Discussion
We wished to identify the characteristics of women who seemed to be resilient or vulnerable to the effects of hot flushes or night sweats. We were able to include a wide range of personal biopsychosocial characteristics in our analyses. Factors independently associated with different group memberships included the following: having had children, smoking, education, self-reported poor physical health, history of menstrual problems, different symptoms (somatic symptoms, night sweats, hot flushes, sleep difficulties, and musculoskeletal symptoms), and perceived consequences of menopausal symptoms. The characteristics of women differed in each multivariate model, except for "perceived consequences," which emerged as an independent factor in each model. Some factors—including social support and marital status—were notable for their absence in the final models. However, it should be noted that almost 80% of women in our study were married, and so we lacked power to examine the effect of marital status on resilience and vulnerability to hot flushes and night sweats.
There has been only one other community-based study, conducted in North America, that looked at this issue with similar depth. In contrast to that study, marital status and menopause status were not found in our study to be independently associated with resilience to hot flushes or night sweats, and poor self-rated health was not independently associated with vulnerability to either symptom. On the other hand, as in SWAN, women resilient to night sweats in our study were less likely to report sleep difficulties. Educational attainment was associated with vulnerability to hot flushes in SWAN, whereas this variable was associated with vulnerability to night sweats in our study. Differences between studies may be attributed to how educational attainment and self-rated health were assessed or to the influences of these factors in different countries. Further studies are needed to build the evidence base for factors associated with resilience or vulnerability to these and other menopausal symptoms.
A common factor in all models was the respondents' perceptions about the consequences of their symptom on their lives (one aspect of illness representations). Women who perceived their menopausal symptoms as having little consequence on their lives were more likely to be in the "resilient" group of either symptom, whereas those with perceptions of moderate or high consequences were more likely to be in the "vulnerable" group. The relationship between perceptions about the consequences of symptoms and vasomotor symptom experience has not been reported before. However, Women's perceptions about the consequences of their menopausal symptoms have been linked to help-seeking behavior; women reporting negative consequences have been found to be more likely to have visited their doctor recently than women with less negative illness perceptions. Our results suggest that illness representations may influence the impact of menopausal symptoms. Illness representations can be changed, suggesting a novel way for helping some women manage their menopausal symptoms, perhaps by designing psychoeducational programs for menopause management or by challenging negative perceptions about symptoms through group cognitive-behavioral therapy. Future studies of the menopausal transition should include measures of illness perceptions.
Methodological Strengths and Weaknesses
One of the main strengths of our study was its community base. Results from this study are more likely to be representative of women living in the community than those from studies using samples from healthcare settings such as gynecological clinics.
The analyses assume that frequency of symptoms and reported associated level of bother correctly differentiated women into those who, if asked directly, perceive themselves as "resilient" or "vulnerable" to the effects of these menopausal symptoms. A strength of our approach is that it moves away from groupings based on help-seeking behavior for symptoms. Consulting a healthcare professional may not always identify women who are resilient or vulnerable to their symptoms because a proportion of women bothered by their symptoms do not consult their doctor. Indeed, in our study, only one of the five different types of management strategies examined emerged as an independent variable in one of our multivariate models (not currently using psychological strategies and "resilience" to night sweats). This indicates, at best, only a weak relationship between help-seeking behavior and resilience or vulnerability to symptoms. Although work is needed to ascertain whether our assumptions about our ability to accurately allocate women to appropriate groups are correct, we believe that our study provides an important addition to the paucity of information about resilience and vulnerability during the menopausal transition.
The inclusion of information about illness perceptions was novel. We adapted the IPQ-R in our study after consulting with one of the individuals (Professor John Weinman) who devised the original measure. The adapted questionnaire included two items for each domain to restrict the size of the questionnaire and referred to menopausal symptoms, rather than illness, throughout because many women are unlikely to regard menopause as an illness. Although perceived consequences of menopausal symptoms emerged as a significant factor in all four models, none of the other psychological elements emerged as important due to the fact that we only included two items per domain in our version of the IPQ-R. Although further work is needed to test the validity of our IPQ-R, our observations show that psychological factors probably have an important role in symptom experience and thus should be measured.
Like others, we separately categorized women who had a hysterectomy as surgically postmenopausal. We did this even though women who had had the operation without concomitant bilateral oophorectomy will not have had an immediate cessation of ovarian function. In our study, among women with hysterectomy, roughly two thirds (65.4%) reported having had hysterectomy without bilateral oophorectomy. We have previously shown that women classified as surgically postmenopausal in this way had a similar frequency of hot flushes and night sweats and a similar level of associated bother as women postmenopausal from natural causes. This pattern occurred even though more surgically postmenopausal women were current users of HT than naturally postmenopausal women (21% vs 8%, respectively). It is probable, therefore, that a high proportion of women classified in our study as surgically postmenopausal had cessation of ovarian function. This concurs with other research reporting that hysterectomy with ovarian conservation in premenopausal women is associated with loss of ovarian function almost 4 years earlier than women without hysterectomy. Differences in the use of HT or other medications were allowed for in the statistical models and thus cannot explain our findings.
Our questionnaire response rate of 55% was less than desired but higher than those achieved by others in recent UK-based lifestyle surveys. Respondents were significantly more affluent than nonrespondents. Response bias, therefore, may have occurred. Some studies have found that women from lower socioeconomic groups report more menopausal symptoms than those from higher socioeconomic groups. Alternatively, women experiencing menopausal symptoms may have been more likely to complete the questionnaire than asymptomatic recipients. Although response bias might have influenced our estimates of the prevalence of hot flushes and night sweats, it is unclear how this might have affected our comparisons of resilience and vulnerability. A further limitation of our work was its cross-sectional nature. Although it was possible to investigate different patterns of associations, temporal sequences of cause and effect could not be established. Furthermore, although a wide range of sociodemographic, physical, and psychological variables were assessed in the questionnaire, others, such as diet, were not assessed to avoid response fatigue. Some of the associations, therefore, may be attributed to confounding. A further limitation was the investigation of numerous variables, making the study potentially prone to type I errors. On the other hand, some of the comparisons were based on a small number of women with the outcome of interest, so some factors related to resilience or vulnerability may not have been detected.