Background
Adenomyosis symptoms are disabling. Population-based data on incidence and prevalence of adenomyosis are lacking that could guide future evidence-based treatments and clinical management.
Objective
To evaluate the incidence, 10-year secular trends, and prevalence of adenomyosis diagnoses and to describe symptoms and treatment patterns in a large U.S. cohort.
Study Design
We performed a retrospective population-based cohort study of women aged 16–60 years in 2006–2015, enrolled in Kaiser Permanente Washington, a mixed-model health insurance and care delivery system. Adenomyosis diagnoses identified by ICD codes from the International Classification of Diseases 9th and 10th editions and potential covariates were extracted from computerized databases. Women with prior hysterectomy, and for incidence estimates women with prior adenomyosis diagnoses, were excluded. Linear trends in incidence rates over the 10-year study period were evaluated using Poisson regression. Rates and trend tests were examined for all women adjusting for age using direct standardization to the 2015 study population, by age groups, and by race/ethnicity. Chart reviews were performed to validate diagnostic accuracy of ICD codes in identifying adenomyosis incidence. Symptoms and treatment patterns at diagnosis and in the following 5 years were assessed.
Results
A total of 333,693 women contributed 1,185,855 woman-years (2006–2015) for incidence calculations. Associated symptom-related codes (menorrhagia or abnormal uterine bleeding, dysmenorrhea or pelvic pain, dyspareunia, and infertility) were observed in 90.8%; 18.0% had co-occurrent endometriosis codes and 47.6% had co-occurrent uterine fibroid codes. The overall adenomyosis incidence was 1.03% or 28.9 per 10,000 woman-years, with a high of 30.6 in 2007 and a low of 24.4 in 2014. Overall age-adjusted estimated incidence rates declined during the 10-year study interval (linear trend P < .05). Incidence was highest for women aged 41–45 years (69.1 per 10,000 woman-years in 2008) and was higher for black (highest 44.6 per 10,000 woman-years in 2011) vs white women (highest 27.9 per 10,000 woman-years in 2010). Overall prevalence in 2015 was 0.8% and was highest among women aged 41–45 years (1.5%). Among the 624 potential adenomyosis cases identified by diagnostic codes in 2012–2015 and with sufficient information in the medical record to determine true case status, 490 were confirmed as incident cases, yielding a 78.5% (95% confidence interval, 75.1%, 81.7%) positive predictive value of adenomyosis ICD-9/ICD-10 codes for identifying an incident adenomyosis case. Health care burden was substantial: 82.0% of women had hysterectomies, nearly 70% had imaging studies suggestive of adenomyosis, and 37.6% used chronic pain medications.
Conclusion
Adenomyosis burden to the individual and the health care system is high. Incidence rates are disproportionately high among black women. These findings are of concern, as currently available long-term medical therapies remain limited beyond hysterectomy. Our data and methodologies are novel and could serve as a foundation to guide clinicians and health care systems to develop clinical management plans and track outcomes for women with adenomyosis.
Adenomyosis is the aberrant location of endometrial glandular tissue within the uterine myometrium often associated with cyclical uterine pain, dyspareunia, abnormal uterine bleeding such as menorrhagia, spotting or bleeding before and after menses, and infertility. Adenomyosis may be associated with endometriosis and uterine fibroids. , Until the last decade adenomyosis was considered a surgical diagnosis made at the time of hysterectomy. But increasingly, imaging studies, particularly pelvic ultrasound, have defined features indicative of adenomyosis, including a globular enlarged uterus, indistinct or irregular endometrial myometrial junction, heterogeneous myometrium, and myometrial cysts. , Diagnostic accuracy of ultrasound for adenomyosis is unknown, as prior studies evaluating this outcome were performed in select populations.
Why was the study conducted?
Adenomyosis symptoms are disabling. Population-based cohort studies of incidence, prevalence, trends, and treatment of adenomyosis are lacking.
Key findings
Overall incidence among 333,693 women aged 16–60 years (2006–2015) was 1%; incidence was higher for black vs white women and highest for ages 41–45. Ninety-one percent of incident cases had ICD-9 symptom-related codes. Adenomyosis co-occurrence with endometriosis and uterine fibroids was 18% and 47%, respectively. Eighty-two percent of women had hysterectomies, almost 70% had imaging studies suggestive of adenomyosis, and 38% used chronic pain medications.
What does this add to what is known?
Women in their early 40s are at highest risk for symptomatic adenomyosis. Incidence rates are disproportionately high among black women. Co-occurrence with uterine fibroids and endometriosis is high. Health care burden is substantial.
Adenomyosis symptoms can be disabling and have been treated medically, despite no U.S. Food and Drug Administration–approved therapies, and surgically. Better data on efficacy of medical treatment would assist women who prefer not to have hysterectomy. Tools that can track adenomyosis incidence, prevalence, and treatment response will be important as new therapies targeted at the pathogenesis of the disease—sex steroid regulation, inflammation, apoptosis, and neuroangiogenesis manipulation , —are developed. Despite considerable public health burden, associated costs of care, and impacts on the lives of many women, reliable population-based incidence estimates of adenomyosis do not exist and studies on prevalence vary widely. To address this gap, we conducted a retrospective cohort study using electronic health records (EHR) to estimate the incidence of symptomatic adenomyosis over a 10-year period (2006–2015) and symptomatic adenomyosis prevalence (2015) in a U.S. population. Secondary aims were to estimate incidence rates by age and by race/ethnicity, to evaluate trends, and to describe symptoms and treatment practice patterns. Furthermore, we performed chart reviews to assess the accuracy of diagnosis codes in identifying incident adenomyosis. We also estimated the proportion of cases that had imaging changes prior to their diagnosis that could be indicative of adenomyosis, thus estimating the proportion of women that might benefit from medical therapies.
Methods
Study setting and cohort
This retrospective cohort study was conducted at Kaiser Permanente Washington (KPWA), a mixed-model health insurance and care delivery system based in Seattle, Washington. KPWA provides comprehensive care on a prepaid basis to approximately 650,000 individuals in 22 Washington counties. It contracts with the KP Physicians group to provide care within an integrated group practice division (GPD) for approximately 70% of enrollees. The remaining 30% are insured by this health plan and receive care from non-KP provider networks located in geographic areas not served by KPWA medical centers. The KPWA population generally reflects the underlying community it serves with respect to age, race, and sex. The cohort consisted of all women aged 16–60 years in 2006–2015 enrolled at KPWA for a minimum of 2 years with at least 1 health care utilization at KPWA in the 2 years before cohort entry on January 1, 2006, through December 31, 2015. We further restricted to women who did not have a record of hysterectomy at least 61 days (or 2 months) prior to cohort entry. All study methods received approval from KPWA’s Human Subjects Institutional Review Board.
Data collection
We utilized KPWA electronic health care data sources. A notable feature of KPWA is the depth and longevity of its multiple computerized databases, extensively used for patient care and research for nearly 50 years. Information on enrollment, demographics, health care utilization, height and weight, diagnoses, procedures, pharmacy dispensings, and radiology and laboratory results have been maintained in automated databases since 1977. A fully integrated EHR that documents all patient care and contacts, including clinic notes and phone and e-mail communications, in KPWA-owned clinics began in 2005. All automated data sources are linked using the member’s unique health record number.
Race/ethnicity data were not complete in all years and for women who were not enrolled in the GPD. As a result, the analyses by race/ethnicity were restricted to GPD enrollees. To obtain more complete race/ethnicity data on women enrolled in the GPD, we augmented race/ethnicity data from the EHR with data extracted via Natural Language Processing. Our prior research showed a reduction from 19% to 13% in women with unknown race/ethnicity using these methods.
Identification of adenomyosis cases and potentially associated symptoms
We identified incident adenomyosis cases by selecting all women with International Classification of Diseases, 9th revision (ICD-9) diagnosis code 617.0 or 10th revision (ICD-10) code N80.0. We restricted the analyses of potential incident cases to women without an adenomyosis diagnosis in the 2 years prior to study entry. To assess the current burden of disease, we estimated adenomyosis prevalence among women enrollees in 2015 regardless of their history of adenomyosis diagnosis.
Symptoms and 2 conditions potentially associated with adenomyosis were identified from ICD-9 diagnosis codes ( Table 1 ).
| ICD-9 codes | |
|---|---|
| Symptoms | |
| Menorrhagia or abnormal uterine bleeding | 626.2 excessive bleeding 626.8 other abnormal bleeding 626.9 abnormal bleeding, unspecified 627.0 premenopausal menorrhagia: excessive bleeding associated with onset of menopause |
| Dysmenorrhea or pelvic pain | 625.3 dysmenorrhea, painful menstruation 625.9 pelvic pain 626.4 irregular periods 626.6 metrorrhagia, bleeding between menses |
| Dyspareunia | 625.0 pain with sex (dyspareunia) |
| Infertility | 628.0 infertility, female, associated with anovulation 628.2 infertility, female, of tubal origin 628.8 infertility, female, of other specified origin 628.9 infertility, female, of unspecified origin |
| Related conditions | |
| Uterine fibroids | 218 uterine leiomyoma 218.0 submucous leiomyoma of uterus 218.1 intramural leiomyoma of uterus 218.2 subserous leiomyoma of uterus 218.9 leiomyoma of uterus, unspecified |
| Endometriosis a | 617.1/N80.1 endometriosis, ovary 617.2/N80.3 endometriosis, fallopian tubes 617.3/N80.3 endometriosis, pelvic peritoneum 617.4/N80.4 endometriosis, vagina 617.5/N80.5 endometriosis, intestine 617.6/N80.6 endometriosis in scar of skin 617.8/N80.8 endometriosis, other specified sites 617.9/N80.9 endometriosis, site unspecified |
Description of treatment and utilization patterns
We assessed treatment and utilization patterns on the incident diagnosis date and in the following 5 years among women who had an ICD-9 adenomyosis code in 2006–2010. Four treatments were of interest: (1) hysterectomy (61 days prior to diagnosis through 5 years postdiagnosis) as identified by procedure codes, (2) laparoscopy and/or laparotomy from procedure codes, (3) dispensing of pain medications including opioid and nonsteroidal antiinflammatory drugs in the pharmacy data, and (4) dispensing of hormone medications including progesterone, oral contraceptives, danazol, progesterone intrauterine devices, and gonadotropin-releasing hormone antagonists. For pain medication use, we further identified chronic users as women who had at least 7 fills of opioid and/or nonsteroidal antiinflammatory drugs, which equated to approximately 7 months of use. For hormone medication use, we defined chronic users as those women who had at least 3 fills of oral progesterone and/or oral contraception, at least 3 fills of danazol, at least 3 injectable progesterone fills, any implant progesterone, any progesterone intrauterine devices, or gonadotropin-releasing hormone injections which equated to over 6 months of use.
Case validation
Two trained abstractors and a study clinician (S.D.R.) reviewed medical records of enrollees with adenomyosis diagnosis codes in years 2012–2015. True incident cases identified at chart review were women who had surgical or imaging diagnosis of adenomyosis in the index period (defined as 60 days before and 60 days after the automated index diagnosis date) without a prior adenomyosis code and without prior imaging studies with possible or probable adenomyosis. By definition, incident cases diagnosed by imaging had the word “adenomyosis” in the body or in the impression of the imaging report. Cases defined by imaging could include words like “possible,” “probable,” or “unable to rule out” adenomyosis. Incident cases may have had prior imaging with characteristics suggestive of adenomyosis but without the word “adenomyosis” in the imaging report.
Statistical analyses
Women contributed person-time from the date when they became eligible for the study through the earliest date of disenrollment from KPWA (their 61st birthday, date of hysterectomy (if it occurred after study entry), or study end date of December 31, 2015. Annual incidence rates of adenomyosis were calculated for all women (16–60 years), age-adjusted using direct standardization to the 2015 study population. Linear trends in annual adenomyosis incidence rates over the 10-year study period (2006–2015) were evaluated using Poisson regression. Rates and linear trends were also examined for each 5-year age group (16–20 years through 56–60 years).
We examined annual adenomyosis incidence rates by race/ethnicity among women enrollees in GPD only (owing to more complete race/ethnicity capture). Groups included Hispanic; non-Hispanic groups: black, white, Asian, Hawaiian/Pacific Islander, and Native American; and other or unknown race/ethnicity. Annual incidence rates were calculated for all women in each race/ethnicity group, age-adjusted using direct standardization to the 2015 race/ethnicity-specific study cohort.
We estimated the overall and age-specific prevalence of adenomyosis for the most recent study year (2015). The denominator consisted of all women who contributed any person-time in 2015; women who also received an adenomyosis diagnosis during or before 2015 comprised the numerator.
Among women with an incident adenomyosis diagnosis in 2006–2010, the proportions with symptoms potentially associated with adenomyosis in the 2 years prior to and 5 years following the diagnosis were calculated. We also estimated the proportion of incident cases that had an ICD-9 code for uterine fibroids or endometriosis. Treatment and utilization patterns were assessed by determining the proportions of women who experienced a surgical procedure or had a medication fill on the day of diagnosis or in the following 5 years.
To determine the accuracy of ICD codes in identifying incident cases of adenomyosis, positive predictive values (PPV) were calculated from chart review (proportion of true cases among all potential cases identified by ICD codes). We estimated the proportion of incident cases that had prior ultrasounds with characteristic features of adenomyosis, but without the word “adenomyosis” in the imaging report. The proportion of cases identified by imaging with surgical confirmation was estimated.
Results
Adenomyosis incidence and prevalence
A total of 333,693 women without an adenomyosis diagnosis in the past 2 years contributed 1,185,855 woman-years during the 10-year study period. Of these, 3425 women received a first diagnosis of adenomyosis and were considered potential incident cases. Mean age at study cohort entry was 41.5 years for case women and 37.5 years for non-case women ( Table 2 ). Women with adenomyosis diagnoses were more likely to be non-Hispanic black. Among 3425 women with an incident diagnosis of adenomyosis, 1633 (47.7%) received the diagnosis during an inpatient stay and 1380 (40.3%) during an outpatient visit. On the date of adenomyosis diagnosis, 18.0% were also diagnosed with endometriosis and 47.6% were also diagnosed with uterine fibroid(s).
| Adenomyosis diagnosis | ||||
|---|---|---|---|---|
| Yes (n = 3425) | No (n = 330,268) | |||
| N | % | N | % | |
| Age at study entry a | ||||
| Mean age (SD), years | 41.5 | (7.7) | 37.5 | (14.3) |
| Median age, years | 42 | 39 | ||
| 16–20 years | 34 | 1 | 53,804 | 16.3 |
| 21–25 years | 42 | 1.2 | 25,997 | 7.9 |
| 26–30 years | 211 | 6.2 | 32,784 | 9.9 |
| 31–35 years | 445 | 13 | 32,682 | 9.9 |
| 36–40 years | 710 | 20.7 | 32,435 | 9.8 |
| 41–45 years | 936 | 27.3 | 34,844 | 10.6 |
| 46–50 years | 673 | 19.6 | 38,633 | 11.7 |
| 51–55 years | 289 | 8.4 | 41,305 | 12.5 |
| 56–60 years | 85 | 2.5 | 37,784 | 11.4 |
| Race/ethnicity b | ||||
| Hispanic | 127 | 6.7 | 13,648 | 6.2 |
| Non-Hispanic white | 1316 | 69.3 | 136,404 | 61.9 |
| Asian | 151 | 8 | 21,851 | 9.9 |
| Native American | 55 | 2.9 | 3879 | 1.8 |
| Non-Hispanic black | 146 | 7.7 | 11,321 | 5.1 |
| Hawaiian/Pacific Islander | 16 | 0.8 | 2641 | 1.2 |
| Other/unknown | 87 | 4.6 | 30,788 | 14 |
| Diagnosis year | ||||
| 2006 | 371 | 10.8 | n/a | |
| 2007 | 380 | 11.1 | ||
| 2008 | 332 | 9.7 | ||
| 2009 | 320 | 9.3 | ||
| 2010 | 364 | 10.6 | ||
| 2011 | 351 | 10.2 | ||
| 2012 | 371 | 10.8 | ||
| 2013 | 372 | 10.9 | ||
| 2014 | 289 | 8.4 | ||
| 2015 | 275 | 8 | ||
| Visit setting at diagnosis | ||||
| Inpatient | 1633 | 47.7 | n/a | |
| Emergency department | 11 | 0.3 | ||
| Urgent care | 11 | 0.3 | ||
| Outpatient | 1380 | 40.3 | ||
| Radiology | 64 | 1.9 | ||
| Other care | 326 | 9.5 | ||
| Other ICD-9/10 diagnosis codes on adenomyosis diagnosis date (not mutually exclusive) | ||||
| Endometriosis c | 617 | 18.0 | n/a | |
| 617.1/N80.1 Ovary | 275 | 8.0 | ||
| 617.2/N80.2 Fallopian tubes | 70 | 2.0 | ||
| 617.3/N80.3 Pelvic peritoneum | 344 | 10.0 | ||
| 617.4/N80.4 Vagina | 2 | 0.1 | ||
| 617.5/N80.5 Intestine | 29 | 0.8 | ||
| 617.6/N80.6 Scar of skin | 3 | 0.1 | ||
| 617.8/N80.8 Other sites | 40 | 1.2 | ||
| 617.9/N80.9 Unspecified | 441 | 12.9 | ||
| Uterine fibroids 218.0, 218.1, 218.2, 218.9 | 1630 | 47.6 | ||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
