Hartmann, Katherine E. MD, PhD1; Birnbaum, Howard PhD2; Ben-Hamadi, Rym MSc2; Wu, Eric Q. PhD2; Farrell, Max H.2; Spalding, James PharmD, MS3; Stang, Paul PhD4
Uterine leiomyomata, or fibroids, are benign tumors of the uterus made up of smooth muscle, collagen, and elastin. Studies based on imaging and/or pathology specimens suggest that as many as 77% of women develop leiomyomata during their reproductive lifetime (18–44 years old).1,2 Leiomyomata are more common among black women, and prevalence increases with age across all races.2–4 Other associated factors include early menarche, low parity, late age at first birth, years since last birth, and early use of oral contraceptives (ages 13–16 compared with later ages).3,4 Incidence of leiomyomata is estimated to be 9.2 per 1,000 person-years for U.S. women of all races, aged 25 to 44, which means that there are approximately 400,000 new cases each year.3
Most women with leiomyomata have no concerning symptoms. When leiomyoma symptoms do occur, they commonly include menorrhagia, anemia, pelvic pain or pressure, and abdominal distention.5–7 Self-report of heavy bleeding appears to be related primarily to leiomyoma location and size, although size and location alone are not strong predictors of whether a woman will be symptomatic.5 Symptoms associated with leiomyomata can be managed medically with drugs, including nonsteroidal anti-inflammatory drugs (NSAIDs), oral contraceptive pills (OCPs), and progestational agents. Some agents, such as gonadotropin-releasing hormone (GnRH) agonists and the antiprogestin mifepristone, address symptoms by directly influencing leiomyoma biology.8 Therefore, another approach to managing symptoms is to use GnRH agonists to reduce leiomyoma size or minimize bleeding and therefore allow hematologic parameters time to stabilize before surgery.9 Surgical procedures have traditionally included hysterectomy and myomectomy. Newer approaches to these procedures include hysteroscopic resection, laparoscopic myomectomy or other destructive techniques, uterine artery embolization, and most recently, magnetic resonance imaging–guided focused ultrasonic ablative therapy.1,10
Very little research has examined the relationship between clinically diagnosed leiomyomata and overall use, content, and costs of medical care. In most instances, investigators have examined the costs associated with specific procedures (hysterectomy, myomectomy, uterine artery embolization).11,12 Because leiomyomata are so common and the approaches to treatment so varied, more global assessment of the costs of current patterns of care, including both the costs associated with the care itself and those that result from lost productivity, are needed. Therefore, our objective was to conduct an analysis of the care received, the direct health care costs, and the indirect work loss costs for women diagnosed with uterine leiomyomata.
MATERIALS AND METHODS
We analyzed de-identified claims from an administrative database that includes nine self-insured companies across the United States. Together, these companies have nationwide coverage of approximately 1.2 million lives in an array of industries and occupations. The data include medical and prescription drug claims and disability claims for employees of these companies. Our inclusion criteria limited the sample to female employees ages 18 to 64 and their medical, prescription drug, and disability claims for services provided between January 1999 and December 2003. Inclusion also required at least 12 months of continuous coverage for the analyses; retirees, spouses of women with leiomyomata, and dependants were excluded. Data were analyzed with SAS 9.1 (SAS Institute Inc, Cary, NC). Because the data conform with Health Insurance Portability and Accountability Act of 1996 (HIPAA) confidentiality requirements, no informed consent or institutional review board approval was necessary.
Women were classified as having a diagnosis of uterine leiomyomata if they had one or more relevant International Classification of Diseases, 9th Revision, diagnoses recorded (ICD-9 218.xx, 654.1x). An “index date” was assigned as the date of the first observed diagnosis of uterine leiomyomata, and the study period was defined as the 12-month period beginning with the index date. A control sample of women without leiomyomata was randomly selected and matched 1:1 on the basis of age, employment, region of residence, health plan type, and length of enrollment. The matching algorithm first looked for an exact age match and was then relaxed for a broader age range (to ±1 year, up to ±5 years). The other matching characteristics were perfectly matched. Patients for whom a matching control could not be found were dropped out of the study sample. Control patients were identified from the insurance eligibility data and may or may not have filed claims. None of the controls had a diagnosis of uterine leiomyomata. Because control patients cannot have a diagnosis of leiomyomata, they do not have a natural index date. The comparable lag time between initial insurance enrollment and the index date for the matched patient with leiomyomata was used to derive proxy index dates for the controls.13
Medical services were classified as either outpatient services or inpatient hospitalization. Outpatient services were grouped into leiomyoma-related procedures, emergency room visits, and clinical visits (defined as nonprocedural, non–emergency room claims). Specific procedures commonly used in the diagnosis and treatment of women with leiomyomata were identified by current procedural terminology code and examined individually. Diagnosis and evaluation procedures of interest included complete blood count, ultrasonography, computed tomography, magnetic resonance imaging, X-ray, endometrial biopsy/dilation and curettage (D&C), diagnostic laparoscopy, and diagnostic hysteroscopy. Surgical procedures of interest included hysterectomy, myomectomy, surgical laparoscopy, surgical hysteroscopy, uterine artery embolization, and other treatments involving ablation and cryoablation of endometrium and of leiomyomata. We calculated the proportion of women with leiomyomata and of controls who received each type of medical service and procedure and compared them using relative risk ratios. We also computed and compared the proportion of women with and of those without diagnosed leiomyomata who had a disability claim. We assessed significance of differences in usage rates with Pearson χ2 tests and report 95% confidence intervals.
We identified a subset of women as “newly diagnosed” with leiomyomata characterized as follows: 1) continuous eligibility for the 12 months prior and 12 months after the index date; 2) a diagnostic code for leiomyomata in claims during 2000–2002; and 3) no diagnosis of leiomyomata in the year before their index date. This subset analysis was conducted to describe treatment patterns for the 12-month period following a “new” diagnosis. Use of leiomyoma-related diagnostic and surgical procedures and hospital inpatient stays (with or without surgical procedures) were analyzed. In addition, we also analyzed the pharmaceutical treatment pattern for selected drugs, including hormone-based treatments, NSAIDs, anemia treatments, and GnRH agonists. We did not attempt to estimate excess use of these medications and therefore did not compare their use among patients with diagnosed leiomyomata and controls.
Costs incurred during the year after the index date were measured in the full sample. Health care costs (ie, direct costs) were estimated by using third party payments to health care providers for medical and prescription drug claims rather than billed charges and do not reflect patient out-of-pocket payments. All health care costs were included.
Health care costs were disaggregated into drug, inpatient, and outpatient costs. Because approximately 2% of the health care claims were missing the payment amount, stratified hot-deck imputations14 were applied to a random selection of observed claims payment amounts with the same procedure code, specialty, place of service, and type of service to estimate these payments. Claims with missing medical costs that could not be reliably imputed using this method were dropped from the cost study (approximately 0.01%).
We calculated work loss costs (ie, indirect costs) based on disability claims, absenteeism due to medical conditions, and industry average wage information. Both short- and long-term disability claims were identified. When estimating medically related absenteeism, we assumed that each hospital day represented a full day of work loss, and one or more outpatient visits in the same day accounted for a half day of work loss. Sick days not associated with a medical visit were not included. Industry-specific average daily salaries are derived from the Merged Outgoing Rotation Groups from the Current Population Survey conducted by the Bureau of Labor Statistics. These average wages were applied to calculate work loss in dollars. Average per-capita costs were compared across samples using paired t tests, and 95% confidence intervals are reported. The excess (ie, incremental) cost for women with leiomyomata was calculated as the difference between the average costs for women with a diagnosis of leiomyomata and that for the controls without a diagnosis of leiomyomata.
Additionally, we used a multivariable, two-part regression model to further control for potential confounding factors. The underlying sample used in the regression model consisted of the 1:1 matched sample described above.
A prediction model was used to estimate the costs for control patients, adjusting for age, region of residence, health plan type, and comorbidities measured with the Charlson comorbidity index score.15 This model was then used to estimate what the costs for patients would have been had they not had uterine leiomyomata. The difference between the observed costs for women with diagnosed leiomyomata and this estimated amount was defined as the excess costs. Confidence intervals were estimated using the nonparametric bootstrap method.16
We identified 5,122 women with diagnoses of leiomyomata and 5,122 matched control subjects. Table 1 summarizes the sample’s age distribution. By design, the control sample had the same age distribution. Eighty-two percent of women with leiomyomata had some diagnostic evaluation: 42% had one or more complete blood count drawn, 66% had pelvic imaging (predominantly ultrasonography), and 38% had surgical procedures for either diagnosis or treatment (Table 2).
In the subset of 2,957 women whom we classified as newly diagnosed, patterns of care were very similar to those in the full sample. Most had one or more procedures or laboratory tests (84%) within the year following the initial diagnosis. Average use of diagnostic and evaluation tests included 2.4 procedures, mainly the repeated use of complete blood counts. Thirty-two percent of newly diagnosed women had a complete blood count before the index diagnosis date, and 24% had an ultrasound examination. In the year following the index date, ultrasound examinations were the most common procedure (65%), followed by endometrial biopsies/D&C, both before (7%) and after (11%) the date of leiomyoma diagnosis.
Over half of the newly diagnosed patients had a prescription for at least one of the drugs we selected for possible use in managing leiomyomata or their symptoms (Table 3). Hormone-based therapies (including oral contraceptives and other hormonal treatments) and NSAIDS were the most common drugs prescribed (34% and 28%, respectively). Use of anemia treatments, such as erythropoietin, and GnRH agonists was very low.
Thirty percent of newly-diagnosed patients had at least one claim for a surgical procedure, most commonly hysterectomy (20%). One-third had at least one hospital inpatient stay with an average duration of 2.8 days. Overall, 94% of newly-diagnosed patients with leiomyomata had at least one claim for a procedure (diagnostic or surgical) in the year following the index date.
In the entire sample, women with a diagnosis of leiomyomata had consistently higher use of each of the resources investigated compared with controls (Table 2). Women with leiomyomata were 3.1 times more likely (95% confidence interval [CI] 2.9–3.2) to incur costs for a diagnostic or evaluation procedure (82% versus 27%) and more than 10 times as likely to have ultrasonography, diagnostic hysteroscopy, or diagnostic laparoscopy than matched controls. Although the surgical procedures of interest, with the exception of myomectomy and uterine artery embolization, have many other indications, total use of surgical procedures was 35-fold greater among women with leiomyomata (30% versus 1%). Twenty-one percent of women with leiomyomata had hysterectomy, which is 50 times greater than the rate of hysterectomy among women without a diagnosis of leiomyoma during the time window of study. Assessing all types of hospitalizations, women with leiomyomata were also 5.6 times (95% CI 5.0–6.3) more likely to have one or more inpatient hospitalizations (36% versus 6%).
Leiomyomata and care related to symptoms caused by leiomyomata also influenced work loss. Women with diagnoses of leiomyomata were more than three times as likely to have disability claims as controls (relative risk [RR] 3.1, 95% CI 2.7–3.6). Among the 804 women with disability claims, 572 (71%) also had at least one claim for a leiomyoma-related surgery, representing 599 surgeries in total, 82% of which occurred in the week before the disability claim or on the same day.
Our cost analysis was conducted using the full sample of 5,122 women with a diagnosis of leiomyomata and their matched controls. The average annual health care (including inpatient, outpatient, and drug) cost for a woman with a diagnosis of leiomyomata was $5,989 or 3.2 times the cost of a matched woman without a diagnosis of leiomyomata (Table 4). Outpatient services accounted for the largest portion of total costs, among both women with and those without a diagnosis of leiomyomata (40% and 46%, respectively), but excess clinical care cost for women with leiomyomata ($4,143) was statistically significant (P<.01). Drug costs (not limited to drugs related to leiomyoma treatment) added a further $105 to patients’ excess medical costs, bringing the total excess direct health care costs for those women with a diagnosis of leiomyomata to $4,248 (95% CI $3,929–$4,568). Costs of inpatient hospitalizations were 6.6 times the costs of controls, accounting for an excess cost of $2,325 (95% CI $2,112–$2,538).
Excess indirect cost for women with a diagnosis of leiomyomata was $833 (95% CI $729–$936), which included excess absenteeism ($210, 95% CI $143–$278) and disability claims ($622, 95% CI $543–$701). Total costs for those with diagnoses of leiomyomata ($8,192) were 2.6 times that for matched controls and resulted in excess total direct and indirect costs of $5,081 (95% CI $4,710–$5,452).
The fully adjusted estimates that predict excess annual cost for women with a diagnosis of leiomyomata are slightly lower and include $3,885 for direct health care costs (95% CI $3,467–$4,225), $771 for indirect work loss costs (95% CI $661–$881), and $4,624 for total excess costs (95% CI $4,132–$5,042).
We found that clinically detected uterine leiomyomata are a costly condition, both in terms of direct health care and indirect work loss costs. All components of these costs were consistently and significantly higher for women with a diagnosis of leiomyomata compared with a carefully matched group of women without such a diagnosis—in total an adjusted average of $4,624 dollars per woman per year. The excess cost estimates from the multivariate, two-part regression model are modestly lower than the costs estimated directly for health care, indirect and total, respectively, because the multivariate regression model controls for comorbidities in addition to other potential confounding factors. To some degree the indirect economic costs are underestimates. We could not capture reduced productivity at work or sick days unless the lost day was associated with a medical visit. We were also unable to capture instances in which vacation days may have been used in addition to or in place of disability days while recovering from surgery. Therefore, for indirect costs, these results should be seen as conservative.
For direct costs of care, these data are robust and reflect a large and diverse population, but all women included were insured and employed. As has been shown in a number of other conditions and medical payment support systems,17,18 we can infer that women in more vulnerable populations with poor access to care may incur substantially greater morbidity before seeking or receiving care and may make greater use of costly emergency services, or because of their disease, may be unable to fully or effectively participate in the job market. Although it is not possible to know whether their costs would be comparable or higher in a given year, it is unlikely that their costs would be lower when viewed across a lifetime.
We note that these costs are cross-sectional and reflect both women with new and those with longer-term diagnoses of leiomyomata. Women may be most likely to receive a diagnosis of leiomyomata when they seek care specifically aimed at managing leiomyomata or related symptoms or may learn of them from incidental detection during obstetric ultrasonography or from pelvic examinations without acquiring a coded diagnosis in claims records. If awareness of having a leiomyoma influences care seeking and drives diagnosis predominantly when symptoms worsen, this analysis overestimates costs. Knowledge of leiomyoma status, however, has been reported not to influence symptom perception in terms of bleeding and pain.19 As a result, we anticipate that this economic analysis is not substantially flawed by differences in the likelihood that a woman with leiomyomata seeks care in the absence of symptoms. The excess costs documented are appropriately viewed as the excess costs predominantly, although not exclusively, associated with women with symptomatic leiomyomata, not the costs for any woman with leiomyomata. It is possible that there are women included in the matched sample who have a leiomyoma history before our window of analysis. Because nearly 80% of women have leiomyomata,2 it is likely that even some controls also had uterine leiomyomata and had not been diagnosed during the index period. However the purpose of this analysis is to assess the economic consequences of symptomatic or clinically detected leiomyomata.
Because leiomyomata are common and incidence of leiomyoma symptoms rises steadily across the reproductive lifetime, a continuous stream of women is being diagnosed with leiomyomata and associated with related care. We have provided an annualized snapshot of the associated costs. We have also noted that costs are elevated, even after menopause when costs might be presumed to be very low (data not shown). Further investigation specifically aimed at understanding care patterns for women with leiomyomata and leiomyoma-related cost across the reproductive lifespan and after menopause is warranted.
Women, their care providers, those who cover the costs of care, and employers who absorb the lost productivity associated with work absences, all have a vested interest in understanding how to achieve the best possible outcomes of leiomyoma-related care. This analysis is simplified in that it reduces the “costs” to monetized measures. We regret that we are not able to capture other personal and emotional costs. Leiomyoma-related bulk symptoms and associated bleeding can interfere with sexual function, result in embarrassing soaking-type bleeding, and cause pain or bleeding that restricts women’s social and physical activities or interferes with roles that they value in the family, the community, and the workplace.20 The sample selection criteria reflect the population of interest in the study, ie, female employees, aged 18–64 years, because the purpose of the study is to assess direct and indirect costs associated with the diagnosis of uterine leiomyomata among the working-age population. Although the results are not generalizable to the entire U.S. population, they are relevant to the employed population with health insurance coverage.
This large-scale look at direct and indirect costs is only a first step that highlights the need for ongoing research. Beyond treatment, diagnostic procedures contribute substantially to cost. For both the entire sample of women who have a diagnosis of leiomyomata in this study and those classified as newly diagnosed, use of ultrasonography and complete blood counts predominated, followed by endometrial biopsies and D&C. Research data to guide the use of such tools in management of leiomyomata is meager. Moreover, no evidence exists at present to directly compare and contrast the benefits, costs, and harms of medical treatments to resolve leiomyomata compared with surgical approaches or symptom management approaches. Indeed, when such research is conducted, it may find that individual patient outcomes and cost-effective care are not concomitant. Nonetheless, future research that encompasses the entire spectrum—from discovering the causes of leiomyomata and ways of preventing their occurrence to building a better evidence base for current treatments and discovering novel diagnostic procedures and therapies—will support more informed decisions by clinicians, payers, and patients and, hopefully, lead to improved care for the many affected women.
1.Nowak RA. Fibroids: pathophysiology and current medical treatment. Bailleres Clin Obstet Gynaecol 1999;13:223–38.
2.Day Baird D, Dunson DB, Hill MC, Cousins D, Schectman JM. High incidence of uterine leiomyoma: ultrasound evidence. Am J Obstet Gynecol 2003;188:100–7.
3.Marshall LM, Spiegelman D, Barbieri RL, Goldman MB, Manson JE, Colditz GA, et al. Variation in the incidence of uterine leiomyoma among premenopausal women by age and race. Obstet Gynecol 1997;90:967–73.
4.Wise LA, Palmer JR, Harlow BL, Spiegelman D, Stewart EA, Adams-Campbell LL, et al. Reproductive factors, hormonal contraception, and risk of uterine leiomyomata in African-American women: a prospective study. Am J Epidemiol 2004;159:113–23.
5.Wegienka G, Baird DD, Hertz-Picciotto I, Harlow SD, Steege JF, Hill MC, et al. Self-reported heavy bleeding associated with uterine leiomyomata. Obstet Gynecol 2003;101:431–7.
6.Kjerulff KH, Langenberg P, Seidman JD, Stolley PD, Guzinski GM. Uterine leiomyomas: racial differences in severity, symptoms and age at diagnosis. J Reprod Med 1996;41:483–90.
7.Akinyemi BO, Adewoye BR, Fakoya TA. Uterine fibroid: a review. Niger J Med 2004;13:318–29.
8.Agency for Healthcare Research and Quality. Management of uterine fibroids. Evidence Report/Technology Assessment: Number 34. Rockville (MD): AHRQ; 2001.
9.Zullo F, Pellicano M, De Stefano, R, Zupi E, Mastrantonio P. A prospective randomized study to evaluate leuprolide acetate treatment before laparoscopic myomectomy: efficacy and ultrasonographic predictors. Am J Obstet Gynecol. 1998;178:108–12.
10.Stewart EA, Gedroyc WM, Tempany CM, Quade BJ, Inbar Y, Ehrenstein T, et al. Focused ultrasound treatment of uterine fibroid tumors: safety and feasibility of a noninvasive thermoablative technique. Am J Obstet Gynecol 2003;189:48–54.
11.Beinfeld MT, Bosch JL, Gazelle GS. Hospital costs of uterine artery embolization and hysterectomy for uterine fibroid tumors. Acad Radiol 2002;9:1300–4.
12.Al-Fozan H, Dufort J, Kaplow M, Valenti D, Tulandi T. Cost analysis of myomectomy, hysterectomy, and uterine artery embolization. Am J Obstet Gynecol 2002;187:1401–4.
13.Wu EQ, Birnbaum H, Marynchenko M, Mareva M, Williamson T, Mallet D. Employees with overactive bladder: work loss burden. J Occup Environ Med 2005;47:439–46.
14.Little R, Rubin D. Statistical analysis with missing data. New York (NY): John Wiley & Sons; 1987.
15.Charlson ME, Pompei P, Ales KL, Mackenzi CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40:373–83.
16.Efron B, Tibshirani R. Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy. Statist Sci 1986;1:54–77.
17.Long SK, Coughlin T, King J. How well does Medicaid work in improving access to care? Health Serv Res 2005;40:39–58.
18.Lu N, Samuels ME, Wilson R. Socioeconomic differences in health: how much do health behaviors and health insurance coverage account for? J Health Care Poor Underserved 2004;15:618–30.
19.Wegienka G, Baird DD, Hertz-Picciotto I, Harlow SD, Hartmann KE. Uterine leiomyomata (fibroids): Are bleeding symptoms more likely to be reported after diagnosis? J Clin Epidemiol 2004;57:318–20.
20.Spies JB, Cooper JM, Worthington-Kirsch R, Lipman JC, Mills BB, Benenati JF. Outcome of uterine embolization and hysterectomy for leiomyomas: results of a multicenter study. Am J Obstet Gynecol 2004;191:22–31.