IN RECENT DECADES, Chlamydia trachomatis infections have been especially prevalent among sexually active adolescents in the United States, with rates ranging from 3.7% to 12.4% among 15- to 19-year-old adolescents. 1 The medical consequences and costs of infection are greatest among women. Untreated and uncured chlamydia infection can result in pelvic inflammatory disease (PID), chronic pelvic pain, ectopic pregnancy, and infertility. Because chlamydia infections are asymptomatic in most women, efforts to prevent medical complications and decrease transmission have focused on screening. Most large-scale chlamydia screening programs are conducted in family planning and prenatal care clinics; very few primary care providers routinely offer screening services. 2–6 Some schools have recently expanded their health services to include STD care. More than 10% of middle/junior high schools and 20% of senior high schools provide screening and treatment for sexually transmitted diseases (STDs). 7
One major school-based STD screening and treatment program was implemented in eight New Orleans public senior high schools during 3 consecutive school years (1995–1998). The program began in three urban senior high schools. All 9th through 12th grade students in these schools, regardless of symptoms or sexual history, were given the opportunity to be tested for chlamydia and gonorrhea with urine-based ligase chain reaction (LCR) tests. Parental consent was required for testing. In the third school year (1997–1998), the program was expanded to five additional senior high schools. Recent studies 8,9 of this program showed that it was associated with declines in chlamydia prevalence among students.
While STD screening in school settings has been recognized as an important part of programs to control STDs nationally, no research has been done to examine its cost-effectiveness. In this cost-conscious health care delivery era, determining that a preventive measure is effective is no longer adequate to justify its use. Policy-makers and program planners also are concerned about cost (i.e., whether a particular prevention program is affordable) and cost-effectiveness (i.e., whether the effects of a program justify the cost of its implementation).
In this study report, we present a retrospective economic evaluation of the New Orleans STD screening program by comparing the school-based screening scenario with a non-school-based screening scenario (existing practice). This study addresses two research questions: (1) How much does it cost to implement the program? and (2) What is the incremental cost-effectiveness of replacing the non-school-based screening with the school-based screening program? The results are intended to inform future efforts to implement school-based STD screening programs and, more generally, to inform resource-allocation decisions in the health care system.
We chose to use the New Orleans program for our economic study for two reasons. First, to guide implementation and expansion of school-based STD screening programs, we needed to conduct an economic study of a program already in place. Second, to conduct our economic study, we needed information on the cost of program delivery, the results of the screening (prevalence), and the circumstances preceding screening, such as the presence of STD-related symptoms, health-care-seeking behaviors, and off-site screening services received in the previous 12 months. The studies of the New Orleans program conducted by Cohen et al 8,9 provided this information.
Although the school-based screening program offered both chlamydia and gonorrhea testing in eight high schools, this study focused on the first series of chlamydia screening conducted in the five high schools, added in the 1997 to 1998 school year. In those five schools, 5063 students were given the opportunity to be tested. Parental consent was obtained by 3228 students (63.8%), and among these, tests were completed on 2653 (82.2%; 1402 females and 1251 males, or 52.4% of all students offered testing). Students who had positive test results were offered counseling and a single dose of azithromycin, a recommended regimen for chlamydia. 10 Although one of the five high schools had a school-based health center, tests performed in the school-based screening program were independent of school health center services. Among students tested, 1395 females and 1242 males were surveyed about their STD-related symptoms, their health-care-seeking behaviors, and off-site screening services they received in the previous 12 months (off-site services are defined as STD services received outside the school-based STD screening program).
We used standard cost-effectiveness analysis methods 11 to estimate the incremental cost-effectiveness of the school-screening scenario compared with a non-school-based screening scenario. According to the student survey responses, in the non-school-screening scenario, a certain proportion of symptomatic students sought diagnostic evaluation off-site and a certain proportion of asymptomatic students were offered off-site screening at preventive care visits. Using a health care system perspective, we considered the effects of the screening program not only on the students tested but also on the female partners of male students. Health outcome was measured as cases of PID prevented by replacing the non-school-based screening with the school-based screening program. Costs were measured as costs associated with testing and treating chlamydia infection and with treating PID and its sequelae. The incremental cost-effectiveness ratio was quantified and measured as cost per case of PID prevented.
The base-case analysis was conducted in five steps. First, we developed a decision model (Figure 1) to estimate the number of chlamydia infections that would occur in each scenario. Second, we estimated the number of expected PID cases, based on the number of chlamydia infections expected among female students and female partners of infected male students in each scenario (N1 and N0: 1 refers to school-based screening, 0 refers to non-school-based screening). Third, we estimated the costs associated with testing and treating chlamydia infections in each scenario (P1 and P0). Fourth, we estimated the costs associated with treating PID and its sequelae in each scenario (C1 and C0). Fifth, we calculated the incremental cost-effectiveness ratio of the school-based screening program compared with non-school-based screening (CE ratio = [(P1 − P0) − (C0-C1)]/(N0 − N1)).
The decision tree diagram (Figure 1) shows the probability structure used in the two scenarios. In the school-based screening scenario, the probability variables include prevalence, test sensitivity, test specificity, rate of return for treatment, and effectiveness of treatment. In the non-school-based screening scenario, the variables include prevalence, probability of an infected student having symptoms, probability of an uninfected student having symptoms, probability of a symptomatic student having an off-site diagnostic evaluation, probability of an asymptomatic student having off-site STD screening, test sensitivity and specificity of off-site chlamydia tests, rate of return for treatment, and effectiveness of treatment. The values and sources of the above variables are summarized in Table 1. The means by which we calculated some of those estimates for the school-based scenario are as follows.
Because test sensitivity and specificity are less than 100%, the number of students with positive test results was not equal to the number of students with infection. Using the number of students tested (1402 females, 1251 males), the number of positive test results (167 for females, 80 for males), and the sensitivity (89%) and specificity (99%) of the urine-based LCR test, we estimated the true prevalence as 12.4% among females and 6.1% among males (see Appendix).
Rate of return for treatment.
Among infected students, 91.6% of females (153/167) and 97.5% of males (78/80) returned for treatment.
The means by which we calculated some estimates for the non-school-based screening scenario are as follows.
Probability of an infected student having STD symptoms.
Among students surveyed, 165 females and 79 males had positive results, and 16 females and 3 males reported symptoms (discharge, dysuria, sores, or ulcers) at the time of school screening. Thus, the probability of an infected student having symptoms was 9.7% among females (16/165) and 3.8% among males (3/79).
Probability of an uninfected student having STD symptoms.
At the time of the school screening, among the 1395 females and 1242 males who were tested and surveyed, 90 females and 23 males reported symptoms. Thus, the probability of an uninfected student having symptoms was 6.0% ([90 − 16]/[1395 − 165]) among females and 1.7% ([23 − 3]/[1242 − 79]) among males.
Probability of a symptomatic student having an off-site diagnostic evaluation.
In the preceding 12 months, among students surveyed, 106 females and 40 males reported symptoms; of these, 76 females and 28 males reported off-site diagnostic evaluations. Thus, the probability of a symptomatic student having an off-site diagnostic evaluation was 71.7% (76/106) among females and 70% (28/40) among males.
Probability of an asymptomatic student having an off-site STD screening.
In the preceding 12 months, among students surveyed, a total of 372 asymptomatic females and 157 asymptomatic males underwent an off-site chlamydia screening. Thus, the probability of an asymptomatic student having off-site screening was 28.86% (372/[1395 − 106]) among females and 12.96% (157/[1251 − 40]) among males.
Test sensitivity and specificity of off-site chlamydia evaluations.
The performance of off-site chlamydia evaluations could vary, depending on the type of chlamydia test used and the service setting. According to a provider survey (Janet St. Lawrence, personal communication), the most frequently used chlamydia tests are DNA probe; cell culture; urine nucleic acid amplification test; and enzyme immunoassay, enzyme-linked immunosorbent assay, or direct fluorescent antibody test (EIA-ELISA/DFA). Table 2 displays information on each test, including the frequency with which each test is used and test sensitivity and specificity. On the basis of these data, we estimated the weighted average sensitivity and specificity of the four classes of tests as 74.6% and 99%, respectively.
Cases of Expected Chlamydia Infection
Using the decision analysis model and DATA software (DATA, Williamstown, MA), we first estimated the number of chlamydia infections that would occur among female students in each scenario. The triangular nodes in the tree (Figure 1) adjacent to arrow signs represent the probabilities of chlamydia infection occurrence. In the school-screening scenario, the sum of those probabilities is the average likelihood of a student having an infection, which includes possibilities for an infection not detected due to poor test performance, a detected infection not treated, and a treated infection not cured due to treatment failure. In the non-school-screening scenario, the sum of those probabilities, which again is the average likelihood of a student having an infection, includes possibilities for an infected student not having an off-site chlamydia test, as well as an infection that was not detected, not treated, or not cured. The number of chlamydia infections that would occur among female students in each scenario was calculated as the product of the average likelihood of a female student having an infection in each scenario multiplied by 1402, the total number of female students tested.
Using the same approach, we then estimated the number of chlamydia infections that would occur among male students in each scenario. We also estimated the number of chlamydia infections that would occur among female partners of infected male students. Using 1999 National Youth Risk Behavior Survey data, 12 we estimated that sexually active students had an average of 1.6 partners. We assumed that 50% of those female partners attended the same schools and already were screened. For the 50% who were not screened at school, we assumed that 11.5% already were infected (the rate being based on the female chlamydia prevalence at all eight high schools, estimated by Cohen and coworkers 9) and 88.5% were at risk for infection. Since the probability of an uninfected female acquiring chlamydia as a result of intercourse with an infected male is between 50% and 70%, 13 in this analysis, we used a 60% probability of transmission. Using the above assumptions and estimates, we estimated that for every male student infected, 0.42 (1.6 × 50% × 88.5% × 60%) female partners would become infected.
Cases of Expected PID
We estimated the number of expected PID cases on the basis of the number of chlamydia infections expected among female students and female partners of infected male students in each scenario. The probability of PID developing is a function of treatment success. Published studies 14,15 indicate that 20% to 40% of untreated/uncured chlamydia infections will develop into PID. For our analysis, we assumed 30% of untreated/uncured chlamydia infections would develop into PID.
Costs Associated With Testing and Treating Chlamydia Infections
In the school screening scenario, as displayed in Table 3, we considered all costs associated with program delivery. Staff costs included salaries for a project manager, field staff, and a school nurse who provided counseling and treatment. Physician supervision was provided in kind. Lab costs were based on $7.00 per test, and medication costs were based on $10 per 1-g sachet of azithromycin. Data entry costs for tracking students, parental consents, test results, and treatment were estimated at $1600 per school.
In the non-school-based screening scenario, we considered all costs associated with off-site chlamydia evaluations. For off-site diagnostic evaluation services for symptomatic students, we considered both public- and private-setting costs, because students seek health care at both settings. For off-site screening services for asymptomatic students, we considered only the public-setting cost, because most chlamydia screening programs are conducted in family planning and prenatal clinics. As shown in Table 2, public-setting cost estimates were derived from published literature 15,16 and adjusted to 1997 dollars. Private-setting cost estimates were derived from the HealthCare Consultants’ 2000 Physician Fee and Coding Guide. 17 The following Current Procedural Terminology (CPT) codes were used and adjusted to 1997 dollars: CPT 87490 (DNA probe), 87110 (cell culture), 87491 (urine LCR/PCR), 87299 (DFA), and 99213 and 99214 (office visit). Among students who reported seeking off-site diagnostic evaluation within 12 months before the school-based screening, 37.8% of females and 44% of males had their visit at a public setting, and 62.2% of females and 56% of males had their visit at a private setting.
On the basis of the data listed in Table 2 and the percentage of females and males who had diagnostic evaluations in each setting, we first calculated the weighted average cost per off-site chlamydia test as $11.90 in the public settings and $61.70 in the private settings. For off-site screening, we estimated the cost per student tested as $30.30 and the cost per student tested and treated as $47.50. For off-site diagnostic evaluations, we estimated the cost per student tested as $94.10 for females and $87.70 for males and the cost per student tested and treated as $129.30 for females and $121.10 for males. By incorporating the above unit costs into the decision tree, we calculated the total costs associated with testing and treating chlamydia infections in the non-school-screening scenario.
Costs Associated With Treating PID and Its Sequelae
Several studies have previously calculated the medical cost of PID and its sequelae. In an early study, Haddix et al 16 estimated the medical cost of treating a case of PID and its sequelae (including costs for chronic pelvic pain, ectopic pregnancy, and infertility) to be $4575 (in 1993 dollars; $5388 in 1997 dollars). In a later study on adolescent PID patients, Shafer et al 18 found the cost of treating a case of PID only (not its sequelae) to be $2977 (in 1996 dollars; $3154 in 1997 dollars). Rein et al 19 recently estimated that the cost of treating a case of PID and its sequelae to be $1097 (in 1997 dollars), a number much lower than earlier estimates. However, the authors point out that their study was limited by at least five factors that tend to underestimate the costs of PID and its sequelae. For example, the cost of infertility diagnosis and treatment ($4625) was estimated with use of older estimates of infertility costs and older probabilities of treatment. The current cost of infertility can range from two to 20 times the estimate that they used in their study. 19 Thus, the $1097 estimate probably represents a minimum estimate of PID-related costs. Because our study evaluated PID costs for a population similar to that of Shafer et al, 18 we used Shafer's estimate of $3154 in our analysis.
To test whether the results of our base-case analysis were dependent on the accuracy of the probability and cost estimates that we used, we conducted several sensitivity analyses. First we conducted univariate sensitivity analyses, in which we varied the estimates of five key variables over a wide range we considered plausible. We varied the test sensitivity of the urine LCR test from 82% to 96%; the percentage of female partners who were not screened from 25% to 75%; the probability of a female acquiring infection from an infected male from 50% to 70%; the probability of a female with untreated/uncured chlamydia infection developing PID from 20% to 40%; and the cost of treating a case of PID and its sequelae from $1097 to $5388. We examined one variable at a time to assess whether these alterations would affect our results while holding other variable estimates at the same level as in the base-case analysis. We then conducted a multivariate sensitivity analysis by examining all five variables simultaneously to see whether the overall uncertainty of the five variables would affect our results. The results of the multivariate sensitivity analysis are presented for the best-case and worst-case scenario.
To test whether the New Orleans screening program would be cost-effective in other locations, we also conducted a univariate sensitivity analysis to examine the sensitivity of the results to the variation of prevalence. We used ±50% of base-case value to vary the female chlamydia prevalence from 6.2% to 18.6% and the male chlamydia prevalence from 3.05% to 9.15%.
Table 4 displays the results of the base-case analysis. Under base-case assumptions, at an intervention cost of $86,449, the school-based screening program detected and successfully treated a total of 159.8 cases of chlamydia infection (102.7 cases among female students, 57.1 cases among male students) and prevented 24 cases of chlamydia infection among female partners of male students. In comparison with non-school-based screening, the program prevented a total of 38 cases of PID, as well as $119,866 of medical costs for treating PID and its sequelae. Replacing non-school-based screening with the school-based screening resulted in a net savings of $1524 per case of PID prevented.
Table 5 displays the results from the sensitivity analyses. From the univariate sensitivity analysis on each of the five key variables, we found the results remain cost-saving over a reasonable range of four variable estimates: the sensitivity of the urine LCR test, the percentage of female partners of infected male students who were not tested, the probability of chlamydia transmission from an infected male to a female, and the probability of a female with untreated or uncured chlamydia infection developing PID. The cost-effectiveness result is sensitive only to the estimated cost of treating PID and its sequelae. When the variable for the cost of treating a case of PID and its sequelae is lowered to $1097, our analyses indicate that it would cost $540 to prevent a case of PID. From the multivariate sensitivity analysis we found that in the worst-case scenario, it costs $1869 to prevent a case of PID and its sequelae.
The result of the univariate sensitivity analysis of the chlamydia prevalence demonstrated that settings with a higher chlamydia prevalence generate a more cost-effective program. When chlamydia prevalence varies from 6.2% to 18.6% among females and from 3.05% to 9.15% among males, the cost-effectiveness ratio ranges from a saving of $303 to $1975 per case of PID prevented.
From an individual health perspective, the New Orleans screening program helped identify cases of chlamydia infection and prevent PID and its sequelae. From a health care system perspective, the program reduced the chlamydia prevalence, as well as the occurrence of PID and other sequelae in the population, and resulted in a net cost-savings to the health care system. School-based STD screening programs of this type warrant careful consideration by policy-makers and program planners as they allocate resources. From these results, federal- and state-level decision-makers should infer that school-based STD screening programs such as this one are cost-effective uses of public funds. Community planning groups should consider these results as evidence that school-based STD screening programs are likely to be effective uses of available resources.
Although most school-based STD services are delivered in a school-based or school-linked health centers, the majority of schools do not have school-based health centers. This study evaluated a school-based STD screening and treatment program that is independent of a clinical setting. Since students attending schools without clinics can receive these services, a school-based STD screening and treatment program provides increased access for all students. In addition, this type of program incurs lower programmatic costs since it has no clinic-associated costs. Health-screening programs conducted outside the clinic setting have disadvantages of limited service availability and potential confidentiality difficulties. These potential obstacles can be overcome through functional linkages to community resources and the support of school administration and staff.
This study has some clear limitations. First, it was based on a decision analysis that used estimates from various sources, and the generalizability of these estimates is unknown. For example, the case-weighted cost estimates of treating a case of PID and its sequelae found in the research of Haddix et al, 15 Shafer et al, 17 and Rein et al 18 are very different. To resolve this problem, we conducted our sensitivity analyses over a reasonable range of five key variable estimates. Second, the study was retrospective, so costs were estimated rather than prospectively measured. Third, the number of expected chlamydia infections in the non-school-screening scenario was modeled rather than directly measured. Fourth, we did not consider reinfection in either scenario because the complicity related to reinfection was beyond the scope of this study. Fifth, we only considered the benefit of preventing chlamydia sequelae, thus excluding the benefit of preventing gonorrhea sequelae. Sixth, we excluded some intangible benefits of the screening program of treated participants, such as decreased susceptibility to HIV infection and decreased transmission to newborns. Inclusion of these additional benefits would increase the estimated cost-effectiveness of the screening program.
Even with these limitations, we have been cautious in our approach and have carefully examined the robustness of our results. The results of our sensitivity analysis indicate that our findings are generally robust with respect to most of the key sources of uncertainty in the analysis, although they are sensitive to changes in the cost value associated with treating PID and its sequelae. However, as long as the cost of treating PID and its sequelae was over $2,966, even if the other four variables were held at the bottom of their estimated range, the program remained cost-saving. Because none of the three studies that previously reported cost estimates of treating a case of PID and its sequelae included the indirect and intangible costs associated with PID, such as lost time from school, pain and suffering, and susceptibility to HIV, the likelihood of the actual cost being less than $2966 is small. Therefore, it is justifiable to conclude that the STD screening program implemented in the five New Orleans schools was cost-effective and cost-saving under most scenarios considered.
Furthermore, our sensitivity analysis of chlamydia prevalence indicates that the higher the prevalence, the more cost-effective the program is. We found that the program remains cost-effective and cost-saving over a range of prevalence estimates (6.2% to 18.6% among females, 3.05% to 9.15% among males). Considering that the high rates of asymptomatic chlamydia infection among adolescents are likely to fall within such ranges, the program has great potential to be cost-effective in other school settings.
To our knowledge, this is the first cost-effectiveness analysis of a school-based STD screening program. The model developed in this study helped identify what critical data are lacking and what research is needed to assist informed decision-making. More research is needed on the probability of PID developing from an untreated chlamydia infection, the cost of treating a case of PID and its sequelae, adolescent health-care seeking behavior, and community-provider STD-related practices. In addition, future school-based STD screening programs should collect program cost data to facilitate more cost-effectiveness evaluations.
Schools remain the ideal setting for delivering needed health care services to adolescents who would not otherwise receive them, especially adolescents with asymptomatic STDs. The findings of this study suggest that a school-based chlamydia screening and treatment program can be a cost-effective chlamydia control strategy by preventing the development of sequelae and saving future health care costs. Therefore, we recommend that policy-makers routinely consider incorporating school-based STD screening programs as part of an STD control strategy to efficiently reduce the burden of adolescent STDs.
The following equations were used to calculate the true prevalence rates among female students.
- a = 89% × (a +c)
- d = 99% × (b +d)
- a +b +c +d = 1402
- a +b = 167
a = number of female students having true positive results
b = number of female students having false positive results
c = number of female students having false negative results
d = number of female students having true negative results
89% = test sensitivity
89% = test specificity
1402 = total number of female students tested
167 = total number of female students having positive results
As a result,
a = 154.8, b = 12.2, c = 19, d = 1216, and
true prevalence rate = (154.8 + 19) ÷ 1402 = 12.4%
Using the same approach, we estimated the true prevalence rate among male students to be 6.1%.
1. Mertz KJ, McQuillan GM, Levine WC, et al. A pilot study of the prevalence of chlamydia infection in a national household survey. Sex Transm Dis 1998; 25: 225–228.
2. Mertz KJ, Levine WC, Mosure DJ, Berman SM, Dorian KJ. Trends in the prevalence of chlamydia infections: the impact of community-wide testing. Sex Transm Dis 1997; 24: 169–175.
3. Burstein GR, Lowry R, Klein JD, Santelli JS. Primary care preventive care visits and sexually transmitted disease and pregnancy prevention services received by U.S. high school students [abstract]. J Adolesc Health 2001; 28: 127.
4. Burstein GR, Snyder MH, Conley D, Boekeloo BO, Quinn TC, Zenilman JM. Adolescent chlamydia testing practices and diagnosed infections in a large managed care organization. Sex Transm Dis 2001; 28: 477–483.
5. Cook RL, Wiesenfeld HC, Ashton MR, Krohn MA, Zamborsky T, Schoole SH. Barriers to screening sexually active adolescent women for chlamydia: a survey of primary care physicians. J Adolesc Health 2001; 28: 211–221.
6. Mangione-Smith R, McGlynn EA, Hiatt L. Screening for chlamydia in adolescents and young women. Arch Pediatr Adolesc Med 2000; 154: 1108–1113.
7. Leavy SM, Smith MJ, Allensworth DD, Farquhar BK, Kann L, Pateman BC. School health services. J Sch Health 1995; 65: 319–326.
9. Cohen DA, Nsuami M, Martin DH, Farley TA. Repeated school-based screening for sexually transmitted diseases: a feasible strategy for reaching adolescents. Pediatrics 1999; 104: 1281–1285.
10. Centers for Disease Control and Prevention. Guidelines for treatment of sexually transmitted diseases. MMWR Morb Mortal Wkly Rep 1998; 47 (RR-1):53–59.
11. Gold MR, Siegel JE, Russell LB, Weinstein MC. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press, 1996.
12. Centers for Disease Control and Prevention. 1999 National youth risk surveillance data. Available at: http://www.cdc.gov/nccdphp/dash
. Accessed December 4, 2000.
13. Jones RB, Wasserheit JN. Introduction to the biology and natural history of sexually transmitted diseases. In: Judith NW, Sevgi OA, King KH, Penelope JH, eds. Research Issues in Human Behavior and Sexually Transmitted Disease in the AIDS Era. Washington, DC: American Society for Microbiology, 1991: 11–37.
14. Howell MR, Quinn TC, Gaydos CA. Screening for Chlamydia trachomatis
in asymptomatic women attending family planning clinics. Ann Intern Med 1998; 128: 277–284.
15. Howell MR, Quinn TC, Brathwaite W, Gaydos CA. Screening for Chlamydia trachomatis
in family planning clinics. Sex Transm Dis 1998; 25: 108–117.
16. Haddix AC, Hillis SD, Kassler WJ. The cost-effectiveness of azithromycin for Chlamydia trachomatis
infections in women. Sex Transm Dis 1995; 22: 274–280.
17. HealthCare Consultants of America. HealthCare Consultants’ 2000 Physician Fee and Coding Guide. Augusta, Georgia: HealthCare Consultants of America, Inc., 2000.
18. 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.
19. Rein DB, Kassler WJ, Irwin KL, Rabiee L. Direct medical cost of pelvic inflammatory disease and its sequelae: decreasing, but still substantial. Obstet Gynecol 2000; 95: 397–402.