Doxycycline Compared with Azithromycin for Treating Women with Genital Chlamydia trachomatis Infections: An Incremental Cost-Effectiveness Analysis

Chlamydia trachomatis infection, the most common sexually transmitted disease in the United States, is estimated to result in $2.4 billion in direct and indirect costs each year [1]. Early diagnosis and treatment of women with chlamydial infections confined to the lower genital tract are believed to reduce the likelihood of development of pelvic inflammatory disease and its major sequelae: infertility, chronic pelvic pain, and ectopic pregnancy [2, 3]. For many years, the Centers for Disease Control and Prevention have recommended doxycycline as first-line therapy for cervical chlamydial infection [4]. However, the potential effectiveness of doxycycline is reduced if patients do not comply with the recommended twice-daily, 7-day regimen [5-8]. Azithromycin, a recently approved azalide antibiotic, is an effective treatment for C. trachomatis infection when administered as a single 1-g dose [9-15]. Azithromycin represents a major advance in therapy for chlamydia because it allows directly administered single-dose therapy. However, single-dose therapy with azithromycin is considerably more expensive than a 1-week course of doxycycline, a factor that inhibits the widespread adoption of azithromycin for treatment of chlamydial infections [16]. We compared the health outcomes, costs, and incremental cost-effectiveness of conventional therapy with doxycycline with those of single-dose therapy with azithromycin for women who have uncomplicated cervical C. trachomatis infection. Methods Structure of the Decision Model: The Basic Tree Figure 1 shows the decision model used in our analysis. We calculated health outcomes and costs for two hypothetical cohorts of 100 000 nonpregnant women of child-bearing age who had laboratory-confirmed, uncomplicated cervical chlamydial infections. One cohort was treated with the doxycycline strategy, and the other cohort was treated with the azithromycin strategy. The doxycycline strategy consisted of oral doxycycline, 100 mg twice daily for 7 days; the azithromycin strategy consisted of 1 g of azithromycin in a sachet formulation administered as a single oral dose. The doxycycline and azithromycin management strategies were identical in all other respects. The subtrees of Figure 1 (identified by the curly braces) represent the risk for adverse antibiotic reactions, pelvic inflammatory disease and its sequelae, and sequelae of chlamydia in new sexual partners and neonates that occur with both treatment strategies. Figure 1. Decision tree for the azithromycin (AZI) and doxycycline (DOXY) treatment strategies. Chlamydia trachomatis We defined antibiotic reactions as adverse occurrences resulting from antibiotic therapy. Minor reactions to oral doxycycline included nausea, vomiting, diarrhea, abdominal pain, esophagitis, mild hypersensitivity reactions, photosensitivity, and oral and vaginal candidiasis [17-21]. Minor reactions to azithromycin included nausea, vomiting, diarrhea, abdominal pain, headache, dizziness, fatigue, somnolence, palpitations, mild hypersensitivity reactions, and vaginal candidiasis [9-15, 22]. Major reactions to oral doxycycline included enterocolitis, pericarditis, anaphylaxis (including angioedema), the Stevens-Johnson syndrome, severe urticarial reactions, and a lupus-like syndrome [19-21]. Major reactions to azithromycin included anaphylaxis (including angioedema), cholestatic jaundice, and interstitial nephritis [9-1522, 23]. The true frequency of major reactions associated with azithromycin is unknown because of limited experience with this drug; we therefore assumed that major reactions to azithromycin occurred with the same frequency as reactions to doxycycline. We defined chlamydia sequelae as conditions that occurred as a result of persistent cervical C. trachomatis infection. We defined minor sequelae of chlamydial infection and antibiotic reactions as self-limited conditions that could be managed solely on an outpatient basis, whereas major sequelae and reactions were outcomes that could potentially lead to hospitalization or substantial future morbidity. We classified urethritis in a male partner and neonatal conjunctivitis as minor sequelae of chlamydial infection and considered pelvic inflammatory disease, infertility, chronic pelvic pain, ectopic pregnancy, epididymitis in a male partner, and neonatal pneumonia to be major sequelae. Probability Estimates The probability estimates used in the model (Table 1) were obtained from a review of the literature and from a survey of experts using a modified Delphi method. Thirteen physicians expert in managing C. trachomatis infection and its sequelae (specialists in obstetrics and gynecology, pediatrics, and infectious disease, including faculty members of medical schools and physicians practicing in public health clinic settings) were surveyed about their estimates for an extensive set of epidemiologic and clinical variables. For each variable, we determined a best estimate and a range of plausible values representing the degree of uncertainty surrounding the variable. We determined the effectiveness of doxycycline in curing uncomplicated cervical chlamydial infections in a routine clinical setting by combining estimates of compliance with estimates of cure rates at various levels of compliance. Compliance with a 1-week, twice-daily doxycycline regimen was categorized into four levels by the number of pills taken. We assumed that 70% of patients would take 12 to 14 pills and that 96% of these patients would be cured of C. trachomatis infection; 20% would take 7 to 11 pills and 80% would be cured; 5% would take 3 to 6 pills and 40% would be cured; and 5% would take 1 to 2 pills and 10% would be cured. From these estimates, we calculated that the overall baseline estimate of doxycycline effectiveness would be 85.7% (0.70 0.96 + 0.20 0.80 + 0.05 0.40 + 0.05 0.10 = 0.857). Table 1. Estimated Probabilities for the Variables Included in the Decision Model to Determine Optimal Strategies for Treating Women with Cervical Chlamydial Infections Evaluation of the Use of Medical Services We estimated the use of medical services by patients with chlamydia sequelae on the basis of a review of the literature and expert opinion (Appendix). We asked expert respondents to detail the nature and quantity of services and resources used to treat patients with chlamydia sequelae (that is, physician visits, hospitalizations, diagnostic tests, surgical procedures, and medical therapy). Cost Estimates Our analysis was done from the perspective of a payer and was confined to assessing direct medical costs resulting from sequelae of C. trachomatis infections (Table 2). We intended the primary analysis to be applicable to many settings in which patients would be screened and treated for chlamydial infections. For the primary analysis, 1993 payments and allowed charges for medical services (diagnostic tests, physician services, and hospital care) were obtained from Colorado Blue Cross and Blue Shield. We used Blue Cross and Blue Shield payments to estimate the cost of hospital services and used Blue Cross and Blue Shield-allowed charges to estimate the cost of outpatient and physician services. Allowed charges and actual payments are a reasonable proxy for medical costs and should not be confused with the nominal charges of hospitals and physicians. Table 2. Estimated Costs of Management of Sequelae of Chlamydia trachomatis Infection We used the average wholesale price (AWP) to estimate antibiotic acquisition costs in the primary analysis. Although the term wholesale is used in its name, the AWP is actually a price suggested by the manufacturer to the pharmaceutical retailer [48]. We chose to use the AWP because it is a standardized national source for pharmaceutical pricing and because it may overstate the antibiotic cost differential between the azithromycin strategy and the doxycycline strategy, thereby biasing the model in favor of the doxycycline strategy. We determined the costs associated with each medical outcome by aggregating the various cost elements (Appendix). To adjust for time-related preferences in this analysis, costs were discounted at a rate of 5% per year for complications that would probably occur after 1 year of infection (ectopic pregnancy, chronic pelvic pain, and infertility). We then calculated the total discounted cost of each management strategy by adding the costs of each associated outcome. We also did a secondary analysis in which we modeled cost estimates for a public health clinic. Colorado Medicaid reimbursements were used to estimate the cost of inpatient, outpatient, and physician services. Estimates of antibiotic costs were based on the acquisition prices of federally supported public clinics. Because the results of the secondary analysis were almost identical to those of the more general primary analysis, we present only the results of the primary analysis. Health Outcomes For each management strategy, we summed major antibiotic reactions and major chlamydia sequelae to calculate total major complications. Minor antibiotic reactions and minor chlamydia sequelae were added to determine total minor complications. For our analysis, all major complications were considered to be of equal importance, and similarly, all minor complications to be of equal importance. There is controversy over which technique should be used to measure preferences and quality of life [49]. Because there is no standard technique for measuring utilities, we did not attempt to present quality-adjusted outcome measures. Although experts agree that financial costs must be discounted for time effects, there is controversy about whether nonfinancial health benefits should be discounted [50]. Thus, we ran the model with health outcomes both undiscounted and discounted at a 5% rate for complications that would probably occur after 1 year of infection (ectopic pregnancy, chronic pelvic pain, and infertility). Because the results were