Women's Health Disorders in a Coeliac Disease Population After Diagnosis-A Nationwide Cohort Analysis.

Rama Nanah: investigation, formal analysis, writing – original draft, methodology, conceptualization, software. Claire Jansson‐Knodell: writing – review and editing, writing – original draft, methodology, supervision, investigation, validation, conceptualization, visualization. Arjun Chatterjee: writing – review and editing, writing – original draft. Robana Nanah: writing – review and editing, writing – original draft. M. Housam Nanah: writing – original draft, writing – review and editing. Jehad Almasri: writing – review and editing. Andrew Ford: software, formal analysis. Osama Hamid: writing – review and editing. Ahmed Telbany: writing – review and editing, writing – original draft. Alberto Rubio‐Tapia: supervision, conceptualization, methodology, validation, writing – review and editing, visualization, writing – original draft.

This study is a case–control retrospective analysis using the TriNetX database, which pulls from electronic medical records of participating health care organisations (HCO) in the US. TriNetX is Health Insurance Portability and Accountability Act (HIPAA) compliant. Any data displayed on the TriNetX platform contains only de‐identified data as per HIPAA standards. As such, this study was exempt from Institutional Review Board approval. Study data was collected on May 17, 2023. Coeliac disease was defined using ICD‐10 code (K90.0) along with one or more of the following: a diagnosis of Dermatitis Herpetiformis ICD‐10 (L13.0), abnormal histological findings in specimens from digestive organs and the abdominal cavity (R85.7), dietary counselling and surveillance (Z71.3), a prior positive result or numerical value ≥ 30 U/mL of tissue transglutaminase immunoglobulin A (IgA) antibody or deamidated gliadin peptide, or a prior positive result of endomysial antibody IgA titre in serum by immunofluorescence. This case identification methodology is similar to prior published work [ 11 , 12 ]. They could not have ICD‐10 code for non‐coeliac gluten sensitivity (K90.41). Additional inclusion criteria were female sex and age 10–60 years. Exclusion criteria were patients seen > 20 years ago when serology testing was not as reliable, and general knowledge of coeliac disease and testing efforts were not as strong, to attempt to decrease rates of undetected coeliac disease and patient misclassification. The control group was defined as ambulatory women without an ICD‐10 code for coeliac disease (K90.0), positive serology test (tissue transglutaminase IgA, deamidated gliadin peptide, or endomysial antibody), or any of the above‐mentioned ICD‐10 codes charted. We planned for an initial exploratory comparison between women with coeliac disease and controls to identify overall trends in post‐diagnosis women's health outcomes (Appendix  S1 ) requiring more in‐depth study. We then stratified our populations based on different hormonal stages in women's lives. As women tend to have different concerns and prevalent disorders throughout their reproductive lives, we selected age groups to represent these different stages: (1) Paediatric (10–18 years old), (2) Reproductive (19–35 years old), (3) Later reproductive and pre‐menopausal (36–45 years old) and (4) peri‐ and post‐menopausal (46–60 years old). Then, we performed propensity score matching for each group of subjects for age, race, and BMI to minimise the effect of confounders. A cut‐off of 35 years was selected to divide reproductive from later reproductive groups per the obstetrical definition of advanced maternal age. The index event to determine age was coeliac disease diagnosis or an outpatient encounter in non‐coeliac disease patients. Patients with the outcomes (studied women health disorders) prior to the diagnosis of coeliac disease or the outpatient encounter were excluded. Women's health disorders across age groups. Descriptive statistics were used to compare the baseline characteristics of the two groups utilising counts and percentages. The number of patients with the outcomes was obtained, and risk differences were calculated. Chi‐square analyses were performed to compare categorical parameters, and an analysis of variance (ANOVA) was performed to compare the continuous variables, including age and BMI. Odds ratios (OR) with 95% confidence intervals (CI) were calculated to compare the risks of developing the outcomes between the coeliac and non‐coeliac women. p  < 0.05 was used for statistical significance. Propensity score matching for baseline characteristics was performed using a built‐in algorithm, which was based on 1:1 nearest neighbour matching with callipers of 0.1 of the standard deviation. Statistical analysis was completed using the TriNetX platform analytics (Cambridge, MA, US). Initial data access and analysis was completed on May 17, 2023. Additional variables of delayed menarche and premature menopause were analysed on August 24, 2023, to further describe fertile lifespan.

The exploratory study included > 25 million outpatient women without coeliac disease and 9368 women with coeliac disease. The women with coeliac disease were younger at the time of coeliac disease diagnosis or first outpatient encounter (25 vs. 28 years old). Women with coeliac disease had lower BMIs compared to women without coeliac disease (24.6 vs. 26.1) and were predominantly of white race (83% vs. 47.7%) (Table  1 ). Overall sample demographics. Exploratory data showed that women with coeliac disease compared to women without coeliac disease have significantly higher odds of numerous women's health conditions relating to ovarian function, menstruation, and fertility with onset after their coeliac disease diagnosis. Women with coeliac disease had higher rates which translated into higher odds of primary ovarian failure (0.96% vs. 0.16% respectively; OR 6.25, 95% CI: 4.89, 7.41), ovarian dysfunction (4.4% vs. 1.25%; OR 3.64, 95% CI: 3.30, 4.02), and polycystic ovarian syndrome (PCOS) (3.3% vs. 1%; OR 3.2, 95% CI: 2.94, 3.68). More instances of irregular menstruation meaning excessive, frequent, or abnormal uterine or vaginal bleeding (15.5% vs. 6.9%; OR 2.45, 95% CI: 2.32, 2.60), absent/rare menstruation (4.6% vs. 2%; OR 2.34, 95% CI: 2.13, 2.58), and dysmenorrhea (10.9% vs. 4.24%; OR 2.76, 95% CI: 2.58, 2.94) were reported in women with coeliac disease. Infertility was more often coded in women with coeliac disease (1.44% vs. 0.85%; OR 1.69, 95% CI: 1.43, 2.01) and recurrent pregnancy loss (0.26% vs. 0.18%, OR 1.45, 95% CI: 1.01, 2.07). No significant differences were seen for the outcomes of spontaneous abortions or premature births. Endometriosis was more common in women with coeliac disease (2.3% vs. 0.93%; OR 2.53, 95% CI: 2.21, 2.90). Menopausal and peri‐menopausal disorders occurred at higher rates in women with coeliac disease (4.3% vs. 1.56%; OR 2.85, 95% CI: 2.58, 3.15). Given these differences in the overall sample's exploratory analysis, we moved forward with age stratifying women to study pertinent conditions for each age segment (Figure  1 ). This age range captures the paediatric age group and puberty. After propensity score matching for age, race and BMI, 2431 patients from each group were analysed (Table  2 ). Outcomes in paediatric group (10–18 years). Girls with coeliac disease had higher rates of delayed menarche (OR 3.63, 95% CI: 1.8, 7.33) and absent and rare menstruation (OR 2.48, 95% CI: 1.51, 4.09) compared to girls without coeliac disease. PCOS was twice as likely in girls with coeliac disease, with an OR 2.11 (95% CI: 0.99, 4.48). Menstrual pain and cyst issues (classified as non‐inflammatory disorders of the ovary, fallopian tube, and broad ligament) were similar between the groups. No cases of endometriosis were documented. This age group encompasses early adulthood and the reproductive phase of life. After propensity score matching for age, race and BMI, 3488 patients from each group were analysed (Table  3 ). In this age group, absent and rare menstruation (OR 1.75, 95% CI: 1.40, 2.19) and PCOS (OR 2.79, 95% CI: 2.04, 3.80) continued to occur at higher odds in the coeliac disease population. These conditions were diagnosed at higher rates in this adult group compared to the adolescent group. Concerning fertility and peripartum issues, women with coeliac disease had higher rates of infertility (OR 1.74, 95% CI: 1.07, 2.82), but lower rates of labour and delivery complications (OR 0.81, 95% CI: 0.66, 0.99). There were no significant differences between women with coeliac disease and matched controls for recurrent pregnancy loss (OR 1.20, 95% CI: 0.51, 2.78) in this age group in contrast to the overall analysis. Endometriosis emerged in this age group as problematic for women with coeliac disease, with an OR of 2.69 (95% CI: 1.68, 4.32). Irregular menstruation, described as excessive, frequent, and irregular or abnormal uterine or vaginal bleeding, was also noted to be higher in women with coeliac disease (OR 1.84, 95% CI: 1.60, 2.12) as was dysmenorrhea (OR 2.01, 95% CI: 1.71, 2.36). Outcomes in early adulthood/reproductive group (19–35 years). This age range captures the later reproductive and pre‐menopausal years (Table  4 ). After the propensity score matching, we analysed 1527 patients from each group. In this age group, ovarian dysfunction occurred at higher odds in women with coeliac disease compared to women without coeliac disease (OR 4.16, 95% CI: 2.56, 6.10). PCOS occurred more often with OR 3.68 (95% CI: 2.28, 5.94), as did irregular menstruation with OR 2.04 (95% CI: 1.66, 2.50), pain with menstrual cycle with OR 1.92 (95% CI: 1.49, 2.48), and endometriosis with OR 1.79 (95% CI: 1.17, 2.75). In these later fertile years, infertility became a greater issue with higher rates in women with coeliac disease (OR 2.5, 95% CI: 1.56, 4.00). While there were rate differences for recurrent pregnancy loss and labour and delivery complications, these effect sizes were small and not noticeably different between the groups. Menopausal issues began to appear in this age group, including absent or rare menstruation (OR 1.63, 95% CI: 1.13, 2.36) and menopausal or perimenopausal disorders (OR 4.18, 95% CI: 2.08, 8.38). Outcomes in later reproductive and pre‐menopausal group (36–45 years). This final group is characterised by menopause (Table  5 ). After the propensity score matching, 1918 patients from each group were included in the analysis. In this age group, most women's health issues occurred with higher frequency in coeliac disease patients compared to controls. In particular, for premature menopause, the OR was 3.13 (95% CI: 1.53, 6.40); for primary ovarian failure, the OR was 3.0 (95% CI: 1.47, 6.21); for ovarian dysfunction, the OR was 5.22 (95% CI: 2.93, 9.30); and for absent/rare menstruation, the OR was 2.22 (95% CI: 1.30, 3.79). Menopausal and perimenopausal disorders were more common in women with coeliac disease, with OR 2.28 (95% CI: 1.84, 2.83). Rates of PCOS, endometriosis, dysmenorrhea, and irregular menstruation were also significantly higher for women with coeliac disease. Outcomes in perimenopausal, menopausal, and postmenopausal groups (46–60 years).

The main finding of this study is that women with coeliac disease appear to exhibit higher rates of later women's health disorders compared to their counterparts without coeliac disease. Individual age groups representing different hormonal stages showed increased prevalence of many women's health issues throughout all stages of life after propensity score matching for age, race, and BMI. Age‐specific concerns evolved over the study groups, mirroring key obstetric and gynaecologic concerns for each range of years. Putting this nationwide database study in the context of coeliac literature, there are some notable differences. Older studies have reported a significant delay in the age of menarche in girls with coeliac disease, with higher rates among those not compliant with a GFD. This finding, along with reported premature menopause, raises concern for a shortened reproductive period [ 13 , 14 ]. In a single‐centre retrospective survey from Brazil, menarche was slightly delayed in girls with coeliac disease at 12.8 years ±1.22 versus 12.6 years ±1.40 with earlier menopause in women with coeliac disease at 48.3 ± 1.82 versus 49.0 ± 1.56 years [ 15 ]. On the other hand, a Slovenian study showed the age at menarche in girls with coeliac disease to be comparable with girls without coeliac disease [ 16 ]. These studies included 54, 74, 214 and 145 patients, respectively, whereas our nationwide analysis included coeliac disease patients in the thousands, increasing our confidence in the findings. Although we were unable to detect the exact age of menarche or menopause using the database, we found higher odds of delayed menarche in girls with coeliac disease and higher odds of premature menopause in women with coeliac disease. In addition to a suggested shortened window of menstruation, our study also adds to the literature on coeliac disease and fertility. Several studies have explored the prevalence of coeliac disease in infertility with variable results. Most studies focused on coeliac disease screening in women with infertility, instead of comparing infertility rates among matched coeliac disease and non‐coeliac disease patients, setting ours apart. One prospective study from Mayo Clinic screened infertile women to find undiagnosed coeliac disease (biopsy confirmed) in 2.1% of patients. This rate nearly tripled to 5.9% in patients with unexplained infertility [ 17 ]. In contrast, an American prospective cohort study screening 121 women showed no increased coeliac disease prevalence in women with unexplained infertility (0.8%) compared to the general population [ 18 ]. A more recent meta‐analysis pooling data from 11 studies applied strict inclusion criteria of biopsy‐proven coeliac disease (Marsh 3—villous atrophy) and excluded all women whose infertility was explainable in part by male infertility, reporting that the pooled prevalence of biopsy‐confirmed coeliac disease was 0.7% in women with infertility [ 9 ]. We demonstrated a higher odds ratio of infertility in women with coeliac disease at 1.69; more specifically, the OR was 1.74 in the younger [ 8 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 ] adult group and 2.50 in the older (36–45) adult group. Given the decreased reproductive period noted in previous studies and suggested by our results, we evaluated the rates of spontaneous abortions and premature births. A recent paper was published on reproductive health outcomes in women with coeliac disease using the National Inpatient Sample from the US. There were no significant differences in preterm births (OR 0.98, 95% CI: 0.81–1.19) or stillbirths (OR 1.4, 95% CI: 0.45–4.35) in women with coeliac disease compared to controls [ 19 ]. Our results were consistent with these findings for preterm births. We also examined endometriosis, an understudied, yet important topic in coeliac disease given it can cause significant, poorly localised abdominal pain mimicking gastrointestinal conditions. One large Swedish study included women with coeliac disease > 35 years and found slightly higher rates of endometriosis in those with coeliac disease (1% vs. 0.7%) [ 20 ]. Our study had higher rates of endometriosis in both coeliac disease (3.9%) and non‐coeliac disease (2.2%) populations aged 36–45 years compared to the Swedish study, possibly because the Swedish study used more selective inclusion criteria of Marsh 3 biopsy‐proven coeliac disease. An Italian study flipped the predictor and outcome to find a rate of biopsy‐proven coeliac disease of 2.2% in patients with endometriosis vs. 0.8% in controls without endometriosis that was not significantly different [ 21 ]. Our study explored PCOS, a common endocrine disorder, affecting 6%–10% of reproductive age women worldwide with various degrees of infertility [ 22 ]. One study looked for but found no cases of biopsy‐proven coeliac disease in 52 PCOS patients [ 23 ]. PCOS is associated with obesity, while coeliac disease can be associated with malabsorption and low BMI. Interestingly, our study reports higher rates of PCOS in women with coeliac disease of reproductive ages (4.3% vs. 1.6% in the younger and 5.1% vs. 1.4% in the older age group) compared to women without coeliac disease controls. Our study aligns with one small study on dysfunctional uterine bleeding (DUB) that included 50 women with coeliac disease versus 70 without coeliac disease. This Iranian study looked at biopsy‐proven (Marsh Criteria 1‐3c) women with coeliac disease to report higher rates of DUB in 12 women with coeliac disease and 7 without (24% vs. 10%) after matching for demographics and BMI with an OR of 2.84 (95% CI: 1.03, 7.84), similar to our overall OR of 2.45 [ 8 ]. Given the observational nature of our study, we are only able to describe associations and are not able to draw conclusions about causation. Some proposed aetiologies include nutritional deficiencies, inflammatory state, chronic disease status, and autoimmunity via an inhibitory effect of anti‐transglutaminase antibodies on endometrial angiogenesis to explain these associations [ 8 , 24 ]. The amount of adipose tissue for the creation of sex hormones, dysfunction of the hypothalamic–pituitary –ovarian axis, and altered T‐helper 1 cell immune responses may also connect coeliac disease to women's health issues [ 20 , 25 , 26 , 27 , 28 , 29 , 30 ]. Further research is needed to elucidate the underlying mechanisms. The major strength of our study is the large cohort from a nationally representative dataset featuring racial, ethnic, and geographic diversity. Many prior studies were smaller, had less diversity, and were more concentrated geographically. The database used allowed for the selection of additional criteria along with ICD‐10 codes to increase our confidence in identifying coeliac disease patients. Conversely, diagnosis codes may adversely impact the study via coding errors. Using ICD‐10 codes inherently introduces variability since coding practices may differ across institutions, networks, and providers, potentially leading to misclassification of cases. The use of multiple ICD‐10 codes to define coeliac disease may have increased specificity but likely decreased sensitivity, missing individuals with coeliac disease who had few healthcare encounters or a single code. This methodology may explain the low prevalence in the study compared to nationally representative estimates from the United States [ 31 , 32 ]. The prevalence of diagnosed coeliac disease is around 10‐fold higher in that research compared to our data. As our study captured only a subset of women with coeliac disease, it may not be fully representative, which has implications for generalisability. Using a database for this research means it was also subject to ascertainment bias—those with higher healthcare utilisation or higher socioeconomic status in order to access care are over‐represented—which may limit applicability to a subgroup of similar women with coeliac disease. In our quest for specificity, we may have inadvertently selected a population with greater healthcare contact and thus more opportunities to detect additional conditions, biasing results. With this database, we had no ability to review tissue samples or histology results to verify coeliac disease diagnosis; this lack of direct validation of the case definition could influence results as we could not be certain cases truly represented individuals with coeliac disease. Another limitation is that the study was retrospective. Events or outcomes were captured after coeliac disease diagnosis, and this may have underestimated conditions like infertility which can manifest prior to coeliac disease diagnosis and resolve with treatment. This dataset lacked information on whether patients were following a GFD. This limitation is key as prior research indicates that treatment and GFD compliance mitigate some of these conditions. Some research reports delayed menarche in coeliac disease girls not compliant with the GFD [ 33 ], but other work found no effect of GFD on age of menarche [ 34 ]. Ciacci et al. compared untreated women with coeliac disease, women with coeliac disease on a GFD, and healthy volunteers to show a shorter fertile lifespan in untreated women with coeliac disease despite higher rates of alcohol and tobacco use among controls and a longer fertile lifespan in women with coeliac disease on a GFD. Untreated women with coeliac disease reported more severe hot flashes and irritability perimenopausally [ 30 ]. A Brazilian study reported delayed menarche and amenorrhea in girls with coeliac disease not compliant with a GFD but no differences in total number of pregnancies, age at menopause, or duration of reproductive lifespan in women with coeliac disease [ 33 ]. The GFD reportedly improves irregular menstruation, DUB, and endometriosis symptoms in women without known coeliac disease and is considered a novel management strategy [ 35 , 36 ]. Lastly, non‐Caucasian patients were under‐represented in this dataset. This did not allow for an adequately powered racial or ethnic disparities analysis to address this recurring gap without overgeneralising. Differing prevalence may have been due to disparities in access to care, lower coeliac testing rates in non‐Caucasian groups, environmental factors, or genetic predisposition; we can only speculate as the data available did not provide direct insight into causation.

Our study found higher rates of subsequent women's health disorders in women with coeliac disease of all studied age groups. The paediatric group had higher rates of delayed menarche. The early reproductive group had higher rates of infertility and dysmenorrhea. The later reproductive group had higher rates of ovarian dysfunction, endometriosis, and PCOS. The menopausal age group had higher rates of premature menopause and menopausal disorders. This observational study is not able to comment on the mechanism behind the association between coeliac disease and women's health disorders. Large‐scale, prospective studies are needed to better understand the role coeliac disease plays in women's health and elucidate the impact of the GFD in the reproductive and gynaecologic health of women with coeliac disease. Our findings suggest that gastroenterologists should be aware of these higher rates of women's health issues in coeliac disease patients. Early referral to obstetrics and gynaecology may be beneficial for women with coeliac disease experiencing ongoing abdominal symptoms despite the GFD to investigate for these conditions. Additionally, with the possible decreased fertile lifespan, early identification may be key for reproductive success and a family for coeliac disease patients desiring one.

Coeliac disease is an underappreciated immune‐mediated enteropathy triggered by ingestion of foods containing gluten. Coeliac disease is a global problem [ 1 , 2 , 3 ] diagnosed more often (60%–70%) in women [ 4 ]. Global pooled data report biopsy‐proven coeliac disease in 0.6% of females with even higher prevalence for serology‐based coeliac disease diagnosis [ 5 ]. Women also have higher pooled rates of undetected coeliac disease compared to men in data amassed from population screening (0.58% vs. 0.41%) [ 2 ] studies. Our understanding of not just who is affected by coeliac disease, but how they are affected by coeliac disease has also changed. Epidemiology has shifted in adult coeliac disease towards non‐classical presentations. The Rochester epidemiology project's coeliac registry showed a higher overall rate of non‐classical (50.3%) and asymptomatic presentations (5.5%) compared to classical presentations (44.3%) [ 4 ]. Classically, coeliac disease has gastrointestinal manifestations related to malabsorption. However, with rising rates of non‐classical presentations, clinical suspicion arises from extra‐intestinal manifestations including dermatologic disorders, neuropsychiatric features, abnormal liver biochemistry tests, reproductive abnormalities and endocrinological manifestations [ 3 , 6 , 7 , 8 ]. Sex differences in coeliac disease manifestations, including non‐classical ones, have been reported [ 1 , 2 , 9 ]. In prior studies, women present younger and have different rates of both gastrointestinal and extra‐intestinal manifestations [ 1 ]. When compared to men, women had higher rates of depression and osteoporosis [ 10 ]. Other studies found differences in associated autoimmune disorders, with hypothyroidism being more predominant in women with coeliac disease [ 4 ]. Additionally, increased prevalence of coeliac disease has been noted in several reproductive disorders [ 9 ]. Late menarche, secondary amenorrhea, early menopause, pregnancy, and labor and delivery complications have been reported in non‐gluten‐free diet (GFD)‐controlled women with coeliac disease [ 6 ], in the setting of nutritional deficiencies irrespective of body mass index [ 1 , 2 , 6 ]. Given the relatively high prevalence of women with coeliac disease, rising rates of non‐classical coeliac disease presentations, and the impact of sex differences on coeliac disease symptoms, we aimed to examine the rates of women's health disorders affecting women with coeliac disease in the United States (US) to observe whether associations exist after coeliac diagnosis.

The authors declare no conflicts of interest.

Appendix S1.