IN THE UNITED STATES, Chlamydia trachomatis is one of the most common and costly sexually transmitted diseases (STDs), especially among sexually active adolescents and young adults. 1 Approximately 3 million new cases of C trachomatis infection occur annually, and up to 70% of C trachomatis infections in women are asymptomatic. 1,2 C trachomatis infections in women can lead to pelvic inflammatory disease (PID), which can cause ectopic pregnancy, tubal infertility, and chronic pelvic pain. C trachomatis infection and its attributable PID cost about $1.6 billion in the U.S. annually. 3
The Preventive Health Amendments of 1992 provided $8.3 million in fiscal year 1994 for the prevention of infertility associated primarily with C trachomatis infections. 4,5 Essential activities authorized by the legislation include detection, treatment, follow-up, and referral services for at-risk women and their sex partners. These publicly funded programs support C trachomatis screening primarily in family planning clinics and STD clinics. Although these programs have allocated expanded resources for the prevention and control of C trachomatis infection, resources to control C trachomatis infections in many medical care clinics are still limited. 6 Individual programs may not have budgets that are high enough to screen all women with the most effective test assays and treatment. Budget constraints force many clinics to make choices regarding which populations to screen, which testing technologies to use, and which treatments to use. In practice, program managers often have fixed amounts of money that can be spent to provide these services; the effect of this is that a fixed amount of money is available per clinic visit. The goal is to maximize the detection and treatment of C trachomatis infection on this fixed budget.
The objective of this study report is to present a resource allocation model to determine the optimal combination of screening coverage, test selection, and treatment for controlling C trachomatis infection in asymptomatic women clients of a family planning program with a fixed budget.
The resource allocation model developed in this study was based on a binary integer linear programming method used in the field of operations research. Similar techniques have been used to solve a wide variety of optimization problems in areas as diverse as production planning, marketing, police patrol officer scheduling, child immunization, and HIV infection prevention. 7–13 The resource allocation model applied in this study considered both the clinic and healthcare system perspectives. From the clinic perspective, the optimal strategy was defined as the strategy that maximized the number of women with C trachomatis infection who were cured with any given C trachomatis control budget, because a primary goal of family planning programs is to deliver optimal care. From the healthcare system perspective, the optimal strategy was defined as the one that maximized the cost-saving value (total benefit − total cost) with any given C trachomatis control budget, because a goal of the healthcare system often is to deliver appropriate care in a cost-effective manner. 14 The total benefit included the cost-savings of averted sequelae of C trachomatis infections. The total cost included the direct medical costs for testing and treatment. In this analysis, we identified the optimal screening and treatment strategies for a fixed annual program budget for a family planning program of $6 per visit from both the clinic and healthcare system perspective. The fixed annual budget per visit was calculated by taking the total clinic C trachomatis control budget and dividing by the number of visits by women.
We considered age distribution, age-specific prevalence of C trachomatis infection, test assay selection, and treatment regimen selection in this model. We considered three age groups: (1) less than 20 years, (2) 20 to 24 years, and (3) more than 24 years. The age groups analyzed were similar to those classified in the CDC recommendation for screening and treating C trachomatis infection. 1 We considered two test assays (a nucleic acid [DNA] probe test and a ligase chain reaction [LCR] test) and two CDC-recommended treatments (doxycycline and azithromycin) in the model. DNA probe and LCR tests are both commonly used for screening C trachomatis infection, and doxycycline and azithromycin are recommended for treatment of C trachomatis infection by the CDC and are also commonly used.
Parameter values were obtained from various sources. The age distribution and the age-specific prevalence of C trachomatis infection were based on 5078 annual visits by women, universally screened for C trachomatis infection in Philadelphia's publicly funded family planning program (Table 1). 15 The majority of women who tested for C trachomatis infection were those aged <31 years, because Philadelphia's publicly funded family planning program limited screening to women aged <31 years. The values for test sensitivity and specificity we used were also based on an evaluation done in this family planning program (Table 1); these included the DNA probe assay (Pace 2; Gen-Probe, San Diego, CA) and LCR test (LCx LCR; Abbott Laboratories, North Chicago, IL) on cervical specimens only. Of 5078 tests performed using LCR, 302 (6%) were positive. Simply, we used this positive rate as the prevalence of C trachomatis infection in this study. Test cost, treatment effectiveness and cost, and cost of averted sequelae of C trachomatis infection were drawn from the literature (Table 1). 16–28
To determine if the strategy that maximizes the number of women in a given age group who are cured of C trachomatis infection under a fixed budget in one region is different from that in another region under the same budget, we conducted a second analysis in which the age distribution and the age-specific prevalence of C trachomatis infection in family planning programs in Oregon and in Washington were substituted for the same data from Philadelphia. The age distribution was 35% for women aged <20 years; 36% for women aged 20 to 24 years; and 29% for women aged >24 years. Their respective age-specific prevalences of C trachomatis infection were 6.0% for women aged <20 years; 4.1% for women aged 20 to 24 years, and 1.6% for women aged >24 years. 15
We made five assumptions in our models. First, we assumed that all women who visited the family planning clinic were asymptomatic for C trachomatis infection. This simplifying assumption was based on data from Philadelphia's publicly funded family planning program showing that <10% of attendees have symptoms of C trachomatis infection. 15 Second, we assumed that all women who tested positive for C trachomatis infections would be treated. Third, the model allowed for screening all or some age groups. Our model did not require all age groups to be screened. Fourth, we assumed that all women would undergo the same test and treatment if they were screened. Finally, no woman would receive more than one test or treatment for C trachomatis infection at a clinic visit.
The details of the model from the clinic perspective that maximized the number of women with C trachomatis infection who were cured are described in the Appendix. The baseline scenario was based on the baseline value of each parameter listed in Table 1. The model from the healthcare system perspective was identical except that it maximized the cost-saving value. This binary integer linear program contained 28 decision variables, representing 28 potential strategies, that allowed the model to identify the optimal strategy with a combination of screening coverage, test selection, and treatment selection for each of three age groups and an equation that calculated the number of women with C trachomatis infections who were cured under the 2 conditions: (1) the total C trachomatis control budget had to exceed or equal the total costs for all women screened and treated for C trachomatis infections, and (2) only 1 of 28 potential strategies was selected. The 28 potential strategies were derived from 7 screening coverages and 4 combinations of test and treatment selections. The 7 screening coverages included: (1) women aged <20 years; (2) women aged 20 to 24 years; (3) women aged >24 years; (4) women aged <20 years and 20 to 24 years; (5) women aged <20 and >24 years; (6) women aged 20 to 24 years and >24 years; and (7) all women. The four combinations of test and treatment selections included: (1) DNA probe for test and doxycycline for treatment; (2) DNA probe for test and azithromycin for treatment; (3) LCR for test and doxycycline for treatment; and (4) LCR for test and azithromycin for treatment. The model calculated the cost and the number of women with C trachomatis infections who were cured for each of 28 potential strategies, compared each strategy's cost to the budget constraint, and selected the strategy that met the budget requirement and maximized the number of women with C trachomatis infections who were cured.
In addition, we performed multiple sensitivity analyses in which one or more parameter values were varied while holding other parameters at their baseline values. Ranges of the parameter values are summarized in Table 1.
From the clinic perspective, the optimal strategy for a fixed budget of $6 per visit involved screening all women with the DNA probe assay and treating all women with positive tests with azithromycin (Table 2). Under this strategy, the family planning programs would spend $5 per visit for screening and treatment, and 61% of women with C trachomatis infection (183/302) would be cured. This was also the optimal strategy from the healthcare system perspective and resulted in a cost-savings of $140,176.
Optimal strategies under other selected budgets from the clinic are presented in Table 2. Without any budget limitation, screening all women with LCR and treating women who tested positive with azithromycin would result in curing 89% of infected women (270/302). Meeting the clinic perspective goal of maximizing the number of women cured did not require screening all age groups at some budget levels. For example, strategy 7 (which screened all females aged <20 years and 20–24 years with LCR) would cure more infected women than strategy 6 (which screened all women with a DNA probe) if the budget level was <$7.30 per visit, even though strategy 7 would screen fewer women than strategy 6 (2 different age groups versus 3).
The optimal strategy from the clinic perspective did not always match the optimal strategy from the healthcare system perspective. The two perspectives could result in two different optimal strategies under the same budget condition (Table 2). For a budget of <$12.40 per visit, strategy 9 (which screened all women with LCR and treated those who had positive tests with doxycycline) was the optimal strategy from the clinic perspective but not from the healthcare system perspective. The optimal strategy from the healthcare system perspective under a budget of <$12.40 per visit was strategy 8 (which screened all females aged <20 years and 20–24 years with LCR and treated those who tested positive with azithromycin), because strategy 8 resulted in a cost-savings of $166,229 rather than $165,767 under strategy 9.
When the age distribution and the age-specific prevalence of C trachomatis infection from Oregon and Washington state family planning clinics were applied to the model, the optimal strategy for a fixed budget of $6 per visit was the same as that for the Philadelphia's publicly funded family planning program.
Sensitivity analyses showed that two parameters strongly influenced which strategy was optimal under the baseline budget of $6 per visit from the clinic perspective: the cost of the DNA probe assay and the cost of the LCR assay. For example, when the cost of the DNA probe assay was increased from $4.50 to $5.70, the optimal strategy involved screening all women with the DNA probe assay and treating all with positive tests with doxycycline rather than azithromycin. When the cost of the LCR was decreased from $12.00 to $10.20, the optimal strategy involved screening all females aged <20 years and 20 to 24 years with LCR and treating all with positive tests with doxycycline.
Although our sensitivity analysis showed that the cost of C trachomatis sequelae, which ranged from $600 to $1200, did not strongly influence the choice of optimal strategy (with the cost-savings ranging from $85,054 to $194,854) from the healthcare system perspective under the baseline budget, the cost of sequelae would influence choice of optimal strategy under the budget of <$12.40 per visit. When we increased the cost of sequelae from $900 to $920, the optimal strategy under this budget level from the healthcare system perspective changed from strategy 8 to strategy 9 because the cost-savings became $170,561 and $170,555 for strategy 9 and strategy 8, respectively.
Our results suggest that the optimal strategy of screening coverage, test selection, and treatment selection vary greatly as fixed budget levels change. With an increased budget per visit, more infected women can be cured. Our results have shown that it required an additional increase of $0.20 to $0.40 per visit to switch from doxycycline to azithromycin but required an additional increase of $1.40 to $5.10 per visit to switch from DNA probe to LCR, depending on the screening coverage (Table 2). This is why DNA probe was selected at all low budget levels and LCR at all high budget levels in Table 2. Our analysis also showed that under certain budget conditions, the strategy that cures more women may not necessarily involve screening more women. For example, at a fixed budget of $7.00 per visit, a strategy involving LCR screening of young women with higher prevalence but no screening of older women (>24 years) with lower prevalence resulted in curing more women than a strategy involving screening all women with DNA probe tests.
Our sensitivity analysis showed that the optimal strategy at the budget of $6.00 was strongly influenced by test cost but not treatment cost, because more clients were tested than treated and because there was a $1.00 surplus per visit at this budget level that kept the optimal strategy unchanged although treatment cost could increase from $9.50 to $30.0. In fact, at most budget levels, treatment cost was smaller than screening cost. Our results suggested that if the prices of assays and drugs could be negotiated, the clinic managers might pay more attention to reducing assay prices, because more clients are tested than treated.
Realistically, we assumed that all women would receive the same test and treatment if they were screened. However, if we relaxed this assumption and allowed use of a strategy that did not apply uniform testing and treatment methods for all age groups, in some cases it would be possible to cure more women at a given budget level. For example, under the baseline budget of $6.00 per visit, 195 women could be cured if all females aged <20 years were tested with LCR, all aged 20 to 24 years were tested with DNA probe, none aged >24 years were tested, and all with positive tests were treated with azithromycin, in comparison with the 183 women cured with the current optimal strategy. The strategy involving more than one test or more than one treatment may be more complicated to implement. For example, when using two different screening tests in one facility, the clinic would need to ensure that women in different age groups received the appropriate test and that unique handling, storage, transport, and billing instructions for each test were carefully followed. In addition, providers would likely need additional training on the performance parameters of each test, so that they would be able to explain test performance issues to women in each age group. Therefore, clinic managers might need to balance optimal resource allocation with operational aspects when the clinic has the option of using a strategy that does not apply uniform testing and treatment methods for all age groups in the family planning program.
Numerous studies have identified that age and behavioral risks are among the strongest predictors of C trachomatis infection and should be considered as criteria for screening. 19,29–32 However, several studies have shown that in many settings, obtaining risk factor information from clients is done inconsistently or not at all. 33–35 Therefore, using age-based screening criteria that do not require providers to elicit C trachomatis risk factors from clients is easy. For this reason, age was the only variable we used to classify the prevalence of C trachomatis infection in our study. However, in settings where behavioral risk factor data are routinely collected and can be used to define C trachomatis prevalence, the current resource allocation model can be modified to use both age and risk behaviors to identify the optimal strategy that could cure more infected women, especially for at-risk women, and prevent more transmissions.
This resource allocation model has several advantages over cost-effectiveness of C trachomatis screening strategies: (1) it identified cost-effectiveness under the fixed budget conditions that prevail in many family planning clinics and (2) it identified the optimal combination of screening coverage, test selection, and treatment selection for different age groups. In contrast, most cost-effectiveness studies typically considered the impact of changing only one program characteristic at a time (e.g., screening coverage, test selection, or treatment selection). 16,21,36–39
Our model has several limitations. First, it did not consider the potential cost-savings of preventing disease transmissions, side effects of treatment, costs other than screening and treatment, and other possible constraints, such as time required for providing patient counseling and sex partner services. Second, we assumed that the budget for C trachomatis screening and treatment efforts was distinct from other clinic budget allocations. If the clinic budget included both categorical C trachomatis prevention funds and general STD prevention funds, the model should be modified to include clinical management of other STDs, especially gonorrhea. Therefore, in a future study, a more complicated model should be built to better represent clinic practice, that is, to maximize the number of quality-adjusted life years (QALYs) for all patients who visit the clinic under the clinic annual budget constraint. However, constructing a model that includes clinical management of many STDs requires more data for its parameters. Obviously, more epidemiologic studies are needed before such a model can be designed. Third, the number of visits for a future budget period is not known at the time testing and treatment strategies are selected. Past experience could be used to forecast the volume of female clients, but predicted volume may be much different from the real volume. Finally, to estimate the effectiveness or the cost-effectiveness of the chlamydia screening intervention, family planning programs may need periodically to provide universal screening to the entire clinic population to estimate accurately the age-specific prevalence of C trachomatis infections and to adjust their optimal strategy accordingly, especially when the clinic has the option of using a strategy that does not need uniform testing and treatment methods for all age groups in the family planning program.
Clinic managers may use resource allocation models to assist them in decisions about efficient use of limited resources. Our study demonstrated that using a resource allocation model enables clinic managers to identify a control strategy for C trachomatis that cures the maximum number of women or yields the most cost-benefit value within a fixed budget when the age distribution and age-specific prevalence of C trachomatis are known.
The following is a description of an integer linear program that maximizes the number of infected women cured, from the clinic perspective. EQUATION
Amnl =N m *P m *SEN n *DE l
Bmnl =N m * (CT n + (P m *SEN n + (1 -P m)(1 -SPE n)) *CD l)
N m = Number of women at age group m in the population
P m = Prevalence of C trachomatis at age group m
SEN n = Sensitivity of test assay n
SPE n = Specificity of test assay n
CT n = Cost of test assay n
DE l = Effectiveness of medication l
CD l = Cost of medication l
αi 1 = Am 11, i = m, i = 1, 2, 3
αi 2 = Am 12, i = m, i = 1, 2, 3
αi 3 = Am 21, i = m, i = 1, 2, 3
αi 4 = Am 22, i = m, i = 1, 2, 3
α4 j = α1 j + α2 j, j = 1, 2, 3, 4
α5 j = α1 j + α3 j, j = 1, 2, 3, 4
α6 j = α2 j + α3 j, j = 1, 2, 3, 4
α7 j = α1 j + α2 j + α3 j, j = 1, 2, 3, 4
βi 1 = Bm 11, i = m, i = 1, 2, 3
βi 2 = Bm 12, i = m, i = 1, 2, 3
βi 3 = Bm 21, i = m, i = 1, 2, 3
βi 4 = Bm 22, i = m, i = 1, 2, 3
β4 j = β1 j + β2 j, j = 1, 2, 3, 4
β5 j = β1 j + β3 j, j = 1, 2, 3, 4
β6 j = β2 j + β3 j, j = 1, 2, 3, 4
β7 j = β1 j + β2 j + β3 j, j = 1, 2, 3, 4
X ij = 0, 1
i = screening coverage, i = 1,…, 7
j = combination of test and treatment selections, j = 1, 2, 3, 4
m = age group, m = 1, 2, 3
n = test assay, n = 1, 2
l = medication, l = 1, 2
1. Centers for Disease Control and Prevention. 1998 Guideline for treatment of sexually transmitted diseases. MMWR Morbid Mortal Wkly Rep 1998; 47 (RR-1):53–59.
2. Groseclose SL, Zaidi AA, DeLisle SJ, Levine WC, St Louis ME. Estimated incidence and prevalence of genital Chlamydia trachomatis
infections in the United States, 1996. Sex Transm Dis 1999; 26: 339–344.
3. Rein D, Kassler W, Irwin K, Rablee L. Direct medical cost of pelvic inflammatory disease and its sequelae: decreasing, but still substantial. Obstet Gynecol 2000; 95: 397–402.
4. Hillis S, Black C, Newhall J, Walsh C, Groseclose SL. New opportunities for chlamydia prevention: applications of science to public health practice. Sex Transm Dis 1995; 22: 197–202.
5. Division of STD Prevention. Sexually transmitted disease surveillance 1999 supplement, chlamydia prevalence monitoring project. Atlanta: Department of Health and Human Services, Centers for Disease Control and Prevention.
6. American Social Health Association. Update on fiscal year 1994: funding for the STD program of the CDC and NIH. Sex Transm Dis 1994; 21:55–58.
7. Lawrence K, Zanakis S. Production planning and scheduling: mathematic programming applications. Atlanta: Industrial Engineering and Management Press, 1984.
8. Lilien G, Kotler P. Marketing decision models. New York: Harper and Row, 1983.
9. Jacobson SH, Sewell EC, Deuson R, Weniger BG. An integer programming model for vaccine procurement and delivery for childhood immunization: a pilot study. Health Care Manag Sci 1999; 2: 1–9.
10. Richter A, Brandeau ML, Owens DK. An analysis of optimal resource allocation for prevention of infection with human immunodeficiency virus (HIV) in injection drug users and non-users. Med Decis Making 1999; 19: 167–179.
11. Committee on HIV Prevention Strategies in the United States. No time to lose: getting more from HIV prevention. Washington, DC: National Academy Press, 2001.
12. Winston WL. Introduction to mathematical programming: application and algorithms. 2nd ed. Belmont, California: Wadsworth Publishing Company, 1995.
13. Taylor P, Huxley S. A break from tradition for the San Francisco police: patrol officer scheduling using an optimization-based decision support system. Interfaces 1989; 19: 4–24.
14. Clemmer B, Haddix AC. Cost-benefit analysis. In: Haddix AC, Teutsch SM, Shaffer PA, Dunet DO, eds. Prevention effectiveness: a guide to decision analysis and economic evaluation. New York: Oxford University Press, 1996: 85–102.
15. Dicker LW, Mosure DJ, Levine WC, Black CM, Berman SM. Impact of switching laboratory tests on reported trends in Chlamydia trachomatis
infections. Am J Epidemiol 2000; 151: 430–435.
16. Howell MR, Quinn TC, Brathwaite W, Gaydos CA. Screening women for Chlamydia trachomatis
in family planning clinics: the cost-effectiveness of DNA amplification assays. Sex Transm Dis 1998; 25: 108–117.
17. Steece R. National pricing of chlamydia reagents. Washington: Association of State and Territorial Public Health Laboratory Directors, 1997.
18. Ciemins EL, Kent CK, Flood J, Klausner JD. Evaluation of chlamydia and gonorrhea screening criteria: San Francisco sexually transmitted disease clinic: 1997 to 1998. Sex Transm Dis 2000; 27: 165–167.
19. Marrazzo JM, Celum CL, Hillis SD, Fine D, Delisle S, Handsfield HH. Performance and cost-effectiveness of selective screening criteria for Chlamydia trachomatis
infection in women. Sex Transm Dis 1997; 24: 131–141.
20. Dean D, Ferrero D, McCarthy M. Comparison of performance and cost-effectiveness of direct fluorescent-antibody, ligase chain reaction, and PCR assays for verification of chlamydial enzyme immunoassay results for populations with a low to moderate prevalence of Chlamydia trachomatis
infection. J Clin Microbiol 1998; 36: 94–99.
21. Haddix AC, Hillis SD, Kassler WJ. The cost effectiveness of azithromycin for Chlamydia trachomatis
infections in women. Sex Transm Dis 1995; 22: 274–280.
22. Hillis SD, Coles FB, Litchfield B, et al. Doxycycline and azithromycin for prevention of chlamydial persistence or recurrence one month following treatment in women: a use-effectiveness study in public health settings. Sex Transm Dis 1997; 25: 5–11.
23. Cerin A, Grillner L, Persson E. Chlamydia test monitoring during therapy. Int J STD AIDS 1991; 2: S176–S179.
24. Martin DH, Mroczkowski TF, Dalu ZA, et al. A controlled trial of a single dose of azithromycin for the treatment of chlamydial urethritis and cervicitis. N Engl J Med 1992; 327: 921–925.
25. Handsfield HH, Stamm WE. Treating chlamydia infection: compliance versus cost. Sex Transm Dis 1997; 25: 12–13.
26. 1998 Drug Topics Red Book. Montvale, NJ: Medical Economics Company, 1998.
27. Shafer MB, Pantell RH, Schachter J. Is the routine pelvic examination needed with the advent of urine-based screening for sexually transmitted diseases? Arch Pediatr Adolesc Med 1999; 153: 119–125.
28. Centers for Disease Control and Prevention. Recommendations for the prevention and management of Chlamydia trachomatis
infections, 1993. MMWR Morb Mortal Wkly Rep 1993; 42 (RR-12):2–3.
29. Humphreys JT, Henneberry JF, Rickard RS, Beebe JL. Cost-benefit analysis of selective screening criteria for Chlamydia trachomatis
infection in women attending Colorado family planning clinics. Sex Transm Dis 1992; 19: 47–53.
30. Mosure DJ, Berman S, Fine D, Delisle S, Cates W, Boring JR. Genital chlamydia infections in sexually active female adolescents: do we really need to screen everyone? J Adolesc Health 1997; 20: 6–13.
31. Gaydos CA, Howell MR, Pare B, et al. Chlamydia trachomatis infections in female military recruits. N Engl J Med 1998; 339: 739–44.
32. Weinstock HS, Bolan GA, Kohn R, Balladares C, Back A, Oliva G. Chlamydia trachomatis
infection in women: a need for universal screening in high prevalence population? Am J Epidemiol 1992; 135: 41–47.
33. Eubanks C, Lafferty WE, Kimball AM, MacCormack R, Kassler WJ. Privatization of STD services in Tacoma, Washington: a quality review. Sex Transm Dis 1999; 26: 537–542.
34. Bull SS, Rietmeijer C, Fortenberry JD, et al. Practice patterns for the elicitation of sexual history, education, and counseling among providers of STD services: results from the gonorrhea community action project (GCAP). Sex Transm Dis 1999; 26: 584–589.
35. Tao G, Irwin KL, Kassler WJ. Missed opportunities to assess sexually transmitted diseases among US adults during routine medical checkups: results of the 1994 US National Health Interview Survey. Am J Prev Med 2000; 18: 109–114.
36. Sellors JW, Pickard L, Gafni A, et al. Effectiveness and efficiency of selective vs universal screening for chlamydial infection in sexually active young women. Arch Intern Med 1992; 152: 1837–1844.
37. Genc M, Mardh P. A cost-effectiveness analysis of screening and treatment for Chlamydia trachomatis
infection in asymptomatic women. Ann Intern Med 1996; 124: 1–7.
38. Magid D, Douglas JM, Schwartz JS. Doxycycline compared with Azithromycin for treating women with genital Chlamydia trachomatis
infections: an incremental cost-effectiveness analysis. Ann Intern Med 1996; 124: 389–399.
39. Phillips RS, Aronson MD, Taylor WC, Safran C. Should tests for Chlamydia trachomatis
cervical infection be done during routine gynecologic visits? An analysis of the costs of alternative strategies. Ann Intern Med 1987; 107: 188–194.