numbers of people to influence their social norms and collective risky behaviors for a relatively low cost per person. Because HIV transmission is continuing to increase among women in the southern United States, interventions to stem this epidemic are particularly warranted. This study explores whether structural interventions may be a cost-effective way to prevent HIV in this population.
Methods: We used the cost-effectiveness estimator, “Maximizing the Benefit” to determine the relative cost-effectiveness of 6 structural HIV prevention interventions. “Maximizing the Benefit” is a spreadsheet tool using mathematical models to estimate the cost per HIV infection prevented taking into account the epidemiologic contexts, behavioral change as a result of an intervention, and the costs of intervention. We applied estimates of HIV prevalence related to blacks in the southern United States.
Results: All the structural interventions were cost-effective compared with average lifetime treatment costs of HIV, but mass media, condom availability, and alcohol taxes theoretically prevented the largest numbers of HIV infections.
Conclusions: Although the assumptions used in cost-effectiveness estimates have many limitations, they do allow for a relative comparison of different interventions and help to inform policy decisions related to the allocation of HIV prevention resources. Structural interventions hold the greatest promise in reducing HIV transmission among low-prevalence populations.
Although the assumptions used in cost-effectiveness estimates have many limitations, they do allow for a relative comparison of different interventions and help to inform policy decisions related to the allocation of HIV prevention resources.
From *RAND Corporation, Santa Monica, California; and †Tulane School of Public Health and Tropical Medicine, New Orleans, Louisiana
Correspondence: Deborah Cohen, MD, MPH, RAND Corporation, 1700 Main Street, Santa Monica, CA 90405. E-mail: firstname.lastname@example.org
Received for publication August 10, 2004, and accepted December 13, 2004.
IN THE ABSENCE OF A VACCINE, the goal of HIV prevention is to reduce those conditions and high-risk behaviors that expose individuals to the virus. Behavioral interventions are particularly emphasized by HIV prevention programs. Many fields of study, including economics, sociology, anthropology, and psychology, have tried to understand human behavior and how it is influenced, but theoretical models from the field of psychology have dominated interventions to prevent HIV transmission. Within psychology are at least 2 very different schools of therapy: cognitive and behaviorist. Although there are a variety of different methods used by cognitive psychology, they all fundamentally assume that people are rational and will make decisions in their own interest as they perceive it and that these decisions will be healthful if they have the right information, the appropriate attitude, and the necessary skills for action. The methods of cognitive therapy rely on education, counseling, and role modeling and are thought to have maximum impact if they are “client-centered.” 1 In contrast, behavioral psychology views people as a product of their environments, which form behavior by positive or aversive reinforcers. According to Skinner,2 the behaviorist first seeks to change the world that people inhabit rather than “their minds and hearts.” Structural factors comprise the world within which we live and, therefore, structural interventions could be thought of as macrobehavioral therapy. A structural intervention changes the environment or context within which people act for the purpose of influencing individual health behaviors.
Because HIV transmission is continuing to increase among women in the southern United States, interventions to stem this epidemic are particularly warranted.3 However, it is unclear which interventions or combinations of interventions should be implemented. In particular, little is known about the relative cost-effectiveness of different HIV-preventive interventions. In a prior study, we described a tool, “Maximizing the Benefit,” that estimates the cost-effectiveness of 26 different HIV prevention interventions.4 This study showed that for high-prevalence populations (e.g., gay men, injection drug users), individually focused interventions can be cost-effective, but for low-prevalence populations (e.g., black women), it appeared that these individually focused interventions were not cost-effective, suggesting we should look more at structural interventions. Structural interventions are theoretically promising because they have the ability to reach large numbers of people in contrast to individually focused methods, which rarely can be scaled up to large numbers because they are often labor-intensive and require complex logistic planning to bring clients and counselors together to work on behavior change. Furthermore, the Centers for Disease Control and Prevention (CDC) estimates that the number of new HIV infections that occur annually has not declined in more than 5 years.5 Structural interventions have not been a significant part of CDC investment in HIV prevention and are worth considering given that the existing interventions used in the field do not appear to be making a significant difference in the epidemic.
We examined 6 different structural interventions and their potential cost-effectiveness in reducing HIV transmission among women in the South. The structural interventions we considered were: condom availability6; needle availability, through needle exchange programs7 or through deregulation of the laws that prohibit the sale of sterile needles without a prescription8; alcohol taxes9; and mass media campaigns.10 We also estimated the cost-effectiveness of other interventions that have some structural aspects but also combine individual-level approaches, specifically street outreach,11 and opinion leader programs.12 It is not possible to estimate the cost-effectiveness of other interventions designed to change the environment if they have not been evaluated or shown to be effective. Therefore, the structural interventions of closing bathhouses or crack houses were not considered in this study.
The 6 different interventions can be considered structural because their immediate target is the conditions in which people live rather than the people themselves. Condom availability simply provides access to condoms and does not necessarily involve overt motivational or educational messages. Needle/syringe exchange programs provide sterile needles to individuals who return used needles in exchange, thereby reducing the likelihood of reuse of an infected needle. Needle deregulation efforts, by allowing intravenous drug users (IDUs) to purchase their own sterile needles/syringes, should reduce the likelihood that IDUs will reuse infected needles/syringes from others. Alcohol use has been associated with high-risk behaviors in many studies, including high-risk sexual behavior.13–15 Although reducing alcohol availability is not usually considered an HIV prevention strategy for individuals, it may have the effect of reducing risky behavior, as suggested by studies showing reductions in sexually transmitted diseases (STDs) after alcohol tax increases.9 Media campaigns potentially influence norms. Norms are an important part of the social environment and influence values, expectations, and behavior.16 Media campaigns promoting condom use have been very successful in Europe and in developing countries.10,17 Community mobilization/street outreach programs have an individual-level methodology that often includes one-to-one counseling, but the interventions can also affect community norms or serve as additional sources for condoms or sterile needles.18,19 Opinion leader interventions also have structural components in that they also are meant to change social norms from which individual behavior change follows. The program is based on diffusion of innovation/social influence principles, which states that trends and innovations are often initiated by a relatively small segment of opinion leaders in the population.20 Once innovations are visibly modeled and accepted, they then diffuse throughout a population, influencing others.21
We used “Maximizing the Benefit” to estimate the cost-effectiveness of the 6 different interventions likely to reduce high-risk behaviors related to HIV transmission to women.22 Cost-effectiveness is assessed from the perspective of the public health system and includes only resources (purchased, donated, or volunteered) used to implement the intervention, but excludes any cost incurred by the participants unless they are reimbursed. Costs were taken from the literature or estimated. The tool uses several different mathematical models, including a Bernoulli model, for interventions designed to reduce sexual transmission by reducing partner change or increasing condom use,23 a similar Bernoulli model for interventions designed to reduce needle sharing, and a “proportionate change” model for interventions evaluated with STD incidence reduction as outcome. It estimates the cost per HIV case prevented by taking into account the epidemiologic contexts, behavioral change as a result of an intervention, and costs of its implementation.
The duration of the interventions' effectiveness is unknown. Except for the opinion leader programs, our models assumed that the interventions were immediately effective and the effect lasted until their last measurement, but no longer. In the opinion leader programs, we extended the effectiveness to a 6-month period. If the effect lasts longer than when the last measurement was taken, each of the infections will be considerably more cost-effective than we estimated, but determining exactly when the effects of any intervention disappear may not be feasible. Theoretically, structural interventions support continued behavior change as long as the intervention is in force. (Details are fully described in a manual at http://www.rand.org/health/tools/hiv_prevention.html.)
To estimate the cost-effectiveness of these interventions if implemented among women in the South, we obtained HIV prevalence values from the state of Louisiana and assumed they were similar to HIV infection rates in other parts of the South. Among women, HIV rates are highest among blacks, with the prevalence estimated to be 0.006. For black men, the prevalence estimate is 0.016.3,24
To calculate the cost-effectiveness of condom availability, we used the Bernoulli model formula and assumed that condom availability would increase condom use from 40% to 52% in men and 28% to 36% in women based on studies in Louisiana.25 For assessing the cost-effectiveness of needle availability, we assumed the prevalence of HIV in IDUs was 10% based on estimates of HIV prevalence among IDUs in southern cities from Holmberg.26 The cost of a needle exchange program is estimated at $10 per person for a 3-month period.7,27 There is no obvious cost of needle deregulation, but to have it more fully accepted, it may require lobbying and also education of pharmacists.28,29 We arbitrarily chose a one-time cost of $100,000, because prior studies have shown that even when needle sales may be legal, only 66% of pharmacies say they never sell needles to persons they suspect are IDUs.29 (The legality of purchasing needles without a physician prescription in all the southern states is unclear but could be interpreted as legal, except in Georgia, where it is illegal.)28 Similarly, although alcohol taxes generate money, we expected that the revenue would be spent elsewhere and that it would cost at least $100,000 in lobbying efforts to get an alcohol tax passed in each state.
We assumed the media campaign would target black women and would increase condom use from 48% to 57% rather than the increase in rates seen in Switzerland, which had a lower baseline than currently exists in the southern United States. (The figures we assumed reflect the same increase associated with street outreach.11) We used the same 7-year timeframe (based on the HIV prevention media campaign “STOP AIDS” in Switzerland, which included widely placed graphic billboards and television and radio advertisements promoting condom use10) and estimated an annual $1,000,000 budget to reach approximately 1,500,000 persons in the general population with HIV prevalence at 0.3%. To estimate the cost-effectiveness of the opinion leader programs, we estimated the effect would be the same as that seen in gay men as reported by Kelly et al30 but used the HIV prevalence among black women and their sex partners in our formula.
The numbers that would be reached by the intervention are estimated based on the numbers reached by previous evaluations or what could reasonably be expected by a single state. For alcohol tax, we limited the affected population to half of the estimated highest-risk drinkers, approximately 5% to 6% nationally.31 (For Louisiana, we chose 2.5% of the total population, or 100,000.) We estimated the cost-effectiveness of the opinion leader intervention and needle exchange programs using target populations numbers that were associated with previous evaluations. Although these 2 interventions could potentially be scaled up and reach more people, few costs savings would be expected and the cost-effectiveness would remain about the same. The threshold for cost-effectiveness was set at $200,000, the estimated cost of lifetime medical treatment for HIV.32
The most cost-effective structural interventions were alcohol taxes, needle deregulation, and needle exchange, all theoretically costing less than $10,000 per case of HIV infection prevented (Table 1). Street outreach, condom availability, and mass media were also cost-effective. Condom availability and mass media campaigns had the potential to prevent the most potential cases of HIV transmission (45.3 and 40.7 per year, respectively) as a result of the large number of people that can be reached with these structural methods. In comparison, it is not known whether programs like needle exchange can be easily scaled up to reach more than a few thousand IDUs. In the example we used, reaching 1000 IDUs through needle exchange resulted in the prevention of one infection. Because needle deregulation could become a state law, it could potentially reach many more IDUs and thus prevent more infections.
Although opinion leader programs were cost-effective when targeted to men who have sex with men in earlier studies and were estimated to cost approximately $65,000 to prevent a case of HIV where the prevalence of HIV was 9%,33 the cost-effectiveness worsened to $1,000,000 per infection prevented when the program was applied to a population with an HIV prevalence of 0.6%. Even if the opinion leader program's effectiveness could increase condom use to 99% among black women, the cost per infection prevented would still be relatively high, at $120,000, and the number of infections prevented would be less than one for a program that reaches 400 to 500 persons. The scalability of opinion leader programs is unknown, but it may not be feasible to use this type of program to reach several hundreds of thousands of women. Among black women in the South, any structural intervention that reaches fewer than 1000 persons may not even prevent one HIV infection. Any program may need to reach a minimum of 20,000 people to prevent 10 infections.
Although the number of cases of HIV is increasing in minority women in the southern United States, their HIV prevalence is still relatively low (0.006). This means that to make a meaningful difference at the population level within achievable budgets, HIV prevention interventions for this population will have to reach large numbers of persons and cost very little per person reached. Interventions exist that meet these criteria and thus which could be used within reasonable budgets to substantially slow the spread of HIV in women in the South.
Caution should be used when applying the cost-effectiveness estimates we developed. First, our estimates are based on mathematical models, the assumptions of which are clearly open to challenge. This means the results are at best very rough approximations. Nonetheless, the results show differences of orders of magnitude, so these approximations are useful in planning.
Second, except for the opinion leader program, none of the other structural interventions have ever been evaluated using a randomized, controlled intervention design, so their actual efficacy is in some doubt. If they are not efficacious, they will never be cost-effective. However, it is very difficult to conduct rigorous controlled intervention trials for structural interventions, because they are potentially extremely expensive and the conditions are difficult to control. Given that they have to reach huge numbers of people, random assignment may not be politically feasible, and even it were feasible, it may be impossible to find matched populations of that large a scale.
Except for alcohol taxes and needle availability programs, all of the interventions were evaluated based on self-reported sexual behavior. Sometimes self-reports may not be reliable, because the participants may supply answers based on social desirability.
Although our estimates of the cost-effectiveness of these 6 interventions have several limitations, the formulas that estimate HIV transmission do support the general concept that campaigns reaching large volumes of people are needed to favorably impact transmission among low-prevalence populations. Furthermore, in other countries where structural interventions have been used, including the 100% condom campaign in Thailand34 and the STOP AIDS campaign in Switzerland,10 population-level impacts have been measurable. Even if individual-level behavioral interventions were cost-effective, it is highly doubtful that they could ever reach enough people in this low-prevalence population to produce a meaningful reduction in HIV incidence. This precept should be considered in the development of any campaign to stem the HIV/STD epidemic in the southern United States.
1.Kirschenbaum H. Carl Rogers's life and work: An assessment on the 100th anniversary of his birth. J Couns Dev 2004; 82:116–124.
2.Skinner BF. Reflections on Behaviorism and Society. Englewood Cliffs, NJ: Prentice-Hall, 1978.
3.Karon JM, Fleming PL, Steketee RW, et al. HIV in the United States at the turn of the century: An epidemic in transition. Am J Public Health 2001; 91:1060–1068.
4.Cohen D, Wu S, Farley T. Comparing the cost-effectiveness of HIV prevention interventions. J Acquir Immun Defic Syndr 2004; 37:1404–1414.
6.Cohen DA, Farley TA, Bedimo-Etame JR, et al. Implementation of condom social marketing in Louisiana, 1993 to 1996. Am J Public Health 1999; 89:204–208.
7.Heimer R, Khoshnood K, Bigg D, et al. Syringe use and reuse: Effects of syringe exchange programs in four cities. J Acquir Immun Defic Syndr Hum Retrovirol 1998; 18(suppl 1):S37–S44.
8.Groseclose SL, Weinstein B, Jones TS, et al. Impact of increased legal access to needles and syringes on practices of injecting-drug users and police officers—Connecticut, 1992–1993. J Acquir Immun Defic Syndr Hum Retrovirol 1995; 10:82–89.
9.Alcohol policy and sexually transmitted disease rates—United States, 1981–1995. MMWR Morb Mortal Wkly Rep 2000; 49:346–349.
10.Dubois-Arber F, Jeannin A, Konings E, et al. Increased condom use without other major changes in sexual behavior among the general population in Switzerland. Am J Public Health 1997; 87:558–566.
11.Wendell DA, Cohen DA, LeSage D, et al. Street outreach for HIV prevention: Effectiveness of a state-wide programme. Int J STD AIDS 2003; 14:334–340.
12.Kelly JA, St. Lawrence JS, Diaz YE, et al. HIV risk behavior reduction following intervention with key opinion leaders of population: An experimental analysis. Am J Public Health 1991; 81:168–171.
13.Seage GR 3rd, Mayer KH, Wold C, et al. The social context of drinking, drug use, and unsafe sex in the Boston Young Men Study. J Acquir Immun Defic Syndr Hum Retrovirol 1998; 17:368–375.
14.Staton M, Leukefeld C, Logan TK, et al. Risky sex behavior and substance use among young adults. Health Soc Work 1999; 24:147–154.
15.Petry NM. Alcohol use in HIV patients: What we don't know may hurt us. Int J STD AIDS 1999; 10:561–570.
16.Basen-Engquist K, Coyle KK, Parcel GS, et al. Schoolwide effects of a multicomponent HIV, STD, and pregnancy prevention program for high school students. Health Educ Behav 2001; 28:166–185.
17.Nigeria using more condoms. Africa Health 1997; 19:3.
18.Kegeles SM, Hays RB, Coates TJ. The Mpowerment Project: A community-level HIV prevention intervention for young gay men. Am J Public Health 1996; 86:1129–1136.
19.Coyle SL, Needle RH, Normand J. Outreach-based HIV prevention for injecting drug users: A review of published outcome data. Public Health Rep 1998; 113(suppl 1):19–30.
20.Kelly JA, Murphy DA, Sikkema KJ, et al. Randomised, controlled, community-level HIV-prevention intervention for sexual-risk behaviour among homosexual men in U.S. cities. Community HIV Prevention Research Collaborative. Lancet 1997; 350:1500–1505.
21.Backer TE, Rogers EM. Diffusion of innovations theory and work-site AIDS programs. J Health Commun 1998; 3:17–28.
22.Cohen DA, Wu SY, Farley TA. Comparing the cost-effectiveness of HIV prevention interventions. J Acquir Immun Defic Syndr 2004; 37:1404–1414.
23.Weinstein M, Graham J, Siegel J, et al. Cost-effectiveness analysis of AIDS prevention programs: Concepts, complications, and illustration. In: Turner C, Miller H, Moses L, eds. AIDS: Sexual Behavior and Intravenous Drug Use. Washington, DC: National Academy Press, 1989:471–499.
24.Bedimo AL, Pinkerton SD, Cohen DA, et al. Condom distribution: A cost-utility analysis. Int J STD AIDS 2002; 13:384–392.
25.Cohen DA, Wu SY, Farley TF. Maximizing the Benefit: Estimating the Relative Cost-Effectiveness of HIV Prevention Interventions. Paper presented at the 16th biennial meeting of the International Society for Sexually Transmitted Disease Research (ISSTDR), 2003, Ottawa, Canada.
26.Holmberg SD. The estimated prevalence and incidence of HIV in 96 large U.S. metropolitan areas. Am J Public Health 1996; 86:642–654.
27.Lurie P, Gorsky R, Jones TS, et al. An economic analysis of needle exchange and pharmacy-based programs to increase sterile syringe availability for injection drug users. J Acquir Immun Defic Syndr Hum Retrovirol 1998; 18(suppl 1):S126–132.
28.Burris S, Vernick JS, Ditzler A, et al. The legality of selling or giving syringes to injection drug users. J Am Pharm Assoc (Wash) 2002; 42(suppl 2):S13–18.
29.Farley T, Niccolai L, Billeter M, et al. Attitudes and practices of pharmacy managers regarding needle sales to injection drug users. J Am Pharm Assoc 1999; 39:23–26.
30.Kelly JA, St. Lawrence JS, Stevenson LY, et al. Community AIDS/HIV risk reduction: The effects of endorsements by popular people in three cities. Am J Public Health 1992; 82:1483–1489.
32.Holtgrave DR, Pinkerton SD. Updates of cost of illness and quality of life estimates for use in economic evaluations of HIV prevention programs. J Acquir Immun Defic Syndr Hum Retrovirol 1997; 16:54–62.
33.Pinkerton SD, Holtgrave DR, DiFranceisco WJ, et al. Cost-effectiveness of a community-level HIV risk reduction intervention. Am J Public Health 1998; 88:1239–1242.
34.Rojanapithayakorn W, Hanenberg R. The 100% condom program in Thailand. AIDS 1996, 10:1–7.
35.Chesson H, Harrison P, Kassler WJ. Sex under the influence: The effect of alcohol policy on sexually transmitted disease rates in the United States. J Law Econ 2000; XLIII:215–238.