Background
The consequences of an infertility diagnosis extend beyond the pursuit of family building, because women with infertility also face increased risks for severe maternal morbidity, cancer, and chronic disease.
Objective
This study aimed to examine the association between female infertility and all-cause mortality.
Study Design
This retrospective analysis compared 72,786 women with infertility, identified in the Optum Clinformatics Datamart from 2003 to 2019 by infertility diagnosis, testing, and treatment codes, with 3,845,790 women without infertility seeking routine gynecologic care. The baseline comorbidities were assessed using the presence of ≥1 metabolic syndrome diagnoses and the Charlson Comorbidity Index. The primary outcome, which was all-cause mortality, was identified by linkage to the Social Security Administration Death Master File outcomes and medical claims. The association between infertility and mortality was examined using a Cox proportional hazard regression by adjusting for age, hypertension, hyperlipidemia, type II diabetes, year of evaluation, smoking, number of visits per year, nulliparity, obesity, region of the country, and race.
Results
Among 16,473,458 person-years of follow-ups, 13,934 women died. Women with infertility had a 32% higher relative risk for death from any cause (0.42% vs 0.35%, adjusted hazard ratio, 1.32; 95% confidence interval, 1.18–1.48) than women without infertility. The mean follow-up time per patient was 4.0±3.7 years vs 4.2±3.8 years for women with and without infertility, respectively. When stratified by age of <35 or ≥35 years or baseline medical comorbidity, the association between infertility and mortality remained. Women with infertility who delivered a child during the follow-up period faced a similar increased risk for mortality than the overall infertile group. Finally, receiving fertility treatment was not associated with a higher risk for death than receiving an infertility diagnosis or testing alone.
Conclusion
Although the absolute risk for death was low in both groups, women with infertility faced a higher relative risk for mortality than women without infertility. The association remained across all age, race and ethnicity groups, morbidities, and delivery strata. Importantly, infertility treatment was not associated with an increased risk for death. These findings reinforce the disease burden associated with infertility and its potential for long-term sequelae.
Why was this study conducted?
The objective of this study was to examine the association between female infertility and mortality using an insurance claims database.
Key findings
Although the absolute risk for death was low in both groups, we report that women with infertility have a 32% higher relative risk for death from any cause (0.42% vs 0.35%; adjusted hazard ratio, 1.32; 95% confidence interval, 1.18–1.48) than women without infertility. When stratified by age or baseline medical comorbidity, the association between infertility and mortality remained.
What does this add to what is known?
Novel findings include that receiving fertility treatment was not associated with a higher risk for death than receiving an infertility diagnosis or testing alone. Overall, these findings suggest that infertility is not only a distinct disease process, but also provides a window into future health.
Introduction
Infertility, defined by the American Society of Reproductive Medicine as the inability to achieve pregnancy after 12 months of regular, unprotected intercourse, has short- and long-term health implications. After the birth of the first in vitro fertilization (IVF) baby in 1978, the field of assisted reproductive technology (ART) has developed at a remarkable pace, with rapid advances in ovarian stimulation, IVF techniques, and the embryology laboratory. ART has created family building options for patients that were previously unimaginable. Although our ability to address the immediate (periconceptional) aspects of infertility have improved, the long-term health implications of infertility have received less recognition.
Both male and female infertility have been associated with greater risks for cancer and chronic disease, which are the leading causes of morbidity and mortality. These associations suggest that infertility provides a window into future health and has effects beyond the reproductive years. , Although nulliparity and infertility are distinct, the former has been associated with an increased risk for mortality in large, population-based linkage studies. , An independent association between infertility and mortality was reported in a secondary analysis of a multicenter, randomized clinical trial from 1992 to 2001. We sought to build on this previous work by presenting results from a contemporary cohort of women with infertility and examining the differential effects of infertility and fertility treatment on the subsequent risk for mortality.
Materials and Methods
Patients
We analyzed insured patients included in the de-identifed Clinformatics Data Mart (CDM) (Optum, Inc, Eden Prairie, MN), a nationwide commercial and Medicare Advantage claims database, from 2003 to 2019. Patients are enrolled in Optum’s CDM when their insurance coverage begins and when the administrative claims submitted for payment by providers and pharmacies are verified, adjudicated, adjusted, and de-identified before inclusion. Of note, 23% of the US population is covered by a plan included in the CDM, and the distribution of ethnicity and region is similar to the overall US population. This study was approved by the Stanford Institutional Review Board.
Women from Optum’s CDM were identified for inclusion in the infertile and control groups based on the following criteria: (1) age between 20 and 45 years; (2) enrolled in the insurance claims database for at least 6 months before and after inclusion in the study; and (3) the presence of specific diagnostic or procedural codes indicating the fertility status. Characterization of the infertile and noninfertile groups on the basis of the International Classification of Diseases (ICD) ninth and tenth edition diagnosis and procedural (current procedural terminology [CPT]) codes has been described in detail previously by our group. , In brief, the group of women with infertility was composed of women receiving any of the following during enrollment: (1) an infertility diagnosis; (2) fertility testing; or (3) fertility treatment. Women with an infertility diagnosis were identified by the ICD 9 and 10 diagnosis codes ( Supplemental Table 1 ). Fertility testing was identified through the ICD 9 and 10 diagnosis codes or the presence of a CPT code for hysterosalpingogram (HSG). Patients receiving fertility treatment were identified by the presence of a CPT code for intrauterine artificial insemination (IUI), follicular puncture for oocyte retrieval, or intrauterine embryo transfer. Pharmacy claims for a prescription for clomiphene citrate or gonadotropins (follicle stimulating hormone, human menopausal gonadotropin, or human chorionic gonadotropin) were also used to identify patients receiving fertility treatment.
The comparison group was composed of women receiving routine gynecologic care who did not have an infertility diagnosis or any procedural codes for fertility testing or fertility treatment during enrollment. These patients without infertility were identified using the following criteria: (1) the presence of a claim for a well woman visit; (2) an encounter for contraceptive management; (3) placement or removal of an intrauterine device; (4) placement of a contraceptive implant; (5) encounter for contraceptive surveillance; or (6) a pap smear.
Assessment of baseline comorbidities
Baseline comorbidities were assessed using the presence of ≥1 metabolic syndrome (MetS) diagnosis and the Charlson Comorbidity Index (CCI) score. MetS diagnoses included hypertension, hyperlipidemia, type II diabetes, and obesity and were identified using the diagnosis codes from inpatient and outpatient records (hypertension, ICD 9: 401–405 and ICD 10: I10–I16; hyperlipidemia, ICD 9: 270.2–270.4 and ICD 10: E784, E785, E781, E782, E7800; diabetes mellitus, ICD 9: 250 and ICD 10: E08–E13; obesity, ICD 9: 278.0 and ICD 10: E66.9, E66.01, E66.3, E66.2). Although the CCI has traditionally been used in an inpatient setting to evaluate mortality, its use has been validated in both the ambulatory and reproductive settings. , In addition, an updated CCI was calculated for all patients that is appropriate for use with administrative data and has shown good to excellent discrimination in predicting mortality.
Outcome ascertainment
All-cause mortality was determined by Optum using any of the following strategies: (1) linkage to the Social Security Administration Death Master File, which contains information about individuals who had social security numbers and whose deaths were reported to the Social Security Administration from 1962 to the present; if multiple dates were present, verification was performed with death certificates; (2) claims from the Centers for Medicare and Medicaid Services data; (3) facility claims for which the discharge status was designed as “expired”; and (4) member coverage discontinuation owing to death. Death records were removed if claims for insurance coverage were found more than 60 days after the death date.
Subgroup of patients with a delivery
In both groups, women who became pregnant and had a delivery during the follow-up period were identified by diagnosis and procedure codes obtained from literature indicating the conclusion of a pregnancy. ,
Statistical analysis
The association of infertility with mortality was examined using a Cox proportional hazard regression. Upon inclusion into either the infertile or noninfertile group, patients were followed for the duration of their enrollment in the insurance database or until their death, whichever occurred first. The end of enrollment was defined as the time when insured coverage ended or the last period when data were available, and women were censored in the survival analysis at this time point. The analyses were adjusted for age, hypertension, hyperlipidemia, type II diabetes, year of evaluation, smoking, number of visits per year, nulliparity, obesity, region of the country, and race. We also stratified the data by age (<35 or ≥35 year), baseline medical comorbidity quantified as the number of MetS diagnoses (0 or ≥1) and the CCI score (score of 0 or ≥1), follow-up time (<4 or ≥4 years), race, delivery, and receipt of fertility treatment during the follow-up period. Identical Cox proportional hazard models were applied for the subanalyses. All P values were 2-sided with a P value of <.05 considered statistically significant. Analyses were performed using the statistical analysis system (SAS) (version 9.4; SAS Institute Inc, Cary, NC).
Results
Patient demographics
The study population included 72,786 women with infertility and 3,845,790 women without infertility. Women with and without infertility were similar in age at the time of enrollment (mean age, 33.7±5.8 years vs 32.6±7.5 years, respectively) and were followed for comparable time periods (mean follow-up time, 4.0±3.7 years vs 4.2±3.8 years, respectively). Women with infertility were more likely to be nulliparous (16.6% vs 8.1%), obese (20.2% vs 14.9%), and smokers (7.3% vs 5.4%) than women without infertility. Epidemiologic studies vary in their findings, but some have found a higher prevalence of smoking among women with infertility than among women without infertility. A similar distribution in ethnicity, year of evaluation, and region of the country was observed among the women with and without infertility ( Table 1 ). Women with infertility had a median of 4.0 (range, 1.9–7.2) medical visits per year (all types) compared with 2.7 (range, 1.3–4.8) visits per year (all types) among women without infertility. Of note, 37% of the women in the cohort underwent infertility testing, 90% of women in the cohort had an infertility diagnosis and 23% of the infertile cohort underwent fertility treatment. We included patients that were undergoing fertility testing with the understanding that they would already meet the criteria for an infertility diagnosis. However, providers may use testing instead of diagnosis codes for reimbursement purposes, and thus we sought to capture these women for inclusion in the infertile group.
| Characteristics | Women with infertility (n=72,786) | Women without infertility (n=3,845,790) |
|---|---|---|
| Age (y), mean (SD) | 33.7 (5.8) | 32.6 (7.5) |
| 20–24, n (%) | 4331 (6.0) | 698,131 (18.2) |
| 25–29, n (%) | 14,053 (19.3) | 780,473 (20.3) |
| 30–34, n (%) | 21,532 (29.6) | 743,454 (19.3) |
| 35–39, n (%) | 19,714 (27.1) | 719,187 (18.7) |
| 40–45, n (%) | 13,156 (18.1) | 904,545 (23.5) |
| Follow-up time (y), mean (SD) | 4.0 (3.7) | 4.2 (3.8) |
| 0–1, n (%) | 12,593 (17.3) | 632,019 (16.4) |
| 1–2, n (%) | 17,552 (24.1) | 878,338 (22.8) |
| 2–3, n (%) | 10,319 (14.2) | 539,445 (14.0) |
| 3–4, n (%) | 6920 (9.5) | 366,186 (9.5) |
| >4, n (%) | 25,402 (34.9) | 1,429,802 (37.2) |
| Total (y) | 292,659.0 | 16,180,799.6 |
| Nulliparity, n (%) | 12,073 (16.6) | 309,968 (8.1) |
| Obesity, n (%) | 14,689 (20.2) | 572,426 (14.9) |
| Smoking, n (%) | 5313 (7.3) | 206,726 (5.4) |
| Index date, n (%) | ||
| 2003 | 5827 (8.0) | 334,573 (8.7) |
| 2004 | 6109 (8.4) | 339,374 (8.8) |
| 2005 | 4938 (6.8) | 290,230 (7.6) |
| 2006 | 5867 (8.1) | 319,446 (8.3) |
| 2007 | 5114 (7.0) | 276,433 (7.2) |
| 2008 | 4865 (6.7) | 270,675 (7.0) |
| 2009 | 3991 (5.5) | 242,446 (6.3) |
| 2010 | 3350 (4.6) | 202,660 (5.3) |
| 2011 | 3431 (4.7) | 198,358 (5.2) |
| 2012 | 3380 (4.6) | 189,021 (4.9) |
| 2013 | 3549 (4.9) | 180,817 (4.7) |
| 2014 | 3729 (5.1) | 176,551 (4.6) |
| 2015 | 4262 (5.9) | 189,951 (4.9) |
| 2016 | 4646 (6.4) | 206,514 (5.4) |
| 2017 | 4691 (6.4) | 216,822 (5.6) |
| 2018 | 5037 (6.9) | 211,919 (5.5) |
| Visits per person-y, median (range) | 4.0 (1.9–7.2) | 2.7 (1.3–4.8) |
| <1, n (%) | 9013 (12.4) | 723,672 (18.8) |
| 1–2, n (%) | 9766 (13.4) | 752,178 (19.6) |
| >2, n (%) | 54,007 (74.2) | 2,369,940 (61.6) |
| Region of the country, n (%) | ||
| Midwest | 16,479 (22.6) | 954,824 (24.8) |
| Northeast | 8201 (11.3) | 385,803 (10.0) |
| South | 29,962 (41.2) | 1,732,143 (45.0) |
| West | 17,919 (24.6) | 766,041 (19.9) |
| Unknown | 225 (0.3) | 6979 (0.2) |
| Race, n (%) | ||
| Asian | 7129 (9.8) | 197,821 (5.1) |
| Black | 6972 (9.6) | 364,996 (9.5) |
| Hispanic | 9046 (12.4) | 445,851 (11.6) |
| White | 40,029 (55.0) | 2,365,380 (61.5) |
| Unknown | 9610 (13.2) | 471,742 (12.3) |
Risk for mortality between women with and without infertility
Among the 16,473,458 person-years of follow-ups, 13,934 women died. Among the women with infertility, there were 307 deaths (0.42%) during the follow-up period. Among the women without infertility, there were 13,627 deaths (0.35%). Women were 42.2±7.6 years of age on average at the time of death in the infertile group and 43.3±8.2 years of age at the time of death in the noninfertile group. Although the absolute rates of death from any cause were low, women with infertility had a 32% higher relative risk for death (adjusted hazard ratio [aHR], 1.32; 95% confidence interval [CI], 1.18–1.48) than women without infertility during the follow-up period. The risk for all-cause mortality for women with infertility compared with women without infertility during the follow-up period is depicted in the Figure . Using the Schoenfeld residuals method, we concluded that the assumption of proportional hazards was valid. A subgroup of the noninfertile and infertile cohorts were matched based on age, follow-up time, nulliparity, obesity, smoking, year of visit, region, and race. In this analysis, the power was limited because of the smaller sample size and number of outcomes, but the point estimates and interpretations were similar ( Supplemental Table 2 ).
Stratification by age and comorbidities
The analyses were stratified by age and medical comorbidities to further investigate the association between infertility, mortality, and potential confounders ( Table 2 ). When stratified by age as <35 or ≥35 years or baseline medical comorbidity, the association between infertility and mortality remained (aHR, 1.39; 95% CI, 1.15–1.69 and aHR, 1.32; 95% CI, 1.15–1.52, respectively). When stratified by MetS diagnoses, there was a similar increase in the risk for death among patients with infertility compared with patients without infertiltiy across strata (aHR, 1.24; 95% CI, 1.07–1.44 among patients with no previous MetS diagnosis vs aHR, 1.24; 95% CI, 1.05–1.47 among patients with at least 1 previous MetS diagnosis). Baseline comorbidities were also quantified using the CCI with similar results as shown in Table 2 .
