The Cost of Diagnosing Endometrial Cancer: Quantifying the Healthcare Cost of an Abnormal Uterine Bleeding Workup

Abstract

Objective

The evaluation of women with perimenopausal abnormal uterine bleeding (AUB) and postmenopausal bleeding (PMB) to detect endometrial cancer (EC) and its precursors is not standardized and can vary widely. Consequently, costs associated with the workup and management undoubtedly vary. This study aimed to quantify costs of AUB/PMB evaluation to understand the healthcare burden associated with securing a pathologic diagnosis.

Methods

Women ≥45 years of age presenting to a single institution gynecology clinic with AUB/PMB for diagnostic workup were prospectively enrolled February 2013-October 2017 for a lower genital tract biospecimen research study. Clinical workup of AUB/PMB was determined by individual provider discretion. Costs of care were collected from administrative billing systems from enrollment to 90 days post enrollment. Costs were standardized and inflation-adjusted to 2017 US Dollars (USD).

Results

In total, there were 1,017 women enrolled with 5.6% diagnosed with atypical hyperplasia or endometrial cancer (EC). Within the full cohort, 90-day median cost for AUB/PMB workup and management was $2,279 (IQR $512–4,828). Among patients with a diagnostic biopsy, median 90-day costs ranged from $2,203 (IQR $499–3604) for benign or disordered proliferative endometrium (DPE) diagnosis to $21,039 (IQR $19,084–24,536) for a diagnosis of EC.

Conclusions

The costs for diagnostic evaluation of perimenopausal AUB and PMB vary greatly according to ultimate tissue-based diagnosis. Even reassuring benign findings that do not require further intervention—the most common in this study’s cohort—yield substantial costs. The development of sensitive, specific, and more cost-effective diagnostic strategies is warranted.

1. Introduction

One third of all women are affected by abnormal uterine bleeding (AUB) at some point in their life [1]. AUB has shown to diminish a woman’s quality of life significantly with implications on her physical, psychological, and emotional health [2]. In reproductive age women, AUB represents one of the most frequent gynecologic conditions affecting up to 30% of women and contributing to 70% of gynecologic clinic visits in premenopausal women [3]. Postmenopausal bleeding (PMB) contributes to roughly 5% of all office gynecology visits [4]. An estimated 10% of postmenopausal women who present for evaluation of PMB are diagnosed with endometrial cancer (EC) [5–7]. AUB occurring in perimenopausal women or in high risk populations should always be perceived as suspicious for underlying malignancy. As such, priority is placed on ruling out malignancy as part of the diagnostic workup and this often requires a clinical pathology diagnosis. While the American College of Obstetricians and Gynecologists (ACOG) recommends endometrial tissue sampling as a first line diagnostic test for all patients with AUB who are older than 45 years, there is great variability in how providers choose to proceed with diagnostic assessment of AUB without a consensus on the most cost-effective approach [8]. Diagnostic assessment most commonly includes a pelvic ultrasound, office-based endometrial sampling, hysteroscopy with directed biopsies or dilatation & curettage (D&C), or a combination of any or all of these diagnostic approaches. The most common causes for PMB are endometrial atrophy (60–80%), estrogen replacement therapy (15–25%), endometrial polyps (2–12%), endometrial hyperplasia (5–10%), and EC (5– 10%) [9]. As new diagnostic technologies for the assessment of AUB and PMB emerge, there is a need to better understand the cost associated with workup and management, especially in the setting of benign findings that require no further intervention. Such objective information has the potential to also improve our understanding of the impact of this common gynecologic disorder on our healthcare system and inform decisions regarding healthcare spending and resource allocation. To that end, this study was designed to quantify costs associated with the diagnostic workup and management of women presenting with perimenopausal AUB or PMB.

2. Methods

2.1. Patient Population

We performed a prospective cohort study in women ≥45 years of age presenting to a single institution gynecology clinic with perimenopausal AUB or PMB for diagnostic workup. Patients in our cohort were enrolled between February 2013 to October 2017 for a lower genital tract biospecimen research study (NCT01793545). The study was approved by the Mayo Clinic Institutional Review Board (IRB # 16-004660). Women with prior hysterectomy, prior pelvic radiation, or endometrial biopsy within the 3 months prior to clinic presentation were excluded. Specific clinical decisions for the workup and management of AUB or PMB were performed according to individual gynecologic provider discretion. Patient demographics and clinical characteristics were obtained through both the electronic medical record and direct data collection from enrolled women by the study coordinator.

2.2. Costs

Cost data was obtained from the Mayo Clinic Rochester Cost Data Warehouse, which has been described in detail elsewhere. This database contains all billed services occurring at Mayo Clinic Rochester. These services are standardized for generalizability. Appropriate cost-to-charge ratios from Medicare Cost Reports were multiplied by the charges for all hospital billed services. Medicare reimbursement rates were assigned to all related professional billed services to estimate standardized costs with inflation adjustment to 2017 US Dollars (USD). Current Procedural Terminology (CPT) and Uniform Billing (UB) diagnosis codes were abstracted retrospectively at a minimum of 90 days post enrollment. Relatedness of individual codes to AUB or PMB workup and management was determined by three separate reviewers which included a gynecologic oncologist, minimally invasive gynecologic surgeon, and resident physician (JNB, CCD, SKW, respectively) (Appendix 1).

2.3. Data Analysis

Clinical pathology was utilized to assign a final tissue-based diagnosis for each patient. For patients with multiple diagnoses on clinical pathology, a single final tissue-based diagnosis was assigned based on the following disease severity hierarchy: endometrial cancer (EC) > atypical hyperplasia (AH) > hyperplasia without atypia > benign endometrial polyp > benign or disordered proliferative endometrium (DPE). Additional subgroups included were no endometrial biopsy performed or inadequate endometrial biopsy. Patient demographics and clinical characteristics were described for each tissue-based diagnosis. The incidence of polyps in patients with a final tissue-based diagnosis of AH was evaluated. For the subgroup of patients diagnosed with AH on endometrial biopsy or D&C, we completed a retrospective electronic medical record chart review to assign the final surgical pathology diagnoses (AH or EC) at time of hysterectomy.

Cost data was reported as 7-day, 30-day, and 90-day costs. These follow up intervals were chosen because we presumed that the 7-day, 30-day, and 90-day costs would most likely represent the costs associated with an office-based workup alone, office-based and more invasive workup plus short-term management of benign pathology, and all workup plus longer-term and/or definitive surgical management, respectively. Descriptive statistics of costs defined as related at 7-day, 30-day, and 90-day follow up timeframes were completed for each final tissue-based diagnosis. Given the highly skewed nature of healthcare costs, an analysis of non-normal/skewed distribution was performed and cost data was reported as a median value with interquartile ranges (IQR) [10, 11].

Multivariate analyses of all-cause costs were restricted to the two largest endometrial pathology subgroups, which included benign/DPE and polyp, based on tissue-based diagnoses. The multivariate analysis was used to identify covariates that were predictive of cost differences. Covariates incorporated to the model included post-menopausal status (yes/no) and BMI as a categorical variable (<25, 25 to <30, 30 to <40, and ≥40 kg/m²). To account for non-normal skewed distribution of cost, a generalized linear model with gamma distribution and log-link function was used. Predicated mean cost differences were reported for model covariates with statistical significance.

Observed all-cause costs for the complete population were also investigated based on description of services. Services were grouped by CPT-4 procedure code where available and hospital UB diagnosis code for facility services not having a CPT-4 code listed. Clinically relevant costs were defined as those that occurred in at least 50 patients (Appendix 2). After review of all clinically relevant CPT and UB diagnosis codes, we identified the five most common individual costs, most expensive individual costs, and largest drivers of total cost among all patients.

3. Results

In total, there were 1,017 women enrolled during the study period. A total of 57 (5.6%) patients were diagnosed with AH or EC (Table 1). The mean age was 55.7 ± standard deviation (SD) 7.9 years and mean BMI was 30.3 ± SD 8.0 kg/m². In our patient cohort, there were 431 (42.6%) with a BMI ≥ 30 kg/m², 541 (53.3%) postmenopausal, 161 (15.8%) nulliparous, 253 (24.9%) with hypertensive disease, 73 (7.2%) with diabetes mellitus, and 212 (20.8%) using hormone replacement therapy (HRT) or tamoxifen at the time of enrollment. The final tissue-based diagnosis of AH was made in 20 patients. A total of 15 (75%) patients with AH also had polyps noted at the time of tissue sampling. A total of 7 (35%) patients with AH based on final tissue-based diagnosis within the 90-day time frame of our study were later found to have concurrent EC at the time of hysterectomy. The final tissue-based diagnosis of EC was made in 37 patients. A total of 14 (37.8%) patients with EC also had polyps noted at the time of their final tissue-based diagnosis.

Table 1:

Baseline Patient Demographics and Clinical Characteristics by Tissue- Based Diagnosis

Characteristic	Overall	No Biopsy	Inadequate Biopsy	Benign/DPE	Polyp	Hyperplasia without Atypia	Atypical Hyperplasia	Endometrial Cancer
Total #	1017	42	32	693	165	28	20	37
Age (years), Mean (SD)	55.7 (7.9)	56.8 (8.9)	58.5 (7.7)	54.3 (7.2)	57.8 (8.5)	57.6 (7.9)	57.9 (7.6)	65.1 (8.6)
Postmenopausal†	541 (53.3)	24 (60.0)	27 (84.4)	324 (46.8)	105 (63.6)	15 (53.6)	13 (65.0)	33 (89.2)
BMI (kg/m2), Mean (SD)‡	30.3 (8.0)	28.7 (7.6)	30.4 (8.7)	29.7 (7.4)	31.4 (8.7)	31.6 (8.8)	41.1 (10.4)	32.8 (8.1)
BMI ≥30 kg/m2‡	431 (42.6)	14 (35.0)	15 (46.9)	272 (39.4)	80 (48.8)	14 (50.0)	18 (90.0)	18 (48.6)
Nulliparity	161 (15.8)	10 (23.8)	4 (12.5)	93 (13.4)	39 (23.6)	4 (14.3)	4 (20.0)	7 (18.9)
Hormone or Tamoxifen use	212 (20.8)	8 (19.0)	8 (25.0)	148 (21.4)	32 (19.4)	10 (35.7)	3 (15.0)	3 (8.1)
Hypertension	253 (24.9)	15 (35.7)	8 (25.0)	141 (20.3)	51 (30.9)	11 (39.3)	11 (55.0)	16 (43.2)
Diabetes	73 (7.2)	2 (4.8)	3 (9.4)	42 (6.1)	10 (6.1)	4 (14.3)	6 (30.0)	6 (16.2)

^†Missing data for menopausal status: no biopsy=2

^‡Missing data for BMI: no biopsy=2, benign/DPE=2, polyp=1

The overall median costs in USD for perimenopausal AUB or PMB diagnostic workup and management were as follows: 7-day $628 (IQR $333-$2,298), 30-day $1,959 (IQR $431-$3,653), and 90-day $2,279 (IQR $512-$4,828) (Table 2). Among patients with a diagnostic endometrial biopsy (N= 975), median 90-day costs ranged from $2,203 (IQR $499-$3,604) for benign/DPE to $21,039 (IQR $19,084-$24,536) for EC. Among women with an inadequate office-based endometrial biopsy (N= 32), the median 7-day cost was $472 (IQR $285-$2,227) with a substantially higher median 90-day cost of $2,205 (IQR $424–2,972). The median 30-day cost of $4,698 (IQR $1,353- $16,233) in the AH subgroup increased substantially to $14,183 (IQR $4,029-$18,669) by the 90-day follow up interval whereas the patients in the EC subgroup had similar median 30-day and 90-day costs.

Table 2:

Median Related Costs Stratified Based on Final Tissue-Based Diagnosis

		Median (Q1– Q3)†
N	Diagnosis	7-day	30-day	90-day
42	No Biopsy	202 (93 – 548)	254 (130 – 657)	320 (138 – 667)
32	Inadequate Biopsy	472 (285 – 2,227)	739 (370 – 2,343)	2,205 (424 – 2,972)
693	Benign/ DPE	581 (347 – 2,255)	1,179 (428 – 2,530)	2,203 (499 – 3,604)
165	Polyp	1,723 (340 – 2,982)	3,653 (1,619 – 5,273)	4,119 (2,601 – 5,568)
28	Hyperplasia without Atypia	703 (452 – 2,298)	2,162 (512 – 4,335)	2,839 (704 – 5,433)
20	Atypical Hyperplasia	2,029 (408 – 2,508)	4,698 (1,353 – 16,233)	14,183 (4,029 – 18,669)
37	Endometrial Cancer	2,312 (595 – 9,671)	19,688 (15,062 – 21,480)	21,039 (19,084 – 24,536)
1017	All diagnoses	628 (333 – 2,298)	1,959 (431 – 3,653)	2,279 (512 – 4,828)

^†Cost in US Dollars

Multivariate modeling revealed BMI and HRT/tamoxifen use to be predictive of mean cost. As compared to the costs in patients with a BMI <25 kg/m², we observed that patients with a BMI ≥40 kg/m² had a borderline greater mean cost of $793 (p=0.075), BMI 30–39 kg/m² had a greater cost of $1,203 (p=0.001), and BMI 25–29 kg/m² had no significant cost difference (p=0.115) (Table 3). We also observed that patients on HRT or tamoxifen at time of enrollment had a lower associated cost by $−810 (p=0.001) than those not on HRT or tamoxifen.

Table 3:

Mean Predicated Cost and Cost Difference by BMI and Current HRT or Tamoxifen Use Among Patients Classified as Benign/DPE or Polyp

Category		Mean Cost (USD)	95% Confidence Interval	Mean Cost Difference (USD)	95% Confidence Interval	p value
BMI kg/m2	< 25	2,707	(2,350 to 3,065)	reference
	25– 29	3,176	(2,710 to 3,642)	469	(−113 to 1,051)	0.115
	30– 39	3,910	(3,315 to 4,505)	1,203	(509 to 1,896)	0.001
	≥40	3,500	(2,698 to 4,303)	793	(−79 to 1,665)	0.075
HRT or Tamoxifen Use	Yes	2,652	(2,268 to 3,035)	−810	(−1,307 to −313)	0.001
	No	3,462	(3,141 to 3,783)	reference

The five most common clinically relevant individual itemized patient costs were surgical pathology (820 patients, $152 per patient), operative hysteroscopy (624 patients, $1,183 per patient), sterile supplies (380 patients, $861 per patient), hospital-based pharmacy services (361 patients, $508 per patient), and anesthesia charges (350 patients, $264 per patient) (Table 4). The five most expensive clinically relevant itemized patient costs were operating room services ($3,105 per patient, 349 patients), operative hysteroscopy ($1,183 per patient, 624 patients), diagnostic hysteroscopy/ laparoscopy ($1,071 per patient, 143 patients), outpatient surgery room observation ($948 per patient, 77 patients), and sterile supplies ($861 per patient, 380 patients) (Table 5). When considering the whole group of 1,017 women enrolled, the top five clinically relevant drivers of overall patient cost were operating room services ($1,083,586 total, 349 patients), operative hysteroscopy ($737,910 total, 624 patients), sterile supplies ($327,232 total, 380 patients), hospital-based pharmacy services ($183,290 total, 361 patients), and diagnostic hysteroscopy/ laparoscopy ($153,186 total, 143 patients) (Table 6).

Table 4:

Top 5 Most Common Clinically Relevant Patient Costs

Code	Code Type	Cost Description	N	Cost Per Patient (USD)
88305	cpt4	1Surgical Pathology	820	152
58558	cpt4	2Operative Hysteroscopy	624	1,183
0272	ubcode	3Sterile Supplies	380	861
0250	ubcode	4Pharmacy	361	508
00000	cpt4	5Anesthesiology	350	264

¹Gross & microscopic examination of surgical pathology

²Operative hysteroscopy with sampling of the endometrium and/or polypectomy, with or without D&C, may include ambulatory or operating room procedures

³Sterile medical or surgical supplies

⁴General classification of hospital based only pharmacy costs (includes inpatient or outpatient-based hospital events, excludes outpatient prescription medications)

⁵Anesthesiology services

Table 5:

Top 5 Most Expensive Clinically Relevant Patient Costs

Code	Code Type	Cost Description	N	Cost Per Patient (USD)
0360	ubcode	1Operating Room Services	349	3,105
58558	cpt4	2Operative Hysteroscopy	624	1,183
58555	cpt4	3Diagnostic Hysteroscopy or Laparoscopy	143	1,071
0762	ubcode	4Room Observation	77	948
0272	ubcode	5Sterile Supplies	380	861

¹General classification of operating room services

²Operative hysteroscopy with sampling of the endometrium and/or polypectomy, with or without D&C, may include ambulatory or operating room procedures

³Diagnostic hysteroscopic or laparoscopic procedures of the corpus uteri

⁴Room and board charge for hospital admission under observation status

⁵Sterile medical and surgical supplies

Table 6:

Top 5 Clinically Relevant Drivers of Overall Patient Cost

Code	Code Type	Cost Description	N	Overall Patient Cost (USD)
0360	ubcode	1Operating Room Services	349	1,083,586
58558	cpt4	2Operative Hysteroscopy	624	737,910
0272	ubcode	3Sterile Supplies	380	327,232
0250	ubcode	4Pharmacy	361	183,290
58555	cpt4	5Diagnostic Hysteroscopy or Laparoscopy	143	153,186

¹General classification of operating room services

²Operative hysteroscopy with sampling of the endometrium and/or polypectomy, with or without D&C, may include ambulatory or operating room procedures

³Sterile medical or surgical supplies

⁴General classification of hospital based only pharmacy costs (includes inpatient or outpatient-based hospital events, excludes outpatient prescription medications)

⁵Diagnostic hysteroscopic or laparoscopic procedures of the corpus uteri

4. Discussion

In this study, we describe the substantial healthcare cost associated with the diagnostic workup and management of perimenopausal AUB and PMB. The overall costs of this common gynecologic presentation were considerable at 7-day, 30-day, and 90-day intervals regardless of the etiology of AUB and PMB. We found the cost of initial office-based diagnostic workup to be best reflected by the overall 7-day follow up interval, whereas the cost of both workup and management was best reflected by the 30-day and 90-day follow up intervals as these were relatively similar. However, we did find variation in median costs when analyzing according to tissue-based diagnosis. Among women with AH, the cost substantially increased from 30-day to 90-day follow up intervals suggesting a more complex workup was required for definitive diagnosis and management. Clinically, this may include but is not limited to an office endometrial biopsy, an operative hysteroscopy ± D&C, and ultimately a hysterectomy. Indeed, 75% of those ultimately diagnosed with AH had a presumed benign endometrial polyp until their more invasive workup identified the presence of AH. As such, the numerical difference found in 30-day and 90-day cost in women with AH clinically fits with the finding of a diagnosis that requires further management than a D&C.

In the EC subgroup, the 30-day cost and 90-day costs were quite similar suggesting most women diagnosed with EC on initial office based endometrial biopsy were efficiently diagnosed and likely had their definitive surgical management via hysterectomy within 30 days of diagnosis. We found that even in women with a non-diagnostic office-based endometrial biopsy, the additional workup towards a tissue-based diagnosis prompted a substantial increase from 7-day cost to 90-day cost. This likely represents the diagnostic steps taken to convincingly determine the etiology of AUB or PMB after a non-diagnostic endometrial biopsy.

Prior studies attempting to evaluate the healthcare cost burden of such a common gynecologic disorder have relied primarily on estimates rather than specific clinic or hospital charges. In 2007, a systematic review of studies from 1997–2005 evaluated the impact of AUB utilizing estimations of the economic burden from a societal perspective with attention to health-related quality of life (HRQOL) measures [12]. This systematic review identified that one-third of all gynecologic visits are related to AUB with an estimated annual direct cost of $1.55 billion and indirect cost of $36 billion. Their cost assessment was important in quantifying the cost per year for AUB management in the United States. Their data was limited to qualitative measurements given the heterogeneity of data. The included studies did not stratify costs by patient characteristics or tissue-based diagnoses. The data from these studies are also now at least 15 years old. Further, their review included primarily patients at a substantially lower risk of EC by including many women less than 45 years of age. To our knowledge, we present the first study to quantify costs utilizing direct CPT and UB diagnosis codes and to standardize costs by Medicare reimbursement rates with dollars adjusted to a specific year to account for inflation. We also stratified costs by tissue-based diagnosis and were able to identify clinical variables that were associated with a higher cost.

Our study defined the top five drivers of overall cost related to perimenopausal AUB or PMB: operating room services, operative hysteroscopy, sterile supplies, hospital-based pharmacy services, and diagnostic hysteroscopy/laparoscopy. Not surprisingly, the highest costs were those that required operating room services, instruments with sterilization costs, and hospital-based pharmacy services. This highlights the limited cost-effective diagnostic techniques available to providers in the assessment of perimenopausal AUB and PMB. Room observation, a cost associated with hospital admission under outpatient observation status, was the fourth most expensive clinically relevant itemized cost. This cost is most related to patients that required minimally invasive surgery (MIS), such as robotic, laparoscopic, or vaginal hysterectomy, for either AH, EC, or benign indications. As hysterectomies performed through MIS approaches have become increasingly more common, we have seen an increase in same-day dismissals (SDD) for patients and therefore less costs for hospital observation. Given that all patients are not appropriate candidates for SDD, this cost may be more common in patients with medical comorbidities or more complex surgeries that would require further overnight hospital observation.

Our analysis revealed that among women with benign/DPE and polyps, the most common group in the study, higher BMI was associated with higher costs. While the clinical workup was determined by each patient’s provider, one could interpret this finding to be associated with the challenges that may exist in performing a pelvic exam in an ambulatory setting for a patient with a significantly elevated BMI. Compared to normal weight women, there is greater adiposity in the perineum and labia which can result in increased distance between the vulva and cervix [13]. This can result in more difficult pelvic exams that potentially require deeper instruments, significant retraction, or patient positioning that may not be tolerated in the office. Additionally, obesity doubles the lifetime risk of endometrial cancer and the greater cost may reflect the clinicians’ recognition that these women are at higher risk such that reassurances of benign findings required more thorough workup [14].

This study also raises the question as to whether there are alternative diagnostic approaches that may prove more cost effective that should be further studied. The utility of of endometrial cytology has garnered renewed interest given the technical advances of automated cytology and digital pathology though this remains limited by availability of endometrial cytopathologist expertise [15]. Additionally, studies are ongoing in the assessment of even less invasive modalities, such as patient self-sampling via devices such as an intravaginal tampon paired with sensitive molecular testing, as diagnostic tests in the setting of AUB/PMB [16, 17].

Strengths of our study include the large patient cohort size, patient inclusion based on symptoms at presentation rather than final diagnosis, and clinical data obtained prospectively directly from patients. We accounted for variability of reimbursement rates across various third-party payors by standardizing costs to Medicare reimbursement rates and adjusting for differences in inflation across study years.

Limitations of our study include the single institution experience and variability in clinical workup among physicians. In contrast to some practices, hysterosonogram is not commonly performed at our institution therefore this form of assessment was not accounted for in our study. By standardizing costs to Medicare dollars, our study likely underestimates overall healthcare system costs given the reimbursement variability among third-party payers. Additionally, we were unable to include the cost of workup completed prior to the date the patient presented to our gynecology clinic and enrolled in the study, such as any testing completed prior to referral. Another limitation is that our data collection did not include report of TVUS results, such as the presence or absence of findings that would guide clinical decisions for endometrial biopsy. With the inherent heterogeneity in the diagnostic assessment pathways of AUB/PMB, we recognize there may also be an opportunity for institution-specific or society-based guidelines; although such guidelines would require flexibility to still allow for clinical judgement and to account for resource disparities. Also, while we utilized clinical knowledge of what diagnostic tests would be related and unrelated to the AUB/PMB workup, we recognize that patients may have received evaluation for other clinical concerns or health maintenance during the study period. As such, even the best efforts to determine related and unrelated costs may have missed a relevant CPT or UB code.

In conclusion, there are substantial costs associated with the diagnostic assessment required to evaluate for the presence of EC and other endometrial pathologies in women who present with perimenopausal AUB or PMB. By quantifying the costs associated with diagnosis and management of underlying causes of perimenopausal AUB and PMB, we have defined relevant cost benchmarks toward which novel diagnostic strategies for AUB and PMB should aim.

Highlights.

• Cost for diagnostic evaluation and management of abnormal uterine bleeding is substantial

• Costs vary greatly according to ultimate tissue-based diagnosis

• Even reassuring benign findings that do not require further intervention yield substantial costs

• The development of sensitive, specific, and more cost-effective diagnostic strategies is warranted