The sequelae of chlamydial infection cost more than $2.7 billion each year [4]. Screening and timely treatment can decrease the incidence of sequelae [1, 5] and the associated medical costs [6]. Universal screening offers the potential to identify and treat as many chlamydial infections as possible, thus decreasing the incidence of future illness and the economic consequences of sequelae [7]. Alternatively, a selective screening program based on the CDC criteria or age could target the portion of the population at highest risk and identify most chlamydial infections without testing all women [8].

To address the question of who should be screened, we applied three screening strategies in a medical cost and outcome decision model to the predominantly asymptomatic population seen at family planning clinics in Baltimore, Maryland. The three strategies were 1) screening women according to the CDC criteria, 2) screening women younger than 30 years of age, and 3) screening all women. We compared each strategy's ability to identify women at risk for chlamydial infection and thus allow the initiation of testing and treatment, decrease the incidence of sequelae, and decrease overall medical costs.

Methods

Top Methods Results Discussion Author & Article Info References

Study Sample and Setting

We considered 8654 women who attended Baltimore City Health Department (BCHD) family planning clinics between April 1994 and August 1996. As part of the Region III Infertility Prevention Project, these women were tested for Chlamydia trachomatis by polymerase chain reaction (PCR) assays (Roche Diagnostic Systems, Branchburg, New Jersey) that were processed at the Johns Hopkins University Chlamydia Research Laboratory. Women who had at least one clinical sign of chlamydial infection on pelvic examination (mucopus, cervicitis, cervical ectopy, cervical motion tenderness, or friability on cervical contact) and women who presented as the sex partner of a person with chlamydial infection were offered presumptive treatment and were excluded from the study. Women who attended the clinic for reasons that necessitated a pelvic examination (for example, for an annual Papanicolaou test) had PCR testing of cervical specimens. Women who did not need a pelvic examination had PCR testing of urine specimens. Informed consent was obtained from all patients who were attending BCHD family planning clinics to receive standard care. This study was approved by the institutional review board of Johns Hopkins University.

Decision Analysis

Statistical analysis was done by using Intercooled Stata 4.0 (Stata Corp., College Station, Texas). A decision analytic model (Smltree 2.9, Jim Hollenberg, Roslyn, New York) was developed to analyze a series of possible events associated with each strategy (Table 1). Medical outcomes were defined as prevented cases of pelvic inflammatory disease, chronic pelvic pain, ectopic pregnancy, infertility, male urethritis, epididymitis, infantile conjunctivitis, and infantile pneumonia. Economic outcomes were defined as all screening-related program costs and direct medical cost savings of all sequelae (Table 1). Indirect costs were considered in sensitivity analyses. For descriptions of the equations and calculations used in the model, see Appendix.

Overall economic and medical outcomes were compared from the perspective of a health care system to determine which strategy prevented the most disease and had the lowest total costs. Incremental cost-effectiveness ratios (that is, estimates of the additional cost that one strategy requires to prevent one more case of pelvic inflammatory disease than the next most effective strategy) were compared. The reference case estimates were varied in univariate and multivariate sensitivity analyses to determine how changes in estimated values affected the results and to identify break-even prevalence points. Costs were adjusted to 1995 U.S. dollars by using the medical care component of the 1996 consumer price index. Future costs were discounted at an annual rate of 3% [30].

Measurement of Estimated Values

Clinic personnel collected demographic, behavioral, and clinical presentation data by using the Region III Infertility Prevention Program Prevalence Study survey. The administrative time associated with the collection of a patient's data was not included as a program cost because the information would have been collected even in the absence of a screening program. The clinician time associated with a full pelvic examination (28 minutes) was estimated by the direct observation of activities in a clinic patient-flow analysis (BCHD. Unpublished data). Begley and colleagues [9] previously estimated the cost of materials for a pelvic examination. The clinician time required for the collection and administrative processing of an additional cervical swab for screening purposes (1.5 minutes) was based on the manufacturer-recommended protocol (Roche Diagnostic Systems). The cost of specimen collection associated with each screening strategy was weighted by the percentage of women who received a pelvic examination for a medical reason other than screening for chlamydial infection.

This analysis considered the question of testing cervical specimens by PCR. The sensitivity of PCR (87.6%) was determined by modified meta-analysis of the peer-reviewed medical literature [10-17]. A specificity of 99% was used [12]. The cost of the PCR assay for screening purposes was provided by the Maryland State Health Laboratory and was based on costs of materials, manufacturer quotes for public health testing, and the annualized cost of a thermocycler. The cost of laboratory technician time per specimen was obtained by time-in-motion studies of the laboratory technicians in the Johns Hopkins University Chlamydia Research Laboratory.

The follow-up cost of scheduling a return visit for treatment after a positive test result was previously estimated by Begley and colleagues [9]. In the BCHD family planning clinics, approximately 80% of women who are notified of a positive test result return for treatment (BCHD. Unpublished data).

The cost of the treatment visit, including administration time, clinician and counseling time, and distribution of doxycycline, was based on clinic records and direct observation of clinician activities (BCHD. Unpublished data). The effectiveness of doxycycline was previously estimated to be 85.7% [18]. Doxycycline causes minor to major side effects in 16% of persons [18], and the cost of treating these side effects was considered a cost of the screening program.

An estimated 10% to 40% of women with untreated chlamydial infections develop pelvic inflammatory disease [5, 19-21]. We used an overall probability of pelvic inflammatory disease of 30% and assumed that 40% of these cases would be symptomatic in year 1 [22]. Our reference case-weighted cost of sequelae, which considered the risks for and costs of future sequelae in men, women, and infants, was derived from the research of Washington, Marrazzo, and Haddix and coworkers [4, 22, 23].

Role of the Funding Source

The funding source for this study did not play a role in gathering, analyzing, or interpreting the data or in deciding whether to submit the manuscript for publication.

Results

Top Methods Results Discussion Author & Article Info References

Reference Case Analysis

From April 1994 through August 1996, the BCHD family planning clinics provided services to 8354 women. Approximately 7.6% of these women (n = 638) either had clinical evidence of a chlamydial infection (n = 558) or presented because an infected sex partner had informed them of their possible exposure to chlamydia and had advised them to seek medical care (n = 80). These women were excluded from the study. Of the remaining 7716 women, 17 reported never having had vaginal intercourse. None of these women tested positive for chlamydial infection, and all were dropped from the analysis. Of the remaining 7699 women, most were African American and had no medical insurance (Table 2). The median age was 25 years (range, 11 to 68 years). Many had previously had a diagnosis of a sexually transmitted disease, but few reported having previously had a diagnosis of syphilis. Few women reported having more than one sex partner or having a new sex partner in the 90 days before screening. However, more than half reported inconsistent use of condoms.

A total of 5173 women received a pelvic examination for routine reproductive care (that is, they would have received a pelvic examination even in the absence of a chlamydial screening program). These women were tested for chlamydial infection by PCR of cervical specimens; this resulted in an observed prevalence of chlamydial infection of 7.3% (376 of 5173 women). For the purposes of the Region III Prevalence Study, the women without a medical reason for a pelvic examination (for example, women who needed to have a prescription for oral contraceptives refilled) (n = 2526) had PCR screening of urine specimens that resulted in an observed prevalence of chlamydial infection of 5.2% (132 of 2526 women). The overall observed prevalence was therefore 6.6% (508 of 7699 women).

In this family planning setting, a screening program based on the CDC criteria necessitated testing 36.5% of the population. This strategy had the potential to identify 70.4% of the infected women but required unnecessary testing for 34.2% of the uninfected women. The age-based screening strategy required testing 71.4% of the population. It had the potential to identify 93.3% of the infected women and required unnecessary testing for 69.8% of the uninfected women. Universal screening, by definition, required testing 100% of the women. It had the potential to identify 100% of the infected women and required unnecessary testing for 100% of the uninfected women.

Without screening, 152 cases of pelvic inflammatory disease, 27 episodes of chronic pelvic pain, 18 infertility workups, and 12 ectopic pregnancies would have occurred in the BCHD family planning clinic population. Additional sequelae in men and infants would have comprised 138 cases of urethritis, 7 cases of epididymitis, 6 cases of infantile conjunctivitis, and 3 cases of infantile pneumonia. Costs of sequelae would have totaled $676 000 (Table 3). A screening program based on the CDC criteria would have prevented 64 cases of pelvic inflammatory disease and would have saved $231 000 in net costs because of the medical cost savings provided by all prevented sequelae in women, men, and infants. By removing the requirement of reported high-risk sexual behavior from the screening criteria and instead screening all women younger than 30 years of age, 21 more cases of pelvic inflammatory disease (and numerous other sequelae) would have been prevented, providing an additional $74 000 in net savings compared with screening based on the CDC criteria. Universal screening would have prevented an additional 6 cases of pelvic inflammatory disease but would have incurred $19 000 more in net costs compared with age-based screening. Thus, universal screening cost $3000 per case of pelvic inflammatory disease prevented in addition to cases prevented by age-based screening.

Sensitivity Analyses

The differences in cost-effectiveness between age-based screening and universal screening were sensitive to changes in several estimates. In the reference case, age-based screening had the potential to identify 93.3% of all infected women. If age-based screening identified less than 88.5% of the infected women, universal screening would provide the greatest cost-savings. In addition, changes in the prevalence of chlamydial infection affected the comparative cost-saving abilities of age-based and universal screening. If the prevalence of chlamydial infection were to increase from 6.6% to 10.2%, universal screening would become the cost-saving strategy (assuming an age distribution similar to the one in our study). If the prevalence were to be lower, age-based screening would be cost saving unless the prevalence was less than 1.1% (the break-even prevalence at which screening done using any criteria did not provide cost savings) (Figure 1). The break-even prevalence at which universal screening became cost saving depended on the ability of age-based screening to identify women with chlamydial infection (Figure 2). For example, if the prevalence of chlamydial infection was 4.5%, it was only when age-based screening detected less than 82.2% of the infected women that universal screening provided the greatest cost savings.

View larger version (13K): [in this window] [in a new window]   Figure 1. Cost-effectiveness of age-based screening and universal screening at various prevalences of chlamydial infection. Curve = (universal screening $) –(age-based screening $)/(universal testing pelvic inflammatory disease) –(age-based screening pelvic inflammatory disease). As the prevalence of chlamydial infection increases, the age-based screening strategy becomes less cost-saving relative to universal screening. At prevalences greater than 10.2%, universal screening prevents more disease and provides more cost savings than age-based screening. The break-even prevalence for all screening is 1.1%; this is the point at which screening by any criteria ceases to provide cost savings. PID = pelvic inflammatory disease.

View larger version (11K): [in this window] [in a new window]   Figure 2. Screening strategy recommendations based on different prevalences of chlamydial infection and different sensitivities of age-based screening. The line represents the break-even values for prevalence of chlamydial infection and sensitivity of age-based screening, where age-based screening and universal screening provide the same cost savings. For example, at prevalences greater than 4.5%, where age-based screening fails to identify more than 82.2% of infections occurring in the population, universal screening provides greater cost savings than age-based screening.

The reference case model estimated that 80% of all women who were notified of a positive test result would return to the clinic for treatment. Changes in the return rate did not affect the results unless the rate was less than 14.5%, the point at which screening according to the CDC criteria provided the greatest cost savings. If less than 13.7% of women returned for treatment, then too few women received treatment to justify screening under any of the three strategies. At rates of return of 14.5% to 100%, age-based screening provided the greatest cost savings.

For clinicians without a public health or clinical laboratory for processing specimens, the diagnostic costs would be higher than those modeled here. To model the use of a commercial laboratory, we substituted an assay cost of $28, which represents the Maryland Medicare-Medicaid reimbursement rate. At this cost, age-based screening still provided the greatest cost savings. However, universal screening provided the lowest cost savings of the three strategies examined: It cost an additional $8100 per case of pelvic inflammatory disease prevented in addition to those prevented by age-based screening and $800 per case of pelvic inflammatory disease prevented in addition to those prevented by CDC criteria-based screening. Even if the assay costs were to increase to more than $50, the prevalence of chlamydial infection would only need to be more than 5.2% to justify age-based screening.

Application of ligase chain reaction or PCR for use with urine specimens decreases the cost of specimen collection by pelvic examination from $16.60 to $0.09 in women who lack medical indications necessitating a pelvic examination. In the BCHD family planning population, urine-based screening with a diagnostic test sensitivity of 89% [15, 16] decreased the total costs associated with universal screening from $391 000 to $339 000 and decreased the total costs associated with age-based screening from $372 000 to $354 000. Thus, if DNA amplification of urine specimens were to be used, universal screening would provide the greatest cost savings.

The indirect costs of productivity lost because of sequelae were not explicitly modeled in the reference case. Including these costs would have increased the weighted cost of sequelae to $5166. This increase would not have affected the conclusions of the model. If the weighted cost of pelvic inflammatory disease and sequelae were to increase to more than $7372 or if the probability of the development of pelvic inflammatory disease with an untreated chlamydial infection were to increase to more than 50.7% (that is, if the rate of symptomatic pelvic inflammatory disease increased from 12% to 21%), universal screening would provide the greatest cost savings. Magid and colleagues [18] estimated the cost of pelvic inflammatory disease (according to Colorado Blue Cross and Blue Shield payments and allowed charges) as $1917. Even at this cost, age-based screening provided the greatest cost savings. Universal screening at this low cost resulted in an additional $500 per case of pelvic inflammatory disease prevented in addition to those prevented by CDC criteria-based screening and an additional $5600 per case of pelvic inflammatory disease prevented in addition to those prevented by age-based screening. If the weighted sequelae costs were to decrease to less than $790, screening by any strategy would not have provided cost savings.

Individual changes in cost of treatment, cost of follow-up, cost of side effects, cost of treatment visit, sensitivity of the assay, effectiveness of doxycycline, or probability of side effects associated with medication did not affect the results of the model.

Discussion

Top Methods Results Discussion Author & Article Info References

Among asymptomatic patients in a family planning population, screening by any means saved money and prevented disease. When total costs (program expenditures and future costs of all sequelae) were considered, age-based screening provided the greatest cost savings. Universal screening prevented the most disease but cost more than $3000 per case of pelvic inflammatory disease prevented in addition to those prevented by age-based screening. However, if the prevalence of chlamydial infection were greater than 10.2% in this asymptomatic group of women or if age-based screening failed to identify more than 88.5% of infected women, universal screening would provide the greatest cost savings. Screening for chlamydial infection by any criteria is one of the few health interventions that provides cost savings to society even when indirect costs are excluded [31].

Two studies, one conducted in San Francisco and one conducted in Northwest Public Health Region X, analyzed both symptomatic and asymptomatic populations with prevalences of chlamydial infection ranging from 6.6% to 9.2% [7, 22]. These studies assumed that selective screening based on age (20 and 25 years of age) would miss more than 20% of infected women and concluded that all women attending family planning clinics should be screened for C. trachomatis because of the relative insensitivity of the age-based screening criteria. To increase the sensitivity of the screening criteria, we used a higher cut-off (30 years of age) and found that age-based screening missed only 7% of infected women. Through sensitivity analysis, we found that if more than 11.5% of the women with chlamydial infection would not be identified by selective screening, universal screening would provide the greatest cost savings. Thus, our final recommendation agrees with those in other studies [7, 22]. However, the comparability of our study to those studies is limited because of differences in populations and assays used (direct fluorescent antibody or DNA amplification) to identify positive patients.

Our study has several potential limitations. First, it was based on a decision analysis that used estimates from several sources, and the generalizability of these estimates is unknown. For example, methods to estimate sequelae costs other than those used by Washington and Katz and Haddix and colleagues [4, 23] are available, such as that used by Magid and coworkers [18]. In addition, infrequently used therapies, and treatments could decrease the cost of treatment for sequelae [32]. Second, we may have underestimated the cost of sequelae by not modeling decreased pain and suffering and decreased susceptibility to HIV transmission [33]. To overcome these potential biases, all estimates were varied in sensitivity analyses to determine the effect that changes would have on the results. Third, we analyzed a predominantly inner-city minority population that may have prevalences and age distributions different from those of other populations. Chlamydial infection is common among women of all races, and we conducted sensitivity analyses that considered different age and prevalence structures.

Screening done on the basis of the CDC criteria depends on accurate reporting of risk behavior by the patient and young age (<20 years). As Zenilman and coworkers state [34], self-reported condom use may be subject to substantial reporting bias, limiting the value of self-reported sexual behavior as a reliable indicator of risk for infection. The CDC recommends screening all women younger than 20 years of age, but these women represented only 50% of the patients who were positive for chlamydial infection in our study. Screening only those women who met the CDC criteria limited the potential to identify women with chlamydial infection. Screening women younger than 30 years of age in the BCHD family planning population missed only 7% of all asymptomatic infections.

Our results have the potential for application beyond the public health family planning population to managed care populations, where the payer of the screening program is also the direct beneficiary because of the direct savings associated with reductions in pelvic inflammatory disease and other costly sequelae. Given that age-based screening for chlamydial infection was recently included in the Health Employer Data Information Set (HEDIS) used to evaluate managed care organizations, the results are particularly important for decision makers in this setting. In the public sector, the ability to screen for chlamydial infection depends on the availability of programmatic funds. Yet screening benefits society and the entire public health system through the decreased future medical costs associated with averted sequelae. Screening does not directly save money for sexually transmitted disease program offices. When public health allocation decisions about screening for chlamydial infection are made, the long-term benefits to the public health sector as a whole must be considered.

The sequelae that develop in infected women who are never identified and tested and therefore never receive treatment can be costly and can lead to severe reproductive health complications, high medical costs, and a sustained transmission cycle. The CDC recommends screening women for C. trachomatis, and the availability of better diagnostic tools has increased the utility of screening. Although age-based screening provides the greatest cost savings, screening by any criteria provides cost savings compared with no screening. Therefore, clinicians should begin to screen all sexually active young women to prevent future sequelae and to decrease the costs associated with chlamydial infection.

Appendix

Population characteristics were carried out to the first decimal place and rounded to the nearest 0.50. Medical outcomes were rounded to the nearest whole case unit. In text descriptions, cost results were rounded to the nearest thousand and incremental cost-effectiveness ratios were rounded to the nearest hundred. Unrounded costs and ratios are shown in Table 3.

For the meta-analysis, a literature search was conducted in Med2000+ of articles published between 1966 and 1996. Studies that 1) were identified with the keywords C. trachomatis, PCR or Amplicor, and female or women; 2) analyzed female cervical specimens; 3) used culture as a reference test; and 4) conducted discrepant analysis were included.

The following equations were used to determine cost estimates.

Weighted Cost of Specimen Collection

To calculate the cost of clinician time for each type of screening in women receiving routine pelvic examinations, we divided the number of women who would have received pelvic examinations even in the absence of a screening program by the number of women who would be screened and multiplied this number by the cost of clinician time for collection of an additional cervical swab. To calculate the cost of clinician time in women receiving a pelvic examination for screening purposes only, we divided the number of women who would have received a pelvic examination only for chlamydial screening by the number of women who would be screened under each strategy and multiplied this number by the cost of clinician time for a full pelvic examination plus the cost of materials required for a pelvic examination (including speculum, drape, and gloves) [7]. The weighted cost of specimen collection is the sum of these two clinician costs.

Costs of Diagnostic Assay

The technician time associated with the processing of each diagnostic specimen was calculated as the total time required to complete an assay batch divided by the number of diagnostic specimens in a complete batch and multiplied by the average hourly technician wage ($21.33), including benefits. Laboratory overhead was calculated as 69% of the laboratory technician's salary. The cost of a thermocycler was based on an initial cost of $12 000. This cost was annualized to arrive at a cost per year over 3 years and was divided by the number of specimens expected to be run in 1 year. The total diagnostic costs associated with each specimen were calculated as the sum of technician time; laboratory overhead; and assay costs, including the cost of a thermocycler.

Weighted Cost of Sequelae

The cost of sequelae was presented as a weighted cost of all sequelae based on published costs and disease probabilities. Of all women with chlamydial infection, an estimated 30% will develop pelvic inflammatory disease. Approximately 40% will be symptomatic, and 86% will require outpatient care at a cost of $269 in year 1 [4]. The remaining 14% will require inpatient care at a cost of $19 201 [4]. Of all women with pelvic inflammatory disease, 18% will require surgery for chronic pelvic pain in year 2 at a cost of $12 774 [23]. In year 5, 7.8% of women with silent pelvic inflammatory disease will have an ectopic pregnancy at a cost of $12 282 [4] and 12% will be infertile. Twenty-five percent of this 12% will have at least one infertility workup in year 10 at a cost of $5099 [4]. Assuming a minimum of 1 male sex partner per patient, 68% of male sex partners will be infected. Forty percent of these men will have urethritis at a cost of $13 [22]. In addition, 2% will develop epididymitis at a cost of $123 [22]. In our population, 5.7% of the infected population was pregnant. Approximately 20% of infants born to infected mothers will have chlamydial conjunctivitis at a cost of $100 [22], and 10% will have chlamydial pneumonia at a cost of $2501 [22]. The weighted cost of sequelae is the sum of the costs associated with each ailment multiplied by the percentage of persons who will be affected. Indirect costs were estimated at a rate of $80 per day lost [22].