Depression in Female Adolescents with Heavy Menstrual Bleeding.

The University of Michigan Medical School Institutional Review Board approved this retrospective cohort study. Two groups were identified, adolescent females 9–18 years with and without HMB. These groups were ascertained from the electronic medical record (EMR) using DataDirect, a self-serve tool enabling access to clinical data. Subjects in both groups were seen in a general pediatrics clinic between January 1, 2018 – November 19, 2019. For the HMB group, we initially identified 747 adolescent using ICD-10 codes. The electronic medical records of patients were first reviewed to confirm the diagnosis of HMB, defined as bleeding >7 days, changing products more than every 2 hours, presence of large clots 7 . Following initial review, 166 patients were excluded, as they did not meet criteria for a diagnosis of HMB, leaving 581 charts for further evaluation. Using this same process, 587 post-menarchal controls were identified who had presented for a well-child visit in general pediatrics clinic during this same time period and who did not report HMB. The following variables were collected for each patient: age at presentation for HMB or at well child visit for those without HMB, body mass index (BMI), race, ethnicity, diagnosis of depression, diagnosis of anxiety, PHQ-9 scores at diagnosis (when available), treatment for depression as indicated in the medical notes and medication lists (including antidepressant medications, therapy, whether they were seen by psychiatry), hormonal contraception use and type, sexual activity, and substance use. Depression was defined as a diagnosis entered into the medical record by a physician either within the problem summary list, diagnosis list, or within visit notes. Patients who reported concern for depression or depressive symptoms but were not ultimately diagnosed with depression by a physician were excluded. To evaluate the possible effect of HMB and hormonal contraception on the development of depression, the clinical time course was also collected with regards to onset of HMB, depression, and use of hormonal contraception. Age at development of HMB, age at development of depression or anxiety, and age at initiation of hormonal contraception were recorded in years. If multiple events occurred at the same age in years, further detail was recorded regarding the temporal relationship of these events. Events were coded as simultaneous if they occurred within one week of each other. Median follow-up was 12 months. Descriptive statistics were calculated for all variables with means and ranges determined for continuous variables and counts and percentages for categorical variables. Poisson regression, instead of logistic regression, models were fit to the data in three stages. First, we fit two univariable models to assess the individual associations of HMB with depression and hormonal contraception with depression. We then fit a model with both HMB and hormonal contraception together to determine any potential mediation of HC on the association of HMB with depression and vice versa. Our final model included all other characteristics that showed an association with depression, HMB, and/or hormonal contraception to control for potential confounding. No interactions were examined as none were hypothesized a priori . Due to missing values for substance use (956 subjects had data reported) and sexual activity (1010 subjects had data reported), the final multivariable regression model was fit to data from the 909 patients with no missing values. All Poisson regression model results are summarized as risk ratios (instead of odds ratios) and corresponding 95% confidence intervals, using robust standard errors to correct for the inappropriate variance structure assumed by Poisson regression. Statistical significance is defined as a p-value less than 0.05; all statistical analyses were performed in R version 4.0.4 and robust standard errors were computing using the R library sandwich .

For all 1168 patients, the mean age at presentation was 14.5 years and the mean BMI was 25.2 k/m 2 . Breakdown of race and ethnicity was representative of our state population demographics. Patients with and without HMB were similar in age, race, ethnicity, and BMI. Overall, depression was seen in 37.5% of patients, with 50.9% in group with HMB and 24.2% in patients without HMB (p<0.001) ( Table 1 ). When comparing patients with depression to those without ( Table 1 ), depression was associated with older age, higher BMI, HMB, use of hormonal contraception, anxiety, sexual activity, and all types of substance use. In the multivariable regression model, depression was associated with HMB, higher BMI, anxiety, sexual activity, smoking and use of marijuana ( Table 2 ). PHQ-9 scores were available at diagnosis in 67.2% (199/296) of those with HMB and 71.8% (102/142) of those without (p=0.33), with scores of 5–14 interpreted as mild to moderate depression, and 15 – 27 as moderately severe to severe depression 16 . Mean PHQ-9 in the patients with HMB and depression was 14.7 (SD 5.3), compared to 13.6 (SD 4.9) in those without HMB with depression (p=0.08). Of the 438 patients who were depressed, 240 (54.8%) were seen by psychiatry, 289 (66%) were prescribed antidepressants, 378 (86.3%) participated in therapy, and 262 (59.8%) received both antidepressants and therapy. Only 34 (7.8%) of depressed patients received no treatment at all. A greater proportion of patients with HMB and depression saw psychiatry compared to those with depression without HMB (58.5% vs 47.2%p=0.03). Otherwise, there was no difference in treatment for depression between those with HMB and those without (66.9% vs 64.01%, p=0.55 receiving antidepressants, 87.5% vs 83.8%, p=0.29 participating in therapy, 62.2% vs 54.9%, p=0.15 receiving both antidepressants and therapy, 7.7% vs 7.7%, p=1.00 not receiving any treatment at all). As with the diagnosis of depression, the number of patients with a medical record diagnosis of anxiety was also significantly different between those with HMB and those without (46.1% vs 21.5% respectively, p< 0.001). When comparing those with HMB to those without ( Table 1 ) there were statistically significant differences between the two groups regarding the likelihood of depression, anxiety, and use of hormonal contraception. When comparing those who received hormonal contraception to those that did not ( Table 1 ), those on hormonal contraceptives tended to be older, with higher BMI, increased likelihood of depression and anxiety, more likely to engage in sexual activity and substance use, and more likely to have HMB. Most patients who used hormonal contraception in the HMB group and the control group used combined hormonal contraception (434/450 (96.4%) and 185/200 (92.5%), p=0.03), with more patients with HMB utilizing combined hormonal contraception than those without. The number of patients using other forms was too small to compare outcomes between specific hormonal contraception used. On univariable analysis, both HMB and use of hormonal contraception were associated with depression (RRs of 2.10 (95% CI 1.78–2.48) and 1.91 (95% CI 1.61–2.26)) respectively. Bivariable analysis using HMB and use of hormonal contraception demonstrated mediation with decreasing risk ratios for both HMB and use of hormonal contraception (1.78 (95% CI 1.48–2.14) and 1.48 (1.23–1.79) respectively). On multivariable regression, the association between HMB and depression remained (RR 1.66 (95% CI 1.38–2.00) although the association between use of hormonal contraception and depression did not (RR 0.99 (95%CI 0.84–1.17)). The timing of depression diagnosis relative to the development of HMB for those patients with HMB, and relative to hormonal contraception for all patients was reviewed. Of the 296 patients with HMB and depression, 261 (88.2%) developed depression at the time of or following the development of HMB, whereas 35 (11.8%) had already been experiencing depression at the time of HMB onset. Of the 261 patients who developed depression after HMB onset, 199 (76.2%) received hormonal contraception, but only 31(15.6%) developed depression after it was initiated. Thus, of the 296 patients with HMB and depression, only 31 (10.5%) had a clinical time course in which depression could be secondary to, or following, use of hormonal contraception. Of those patients without HMB who had depression, 88 patients (62%) received hormonal contraception. Only one quarter (22 patients, 25%) developed depression following use of hormonal contraception with the majority experiencing depression prior to initiation.

Our study identified HMB as a new association with depression diagnoses, and confirmed prior findings of associations between BMI, sexual activity, and substance use with depression. Over half (50.9%) of the adolescent females with HMB had a depression diagnosis, compared to 24.2% in those without HMB. Prevalance data from the National Survey on Drug Use and Health (NSDUH) found a 21.5% prevalence of major depressive disorder in adolescent females aged 12–17 years 18 . Both groups with depression in our study (those with HMB and those without) had a much higher rate of treatment than reported in the general population with depression(>90% in our study, compared to <40% in the NSDUH). This may be due to the retrospective study design increasing the likelihood of picking up diagnoses of patients being treated. Given the known effects of HMB on HRQOL, it is not surprising that HMB could also have substantial effects on mood. The relationship between depression and HMB is likely complex and bidirectional. Published data suggest that depression and anxiety may precede HMB, although in this study only 11.8% of those with depression developed it prior to HMB onset. One study found that midlife women with a history of major depression were more likely to report HMB 19 . Older women with anxiety also may be more likely to seek care for subjective HMB even in the absence of increased blood loss 20 . Some treatments for depression, such as selective serotonin reuptake inhibitors cause increased bleeding tendency and may contribute to HMB 21 . Obesity, substance use, and sexual acitivtiy have all been found to have bidirectional relationships with adolescent depression 22 , 23 , 24 and in our study were also associated with use of hormonal contraception. Users of hormonal contraceptives were more likely to be sexually active as this information likely contributed to the decision to treat in some patients. Co-occurrence of sexual activity and substance use have been found to cluster in adolescent patients 25 . If obesity, sexual activity and substance use correlate with adolescent depression, and are more frequent in patients prescribed hormonal contraception, these variables may be important confounders in this relationship. Our findings suggest HMB may be an additional risk factor for depression diagnosis. Given that HMB is commonly treated with hormonal contraception, it is possible that previously reported links between hormonal contraception and depression could be partially explained by higher rates of HMB in these patients. The effects of hormonal contraception on mood is an important and controversial issue, with inconsistencies in published data. The reasons for initiation of hormones are varied and complex, ranging from acne treatment, menstrual control, to contraception. Choice of specific hormonal methods has been shown to be affected by underlying mental health conditions, medical comorbidities, family planning, perceptions of risks and benefits, cost and availability, and demographic factors 26 – 28 . All of these factors may confound any previously ascribed association with depression. A large population study in Denmark 8 reported an association between any hormonal contraception and first time use of antidepressants, as well as depression diagnosis. However, this study did not account for reason treatment was initiated so would have missed HMB as a possible confounder. Notably, the association was greater in younger patients, who may be more likely to initiate treatment for HMB rather than for contraceptive purposes. The study also did not include data on substance use or sexual activity, which also may have confounded their findings. This group also reported an association between increased suicide attempts in patients who received hormonal contraception 9 , but did not account for substance use, sexual activity, or reason for initiation. The highest risk for suicide and suicide attempts was seen in former users of hormonal contraception, which could reflect the presence of confounding variables influencing the observed hormonal contraception-depression association rather than direct hormonal effects of HC on mood. A similar study was done in Sweden 10 , which also reported an association between hormonal contraception use and antidepressant prescriptions that was most pronounced in the youngest patients (12–14 years of age) and disappeared in the older ages. “Menstrual disturbances,” which included abnormal bleeding, premenstrual disorders, and endometriosis, were accounted for and were significantly different between users and non-users in the 12–14year age group. Sexual activity, substance abuse, and BMI were not included in their analysis. TRAILS, a large cohort study performed in the Netherlands, had similar results of an association between combined hormonal contraception and depression which was only seen in the 16-year-old group (the youngest in the study). This association decreased when accounting for sexual activity, stressful events, menstrual pain, and acne. The symptoms that were significantly different between users and nonusers at age 16 were crying, hypersomnia, tiredness, and loss of energy. Although other variables were considered, indication for contraceptive initiation and substance use were not included. Although BMI was collected, and was significantly different between users and nonusers, it does not appear to have been included in multivariate analysis. Of note, users and nonusers had statistically significant differences in depressive scores at age 13, an age at which the authors state that hormonal contraception use was unlikely 11 , further supporting the possibility of unmeasured confounding factors. A cohort study in Australia 12 found no increase in depression when hormonal contraception was used for contraception. There was, however, an increase in depression in patients taking hormonal contraceptives for other indications, such as acne or dysmenorrhea. The presence of such confounders in the association of depression and hormonal contraception is further supported by many studies reporting no increased risk of depression, or even decreased risk of depression 13 – 15 . One randomized, double blinded, placebo-controlled study 16 found mixed effects. Combined hormonal contraceptives were associated with an increase in anxiety and mood swings during the inter-menstrual phase. However, significant improvement in depression scores was also seen in the premenstrual period. The proportion of women with clinically significant mood worsening did not differ between users and nonusers. The majority of the patients in our study presenting to primary care providers were treated with combined hormonal contraceptives. Bivariable analysis demonstrates that HMB is a mediator of the effect of hormonal contraceptive use and vice versa. Accordingly, the association between hormonal contraception and depression seen in the youngest patients within large population studies may partially reflect confounding by HMB. This issue is particularly important as hormonal contraceptives are effective in the treatment of HMB 29 . If HMB is proximal, and potentially partially etiologic to depression onset, then effective treatment may result in improvements in depressive symptoms as well. Our findings demonstrate that the majority of patients with HMB and depression develop depression following HMB based upon the timeline recorded in the electronic health records of diagnoses dates and medications prescribed. Of those who developed depression following HMB, that also received hormonal contraception, the majority started hormonal contraception following the diagnosis of depression, suggesting the hormonal treatment was not etiologic. Unfortunately, due to published data suggesting risk of depression with any hormonal contraception, patients presenting to care with HMB may be less likely to receive effective treatment for bleeding. Further, there are data that hormonal contraceptives can be used safely in patients with pre-existing mood disorders. A review of cohort, randomized control, and case control studies of women with depression and bipolar disorder using hormonal contraception found no increased symptoms or hospitalizations 30 . Limitations of our study include its retrospective nature, which relies on the record keeping of providers and thus data may be incomplete, for instance lack of data on sensitive topics, such as substance abuse and sexual activity. Our study is subject to confounding. Specifically, age at menarche, or presence of other menstrual concerns such as dysmenorrhea may confound these results. Further, menarche coincides with a period of increased incidence in depression, and we are only able to determine association rather than causation 31 . We did not collect data on psychosocial risk factors such as poverty and stressful life events that are associated with depression 32 . The vast majority of patients were using combined oral contraceptives so we were unable to compare between types of hormones used and association with depression. Patients were seen by primary care providers associated with a large medical center so findings may not be generalizable to other populations. Depression in adolescence is under-reported and under-recognized 18 with the majority of patients not receiving treatment. Given the prevalence of treatment in our study, this may be an underestimate of the true prevalence of depression in this population.