Chlamydial infections of the genital tract do not invariably cause symptoms that would prompt a person to seek medical aid. Identification and treatment of infected persons is important not only for their own well-being but also to prevent the spread of C. trachomatis in society. We assess the cost-effectiveness of some commonly used diagnostic tests and treatment regimens, including the recently introduced DNA amplification assays and a single oral dose of azithromycin, for identifying and treating asymptomatic female carriers of C. trachomatis.

Methods

Top Methods Results Discussion Author & Article Info References

Eligible Patients

We considered a cohort of 1000 sexually active, nonpregnant women of reproductive age who had no clinical signs or symptoms of a genital tract infection. We assumed that these women attend youth, gynecology, and family planning clinics for either contraceptive advice or routine gynecologic care. We derived the assumptions on their demographic characteristics from the Swedish Women's Health Study [2], which focused on the same group of women considered in our analysis. These assumptions include the following: The mean age of the women is 26 years (range, 15 to 48 years); most are nonparous (68%), are either married or cohabiting (52%), have a steady partner (85%), and have had some training beyond the compulsory 9 years of education (88%); they have had a mean of 11 sex partners; and the prevalence of genital tract infection with C. trachomatis and Neisseria gonorrhoeae is 8.5% and 0.1%, respectively. We believe that these women are representative of the general Swedish female population who are of reproductive age and are at moderate risk for chlamydial infections of the genital tract.

Intervention

We compared a no-screening strategy with tissue cell culture, confirmed enzyme immunoassay, and DNA amplification assays that were based on either polymerase chain reaction (Amplicor C. trachomatis, Roche Diagnostic Systems, Elizabeth, New Jersey) or ligase chain reaction (LCR Assay C. trachomatis, Abbott Diagnostics, North Chicago, Illinois) for screening asymptomatic chlamydial infections of the genital tract. We assumed that all samples with positive or equivocal enzyme immunoassay results would be tested with a blocking assay [3]. Positive results that become blocked would be reported as confirmed positive, and equivocal results that become blocked would be further evaluated with a direct fluorescent assay. If chlamydial elementary bodies are found in the pellet obtained by centrifugation of the sample tested by enzyme immunoassay, the result would again be reported as confirmed positive.

In the screening strategies, endocervical samples are collected from women during a pelvic examination. Any woman with laboratory-confirmed chlamydial infection is referred to a disease intervention specialist, who attempts to confidentially notify infected women and arrange a visit to the clinic. Notification efforts include telephone calls, letters, and visits to the patient's residence. When the patient returns to the clinic, an antibiotic regimen is prescribed, and counseling regarding the disease and its treatment as well as notification of the patient's partner are done. Partner notification is mandatory by law in Sweden; the disease intervention specialist informs the index patient of the importance of revealing her sex partner or partners and interviews her extensively to obtain the names, addresses, and telephone numbers of her sex partners. The disease intervention specialist then contacts these partners to motivate them to be examined at the clinic. This procedure is repeated until the partner comes to the clinic for examination. At this visit, the same steps are followed as with the index patient: physical examination, partner notification, counseling, and prescription of an antibiotic regimen for empiric treatment of uncomplicated chlamydial genital infection.

In our analysis, we assumed that a 7-day, twice-daily course of doxycycline was prescribed to treat uncomplicated chlamydial genital tract infection in both women and their sex partners. We also evaluated the effect of administering a single oral 1-g dose of azithromycin under supervision in the clinic on the outcomes of the screening strategies. We assumed that only persons who seek medical help because of the presence of signs and symptoms of chlamydial diseases and their sequelae receive medical care in the no-screening strategy.

Decision-Analysis Model

We used two decision trees to graphically structure our decision-analysis model Figure 1 and Figure 2. The branches in the decision trees represent the strategies of screening and no screening. The square at the point of branching indicates a decision node: a point at which a decision must be made—for example, whether or not to screen a woman. The nodes represented by circles are used if subsequent outcomes occur by chance—for example, if the physician cannot control the results of the test, which therefore branch from a chance node. We used the Bayes theorem to assign probabilities within the ranges presented in Table 1 to the chance nodes of the decision trees. We varied the prevalence of C. trachomatis among women from 0% to 100% to determine the prevalence above which one strategy is preferred to another [break-even analysis] Figure 1, node 1).

View larger version (21K): [in this window] [in a new window]   Figure 1. Decision tree showing the outcomes of screening strategies for Chlamydia trachomatis among women and the outcomes of no screening. The rectangle indicates the decision nodes; circles, chance nodes; filled circles, end of a branch; and the arrow, untreated chlamydial infection.

View larger version (32K): [in this window] [in a new window]   Figure 2. Decision tree showing the outcomes of tracing and treating contacts of women with a diagnosis of chlamydial infection of the genital tract and the outcomes of no contact tracing. The probabilities of chance node 8 were derived from the probabilities of the corresponding outcomes of screening men, indicated with letters a to f in Figure 1. The rectangle indicates the decision nodes; circles, chance nodes; filled circles, end of a branch; and the arrow, untreated chlamydial infection.

We derived the estimates on the sensitivity and specificity of the laboratory tests from multicenter clinical evaluations in which a positive diagnosis was made when tissue cell culture was positive or when the results of one nonculture test (enzyme immunoassay, direct fluorescence assay, polymerase chain reaction assay, or ligase chain reaction assay) verified those of another. The estimates on follow-up rates for women with positive results and their partners were derived from data obtained from medical centers in North America and Sweden, which do follow-up procedures similar to those considered in our analysis.

We obtained data on the prevalence of infection in male partners of women infected with C. trachomatis from studies in which tissue cell culture or the polymerase chain reaction assay was used for diagnosis. Compliance with a 7-day, twice-daily course of doxycycline is nearly 100% among Swedish patients but may be as low as 50% in other countries. We assumed that administering a single dose of azithromycin ensures full compliance and that noncompliance is synonymous with treatment failure. We also assumed that each woman would disclose one or two male sex partners [13]. Case-finding and partner notification yielded six groups of male partners Figure 2, node 8]. The probabilities for these outcomes were obtained from the probabilities of the corresponding outcomes of screening in women (Figure 1). In the no-screening strategy, the partners can be divided into two groups: contacts of infected women and contacts of healthy women. The probability of a man having an infected female partner is the same as that of a woman being infected Figure 2, node 9).

Costs

Table 2 presents the direct and indirect cost estimates associated with medical services used in our analysis. Direct costs included salaries of health care personnel and costs of hospitals, drugs, equipment, and so forth; indirect costs were lost wages and lost value of household management due to participation in a health care program or sickness. We obtained these cost estimates from reported calculations that were based on average salaries and costs of medical care in Sweden [15, 21]. Using upper and lower limits for input salaries and costs, we made these calculations yield the widest range of monetary value for each cost category. The ranges of direct costs were extended in either direction by 20% to include any regional differences in medical care costs. The cost of sample collection, which included the cost of the sample collection materials, wages of the personnel collecting the endocervical swabs, and the cost of storing and transporting the specimens to the laboratory, were added to the cost of doing a diagnostic test.

We included no cost for the woman's initial clinic visit because the test was not the primary reason for consultation. We assumed that the cost of informing a patient of her test result and arranging a visit to the clinic included the salary of a disease intervention specialist plus the cost of equipment and the setting. Because patient follow-up may be difficult and time-consuming in certain settings, we estimated that a disease intervention specialist spends an average of 30 minutes for the follow-up of each patient. The cost of medical care for uncomplicated genital chlamydial infection includes those of a physician's appointment, antibiotic treatment, a 1-week course of clotrimazole for the 20% of female patients who develop moniliasis from antibiotic therapy, and partner notification.

We obtained the average cost of an untreated chlamydial genital infection from treatment and management of the following: symptomatic cervicitis, pelvic inflammatory disease, ectopic pregnancy, infertility, and chronic pelvic pain in infected women; diseases in the infected neonates of these women, such as conjunctivitis and pneumonia; and symptomatic urethritis and acute epididymitis in the male partners of these women. Assuming that ectopic pregnancy, infertility, and chronic abdominal pain are experienced 5 to 10 years after acquiring chlamydial infection, the direct and indirect future costs associated with these disorders were discounted at a rate of 5% to 10% per year to their present values.

Computer Program

We customized a standard spreadsheet computer program (Microsoft Excel, version 4.0, Microsoft Corp., Redmond, Washington) compatible with an Apple Macintosh computer (Apple Computer, Inc., Cupertino, California) to randomly choose input values within defined ranges and use them in the Equation to compute the outcomes of the decision-analysis model. This was repeated 1000 times; each time, a new combination of input values was used. The outcomes are expressed as 95% intervals on the means of results from all computations.

Results

Top Methods Results Discussion Author & Article Info References

Regardless of the prevalence of chlamydial infection, screening with DNA amplification assay combined with treatment of positive patients using the doxycycline regimen achieved a cure rate of 61.2% to 62.6% for women, 47.7% to 48.9% for their partners, and 54.8% to 56.2% overall. With no screening, the spontaneous cure rate was 7.4% to 7.6% for all categories. The cure rates associated with the enzyme immunoassay screening strategy were less than those associated with DNA amplification assay screening—48.8% to 50.0% for women, 38.4% to 39.4% for their partners, and 43.9% to 44.9% overall. The corresponding values for tissue cell culture screening were 44.7% to 45.7%, 36.4% to 37.2%, and 40.7% to 41.7%, respectively, which were lower than those associated with the other two screening strategies.

If we assume that infected women and their sex partners were treated with the 7-day, twice-daily course of doxycycline, screening with the DNA amplification assay and enzyme immunoassay significantly reduced the cost per cured woman compared with no screening when the prevalence of genital chlamydial infection in women exceeded 9% and 14%, respectively (Figure 3, panel A). Screening with tissue cell culture added costs compared with the other two screening strategies, regardless of the prevalence of infection. The cost associated with tissue cell culture screening was less than that of no screening only when the prevalence of the infection was greater than 40%. When the prevalence was less than 4%, there was a trade-off between the costs and benefits of the DNA amplification assay and enzyme immunoassay screening strategies: The cost per cured woman associated with the DNA amplification assay screening exceeded that of the enzyme immunoassay screening Figure 3, panel A), but the DNA amplification screening also resulted in an overall 10.1% to 10.2% better cure rate. A similar trade-off was seen between DNA amplification assay screening and no screening when the prevalence of chlamydial genital infection was less than 9% and between enzyme immunoassay screening and no screening when the prevalence was less than 14%.

View larger version (21K): [in this window] [in a new window]   Figure 3. Effect of prevalence on the incremental cost per cured woman when switching from no screening to screening with tissue cell culture, enzyme immunoassay, or DNA amplification assays for Chlamydia trachomatis among a cohort of 1000 women. Persons with positive test results received either a 7-day, twice-daily course of doxycycline (panel A) or a single oral dose of azithromycin (panel B). Incremental cost per cured woman equals (total costs of screening minus total costs of no screening)/(number of cured women as a result of screening minus number of women cured spontaneously).

Compared with the aforementioned doxycycline regimen, a single oral dose of azithromycin administered under supervision in the clinic ensured full compliance and improved the overall cure rates of the tissue cell culture, enzyme immunoassay, and DNA amplification assay screening strategies by 11%, 13%, and 15%, respectively. This led to a significantly decreased cost per cured woman associated with each screening strategy Figure 3, panel B). Treatment of positive patients with azithromycin rather than doxycycline reduced the threshold prevalence to 6% for both DNA amplification assay and enzyme immunoassay. At a prevalence of less than 6%, the enzyme immunoassay was also more cost-effective than the DNA amplification assay. When combined with azithromycin treatment, screening with tissue cell culture was cost-effective at a prevalence greater than 14%.

Discussion

Top Methods Results Discussion Author & Article Info References

Our decision analysis suggests that when the prevalence of chlamydial infection exceeds 6%, the DNA amplification assay screening of women and treatment of positive patients with a single oral dose of azithromycin administered under supervision at the clinic is the most cost-effective intervention strategy. Also, at a prevalence greater than 6%, screening with an enzyme immunoassay generates savings and improves the cure rates compared with no screening, but such screening is less cost-effective than screening with DNA amplification assays. Compared with no intervention, tissue cell culture is cost-effective only at a prevalence greater than 14%, which is rarely encountered among the female population considered in our analysis.

At a prevalence of less than 6%, there is a trade-off between the costs and benefits of alternative intervention strategies: First, screening with either DNA amplification assay or enzyme immunoassay adds cost but improves the quality of health care compared with no screening; second, screening with DNA amplification assays results in a greater cost per cured woman but achieves significantly better cure rates than does enzyme immunoassay screening. Therefore, the decision to choose one of these strategies depends on the physician's view of the value of preventing an illness caused by untreated chlamydial infection.

Our conclusions regarding screening with enzyme immunoassay and tissue cell culture for chlamydial infection of the genital tract in asymptomatic women are similar to those of several other studies in which the compliance rate with the standard doxycycline regimen was assumed to be 100% (the same compliance level we assumed for the azithromycin regimen in our study) [19, 22-24]. In these studies, screening with enzyme immunoassay was found to be more cost-effective than tissue cell culture [22, 23]. Compared with no screening, screening with enzyme immunoassay was found to generate savings when the prevalence of genital chlamydial infection in women ranged from 2% to 7% [18, 22, 24]

As suggested by our analysis, treatment of uncomplicated chlamydial infection of the genital tract with a single dose of azithromycin administered under supervision in the clinic is warranted. This procedure overcomes the problem of compliance with the standard 7-day, twice-daily regimen of doxycycline. It therefore improves the cure rates of the intervention strategies and generates savings by preventing chlamydial diseases and their sequelae in otherwise noncompliant patients. In another cost-effectiveness analysis in which 80% of the treated patients were estimated to be compliant with the doxycycline regimen, the cost per prevented case of pelvic inflammatory disease was found to be lower when the single-dose azithromycin regimen was used to treat women with either laboratory-confirmed or suspected chlamydial infection of the genital tract than when the standard doxycycline treatment was used [16].

In settings in which concomitant infections with C. trachomatis and N. gonorrhoeae are common, empiric treatment of gonorrhea is also indicated for patients with chlamydial infection of the genital tract. Washington and associates [25] found that for managing pelvic inflammatory disease, a combined treatment for both organisms was twice as cost-effective as the treatment for C. trachomatis alone. Full compliance and effective treatment of uncomplicated gonococcal genital infection can be ensured by an intramuscular injection of 250 mg of ceftriaxone before the patient leaves the clinic. We did not include any treatment for gonorrhea in our analysis because this infection is extremely rare among Swedish women (State Bacteriology Laboratory, Stockholm, Sweden; 1993).

When the Abbott ligase chain reaction assay was used to screen asymptomatic female carriers of C. trachomatis, first-void urine samples were found to be just as useful as endocervical specimens [5]. One drawback of the use of first-void urine is that other disorders that might be diagnosed during a pelvic examination may be missed. Nevertheless, testing female urine may particularly be useful when a pelvic examination to collect an endocervical sample is not possible.

The direct fluorescence assay is also used to identify asymptomatic female carriers of C. trachomatis. This test is based on the identification of chlamydial elementary bodies in endocervical smears stained with fluorescence-conjugated, Chlamydia-specific antibody. The sensitivity and specificity of direct fluorescence assay depend on the ability of the microscopist to correctly identify the chlamydial particles and distinguish them from artifacts. The diagnostic accuracy (in experienced hands) and the cost of this test are similar to those of enzyme immunoassay [26]. However, direct fluorescence assay is not suitable for testing large numbers of samples daily because it is more laborious and time-consuming than enzyme immunoassay.

In our decision analysis, we assumed that 75% to 90% of women with a diagnosis of chlamydial infection of the genital tract would return for treatment. However, follow-up of these patients may be difficult in certain circumstances. To overcome this problem, physicians may prefer to use an enzyme immunoassay that can be done at the physician's office and can provide a result while the patient is still there. Such rapid enzyme immunoassays are available on the market. In one study, the sensitivity of a rapid enzyme immunoassay, which takes approximately half an hour to complete, was found to be 48% when used in an asymptomatic female population with a chlamydial infection rate of 6.6% [10]. In addition to their poor sensitivity, these tests are not suitable for processing large numbers of samples daily; they become time-consuming and laborious because they cannot be automated.

Most of the savings generated by the screening strategies were attributable to prevented complications and sequelae of chlamydial infection in women. Nevertheless, identification and appropriate treatment of the infected male partners is important to prevent not only reinfections among the treated women but also new infections among healthy women. In addition, all patients and their sexual contacts should be counseled about the disease and its treatment. At-risk persons should be encouraged to modify their sexual behavior and use barrier contraception. Such counseling and effective partner notification are prerequisites for the long-term success of any screening program. Thus, our analysis is limited to those settings in which screening can be combined with these activities.

We assumed that only women who have clinical symptoms of pelvic inflammatory disease are at risk for infertility, ectopic pregnancy, or chronic pelvic pain. However, serologic studies in infertile patients suggest that C. trachomatis infection can cause subclinical pelvic inflammatory disease and result in infertility [27]. Data on the probability of such an outcome are not yet available. Nevertheless, failure to include subclinical pelvic inflammatory disease in our analysis does not affect our conclusions because inclusion of this outcome would favor the benefits rather than the costs generated by the screening strategies.

Some of the women considered in our analysis may have been infected for a long time. Because pelvic inflammatory disease can occur subclinically, some of these women may have had one or more episodes of salpingitis before they were treated for chlamydial endocervical infection. In follow-up studies in women who have laparoscopically verified pelvic inflammatory disease, delay in seeking medical care after the onset of acute salpingitis increases the risk for infertility and ectopic pregnancy [28]. Thus, the pathologic changes caused by tubal infection in some of the women considered in our analysis may have already progressed when treatment was initiated, and eradication of the causative microorganism from the endocervix by the use of antibiotics may not prevent the progress of infertility, ectopic pregnancy, and chronic pelvic pain. Failure to consider such an outcome in our decision analysis may have caused overestimation of the benefits attributable to the screening strategies. Despite this uncertainty, we conclude that screening with DNA amplification assays, combined with treatment of positive patients with azithromycin and partner notification, is the most cost-effective screening strategy for women.