Objective
This study sought to investigate the age at natural menopause and its predictors in a cohort of human immunodeficiency virus (HIV)-infected women in Rio de Janeiro, Brazil.
Study Design
HIV-infected women ≥30 years of age were included. Menopause was defined as having ≥1 year since the last menstrual period. Early age at natural menopause was defined as the onset of menopause at ≤45 years of age. Multivariate Cox proportional hazards analysis was applied.
Results
A total of 667 women were included, and the median age at baseline was 34.9 years (interquartile range, 30.9–40.5 years). In all, 507 (76%) women were premenopausal, and 160 (24%) reached menopause during the observational period; of these, 36 of 160 (27%) had early menopause. The median age at natural menopause was 48 years (interquartile range, 45–50 years). Menarche at <11 years of age (hazard ratio [HR], 2.03; 95% confidence interval [CI], 1.23–3.37), cigarette smoking during the observational period (HR, 1.59; 95% CI, 1.08–2.33), chronic hepatitis C virus (HCV) infection (HR, 2.53; 95% CI, 1.27–5.07), and CD4 count <50 cells/mm 3 (HR, 3.07; 95% CI, 1.07–8.80) were significantly associated with an earlier age at natural menopause. The magnitudes of the effects of menarche at <11 years of age (HR, 2.7; 95% CI, 1.23–5.94), cigarette smoking during the observational period (HR, 3.00; 95% CI, 1.39–6.45), chronic HCV infection (HR, 6.26; 95% CI, 2.12–18.52), and CD4 count <50 cells/mm 3 (HR, 6.64; 95% CI, 1.91–23.20) were much higher and significantly associated with early natural menopause.
Conclusion
Early natural menopause was frequent among the HIV-infected women. In addition to menarche and cigarette smoking, which are menopausal factors among women in general, HIV-related immunodeficiency and chronic HCV were additional predictors for an earlier age at natural menopause. Adequate management of HIV in women is critical, as early onset of menopause has been associated with increased morbidity and mortality.
The worldwide life expectancy has continued to increase in recent decades, even in developing countries, leading to a greater number of individuals aged ≥60 years. The expanded coverage of combination antiretroviral therapy (cART) has led to significant reductions in morbidity and mortality worldwide, as well as in Brazil, effectively turning human immunodeficiency virus (HIV) infection into a chronic condition. From 1998 through 2010 in Brazil, an increase in acquired immune deficiency syndrome (AIDS) cases among individuals aged 50-59 years (9.5-16.3/100,000 inhabitants) and >60 years (2.8-5.1/100,000 inhabitants) has led to an increased number of HIV-infected women entering menopause.
Natural menopause is the permanent cessation of menstruation as a consequence of the loss of ovarian follicular activity and is defined as 12 consecutive months without menstrual periods. Early menopause is the permanent cessation of menstruation between 40-45 years of age. This condition affects 5% of women in the general population, whereas premature menopause occurring <40 years of age affects 1% of women.
Age at menopause varies substantially within and across populations, with the mean age at menopause in the developed world being typically higher than that observed in the developing world. A cross-sectional, population-based study conducted in 1997 through 1998 among 456 Brazilian women between 45-60 years of age selected through area cluster sampling showed a mean age at menopause of 51.2 years, similar to that observed in developed countries. Previous studies conducted worldwide, including a cross-sectional study of 96 HIV-infected Brazilian women, have observed that natural menopause occurs earlier in HIV-infected women, at approximately 46-49 years of age.
A number of factors have been proposed as predictors of natural menopause in the general population of women, including genetics, sociodemographics, lifestyle, smoking history, reproductive history, and adult and early childhood health conditions. However, few studies in the population of HIV-infected women have evaluated factors associated with earlier natural menopause, particularly in low- and middle-income countries.
Earlier menopause has been associated with an increased risk of negative outcomes such as atherosclerosis, cardiovascular disease, stroke, osteoporosis, and fracture in women from the general population. Therefore, identifying the age at menopause and its predictors is critical to the clinical and gynecological management of HIV-infected women, as postmenopausal women living with HIV/AIDS are more vulnerable to comorbidities compared to women in the general population.
The purpose of this study was to investigate the age at natural menopause and the potential predictors of menopause in a large, single-center cohort of HIV-infected women in Rio de Janeiro, Brazil.
Materials and Methods
The Instituto de Pesquisa Clínica Evandro Chagas HIV/AIDS Women’s Cohort and the study population
This study was conducted within the Instituto de Pesquisa Clínica Evandro Chagas (IPEC) HIV/AIDS Women’s Cohort, an open cohort that was established in 1996 at the IPEC, Oswaldo Cruz Foundation, Rio de Janeiro, Brazil. Data from this cohort were published previously. Briefly, study visits occurred once per year from 1996 through 2003 and every 6 months thereafter until 2011. Sociodemographic information, lifestyle behavior, reproductive history, gynecologic history, and laboratory data were collected using structured questionnaires, and specimens were obtained for Pap smears and the detection of sexually transmitted diseases. All women were referred for colposcopy during the baseline visit and further evaluation if abnormal cytology was detected. A total of 1002 women with HIV/AIDS were enrolled in the cohort from May 20, 1996, through Dec. 31, 2010.
Women considered at risk for natural menopause were eligible for the study, and the following inclusion criteria were applied: (1) inclusion in the IPEC Women’s Cohort from May 20, 1996, through Dec. 31, 2010; (2) premenopausal status; (3) age ≥30 years at the time of cohort entry; (4) age ≥30 years on Dec. 31, 2010; and (5) history of at least 1 follow-up visit after study entry.
In all, 728 women met the inclusion criteria (time at inclusion [T0]). However, 31 (4.3%) women were lost to follow-up before reaching 30 years of age, and 30 (4.1%) women did not perform a follow-up visit and were excluded from the analysis. As a result, 667 premenopausal women were considered in the analysis of age at natural menopause. Because women >45 years at the time of cohort enrollment (n = 59) were not eligible for the analysis of early age at natural menopause, only 608 of these premenopausal women were included in the analysis of early menopausal age ( Figure 1 ).
Study outcomes and definitions
The study outcomes included age at natural menopause and early age at natural menopause. Menopausal status was determined prospectively during the cohort interviews.
Natural menopause is defined by the World Health Organization (WHO) as the permanent cessation of menstruation resulting from the loss of ovarian follicular activity. This condition is clinically recognized after at least 12 months of amenorrhea, at which time the final menstrual period (FMP) is characterized with certainty.
Induced menopause followed by surgical removal of both ovaries (with or without hysterectomy) or iatrogenic ablation of ovarian function (eg, by chemotherapy or radiation) was not considered natural menopause in this study.
Early natural menopause is defined as the natural onset of menopause at an age ≤45 years. Premature menopause is defined as menopause occurring age <40 years, according to the WHO definition.
Covariates
Sociodemographic factors
Race/ethnicity, schooling, and monthly family income (in 2011, the Brazilian monthly minimum wage was US$327.92) were self-reported at cohort entry and evaluated as fixed-effect covariates.
Reproductive factors
Age at menarche, parity (which was a time-dependent, categorized variable that also took into account the number of children born before study entry), and oral contraceptive or other exogenous hormone exposure (time-dependent variable) were assessed by questionnaire. For oral contraceptive and exogenous hormone exposure, an answer of “yes” corresponded to use at T0 or during the observational period, and an answer of “no” indicated that the women were never exposed.
Lifestyle factors
Alcohol consumption was assessed with the following question at the time of cohort entry: “When you drink, how many distilled or fermented drinks do you ingest?” Frequency of alcohol intake was not evaluated at entry and during the observational period.
Cigarette smoking was assessed through questions about the date of the first and last cigarettes and the number of cigarettes smoked per day. We computed a woman’s age at the first and the last cigarette exposure using her birth date. Cigarette smoking was analyzed as 2 time-dependent, categorized covariates, thus allowing a unique point change. The first covariate was cigarette-smoking exposure, which took into account the time of cigarette exposure during the observational period. The second covariate was cigarette exposure in pack-years, which was calculated as the average number of cigarettes smoked per day multiplied by the length of smoking time divided by 20. Time in years from smoking initiation to the outcome or censorship was calculated.
Lifetime illicit drug use was assessed by self-report at cohort entry, with “yes” responses indicating that the women had previously used marijuana, cocaine, crack, glue, or lysergic acid diethylamide (LSD). The use of intravenous and snorted cocaine was evaluated in the statistical inferences.
Health-related factors
Anthropometric data were obtained from the patients’ medical charts. The body mass index (BMI) was calculated as the woman’s weight (kilogram) divided by the square of her height (meters) and was reported as kg/m 2 . BMI was categorized according to WHO standards for adults as underweight (<18.5 kg/m 2 ), normal weight (18.5-24.9 kg/m 2 ), and overweight/obese (≥25.0 kg/m 2 ). BMI was analyzed as a time-dependent covariate, allowing for multiple changes in value over the study period. BMI at T0 was defined as the value obtained within 6 months of T0.
All other health-related covariates were assessed using the IPEC HIV/AIDS clinical database and were analyzed as time-dependent variables with a unique point change at the date of the diagnosis.
Type 2 diabetes was diagnosed by a fasting plasma glucose ≥126 mg/dL in 2 samples collected on different days or by a 2-hour plasma glucose ≥200 mg/dL during an oral glucose tolerance test.
Chronic hepatitis C virus (HCV) infection was diagnosed following a positive HCV enzyme-linked immunosorbent assay result and confirmed by recombinant immunoblot assay or 2 detectable HCV RNA assays at least 6 months apart.
Hypothyroidism was diagnosed by the presence of at least 1 of the following: goiter, fatigue, cold intolerance, dry skin, constipation, bradycardia, weight gain, changes in menstrual pattern, and decreased levels of thyroxine and tri-iodothyronine or increased levels of thyroid-stimulating hormone.
HIV/AIDS-related factors
The CD4 + T-cell count was analyzed as a time-dependent and categorized covariate, allowing for multiple point changes over the study period. The CD4 + T-cell count at T0 was described and defined as the value obtained within 6 months of T0. Severe immunodeficiency was defined as a CD4 count <50 cells/mm 3 .
The nadir CD4 + T-cell count was determined according to the lowest CD4 + T-cell count available from the time of HIV diagnosis to the end of the observational period.
AIDS-defining illnesses were defined as the presence of any 1993 Centers for Disease Control and Prevention (CDC)-defined AIDS-defining illnesses at any time during the course of HIV infection to the end of the observational period. AIDS-defining illness was assessed as a time-dependent variable.
cART was defined as any lifetime exposure until the end of the observational period to ≥2 nucleoside reverse transcriptase inhibitors and a nonnucleoside reverse transcriptase inhibitor or at least 1 protease inhibitor.
Statistical analysis
The median (interquartile range [IQR]) and frequency (%) were used to describe the women’s characteristics for continuous and categorical data, respectively. Premenopausal women were observed from the time of their cohort entry (for women aged ≥30 years) or from the time they turned 30 years of age (for women aged <30 years at cohort entry) until the end of the observational period (Dec. 31, 2011). This allowed women enrolled in late 2010 to be observed for natural menopause. Women who presented with induced menopause–for example, from hysterectomy, bilateral oophorectomy, chemotherapy, and/or radiotherapy–were censored from the study at the time of menopause, and women who were lost to follow-up before Dec. 31, 2011, were censored from the study at the time of the last gynecological visit. Women were censored from the analysis of early age at natural menopause as they turned 45 years of age, as women of that age are no longer at risk for early menopause.
A Kaplan-Meier model of natural menopause estimated the reverse survival function between natural menopause and age and was defined as the probability of being in natural menopause at the age of 45 and 50 years.
Incidence rates were estimated for both outcomes and reported per 100 person-years. Cox proportional hazards regression analysis using age as a time scale, which adjusts the effects of other covariates by age, which often leads to coefficients with less bias, was used to assess the role of selected covariates on both outcomes. Collinearity between variables was assessed. We fitted the unadjusted models and considered all covariates as statistically significant at the significance level of 20% for age at natural menopause and at 10% for early age at natural menopause as thresholds for the multivariate analysis. A backward elimination method was used, and covariates with the least significant levels were sequentially removed. Covariates statistically significant at 5% ( P < .05) and those considered as confounders (eg, when removed, a change ≥10% in the hazard ratio [HR] of any other variable of the model was observed) remained in the final model. Proportionality of risks was tested using Schoenfeld residuals analysis.
Software (R, version 3.0.2; R Foundation for Statistical Computing, Vienna, Austria) was used in all analyses.
Ethical statement
The study protocol was reviewed and approved by IPEC, Oswaldo Cruz Foundation (CAE 020/2001) Ethics Committee. Written informed consent was obtained from all the women.
Results
Sample characteristics
In all, 667 women were followed for a total of 3814 person-years with a median follow-up of 5.0 years (IQR, 2.7–8.3). Of the 667 women, 142 (21.4%) were censored from the study during the observational period for the following reasons: 41 (6.1%) died, 23 (3.4%) had surgically induced menopause, 5 (0.7%) had chemotherapy or radiation-induced menopause, 4 (0.6%) were transferred to another facility, and 69 (10.3%) missed their scheduled follow-up gynecological visit for >1 year.
The characteristics of the study population are shown in Table 1 . The median baseline age was 34.9 years (IQR, 30.9–40.5). The majority of women were nonwhite (60.4%) with up to 8 years of schooling (56.2%) and a family income ≤5 Brazilian minimum wages (80.8%). The median age at menarche was 13 years (IQR, 12–14), and 60.9% of the women were multiparous. Forty percent (n = 314) of women reported a lifetime exposure to cigarette smoking, but only 26% (n = 173) remained under exposure to cigarette smoking during the study observation period. Overall, women who had quit smoking did so for a median of 11.7 years (IQR, 5.5–19.0) prior to natural menopause or exclusion from the study. Among the 173 women exposed to cigarette smoking during the observational period, 20.2% (n = 35) reported smoking cessation; of these, 20% experienced natural menopause, and the median time from smoking cessation until the onset of natural menopause was 11.5 years (IQR, 5.8–18.6).
Characteristic | Total (n = 667) | Natural menopausal women (n = 132) | Early natural menopausal women (n = 36) |
---|---|---|---|
Age at baseline, y | 34.9 (30.9–40.5) | ||
Race/ethnicity | |||
White | 264 (39.6) | 66 (50.0) | 20 (55.6) |
Nonwhite | 403 (60.4) | 66 (50.0) | 16 (44.4) |
Schooling, y | |||
>11 | 72 (10.8) | 11 (8.3) | 2 (5.6) |
>8-11 | 220 (33.0) | 44 (33.3) | 13 (36.1) |
≤8 | 375 (56.2) | 77 (58.3) | 21 (58.3) |
Monthly family income a | 560 (300–1000) | ||
>5 | 127 (19.0) | 35 (26.5) | 9 (25.0) |
2-5 | 215 (32.2) | 43 (32.6) | 11 (30.6) |
0-2 | 324 (48.6) | 54 (40.9) | 16 (44.4) |
Missing | 1 (0.2) | – | – |
Age at menarche, y | 13 (12–14) | ||
≥11 | 600 (90.0) | 113 (85.6) | 27 (75.0) |
<11 | 66 (9.9) | 19 (14.4) | 9 (25.0) |
Missing | 1 (0.1) | – | – |
Parity b | 2 (1–3) | ||
0 | 91 (13.6) | 12 (9.1) | 3 (8.3) |
≥1 | 576 (86.4) | 120 (90.9) | 33 (91.7) |
Lifetime oral contraceptive or other exogenous hormone use | |||
Yes | 369 (55.3) | 32 (24.2) | 11 (30.6) |
No | 298 (44.7) | 100 (75.8) | 25 (69.4) |
Alcohol use | |||
No | 415 (62.2) | 87 (65.9) | 18 (50.0) |
1-2 drinks | 75 (11.2) | 9 (6.8) | 4 (11.1) |
3-4 drinks | 64 (9.6) | 11 (8.3) | 3 (8.3) |
≥5 drinks | 113 (16.9) | 25 (18.9) | 11 (30.6) |
Cigarette smoking exposure b | |||
No | 483 (72.41) | 83 (62.9) | 17 (47.2) |
Yes | 173 (25.94) | 49 (37.1) | 19 (52.8) |
Missing | 11 (1.65) | – | – |
Cigarette exposure in pack-y | 9.5 (2.7–23.2) | ||
Never smoked | 353 (52.9) | 51 (38.6) | 11 (30.6) |
<10 | 151 (22.6) | 30 (22.7) | 10 (27.8) |
10-19 | 58 (8.7) | 16 (12.1) | 3 (8.3) |
≥20 | 90 (13.5) | 34 (25.8) | 12 (33.3) |
Missing | 15 (2.2) | 1 (0.8) | – |
Lifetime illicit drug use | |||
No | 550 (82.5) | 111 (84.1) | 27 (75.0) |
Yes | 117 (17.5) | 21 (15.9) | 9 (25.0) |
Lifetime cocaine use (intravenous or snorted) | |||
No | 579 (86.8) | 118 (89.4) | 29 (80.6) |
Yes | 88 (13.2) | 14 (10.6) | 7 (19.4) |
Body mass index, kg/m 2 b,c | |||
Normal weight | 324 (48.6) | 74 (56.1) | 21 (58.3) |
Overweight/obese | 247 (37.0) | 46 (34.8) | 10 (27.8) |
Underweight | 51 (7.6) | 9 (6.8) | 4 (11.1) |
Missing | 45 (6.8) | 3 (2.3) | 1 (2.8) |
Type 2 diabetes | |||
No | 589 (88.3) | 108 (81.8) | 29 (80.6) |
Yes | 78 (11.7) | 24 (18.2) | 7 (19.4) |
Chronic hepatitis C virus b | |||
No | 633 (94.9) | 116 (87.9) | 31 (86.1) |
Yes | 34 (5.1) | 16 (12.1) | 5 (13.9) |
Hypothyroidism b | |||
No | 654 (98.1) | 129 (97.7) | 35 (97.2) |
Yes | 13 (1.9) | 3 (2.3) | 1 (2.8) |
CD4 count nadir, cells/mm 3 | 182 (74–278) | ||
≥350 | 94 (14.1) | 12 (9.1) | 3 (8.3) |
200-349 | 213 (31.9) | 34 (25.8) | 7 (19.4) |
100-199 | 155 (23.2) | 28 (21.2) | 10 (27.9) |
50-99 | 81 (12.1) | 23 (17.4) | 4 (11.1) |
<50 | 123 (18.4) | 35 (26.5) | 12 (33.3) |
Missing | 1 (0.1) | – | – |
CD4 count, cells/mm 3 b,c | |||
≥350 | 334 (50.1) | 52 (39.4) | 12 (33.3) |
200-349 | 153 (22.9) | 31 (23.5) | 11 (30.6) |
100-199 | 81 (12.1) | 24 (18.2) | 7 (19.4) |
50-99 | 33 (5.0) | 12 (9.1) | 3 (8.3) |
<50 | 28 (4.2) | 8 (6.0) | 2 (5.6) |
Missing | 38 (5.7) | 5 (3.8) | 1 (2.8) |
AIDS-defining illnesses | |||
No | 373 (55.9) | 56 (42.4) | 12 (33.3) |
Yes | 294 (44.1) | 76 (57.6) | 24 (66.7) |
Combination antiretroviral therapy exposure | |||
No | 77 (11.5) | 14 (10.6) | 3 (8.3) |
Yes | 590 (88.5) | 118 (89.4) | 33 (91.7) |
b Frequencies presented for baseline but as time-dependent covariate
c Baseline body mass index and CD4 cell counts were defined as values obtained within 6 months (before or after) of enrollment.
The median nadir CD4 count was 182 cells/mm 3 (IQR, 74–278), and 53.7% of women had a nadir CD4 count <200 cells/mm 3 . Lifetime cART exposure was reported by 88.5% of women for a median time of 4.9 years (IQR, 2.4–9.0).
Age at natural menopause and its predictors
Natural menopause was observed in 132 of 667 women, corresponding to an incidence of 3.46 per 100 person-years (95% confidence interval [CI], 2.90–4.09). The probability of reaching menopause at age ≤50 years was .5 (95% CI, 0.40–0.57) ( Figure 2 ). The median age at natural menopause was 48 years (IQR, 45–50).
The results from univariate and multivariate analysis for age at natural menopause are presented in Table 2 . As cigarette smoking exposure and cigarette exposure in pack-years covariates were collinear, the former was chosen for multivariate analysis. Collinearity was observed between the time-dependent CD4 cell count and CD4 cell count nadir covariates, and the former was chosen for multivariate analysis.
Characteristic | Unadjusted analysis | Adjusted analysis | ||||
---|---|---|---|---|---|---|
HR | 95% CI | P value | HR | 95% CI | P value | |
Race/ethnicity | ||||||
White | 1 | |||||
Nonwhite | 0.71 | 0.50–1.01 | .053 | |||
Schooling, y | ||||||
>11 | 1 | |||||
>8-11 | 1.21 | 0.62–2.35 | .578 | |||
≤8 | 1.34 | 0.71–2.52 | .373 | |||
Monthly family income a,b | ||||||
>5 | 1 | |||||
2-5 | 1.34 | 0.85–2.10 | .205 | |||
0-2 | 1.09 | 0.71–1.68 | .684 | |||
Age at menarche, y b | ||||||
≥11 | 1 | 1 | ||||
<11 | 1.98 | 1.20–3.24 | .007 | 2.03 | 1.23–3.37 | .006 |
Parity b | ||||||
0 | 1 | 1 | ||||
≥1 | 0.60 | 0.32–1.13 | .116 | 0.55 | 0.29–1.05 | .071 |
Lifetime oral contraceptive or other exogenous hormone use | ||||||
Yes | 1 | 1 | ||||
No | 1.41 | 0.94–2.12 | .102 | 1.44 | 0.95–2.18 | .085 |
Alcohol use | ||||||
No | 1 | |||||
1-2 drinks | 0.73 | 0.37–1.46 | .377 | |||
3-4 drinks | 0.97 | 0.52–1.83 | .931 | |||
≥5 drinks | 0.81 | 0.52–1.27 | .355 | |||
Cigarette smoking exposure b,c | ||||||
No | 1 | 1 | ||||
Yes | 1.62 | 1.12–2.36 | .011 | 1.59 | 1.08–2.33 | .018 |
Lifetime cocaine use (intravenous or snorted) | ||||||
No | 1 | |||||
Yes | 0.77 | 0.44–1.34 | .348 | |||
Body mass index, kg/m 2 b,c | ||||||
Normal weight | 1 | 1 | ||||
Overweight/obese | 0.68 | 0.48–0.97 | .034 | 0.74 | 0.52–1.07 | .111 |
Underweight | 0.73 | 0.26–2.00 | .538 | 0.75 | 0.27–2.07 | .573 |
Type 2 diabetes | ||||||
No | 1 | |||||
Yes | 0.80 | 0.50–1.27 | .339 | |||
Chronic hepatitis C virus b | ||||||
No | 1 | 1 | ||||
Yes | 2.02 | 1.06–3.88 | .034 | 2.53 | 1.27–5.07 | .009 |
Hypothyroidism b | ||||||
No | 1 | |||||
Yes | 0.37 | 0.05–2.80 | .336 | |||
CD4 count, cells/mm 3 b,c | ||||||
≥350 | 1 | 1 | ||||
200-349 | 1.07 | 0.67–1.72 | .766 | 1.02 | 0.63–1.65 | .934 |
100-199 | 1.07 | 0.48–2.36 | .873 | 0.92 | 0.41–2.10 | .849 |
50-99 | 1.06 | 0.34–3.37 | .918 | 0.82 | 0.25–2.67 | .741 |
<50 | 3.47 | 1.24–9.71 | .018 | 3.07 | 1.07–8.80 | .037 |
AIDS-defining illnesses | ||||||
No | 1 | 1 | ||||
Yes | 1.51 | 1.07–2.15 | .020 | 1.40 | 0.97–2.04 | .075 |
Combination antiretroviral therapy exposure | ||||||
No | 1 | |||||
Yes | 0.74 | 0.42–1.30 | .291 |
b Missing data–monthly family income: n = 1, menarche: n = 1, cigarette smoking: n = 11, body mass index: n = 18, CD4 cell count: n = 1
Early menarche (HR, 2.03; 95% CI, 1.23–3.37), cigarette smoking exposure (HR, 1.59; 95% CI, 1.08–2.33), chronic HCV (HR, 2.53; 95% CI, 1.27–5.07), and a CD4 count <50 cells/mm 3 (HR, 3.07; 95% CI, 1.07–8.80) remained significantly associated with age at natural menopause in the final multivariate model.
Predictors of early age at natural menopause
In all, 608 women were evaluated for the outcome of early age at natural menopause and were followed up for a total of 3299 person-years with a median follow-up of 4.6 years (IQR, 2.5–7.7). Of these, 112 (18.4%) were censored from the study for the following reasons: 33 (5.4%) died, 15 (2.5%) had surgically induced menopause, 4 (0.7%) had chemotherapy or radiation-induced menopause, 4 (0.7%) transferred out of the study, and 56 (9.2%) missed their scheduled follow-up gynecological visit for >1 year.
Early natural menopause was observed in 36 women, with an incidence of 1.09 (95% CI, 0.77–1.49) per 100 person-years. The probability of reaching menopause at age ≤45 years was .1 (95% CI, 0.06–0.15) ( Figure 2 ). Only 3 women experienced premature menopause.
The results from univariate and multivariate analyses for early age at natural menopause are presented in Table 3 . The same collinearity profile as that for age at natural menopause outcome was observed. Cigarette smoking exposure and time-dependent CD4 cell count covariates were entered into the initial multivariate model.
Characteristic | Unadjusted analysis | Adjusted analysis | ||||
---|---|---|---|---|---|---|
HR | 95% CI | P value | HR | 95% CI | P value | |
Race/ethnicity | ||||||
White | 1 | |||||
Nonwhite | 0.59 | 0.30–1.13 | .113 | |||
Schooling, y | ||||||
>11 | 1 | |||||
>8-11 | 2.57 | 0.58–11.4 | .214 | |||
≤8 | 2.55 | 0.60–10.87 | .207 | |||
Monthly family income a,b | ||||||
>5 | 1 | |||||
2-5 | 1.03 | 0.43–2.49 | .943 | |||
0-2 | 1.17 | 0.52–2.65 | .707 | |||
Age at menarche, y b | ||||||
≥11 | 1 | 1 | ||||
<11 | 3.07 | 1.44–6.54 | .004 | 2.70 | 1.23–5.94 | .014 |
Parity b | ||||||
0 | 1 | |||||
≥1 | 1.32 | 0.40–4.30 | .648 | |||
Lifetime oral contraceptive or other exogenous hormone use | ||||||
Yes | 1 | |||||
No | 1.80 | 0.89–3.67 | .104 | |||
Alcohol use | ||||||
No | 1 | 1 | ||||
1-2 drinks | 1.33 | 0.45–3.92 | .611 | 1.18 | 0.38–3.64 | .779 |
3-4 drinks | 1.11 | 0.33–3.76 | .873 | 0.65 | 0.18–2.41 | .523 |
≥5 drinks | 2.13 | 1.01–4.52 | .048 | 1.32 | 0.55–3.16 | .534 |
Cigarette smoking exposure b,c | ||||||
No | 1 | 1 | ||||
Yes | 4.08 | 2.11–7.89 | < .001 | 3.00 | 1.39–6.45 | .005 |
Lifetime cocaine use (intravenous or snorted) | ||||||
No | 1 | |||||
Yes | 1.91 | 0.84–4.36 | .125 | |||
Body mass index, kg/m 2 b,c | ||||||
Normal weight | 1 | 1 | ||||
Overweight/obese | 0.45 | 0.23–0.91 | .027 | 0.53 | 0.25–1.10 | .086 |
Underweight | 0.49 | 0.07–3.63 | .485 | 0.32 | 0.04–2.53 | .280 |
Type 2 diabetes | ||||||
No | 1 | |||||
Yes | 1.08 | 0.47–2.47 | .853 | |||
Chronic hepatitis C virus b | ||||||
No | 1 | 1 | ||||
Yes | 5.39 | 2.09–13.89 | < .001 | 6.26 | 2.12–18.51 | .001 |
CD4 count, cells/mm 3 b,c | ||||||
≥350 | 1 | 1 | ||||
200-349 | 1.14 | 0.46–2.85 | .773 | 1.06 | 0.41–2.74 | .90 |
100-199 | 1.99 | 0.68–5.83 | .212 | 1.31 | 0.41–4.20 | .647 |
50-99 | 2.60 | 0.61–11.15 | .199 | 1.18 | 0.23–5.94 | .842 |
<50 | 8.24 | 2.77–24.48 | < .001 | 6.64 | 1.91–23.16 | .003 |
AIDS-defining illnesses | ||||||
No | 1 | 1 | ||||
Yes | 2.38 | 1.19–4.75 | .014 | 1.49 | 0.67–3.33 | .327 |
Combination antiretroviral therapy exposure | ||||||
No | 1 | |||||
Yes | 0.86 | 0.26–2.79 | .795 |