Discussion
This is the first study to examine VMS profiles in relation to diabetes using data collected across 15 years. We used four distinct VMS profiles identified previously, including the early severe profile, which is characterized by symptoms being reported while still premenopausal, with a peak at menopause and a steady decline in postmenopause. Women with this profile were statistically significantly more likely to have diabetes than women with the mild profile.
To date, few studies have examined the links between VMS and surrogate measures as indices of diabetes, and the results have been mixed. Whereas the proposed impaired glucose delivery hypothesis suggests that hot flushes emerge as blood glucose falls, other studies have shown the opposite. One of these studies found that menopausal sweats were associated with elevated glucose levels, although this association did not remain statistically significant after adjustment for BMI. Our finding of a higher prevalence of diabetes in women with an early severe profile is in line with recent findings from the Study of Women's Health Across the Nation, which show that hot flushes were associated with a higher homeostasis model assessment index and glucose level. The majority of the symptomatic women in that study was premenopausal/early perimenopausal and is therefore comparable to our group of women with early severe VMS that develop symptoms for the first time in premenopausal years. However, our finding of a weak positive association between the late severe profile and diabetes is not in keeping with another study that showed that hot flushes were not related to insulin resistance among a comparable population of postmenopausal women. Nevertheless, none of the previous studies differentiated between early and late VMS and examined diabetes events.
To date, only the WHI has taken the timing of VMS into account and has observed that late VMS, but not early VMS, were associated with increased coronary heart disease risk and all-cause mortality. In the present study, we found positive associations between diabetes and the early severe profile, whereas the findings were only weakly suggestive of an association with the late severe profile. However, comparison between our study and the WHI is difficult because of heterogeneity in study populations and in defining VMS profiles. Whereas our analysis focused on women undergoing natural menopause at the age of 45 to 50 years, the WHI included older postmenopausal women (aged 50-70 y at entry), of which 20% reported surgical menopause and more than 45% used hormones. A subgroup analysis limited to women undergoing natural menopause did not reveal any statistically significant associations. Furthermore, with our approach of using latent class analysis, we were able to identify variations in symptoms through midlife; this is a data-driven approach and therefore different from the a priori grouping used in the WHI. Moreover, in the WHI, symptoms were analyzed relative to both age at menopause and chronological age; this permitted less detailed characterization than our profiles, which were analyzed according to age relative to menopause only.
The physiology linking VMS to diabetes is not entirely clear. Women with an early severe profile were more likely to be overweight or obese. Heavier women tend to have more VMS and an increased risk for diabetes. However, controlling for BMI did not eliminate the observed association. This is in agreement with the findings from the Study of Women's Health Across the Nation. Another possible mechanism linking VMS to diabetes is sympathetic nervous system activity, which is not only thought to be altered in women with VMS but also believed to be involved in the development of diabetes. Inflammation and endothelial dysfunction could also be possible links between VMS and diabetes.
We do not have a clear explanation as to why the association with the early severe profile is stronger than the association with the late severe profile. Perhaps, VMS that develop for the first time in the premenopausal years are physiologically different from VMS that occur later in menopause. This small group of women may represent the group of women with the most severe symptoms. However, because this is the first study, to date, to examine the associations between timing of VMS and diabetes, further research is required to determine if our findings can be confirmed in other populations.
Our study has a few limitations. First, VMS and diabetes cases were self-reported. However, any misclassification is most likely to be random, which may have led us to underestimate any association. Furthermore, a validation study in this cohort showed that 70% of self-reported diabetes cases were verified by Medicare and Pharmaceutical Benefits Scheme records. This is consistent with findings from one study in which 72% of the self-reported cases were confirmed by general practitioners and compares favorably with findings from another study in which 64% of the positive reports of diabetes were validated with medical records. Second, because the VMS profiles reflect the prevalence of symptoms in the six available surveys, it was difficult to establish whether the symptoms occurred before the diagnosis of diabetes. We were therefore not able to study diabetes "incidence" as outcome and studied "prevalence" instead. Further research is required to study whether VMS profiles precede or follow the development of diabetes. Third, the obtained VMS profiles may be unique to this particular study. Although a previous study found similar longitudinal profiles, future studies in other study populations should be performed to see whether similar profiles emerge. Fourth, although we adjusted for a range of confounders, residual confounding caused by unmeasured factors, such as family history of diabetes, cannot be excluded. Finally, we assessed adiposity through BMI measurement. Although BMI is correlated with percentage of body fat, other clinical measures of adiposity, such as waist circumference or waist-to-hip ratio, may be better predictors of diabetes and should be examined as confounders in future studies.
Our study has a number of strengths. Our study population was large and community-based, improving the generalizability of our findings to other middle-aged women. Our analysis was limited to women undergoing natural menopause. However, the inclusion of women who underwent hysterectomy and surgical menopause and women on HT in the analysis would have to alter the characteristics or shapes of profiles, rather than just the level of severity, to substantively affect the main findings. Further strengths lie in the longitudinal approach of the ALSWH (with regular survey of symptoms through the menopausal transition) and the ability to adjust for a wide range of sociodemographic and lifestyle factors.