This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Benyamini, Y. Articles by Leventhal, H. Search for Related Content PubMed PubMed Citation Articles by Benyamini, Y. Articles by Leventhal, H. Related Collections Neuropsychology

Gender Differences in Processing Information for Making Self-Assessments of Health

From the Bob Shapell School of Social Work (Y.B.), Tel-Aviv University, Tel-Aviv, Israel; Robert Wood Johnson School of Medicine (E.A.L.), University of Medicine and Dentistry of New Jersey, Piscataway, NJ; and Institute for Health, Health Care Policy, and Aging Research and Department of Psychology (H.L.), Rutgers University, New Brunswick, NJ.

Address reprint requests to: Yael Benyamini, PhD, Bob Shapell School of Social Work, Tel-Aviv University, Tel-Aviv 69978, Israel. Email: benyael{at}post.tau.ac.il

	ABSTRACT

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
OBJECTIVE: This study proposes that women’s greater inclusiveness of various sources of information when making self-assessed health (SAH) judgments accounts for the finding that SAH is a weaker predictor of mortality in women than in men.

METHODS: Data from a sample of 830 elderly residents of a retirement community and a 5-year mortality follow-up study were used to examine the bases for women’s and men’s reports of negative affect (NA) and judgments of SAH. The degree to which each health-related measure accounts for the SAH-mortality association in each gender group was examined.

RESULTS: The findings support two possible explanations for the lower accuracy of SAH as a predictor of mortality among women: 1) In both men and women, NA is associated with poorer SAH, but in men, NA is more closely linked to serious disease in conjunction with other negative life events, whereas in women, NA reflects a wider range of factors not specific to serious disease. 2) Men’s SAH judgments reflect mainly serious, life-threatening disease (eg, heart disease), whereas women’s SAH judgments reflect both life-threatening and non–life-threatening disease (eg, joint diseases).

CONCLUSIONS: Women’s SAH judgments and NAs are based on a wider range of health-related and non–health-related factors than are men’s. This difference can explain gender differences in the accuracy of SAH judgments and may be related to other documented differences in women’s physical and mental health and illness behavior. The findings emphasize the need to study the bases of NA and other self-evaluations separately for women and men.

Key Words: self-assessments of health • gender differences • negativeaffect • physical functioning • mortality • attitude toward health

Abbreviations: CI = confidence interval; NA = negative affect; OR =odds ratio; PA = positive affect; SAH = self-assessed health.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Our primary hypothesis is that women, in comparison to men, make use of a broader set of inputs (ie, they are more inclusive) when making global judgments, such as SAH judgments. Differences in how women and men make SAH judgments are of more than passing interest, given that individuals assessing their health as poor are more likely to die sooner than others. For example, those assessing their health as poor are typically two to five times more likely to die within 2 to 13 years than those who give themselves high ratings (1, 2). This extraordinarily high risk for mortality persists after controlling for a variety of factors, including age and diagnosed medical conditions, and is often substantially higher for men than for women.

A wide range of psychological data suggest that women are more inclusive and/or attentive to contextual stimuli when making global, perceptual judgments (3–6). We propose that this greater inclusiveness also characterizes women’s ratings of their health and their evaluations of factors underlying their health ratings. Thus, although we expect both men’s and women’s SAH ratings to be lowered by factors clearly related to their physical health status such as serious health problems, and therefore to predict future mortality we expect women to take into account additional factors that are not directly related to physical health and mortality. One such factor is emotional distress, which reflects contextual factors unrelated to their physical health (and therefore unrelated to mortality). A second factor is the presence of non–life-threatening illnesses (eg, musculoskeletal problems). Because both sets of factors lower women’s global SAH ratings but are not expected to predict mortality, women’s SAH ratings will have a weaker relationship to mortality than do the SAH ratings of men.

	EMOTIONAL DISTRESS AND GENDER DIFFERENCES IN SAH

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Of the 17 studies reporting relationships between SAH and mortality separately for women and men, including the present one, only 5 found the association of SAH to mortality to be greater for women than for men (7–11), whereas 12 found the association to be greater for men (12–22). The association of SAH to mortality in these studies is reported as an OR, that is, the likelihood of mortality in participants rating their health as poor or fair in comparison to the likelihood of mortality in participants rating their health as excellent or very good. Most importantly, the effects of emotional distress were partialed out in 3 of the 5 studies finding a stronger relationship of SAH to mortality in women but in only 4 of the 12 studies finding a stronger association in men. Controlling for emotional distress in multivariate models predicting mortality would remove the "noise" from women’s SAH ratings and thus should increase the relationship of their SAH ratings to their mortality. These findings support our basic propositions and led to our first hypothesis.

	HYPOTHESIS 1

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
a) We hypothesized that controlling for NA would strengthen the SAH-mortality association in women and weaken it in men, because b) both women and men would downgrade their SAH ratings in response to NAs and c) women’s NAs would be based on a wide range of non–health-related as well as health-related factors, whereas men’s NAs would be more closely tied to their physical status.

There is evidence showing that both women and men who report higher levels of NA also report lower SAH status (23), and we attempted to replicate this finding (hypothesis 1b). Regarding the sources of NA, there are two types of evidence supporting the suggestion that women’s NAs are influenced by a wide range of non–health-related factors. First, although women are more likely than men to report high levels of negative feelings of anxiety and depression (23–26), the association of psychological distress to life-threatening illnesses seems to be stronger in men. For example, significant correlations were found between anxiety and illness in men, compared with negligible correlations in women (27, 28). Similarly, depression was found to be significantly associated with physical health in older men but not in older women (29, 30). A related finding showed that anxiety after coronary artery bypass surgery was focused on symptoms of immediate physical recovery in men and on who would care for them at home in women (31).

Second, in comparison to elderly men, elderly women are exposed to a greater number of factors that cause NA, such as anxiety and depression, and that are unrelated to life-threatening illness (32, 33). For example, although loss of a spouse has similar, negative effects on the physical and emotional health of both women and men (34), it is more frequently experienced by women. In addition, a larger percentage of women than men serve as caregivers to an unhealthy spouse, and women typically report greater burden and distress in the caregiver role than do men (35–37). Also, in comparison to married men and unmarried women, married women report the lowest levels of health and the highest levels of stress (38). In summary, if in comparison to men, women are more involved in the lives of those around them (39, 40) and are more distressed by events in these other lives, a phenomenon Helgeson (41) has labeled "unmitigated communion," these other sources of NA lower their SAH ratings, thus lowering the relationship of women’s SAH to their own mortality.

To test for women’s greater inclusiveness in the sources of their NA (hypothesis 1c), we examined the effect of non–health-related events on women’s and men’s NA. We tested the association of NA with measures of disease, recent negative (non–health-related) life events, and interactions among life events and disease. We expected women’s NA to be lower when they were experiencing more negative life events, regardless of their illness status, and men’s NA to be affected by these stressors mainly when they occurred in the presence of illness.

	MILD DISEASE AND GENDER DIFFERENCES IN SAH

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Many elderly people suffer from more than one disease and experience a variety of health changes over time. Although some of these changes are potentially life-threatening, others may be disturbing in terms of daily function yet are not life-threatening. Our second hypothesis parallels our first one and suggests a similarity between women’s inclusiveness in responding to a broadly based emotional distress when making judgments of health and the possibility that they will attend to a wider range of diseases, both serious and mild, when rating their health. In contrast, men are expected to focus once again on the more serious diseases, discounting the milder ones.

	HYPOTHESIS 2

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
We hypothesized that a) controlling for mild diseases would strengthen the SAH-mortality association in women and weaken it in men, because b) when making SAH ratings, women attend to and give similar weight to both life-threatening (eg, heart disease) and non–life-threatening (eg, arthritis) conditions, whereas men focus on and give greater weight to life-threatening illnesses and less weight to non–life-threatening illnesses.

We expected both women and men to report poorer SAH if they suffered from serious disease (42). If women’s judgments of their health are indeed more inclusive, this would also apply to their greater inclusiveness in using information bearing directly on their physical health and functioning. Data show that women are more likely to suffer from and to report symptoms of a larger number of bothersome, non–life-threatening chronic conditions, such as arthritis (24, 43), that impair health-related quality of life (44, 45). Thus, we expected women, in comparison to men, to more readily access information about a wider range of physical changes and functionally disturbing but non–life-threatening conditions (eg, arthritis) in making their SAH judgments. Women’s SAH judgments then would be based on both conditions that do and conditions that do not predict mortality; therefore, their SAH would be more weakly associated with mortality in comparison to the association of SAH ratings and mortality in men. Adjusting for mild diseases should diminish this effect (ie, should strengthen the SAH-mortality association) for women but should not affect this association for men.

It was not possible to test this hypothesis in prior studies because most studies used relatively brief lists of major diseases or a simple count of number of diseases without taking disease severity into account (eg, Refs. 17 and 19). The present data set included an extremely detailed report of participants’ medical histories weighted by disease severity, which allowed for separate tests of the relationship of mild and severe diseases to SAH and mortality.

As in prior research on global judgments of health and on gender differences in self-reports of health status, we controlled for demographic factors (age and education), functional status (eg, ability to perform activities of daily living), and social factors (social support) in testing our hypotheses, respecting the relationship of mild and severe indices of illnesses and psychological distress to SAH. Although we expected these additional factors to be associated with SAH, as has been reported elsewhere in the literature (46), we were not aware of any findings suggesting gender differences in the association of these factors with SAH; therefore, there are no specific hypotheses associated with them. We also tested hypothesis 1 using measures of PA. Because there are fewer data on the sources of PA among women and men and on its association with SAH, we did not propose a specific hypothesis.

	METHODS

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Sample
Of the 851 original participants in our study of community-dwelling older adults, 830 (497 women and 333 men) could be traced as alive or deceased at the end of the 5-year follow-up period (July 1991–July 1996). Mean age at baseline was 73 years (see Table 1 for demographic characteristics of the sample). Recruitment procedures are described elsewhere (47).

SAH and Mortality
SAH.
The lead question in each interview asked, "In general, would you say your health is excellent, very good, good, fair, or poor?" (1 = poor, 5 = excellent). The distribution of responses on this question was similar for both women and men; altogether, 2% rated their health as poor; 11%, fair; 38%, good; 31%, very good; and 17%, excellent.

Mortality follow-up.
By July 1996, 525 of the original 851 subjects (62%) were still taking part in annual interviews. Intensive efforts to trace the status of the 326 former participants revealed that 110 were deceased, 195 were alive, and mortality status could not be determined for 21, who were dropped from the analyses. Deaths were identified by obituaries in the community newspaper and verified by family members and/or the central office, which maintains a complete list of residents for legal purposes; care was taken to verify both first and last names and last known address. People who were alive were contacted directly; the status of those who had moved was verified by close family members.

Predictors of SAH
Physical health measures.
The respondent’s medical history and the presence of life-threatening and disabling disorders was assessed by a detailed review of 70 diseases from 19 illness categories, with open-ended probes used to identify additional illnesses in each category (eg, "Have you ever had any of the following heart or cardiovascular diseases?"). A series of prompts followed each category, ending with, for example, "Any other heart disease?". The categories included cardiovascular, lung, allergies/hay fever, infections, cancer, noncancerous tumors/cysts, stomach/intestinal, immune, nervous system, genital/urinary, joint/bone/muscle, kidney, blood, skin, diabetes, thyroid, eye, ear, and mental illnesses. Six physicians rated the severity of every disease reported (from a low of 1 to a high of 99 if it was extremely life-threatening; ratings were made without knowledge of the characteristics of the reporting participants). Eleven of these scale points (eg, 0, 10, 20, 30, ... 99) were given labels to be used as guidelines for the ratings; for example, the physicians agreed on assigning plantar warts a severity of 0, cirrhosis a severity of 50, and arterial aneurysm with rupture a severity of 99. The reliability coefficient (Cronbach’s ) for these six ratings across 427 disease codes was .97. The correlation between the rating by each judge and the average ratings of the other judges ranged from .89 to .93.

The weighted, total illness burden score (ie, sum of illnesses reported in each participant’s medical history weighted by the mean severity rating for that illness) was similar for women (mean ± SD, 162 ± 99) and men (171 ± 110). A board-certified geriatrician (E.A.L.) divided the diseases into a potentially life-threatening group (severity rating 30) and a non–life-threatening group (rating < 30). All cardiovascular diseases and cancers were in the first group, musculoskeletal conditions (eg, osteoarthritis and back pain) and other non–life-threatening but disruptive diseases (eg, irritable bowel syndrome and chronic sinusitis) were in the second. Two indices were computed: one representing diseases in the severe group, each weighted by its severity, and the other representing diseases in the mild group, also weighted by severity. In comparison to men, women had a lower score on the serious diseases index (t = 3.05, p < .01) and a higher score on the mild diseases index (t = 4.28, p < .001).

Other medical indicators (eg, recent illnesses and medication use) were obtained to test hypotheses unrelated to the present study but are not reported here. They were examined to ensure that they would not affect any of the data reported here.

Physical functioning.
Limitations in daily function were assessed with responses to four items ( = .71): 1) "Does your health limit the kinds or amounts of vigorous activities you can do, such as running, lifting heavy objects, or participating in strenuous sports or activities?" 2) "Does your health limit the kinds or amounts of moderate activities you can do, such as moving a table, carrying groceries, bending, or lifting?" 3) "Do you have any trouble walking one block, uphill, or up a few flights of stairs?" 4) "Do you have any trouble eating, dressing, bathing, or using the toilet?" The response scale was from 1 (not at all) to 5 (very much). A mean of the four items was computed. Responses on this four-item scale were highly correlated (r = .84) with responses on an 18-item scale of activity limitations (48), which was completed by 522 of our participants.

Two additional measures of functioning were assessed: Participants reported the total number of hours, on average, per week that they spent in strength-building activities, in strenuous sports, and in light sports (the three items were summed to form a single exercise score reflecting the number of hours per week that they engaged in exercise) and rated how physically active they were in general, aside from exercise (1 = not at all, 5 = very active). The three functioning measures were combined into one index by computing a mean of the standard scores on these measures: limitations in functioning (reversed), physical activity, and exercise. Standard scores were computed for each gender separately.

NA and PA.
Three scales were used to assess negative moods (depression, anxiety, and fatigue), and two scales were used to assess positive feelings (energy and happiness). Each mood scale used six items in a format asking, "How depressed are you usually?" (1 = not at all, 5 = very much). The items were adapted from a study that tested items from commonly used affect scales in an elderly sample (49). Cronbach’s values ranged from .88 to .93. Principal components factor analysis with varimax rotation yielded two factors, one loading on the NA measures and one loading on the PA measures. The measure of NA was the average of the standard scores (computed for each gender separately) of the three NA scores; the measure of PA was computed similarly using the two PA scores.

Social support.
Nine items ( = .81), formulated from Fischer’s (50) study, measuring emotional support (eg, "How often is there someone you can count on to listen to you when you need to talk?"), instrumental support (eg, "How often is there someone you can count on to help you do things that need to be done?"), and supportive presence (eg, "How often is there someone with whom you can get together for relaxation?"). All items were rated on a five-point scale (1 = never, 5 = always).

Life events.
A negative life events checklist included 26 events: 10 threat events (eg, marital problems or someone close was seriously ill), 15 loss events (eg, bereavement or someone close moved away), and 1 item asking whether they had any other serious problem not in the list (for classification of threat vs. loss events, see Ref. 51). In addition, participants were asked whether their spouse or partner had any serious problem. The time frame for questions about death events was the past 5 years and for all other events was the past 12 months. An item asking about recent disability was excluded from the life event measure to avoid confounding with the health measures. Altogether, 86.5% of the sample reported between 1 and 6 events (range, 0–11).

Statistical Analyses
Logistic regression (52) was used to predict mortality (because the majority of our sample was alive at the end of the 5-year follow-up period, we were not interested in survival times of those who were deceased but mainly in the distinction of alive vs. deceased, a binary outcome, for which logistic regression is a commonly used method; see Ref. 1). Age was controlled in all models. Separate logistic regression models were run for women and for men to assess the relationship of SAH and each disease and NA measure to 5-year mortality. To assess whether the gender difference in the relationship of SAH to mortality was due to women giving excessive weight to mild medical conditions or to non–health-related sources of NAs, the same models were run with SAH added to each model (ie, the models for severe diseases, mild diseases, and NA). In the logistic regression models, SAH was grouped into three categories (poor/fair, good, and very good/excellent), and all other measures were recoded into three categories, representing three roughly equal thirds for each gender group. Each factor was coded such that the high end was defined as the direction expected to predict mortality according to its correlation with SAH; the percentage deceased at the high relative to that at the low end defined the OR (ie, the relationship of the variable to mortality).

Multiple linear regression models were conducted separately for women and men to test the relationship of disease and life events to NA and of disease and affect to SAH.

	RESULTS

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Gender Difference in the Relationship of SAH to Mortality
Mean ratings and distributions of SAH were similar for women (mean ± SD, 3.6 ± 1.0) and men (3.5 ± 0.9), in correspondence with figures reported in national surveys (eg, Ref. 43). More men (19%) than women (10%) died during the 5-year mortality follow-up period (see Table 2). Men’s SAH was a strong predictor of future mortality: Those who rated their health as fair or poor were 4.8 times more likely to die than those who rated their health as very good or excellent (adjusted for age; 95% CI, 2.1–11.1, p < .001). Women who rated their health as fair or poor were only 2.2 times more likely to die than women who rated their health as very good or excellent (adjusted for age; 95% CI, 1.0–5.1, p = .06).

Hypothesis 1: NA and Gender Difference in SAH-Mortality Association
According to the first part of our hypothesis (1a), we expected the OR describing the relationship of SAH to mortality to decline when factors that are correlated with SAH and related to the seriousness of respondents’ health status were introduced into the model. Conversely, we expected the OR to increase when factors correlated with SAH but unrelated to the seriousness of respondents’ health status were introduced. Table 3 presents the results of the logistic regression analyses, comparing the adjusted likelihood of mortality for participants in the "poor" third of each measure, compared with those in the "positive" third of the scale. To simplify the presentation of the mortality data, we do not report the OR for the middle category; this category was included in each analysis, and its ORs were typically lower than those for the high end. Adjustments for age were made in all analyses. Although there are no specific hypotheses about gender differences in the effects of physical functioning and PA, similar models were run for these factors. Social support and education were dropped from these analyses because they were unrelated to mortality in our sample.

For PAs, the picture is partially similar to that for NAs. Low PAs are related to mortality in men (OR = 2.9, p < .01) but not in women (OR = 1.7, NS). Controlling for PA has no effect on the OR relating SAH to mortality in women and lowers it in men (from 4.8 to 3.9). The pattern for function is similar to that reported in most studies and contrasts sharply with that for NA. Poor function is predictive of mortality in women (OR = 4.3) and men (OR = 3.2). In addition, because poor function is related to lower SAH, entering function in the model lowers the association of SAH to mortality in both women (OR decreases from 2.2 to 1.2) and men (from 4.8 to 3.3).

As suggested in the second part of our hypothesis (1b), both women and men downgrade their SAH in face of NAs. Table 4 presents the zero-order associations of each of our measures with SAH and the unique contribution of each measure to SAH when they are entered together in linear regression models, separately for women and men. Note that nearly all of the zero-order correlations between SAH and the other control variables are essentially the same for women and men, with the exception of education (women, r = .16; men, r = -.04) and mild diseases (discussed below). High levels of NA have robust and virtually identical relationships to lower levels of SAH in women (r = -.38) and men (r = -.39). Thus, the contribution of NA to the gender difference in the association of SAH to mortality is not due to a gender difference in the impact of NA on SAH but to a gender difference in the source of these NAs. p < .05; ** p < .01; *** p < .001.

View larger version (14K): [in this window] [in a new window]   Fig. 1. NA as a function of burden of serious diseases (low vs. high) and number of recent negative life events among women and men (N = 497 and 333, respectively; the serious diseases measure is dichotomized).

Hypothesis 2: Serious and Mild Diseases and Gender Difference in SAH-Mortality Association
As expected, serious illnesses were similarly and negatively related to SAH for women (r = -.38) and men (r = -.29; difference between correlations, NS). These correlations are similar to those found in many other studies that examined the association between SAH and diagnoses (47). Consistent with our expectations, mild illnesses were significantly negatively related to SAH for women (r = -.31) but only slightly negatively related to SAH for men (r = -.11; difference between correlations, p < .0001; Table 4). These findings lend support to the proposal that women are more inclusive in performing self-evaluations.

High levels of serious disease are similarly related to mortality in both women and men and, as expected, lower the OR for SAH to mortality in both genders (Table 3). Mild diseases are inversely but not significantly related to mortality in both genders (OR = 0.7), again confirming our assumption that these are indeed mild, non–life-threatening diseases. Entering mild diseases into the model with SAH raises the OR for SAH in women (from 2.2 to 3.0), as predicted. Unexpectedly, it also raises the OR for SAH in men, although to a lesser extent (from 4.8 to 5.2). Thus, although the correlation of mild disease to women’s ratings of SAH (r = -.31) lowers the OR for SAH to mortality, this effect is not much larger than the one that occurs in men; therefore, it does not account for the gender difference in the relationship of SAH to mortality, disproving hypothesis 2a. Although the findings confirm the second part of this hypothesis (2b), which states that women making SAH ratings give similar weight to both life-threatening and non–life-threatening conditions, whereas men give greater weight to life-threatening conditions, this does not help to explain the gender difference in the SAH-mortality association.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
Prior data showing a stronger relationship of SAH to mortality in men than in women (eg, Ref. 17) suggested that women and men may differ in how they use health-related information in making global judgments of their health status (19). We hypothesized that gender differences in strategies for acquiring information and making global judgments of health would appear in two areas: 1) NA, which influences SAH in both genders but may draw on different sources among women and men, and 2) disruptive but mild diseases, which are unrelated to future mortality yet affect women’s, but not men’s, SAH.

Our first hypothesis, regarding the affective domain, was that the genders would not differ in the degree to which their SAH judgments were lowered by NAs but in the degree to which NAs were related to future mortality. Specifically, we expected that men’s NAs would mainly reflect serious health problems or other negative life events occurring in conjunction with poor health and that women’s NAs would reflect both health-related and non–health-related life events. Moreover, these differences in sources of NA would also determine the effect that controlling for NAs has on the SAH-mortality association, lowering it in men and increasing it in women. The data supported this hypothesis: High levels of NA were related to lower SAH in both men and women and were related to higher risk of mortality in men but lower risk of mortality in women. The positive association of NA with mortality in men reflected the degree to which their NA was linked to serious diseases. Nondisease life stressors increased NA only in men who had serious diseases, indicating perhaps limitations in their resources to deal with life stresses when also dealing with their ailments. In contrast, women’s NA was also affected by other negative life events, regardless of disease status. Married women were not only troubled by their own problems but also more distressed if their spouse had had any serious problem in the past year. These findings are consistent with the suggestion that women’s evaluations are sensitive to the overall quantity of their negative feelings rather than to their specific source.

Our second hypothesis stated that men would attend mainly to serious illness and that women would attend to both serious and disruptive but non–life-threatening illness. Our detailed questioning about medical history enabled us to develop separate illness burden indices for mild and serious diseases, allowing us to examine the relationship of both severe and mild illness to both SAH and mortality. In contrast, illness checklists in most studies have focused on major diseases and lack information about milder conditions, as emphasized by Jylhä et al. (19). The data supported our expectation: Men’s SAH was associated with serious but not mild diseases, and women’s SAH was associated with both serious and mild diseases. As expected, women were significantly more likely to include mild diseases in their global health judgments, and doing so may have contributed to the weaker relationship of their SAH to mortality. When mild diseases were controlled for in the prediction of women’s mortality from their SAH, the effect of SAH increased. However, in contrast with our expectation, this increase occurred in men also, although to a lesser extent. Thus, although the differential impact of mild diseases on women’s and men’s SAH is in line with our general hypothesis regarding the greater inclusiveness of women’s self-evaluations, it does not account for the gender difference in the relationship of SAH to mortality. In general, our findings regarding the sources of NA and the impact that mild diseases have on SAH are consistent with Jylhä et al.’s suggestion (19) that people "are also likely to take into account dimensions of health that are not directly related to mortality. If these are more important constituents of self-ratings in women than in men, the relation of [their] self-rated health with mortality will, consequently, be weaker."

Why is affect sensitive to a wider range of factors among women, or why is it more sensitive to serious disease among men? First, there are differences in the life circumstances of older women and men. Men of this culture typically marry younger women. Men are also likely to die earlier than women their age. The combination of these factors results in older women being much more likely to be giving care to an ailing husband or to be widowed. Although marriage is protective of health over the life span (53), this effect is more pronounced in men: Unmarried men display higher mortality from all causes than married men (54). Social support seems to protect men from many adverse experiences and is largely provided or mediated by their spouses. In contrast, older women living with their spouse are more likely to be giving care, thus carrying a higher burden, receiving less spousal support, and being more limited in socializing with others. In our sample, 53% of the women, compared with only 15% of the men, were living alone. The differing circumstances of elderly men (ie, that most live with spouses) may be responsible for the lesser impact of life events in the absence of serious diseases.

Second, women’s response styles may also be responsible for the tendency of a variety of events to influence health perceptions: Women are more likely to have a ruminative response style, which is related to more severe periods of depression (55, 56). If women ruminate more when depressed, and men are better able to distract, then any source of depression, even a non–health-related source, may more easily infiltrate women’s self-perceptions of health (or other domains).

Third, the differential sensitivity of affect to life circumstances in the two genders may be related to gender differences across the life span. Affects serve a regulatory function by indicating readiness for action (57, 58). As such, from an evolutionary point of view, it may be adaptive for women, as the traditional family caregivers, to be responsive to all surrounding events and to any disruptive health problem, life-threatening or not. In contrast, men, the traditional hunters, may be "programmed" to attend to health conditions that reduce their ability to manage external threats to self and family and to limit their emotional reactions and physical resources in response to minor health-related or non–health-related events. This suggestion is in line with research showing that women’s depression is more closely linked to negative events that happen to important others (39, 40).

It is also possible that these findings are not universal but are bound to the cultural context. In our case, the context is a culture dominated by isolated nuclear monogamous families. In such a culture, when residing in an independent-living community, loss of health and function is highly stressful. In cultures where people live with extended families, health-related loss of independence may have a lesser effect on men’s psychological well-being.

Given that the range of social as well as medical factors affecting women’s NA is wider than that affecting men’s NA, we can account for the stronger relationship of NA to mortality among men. Among women, NA significantly lowered mortality risk. Because a similar effect was observed in the National Health and Nutrition Examination Survey (NHANES), which involved a large, representative sample and yielded higher mortality for women scoring high on well-being and for men scoring low on well-being (Idler and Leventhal, unpublished manuscript), we do not believe the effect is due to chance or the nonrepresentative nature of our sample. It may be that, paradoxically, NA among women indicates higher social involvement, which, in turn, is related to a lower risk of mortality.

There are several limitations to our study. Our sample size is not large for assessing mortality risk. Such risk is typically examined in epidemiological studies using samples of thousands of people. In addition, our sample is overwhelmingly white and educated: Its nonrepresentativeness may limit some generalizations. All our participants volunteered for this study, and we have no way of assessing whether the motivations of women volunteering for a study on health and aging are different from those of men. Therefore, it is clear that replicating these findings in additional samples, representative or at least different from the one presented here, could contribute to the ability to generalize these findings and to determine the social and cultural conditions under which they apply. Examining the relationship of SAH to various outcomes in different contexts may be an effective way of understanding how SAH is affected by different life situations and how these life situations affect gender differences in self-evaluations.

There is a positive side to the characteristics of our sample: their willingness to volunteer and cooperate with a very long interview, covering in great detail the sensitive issues of physical and mental health. Collecting such detailed information from a larger and more representative sample would be much more difficult.

The broad connotation of health, as can be seen from the variety of factors that affect subjective perceptions of health, has important practical implications. The interactions of social-psychological and physiological factors in forming SAH may lead people to form biases in estimating the health relevance of life events and somatic sensations. Non–health-related distress can lead people to seek health care for nonmedical problems, or, conversely, they may misattribute ambiguous symptoms caused by physical illness to emotional stress if they are facing stressful life events (59). Indeed, research shows that such difficulties affect physicians’ assessments of men’s and women’s health complaints: Although the difference in actual complaints is often not large (60), women are more likely to be diagnosed with mental problems (61), and men are more likely to be diagnosed with cardiovascular problems (62, 63).

Our findings suggest implications for psychological research on gender differences in the factors that influence assessments of health and affect. First, it is important to consider the social and cultural contexts within which women and men make these assessments. Second, in studying the meaning of health for older people, it may be worthwhile to look at large groups (eg, men vs. women, young-old vs. old-old, and those living alone vs. those living with a spouse) and ask what health means to them and how the context changes the relevance of various factors to them. Third, it is important to search for gender differences when studying the sources of affect. Much of the classic research on life experiences (eg, losses and widowhood) leading to depression was conducted in women (eg, 64). Finally, it is important to note that although non–health-related distress may function as "noise" for judgments of physical health status or predictions of mortality, it may be an important source of information for other judgments (eg, quality of life). Thus, although women’s SAH is less "accurate" than men’s in one respect (predicting mortality), it may be as accurate, or more accurate, in relation to other criteria. In any case, the data suggest that, at least for older women, involvement with others and NA are an integral part of living and do not necessarily predict death, a serious and definitive health outcome.

	ACKNOWLEDGMENTS

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 
This research was supported by Grant AG03501 from the National Institute on Aging. We express our thanks to Ellen Idler for her insightful comments. We also thank Susan Brownlee, Michael Diefenbach, Linda Patrick-Miller, Chantal Robitaille, and Frances Sisack for their assistance on various aspects of the research reported in this article.

Received for publication February 12, 1999.

Revision received September 28, 1999.

	REFERENCES

TOP ABSTRACT INTRODUCTION EMOTIONAL DISTRESS AND GENDER... HYPOTHESIS 1 MILD DISEASE AND GENDER... HYPOTHESIS 2 METHODS RESULTS DISCUSSION ACKNOWLEDGMENTS REFERENCES

 

1. Idler EL, Benyamini Y. Self-ratings of health and mortality: a review of 27 community studies. J Health Soc Behav 1997; 38: 21–37.[Medline]
2. Benyamini Y, Idler EL. Community studies reporting association between self-rated health and mortality: additional studies, 1995–1998. Res Aging 1999; 21: 392–401.[Free Full Text]
3. Roberts TA, Pennebaker JW. Gender differences in perceiving internal state: toward a his-and-hers model of perceptual cue use. Adv Exp Soc Psychol 1995; 27: 143–75.
4. Bauermeister M, Wapner S, Weiner H. Sex differences in the perception of apparent verticality and apparent body position under conditions of body tilt. J Pers 1963; 31: 394–407.[Medline]
5. Leventhal H, Mace W. The effect of laughter on evaluation of a slapstick movie. J Pers 1970; 38: 16–30.[Medline]
6. Cupchik GC, Leventhal H. Consistency between expressive behavior and the evaluation of humorous stimuli: the role of sex and self-observation. J Pers Soc Psychol 1974; 30: 429–42.
7. Grant MD, Piotrowski ZH, Chappell R. Self-reported health and survival in the Longitudinal Study of Aging, 1984–1986. J Clin Epidemiol 1995; 48: 375–87.[Medline]
8. McCallum J, Shadbolt B, Wang D. Self-rated health and survival: a 7-year follow-up study of Australian elderly. Am J Public Health 1994; 84: 1100–5.[Abstract/Free Full Text]
9. Miilunpalo S, Vuori I, Oja P, Pasanen M, Urponen H. Self-rated health status as a health measure: the predictive value of self-reported health status on the use of physician services and on mortality in the working-age population. J Clin Epidemiol 1997; 50: 517–28.[Medline]
10. Simons, LA, McCallum J, Friedlander Y, Simons J. Predictors of mortality in the prospective Dubbo study of Australian elderly. Aust NZ J Med 1996; 26: 40–8.[Medline]
11. Wolinsky FD, Johnson RJ. Perceived health status and mortality among older men and women. J Gerontol 1992; 47: S304–12.[Medline]
12. Goldman N, Korenman S, Weinstein R. Marital status and health among the elderly. Soc Sci Med 1995; 40: 1717–30.
13. Hays JC, Schoenfeld D, Blazer DG, Gold DT. Global self-ratings of health and mortality hazard in the North Carolina Piedmont. J Clin Epidemiol 1996; 49: 969–79.[Medline]
14. Franks P, Gold MR, Clancy CM. Use of care and subsequent mortality: the importance of gender. Health Serv Res 1996; 31: 347–63.[Medline]
15. Idler EL, Angel RJ. Self-rated health and mortality in the NHANES-I Epidemiologic Follow-up Study. Am J Public Health 1990; 80: 446–52.[Abstract/Free Full Text]
16. Idler EL, Kasl SV, Lemke JH. Self-evaluated health and mortality among the elderly in New Haven, Connecticut, and Iowa and Washington counties, Iowa, 1982–1986. Am J Epidemiol 1990; 131: 91–103.[Abstract/Free Full Text]
17. Idler EL, Kasl S. Health perceptions and survival: do global evaluations of health status really predict mortality? J Gerontol 1991; 46: S55–65.[Medline]
18. Idler EL, Russell LB, Davis D. Survival, functional limitations, and self-ratings of health in the NHANES EpidemiologicFollow-up Study, 1992. Am J Epidemiol. In press 2000.
19. Jylhä M, Guralnik JM, Ferrucci L, Jokela J, Heikkinen E. Is self-rated health comparable across cultures and genders? J Gerontol 1998; 53B: S144–52.
20. Kaplan G, Barell V, Lusky A. Subjective state of health and survival in elderly adults. J Gerontol 1988; 43: S114–20.[Medline]
21. Krzyzanowski M, Wysocki M. The relation of thirteen-year mortality to ventilatory impairment and other respiratory symptoms: the Cracow Study. Int J Epidemiol 1986; 15: 56–64.[Abstract/Free Full Text]
22. van Doorn C, Kasl SV. Can parental longevity and self-rated life expectancy predict mortality among older persons? Results from an Australian cohort. J Gerontol 1998; 53B: S28–34.
23. Schulz R, Mittelmark M, Kronmal R, Polak JF, Hirsch CH, German P, Bookwala J. Predictors of perceived health status in elderly men and women. J Aging Health 1994; 6: 419–47.[Abstract/Free Full Text]
24. Deeg DJH, Westendorp-de Serière M, editors. Autonomy and well-being in the aging population. I. Report from the Longitudinal Aging Study Amsterdam, 1992–1993. Amsterdam: VU Uitgeverij; 1994.
25. Diener E, Sandvik E, Larsen RJ. Age and sex effects for emotional intensity. Dev Psychol 1985; 21: 542–6.
26. Culbertson FM. Depression and gender. Am Psychol 1997; 52: 25–31.[Medline]
27. Cochran CD, Hale WD. Gender differences in the relationship between health and psychological distress in the elderly. Clin Gerontol 1984; 3: 62–4.
28. Hale WE, Cochran CD. Gender differences in health attitudes among the elderly. Clin Gerontol 1986; 4: 23–7.
29. Beekman ATF, Kriegsman DMW, Deeg DJH, van Tilburg W. The association of physical health and depressive symptoms in the older population: age and sex differences. Soc Psychiatry Psychiatr Epidemiol 1995; 30: 32–8.[Medline]
30. Cadoret RJ, Widmer RB. The development of depressive symptoms in elderly following onset of severe physical illness. J Fam Pract 1988; 27: 71–6.[Medline]
31. Moore SM. A comparison of women’s and men’s symptoms during home recovery after coronary artery bypass surgery. Heart Lung 1995; 24: 495–501.[Medline]
32. Bebbington PE, Brugha T, MacCarthy B, Potter J, Sturt E, Wykes T, Katz R, McGuffin P. The Camberwell Collaborative Depression Study. I. Depressed probands: adversity and the form of depression. Br J Psychiatry 1988;152:754–65.
33. Cooke DJ, Hole DJ. The aetiological importance of stressful life events. Br J Psychiatry 1983; 143: 397–400.[Abstract/Free Full Text]
34. Thompson LW, Breckenridge JN, Gallagher D, Peterson J. Effects of bereavement on self-perceptions of physical health in elderly widows and widowers. J Gerontol 1984; 39: 309–14.[Medline]
35. Barusch AS, Spaid WM. Gender differences in care giving: why do wives report greater burden? Gerontologist 1989; 29: 667–76.[Abstract]
36. Borden W, Berlin S. Gender, coping, and psychological well-being in spouses of older adults with chronic dementia. Am J Orthopsychiatry 1990; 60: 603–10.[Medline]
37. Wallsten SM. Effects of caregiving, gender, and race on the health, mutuality, and social supports of older couples. J Aging Health 2000; 12: 90–111.[Abstract/Free Full Text]
38. Preston DB, Dellasega C. Elderly women and stress: does marriage make a difference? J Gerontol Nurs 1990; 16: 26–32.
39. Kessler RC, McLeod JD. Sex differences in vulnerability to undesirable life events. Am Sociol Rev 1984; 49: 620–31.
40. Turner RJ, Avison WR. Gender and depression: assessing exposure and vulnerability to life events in a chronically strained population. J Nerv Ment Dis 1989; 177: 443–55.[Medline]
41. Helgeson VS. Relation of agency and communion to well-being: evidence and potential explanations. Psychol Bull 1994; 116: 412–28.
42. Moum T. Self-assessed health among Norwegian adults. Soc Sci Med 1992; 35: 935–47.
43. Verbrugge LM. Gender and health: an update on hypotheses and evidence. J Health Soc Behav 1985; 26: 156–82.[Medline]
44. Dunlop DD, Hughes SL, Manheim LM. Disability in activities of daily living: patterns of change and a hierarchy of disability. Am J Public Health 1997; 87: 378–83.[Abstract/Free Full Text]
45. Kaplan RM, Anderson JP, Wingard DL. Gender differences in health-related quality of life. Health Psychol 1991; 10: 86–93.[Medline]
46. Bjorner JB, Kristensen TS, Orth-Gomér K, Tibblin G, Sullivan M, Westerholm P. Self-rated health: a useful concept in research, prevention, and clinical medicine. Stockholm: Swedish Council for Planning and Coordination of Research; 1996.
47. Benyamini YB, Leventhal EA, Leventhal H. Self-assessments of health: what do people know that predicts their mortality? Res Aging 1999; 21: 477–500.[Abstract/Free Full Text]
48. Johnson RJ, Wolinsky FD. Gender, race and health: the structure of health status among older adults. Gerontologist 1994; 34: 24–35.[Abstract]
49. Usala PD, Hertzog C. Measurement of affective states in adults: evaluation of an adjective rating scale instrument. Res Aging 1989; 11: 403–26.[Abstract]
50. Fischer CS. To dwell among friends: personal networks in town and city. Chicago: University of Chicago Press; 1982.
51. Finlay-Jones R, Brown GW. Types of stressful life events and the onset of anxiety and depressive disorders. Psychol Med 1981; 11: 803–15.[Medline]
52. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons; 1989.
53. Ren XS. Marital status and quality of relationships: the impact on health perception. Soc Sci Med 1997; 44: 241–9.
54. Rogers RG. Marriage, sex, and mortality. J Marriage Fam 1995; 57: 515–26.
55. Nolen-Hoeksema S. Responses to depression and their effects on the duration of depressive episodes. J Abnorm Psychol 1991; 100: 569–82.[Medline]
56. Nolen-Hoeksema S, Morrow J, Fredrickson BL. Response styles and the duration of episodes of depressed mood. J Abnorm Psychol 1993; 102: 20–8.[Medline]
57. Frijda NH. The emotions. Cambridge, England, UK: Cambridge University Press; 1986.
58. Leventhal H, Mosbach PA. The perceptual-motor theory of emotion. In: Cacioppo J, Petty R, editors. Social psychophysiology. New York: Guilford Press; 1983. p. 353–88.
59. Cameron LC, Leventhal EA, Leventhal H. Seeking medical care in response to symptoms and life stress. Psychosom Med 1995; 57: 37–47.[Abstract/Free Full Text]
60. Redman S, Webb GR, Hennrikus DJ, Gordon JJ, Sanson-Fisher RW. The effects of gender on diagnosis of psychological disturbance. J Behav Med 1991; 14: 527–40.[Medline]
61. Floyd BJ. Problems in accurate medical diagnosis of depression in female patients. Soc Sci Med 1997; 44: 403–12.
62. Beery TA. Gender bias in the diagnosis and treatment of coronary artery disease. Heart Lung 1995; 24: 427–35.[Medline]
63. Sayer GP, Britt H. Sex differences in morbidity: a case of discrimination in general practice. Soc Sci Med 1996; 42: 257–64.
64. Brown GW, Harris T. Social origins of depression: a study of psychiatric disorders in women. London: Tavistock; 1978.

This article has been cited by other articles:

				M. Huisman, F. van Lenthe, and J. Mackenbach The predictive ability of self-assessed health for mortality in different educational groups Int. J. Epidemiol., December 1, 2007; 36(6): 1207 - 1213. [Abstract] [Full Text] [PDF]

				S. Shooshtari, V. Menec, and R. Tate Comparing Predictors of Positive and Negative Self-Rated Health Between Younger (25-54) and Older (55+) Canadian Adults: A Longitudinal Study of Well-Being Research on Aging, November 1, 2007; 29(6): 512 - 554. [Abstract] [PDF]

				S. Cohen, C. M. Alper, W. J. Doyle, J. J. Treanor, and R. B. Turner Positive Emotional Style Predicts Resistance to Illness After Experimental Exposure to Rhinovirus or Influenza A Virus Psychosom Med, November 1, 2006; 68(6): 809 - 815. [Abstract] [Full Text] [PDF]

				M. Emmelin, A. N. Nafziger, H. Stenlund, L. Weinehall, and S. Wall Cardiovascular risk factor burden has a stronger association with self-rated poor health in adults in the US than in Sweden, especially for the lower educated Scand J Public Health, March 1, 2006; 34(2): 140 - 149. [Abstract] [PDF]

				M. Westin and R. Westerling Health and healthcare utilization among single mothers and single fathers in Sweden Scand J Public Health, March 1, 2006; 34(2): 182 - 189. [Abstract] [PDF]

				S. D. Barger Do Psychological Characteristics Explain Socioeconomic Stratification of Self-rated Health? J Health Psychol, January 1, 2006; 11(1): 21 - 35. [Abstract] [PDF]

				A. Nyamathi, J. Berg, T. Jones, and B. Leake Predictors of Perceived Health Status of Tuberculosis-Infected Homeless West J Nurs Res, November 1, 2005; 27(7): 896 - 910. [Abstract] [PDF]

S. Macintyre, L. McKay, and A. Ellaway Who is more likely to experience common disorders: men, women, or both equally? Lay perceptions in the West of Scotland Int. J. Epidemiol., April 1, 2005; 34(2): 461 - 466. [Abstract] [Full Text]