The best magazine
Methods
Study Population
Participants were recruited to the study from the Roy Morgan Research Single Source Database. This dynamic database has more than 300,000 respondents, with 50,000 new recruits annually. People stay on the database for varying periods, but the average retention is 2 years. Recruitment to the database is based on the Australian electoral roll. In Australia, where voting is compulsory, every adult is registered on this roll. Australia is divided into 150 federal electorates containing roughly equal numbers of registered electors and, thus, approximately equal numbers of eligible adults. Each federal electorate is divided into four approximately equal sections, resulting in about 600 sampling areas. Recruitment is carried out each weekend, such that each of the 600 sampling areas is sampled monthly.
Within each sampling area, the starting address is selected randomly from a list of all known addresses. This list is based on, and is maintained using, the Electoral Commission's "Street Segment" file, supplemented by Pacific Micromarketing's Australian Residential Address List, itself based on Australia Post's database of residential addresses. This provides better coverage of new residential development areas than reliance on census or electoral register information.
From the starting address, individuals are recruited to the database by interviews conducted in person on Saturdays and Sundays between 9:00AM and 4:00PM. Eight interviews are conducted per sampling point, and only one eligible person is recruited per household, starting with the youngest person available aged at least 14 years. From any one recruitment point, on average, four men and four women are recruited (ranging in age from 14 to >90 y), and those individuals would stay on the database for varying periods of time.
For the current study, women on the Roy Morgan Research Single Source Database who were aged between 40 and 65 years were approached by telephone and asked if they would like to participate in a survey of health of women at midlife. Women were excluded only if they were unable to complete a questionnaire in English. After verbal consent had been obtained, a study questionnaire with a reply-paid envelope was posted to each participant. Participants were asked to provide us permission to recontact them by telephone for essential data clarification. A free-call telephone number was provided if participants wanted assistance in completing the questionnaire. Deidentified questionnaires were electronically scanned and incorporated into an SPSS database.
We have used this robust approach to recruiting a representative sample of Australian women in other seminal studies. The study was approved by the Monash University Human Research Ethics Committee. Return of a completed questionnaire was accepted as written informed consent.
Assessment of Menopause Status
Women were classified as premenopausal, perimenopausal, or postmenopausal using published algorithms derived from the Stages of Reproductive Aging Workshop +10 classification. Each woman was classified based on age, history of bilateral oophorectomy, hysterectomy, menstrual bleeding, VMS, and use of systemic hormonal contraception or HT. The postmenopausal group was further divided into women younger than 55 years, women aged 55 to 59 years, and women aged 60 to 65 years.
Assessment of Menopausal Symptoms and Management of Incomplete Data
Menopausal symptoms were assessed using the Menopause-Specific Quality of Life Questionnaire (MENQOL), a validated self-administered instrument. The MENQOL contains 29 items composing four symptom domains: VMS, physical, psychological, and sexual. Women report whether they have experienced each symptom in the last 4 weeks (yes/no); if yes, they record the extent to which they are bothered (0, no bother; 6, extreme bother). Each individual question is answered on a scale from 1 to 8 (1, no symptoms). Each domain includes between 3 and 16 questions, with the domain score calculated as the mean of item scores within that domain.
"Yes" answers to symptoms without degrees of bother provided were treated as missing data. No group imputation was used for missing data. For missing responses, the mean score of the items answered was used if at least two of the three possible questions were completed for the vasomotor and sexual domains, and more than half of those for the psychological and physical domains.
Assigning Severity of VMS
As each VMS (hot flashes, night sweats, and sweating) is rated on a seven-point Likert scale, women who ranked their degree of bother as more than the midpoint of the item were considered to have moderate to severe symptoms. For hot flashes and night sweats, each symptom was scored as none (1), mild (≥2to ≤5), or moderate to severe (>5-8). Therefore, the overall VMS domain score was categorized as none (1), mild (≥2to ≤5), or moderate to severe (>5-8).
Sample Size
Calculation of total sample size was based on a 95% CI of ±2% around a prevalence estimate of 30% for moderate to severe VMS. The total number of completed questionnaires required was 2,000. We purposefully sampled women so that the age distribution of our sample population mimicked the age distribution of the adult Australian female population aged 40 to 65 years in 2011.
Statistical Analyses
Descriptive analysis was performed using numbers (percentages) for categorical data and means (SDs) for continuous data. Women using systemic hormonal contraception (combined hormonal contraception or systemic progestogen), HT (estrogen therapy, estrogen-progestogen therapy, or tibolone), "bioidentical" estrogen, or nonhormonal prescription medication specifically for VMS (gabapentin, sertraline, clonidine, and venlafaxine) were excluded from analyses of menopausal symptoms. Women using vaginal estrogen were also excluded from analyses of sexual symptoms. As only five women were younger than 41 years, these women were excluded from the diagrams of VMS graphed against age.
MENQOL domain scores by menopause status were tested using one-way analysis of variance, with Bonferroni test performed if analysis of variance was significant at an α of 0.05. The association of systemic hormone therapy use between postmenopausal women younger than 60 years and postmenopausal women aged 60 years or older was tested using χ test. A multiple logistic regression model was used to assess factors associated with any VMS, with the outcome defined as "any VMS" versus "no VMS." A second model was constructed, with the outcome defined as "moderate to severe VMS domain score" versus "mild/none." The factors associated with moderate to severe VMS and any VMS in bivariate analysis at P < 0.05 were included in the final regression models. All analyses were performed using STATA version 12.0 (StataCorp). All tests were two sided, and the α level was set at 0.05 for all statistical tests.
Source: ...