AIDS and Behavior

, Volume 16, Issue 5, pp 1115–1120

Cost–Utility Analysis of A Female Condom Promotion Program in Washington, DC


    • Department of Health, Behavior & SocietyJohns Hopkins Bloomberg School of Public Health
  • Catherine Maulsby
    • Department of Health, Behavior & SocietyJohns Hopkins Bloomberg School of Public Health
  • Michael Kharfen
    • HIV/AIDS, Hepatitis, STD & TB AdministrationD.C. Department of Health
  • Yujiang Jia
    • HIV/AIDS, Hepatitis, STD & TB AdministrationD.C. Department of Health
  • Charles Wu
    • HIV/AIDS, Hepatitis, STD & TB AdministrationD.C. Department of Health
  • Jenevieve Opoku
    • HIV/AIDS, Hepatitis, STD & TB AdministrationD.C. Department of Health
  • Tiffany West
    • HIV/AIDS, Hepatitis, STD & TB AdministrationD.C. Department of Health
  • Gregory Pappas
    • HIV/AIDS, Hepatitis, STD & TB AdministrationD.C. Department of Health
Original Paper

DOI: 10.1007/s10461-012-0174-5

Cite this article as:
Holtgrave, D.R., Maulsby, C., Kharfen, M. et al. AIDS Behav (2012) 16: 1115. doi:10.1007/s10461-012-0174-5


A retrospective economic evaluation of a female condom distribution and education program in Washington, DC. was conducted. Standard methods of cost, threshold and cost–utility analysis were utilized as recommended by the U.S. Panel on cost-effectiveness in health and medicine. The overall cost of the program that distributed 200,000 female condoms and provided educational services was $414,186 (at a total gross cost per condom used during sex of $3.19, including educational services). The number of HIV infections that would have to be averted in order for the program to be cost-saving was 1.13 in the societal perspective and 1.50 in the public sector payor perspective. The cost-effectiveness threshold of HIV infections to be averted was 0.46. Overall, mathematical modeling analyses estimated that the intervention averted approximately 23 HIV infections (even with the uncertainty inherent in this estimate, this value appears to well exceed the necessary thresholds), and the intervention resulted in a substantial net cost savings.


Female condomHIV preventionEconomicsCost-effectiveness analysisPolicy analysis


When used consistently and correctly, male and female condoms are highly effective strategies for the prevention of HIV transmission [1, 2]. It is critical that there are HIV prevention interventions entirely under the control of women, and female condoms are the only such woman initiated intervention currently on the open market. Unfortunately, female condoms are not yet in as wide-spread use as are male condoms. There are at least two apparent reasons for this slower uptake. First there is less available information in the general public about the female condom, and there exists some misinformation about the product (e.g., how it is used, whether it is “noisy” during use, etc.) [3]. Second, traditionally the unit cost for a female condom has been higher than for the male condom and there exists a perception among some that the female condom is “too expensive” for use; however, a new production process has allowed for the manufacturing of a female condom (FC2) that is substantially lower in per-unit cost yet retains its product effectiveness [2].

In 2009, the District of Columbia formed a public–private partnership with the DC Department of Health, Washington AIDS Partnership, Female Health Company and CVS/Caremark (first retail sale of FC2 in the U.S.) to conduct a project to (a) disseminate female condoms in areas with disproportionately high HIV prevalence among women; and (b) offer educational services on the use of the product. The MAC AIDS Fund provided funding support for the project, which subsequently engaged five community based organizations for the education and distribution activities. The basis for the project was the DC National HIV Behavioral Surveillance (NHBS) system study on high risk heterosexuals that found high HIV prevalence rates (5.2 % overall and 6.3 % among women), high rates of concurrency and low condom use [4]. The project targeted the heterosexual study areas for female condom promotion.

The project launched in March 2010 and has undertaken extensive activities. The project has provided brief educational interventions (less than 15 min) to 38,000 women and men, extended education sessions (more than 15 min) to 8,000, conducted more than 300 group education sessions, trained nearly 500 peers at health and non-health locations and distributed more than 300,000 FC2s. Evaluation has been a key component of the project. The project utilized a pre- and post-test evaluation design in the education sessions and found a significant increase in knowledge about the FC2 from an average 2.9 out of 5 in the pre-test to 4.6 out of 5 in the post-test. The department of health added female condom questions to the next NHBS heterosexual cycle and found a significant use of the female condom. Further, the project is currently conducting a longitudinal utilization study on frequency of FC2 use.

However, the original evaluation did not include an economic component, and that is the purpose of the present analysis. We conducted a retrospective cost, threshold and cost–utility analysis to examine the affordability, performance standards, and relative cost-effectiveness of the FC2 product provision and educational services.


Standard methods of cost, threshold and cost–utility analysis were employed in accordance with the U.S. Panel on cost-effectiveness in health and medicine [5] and as adapted to HIV prevention by Holtgrave and Pinkerton [6]. In a prior publication, these methods were adapted to the potential roll out of female condoms in Brazil and South Africa [2]. A 1 year time horizon was used to capture both costs and benefits. Both societal and payor perspectives were employed. The Johns Hopkins Bloomberg School of Public Health Institutional Review Board determined this retrospective analysis to be not human subjects research.

Cost Analysis

A standard micro-costing approach was used to carefully account for the resources consumed by the female condom provision and education intervention [57]. All program costs are expressed in 2010 U.S. dollars. Table 1 lists all resource categories consumed by the intervention. These costs were captured by a flexible cost analysis spreadsheet tool adapted from one previously employed in other projects [8]. The cost analysis sought to assess the actual resources consumed regardless of whether that amount was above or below the grant support provided by the Female Health Company and MAC AIDS Fund. These costs include the expenses related to procuring and distributing female condoms, educating people about how to use the condoms (and about HIV prevention more broadly), marketing these services, and the associated central staffing costs. The approach of the DC Female Condom program was to build competency and capacity in organizations that already work in the field of women’s health and HIV and/or sexually transmitted infection prevention (for instance providing outreach and education in community settings such as beauty salons and health centers); the purpose was not to establish brand new, entirely stand alone service delivery programs. Hence, it was assumed that the FC2 education and provision was delivered on the margin of thriving extant service delivery programs. Further, it should be noted that the Female Health Company provides initial training (here 3 days of effort) when a health department or other entity establishes a new FC2 program in a jurisdiction; the cost of this initial training effort is subsumed in the procurement costs of 200,000 FC2.
Table 1

Input parameter values for cost, threshold and cost–utility analyses


Parameter value


Cost elements:


 Education costs


Washington, DC DOH

 Staff costs


Washington, DC DOH

 Distribution costs


Washington, DC DOH

 Marketing costs


Washington, DC DOH

 Procurement costs


Washington, DC DOH

Condoms procured


Washington, DC DOH

Percent of condoms used during sex

65 %

Washington, DC DOH

Percent of condoms used for other purposes (e.g., education sessions)

17 %

Washington, DC DOH

Percent of condoms not used

18 %

Washington, DC DOH

Lifetime HIV medical care costs



No. QALYs saved per infection averted



Percent HIV medical care costs in public sector

75 %


Female HIV prevalence

8.8 %

Washington, DC DOH

Male HIV prevalence

6.6 %

Washington, DC DOH

HIV- women with non-ulcerative STI

8.8 %

Washington, DC DOH

HIV- Women with ulcerative STI

2.3 %

Washington, DC DOH

Female condom effectiveness

95 %


Per act HIV transmission probabilities


 No STI, female to male


[2, 16]

 No STI, male to female


[2, 16]

 Non-ulcerative STI, male to female


[2, 16]

 Ulcerative STI, male to female


[2, 16]

Crowd out of male condom use

0–13 %

[2, 17]

Table 1 also lists the number of FC2 distributed and estimated to be actually used; therefore, the cost per female condom actually used could be calculated. Such cost and cost-per-unit-used estimates provide information to answer questions about the overall resources consumed by, and about the affordability of such a program.

Threshold Analysis

Once the cost of an intervention is known, we can divide that cost by the net present value (discounted at 3 %) of the lifetime medical care costs of a case of HIV [7]. This is the societal perspective cost-saving threshold for the intervention. However, about 25 % of HIV care is paid for in the private sector [9]. Therefore, adjusting the lifetime medical costs by .75 (and then dividing the intervention costs by the public sector medical cost savings) yields the public sector payor perspective cost-saving threshold. Table 1 displays the estimated lifetime medical care costs used in this analysis (our value for this parameter is the same as has been employed by CDC in prior analysis of other HIV prevention interventions) [10].

However, many health care and public health interventions are not cost-saving yet society recognizes them as “cost-effective” and essential services [11]. Examples include screening mammography, type 2 diabetes screening, and kidney dialysis services. In order to calculate a cost-effectiveness threshold we employ the following formula:
$$ {\text{Cost-effective threshold }} = \, \frac{C \, }{{\left[ {T \, + \, \left( {Q*W} \right)} \right]}} \,, $$
where C is the cost of the intervention (calculated in the cost analysis described above), T is the net present value of lifetime medical care costs averted for each infection averted, Q is the number of quality adjusted life years (QALYs) saved each time an infection is averted, and W is society’s willing to pay to save one QALY. A number of different values for W have been expressed in the literature, but one value receiving much current attention is $100,000 per QALY saved [1214] therefore we employ that value here yet note that different analysts might reasonably employ different “cut off” values for W. Table 1 displays the input parameter values for T, Q and W (C is displayed in the results oriented Table 2). The cost-effectiveness threshold is only taken from the societal perspective since the construct Q is not applicable in the payor perspective.
Table 2

Results of cost, threshold and cost–utility analyses

Costs analysis results

 Total overall program cost


 Cost per FC2 used during sex


Threshold analysis results (necessary infections averted)

 Societal perspective cost-saving threshold


 Payor perspective cost-saving threshold


 Societal perspective cost-effective threshold


Estimated HIV infections averted

 Female to male transmission


 No STI, male to female


 Non-ulcerative STI, male to female


 Ulcerative STI, male to female


 Total HIV infections averted


Overall cost–utility analysis result

 Societal perspective

Net savings of $8.160 M

 Payor perspective

Net savings of $6.017 M

Impact of allowance for male condom crowd-out

 Total HIV infections averted


 Societal perspective cost–utility analysis

Net savings of $7.046 M

 Payor perspective cost–utility analysis

Net savings of $5.181 M

These threshold analyses serve to set performance standards for the female condom provision and educational services. If we wish to label these products and services “cost-saving” or “cost-effective,” it is important to determine what level of performance is necessary to warrant use of those labels. Threshold analyses provide exactly this information.

Cost–utility analysis (CUA)

The central formula in CUA is expressed as follows:
$$ R \, = \, \frac{{\left[ {C \, - \, \left( {A*T} \right)} \right]}}{{\left[ {A*Q} \right]}}, $$
where R is the cost–utility ratio (or net cost per quality adjusted life year saved); C, T and Q are as defined above; and A is the number of HIV infections averted by the intervention. In order to calculate A, we estimated the HIV infections that would have occurred with and without the provision of FC2 products and education. To do so, we took into account male and female HIV prevalence in Washington DC, STI rates among sexually active heterosexual women, per sexual act transmission probabilities, and overall effectiveness of FC2. Because we did not have sexual network information on all users of FC2, we made an assumption that all sex acts were randomly distributed among women in the targeted communities and their sexual partners. We conservatively only modeled “first generation” infections and transmissions, and did not attempt to model the routes of new infection downstream to future partners and their partners. Input values and sources for all of these parameters are displayed in Table 1 and are sufficient to provide a CUA base case result.

The calculations for the parameter A used the Table 1 input parameter values in the following way. Two hundred thousand female condoms were procured and 65 % (130,000) were used during sex according to the experience of the DC Female Condom program. We assumed that this implies coverage of 130,000 separate heterosexual acts of vaginal intercourse. In the base case we assume that none of these acts would have otherwise been covered by condom use because (a) we have no reason to believe that female condoms would have been used during these acts had the current project not been in place, and (b) we have no reason to believe that male condoms would have been used either (however, as described in the sensitivity analysis section below, we subject our analysis to uncertainty in this assumption).

The epidemiologic parameters in Table 1 lead to the following estimates (it is important to note the sources of the parameter values in Table 1 as many of the parameters values are empirical point estimates from the DC Department of Health’s surveillance system and their experience with the DC Female Condom program). First, that of the 130,000 protected acts, 8.8 % were among serostatus discordant couples in which the female partner was living with HIV, 6.6 % of the male partners were living with HIV, the female to male per act HIV transmission probability (to seronegative male partners) was .0005, and this probability was reduced by 95 % due to FC2 use. In other words, there 11,440 FC2 products used in such situations (130,000 * 8.8 %) and that use lead to 5.08 female to male transmissions averted (11,440 * .0005 * .934 * .95).

There may also be the possibility of male to female transmissions averted. It was estimated that 91.2 % of the women were HIV seronegative (100–8.8%). Of these HIV seronegative women, 88.9 % had no other sexually transmitted infection (STI), 8.8 % had a non-ulcerative STI (it is coincidental that this percentage is the same as HIV seroprevalence among women), and 2.3 % had a genital ulcerative STI. For HIV seronegative women with no STI, 105,400 FC2 were used (130,000 * 91.2 % * 88.9 %) and that use lead to 6.61 male to female transmissions averted (105,400 * .001 transmission probability * .066 male HIV prevalence rate * .95 FC2 effectiveness).

For HIV seronegative women with a non-ulcerative STI, 10,443 FC2 were used (130,000 * 91.2 % * 8.8 %); this provides the basis for an estimate of 6.54 HIV infections averted (10,433 * .01 transmission probability * .066 male HIV prevalence rate * .95 FC2 effectiveness). For HIV seronegative women with a genital ulcerative STI, 2,727 FC2 were used (130,000 * 91.2 % * 2.3 %), leading to an estimate of 5.13 HIV infections averted (2,727 * .03 transmission probability * .066 male HIV prevalence rate * .95 FC2 effectiveness). The total number of HIV infections averted is the sum across the four types of partnerships described here.

If the numerator of the cost–utility ratio is negative, we would assert that the costs saved by the program exceed the cost of its provision, and label the program “cost-saving.” If the numerator is positive (program costs exceed savings), we would label the program “cost-effective” compared to other medical and public health programs if the cost per QALY saved is less than $100,000 (of course, reasonable parties could disagree about the precise cut off to be employed here).

Sensitivity Analysis

Of course, all retrospective CUAs are subject to some uncertainty. Therefore, we performed a number of sensitivity analyses. Two analyses of particular importance are as follows: (a) an analysis that allows for the possibility that female condom use “crowds out” some amount of current male condom use, and (b) an analysis that determines how low the effectiveness of FC2 can fall before the intervention ceases to be cost-saving or cost-effective. Table 1 displays the crowd out parameter employed in a prior analysis in Brazil and South Africa [2], and to our knowledge is the maximum crowd out estimate seen in the literature (and that is based on one particular analysis in one study of female condom use among Madagascar sex workers) [17].


Table 2 contains the results for the cost, threshold, CUA and sensitivity analyses. The basic investment in the delivery of the intervention was $414,186. Expressed as a cost per product used during sex this is $3.19.

The societal perspective cost-saving threshold analysis result was 1.13 infections that would have to be averted to be cost-saving, and the public sector payor perspective cost-saving threshold result 1.50. Both threshold results are remarkably low due to the high cost of HIV care over the lifespan, and both thresholds would appear to be readily achievable. The cost-effective threshold result (from the societal perspective only) is 0.46, and is even more plausibly achievable.

The base case estimate for parameter A is approximately 23.35 infections averted (well above the threshold values described above). [This estimate of 23.35 infections averted is the sum of HIV infections averted across the four types of partnerships described in the methods section above (5.08 + 6.61 + 6.54 + 5.13 with rounding error to the second decimal place).] The base case CUA ratio indicates cost savings (and by tradition, if the numerator of R is negative, the ratio itself is not presented).

If we allow for crowd out of male condom use by female condom use at the specified level, the overall CUA result still indicates cost-saving. Further, even if we allow crowd out and FC2 effectiveness were to drop as low as 7.04 %, the result will still indicate cost-savings. Hence, the results of this analysis are highly robust to uncertainty in the input parameters.


The economic evaluation presented here indicates that the provision and promotion of female condoms in Washington DC is a highly productive use of public health investment.

Further the results are very robust to uncertainty in the input parameter values.

Of course, this analysis is subject to several limitations. First, the analysis was done retrospectively and could not benefit from prospective cost analysis; however, given the robustness of the findings, any marginal uncertainty in cost estimates would appear to have little impact on the major qualitative conclusion. Second, it would be ideal to measure directly the impact of the intervention on observed HIV incidence; however, the geographic cluster randomized trial that would be necessary to field such a study would be extremely expensive and likely prohibitively large in size. Therefore, we needed to rely on mathematical modeling techniques. Clearly there is uncertainty in the input parameters used for the estimation of numbers of HIV infections averted. However, despite these sources of uncertainty, the base case estimate is very much in excess of the necessary levels indicated by the threshold analyses. This observation combined with the sensitivity analyses presented here gives additional confidence in the overall conclusion even if the base case estimate of infections averted is imprecise in some way.

Third, further information about sexual networks in the targeted communities would have been useful for building a more complex mathematical model, yet the simplifying assumption of random distribution of sexual acts is seemingly the best assumption that can be made given the information available. Fourth, our modeling of only first generation transmissions builds in an inherent conservativeness to the analyses. Fifth, this analysis only considers the FC2 provision and education program offered in Washington DC relative to the absence of the program. This analysis does not attempt to answer the optimization questions of whether female condoms are the most cost-effective possible intervention in the District, or what is the ideal mix of female condoms and other HIV prevention interventions. Answering such questions should be the subject of future, additional research.

Despite these inherent limitations, overall this retrospective use of mathematical modeling and CUA would seem to give sufficient justification that provision and education of FC2 is deserving of a place on the menu of HIV prevention interventions considered worthy of potential investment [14].


Johns Hopkins University received support for this economic evaluation from the Female Health Company, the producer of the female condom product, FC2. The Washington DC Department of Health received support for the FC2 dissemination and education project from the Female Health Company; the educational project was also supported by the MAC AIDS Fund. Further details on this public/private sector partnership are provided in the text. Final control of the analysis and publication rested with the authors, not with any sponsor.

Copyright information

© Springer Science+Business Media, LLC 2012