Figures
Abstract
Background
Although rates have declined, hysterectomy is still a frequent gynaecological procedure. To date, there has been no systematic quantification of the relationships between early/mid-life exposures and hysterectomy. We performed a systematic review and meta-analyses to quantify the associations between age at menarche, education level, parity and hysterectomy.
Methods
Eligible studies were identified by searches in PubMed and Embase through March 2015. Study-specific estimates were summarised using random effects meta-analysis. Heterogeneity was explored using sub-group analysis and meta-regression.
Results
Thirty-two study populations were identified for inclusion in at least one meta-analysis. Each year older at menarche was associated with lower risk of hysterectomy—summary hazard ratio 0.86 (95% confidence interval: 0.78, 0.95; I2 = 0%); summary odds ratio 0.88 (95% confidence interval: 0.82, 0.94; I2 = 61%). Low education levels conferred a higher risk of hysterectomy in the lowest versus highest level meta-analysis (summary hazard ratio 1.87 (95% confidence interval: 1.25, 2.80; I2 = 86%), summary odds ratio 1.51 (95% confidence interval: 1.35, 1.69; I2 = 90%)) and dose-response meta-analysis (summary odds ratio 1.17 (95% confidence interval: 1.12, 1.23; I2 = 85%) per each level lower of education). Sub-group analysis showed that the birth cohort category of study participants, the reference category used for level of education, the year the included article was published, quality of the study (as assessed by the authors) and control for the key variables accounted for the high heterogeneity between studies in the education level meta-analyses. In the meta-analyses of studies of parity and hysterectomy the results were not statistically significant.
Citation: Wilson LF, Mishra GD (2016) Age at Menarche, Level of Education, Parity and the Risk of Hysterectomy: A Systematic Review and Meta-Analyses of Population-Based Observational Studies. PLoS ONE 11(3): e0151398. https://doi.org/10.1371/journal.pone.0151398
Editor: Sharon Cameron, NHS lothian and University of Edinburgh, UNITED KINGDOM
Received: July 3, 2015; Accepted: February 26, 2016; Published: March 10, 2016
Copyright: © 2016 Wilson, Mishra. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The Australian Longitudinal Study on Women’s Health was conceived and developed at the Universities of Newcastle and Queensland and is funded by the Australian Government Department of Health. G. D. M. was supported by an Australian Research Council Future Fellowship (FT120100812). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Hysterectomy continues to be one of the most frequent gynaecological procedures performed in more economically developed countries [1, 2] and has been associated with lower quality of life [3, 4] and poorer health outcomes [5–7]. Approximately 20–45% of women in these countries will have a hysterectomy by the time they are between 60–70 years of age [8–10], with the majority performed for benign indications such as uterine fibroids, dysfunctional uterine bleeding, prolapse and endometriosis [2].
With such high prevalence, it is important to understand the exposures that occur in earlier life that might predict the risk of hysterectomy to alert practitioners of the need to proactively monitor these women for the benign indications of hysterectomy and implement appropriate treatment pathways. Lower socio-economic status (SES), particularly lower levels of education, have been associated with a higher risk of hysterectomy, however, this varies by geographic location [11, 12], ethnicity [13–15] and birth cohort [16]. The associations between a range of reproductive factors and hysterectomy have also been studied. In general, an earlier age at menarche [17–19] and age at first birth [18, 20] have been associated with an increased risk of hysterectomy; while the associations with other adult reproductive factors such as parity [18, 21] and number of miscarriages [19, 22] are less clear.
To date however, there has been no systematic quantification of the relationships between these early/mid-life exposures and hysterectomy. The aim of our study therefore was to systematically review and quantify the evidence from observational studies, through meta-analysis, on the relationship between SES and reproductive factors and hysterectomy.
Materials and Methods
The systematic review and meta-analyses were done in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) Guidelines [23, 24]. An official protocol, however, was not published or registered, although the strategy for this review was discussed and agreed. The clinical question posed was: what is the relationship between SES and reproductive factors and hysterectomy. Meta-analysis was possible for three exposures: age at menarche, level of education and parity.
Search Strategy
We conducted three separate systematic searches in Embase and PubMed: 1) age at menarche, 2) SES, and 3) adult reproductive factors and hysterectomy. The databases were searched for records from inception until March 2015. Full details of the three search strategies are available online in S1 File: Search Strategy. For all searches we used both Medical Subject Headings (MeSH terms) and text words in PubMed, and Emtree terms and text words in Embase. No language or publication data restrictions were imposed. Reference lists of articles were checked for additional articles.
Study selection
One investigator (L.W.) did the initial screen of titles and the abstracts of articles identified through the initial screening process. Two investigators (L.W. and G.M.) independently reviewed the full text of potentially relevant articles for final inclusion, with any disagreement resolved through discussions.
Articles were eligible for inclusion if they 1) reported measures of association and 95% confidence intervals (CI) (or the raw data to calculate these) for associations between any SES factor, age at menarche or adult reproductive factor and hysterectomy, and 2) had population-based samples or cohorts. Studies that focused on population sub-groups (e.g. veterans), or used hospital-based controls, were excluded.
We also contacted several authors [17, 20, 25] for additional information or results. All responded; with two [20, 25] able to provide some or all of the requested information.
Quality evaluation
Study quality was assessed independently by L.W. and G.M. using the Newcastle-Ottawa Scale for cohort studies. The scale includes 9 items under three categories: selection, comparability and outcome. Three of the items were not relevant to assessment of cross-sectional studies (outcome not present at start of study, sufficient follow-up for outcome to occur, and adequacy of cohort follow-up). Unique scales were developed for each review as elements of the selection and comparability sections are exposure specific (available online in S2 File: Quality Assessment Scales).
Data extraction
Data was extracted by L.W. and independently checked by G.M. for accuracy and completeness. Any disagreements were resolved through discussion. For each study we extracted: first author, year of publication, country of study, study design, study description, survey/baseline year, length of follow-up (cohort studies only), age group of sample (from which we derived the birth cohort of the sample), prevalence of hysterectomy in sample, outcome/comparator groups, measures of association (for the relevant exposure variables) and the covariates used in the model.
Statistical analysis
For all meta-analyses, measures of association were summarised using a random-effects model due to the variability in characteristics between studies. The majority of studies reported either hazard ratios (HR) or odds ratios (OR). We summarised these two types of measures in separate meta-analyses. When articles reported measures of association with different covariates, the most fully-adjusted model was included in the meta-analysis.
Lowest versus Highest Education Level.
We conducted meta-analyses on all eligible studies, and also those that only reported adjusted results. If the reference category in the published study was the lowest education level, the inverse of the HR/OR, and the upper and lower confidence intervals, were calculated for inclusion.
Dose-response meta-analyses.
Studies were eligible for inclusion if they reported a dose-response association or categorical variable with at least three exposure levels. Dose-response measures of association were calculated for each study assuming a log-linear relationship between the exposure and hysterectomy. For the education meta-analysis, the lowest education level category had to be equivalent to “high school or less” and the highest education level category equivalent to “university/college degree or higher”. Highest education level was used as the reference category. Where the lowest education level was used as the reference category, the inverse of the calculated dose-response effect estimates and 95% confidence intervals were used. The HR/OR for a one category change in education level was assumed to approximate the HR/OR for one level lower in education.
For age at menarche, the HR/OR for a one category increase was assumed to approximate a per year increase in age at menarche; while for parity, this approximated the HR/OR for a per child increase (with no children as the reference category).
Exploration of heterogeneity.
Between-study heterogeneity was assessed using the chi2 (Cochrane Q) and I2 statistics. We considered I2 values of 25%, 50% and 75% as cut-off points for evidence of low, moderate and high heterogeneity respectively [26]. Meta-regression and sub-group analyses were undertaken for the level of education meta-analyses if at least ten studies were included [27]. For the age at menarche and parity meta-analyses, where there were too few studies for meta-regression, sensitivity analyses were conducted omitting each study in turn to assess whether any single study contributed significantly to heterogeneity.
A priori study characteristics considered in our analysis of heterogeneity in the education meta-analyses included: year of publication (before 2000; 2000 and later), study design (prospective or cross-sectional), whether the definition of hysterectomy included, excluded or did not specify women with malignant conditions, the region in which the study population was selected (United States, Europe/United Kingdom, Australia/New Zealand, Asia), birth cohort of the study population (pre-1945, baby-boomers (1945–1965), broad population cohort), and cut-off for highest education level (high school, college/university degree, other). Post hoc, on the basis of additional expert input, prevalence of hysterectomy, adjustment for key confounding variables (identified through the quality assessment—see S2 File: Quality Assessment Scales) and quality of study (see online S5 File: Quality Assessment) were also considered.
We first undertook univariable random-effects meta-regression to investigate the role of each of the study characteristics described above on between-study heterogeneity. Where variables had more than one category overall effects were assessed (with 1000 permutations). Characteristics with a p-value of ≤0.2 were included in the multivariable model [28, 29].
Publication bias was assessed when there were at least ten studies included in the meta-analysis[27]. This was done through visual inspection of a funnel plot and through formal assessment undertaken by performing both Egger’s test and Begg and Mazumdar’s Rank Correlation test.
All analyses were conducted in Stata version 12.1.
Results
Literature search
The numbers of identified and included articles and study populations are summarised in Fig 1. Meta-analysis was possible for three exposures: age at menarche, level of education and parity. For other SES variables (occupation, employment and income) and adult reproductive factors (age at first birth, number of pregnancies and number of miscarriages) there were either an insufficient number of studies and/or incomparable categories or populations for a meta-analysis to be performed. Across the three searches, 4,102, 7,821 and 17,819 records were identified for age at menarche, SES and reproductive factors respectively. After excluding duplicates and undertaking an initial screen of titles, 104 article abstracts were reviewed for age at menarche, 87 article abstracts for SES factors and 133 for reproductive factors. Across the three searches there was overlap in the full-text articles identified for complete assessment. No non-English language papers were eligible for inclusion. In total, the full-texts of 73 articles were obtained. Of these, 29 articles [11–22, 25, 30–45] with results from 32 study populations contributed data to one or more meta-analysis. The age at menarche meta-analysis included seven study populations [17–19, 30–32]; the lowest versus highest education meta-analysis included 31 study populations [11–16, 20–22, 25, 30–45]; the dose-response education level meta-analysis included ten study populations [14, 16, 32, 39–42, 45], and the parity meta-analysis included eight study populations [11, 12, 14, 18, 19, 21, 44, 45].
Studies were published up until March 2015.
Study characteristics
The characteristics of the 32 study populations that were included in at least one of the meta-analyses are summarised in Table 1.
Abbreviations: LH = lowest versus highest; DR = dose-response.
Of the identified study populations, five were cohort and 27 were cross-sectional. Study populations were drawn from nine countries—all were more economically developed locations. The United States was the most common study population setting (n = 14). Study population samples ranged from 276 to 180,982 participants, and hysterectomy prevalence from 3% to 56% (average 20%).
For each exposure, data on associations were reported for up to 12 study populations but were not included in our meta-analyses because they used a different reference category, exposure variables were categorized in ways that prevented comparison, or there were an insufficient number of categories to enable a dose-response association to be calculated (details of these study populations are available online in S3 File: Tables).
Age at menarche
Seven study populations [17–19, 30–32] were eligible for inclusion in the dose-response meta-analysis. Three study populations [17, 30, 32] reported a dose-response estimate (per year older at menarche) and four [18, 19, 31] reported categorical variables (with the number of categories either 3 or 4). The summary result for the two studies reporting hazard ratios [18, 19] was 0.86 (95% CI: 0.78, 0.95; I2 = 0.0%) per year older at menarche (Fig 2). Random effects meta-analysis of the five cross-sectional studies reporting odds ratios [17, 19, 30–32] gave a summary estimate of 0.88 (95% CI: 0.82, 0.94; I2 = 61%) per year older at menarche (Fig 2). Omitting the Danish Study[19] reduced heterogeneity to 27%, while omitting the results from the Black Women’s Health Study[31] removed all heterogeneity; however in both cases there was minimal difference to the summary result.
Squares represent study-specific estimates (the size of the square reflects the study-specific statistical weight); horizontal lines represent 95% confidence intervals (CI); diamonds represent the summary estimate with corresponding 95% confidence interval. See Table 1 for details of Study abbreviations.
Publication bias was not performed as there were insufficient included studies to properly assess a funnel plot or the results of formal assessments.
Level of Education
Lowest versus Highest education level meta-analysis.
Ten studies reported hazard ratios [11, 21, 22, 33–38]. Random-effects meta-analysis gave a summary HR of 1.87 (95% CI: 1.25, 2.80; I2 = 86%) (Fig 3). Results of sub-group analysis and univariable meta-regression are described in Table 2. Together, those characteristics with a p-value of ≤ 0.2 in univariable meta-regression i.e. year of publication, birth cohort category of study participants and the reference category used for education level explained all of the between-study heterogeneity in a multivariable meta-regression model. Begg and Mazumdar’s correlation was 1.07 (continuity corrected), P-value = 0.28. Egger’s regression intercept was 0.151 (P-value = 0.33). Thus, there was no evidence of publication bias (available online in S4 File: Funnel Plots).
Squares represent study-specific estimates (the size of the square reflects the study-specific statistical weight); horizontal lines represent 95% confidence intervals (CI); diamonds represent the summary estimate with corresponding 95% confidence interval. See Table 1 for details of Study abbreviations.
Abbreviations: CI, Confidence Interval.
Twenty-one studies reported odds ratios [12–16, 20, 25, 30–32, 38–45]. Random-effects meta-analysis gave a summary OR of 1.51 (95% CI: 1.35, 1.69; I2 = 90%) (Fig 3). When studies that only had unadjusted results were excluded from the meta-analysis [12, 14, 20, 43] the summary estimate was similar with little change in heterogeneity (OR = 1.56; 95% CI: 1.39, 1.75; I2 = 87%). Results of sub-group analysis and univariable meta-regression on the full set of studies are described in Table 2. Birth cohort category of study participants, adjustment for key variables and quality of study were significant in the univariable regression analysis. However, when these characteristics were included in multi-variable meta-regression only 28% of between-study variance was explained. Begg and Mazumdar’s correlation was 0.15 (continuity corrected), P-value = 0.88. Egger’s regression intercept was -0.85 (P-value = 0.44). Thus, there was no evidence of publication bias (available online in S4 File: Funnel Plots).
Education Level Dose-response meta-analysis.
Ten studies [14, 16, 32, 39–42, 45] were included in the dose-response meta-analysis. All of the studies reported categorical variables (the average number of categories was 4; ranging from 3 to 6). All of these studies reported odds ratios. Random-effects meta-analysis gave a summary OR of 1.17 (95% CI: 1.12, 1.23; I2 = 85%) per each level lower of education (Fig 4). Results of sub-group analysis are described in Table 3. Together, the birth cohort of study participants and adjustment for at least one reproductive factor explained the heterogeneity between studies. Begg and Mazumdar’s correlation was 0.45 (continuity corrected), P-value = 0.65. Egger’s regression intercept was -0.36 (P-value = 0.77). Thus, there was no evidence of publication bias (available online in S4 File: Funnel Plots).
Squares represent study-specific estimates (the size of the square reflects the study-specific statistical weight); horizontal lines represent 95% confidence intervals (CI); diamonds represent the summary estimate with corresponding 95% confidence interval. See Table 1 for details of Study abbreviations.
Abbreviations: CI, Confidence Interval.
Parity
Random effects meta-analysis of the four studies [11, 18, 19, 21] reporting hazard ratios gave a non-significant summary estimate of 1.10 (95% CI: 0.86, 1.41; I2 = 62%) per child (Fig 5). Three of the studies, from Japan [11], the USA [21] and Denmark [19], had study populations where the majority of participants were aged less than 45 years (mean age 43, 40 and 35 respectively) when responding to questions about parity. When only these three studies [11, 19, 21] were included in the meta-analysis, heterogeneity was reduced (I2 = 21%; HR 0.98, 95% CI 0.76, 1.27). When only the Japanese study [11] was omitted (the only study reporting a HR <1), heterogeneity (I2) was 14%, and the association between parity and hysterectomy increased and was statistically significant (HR = 1.26 95% CI 1.09, 1.45 per child). Publication bias was not performed as there were insufficient included studies to properly assess a funnel plot or the results of formal assessments.
Squares represent study-specific estimates (the size of the square reflects the study-specific statistical weight); horizontal lines represent 95% confidence intervals (CI); diamonds represent the summary estimate with corresponding 95% confidence interval. See Table 1 for details of Study abbreviations.
Random effects meta-analysis of the four studies [12, 14, 44, 45] reporting odds ratios gave a non-significant summary estimate of 1.04 (95% CI: 0.94, 1.14) per child with high heterogeneity (I2 = 96%) (Fig 5). In the sensitivity analysis, no single study reduced heterogeneity below 80%. Publication bias was not performed as there were insufficient included studies to properly assess a funnel plot or the results of formal assessments.
Quality analysis
The results of the quality assessments are available online in S5 File: Quality Assessment. Overall, the majority of studies were assessed as being of moderate quality. Main sources of potential bias included: 1) representativeness of the sample in the parity meta-analysis (where a number of studies had young study participants); 2) ascertainment of age at menarche, with only one study[18] obtaining this information during adolescence; 3) comparability of exposed and non-exposed groups, with the majority of studies not adjusting for key confounding variables (parental/early childhood socio-economic factors for age at menarche and reproductive factors for level of education); and 4) assessment of outcome, with the majority of studies ascertaining hysterectomy outcome by self-report and not hospital records or medical imaging.
Discussion
Principal Findings
Meta-analyses of the above-mentioned population-based studies revealed a 14% and 12% lower risk of hysterectomy for each year older at menarche in studies reporting hazard ratios and odds ratios respectively. Women with the lowest levels of education had a higher risk of hysterectomy when compared with women with the highest education levels; and there was a dose-response relationship with a 17% higher risk of hysterectomy with each level lower of education. In meta-analyses of studies of parity and hysterectomy, the results (hazard ratios and odds ratios) were not statistically significant.
Early menarche has been associated with an increased risk of uterine fibroids [46–48] and endometriosis [49–51], two common indications for hysterectomy. Although the biologic mechanisms are not clearly understood, it has been postulated that this increased risk may be due to increased hormonal stimulation through increased exposure to menstrual cycles [46, 48, 50], or there may be pre-natal or early life factors that are associated with both early menarche and adult disease [46, 48]. Early menarche may also be a marker for early sexual activity and influence marriage and reproductive patterns, especially age at first birth [52], although the majority of studies adjusted for reproductive variables indicating that an association independent of these factors is present.
Lower levels of education may impact upon general health and health literacy levels, the capacity to navigate the health system and decision-making skills [53] or impact upon the quality of care received through resulting locational or socio-economic disadvantage [54]. Women with lower education levels may work in lower income occupations that may negatively impact on health; or be employed on casual or inflexible terms necessitating quick and complete resolution of medical issues. Additionally, women with lower levels of education may be at increased risk of being overweight and obese (a risk factor for hysterectomy [21, 44, 45, 55] and adjusted for by less than half of the studies included in the education meta-analyses) and having poorer health outcomes overall [56].
Exploration of heterogeneity
Moderate to high heterogeneity was prevalent in these meta-analyses, in particular, the education level meta-analyses all had high levels of heterogeneity, indicating that the summary results should be interpreted with caution. Sub-group analysis showed that several factors seem to account for this heterogeneity, including the birth cohort category of study participants, the reference category used for level of education, the year the included article was published, quality of the study (as assessed by the authors) and control for the key variables of age and reproductive factors. While most of these explanatory factors relate to the design of the study and measurement of variables, the birth cohort category of study participants as an explanatory factor indicates that education attainment levels may vary over time and may also be a reflection of changing reproductive trends, increased access to alternative treatments for benign hysterectomy indications and changing treatment preferences among different SES groups [16].
Heterogeneity was less of an issue in the age at menarche meta-analyses. There was no heterogeneity in the age at menarche meta-analysis of study populations reporting hazard ratios (although only two studies were able to be included); and in the meta-analysis of study populations reporting odds ratios, the summary result was robust even after omitting studies that showed evidence of contributing to heterogeneity. In the parity meta-analysis of study populations reporting hazard ratios (with moderate heterogeneity) omission of either the UK[18] or Japanese[11] study population reduced heterogeneity to 21% and 14% respectively. In the meta-analysis of study populations reporting odds ratios no single study reduced heterogeneity below 80%; however, in both cases only a small number of studies were able to be included.
Strengths and limitations
To our knowledge, these are the first meta-analyses to quantify the associations between age at menarche, level of education, parity and risk of hysterectomy. These analyses included only population-based studies to increase generalisability.
Our meta-analyses have several limitations. First, all of the included studies were observational so the results cannot be considered causal evidence. Second, the majority of included studies were cross-sectional with the attendant limitations of this design. Recall bias may have impacted in particular on the accuracy of the measurement of age at menarche. In all but one study [18] this was measured by self-report when women were in adulthood (rather than in adolescence) and less likely to reliably recall menarcheal age [57]. This may have resulted in exposure misclassification although this is unlikely to be differential across hysterectomy and non-hysterectomy groups. Third, in the parity meta-analysis, three of the four studies reporting hazard ratios had populations where the majority of participants were aged less than 45 years at the time of survey, potentially biasing an assessment of the association with hysterectomy (towards the null), as many women delay hysterectomy until they consider their child-bearing years complete.
The process of review and data extraction also demonstrated the differences in outcome and exposure definition across the studies. For example differences in definition and categories in the exposure variables prevented meta-analysis of a number of the SES and adult reproductive factors and contributed to the heterogeneity in the lowest versus highest education meta-analysis. There was very little consistency in adjusting variables across the studies, which also contributed to variance. We anticipated that region of study could have contributed to heterogeneity. This was not the case, however the majority of studies were located in more economically developed countries. Ethnicity may have been a more appropriate measure to assess cultural and contextual differences, however, there was insufficient information across studies to explore this.
Conclusions and implications
In conclusion, our systematic review and meta-analyses indicate that there is a dose-response relationship between younger age at menarche and lower education levels and a higher risk of hysterectomy. As younger age at menarche, lower education levels and hysterectomy are common risk factors for adverse health outcomes such as diabetes and cardiovascular disease [5, 6, 43, 58–62] a better understanding of the causal pathways and the potential mediating role that hysterectomy might play is of public health importance. Future research needs to explore the relationship between SES and reproductive factors utilising data from large prospective cohort studies that measure SES and reproductive variables prior to hysterectomy and can adjust for key confounding variables. Studies that pool individual-level data from populations of different ethnicities and locations, align outcome and exposure definitions, and can adjust for the same set of confounders are also required.
Supporting Information
S1 File. Search Strategy.
This file provides full details of the three search strategies undertaken in Pubmed and Embase.
https://doi.org/10.1371/journal.pone.0151398.s001
(PDF)
S2 File. Quality Assessment Scales.
This file includes the Newcastle-Ottawa Quality Assessment Scales used for each meta-analysis.
https://doi.org/10.1371/journal.pone.0151398.s002
(PDF)
S3 File. Tables.
This file includes details of the studies that reported data on relevant associations but were not included in the meta-analyses.
https://doi.org/10.1371/journal.pone.0151398.s003
(PDF)
S4 File. Funnel Plots.
This file includes the funnel plots assessing publication bias.
https://doi.org/10.1371/journal.pone.0151398.s004
(PDF)
S5 File. Quality Assessment.
This file includes the results of the quality assessments undertaken for each meta-analysis
https://doi.org/10.1371/journal.pone.0151398.s005
(PDF)
Author Contributions
Conceived and designed the experiments: LFW GDM. Performed the experiments: LFW GDM. Analyzed the data: LFW. Contributed reagents/materials/analysis tools: LFW GDM. Wrote the paper: LFW GDM.
References
- 1. Garry R. The future of hysterectomy. BJOG. 2005;112(2):133–9. pmid:15663575.
- 2. Merrill RM. Hysterectomy surveillance in the United States, 1997 through 2005. Med Sci Monit. 2008;14(1):CR24–31. pmid:18160941.
- 3. Kuh DL, Wadsworth M, Hardy R. Women's health in midlife: the influence of the menopause, social factors and health in earlier life. Br J Obstet Gynaecol. 1997;104(8):923–33. pmid:9255084.
- 4. Cooper R, Mishra G, Hardy R, Kuh D. Hysterectomy and subsequent psychological health: findings from a British birth cohort study. J Affect Disord. 2009;115(1–2):122–30. pmid:18835497.
- 5. Michelsen TM, Dorum A, Cvancarova M, Liavaag AH, Dahl AA. Association between hysterectomy with ovarian preservation and cardiovascular disease in a Norwegian population-based sample. Gynecol Obstet Invest. 2013;75(1):61–7. pmid:23220872.
- 6. Yeh JS, Cheng HM, Hsu PF, Sung SH, Liu WL, Fang HL, et al. Hysterectomy in young women associates with higher risk of stroke: a nationwide cohort study. Int J Cardiol. 2013;168(3):2616–21. pmid:23587399.
- 7. Parker WH, Broder MS, Chang E, Feskanich D, Farquhar C, Liu Z, et al. Ovarian conservation at the time of hysterectomy and long-term health outcomes in the nurses' health study. Obstet Gynecol. 2009;113(5):1027–37. pmid:19384117; PubMed Central PMCID: PMC3791619.
- 8.
Australian Bureau of Statistics. National Health Survey: Summary of Results. 4364.0. 2004–05. Canberra: Asutralian Bureau of Statistics, 2006.
- 9. Rositch AF, Nowak RG, Gravitt PE. Increased age and race-specific incidence of cervical cancer after correction for hysterectomy prevalence in the United States from 2000 to 2009. Cancer. 2014;120(13):2032–8. pmid:24821088; PubMed Central PMCID: PMC4073302.
- 10. Redburn JC, Murphy MF. Hysterectomy prevalence and adjusted cervical and uterine cancer rates in England and Wales. BJOG. 2001;108(4):388–95. pmid:11305546.
- 11. Nagata C, Takatsuka N, Kawakami N, Shimizu H. Soy product intake and premenopausal hysterectomy in a follow-up study of Japanese women. Eur J Clin Nutr. 2001;55(9):773–7. Epub 2001/08/31. pmid:11528492.
- 12. Koepsell TD, Weiss NS, Thompson DJ, Martin DP. Prevalence of prior hysterectomy in the Seattle-Tacoma area. Am J Public Health. 1980;70(1):40–7. Epub 1980/01/01. pmid:6965339; PubMed Central PMCID: PMCPMC1619330.
- 13. Hautaniemi SI, Leidy Sievert L. Risk factors for hysterectomy among Mexican-American women in the US southwest. Am J Hum Biol. 2003;15(1):38–47. Epub 2003/01/29. pmid:12552577.
- 14. Powell LH, Meyer P, Weiss G, Matthews KA, Santoro N, Randolph JF Jr., et al. Ethnic differences in past hysterectomy for benign conditions. Womens Health Issues. 2005;15(4):179–86. Epub 2005/07/30. pmid:16051109.
- 15. Meilahn EN, Matthews KA, Egeland G, Kelsey SF. Characteristics of women with hysterectomy. Maturitas. 1989;11(4):319–29. Epub 1989/12/01. pmid:2615667.
- 16. Cooper R, Lucke J, Lawlor DA, Mishra G, Chang JH, Ebrahim S, et al. Socioeconomic position and hysterectomy: a cross-cohort comparison of women in Australia and Great Britain. J Epidemiol Community Health. 2008;62(12):1057–63. Epub 2008/04/17. pmid:18413433; PubMed Central PMCID: PMCPMC2582341.
- 17. Bower JK, Schreiner PJ, Sternfeld B, Lewis CE. Black-White differences in hysterectomy prevalence: the CARDIA study. Am J Public Health. 2009;99(2):300–7. Epub 2008/12/09. pmid:19059854; PubMed Central PMCID: PMCPMC2622766.
- 18. Cooper R, Hardy R, Kuh D. Timing of menarche, childbearing and hysterectomy risk. Maturitas. 2008;61(4):317–22. Epub 2008/11/18. pmid:19013032; PubMed Central PMCID: PMCPMC3500690.
- 19. Settnes A, Lange AP, Jorgensen T. Gynaecological correlates of hysterectomy in Danish women. Int J Epidemiol. 1997;26(2):364–70. Epub 1997/04/01. pmid:9169172.
- 20. Marks NF, Shinberg DS. Socioeconomic differences in hysterectomy: the Wisconsin Longitudinal Study. Am J Public Health. 1997;87(9):1507–14. Epub 1997/10/07. pmid:9314805; PubMed Central PMCID: PMCPMC1380978.
- 21. Harlow BL, Barbieri RL. Influence of education on risk of hysterectomy before age 45 years. Am J Epidemiol. 1999;150(8):843–7. Epub 1999/10/16. pmid:10522655.
- 22. Santow G, Bracher M. Correlates of hysterectomy in Australia. Soc Sci Med. 1992;34(8):929–42. Epub 1992/04/01. pmid:1604382.
- 23. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate healthcare interventions: explanation and elaboration. BMJ. 2009;339:b2700. pmid:19622552; PubMed Central PMCID: PMC2714672.
- 24. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009;339:b2535. pmid:19622551; PubMed Central PMCID: PMC2714657.
- 25. Stang A, Kluttig A, Moebus S, Volzke H, Berger K, Greiser KH, et al. Educational level, prevalence of hysterectomy, and age at amenorrhoea: a cross-sectional analysis of 9536 women from six population-based cohort studies in Germany. BMC Womens Health. 2014;14:10. pmid:24433474; PubMed Central PMCID: PMC3898063.
- 26. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–60. pmid:12958120; PubMed Central PMCID: PMC192859.
- 27.
Higgins JPT, Green S (editors),. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0: The Cochrane Collaboration; 2011 [21 February 2015]. Available from: www.cochrane-handbook.org.
- 28. Harbord R.M. H JPT. Meta-regression in Stata. The Stata Journal. 2008;8(4):493–519.
- 29. Schoenaker DA, Jackson CA, Rowlands JV, Mishra GD. Socioeconomic position, lifestyle factors and age at natural menopause: a systematic review and meta-analyses of studies across six continents. Int J Epidemiol. 2014;43(5):1542–62. pmid:24771324; PubMed Central PMCID: PMC4190515.
- 30. Dennerstein L, Shelley J, Smith AM, Ryan M. Hysterectomy experience among mid-aged Australian women. Med J Aust. 1994;161(5):311–3. Epub 1994/09/05. pmid:7830667.
- 31. Palmer JR, Rao RS, Adams-Campbell LL, Rosenberg L. Correlates of hysterectomy among African-American women. Am J Epidemiol. 1999;150(12):1309–15. Epub 1999/12/22. pmid:10604773.
- 32. Sievert LL, Murphy L, Morrison LA, Reza AM, Brown DE. Age at menopause and determinants of hysterectomy and menopause in a multi-ethnic community: the Hilo Women's Health Study. Maturitas. 2013;76(4):334–41. Epub 2013/09/24. pmid:24054435; PubMed Central PMCID: PMC3840033.
- 33. Brett KM, Marsh JV, Madans JH. Epidemiology of hysterectomy in the United States: demographic and reproductive factors in a nationally representative sample. J Womens Health. 1997;6(3):309–16. Epub 1997/06/01. pmid:9201665.
- 34. Cooper R, Lawlor DA, Hardy R, Ebrahim S, Leon DA, Wadsworth ME, et al. Socio-economic position across the life course and hysterectomy in three British cohorts: a cross-cohort comparative study. BJOG. 2005;112(8):1126–33. Epub 2005/07/28. pmid:16045529.
- 35. Dharmalingam A, Pool I, Dickson J. Biosocial determinants of hysterectomy in New Zealand. Am J Public Health. 2000;90(9):1455–8. Epub 2000/09/13. pmid:10983207; PubMed Central PMCID: PMCPMC1447627.
- 36. Santow G. Education and hysterectomy. Aust N Z J Obstet Gynaecol. 1995;35(1):60–9. Epub 1995/02/01. pmid:7772004.
- 37. Schofield MJ, Hennrikus DJ, Redman S, Sanson-Fisher RW. Prevalence and characteristics of women who have had a hysterectomy in a community survey. Aust N Z J Obstet Gynaecol. 1991;31(2):153–8. Epub 1991/05/01. pmid:1930039.
- 38. Settnes A, Jorgensen T. Hysterectomy in a Danish cohort. Prevalence, incidence and socio-demographic characteristics. Acta Obstet Gynecol Scand. 1996;75(3):274–80. Epub 1996/03/01. pmid:8607343.
- 39. Brett KM, Higgins JA. Hysterectomy prevalence by Hispanic ethnicity: evidence from a national survey. Am J Public Health. 2003;93(2):307–12. Epub 2003/01/30. pmid:12554591; PubMed Central PMCID: PMCPMC1447735.
- 40. Ceausu I, Shakir YA, Lidfeldt J, Samsioe G, Nerbrand C. The hysterectomized woman. Is she special? The women's health in the Lund area (WHILA) study. Maturitas. 2006;53(2):201–9. Epub 2005/12/22. pmid:16368473.
- 41. Erekson EA, Weitzen S, Sung VW, Raker CA, Myers DL. Socioeconomic indicators and hysterectomy status in the United States, 2004. J Reprod Med. 2009;54(9):553–8. pmid:19947032; PubMed Central PMCID: PMC2883776.
- 42. Kjerulff K, Langenberg P, Guzinski G. The socioeconomic correlates of hysterectomies in the United States. Am J Public Health. 1993;83(1):106–8. Epub 1993/01/01. pmid:8417592; PubMed Central PMCID: PMCPMC1694507.
- 43. MacLennan AH, MacLennan A, Wilson D. The prevalence of hysterectomy in South Australia. Med J Aust. 1993;158(12):807–9. Epub 1993/06/21. pmid:8326889.
- 44. Progetto Menopausa Italia Study Group (PMISG). Determinants of hysterectomy and oophorectomy in women attending menopause clinics in Italy. Maturitas. 2000;36(1):19–25. Epub 2000/09/16. pmid:10989238.
- 45. Qi L, Nassir R, Kosoy R, Garcia L, Waetjen LE, Ochs-Balcom HM, et al. Relationship between hysterectomy and admixture in African American women. Am J Obstet Gynecol. 2013;208(4):279 e1–7. Epub 2013/01/22. pmid:23333549; PubMed Central PMCID: PMCPMC3613241.
- 46. Laughlin SK, Schroeder JC, Baird DD. New directions in the epidemiology of uterine fibroids. Semin Reprod Med. 2010;28(3):204–17. pmid:20414843.
- 47. Okolo S. Incidence, aetiology and epidemiology of uterine fibroids. Best Pract Res Clin Obstet Gynaecol. 2008;22(4):571–88. pmid:18534913.
- 48. Velez Edwards DR, Baird DD, Hartmann KE. Association of age at menarche with increasing number of fibroids in a cohort of women who underwent standardized ultrasound assessment. Am J Epidemiol. 2013;178(3):426–33. pmid:23817917; PubMed Central PMCID: PMC3727338.
- 49. Nnoaham KE, Webster P, Kumbang J, Kennedy SH, Zondervan KT. Is early age at menarche a risk factor for endometriosis? A systematic review and meta-analysis of case-control studies. Fertil Steril. 2012;98(3):702–12 e6. pmid:22728052; PubMed Central PMCID: PMC3502866.
- 50. Missmer SA, Hankinson SE, Spiegelman D, Barbieri RL, Malspeis S, Willett WC, et al. Reproductive history and endometriosis among premenopausal women. Obstet Gynecol. 2004;104(5 Pt 1):965–74. pmid:15516386.
- 51. Cramer DW, Missmer SA. The epidemiology of endometriosis. Ann N Y Acad Sci. 2002;955:11–22; discussion 34–6, 396–406. pmid:11949940.
- 52. Udry JR, Cliquet RL. A cross-cultural examination of the relationship between ages at menarche, marriage, and first birth. Demography. 1982;19(1):53–63. pmid:7067870.
- 53.
Bates LM, Berkman L.F., Glymour M.M. Socioeconomic Determinants of Women's Health: The Changing Landscape of Education, Work and Marriage. In: Goldman MB, Troisi R., Rexrode K.M., editor. Women Health. 2nd edition ed. Boston: Elsevier Science and Technology Books, Academic Press; 2013. p. 671–83.
- 54. Harris MF, Furler J., Valenti L., Harris E., Britt H. Matching care to need in general practice: A secondary analysis of Bettering the Evaluation and Care of Health (BEACH) data. Australian Journal of Primary Health. 2004;10(3):151–5.
- 55. Fitzgerald DM, Berecki-Gisolf J, Hockey RL, Dobson AJ. Hysterectomy and weight gain. Menopause. 2009;16(2):279–85. pmid:18971792.
- 56.
Groot WavdB, H. What does education do to our health Copenhagen: OECD/CERI; 2006 [3 October 2013]. Available from: http://www.oecd.org/edu/innovation-education/37425763.pdf.
- 57. Cooper R, Blell M, Hardy R, Black S, Pollard TM, Wadsworth ME, et al. Validity of age at menarche self-reported in adulthood. J Epidemiol Community Health. 2006;60(11):993–7. pmid:17053289; PubMed Central PMCID: PMC2465480.
- 58. Charalampopoulos D, McLoughlin A, Elks CE, Ong KK. Age at menarche and risks of all-cause and cardiovascular death: a systematic review and meta-analysis. Am J Epidemiol. 2014;180(1):29–40. pmid:24920784; PubMed Central PMCID: PMC4070937.
- 59. Kavanagh A, Bentley RJ, Turrell G, Shaw J, Dunstan D, Subramanian SV. Socioeconomic position, gender, health behaviours and biomarkers of cardiovascular disease and diabetes. Soc Sci Med. 2010;71(6):1150–60. pmid:20667641.
- 60. Mueller NT, Duncan BB, Barreto SM, Chor D, Bessel M, Aquino EM, et al. Earlier age at menarche is associated with higher diabetes risk and cardiometabolic disease risk factors in Brazilian adults: Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). Cardiovasc Diabetol. 2014;13:22. pmid:24438044; PubMed Central PMCID: PMC3899384.
- 61. Stockl D, Doring A, Peters A, Thorand B, Heier M, Huth C, et al. Age at menarche is associated with prediabetes and diabetes in women (aged 32–81 years) from the general population: the KORA F4 Study. Diabetologia. 2012;55(3):681–8. pmid:22170465.
- 62. Canoy D, Beral V, Balkwill A, Wright FL, Kroll ME, Reeves GK, et al. Age at Menarche and Risks of Coronary Heart and Other Vascular Diseases in a Large UK Cohort. Circulation. 2015;131(3):237–44. pmid:25512444.