Pregnancy, Estrogen Exposure, and the Development of Otosclerosis: A Case-Control Study of 1196 Women

Abstract

Objective.

This study sought to determine whether a history of pregnancy or bilateral oophorectomy is associated with subsequent otosclerosis development or disease severity.

Study Design.

Population-based case-control study.

Setting.

Olmsted County, Minnesota.

Methods.

Women diagnosed with otosclerosis were matched to 3 women without otosclerosis based on age and historical depth of medical records. Associations of prior delivery and bilateral oophorectomy with subsequent development of otosclerosis and with pure-tone average (PTA) at the time of otosclerosis diagnosis were evaluated.

Results.

We studied 1196 women: 299 cases of otosclerosis and 897 matched controls. The odds ratio for the association of ≥1 delivery with otosclerosis was 1.16 (95% confidence interval [CI] 0.85–1.60; P = .35). Odds ratios for the associations of 1, 2, 3, or ≥4 deliveries with otosclerosis were 1.22 (0.83–1.80), 1.09 (0.71–1.68), 1.28 (0.77–2.12), and 1.00 (0.54–1.84), respectively. The odds ratio for the association of prior bilateral oophorectomy with otosclerosis was 1.12 (0.58–2.18; P = .73). In cases with otosclerosis, PTA at diagnosis was not significantly higher for women with ≥1 delivery as compared with those without (median 45 dB hearing loss [HL] [interquartile range {IQR} 36–55] vs 43 [IQR 34–53]; P = 0.18) but was significantly higher for women with bilateral oophorectomy compared with those without (median 54 dB HL [IQR 44–61] vs 44 [IQR 34–53]; P = .03).

Conclusion.

These data do not support a relationship between endogenous estrogen exposure and development of otosclerosis. Women with otosclerosis who had a history of pregnancy did not have significantly worse hearing at the time of diagnosis, suggesting that pregnancy is not associated with disease severity.

Perhaps one of the most characteristic epidemiological features of otosclerosis is that it disproportionally affects women as compared with men, in an almost 2:1 ratio.^1–3 Since the mid-20th century, pregnancy has been posited as a contributory mechanism behind the observed discrepancies in disease incidence between women and men.⁴ Pathophysiologically, it is thought that systemically elevated circulating levels of estrogen experienced through-out pregnancy may facilitate progression to clinically significant otosclerosis.^1,2 The theoretical basis for this hypothesis stems from the large body of evidence linking estrogen to the inhibition of bone resorption and promotion of osteoblastic function.^5–7

Within the literature, there is continued controversy about whether pregnancy (or any estrogen exposure) accounts for the difference in incidence rates between women and men.^1,2 Although prior studies have found that women recall experiencing a subjective decrease in hearing surrounding pregnancy,^3,8 other institutional series have disagreed with this finding based on a woman’s history of childbirth.^9–11 Furthermore, studies evaluating the influence of oral contraceptives on the development of otosclerosis found no significant association.¹² Amidst these conflicting reports on estrogen’s role in the development of otosclerosis, large population-based studies grounded in objective data remain lacking.

For this reason, the current study was envisioned with the primary objective of determining whether pregnancy portends a higher risk of developing otosclerosis through performing a large, population-based case-control study. To further investigate the role of circulating estrogen on the development of otosclerosis, we also investigated whether women who had undergone bilateral oophorectomy—and therefore removal of the primary estrogen-synthesizing organ—were at decreased risk of developing otosclerosis compared with matched controls. As a secondary analysis, we sought to determine whether prior pregnancy or bilateral oophorectomy was associated with more or less severe disease based on pure-tone average (PTA), respectively.

Methods

Patient Selection and Matching

Institutional Review Board (IRB) approval was sought and obtained (Mayo Clinic IRB protocol No. 18–005225 and Olmsted Medical Center IRB protocol No. 021-OMC-18) prior to study commencement. Incident cases diagnosed with otosclerosis were identified through the Rochester Epidemiology Project (REP) medical records linkage system over the time period from January 1, 1970, through December 31, 2017.^13–15 The medical records for all persons were evaluated to confirm the diagnosis of otosclerosis surgically, the primary method of diagnosis in this group, or through clinical, audiologic, and imaging assessments. Clinical and audiologic assessments included the presence of a Carhart notch, absent acoustic reflexes, and a type A or A_s tympanogram.¹⁶ Imaging criteria included radiolucency in the otic capsule at the fissula ante fenestram or surrounding the otic capsule.¹⁷ Each woman with otosclerosis was matched to 3 women without otosclerosis who were also residents of Olmsted County at the index date, were of similar age (±1 year), and who had similar historical depth of medical record (ie, duration between the earliest diagnosis date in the medical record and the index date). We included 3 matched controls for each case to increase the statistical power for our analyses.

Pregnancy

Over the study period from 1970 to 2017, the REP includes various historical coding systems for diagnoses and procedures.^13,14 Thus, Berkson, Hospital Adaptation of the International Classification of Diseases (HICDA), International Classification of Diseases, 9th Revision (ICD-9), and International Classification of Diseases, 10th Revision (ICD-10) diagnosis codes; Berkson, ICD-9, and ICD-10 procedure codes; CPT service codes; and supplemental birth data from the REP were used to identify distinct delivery dates for all cases and controls who had a recorded birth. If a discrepancy was present among any of these sources, the paper or electronic charts were manually reviewed to verify the correct date of delivery. To ensure consistent risk factor assessment for cases and controls, only completed pregnancies that resulted in birth were considered in the analysis.

Bilateral Oophorectomy

In a similar fashion, Berkson, HICDA, ICD-9, and ICD-10 diagnosis codes; Berkson, ICD-9, and ICD-10 procedure codes; and CPT service codes were used to identify cases and controls who underwent oophorectomy or related gynecologic procedures that may include oophorectomy. We manually reviewed each woman’s surgical report to ensure completion of bilateral oophorectomy and to determine the exact date of surgery.

Statistical Analysis

Continuous features were summarized with medians and interquartile ranges (IQRs), and categorical features were summarized with frequency counts and percentages. Associations of the number of deliveries and bilateral oophorectomy prior to the index date with development of otosclerosis were evaluated using conditional logistic regression and summarized with odds ratios and 95% confidence intervals (CIs). Among cases, associations of the number of deliveries and bilateral oophorectomy with PTA at the time of otosclerosis diagnosis were evaluated using Wilcoxon rank-sum tests. Statistical analyses were performed using version 9.4 of the SAS software package (SAS Institute, Cary, North Carolina), and P values <.05 were considered statistically significant.

Results

Pregnancy

A total of 299 women with a confirmed diagnosis of otosclerosis (cases) were matched to 897 women from the same population without otosclerosis (controls; a ratio of 1 case to 3 controls). The median age at diagnosis for incident cases was 42 years (IQR 32–53, range 18–87), and the median age at the index date for matched controls was 41 years (IQR 32–53, range 18–88). The median depth of the medical record for cases was 15 years (IQR 4–29) and for controls was 16 years (IQR 6–29). Among all 1196 women studied, there were 565 (47%) with at least 1 delivery prior to the index date, including 147 (49%) cases and 418 (47%) controls (Table 1). The odds ratio for the association of at least 1 delivery with development of otosclerosis was 1.16 (95% CI 0.85–1.60; P =.35). The median age at first delivery for the cases was 25 years (IQR 22–29), which occurred at a median of 15 years (IQR 6–28) before the otosclerosis diagnosis date. The median age at the first delivery for the controls was 25 years (IQR 22–29), which occurred at a median of 17 years (IQR 8–28) before the index date. Among all women studied, there were 199 (17%) women with a single delivery prior to the index date, 173 (14%) with 2 deliveries, 108 (9%) with 3 deliveries, and 85 (7%) with 4 or more deliveries. The number of deliveries prior to the index date did not differ significantly between the cases and controls (Table 1). Of the 299 women with otosclerosis, 291 had an air-conduction PTA recorded at diagnosis. Although PTA at diagnosis was higher for women with at least 1 prior delivery (median 45 dB hearing loss [HL]; IQR 36–55) compared with women without any prior deliveries (median 43 dB HL; IQR 34–53), this difference was not statistically significant (P = .18) nor clinically significant.

Table 1.

Association of Number of Prior Deliveries With Development of Otosclerosis.

	n (%)
Number of deliveries	Controls (n = 897)	Cases (n = 299)	Odds ratio (95% CI)	P value
0	479 (53)	152 (51)	1.00 (reference)
≥1	418 (47)	147 (49)	1.16 (0.85–1.60)	.35
0	479 (53)	152 (51)	1.00 (reference)
1	145 (16)	54 (18)	1.22 (0.83–1.80)	.32
2	130 (14)	43 (14)	1.09 (0.71–1.68)	.70
3	78 (9)	30 (10)	1.28 (0.77–2.12)	.34
≥4	65 (7)	20 (7)	1.00 (0.54–1.84)	.99

Bilateral Oophorectomy

Among all 1196 women studied, there were 52 (4.4%) with a history of bilateral oophorectomy before the index date, including 14 (4.7%) cases and 38 (4.2%) controls. The odds ratio for the association of a history of bilateral oophorectomy with development of otosclerosis was 1.12 (95% CI 0.58–2.18; P = .73). The median age at bilateral oophorectomy for the cases was 50 years (IQR 45–57), which occurred at a median of 7 years (IQR 3–11) before the otosclerosis diagnosis date. The median age at bilateral oophorectomy for the controls was 46 years (IQR 39–51), which occurred at a median of 6 years (IQR 2–14) before the index date. Of the women with otosclerosis, PTA at diagnosis was significantly higher for those with a history of bilateral oophorectomy compared with those without (median 54 dB HL [IQR 44–61] vs 44 [IQR 34–53]; P = .03).

Discussion

The influence of pregnancy and estrogen exposure on the development of otosclerosis constitutes one of the longest-standing controversies in the study of otosclerosis epidemiology. Over the years, numerous studies have reported conflicting results, but large population-based objective data (ie, not survey based or self-reported) are lacking. The current study involved almost 1200 women—nearly 300 of whom were diagnosed with otosclerosis—and found no significant association between childbirth and the development of otosclerosis. Furthermore, we found no dose-response relationship with multiple deliveries. Lastly, women with at least 1 prior delivery did not have clinically or statistically significantly worse PTA at the time of otosclerosis diagnosis.

These findings corroborate several prior institutional series that demonstrated no significant association with pregnancy and the development or worsening of otosclerosis.^9–11 Of note, although previous work has implicated pregnancy in the development of otosclerosis, many of these studies employed surveys to ascertain pregnancy and symptom information.^3,8 Unfortunately, it has been well established that survey-based studies bear significant limitations surrounding susceptibility to recall bias, wherein cases are more prone to report historical risk factors related to the condition of interest compared with controls. For example, this bias is commonly observed in the vestibular schwannoma literature, in which persons will report significant exposures related to hearing (eg, occupational noise exposure) after learning that they have developed a tumor on their hearing nerve; however, prior objective, population-based studies that do not rely on surveys demonstrate no such association.^18–20 Considering these methodological limitations of prior research, we believe the data from the current study do not support the assertion that pregnancy increases the risk of developing otosclerosis nor does it contribute to worsening hearing loss.

To further investigate the role of estrogen exposure on the development of otosclerosis, the current work also studied bilateral oophorectomy as a risk factor for otosclerosis. If systemically elevated circulating estrogen influences the development of otosclerosis, then women who have undergone bilateral oophorectomy—thereby removing the chief source of endogenous estrogen—should theoretically exhibit decreased likelihood of developing otosclerosis. However, this hypothesis was not supported by our data. Furthermore, women who had undergone bilateral oophorectomy demonstrated worse PTA at the time of otosclerosis diagnosis compared with women without such a history. These observations further support the conclusion that the hormonal influence of estrogen is unlikely to account for the prior reported association in otosclerosis epidemiologic literature.

Although the historical precedent for differing incidence rates of otosclerosis between women and men is incontrovertible, it is worth noting that recent trends in disease incidence suggest that this may no longer be the case. For instance, in a recent incidence study of otosclerosis in the same population as the current study, the incidence rates of otosclerosis per 100,000 person-years in the first 3 years of the study (1970–1972) were 24.6 in women and 13.6 in men.¹⁵ By contrast, in the most recent 3 years (2015–2017), the incidence rates of otosclerosis were 3.2 in women and 3.1 in men.¹⁵ These trends suggest that the differences in incidence rates between women and men have all but disappeared over the past 45 years. However, the mechanism responsible for the dramatic decline in the incidence of otosclerosis remains unknown. Although speculative, potential etiologies could include incorrect diagnoses in earlier years, greater health care utilization among women compared with men, and the more recent identification of conditions that may mimic otosclerosis such as superior semicircular canal dehiscence. Of note, we explored whether the associations of pregnancy and estrogen exposure with otosclerosis differed across calendar year (2 strata: 1970–1989 vs 1990–2017), but these analyses did not suggest differences (data not shown).

There are several limitations of the current work that warrant mention. First, only pregnancies that resulted in live births were considered. There is potential that exclusion of abortions and miscarriages could have affected these data by incompletely representing a portion of the study participants who were exposed to elevated systemic estrogen during abbreviated pregnancy duration. In addition, there is a possibility that not all births were captured in the REP, given that only births within Olmsted County or at participating REP providers are indexed. Births in women who moved into the Olmsted County region after childbearing age would also not be captured. However, births outside of the REP system and geographic movement into the region should be similar across cases and controls (nondifferential misclassification). Similar limitations apply to the inclusion of women undergoing bilateral oophorectomy. Lastly, there is the potential for hormonal therapy after oophorectomy that could further affect the results as reported, but this is likely limited given the largely postmenopausal group undergoing this procedure.

Conclusion

As a correlate for pregnancy, a history of at least 1 delivery and the number of deliveries did not incur significantly higher risk of development of otosclerosis. Moreover, bilateral oophorectomy was not protective against disease development. Taken together, these data do not support an association between these specific factors and otosclerosis development or disease severity.