PELVIC INFLAMMATORY DISEASE (PID) is a common and damaging disease of young women caused most frequently by sexually transmitted infections.1,2 Infertility, ectopic pregnancy, chronic pelvic pain, and other complications often occur after episodes of PID.1 Hospitalization to treat PID, formerly the norm, has become more uncommon as a result of financial constraints against its use3 and, more recently, as a result of evidence showing its lack of efficacy for complication prevention in women with mild to moderate PID.4,5
At present, the Centers for Disease Control and Prevention (CDC) recommends hospitalization for PID treatment (based on observational data and theoretical concerns) when surgical emergencies cannot be excluded or when the patient is pregnant; does not respond clinically to oral antimicrobial therapy; is unable to follow or tolerate an outpatient oral regimen; has severe illness, nausea and vomiting, or high fever; or has a tuboovarian abscess.1 The CDC also adds that “No data are available that suggest that adolescent women benefit from hospitalization for treatment of PID.”1 Despite this, some physicians believe that hospitalization decreases the risk of infertility and other complications in adolescents and young adult women with available evidence being insufficient to change their practice.6 As a result, hospitalization continues as a treatment strategy used in hope of preserving fertility in nulliparous young females with PID regardless of its clinical severity. Data from 1995–2001 show that, of all women newly diagnosed with PID, 9.4% of 15 to 19 year olds are hospitalized (population hospitalization rate: 1.4 per 1000 women) and 7.8% of 20 to 24 year olds are hospitalized (1.6 per 1000) compared with 5.4% of 25 to 29 years olds (1.1 per 1000) (P <0.01 for all comparisons of proportion hospitalized).7 Furthermore, because the hospitalization rate denominator is the population of all 15 to 19 year olds, not just those who are sexually active, the true hospitalization rate could actually be double this estimate if less than 50% are sexually active, as recent CDC data show.7,8
Much of the evidence comparing hospital and ambulatory treatment of PID comes from the PID Evaluation and Clinical Health (PEACH) Study.4,5 This multicenter, randomized, controlled trial of ambulatory treatment compared with hospitalization for mild to moderate PID showed no differences in fertility outcomes in the entire cohort or in clinically important subgroups such as younger women. Continued follow up over an average of 84 months showed no differences in any complications of PID with the study powered to detect 20% to 30% differences in most outcomes.5 A cost-effectiveness analysis comparing hospitalization with ambulatory treatment for mild to moderate PID using PEACH Study data would show that hospitalization was unwarranted as a result of unchanged effectiveness at higher cost. However, because the study was not powered to detect differences of less than 20% to 30% in most outcomes, and had less power to detect differences in less frequent outcomes and between some subgroups,5 inpatient therapy could still be reasonable, compared with ambulatory treatment, as a result of significant but undetected improvements in outcomes. The purpose of this analysis is to extend prior PEACH Study analyses to determine the economic feasibility of hospitalizing adolescents and young women for PID.
Materials and Methods
We constructed a Markov model to estimate the cost-effectiveness of hospitalization compared with outpatient therapy for mild to moderate PID under various assumptions about hospitalization effects on complications. In our base case analysis, we considered 18-year-old women from a societal perspective9 using a 10-year time horizon, a 3% discount rate, and converting all monetary costs to 2004 U.S. dollars using the U.S. Consumer Price Index.10 In sensitivity analyses, we also considered 15-year-old patients and longer time horizons. Effectiveness was calculated using quality-adjusted life-years (QALY) to account for changes in life expectancy and quality of life. Quality of life for PID-related health states was modeled using utility value weights, which range from zero (death) to one (perfect health) with QALY being the product of time spent in a health state and its utility.9 The Markov cycle length was 1 month.
Our base case assumption was if hospitalization decreases PID complications, then infertility, ectopic pregnancy, and chronic pain rates were equally affected. For example, if hospitalization decreased complication rates by 10%, then the relative risk of all 3 complications decreased by 10%. This assumption was then tested in sensitivity analyses.
Figure 1 shows a schematic diagram of the Markov model. At the start, identical hypothetical cohorts of 18-year-old women clinically diagnosed with and treated for PID can either truly have or not have PID, because studies show a false-positive diagnosis in approximately one third of patients.11 Treatment takes place either in the hospital or in the ambulatory setting. After the acute episode, women with PID transition to the post-PID state, in which complications of chronic pain, infertility, and ectopic pregnancy can occur over time. Embedded within the PID health state are risks of medication nonadherence, treatment side effects, and abscess formation; embedded within the ectopic pregnancy state is the risk of rupture, and the infertility state accounts for the likelihood of seeking and receiving infertility interventions. Transition to the dead state (not shown in the figure) can occur from all other states based on age and sex-specific U.S. mortality rates12 and the low risk of mortality from PID-related causes.
Parameter input values in the model are shown in Table 1. Yearly incidence data for PID complications and recurrences are derived from the PEACH Study4,5 combining data from both inpatient and ambulatory treatment groups (because no statistically significant differences were seen between groups) and converting 7-year incidence data to yearly risks. For example, a yearly infertility risk of 2.8% corresponds to a 25.1% risk of infertility over the 10-year time horizon of the model. Other incidence distribution assumptions, including delays in complication onset after PID episodes,13 were examined in sensitivity analyses. Costs are from the refined estimates of lifetime average PID costs established by Yeh et al13 inflated to 2004 U.S. dollars. Health state utilities are from a study using Health Utility Index values assigned by a committee convened by the Institute of Medicine14; these are varied over relatively broad ranges in sensitivity analyses. Likelihood of PID complications doubled in adolescents and young women with 2 or more prior episodes of PID. In those with one episode of ectopic pregnancy, risk of recurrent ectopic pregnancy increased to 22% thereafter.13,15
One-way sensitivity analyses and probabilistic sensitivity analyses were performed. In the probabilistic sensitivity analyses, all parameters listed in Table 1 were varied simultaneously over their listed ranges with 1000 recalculations of incremental cost-effectiveness ratios using random draws from their distributions. In these analyses, incidence data were varied over normal distributions, probabilities and costs over triangular distributions, and utilities and time-related parameters were varied over uniform distributions, examining scenarios in which complications were decreased 10%, 20%, or 30% by hospital treatment of PID compared with ambulatory therapy.
Figure 2 depicts the main results of the analysis. If hospitalization decreases all PID complications by 10%, hospitalization gains 0.051 QALYs (approximately 18.6 quality-adjusted days) compared with ambulatory treatment at an incremental cost of $7380 producing an incremental cost-effectiveness ratio (ICER) of $145,000 per QALY gained. If complications decrease 20% with hospital treatment, hospitalization costs $67,400 per QALY, and if complications decrease 30%, hospitalization costs $42,400 per QALY gained. Although no absolute cost-effectiveness criterion exists,9 interventions costing less than $20,000 per QALY gained are commonly considered to have strong evidence for adoption, those between $20,000 and $100,000 per QALY have moderate evidence, and those costing more than $100,000 per QALY have weak evidence for adoption.16
Individual variation of other parameters in one-way sensitivity analyses did not have a large impact on the results of our base case analyses. When considering a 10% decrease in complications with hospitalization (which costs $145,000 per QALY gained in the base case analysis), the hospitalization strategy cost less than $50,000 per QALY if the cost of hospitalization is less than $2300 (base case cost $6500). Assuming the same complication rate, the cost per QALY gained with hospitalization was: $119,000 if the use of infertility was reduced to 0.3 (base case value 0.82), $135,000 if the yearly incidence of infertility was increased to 5% (base case 2.8%), and was $203,000 if chronic pain incidence was reduced to 3% per year (base case 7.3%). Increasing the average discounted cost of infertility treatment to $10,000 decreased the incremental cost-effectiveness ratio to $143,500; increasing the average discounted cost of ectopic pregnancy to $20,000 decreased the cost per QALY gained by hospitalization by less than $100. Alternative PID complication incidence distribution assumptions such as incorporation of a 5-year delay in infertility onset after PID occurrence13 had minimal impact on results.
Changing the initial age considered in the model from 18 to 15 had little impact on analysis results, but results were sensitive to length of time considered in the model. In the base case analysis (with a 10-year time horizon and 10% decrease in complications), hospitalizing 15 year olds costs $144,000 per QALY gained. When 15 year olds are considered over longer time periods, hospitalization costs $78,900 over a 30-year time horizon and $43,100 over a lifetime time horizon.
In probabilistic sensitivity analyses, all parameters were varied simultaneously in the ranges shown in Table 1. Figure 3 displays the results of those analyses in the form of cost-effectiveness acceptability curves, showing the proportion of the 1000 random draw cost-effectiveness calculations in which hospitalization would be considered acceptable from a societal standpoint for a series of willingness to pay thresholds. If an acceptability threshold of $50,000 per QALY is chosen, the likelihood of hospitalization being acceptable is only 1% if complications are reduced by 10%, approximately 44% if complications are decreased by 20%, and approximately 76% if complications decrease 30%. If the acceptability threshold is $100,000, hospitalization is more likely to be acceptable if all complications are decreased by 20% or more.
Clinical trial data and physician opinion are least convincing that hospitalization affects the future development of chronic pelvic pain, a frequent and costly complication of PID.3–5,13 Assumptions about hospitalization effects on the development of chronic pelvic pain heavily weight the analysis; costs per QALY gained by hospitalization increase considerably if chronic pain is unaffected. For example, if chronic pelvic pain risk is unaffected by hospitalization but the other PID complications are, the acceptability curves change considerably (Fig. 4) with a very low likelihood of acceptability if the cost-effectiveness criterion is <$100,000 per QALY gained and a <50% acceptability likelihood if the cost-effectiveness criterion is $500,000 per QALY or less. Additionally, if a lifetime time horizon for 15-year-old patients is considered and no effects on chronic pain are seen, hospitalization will cost $238,000 per QALY or more if other complications are reduced by 20% or less. Other than for chronic pelvic pain, relaxing the assumption of equal therapeutic effect on complication relative risk reduction and allowing differential impact across complications has little influence on results.
In our analysis, we show that hospitalizing young, nulliparous women with mild to moderate PID in hopes of preserving fertility is unlikely to be economically reasonable. Several factors lead to this conclusion. First, in this analysis, we biased the model toward hospitalization by assuming that the likelihood of PID complications is decreased by hospital treatment compared with ambulatory therapy, which is unsupported by clinical trial data but could be present based on the limited power to detect differences in complication rates in those trials.4,5 Second, if all complications are equally affected by hospital therapy (i.e., relative risks are all decreased by the same degree), decreases in complication rates of 20% to 30% would need to be seen for costs per QALY gained to approach economically reasonable ranges. Current studies were powered to detect this level of difference for all complications combined and for the most frequently observed individual complication, chronic pelvic pain.4,5 Finally, the cost-effectiveness of hospitalization substantially depends on effects on chronic pelvic pain. If chronic pain risk is unaffected by hospitalization, then the decision to admit to the hospital is very expensive, costing more than $500,000 per QALY gained if other complications are decreased by 30% or less.
Chronic pelvic pain is a common, morbid, and expensive sequela of PID; thus, its influence in the analysis is not surprising. Furthermore, its high relative frequency (affecting 20–41% of women with a history of PID) should make clinical trial detection of differences in pelvic pain rates between hospital and ambulatory treatment regimens easier than for less common complications. Despite this, no differences in pelvic pain rates based on treatment strategy have been described.
Most arguments for hospital treatment of young females with PID center, along with the ability to monitor antibiotic therapy, on the hospital stay being an opportunity to educate and emphasize the importance of antibiotic adherence to preserve future fertility.6 Unfortunately, clinical trial data suggest that a hospital stay for PID has no effect on antibiotic adherence after hospital discharge.17
Our study has limitations. Data from the PEACH Study, the major source of data for our model, may not be generalizable to all segments of the female population at risk for PID. However, as a result of its multicenter, prospective approach and the number of patients enrolled, the PEACH Study gives the most accurate picture available of the impact of PID. Observed rates of infertility and ectopic pregnancy in the PEACH Study were less than those seen in some other studies; however, variation of these rates in sensitivity analyses did not significantly alter analysis results. Finally, we did not consider the ambulatory use of intravenously antibiotics, which could conceivably improve outcomes at less cost than hospitalization. However, no data to support or refute this contention are available.
Decision analysis techniques allow structured and systematic use of available data to examine where data are sufficient to reach reasoned conclusions within the bounds of uncertainty presented by that data or to point out where further data would be helpful or necessary to reach more robust conclusions.18 Based on our analysis, we conclude that hospitalizing young women for treatment of mild to moderate PID is very unlikely to be clinically effective or economically reasonable even when accounting for uncertainly based on the power of existing clinical trials. The cost-effectiveness of hospitalization depends on substantial reductions in PID complications and, of those complications, depends mainly on an effect on chronic pelvic pain. Adequately powered clinical trials have shown no benefit of hospitalization for mild to moderate PID for any patient group, and effects large enough to influence the decision should have already been seen.
1. Centers for Disease Control and Prevention. Sexually transmitted diseases treatment guidelines 2002. MMWR Morb Mortal Wkly Rep 2002; 51:1–78.
2. Westrom L, Joesoef R, Reynolds G, et al. Pelvic inflammatory disease and fertility. A cohort study of 1,844 women with laparoscopically verified disease and 657 control women with normal laparoscopic results. Sex Transm Dis 1992; 19:185–192.
3. Berger GS, Westrom L. Pelvic Inflammatory Disease. New York: Raven Press, 1992.
4. Ness RB, Soper DE, Holley RL, et al. Effectiveness of inpatient and outpatient treatment strategies for women with pelvic inflammatory disease: Results from the Pelvic Inflammatory Disease Evaluation and Clinical Health (PEACH) randomized trial. Am J Obstet Gynecol 2002; 186:929–937.
5. Ness RB, Trautmann G, Richter HE, et al. Effectiveness of treatment strategies of some women with pelvic inflammatory disease: A randomized trial. Obstet Gynecol 2005; 106:573–580.
6. Hemsel DL, Ledger WJ, Martens M, et al. Concerns regarding the Centers for Disease Control’s published guidelines for pelvic inflammatory disease. Clin Infect Dis 2001; 32:103–107.
7. Sutton MY, Sternberg M, Zaidi A, et al. Trends in pelvic inflammatory disease hospital discharges and ambulatory visits, United States, 1985–2001. Sex Transm Dis 2005; 32:778–784.
8. Centers for Disease Control and Prevention. Trends in sexual risk behaviors among high school students—United States, 1991–2001. MMWR Morb Mortal Wkly Rep 2002; 51:856–859.
9. Gold MR, Siegel JE, Russell LB, Weinstein MC, eds. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press, 1996.
10. US Department of Labor, Bureau of Labor Statistics web site. Consumer Price Index. Available at: http://www.bls.gov/data/
. Accessed March 1, 2006.
11. Beigi RH, Wiesenfeld HC. Pelvic inflammatory disease: New diagnostic criteria and treatment. Obstet Gynecol Clin North Am 2003; 30:777–793.
12. Arias E. United States life tables, 2002. National Vital Statistics Reports, vol 53, no 6. Hyattsville, MD: National Center for Health Statistics, 2004.
13. Yeh JM, Hook EW 3rd, Goldie SJ. A refined estimate of the average lifetime cost of pelvic inflammatory disease. Sex Transm Dis 2003; 30:369–378.
14. Appendix 2: Chlamydia. In: Stratton KR, Durch J, Lawrence RS, eds. Vaccines for the 21st Century: A Tool for Decision Making. Washington, DC: National Academy Press, 2000.
15. Joesoef MR, Westrom L, Reynolds G, Marchbanks P, Cates W. Recurrence of ectopic pregnancy: The role of salpingitis. Am J Obstet Gynecol 1991; 165:46–50.
16. Laupacis A, Feeny D, Detsky AS, et al. How attractive does a new technology have to be to warrant adoption and utilization? Tentative guidelines for using clinical and economic evaluations. CMAJ 1992; 146:473–481.
17. Dunbar-Jacob J, Sereika SM, Foley SM, Bass DC, Ness RB. Adherence to oral therapies in pelvic inflammatory disease. J Womens Health 2004; 13:285–291.
18. Pauker SG, Kassirer JP. Decision analysis. N Engl J Med 1987; 316:250–258.
19. Washington AE, Katz P. Cost of and payment source for pelvic inflammatory disease. Trends and projections, 1983 through 2000. JAMA 1991; 266:2565–2569.
20. Rein DB, Kassler WJ, Irwin KL, et al. Direct medical cost of pelvic inflammatory disease and its sequelae: Decreasing, but still substantial. Obstet Gynecol 2000; 95:397–402.
21. Landers DV, Sweet RL. Tubo-ovarian abscess: Contemporary approach to management. Rev Infect Dis 1983; 5:876–884.
22. Wiesenfeld HC, Sweet RL. Progress in the management of tuboovarian abscesses. Clin Obstet Gynecol 1993; 36:433–444.
23. Washington AE, Arno PS, Brooks MA. The economic cost of pelvic inflammatory disease. JAMA 1986; 255:1735–1738.
24. Schwebke JR, Sadler R, Sutton JM, et al. Positive screening tests for gonorrhea and chlamydial infection fail to lead consistently to treatment of patients attending a sexually transmitted disease clinic. Sex Transm Dis 1997; 24:181–184.
25. Hu D, Hook EW 3rd, Goldie SJ. Screening for Chlamydia trachomatis
in women 15 to 29 years of age: A cost-effectiveness analysis. Ann Intern Med 2004; 141:501–513.
26. Magid D, Douglas JM Jr, 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.