{"paper_id":"8d4bc18e-0e26-401b-9d7b-eabd30fc4263","body_text":"ORIGINAL RESEARCH\nImpact of Endometriosis Diagnostic Delays\non Healthcare Resource Utilization and Costs\nEric Surrey . Ahmed M. Soliman . Helen Trenz . Cori Blauer-Peterson .\nAshley Sluis\nReceived: November 19, 2019 / Published online: January 20, 2020\n/C211The Author(s) 2020\nABSTRACT\nIntroduction: Endometriosis symptoms are\nnonspecific and overlap with other gynecologic\nand gastrointestinal diseases, leading to long\ndiagnostic delays. The burden of endometriosis\nhas been documented; however, little is known\nabout the impact of diagnostic delays on\nhealthcare costs leading up to diagnoses. The\npurpose of this study was to examine the eco-\nnomic impact of diagnostic delays on pre-diag-\nnosis healthcare utilization and costs among\npatients with endometriosis.\nMethods: This was a retrospective database\nstudy of adult patients with a diagnosis of\nendometriosis from 1 January 2004 to 31 July\n2016. Patients had continuous health plan\nenrollment 60 months prior to and 12 months\nfollowing the earliest endometriosis diagnosis\nand C 1 pre-diagnosis endometriosis symptom\n(dyspareunia, generalized pelvic pain, abdomi-\nnal pain, dysmenorrhea, or infertility). Patients\nwere assigned to short ( B 1 year), intermediate\n(1–3 years), or long (3–5 years) delay cohorts\nbased on the length of their diagnostic delay\n(time from first symptom to diagnosis).\nHealthcare resource utilization and costs were\ncalculated and compared by cohort in the\n60-month pre-diagnosis period.\nResults: A total of 11,793 patients were included\nin the study, of which 37.7% (4446/11,793),\n27.0% (3179/11,793), and 35.3% (4168/11,793)\nhad short, intermediate, and long delays,\nrespectively. Patients with intermediate or long\ndiagnostic delays had consistently more all-cause\nand endometriosis-related emergency visits and\ninpatient hospitalizations in the pre-diagnosis\nperiod than patients with short delays. Pre-diag-\nnosis all-cause healthcare costs were significantly\nhigher among patients with longer diagnostic\ndelays, averaging $21,489, $30,030, and $34,460\namong patients with a short, intermediate, and\nlong delay, respectively ( p \\ 0.001 for all pair-\nwise comparisons). Endometriosis-related costs\naccounted for 12.5% ($3553/$28,376) of all-\ncause costs and followed a similar pattern.\nConclusion: Patients with endometriosis who\nhad longer diagnostic delays had more pre-di-\nagnosis endometriosis-related symptoms and\nhigher pre-diagnosis healthcare utilization and\ncosts compared with patients who were\nEnhanced Digital Features To view enhanced digital\nfeatures for this article go to: https://doi.org/10.6084/\nm9.figshare.11417625.\nE. Surrey\nColorado Center for Reproductive Medicine, Lone\nTree, CO, USA\ne-mail: ESurrey@colocrm.com\nA. M. Soliman\nHealth Economics and Outcomes Research, AbbVie\nInc., North Chicago, IL, USA\nH. Trenz ( &) /C1C. Blauer-Peterson /C1A. Sluis\nHealth Economics and Outcomes Research, Optum,\nEden Prairie, MN, USA\ne-mail: helen.trenz@uhg.com\nAdv Ther (2020) 37:1087–1099\nhttps://doi.org/10.1007/s12325-019-01215-x\n\ndiagnosed earlier after symptom onset, provid-\ning evidence in support of earlier diagnosis.\nKeywords: Diagnostic delay; Endometriosis;\nHealthcare costs; Healthcare utilization;\nWomen’s health\nKey Summary Points\nWhy carry out this study?\nDue to non-specific symptoms that\noverlap with other gynecologic, urologic,\nand gastrointestinal issues, long\ndiagnostic delays are prevalent among\npatients with endometriosis\nLittle information is known about the\nimpact diagnostic delays may have on\nhealthcare costs leading up to diagnosis\nThe purpose of this study was to examine\nthe economic impact of diagnostic delays\non pre-diagnosis healthcare utilization\nand costs among patients with\nendometriosis\nWhat was learned from the study?\nPre-diagnosis all-cause and endometriosis-\nrelated healthcare costs were higher\namong patients with longer diagnostic\ndelays\nPatients with intermediate or long\ndiagnostic delays had consistently more\nall-cause and endometriosis symptom-\nrelated emergency visits and inpatient\nhospitalizations in the pre-diagnosis\nperiod than patients with short delays\nINTRODUCTION\nEndometriosis affects approximately 10% of\nreproductive-aged women [ 1–3] with symptoms\nof abdominal or pelvic pain, dysmenorrhea,\nmenstrual abnormalities, constipation, dysche-\nzia, dysuria, urinary frequency and urgency,\nand dyspareunia [ 4–6]. Chronic pain and infer-\ntility due to endometriosis have been shown to\nsignificantly decrease quality of life and increase\nphysical and psychologic morbidity [ 6–10].\nEndometriosis has also been associated with a\ntwofold or higher increased risk of developing\ncomorbidities including ovarian cysts, uterine\nfibroids, pelvic inflammatory disorder, intersti-\ntial cystitis, irritable bowel syndrome, consti-\npation, and ovarian and endometrial cancers\n[11].\nEndometriosis diagnosis often presents\nchallenges as symptoms are nonspecific and\noverlap with other gynecologic, urologic, and\ngastrointestinal issues resulting in long diag-\nnostic delays. Nnoaham et al. documented an\naverage diagnostic delay of 6.7 years among\npatients with endometriosis mainly due to\ndelays in referral from primary care to a spe-\ncialist [ 12]. Soliman et al. reported an average\ndiagnostic delay of 4.4 years with 89% of diag-\nnoses made by obstetricians/gynecologists [ 13].\nSeveral reasons for endometriosis diagnostic\ndelays have been identified including both\npatient-centered causes such as embarrassment,\nstigmatization, tolerance, or uncertainty of\nnormal versus abnormal symptoms and physi-\ncian-centered causes such as normalization of\npatient symptoms and reliance on inadequate\ndiagnostic methods [ 14]. Clinical guidelines do\nnot provide a consistent approach to the diag-\nnosis and management of endometriosis with\nlittle attention given to the presence of comor-\nbidities, which may contribute to diagnostic\ndelays [ 15, 16].\nPatients with endometriosis experience sig-\nnificant healthcare expenses. In a study of\nendometriosis patients and matched controls,\nmean annual adjusted direct healthcare costs\nwere more than three times higher in\nendometriosis patients than controls during the\n12 months following diagnosis ($16,573 versus\n$4733, p \\ 0.005) [ 17]. Fuldeore et al. found\nthat costs were highest in the first year follow-\ning an endometriosis diagnosis, costing $13,199\ncompared with $6041 in the year prior to\ndiagnosis and $6720 in the year following the\nindex year. Additionally, in the 5 years prior to\nan endometriosis diagnosis, costs were $7028\nhigher among patients with endometriosis\n1088 Adv Ther (2020) 37:1087–1099\n\ncompared with matched controls without\nendometriosis [ 18].\nThe economic burden of endometriosis has\nbeen well documented in the literature; how-\never, little is known about the impact a diag-\nnostic delay may have on healthcare costs\nleading up to diagnosis. This study focused on\nthe time period prior to diagnosis to attempt to\naddress this gap. The purpose of this study was\nto examine the economic impact of diagnostic\ndelays on pre-diagnosis healthcare utilization\nand costs among patients with endometriosis.\nMETHODS\nStudy Design and Data Source\nThis retrospective study used the Optum\nResearch Database, a geographically diverse US\ndatabase representing approximately 67 million\nindividuals from 1993 to present. Medical and\npharmacy claims and enrollment information\nwere obtained from 1 January 1999 to 31 July\n2017 (study period). Medical claims consisted of\nInternational Classification of Disease, Ninth\nand Tenth Revisions, Clinical Modification\n(ICD-9-CM and ICD-10-CM) diagnosis and\nprocedure codes, Healthcare Common Proce-\ndure Coding System (HCPCS) codes, and rev-\nenue codes. Pharmacy claims included National\nDrug Codes for filled prescriptions and days and\nquantity of drug supplied. The Optum Research\nDatabase is fully de-identified and HIPAA com-\npliant and did not require Institutional Review\nBoard approval or waiver of authorization.\nStudy Population\nPatients were required to be aged 18–49 years\nand have C 1 medical claim for endometriosis\nin any position (ICD-9-CM/ICD-10-CM diag-\nnosis code 617.x/N80.x) from 1 January 2004 to\n31 July 2016 (identification period). The date of\nthe first medical claim with an endometriosis\ndiagnosis code was considered the index date.\nPatients were required to have continuous\nhealth plan enrollment with medical and\npharmacy benefits for C 60 months (1825 days)\nprior to the index date (pre-diagnosis period)\nand C 12 months (365 days) following the\nindex date and a medical claim for C 1\nendometriosis symptom in any position (dys-\npareunia, generalized pelvic pain, abdominal\npain, dysmenorrhea, or infertility) during the\npre-diagnosis period. A pre-diagnosis period of 5\nyears was selected based on recent evidence\nfrom Soliman et al. who reported the average\ndelay from first symptom to endometriosis\ndiagnosis was approximately 4.4 years [ 13].\nEndometriosis symptoms were selected based\non published literature [ 4, 5] and guidance from\nthe clinician author. Patients with a medical\nclaim for endometriosis or malignancy prior to\nthe index date were excluded from the study. To\nrule out patients with conditions that have\nsymptoms similar to endometriosis, patients\nwith an ICD-9/ICD-10 diagnosis code for geni-\ntourinary or intra-abdominal infection (e.g.,\nchlamydia, gonorrhea), inflammatory bowel\ndisease, diverticulitis, appendicitis, peritonitis,\nother genitourinary conditions (cystitis,\nurethritis), or kidney stones any time prior to\nthe index date were also excluded.\nPatients were assigned to delay cohorts based\non the length of time from the date of the first\nmedical claim for a non-diagnostic service for\nan endometriosis symptom to the index date\ncategorized as: short delay ( B 1 year), interme-\ndiate delay (1–3 years), and long delay (3–-\n5 years). These cutoffs are conservative\nestimates of disease burden and were chosen to\nbalance the clinical burden and sample size\nrequirements for the study.\nStudy Measures\nPatient Characteristics\nDemographic and clinical characteristics that\nincluded patient age, geographic region, insur-\nance type (commercial or Medicare Advantage),\nlength of diagnostic delay, disease severity\n(proxy based on the annualized count of\nendometriosis symptoms), and targeted\nendometriosis-related comorbid conditions\nwere measured using claims data during the pre-\ndiagnosis period.\nAdv Ther (2020) 37:1087–1099 1089\n\nPre-Diagnosis Healthcare Resource Utilization\nAll-cause and endometriosis-related healthcare\nresource utilization was calculated as the mean\nnumber of ambulatory (office and outpatient)\nvisits, emergency visits, and inpatient stays\nduring the 60-month pre-diagnosis period.\nUtilization was considered endometriosis-re-\nlated if the medical claim included a diagnosis\ncode for an endometriosis symptom (dyspareu-\nnia, generalized pelvic pain, abdominal pain,\ndysmenorrhea, or infertility), a Current Proce-\ndural Terminology (CPT) or HCPCS code for an\nendometriosis treatment procedure (e.g.,\nlaparoscopy), or a HCPCS code for an\nendometriosis-related medication administra-\ntion (e.g., neuropathic pain agent, progestin,\nhormonal contraceptive, non-steroidal anti-in-\nflammatory drug) in a physician’s office.\nPre-Diagnosis Healthcare Costs\nAll-cause and endometriosis-related healthcare\ncosts were calculated as the combined health\nplan and patient-paid amounts during the\n60-month pre-diagnosis period adjusted for\ninflation from 1999 to 2016 using the annual\nmedical care component of the Consumer Price\nIndex (CPI) [ 19]. Costs were considered\nendometriosis-related if the medical claim\nincluded a diagnosis for an endometriosis\nsymptom, a CPT or HCPCS code for an\nendometriosis treatment, or a pharmacy claim\nfor a medication used to treat endometriosis or\nits symptoms (e.g., neuropathic pain agent,\nprogestin, hormonal contraceptive, non-ster-\noidal anti-inflammatory drugs). The diagnosis\ncode on the medical claim for the endometrio-\nsis symptom must have been in the primary\nposition to be considered endometriosis-related\nemergency or inpatient costs.\nStatistical Analyses\nAll study variables were analyzed descriptively,\nand comparisons between delay cohorts were\nmade. Mean healthcare utilization and costs\nover the 60-month pre-diagnosis period were\ncalculated and presented separately for each of\nthe three diagnostic delay cohorts. Statistical\ntests of significance for differences across the\nthree cohorts were conducted using chi-square\ntests for categorical variables and ANOVA and\nt test for continuous variables. For endometrio-\nsis-specific healthcare resource utilization and\ncosts, the hypothesis tested was whether\npatients who have had endometriosis symp-\ntoms for a longer period of time would have\nmean utilization and costs equal to patients\nwho have had endometriosis symptoms for a\nshorter period of time. Calculated p-values were\nadjusted for multiple comparisons (Bonferroni\ncorrection) with a threshold of statistical sig-\nnificance of p \\ 0.017.\nRESULTS\nDemographic and Clinical Characteristics\nAfter applying inclusion and exclusion criteria,\n11,793 patients were included in the study, of\nwhich 37.7% ( n = 4446) had a short delay,\n27.0% ( n = 3179) had an intermediate delay,\nand 35.3% ( n = 4168) had a long delay (Fig. 1).\nPatients with a short delay were slightly older\n(39.8 ± 7.0) than patients who had intermedi-\nate (38.9 ± 7.8) or long delays (38.9 ± 7.6)\n(p \\ 0.001 for both comparisons) (Table 1).\nApproximately half of all patients resided in the\nFig. 1 Patient sample selection\n1090 Adv Ther (2020) 37:1087–1099\n\nSouth, and 66% had a point of service health-\ncare plan ( p value not significant).\nPatients in this study had a mean diagnostic\ndelay of 763.9 ± 631.0 days (2.09 ± 1.77 years)\n(Table 1). Patients with a short, intermediate, or\nlong delay averaged 90.2 days, 733.4 days, and\n1505.9 days, respectively, from the onset of\nendometriosis symptoms until diagnosis. Com-\nmon symptoms identified were abdominal pain\n(67.3%), dysmenorrhea (52.0%), and dyspareunia\n(13.0%) (Table 1). Patients with a short delay were\nleast likely to have abdominal pain and infertility,\nbut were most likely to have dysmenorrhea com-\npared with patients who had intermediate and\nlong delays. The proportion of patients with\nabdominal pain increased significantly with\nincreasing diagnostic delay ( p \\ 0.001 for all\ncomparisons). Using the proxy for disease sever-\nity, patients with a long delay had a less concen-\ntrated presence of endometriosis symptoms than\nthose with shorter delays ranging from a symp-\ntom severity of 0.3–1.0 (p \\ 0.001 for all compar-\nisons; Table 1).\nAlmost all patients (95.8%) had C 1 comor-\nbid condition (Table 1), with the most common\nbeing fatigue/neurasthenia (49.2%), headache\nand migraine (42.8%), ovarian cysts (40.6%),\nurinary tract infections (38.8%), depression and\nanxiety (37.6%), and uterine fibroids (34.2%).\nComorbidities tended to be the highest among\npatients with longer delays.\nPre-Diagnosis Healthcare Utilization\nAll-Cause Utilization\nAlmost all patients had C 1 all-cause ambula-\ntory visit during the pre-diagnosis period\n(Table 2). The mean number of ambulatory\nvisits increased with longer diagnostic delays\nfrom 47.3 visits among patients with a short\ndelay, 61.0 visits in patients with an interme-\ndiate delay, to 69.1 visits among patients with a\nlong delay ( p \\ 0.001 for all comparisons).\nAlmost 66% of patients had an emergency room\nvisit, and there was an average of 4.2 visits over\nthe 60-month pre-diagnosis period. The pro-\nportion of patients with an emergency room\nvisit increased significantly with longer diag-\nnostic delays ranging from 58.0% in those with\na short delay, 69.0% in those with an interme-\ndiate delay, and 71.9% in patients with a long\ndelay (p B 0.007 for all comparisons). The mean\nnumber of emergency room visits was signifi-\ncantly lower in patients with a short delay\ncompared with patients with an intermediate or\nlong delay ( p \\ 0.001 for both comparisons).\nApproximately 22% of patients had an inpa-\ntient stay during the pre-diagnosis period. Both\nthe proportion of patients with an inpatient\nstay and the mean number of stays during the\npre-diagnosis period increased as the length of\ndiagnostic delay increased ( p \\ 0.001 for all\ncomparisons).\nEndometriosis-Related Utilization\nApproximately 92% of patients had an\nendometriosis-related ambulatory visit during\nthe pre-diagnosis period (Table 2). The number\nof ambulatory visits increased with longer\ndiagnostic delays ranging from 2.4 visits among\nshort delay patients, 5.0 visits among interme-\ndiate delay patients, and 6.6 visits among long\ndelay patients ( p \\ 0.001 for all comparisons).\nOverall, one in six patients had an\nendometriosis-related emergency visit during\nthe pre-diagnosis period. Patients with longer\ndelays were more likely to have an\nendometriosis-related emergency room visit\n(p \\ 0.001 for all comparisons) and a greater\nnumber of endometriosis-related emergency\nvisits ( p \\ 0.001 for all comparisons).\nEndometriosis-related inpatient stays were rare\nand increased with longer diagnostic delays ( p\nB 0.007 for all comparisons).\nPre-Diagnosis Healthcare Costs\nAll-Cause Costs\nAll-cause healthcare costs during the pre-diag-\nnosis period averaged $28,376 (Fig. 2). Ambu-\nlatory costs were the major cost driver\naccounting for 57.7% of total all-cause costs.\nAll-cause costs were significantly higher in\npatients with longer diagnostic delays. Mean\ntotal costs in patients with a long delay were\n60.4% and 14.8% higher than costs in patients\nwith a short and intermediate delay, respec-\ntively ( p \\ 0.001 for all comparisons). All-cause\nAdv Ther (2020) 37:1087–1099 1091\n\nTable 1 Patient demographic and clinical characteristics during the 60-month pre-diagnosis period\nTotal\n(n = 11,793)\nShort delay\n(n = 4446)\nIntermediate delay\n(n = 3179)\nLong delay\n(n = 4168)\nShort versus\nintermediate\np value\nShort\nversus long\np value\nIntermediate\nversus long\np value\nAge, mean (SD) 39.3 (7.4) 39.8 (7.0) 38.9 (7.8) 38.9 (7.6) \\ 0.001 \\ 0.001 0.940\nRegion, n (%) 0.747 0.025 0.188\nNortheast 850 (7.2) 291 (6.6) 218 (6.9) 341 (8.2)\nMidwest 3301 (28.0) 1263 (28.4) 885 (27.8) 1153 (27.7)\nSouth 5892 (50.0) 2246 (50.5) 1591 (50.1) 2055 (49.3)\nWest 1748 (14.8) 646 (14.5) 485 (15.3) 617 (14.8)\nOther 2 (0.0) 0 (0.0) 0 (0.0) 2 (0.1)\nLength of diagnostic delay (days),\nmean (SD)\n763.9 (631.0) 90.2 (103.6) 733.4 (213.7) 1505.9 (212.0) \\ 0.001 \\ 0.001 \\ 0.001\nPresence of endometriosis symptoms, n (%)\nDysmenorrhea 6132 (52.0) 2420 (54.4) 1578 (49.6) 2134 (51.2) \\ 0.001 0.003 0.185\nDyspareunia 1536 (13.0) 562 (12.6) 413 (13.0) 561 (13.5) 0.651 0.259 0.558\nPelvic pain 346 (2.9) 145 (3.3) 81 (2.6) 120 (2.9) 0.070 0.305 0.389\nAbdominal pain 7935 (67.3) 2156 (48.5) 2354 (74.1) 3425 (82.2) \\ 0.001 \\ 0.001 \\ 0.001\nInfertility 1414 (12.0) 374 (8.4) 421 (13.2) 619 (14.9) \\ 0.001 \\ 0.001 0.050\nCount of endometriosis symptoms,\na\nmean (SD)\n1.5 (0.7) 1.3 (0.5) 1.5 (0.7) 1.7 (0.7) \\ 0.001 \\ 0.001 \\ 0.001\nSymptom severity proxy (annualized\ncount), mean (SD)\n0.6 (0.5) 1.0 (0.7) 0.5 (0.2) 0.3 (0.1) \\ 0.001 \\ 0.001 \\ 0.001\nMost common endometriosis-related comorbidities, n (%)\nFatigue neurasthenia 5804 (49.2) 1912 (43.0) 1617 (50.9) 2275 (54.6) \\ 0.001 \\ 0.001 0.002\nHeadache and migraine 5046 (42.8) 1566 (35.2) 1417 (44.6) 2063 (49.5) \\ 0.001 \\ 0.001 \\ 0.001\nOvarian cysts 4790 (40.6) 1604 (36.1) 1314 (41.3) 1872 (44.9) \\ 0.001 \\ 0.001 0.002\n1092 Adv Ther (2020) 37:1087–1099\n\nTable 1 continued\nTotal\n(n = 11,793)\nShort delay\n(n = 4446)\nIntermediate delay\n(n = 3179)\nLong delay\n(n = 4168)\nShort versus\nintermediate\np value\nShort versus\nlong\np value\nIntermediate\nversus long\np value\nUrinary tract\ninfection\n4570 (38.8) 1374 (30.9) 1350 (42.5) 1846 (44.3) \\ 0.001 \\ 0.001 0.118\nDepression and\nanxiety\n4437 (37.6) 1428 (32.1) 1223 (38.5) 1786 (42.9) \\ 0.001 \\ 0.001 \\ 0.001\nUterine fibroids 4034 (34.2) 1515 (34.1) 1089 (34.3) 1430 (34.2) 0.870 0.819 0.962\nCount of comorbid conditions, n (%)\n0 491 (4.2) 289 (6.5) 97 (3.1) 105 (2.5) \\ 0.001 \\ 0.001 0.167\n1 1467 (12.4) 766 (17.2) 343 (10.8) 358 (8.6) \\ 0.001 \\ 0.001 0.001\n2? 9835 (83.4) 3391 (76.3) 2739 (86.2) 3705 (88.9) \\ 0.001 \\ 0.001 \\ 0.001\nSD standard deviation\na Endometriosis symptoms included dysmenorrhea, dyspareunia, pelvic pain, abdominal pain, and infertility\nAdv Ther (2020) 37:1087–1099 1093\n\nTable 2 Healthcare resource utilization during the 60-month pre-diagnosis period\nTotal\n(n = 11,793)\nShort delay\n(n = 4446)\nIntermediate\ndelay\n(n = 3179)\nLong delay\n(n = 4168)\nShort versus\nintermediate\np value\nShort\nversus\nlong\np value\nIntermediate\nversus long\np value\nAll cause\nAmbulatory\nvisit count,\nmean (SD)\n58.7 (44.6) 47.3 (35.9) 61.0 (45.3) 69.1 (49.4) \\ 0.001 \\ 0.001 \\ 0.001\nProportion\nwith C 1\nvisit, n (%)\n11,790 (100.0) 4443 (99.9) 3179 (100.0) 4168 (100.0) 0.143 0.093 –\nEmergency visit\ncount, mean\n(SD)\n4.2 (12.0) 3.3 (9.7) 4.6 (12.2) 5.0 (13.9) \\ 0.001 \\ 0.001 0.177\nProportion\nwith C 1\nvisit, n (%)\n7769 (65.9) 2580 (58.0) 2193 (69.0) 2996 (71.9) \\ 0.001 \\ 0.001 0.007\nInpatient stay\ncount, mean\n(SD)\n0.3 (0.7) 0.2 (0.6) 0.3 (0.7) 0.4 (0.9) \\ 0.001 \\ 0.001 \\ 0.001\nProportion\nwith C 1\nstay, n (%)\n2541 (21.6) 763 (17.2) 685 (21.6) 1093 (26.2) \\ 0.001 \\ 0.001 \\ 0.001\nEndometriosis\nrelated\nAmbulatory\nvisit count,\nmean (SD)\n4.6 (5.8) 2.4 (2.9) 5.0 (5.5) 6.6 (7.4) \\ 0.001 \\ 0.001\n\\ 0.001\nProportion\nwith C 1\nvisit, n (%)\n10,829 (91.8) 3599 (81.0) 3130 (98.5) 4100 (98.4) \\ 0.001 \\ 0.001 0.760\nEmergency visit\ncount, mean\n(SD)\na\n0.2 (0.8) 0.1 (0.5) 0.3 (0.9) 0.4 (0.9) \\ 0.001 \\ 0.001 \\ 0.001\nProportion\nwith C 1\nvisit, n (%)a\n1985 (16.8) 422 (9.5) 594 (18.7) 969 (23.3) \\ 0.001 \\ 0.001 \\ 0.001\n1094 Adv Ther (2020) 37:1087–1099\n\npharmacy costs in patients with a short delay\n($4351) were significantly lower than costs in\npatients with an intermediate ($5565) or long\ndelay ($6106) ( p \\ 0.001 for both comparisons).\nAll-cause medical costs were also significantly\nhigher with longer diagnostic delays ($17,138,\n$24,465, and $28,354 in patients with a short,\nintermediate, and long delay, respectively)\n(p \\ 0.001 for all comparisons).\nEndometriosis-Related Costs\nPre-diagnosis endometriosis-related healthcare\ncosts accounted for 12.5% of all-cause costs.\nThis proportion was highest among patients\nwith longer diagnostic delays with values of\n9.7%, 13.3%, and 13.9% in patients with short,\nintermediate, and long delays, respectively. The\nTable 2 continued\nTotal\n(n = 11,793)\nShort delay\n(n = 4446)\nIntermediate\ndelay\n(n = 3179)\nLong delay\n(n = 4168)\nShort versus\nintermediate\np value\nShort\nversus\nlong\np value\nIntermediate\nversus long\np value\nInpatient stay\ncount, mean\n(SD)\na\n0.03 (0.2) 0.02 (0.2) 0.03 (0.2) 0.05 (0.3) 0.007 \\ 0.001 0.002\nProportion\nwith C 1 stay,\nn (%)a\n353 (3.0) 85 (1.9) 90 (2.8) 178 (4.3) 0.008 \\ 0.001 0.001\nSD standard deviation\na Diagnosis code for the endometriosis symptom had to in the primary position on the claim\nFig. 2 All-cause healthcare costs over the 60-month pre-\ndiagnosis period. ap \\ 0.017 in comparison of the short\nand intermediate delay cohorts. bp \\ 0.017 in comparison\nof short and long delay cohorts. cp \\ 0.017 in comparison\nof intermediate and long delay cohorts\nFig. 3 Endometriosis-related healthcare costs over the\n60-month pre-diagnosis period. aDiagnosis code for the\nendometriosis symptom had to in the primary position on\nthe claim to be considered endometriosis-related.\nbp \\\n0.017 in comparison of short and intermediate delay\ncohorts. cp \\ 0.017 in comparison of short and long delay\ncohorts. dp \\ 0.017 in comparison of intermediate and\nlong delay cohorts\nAdv Ther (2020) 37:1087–1099 1095\n\nmajor cost driver was ambulatory visits\naccounting for 59.1% of total endometriosis-\nrelated costs (Fig. 3). Compared with patients\nwith a short delay, pre-diagnosis endometriosis-\nrelated costs were almost double and more than\ndouble those in patients with intermediate or\nlong delays, respectively ( p \\ 0.001 for both\ncomparisons). Endometriosis-related pharmacy\ncosts were also highest among patients with a\nlong delay and lowest among those with a short\ndelay ($683 versus $568, p \\ 0.001).\nDISCUSSION\nDiagnostic delays among patients with\nendometriosis have been well documented, but\nlittle information is known about the economic\nimpact these delays have on the patient and\nhealthcare system. This study identified\npatients with endometriosis stratified into\ncohorts defined by the length of time from the\nfirst claim for an endometriosis symptom to an\nendometriosis diagnosis. Patients with longer\ndiagnostic delays had a significantly higher\nclinical burden with more endometriosis-re-\nlated symptoms and comorbidities and a greater\neconomic burden due to significantly higher\nhealthcare resource utilization and costs com-\npared with patients with shorter delays.\nAll-cause and endometriosis-related health-\ncare resource utilization increased with longer\ndiagnostic delays. Patients had an average of\n11.7 all-cause annualized ambulatory visits, 0.8\nall-cause annualized emergency visits, and 0.1\nall-cause annualized inpatient stay during the\n60-month pre-diagnosis period. Similarly, Soli-\nman et al. [ 17] reported that patients with\nendometriosis averaged 9.9 office/obstetrics-gy-\nnecology visits, 0.6 emergency visits, and 0.1\ninpatient admissions in the 12 months prior to\ndiagnosis. Utilization of healthcare services was\nalso comparable to results described by Fuldeore\net al. [ 18] who found emergency visits, outpa-\ntient visits, and physician visits increased over a\n5-year period prior to endometriosis diagnosis\npeaking in the year immediately prior to diag-\nnosis. In our study, patients with the longest\ndiagnostic delay had 38% more all-cause\nambulatory visits, 52% more all-cause\nemergency visits, and 100% more all-cause\ninpatient stays during the 60-month pre-diag-\nnosis period compared with patients who had\nthe shortest diagnostic delay. Future studies are\nneeded to assess the economic impact of diag-\nnostic delays post-endometriosis diagnosis.\nAll-cause and endometriosis-related health-\ncare costs increased with longer diagnostic\ndelays. Patients with long diagnostic delays had\n60% higher mean all-cause costs compared with\npatients with a short delay and 15% higher costs\ncompared with patients with an intermediate\ndelay. Mean annual costs in the 5 years prior to\ndiagnosis ranged from $4298 in patients with a\nshort delay to $6892 in patients with a long\ndelay. These costs were similar to costs pre-\nsented by Fuldeore et al., which ranged from\n$3730 (adjusted to 2016 USD) in the 5th year\nprior to endometriosis diagnoses to $6649 (ad-\njusted to 2016 USD) in the year immediately\nprior to diagnoses [ 18]. It is possible that the\nhigher costs seen in patients with a long diag-\nnostic delay in this study could be a result of the\ngreater number of endometriosis-related\ncomorbidities found in those patients.\nEndometriosis-related costs accounted for\napproximately 12.5% of total all-cause costs in\nthe pre-diagnosis period. Ambulatory costs\naccounted for more than half of total\nendometriosis-related costs. Similar to all-cause\ncosts, patients with the longest diagnostic\ndelays experienced 130% higher endometriosis-\nrelated costs compared with patients with the\nshortest delays.\nPatients with long diagnostic delays had\nmore claims for endometriosis symptoms and\nendometriosis-related comorbidities over the\n60-month pre-diagnosis period. The increased\npresence of comorbidities with similar sympto-\nmology to endometriosis may have further\ncomplicated the diagnosis of endometriosis\nleading to a longer diagnostic delay. Patients\nwith a long delay also had significantly more\nendometriosis symptoms over the pre-diagnosis\nperiod, most notably abdominal pain and\ninfertility. Nnoaham et al. [ 12] found that\ndiagnostic delays were significantly longer in\npatients with more pelvic symptoms, which was\nconsistent with results observed in our study.\nWhile patients with long delays experienced\n1096 Adv Ther (2020) 37:1087–1099\n\nmore symptoms and comorbidities over the\n5-year pre-diagnosis period, patients with a\nshort delay had more concentrated\nendometriosis symptoms in the year they\nexperienced symptoms, which may have facili-\ntated an earlier diagnosis.\nThe results of this study highlight the sig-\nnificant pre-diagnostic clinical and economic\nimpact of diagnostic delays on patients with\nendometriosis. Several approaches have been\ninvestigated to shorten the diagnostic delay,\nincluding earlier detection of endometriosis\nsymptoms through increased physician aware-\nness and training, use of non-surgical methods\nof diagnosis (i.e., transvaginal ultrasound), and\nearly treatment interventions based on symp-\ntoms, signs, and clinical findings prior to con-\nfirmation with laparoscopy [ 6, 13, 20–22]. Due\nto the hidden economic burden associated with\nthe delay in the diagnosis of endometriosis and\nthe important implications it has for healthcare\ndecision makers, physicians, and payers, future\nresearch is needed in this area to determine if\nearlier detection of endometriosis using the\nabove approaches can reduce this burden.\nLimitations\nThere are several limitations to this study.\nHealthcare claims are collected for the purpose\nof payment, not research, which leads to several\ninherent limitations. The presence of an\nendometriosis diagnosis code on a medical\nclaim is not proof of the presence of disease. The\ndiagnosis code may be incorrectly coded or\nincluded as rule-out criteria rather than actual\ndisease. It is possible that patients may have had\na diagnosis of endometriosis prior to the base-\nline period, which may explain the older age of\nonset found in this study. The baseline period\nwas extended to 5 years to minimize this risk.\nAdditionally, endometriosis symptoms may not\nhave been fully captured in the claims database.\nEndometriosis-related healthcare utilization\nand costs in the pre-diagnosis period were based\non the presence of claims for five common\nendometriosis symptoms and endometriosis-\nrelated surgical and pharmacologic treatment.\nWhile this was done to provide conservative\nestimates, it is possible that the true costs and\nutilization due to endometriosis were higher in\nthis population. This study included a managed\ncare population and may not be generalizable to\nother populations. Additionally, due to the\ncoverage of the health plan underlying the\nclaims database, about half of the study patients\nwere from the South. Since racial and ethnic\ndata were not collected, we cannot know if this\nmay have skewed study results. Lastly, due to\nthe observational nature of this study and the\ndescriptive analyses performed, it is possible\nthat confounding factors not accounted for\ncould contribute to the difference in costs and\nutilization between the diagnostic delay\ncohorts.\nCONCLUSIONS\nPatients with endometriosis with longer diag-\nnostic delays had more pre-diagnosis\nendometriosis-related symptoms and comor-\nbidities and higher pre-diagnosis healthcare\nresource utilization and costs compared with\npatients who were diagnosed sooner after\nsymptom onset. Future research is needed to\ndifferentiate costs related to comorbidities\nassociated with endometriosis versus those\nrelated to disease management and how thera-\npeutic interventions directed specifically at dis-\nease management and early diagnosis can\nimpact these costs. Further future research is\nneeded to determine the impact of diagnostic\ndelays in patients with endometriosis on quality\nof life, productivity losses, and relationships.\nACKNOWLEDGEMENTS\nFunding. This study and the journal’s Rapid\nService and Open Access Fees were funded by\nAbbVie Inc. AbbVie participated in the study\ndesign; data collection, analysis, and interpre-\ntation; and review and approval of the final\nmanuscript for publication. All authors had full\naccess to all of the study results and take com-\nplete responsibility for the integrity of the\nresults and accuracy of the data analysis.\nAdv Ther (2020) 37:1087–1099 1097\n\nMedical Writing, Editorial, and Other\nAssistance. Medical writing and editorial assis-\ntance was provided by Deja Scott-Shemon,\nMPH, an employee of Optum. This assistance\nwas funded by AbbVie Inc. Authors would also\nlike to acknowledge Carolyn Martin for her\nassistance with analytic interpretation and\nmanuscript review and Susan Peckous for her\nassistance with project management and dis-\nsemination of study results.\nAuthorship. All named authors meet the\nInternational Committee of Medical Journal\nEditors (ICMJE) criteria for authorship for this\narticle, take responsibility for the integrity of\nthe work as a whole, and have given their\napproval for this version to be published.\nDisclosures. Eric Surrey has served as a con-\nsultant for AbbVie, has been a member of the\nAbbVie Inc. and Ferring speakers bureau, and\nserves on an advisory board for DOT Laborato-\nries. Ahmed M. Soliman is an employee of and\nowns stock in AbbVie Inc. Cori Blauer-Peterson\nand Ashley Sluis are employees of Optum and\nwere funded by AbbVie Inc. to conduct the\nstudy. Helen Trenz was an employee of Optum\nat the time this study was conducted and is\ncurrently employed by UnitedHealth Group.\nCompliance with Ethics Guidelines. The\nOptum Research Database is fully de-identified\nand HIPAA compliant and did not require\nInstitutional Review Board approval or waiver\nof authorization.\nData Availability. The data contained in\nour database contain proprietary elements\nowned by Optum and therefore cannot be\nbroadly disclosed or made publicly available at\nthis time. The disclosure of these data to third\nparty clients assumes certain data security and\nprivacy protocols are in place and that the third\nparty client has executed our standard license\nagreement which includes restrictive covenants\ngoverning the use of the data.\nOpen Access. This article is licensed under a\nCreative Commons Attribution-NonCommer-\ncial 4.0 International License, which permits\nany non-commercial use, sharing, adaptation,\ndistribution and reproduction in any medium\nor format, as long as you give appropriate credit\nto the original author(s) and the source, provide\na link to the Creative Commons licence, and\nindicate if changes were made. The images or\nother third party material in this article are\nincluded in the article’s Creative Commons\nlicence, unless indicated otherwise in a credit\nline to the material. If material is not included\nin the article’s Creative Commons licence and\nyour intended use is not permitted by statutory\nregulation or exceeds the permitted use, you\nwill need to obtain permission directly from the\ncopyright holder. To view a copy of this licence,\nvisit http://creativecommons.org/licenses/by-\nnc/4.0/.\nREFERENCES\n1. Quan H, Li B, Couris CM, et al. Updating and val-\nidating the Charlson comorbidity index and score\nfor risk adjustment in hospital discharge abstracts\nusing data from 6 countries. Am J Epidemiol.\n2011;173:676–82.\n2. Eskenazi B, Warner ML. Epidemiology of\nendometriosis. Obstet Gynecol Clin North Am.\n1997;24:235–58.\n3. Cramer DW, Missmer SA. The epidemiology of\nendometriosis. Ann N Y Acad Sci. 2002;955:11–22\ndiscussion 34-6, 396-406 .\n4. Fuldeore MJ, Soliman AM. Prevalence and symp-\ntomatic burden of diagnosed endometriosis in the\nUnited States: National estimates from a cross-sec-\ntional survey of 59,411 women. Gynecol Obstet\nInvest. 2017;82:453–61.\n5. De Graaff AA, D’Hooghe TM, Dunselman GA, et al.\nThe significant effect of endometriosis on physical,\nmental and social wellbeing: results from an inter-\nnational cross-sectional survey. Hum Reprod.\n2013;28:2677–85.\n6. Angalakuditi M, Edgell E, Beardsworth A, Buysman\nE, Bancroft T. Treatment patterns and resource\nutilization and costs among patients with pul-\nmonary arterial hypertension in the United States.\nJ Med Econ. 2010;13:393–402.\n7. Dunselman GA, Vermeulen N, Becker C, et al.\nESHRE guideline: management of women with\nendometriosis. Hum Reprod. 2014;29:400–12.\n1098 Adv Ther (2020) 37:1087–1099\n\n8. Soliman AM, Coyne KS, Gries KS, Castelli-Haley J,\nSnabes MC, Surrey ES. The effect of endometriosis\nsymptoms on absenteeism and presenteeism in the\nworkplace and at home. J Manag Care Spec Pharm.\n2017;23:745–54.\n9. Soliman AM, Coyne KS, Zaiser E, Castelli-Haley J,\nFuldeore MJ. The burden of endometriosis symp-\ntoms on health-related quality of life in women in\nthe United States: a cross-sectional study. J Psycho-\nsom Obstet Gynaecol. 2017;38:238–48.\n10. Winkel CA. Evaluation and management of women\nwith endometriosis. Obstet Gynecol. 2003;102:\n397–408.\n11. Surrey ES, Soliman AM, Johnson SJ, Davis M, Cas-\ntelli-Haley J, Snabes MC. Risk of developing\ncomorbidities among women with endometriosis: a\nretrospective matched cohort study. J Womens\nHealth (Larchmt). 2018;27:1114–23.\n12. Nnoaham KE, Hummelshoj L, Webster P, et al.\nImpact of endometriosis on quality of life and work\nproductivity: a multicenter study across ten coun-\ntries. Fertil Steril. 2011;96(366–73):e8.\n13. Soliman AM, Fuldeore M, Snabes MC. Factors\nassociated with time to endometriosis diagnosis in\nthe United States. J Womens Health (Larchmt).\n2017;26:788–97.\n14. Ballard K, Lowton K, Wright J. What’s the delay? A\nqualitative study of women’s experiences of reach-\ning a diagnosis of endometriosis. Fertil Steril.\n2006;86:1296–301.\n15. Ay W. The significance of diagnostic delay in\nendometriosis. MOJ Womens Health. 2016;2:10–1.\n16. Seear K. The etiquette of endometriosis: stigmati-\nsation, menstrual concealment and the diagnostic\ndelay. Soc Sci Med. 2009;69:1220–7.\n17. Soliman AM, Surrey E, Bonafede M, Nelson JK,\nCastelli-Haley J. Real-world evaluation of direct and\nindirect economic burden among endometriosis\npatients in the United States. Adv Ther. 2018;35:\n408–23.\n18. Fuldeore M, Yang H, Du EX, Soliman AM, Wu EQ,\nWinkel C. Healthcare utilization and costs in\nwomen diagnosed with endometriosis before and\nafter diagnosis: a longitudinal analysis of claims\ndatabases. Fertil Steril. 2015;103:163–71.\n19. Clinical Classification Software (CCS) for ICD-9-\nCM. Agency for Healthcare Research and Quality,\nRockville, MD. http://www.hcup-us.ahrq.gov/\ntoolssoftware/ccs/ccs.jsp. Comorbidities defined by\nHealthcare Cost and Utilization Project (HCUP)\ncomorbidity software. 2015. Agency for Healthcare\nResearch and Quality (AHRQ), Rockville, MD. www.\nhcup-us.ahrq.gov/toolssoftware/comorbidity/\ncomorbidity.jsp. Accessed 1 Aug 2019.\n20. Agarwal SK, Chapron C, Giudice LC, et al. Clinical\ndiagnosis of endometriosis: a call to action. Am J\nObstet Gynecol. 2019;220:354.e1–12.\n21. Nnoaham KE, Hummelshoj L, Kennedy SH, Jenk-\ninson C, Zondervan KT, World Endometriosis\nResearch Foundation Women’s Health Symptom\nSurvey C. Developing symptom-based predictive\nmodels of endometriosis as a clinical screening tool:\nresults from a multicenter study. Fertil Steril.\n2012;98:692–701 e5.\n22. Nisenblat V, Bossuyt PM, Farquhar C, Johnson N,\nHull ML. Imaging modalities for the non-invasive\ndiagnosis of endometriosis. Cochrane Database Syst\nRev. 2016;2:CD009591.\nAdv Ther (2020) 37:1087–1099 1099","source_license":"CC0","license_restricted":false}