The Spectrum of Adenomyosis Identified in a Large US Electronic Health Record Database Anita M Loughlin1,2, Gally Reznor1, Michael C Doherty1, Stephanie E Chiuve3 and Cheryl Enger1* 1Optum Epidemiology, Boston, MA/Ann Arbor, MI, US ; 2Corrona, Waltham, MA, US; 3AbbVie Inc, North Chicago, IL, US *Corresponding author: [email protected]. This work was funded by a research contract between Optum Epidemiology and AbbVie.

Background Objective

Adenomyosis, the presence of endometrial tissue within the myometrium, To examine the spectrum of the clinical presentation of adenomyosis. can co-exist with conditions such as endometriosis and leiomyomas that have symptoms in common; thus it is difficult to delineate symptom profiles specific to adenomyosis. Until recently, a definitive diagnosis was Methods made by hysterectomy; more recently, use of MRI and transvaginal . Accrual period was January 1, 2014 through December 31, 2018 ultrasound allow for non-surgical diagnosis. . Underlying population was women aged 18 – 55 years, who had clinical Data Source notes . A record of adenomyosis in the NLP consistent with a clinical diagnosis Optum Electronic Health Records (EHR) Database . Anchor date set as date of first qualifying adenomyosis mention . Patient-level database that combines electronic medical record data . Characteristics determined in 12-month period prior to anchor date (medical claims, prescription, and practice management data) from over 60 US hospitals and medical groups . Predictive probability (PP) model compared confirmed adenomyosis cases to a suitable pool of similarly symptomatic women without affirmed . Information from full-text notes (clinical notes, visit summaries, adenomyosis using logistic regression letters, and reports) are extracted using a generalized natural language processing (NLP) system and organized into concepts with . Adenomyosis cases stratified by PP pertinent attributes, sentiments, and other modifiers . Baseline characteristics assessed across PP strata . Full-text notes are available with restrictions, including privacy control . Refer to Abstract # 3722 for more detail of defining affirmed mention of and approval processes, but were not utilized for this study adenomyosis Results Figure 1. Attrition Table: Algorithm to Identify Patients with Identifying a suitable pool of comparators Adenomyosis in the Optum EHR Database . Select the subpopulation that have characteristics in common with adenomyosis population (suitable pool). Population in Optum EHR Database Exclusions . and who have Clinical Notes ICD-10 diagnosis codes N92, N94, R10, N80, D25 N=23,347,167 Table 1. Most Frequent Diagnoses Identified in Baseline - Adenomyosis No Mentions Cohort N=23,258,369 Adenomyosis Cohort (N=19,503) Population in Optum EHR Database ICD-10 Rank Description N % and who have Mention of Adenomyosis CM N=88,716 1 Z01 Encounter for Other Special Examination Without 11,044 56.6 Possible N=37,249 Complaint, Suspected or Reported Diagnosis Unclear N=7,873 2 N92 Excessive, Frequent and Irregular Menstruation 10,957 56.2 Negative N=4,076 3 Z12 Encounter for Screening for Malignant 8,816 45.2 Population with an Algorithm Affirmed Neoplasms Mention of Adenomyosis1 4 N94 Pain and Other Conditions Associated with 7,793 40.0 N=39,518 Female Genital Organs and Menstrual Cycle Not Adequate Baseline 5 R10 Abdominal and Pelvic Pain 7,776 39.9 6 N80 Endometriosis 6,117 31.4 N=12,397 7 D25 Leiomyoma of Uterus 6,037 31.0 Affirmed Mention of Adenomyosis with 8 Z00 Encounter for General Examination Without 5,997 30.7 12 months Prior Clinical Data Complaint, Suspected or Reported Diagnosis N=27,121 9 N93 Other Abnormal Uterine and Vaginal Bleeding 4,578 23.5 No Outpatient Visit 10 Z79 Long Term (Current) Drug Therapy 4,576 23.5 Available in Baseline ICD-10 CM, international classification of diseases,10th revision, clinical modification. N=3,812 Affirmed Mention of Adenomyosis with an Outpatient Visit in Baseline N=23,309 Not in Age Range of Underlying Population Female 18-55 years Interest ~11.4 million N=3,806 Cohort: Females Aged 18-55 Years with . Excessive, Frequent and Irregular Menstruation Affirmed Mention of Adenomyosis . Pain and Other Conditions Associated with N=19,503 Female Genital Organs and Menstrual Cycle . Abdominal and Pelvic Pain EHR, electronic health records. . Endometriosis 1 Adenomyosis mentions were ranked hierarchically: definite, probable (highly . Leiomyoma of Uterus likely), possible, unclear, unlikely/negated. Those mentions ranked either definite or highly likely were defined to be affirmed mentions, and anchor date was assigned to date of 1st affirmed mention of highest rank.

Table 2. Frequency of Select Baseline Characteristics for Adenomyosis by Predictive Probability Adenomyosis Adenomyosis Adenomyosis Adenomyosis Adenomyosis PP ≤ 0.1 0.1 < PP < 0.3 0.3 ≤ PP < 0.6 0.6 ≤ PP < 0.9 PP ≥ 0.9 Baseline Characteristics % % % % % Total 5,136 3,857 3,523 3,651 3,336 Demographic and Lifestyle Characteristics Age Group (years) 18-35 28.5 17.4 11.6 10.3 7.1 36-45 37.2 41.3 44.6 43.2 43.6 46-55 34.2 41.3 43.7 46.5 49.2 Pregnancy History Ever Pregnant No 67.7 44.3 35.7 34.6 26.4 Yes 32.3 55.7 64.3 65.4 73.6 Comorbid Conditions Any Endometriosis 5 10.1 18.3 47.3 93.1 Other Conditions Uterine Leiomyomas 8.3 21.6 34.1 45.9 57.4 Polyp of the Corpus Uteri 1.3 3.5 5.1 8 13.1 Endometrial Hyperplasia 0.3 1.3 2.9 4.6 5.9 Endometriosis/Adenomyosis Symptom History Heavy Menstrual Bleeding 22.8 55.8 73.5 78 86.9 Irregular Menstrual Bleeding 10.3 18.5 26.7 29 32.7 Pain (pelvic, lower abdomen) 21.7 32.7 38.5 44.3 52.4 Pain (other abdomen, back) 41.2 44.7 47.3 49.7 53.1 Pain on Micturition and/or Urinary Frequency 17.9 18.9 20.9 20.5 23 Dysmenorrhea 10 18.8 28.8 35.9 46.4 Dyspareunia 4.6 7.3 10 12.1 15.7 Hypertrophy of Uterus (without pregnancy) 1.8 4.6 8.2 11.6 15.9 Gastrointestinal Symptoms 37.7 41.1 43.9 47.4 50.5 Procedures Related to Adenomyosis Transvaginal Ultrasound (other ultrasound of uterus) 14.5 41.8 62.8 70.1 77.8 Magnetic Resonance Imaging (abdomen/pelvis) 1.7 4.3 6 9 11.7 Hysterectomy 1.6 4.7 10.6 23 50.8 Select Medication Use Current Use of Hormonal Contraceptive (oral, Depo-Provera, IUD) 12.5 16.9 17.1 18.8 18.1 Other Hormonal Treatment (HRT, progestin, estrogens) 5.4 11.2 15.2 15.7 17 Narcotic-Opioid Analgesic 27.9 34.6 41.1 50.2 66.9 Any Use of Prescription Nonsteroidal Anti-inflammatory Drugs 26.9 33.6 38.7 46.2 59.4 HRT, hormonal replacement therapy; IUD, intra-uterine device; PP, predictive probability Discussion . Using PP stratification of the cases, we show that the adenomyosis clinical presentation can range from mildly symptomatic to severely symptomatic, making recognition of disease and management of adenomyosis often difficult. . Compared to women with higher PP of adenomyosis, women in strata of lower PP were younger, less likely to have been pregnant or to have a related condition, such as endometriosis and uterine fibroids. . Among women in higher strata of PP, the most common symptoms were heavy menstrual bleeding, pelvic pain, and dysmenorrhea. Many of these women had had a related diagnostic procedure and were taking medication for pain. . Women in lower PP strata presented with less specific symptoms, such as GI symptoms and back or abdomen pain. . Understanding symptom profiles may improve diagnostic suspicion, early detection, and lead to more treatment options for women with adenomyosis.

To download a PDF copy of the poster, please scan the QR code:

Presented at the 36th International Conference on Pharmacoepidemiology and Therapeutic Risk Management (ICPE All Access); September 16-17, 2020, Poster #PO-3741