Increased Prevalence of Adverse Health Outcomes Across the Lifespan in Those Affected by Polycystic Ovary Syndrome: A Canadian Population Cohort.

In this retrospective cohort study, the administrative data from the regional health authority, Alberta Health Services, were used to identify both a qualifying cohort with PCOS and controls (defined below) who had heath system encounters in Alberta, Canada from April 1, 2002 to March 31, 2018. Alberta has a single-payer government-funded healthcare system that provides universal access to essential services to all residents. Each individual patient is assigned a personal healthcare number that acts as a unique lifetime identifier (ULI) that is captured in all health system encounters. Activities across the healthcare system are collected in different databases and are linked using this ULI. These databases include data for approximately 4.3 million people (3.1 million adults) across Alberta. These databases have been used in various studies and have been proven to be representative of a Canadian province, and have been validated for multiple morbidities. 46 , 47 , 48 The study protocol was approved by the Human Research Ethics Board of the University of Alberta (Pro00094712). Four databases used in this project include the Discharge Abstract Database (DAD), the National Ambulatory Care Reporting System (NACRS) database, and the Physician Claims and Alberta Health Care Insurance Plan (AHCIP) registry. 49 , 50 The Discharge Abstract Database captures admissions to acute care hospitals with an event date and includes up to 25 International Classification of Diseases (ICD), 10th revision (-10) Canadian (CA) diagnosis codes and up to 20 procedures (Canadian Classification of Health Interventions codes). The National Ambulatory Care Reporting System database includes visits to hospital-based ambulatory care centres, including emergency departments, in Alberta. Each record contains the date of visit, up to 10 ICD-10-CA diagnosis codes, and up to 10 procedure codes (Canadian Classification of Health Interventions codes). Physician billing claims data capture information on fee-for-service claims by physicians and include relevant dates, up to 3 ICD, 9th revision (-9) diagnosis codes, and a procedure or billing code. The AHCIP provides basic medical and hospital insurance coverage for Albertans under the Canada Health Act, and it is used to represent the Alberta population. This database was used to select the matching controls for the PCOS cohort. Patient diagnosis and health outcomes were identified using ICD-9 and ICD-10 codes (see Supplemental Table S1 ) in any hospitalization or ambulatory encounter during the observation period. 47 , 51 , 52 The date of a health outcome diagnosis, including PCOS and morbidities, was determined using the earliest date of the first hospital encounter with a PCOS diagnosis, or the second time a PCOS diagnosis was recorded in a community physician or specialist encounter. The Charlson Comorbidity Index using 19 validated comorbid conditions was used, along with the associated ICD-9 and ICD-10 codes for operationalization of the administrative data. 51 , 53 , 54 We conducted a retrospective observational cohort study in adolescents and women aged ≥ 12 years who were diagnosed with PCOS in the period from 2002 to 2018, using the Alberta Health Services administrative health database, and in a ratio of 1:3 with an age-matched control non-PCOS population. PCOS is a disease that affects exclusively the female sex, and the role of gender in the diagnosis and prevalence of PCOS was not explored in this study, as this variable is not available in the administrative database. We identified all patients aged ≥ 12 years with a PCOS diagnosis, using the ICD-9 code (256.4) or the ICD-10 code (E28.2) for PCOS, excluding endocrine or other disorders that present like PCOS. 55 , 56 , 57 , 58 The date of the first diagnosis of PCOS was considered to be the index date. Patients with an index date during the study period (2002-2018) were included. Patients with an index date before the start of the study period (preexisting condition) were excluded from the exposure cohort as well as the control cohort. The unexposed or controls in this study were drawn randomly from the AHCIP registry using the 1:3 case-to-control ratio. Controls were identified by their ULI and were linked to other databases for demographics and health outcomes. The unexposed or controls in this study were drawn randomly from the AHCIP registry using the 1:3 case-to-control ratio. Controls were identified by their ULI and were linked to other databases for demographics and health outcomes. The greedy matching algorithm was used, and for each patient with PCOS, 3 controls born within 3 years before or after the start of a PCOS case were matched. 59 , 60 Each patient was matched only once. Each control was assigned a pseudo-index date, which is the date of the first PCOS diagnosis for their matched PCOS case. The primary outcome was the prevalence of morbidities in PCOS and non-PCOS controls from 2002-2018. The secondary outcome was the age of diagnosis of select morbidities. The proportion of individuals with each morbidity was expressed as a percentage. Χ 2 tests were applied to examine the statistical significance of the difference between PCOS-exposed and PCOS-unexposed groups. Significance was set at P < 0.05. Statistical analyses used STATA 13.0 (StataCorp, College Station, TX).

We found that 16,531 cases of PCOS were diagnosed in Alberta in the period 2002-2018, with the median age of first diagnosis being 28 years. The PCOS-exposed cases were matched with 49,335 non-exposed controls without PCOS ( Table 1 ). The diagnosis of PCOS increased 8-fold, from 359 cases in 2002, to 2861 cases in 2016 ( Fig. 1 ). This change is likely attributable to both the increased awareness of PCOS over this period and the development and implementation of international diagnostic criteria in adolescents and adults. 1 , 61 , 62 , 63 , 64 Table 1 Prevalence of adverse health outcomes in patients with polycystic ovary syndrome (PCOS) and non-PCOS controls across the lifespan Health outcomes Non-PCOS controls, n (%) PCOS patients, n (%) P N 49,335 16,531 Gynecological-endocrine  Infertility 2879 (5.83) 4474 (27.06) < 0.0001  Menstrual dysfunction 17,731 (35.94) 12,259 (74.16) < 0.0001  Hyperandrogenism 8784 (17.80) 15,308 (92.60) < 0.0001  Polycystic ovaries 1699 (3.44) 14,980 (90.62) < 0.0001  Endometriosis 4114 (8.34) 3747 (22.67) < 0.0001 Circulatory  Hypertension 18,216 (36.92) 8846 (53.51) < 0.0001  Cardiovascular disease 1235 (2.50) 614 (3.71) < 0.0001  Myocardial infarction 226 (0.31) 134 (0.55) < 0.0001  Congestive heart failure 325 (0.47) 233 (0.95) < 0.0001  Cerebrovascular disease 4094 (8.29) 2332 (14.11) < 0.0001  Peripheral vascular disease 189 (0.36) 261 (0.76) < 0.0001 Endocrine-metabolic  Dyslipidemia 1721 (3.48) 1264 (7.64) < 0.0001  Obesity 3287 (6.66) 4203 (25.42) < 0.0001  Metabolic syndrome 2065 (4.18) 643 (3.89) < 0.097  Non-alcoholic fatty liver disease 611(1.12) 443 (2.68) < 0.0001  Type 1 diabetes 1414 (2.87) 1442 (8.72) < 0.0001  Type 2 diabetes 1504 (3.05) 1649 (9.98) < 0.0001 Renal disease 330 (0.67) 213 (1.29) < 0.0001 Respiratory 43,167 (57.86) 20,208 (79.97) < 0.0001  Bronchitis 27,824 (56.40) 12,223 (73.93) < 0.0001  Influenza-pneumonia 4674 (9.47) 2396 (14.49) < 0.0001  CPD 1212 (14.99) 826 (27.67) < 0.0001 Cancer  Any malignancy 1519 (3.08) 998 (6.04) < 0.0001  Metastatic solid tumour 296 (0.60) 176 (1.08) < 0.0001 Mental health  Mental illness 14,196 (28.77) 7193 (43.51) < 0.0001  Depression-anxiety 18,887 (38.28) 9222 (55.78) < 0.0001  Eating disorders 1573 (3.19) 937 (5.67) < 0.0001  Dementia (unspecified) 132 (0.22) 106 (0.48) < 0.0001 Gastrointestinal disease  Peptic ulcer 799 (1.62) 540 (3.27) < 0.0001  Nonalcoholic fatty liver disease 611(1.12) 443 (2.68) < 0.0001 Muscoloskeletal  Rheumatoid disease 617 (1.30) 330 (2.00) < 0.0001 Communicable diseases  AIDS/HIV 889 (1.80) 371 (2.20) 0.0003  STD 3066 (1.30) 1291 (7.80) < 0.0001  Tuberculosis 49 (0.10) 16 (0.10) 0.9285 See Supplemental Table S1 for health outcomes using International Classification of Diseases codes. AIDS/HIV, acquired immune deficieny syndrome/human immunodeficiency virus; CPD, chronic pulmonary disease; STD, sexually transmitted disease. Figure 1 The prevalence of diagnosis of polycystic ovary syndrome (PCOS) in the cohort from 2002-2018. The cohort was determined using the Alberta Health Services Provincial Research Data Service, which links administrative datasets including the Discharge Abstract Database (inpatient), the National Ambulatory Care Reporting System, Physician Claims, and a province-wide laboratory repository for all residents of Alberta. 53 , 54 PCOS diagnosis was defined using the International Classification of Diseases, 9th and 10th revision codes. Prevalence of adverse health outcomes in patients with polycystic ovary syndrome (PCOS) and non-PCOS controls across the lifespan See Supplemental Table S1 for health outcomes using International Classification of Diseases codes. AIDS/HIV, acquired immune deficieny syndrome/human immunodeficiency virus; CPD, chronic pulmonary disease; STD, sexually transmitted disease. The prevalence of diagnosis of polycystic ovary syndrome (PCOS) in the cohort from 2002-2018. The cohort was determined using the Alberta Health Services Provincial Research Data Service, which links administrative datasets including the Discharge Abstract Database (inpatient), the National Ambulatory Care Reporting System, Physician Claims, and a province-wide laboratory repository for all residents of Alberta. 53 , 54 PCOS diagnosis was defined using the International Classification of Diseases, 9th and 10th revision codes. The prevalence of hypertension and CVD was 30% higher in those with PCOS, compared to the non-PCOS control population ( Table 1 ). Peripheral vascular disease and cerebrovascular disease had a 50% and 40% higher prevalence in those with PCOS, compared to non-PCOS controls, and these diseases occurred on average 3 years earlier in those with PCOS, compared to non-PCOS controls (at ages 35 vs 37 years, respectively, for peripheral vascular disease, and at ages 38 vs 40 years, respectively, for cerebrovascular disease). The prevalence of myocardial infarction was 40% higher in those with PCOS, with a median age of diagnosis at age 43 years in PCOS patients, compared to age 47 years in non-PCOS controls. Congestive heart failure had a 50% higher prevalence in those with PCOS, with a median age of diagnosis of 43 years, compared to age 42 years in non-PCOS controls. Obesity had a 3.8-fold higher prevalence in those with PCOS, occurring in 25.4% of the cohort, compared to 6.7% of non-PCOS controls ( Table 1 ). The prevalence of the metabolic syndrome was not different between the PCOS patients and the non-PCOS controls, occurring in approximately 4.0% of the population. Type 2 diabetes had a 3-fold higher prevalence in those with PCOS (10%), compared to non-PCOS controls (3%), and no difference was present in the age of onset, at 32 years for both. Dyslipidemia had a 2-fold higher prevalence in PCOS patients (7.6%), compared to non-PCOS controls (3.5%), and the diagnosis of dyslipidemia occurred 3 years earlier in PCOS patients, at age 35 years, vs age 38 years in non-PCOS controls. Type 1 diabetes was 3-fold more prevalent in PCOS patients (8.7%), compared to the non-PCOS controls (2.9%). Renal disease was 30% higher in those with PCOS, yet it occurred 3 years earlier in non-PCOS controls, at age 35 years, compared to age 37 years in those with PCOS ( Table 1 ). All types of gastrointestinal disorders, including inflammatory bowel disorders, ulcerative colitis, crohn’s disease, and celiac disease, had a more than 3.5-fold higher prevalence in PCOS patients (94.1%) than in non-PCOS controls (24.9%) ( Table 1 ). Non-alcoholic fatty liver disease (NAFLD) had a 2-fold higher prevalence in PCOS patients (2.7%), compared to non-PCOS controls (1.2%). All respiratory disease had a 30% higher prevalence in those with PCOS (79%) compared to non-PCOS controls (57%). Bronchitis had a 20% higher prevalence in those with PCOS (73.9%), compared to non-PCOS controls (56.4%). The prevalences of influenza and pneumonia were 1.5-fold higher in those with PCOS, compared to non-PCOS controls. Chronic pulmonary disease (CPD), which includes all conditions related to breathing or respiratory symptoms (asthma, bronchitis, emphysema) and is commonly coded in physician claims, had a 1.8-fold higher prevalence in those with PCOS (27.6%), compared to non-PCOS controls (14.9%) ( Table 1 ). All malignant cancers had a 50% higher prevalence in those with PCOS (6.0%), compared to non-PCOS controls (3.1%) ( Table 1 ). All types of mental disorders, including bipolar disorder, psychosis, and schizophrenia, had a 1.5-fold higher prevalence in those with PCOS (43.5%), compared to non-PCOS controls in the cohort (28.8%), with a median age of diagnosis at age 27 years and 28 years, respectively. Depression-anxiety disorders had a 1.4-fold higher prevalence in those with PCOS (55.8%), compared to non-PCOS controls (38.3%) ( Table 1 ). Eating disorders are classified as psychiatric disorders, and the prevalence was 40% higher in those with PCOS (5.7%), compared to non-PCOS controls (3.2%). Dementia, which may include coding for several related symptoms in primary care, and for symptoms associated with other neurological diseases such as Parkinson’s disease and unspecified dementia, occurred 19 years earlier in those with PCOS, at a median age of 43 years, compared to age 62 years in non-PCOS controls, and the prevalence was 2-fold higher in those with PCOS (0.48%), compared to non-PCOS controls (0.22%). Rheumatoid disease had a 30% higher prevalence in those with PCOS (1.3%) vs non-PCOS controls (2.0%) ( Table 1 ). Only very limited reports have been published on acquired immune deficiency syndrome/human immunodeficiency virus (AIDS/HIV) and sexually transmitted diseases in those with PCOS, and thus, this area requires further investigation. We found that AIDS/HIV had a 20% higher prevalence in those with PCOS, compared to non-PCOS controls, with prevalences of 2.2% and 1.8%, respectively ( Table 1 ). Sexually transmitted diseases occurred at a 6-fold higher rate in those with PCOS, compared to non-PCOS controls.

The results of this study have shown in this Canadian cohort for the first time that those with PCOS have an increased prevalence of morbidities that impact them across their lifespan. The prevalences of cardiometabolic risk factors, mental illness, diabetes, and CVD were significantly increased in those with PCOS. These findings provide evidence-based data on the potential economic and healthcare burden of those affected by PCOS, and highlight the need for the development of early screening, monitoring, and prevention of morbidities and long-term chronic disease in this high-risk population. Our data also highlight the imminent need for increased research funding and resource allocation to target the healthcare burden for those affected by PCOS in Canada.

Our results have shown for the first time in a Canadian cohort that women with PCOS have an increased prevalence of adverse health outcomes across the lifespan, compared to non-PCOS controls. These findings are consistent with those of other retrospective cohort studies demonstrating a predisposition in those with PCOS to the development of morbidities with potentially, shared pathophysiological pathways. 2 , 6 , 17 , 18 , 31 , 32 The likelihood of onset of noncommunicable diseases increases with age, and in those with PCOS, the onset of circulatory disease occurred strikingly earlier, by 3-4 years, including end-stage myocardial infarction, and peripheral and cerebrovascular disease. Mental health disorders were 30%-40% more prevalent in those with PCOS across the lifespan, including depression, and anxiety and eating disorders. The most striking finding was the 2-fold higher prevalence of dementia and related symptoms in those with PCOS, and the finding that a diagnosis occurred, on average, 19 years earlier in those with PCOS. A well established finding is that women with PCOS have increased cardiometabolic risk, and this was affirmed in our Canadian cohort. 1 In our cohort, the prevalence of obesity in those with PCOS was 25.4%, compared to 6.7% in non-PCOS controls, which is comparable to other population cohort studies of PCOS. In an Australian cohort, those with PCOS had an obesity prevalence of 16.0%, compared to 3.7% in non-PCOS controls. 6 In a US cohort, 39.5% of those with PCOS had obesity, compared to 10.1% of non-PCOS controls. 31 In other case-control and epidemiologic studies, the prevalence of overweight and obesity in those with PCOS is estimated at 50%, and it ranges from 33% to 88%. 65 , 66 Obesity is associated with a more adverse PCOS phenotype, particularly type A (classic PCOS), which includes hyperandrogenism, menstrual-ovulatory dysfunction, and polycystic ovary morphology. 65 , 66 , 67 In gene-wide association studies, cross-trait analysis has shown that obesity-related traits and body mass index share a genetic basis with PCOS. 68 However, genetic predisposition to PCOS may not predict obesity-related traits, and PCOS has been shown to be independent of obesity and to be dependent on ethnicity, nationality, socioeconomic status, and demographics. 65 , 66 , 68 Our data showed no difference in the prevalence of the metabolic syndrome between those with PCOS and controls, despite the higher prevalence of obesity and diabetes in those with PCOS. The metabolic syndrome had a low prevalence (4%) compared to that identified in other Canadian studies that have found prevalences in adult women of 16.5% (age ≥18 years), 31.5% (age ≥18 years), and 13.4% (age 40-79 years). 48 , 69 , 70 PCOS is diagnosed in young women of reproductive age (18-35 years), and the metabolic syndrome may not be coded simultaneously at the time of PCOS diagnosis or at follow-up encounters recorded in physician claims and the acute care administrative databases. Furthermore, studies use different methods for defining the metabolic syndrome, including the documentation of a specific ICD code, documentation of a cluster of ICD codes, and/or individual laboratory results. 48 , 70 , 71 , 72 These factors may explain the apparent lower prevalence of the metabolic syndrome in this cohort. Dyslipidemia is common in PCOS and is estimated to occur in 70% of those with PCOS. 73 , 74 , 75 , 76 We and others have reported that those with PCOS have a more adverse fasting lipid profile, and this is positively associated with hyperandrogenemia, independent of obesity. 73 , 74 , 75 , 77 , 78 , 79 Atherogenic dyslipidemia in the fasted and non-fasted state has been described in those with PCOS and commonly includes elevated plasma triglycerides, apoB-lipoproteins, remnant cholesterol and lowered high-density lipoprotein cholesterol, and impairments in postprandial clearance of triglycerides and apoB48-lipoproteins. 73 , 76 , 77 , 78 , 80 , 81 , 82 , 83 Atherogenic dyslipidemia, in particular elevated triglycerides and apoB-associated remnant cholesterol, is a causative factor in atherosclerotic CVD and is associated with end-stage cardiovascular ischemic events. 84 , 85 Young women and adolescents with PCOS have been shown to have early atherosclerotic CVD, including increased carotid intimal medial thickness and carotid plaque. 86 , 87 , 88 , 89 Our studies have shown that these atherosclerotic vascular indices are positively correlated with apoB-lipoproteins and remnant cholesterol in young women aged 18-45 years. 90 We have found the prevalence of dyslipidemia in this cohort to be 2-fold higher (7.60%) in those with PCOS, compared to non-PCOS controls (3.5%). A higher prevalence of dyslipidemia predisposes those with PCOS to having an increased risk of atherosclerotic CVD, and in those with PCOS, dyslipidemia was diagnosed earlier, at age 35 years, compared to age 38 years in non-PCOS controls. In a US population cohort of young women aged 15-45 years, dyslipidemia had a prevalence of 20.7%, compared to 6.8% in those with PCOS and in non-PCOS controls, respectively. 31 In a Danish cohort study, the prevalence of dyslipidemia, defined as hypercholesterolemia only, using ICD-9 (code E78) was double that in those with PCOS, with a prevalence of 2%, compared to 1% in non-PCOS controls. 18 We and others have shown that dyslipidemia, particularly hypertriglyceridemia and apoB-lipoproteinemia, 73 , 75 , 79 occurs in youth and young adults with PCOS and obesity. 77 , 78 , 83 Currently, PCOS is not listed as a cardiovascular risk factor in Canadian Cardiovascular Society dyslipidemia guidelines. 91 However, dyslipidemia and atherosclerosis in PCOS are very early subclinical indicators of CVD risk and highlight the need for early screening and intervention in the prevention of long-term CVD in this population. The prevalence of gastrointestinal disorders, inclusive of gastric, intestinal, splenic, pancreatic, and hepatic dysfunction, was very high in PCOS patients, compared to controls, indicating that increased symptom presentation of these disorders occurs in primary, acute, and ambulatory care settings. 47 The higher prevalence of gastrointestinal disorders in PCOS may be associated with several factors, including the following: a higher prevalence of inflammatory bowel disorders in Canada and Alberta 92 , and globally 93 , 94 , 95 ; a higher prevalence of gut dysbiosis in those with PCOS 96 , 97 ]; a high prevalence of metformin use in those with PCOS, which causes gastrointestinal side effects 58 , 83 , 98 ; a higher prevalence of NAFLD and bariatric surgery in those with PCOS, which may be associated withgastrointestinal symptoms. 99 , 100 , 101 , 102 The prevalence of NAFLD has been reported to be either not different or higher in prevalence in adolescent and adult individuals with PCOS, and it has been associated with higher rates of obesity, insulin-resistance, hypertriglyceridemia, and hyperandrogenemia indices. 103 , 104 , 105 , 106 , 107 NAFLD also is associated with an increased transition and incidence of type 2 diabetes in the general population and in those with PCOS. 108 The 4.27% prevalence of NAFLD in those with PCOS in our results is very low compared to findings in other population cohort studies, which show a 40%-50% prevalence in those with PCOS. 100 , 103 This difference may be reflective of the low level of screening for NAFLD in Canada. Early screening for NAFLD has been recommended to prevent progression to nonalcoholic steatohepatitis (NASH) and advanced liver disease in those with PCOS. 100 , 103 Our data showed that type 2 diabetes had a 3-fold higher prevalence in those with PCOS in Canada (10% vs 3% in controls), consistent with other epidemiologic studies. 6 , 17 , 27 , 31 , 109 In our cohort, the median age of diagnosis of type 2 diabetes was 32 years, in both those with PCOS and non-PCOS controls. This finding is in contrast to results in a Danish cohort of women aged 12-60 years, who showed an earlier median age of diagnosis of 33 years in those with PCOS, compared to 38 years in non-PCOS controls. 17 Similar to our findings, in those that developed type 2 diabetes, a higher rate of risk factors, including body mass index, and levels of fasting blood HbA1c, glucose, insulin, and triglycerides at baseline, was shown in the Danish cohort. 17 At present, we do not know the economic burden of diabetes or gestational diabetes in PCOS in Canada. In the US, the excess economic burden in those with PCOS and type 2 diabetes is estimated at $1.5 billion, and this figure is based on an odds ratio of 2.4 and a prevalence of 0.5% to 21.4% in those with PCOS, compared to the control population, across the lifespan. 27 Type 1 diabetes has been observed to have a higher prevalence (7%-25%) in those with PCOS, and in those with menstrual and reproductive dysfunction, and this has been associated with hyperandrogenemia. 110 , 111 , 112 , 113 , 114 Our results are consistent with these findings and support the concept that type 1 diabetes may predispose individuals to the development of PCOS. The predisposing risk factors and mechanisms remain unclear, but studies suggest that insulin therapy may increase ovarian androgen synthesis ensuing the PCOS phenotype. 113 , 115 Our results show that the incidence of a diagnosis of hypertension and CVD was 1.5-fold higher in those with PCOS, compared to non-PCOS controls, and ischemic CVD events occurred in young women, consistent with findings in other population studies. 6 , 11 , 13 , 18 , 27 , 31 , 32 , 116 , 117 Of note, our results showed that diagnosis of myocardial infarction, and peripheral and cerebrovascular disease occurred in young premenopausal women, and 3-4 years earlier in those with PCOS. In another Canadian cohort, 5.6% and 8.4% of women with a median age of 51.7 years were reported to have had a previous myocardial infarction and spontaneous coronary artery dissection, respectively. 118 In a Danish cohort, CVD incidence, including hypertension, occurred at a median age of 35 and 36 years in those with PCOS and in non-PCOS controls, respectively, and > 95% of those diagnosed with CVD were aged < 50 years. 18 In an Australian cohort, the prevalence of coronary heart disease was 0.0% and 0.8% in controls and those with PCOS, respectively, for whom the median age was 35.8 years. 6 The reproductive and premenopausal years represent periods of increased cardiovascular risk in those with PCOS, and this may be associated with a higher prevalence of obesity and diabetes. 11 The increased incidence of obesity and type 2 diabetes in those with PCOS may contribute to the higher prevalence and earlier age of diagnosis of cardiovascular events. In a Danish cohort with PCOS, diabetes and obesity predicted a 4.9 and a 2.7 higher hazard ratio for CVD ischemic events, respectively. 18 To our knowledge, our study is the first Canadian study to show an increased CVD risk in young premenopausal women with PCOS, and it highlights the need for early screening of cardiac dysfunction and atherosclerotic CVD in this population. An increased prevalence of menstrual dysfunction, infertility, obesity, and insulin resistance has been reported in women with asthma. 119 , 120 , 121 The link between asthma and PCOS is proposed to occur via inflammatory pathways that are shared in obesity and insulin resistance. 119 , 120 , 121 Those with PCOS appear to have a higher risk of developing asthma, and the level of use of asthma medication prescriptions is reported to be higher in those with PCOS. 32 , 122 In a Danish cohort, asthma had a prevalence of 25% in those with PCOS who were aged 12-60 years. 32 In an Australian study, a hazard ratio of 2.5 was reported for asthma-related hospitalizations in those with PCOS, as compared to non-PCOS controls. 6 The same study showed a higher rate of hospitalization in those with PCOS associated with respiratory disease, with a hazard ratio of 1.7 in those with PCOS. 6 Consistently, our data show that the incidence of respiratory disease is 25%-35% higher in those with PCOS, including bronchitis and influenza-pneumonia. The prevalence of malignant cancer has been reported to be 2-fold greater or not different in those with PCOS. 6 , 32 The prevalence of endometrial cancer has been reported to be increased 3-5 fold in those with PCOS in the postmenopause years, and an increased risk for ovarian cancer has been reported in younger women with PCOS in the premenopause years. 6 , 123 , 124 , 125 In our cohort, we have shown that all malignancy was increased by 30% in those with PCOS, suggesting that increased awareness and screening may be warranted. Our data showed that mental health and psychiatric disorders had a 30% higher prevalence in those with PCOS, consistent with findings of previous population cohort studies. 6 , 29 , 31 , 32 , 126 , 127 In particular, the prevalence of depression and anxiety in those with PCOS has been reported to be markedly elevated, with a 30%-50% prevalence and an odds ratio of 1.4 to > 2.5 in those with PCOS. 7 , 29 , 32 , 127 , 128 Eating disorders can be a product of depression and anxiety in the PCOS patient, and we have shown that eating disorders have a 5.7% prevalence in those with PCOS, and a 40% higher prevalence, as compared to non-PCOS controls. This prevalence of eating disorders is relatively high compared to that found in other population studies that report a 0.3%-2.45% prevalence in those with PCOS. 31 , 126 , 129 , 130 However, in meta-analyses, a prevalence ratio of 1.2 and an odds ratio of 3.8 have been reported for eating disorders in those with PCOS, compared to controls, 29 , 131 and in case control studies, the prevalence of an eating disorder has been reported to range from 11% to 62%. 132 , 133 , 134 , 135 , 136 We have reported that the eating disorders seen in those with PCOS are most commonly bulimia nervosa and binge eating disorder. 137 The healthcare costs associated with anxiety, depression, and eating disorders in the US were estimated to be $4.26 billion per year in 2021. 29 Our data show that dementia-related symptomology coded in the administrative database was increased 2-fold in those with PCOS and that it occurs 19 years earlier in those with PCOS, compared to the age in controls. This finding requires further investigation into the symptoms vs the early onset of cognitive decline specific to dementia or Alzheimer’s disease. Studies of PCOS and risk of cognitive decline, dementia, and Alzheimer’s disease are lacking, but the postulation has been made that increased risk may be associated with chronic inflammation, obesity, insulin resistance, and androgenic-mediated effects on the central nervous system that affect cognitive function. 138 We have established the first Canadian cohort of those with PCOS, using a provincial administrative health database. A strength of the study is the large size of the cohort and that prevalence of disease was determined for a long time period. We had the ability to match PCOS cases to non-PCOS controls using administrative databases that are linked together, which provided a comprehensive ability to capture cases of or those exposed to PCOS, and to match them with non-PCOS controls, using the ULI. We used validated ICD coding for diseases using the Charlson Comorbidity Index and specific ICD codes for disease. For example, we used separate coding for hypertension, myocardial infarction, and CVD, whereas in other epidemiologic studies, these are often combined within one definition of CVD. 18 , 51 , 52 , 53 , 139 The diagnosis of PCOS included the ICD-9 and ICD-10 codes, and these have been validated and used in previous epidemiologic studies. 6 , 10 , 17 , 27 , 32 , 55 , 56 , 57 The limitations of the study are that the administrative database does not contain information on quality-of-life indices, gender, socioeconomic status, lifestyle, or dietary habits, all of which can significantly impact health outcomes in different populations and across the lifespan. Individual patient laboratory data used as biomarker risk factors were not accessed to investigate their potential mechanisms and association with disease outcomes, to determine potential causality in this study. We used validated ICD codes for disease prevalence, based on the ICD code first inputted into the administrative database, but this coding can be based on symptomology at the primary-care or acute-care level, which may cause overestimation of prevalence rates. In future studies, a secondary or tertiary inputting of the ICD code to confirm a diagnosis of a disease, such as a malignant cancer or dementia, rather than related symptomology, is recommended. In addition, primary-care claims may have codes that were inputted for the purposes of billing, based on symptomology, and this may impact assessment of prevalence rates, particularly for gastrointestinal and respiratory complaints. Health system–encounter activity may also impact prevalence rates for all conditions, owing to the greater frequency of code input, and in future studies, we will address the matching of cases and controls for this activity. The diagnosis of PCOS included the ICD-9 and ICD-10 codes, but it did not include codes for symptomology of PCOS, including hyperandrogenism (hirsutism) and menstrual dysfunction, which may be coded separately and are criteria that are used to diagnose PCOS. 58 Not including these symptoms in the PCOS exposure definition may underestimate the number of cases in epidemiologic studies, but it may also prevent inclusion of nonconfirmed PCOS and non-PCOS cases that do not present with 2 out of 3 symptoms of the diagnostic criteria. 18 , 55 Different phenotypes of PCOS have been linked to severity of symptoms and risk for adverse health outcomes. 55 We have not explored the phenotypes of PCOS in this study, as this information is not directly available in the databases and requires validation on a case-by-case basis using patient chart review, which was beyond the scope of this cohort study. 55 Individual symptoms were not used to define different conditions, such as the metabolic syndrome; therefore, the prevalence of this syndrome may have been underestimated in the population cohort we examined. We have not explored the prevalence of comorbidities in those with PCOS compared to non-PCOS controls or to those with one of the entire ICD of diseases, but rather, we have provided an epidemiologic snapshot into common diseases over the lifespan. In this study, we have not investigated medication use and its relationship to disease outcomes, or economic health costs, but these research questions can be explored in future studies using this population cohort.