The potential effects of making a marital transition on subsequent physical activity were evaluated across a ten-year period in a population-based sample of 302 women and 256 men ages 25 to 75 years. Subjects completed a structured interview at five timepoints throughout the ten-year period during which they reported on their physical activity level as well as marital status. The transition from a married to a single state did not affect physical activity relative to remaining married when analyses of either slopes or mean values were used, In contrast, the transition from a single to a married state resulted in significant positive changes in physical activity relative to remaining single throughout the study period when physical activity slopes, though not means, were compared. The results suggest that marriage may potentially set the stage for natural changes in physical activity that could be capitalized on through appropriate intervention, but additional research is needed to verify this in light of the inconsi~ent pattern of findings.
(Ann Behav Med 1998, 20(2):64-69) INTRODUCTION
The health effects of regular physical activity are now well-established for a variety of chronic diseases and conditions, including coronary heart disease, stroke, some forms of cancer, non-insulin-dependent diabetes mellitus, osteoporosis, and obesity (1). Despite these known benefits, the majority of the U.S. population remains underactive (2). Identifying strategies for facilitating sustained exercise participation at a level sufficient to provide such health benefits across a person’s lifetime constitutes a major public health challenge (2,3). Increasingly, health behavior change has been conceptualized as a series of psychological processes or stages (4).
A growing amount of evidence demonstrates how stages of motivational readiness for change may influence initial adoption as well as ongoing maintenance of physical activity (5). Less systematic attention has been paid to how stages of human development may influence physical activity adoption and maintenance. Specifically, a life span perspective encourages an increased focus on periods and transitions in life when behaviors such as physical activity may
i Preparation of this manuscript was supported in part by PHS grant #IlL 21906 (Five-City Project) from the National Heart, Lung, and Blood Institute and PHS grant #AG 12358 from the National Institute on Aging. 2 The authors would like to thank Helena C. Kraemer, Ph.D. for her helpful suggestions related to statistical analysis and Stephen P. Fortmann, M.D. for his helpful comments on earlier drafts of this manuscript.
Reprint Address: A. C. King, Ph.D., Stanford University School of Medicine, 730 Welch Road, Suite B, Palo Alto, CA 94304-1583. 9 1998 by The Society of Behavioral Medicine.
be significantly altered (6-8). Examples of such transitions include puberty, changes in work status and employment, parenthood, and changes in marital status (7). A transition or milestone approach to prevention (i.e. administering services to persons who have reached a particular, predefined point or stressful life event ) often has been applied to acute, time-limited illnesses or interventions, as in the case of preparatory programs for surgical and dental procedures (10). Conceivably, such an approach is also applicable to those transition points in a person’s life, such as puberty, marriage, childbearing, retirement, and menopause, when health behaviors influencing more chronic diseases and conditions may be especially affected (11). Such health behaviors include cigarette smoking, alcohol use, dietary habits, and physical activity. Marital status, in particular, has been shown to have important effects on health outcomes. Mortality rates are generally lower for married than unmarried persons, with the protective effects of marriage especially notable for men (12,13).
Furthermore, loss of the marital relationship, either through widowhood or divorce, contributes to a decline in physical health and an increase in mortality in both sexes, with the relationship again appearing to be particularly strong for men (14-18). One potential explanation for these relationships resides with a social integration/social control model that suggests that social relationships, including marital status, may exert an effect on a person’s desire or inclination to engage in detrimental or healthful behaviors (19). For example, the shift from being married to unmarried may be deleterious to health, at least in part, because there is no longer a partner available to monitor or support one’s health behavior patterns (20,21). In contrast, marriage may provide a support system conducive to the initiation or maintenance of healthful behaviors (21-25). Such transition periods may provide a potentially useful window of opportunity with respect to intervention.
If, for instance, a relationship was found between marital disruption (e.g. divorce, widowhood) and decreases in physical activity, these population segments could be targeted with interventions aimed at preventing the decline in physical activity. Alternatively, if getting married is found to be a potential enabling factor for increases in physical activity, further investigations of the mechanisms through which it exerts an influence (e.g. enhanced health information exchange, ongoing social support, enhanced financial or material resources) could potentially strengthen subsequent physical activity interventions.
Despite the potential significance of transitions in marital status for health behavior change, few attempts have been made to study this relationship systematically. In a national two-wave panel survey conducted in 1986 and 1989, shifting from a single to a married state appeared to have little effect on a variety of health behaviors, including physical activity, in women and men (20). In contrast, shifting from a married to an unmarried state (through either divorce or widowhood) was associated with higher tobacco and alcohol consumption and greater reported weight loss in men and reduced hours of sleep per night and greater reported weight loss in women. Despite the weight loss among both sexes, reported physical activity levels did not change. However, the three-year duration of this study limited the number of data points evaluated as well as the type of analysis that could be undertaken (i.e. a posttest comparison only, adjusting for initial physical activity levels). These factors may have reduced the study’s power to detect significant differences for those persons who made a marital transition.
In one of the few other studies to systematically evaluate the effects of changes in marital status on health behaviors, U.S. men, sampled via the Health and Nutrition Examination Survey, who either became married between baseline and ten-year follow-up or remained unmarried had a higher risk of weight gain than consistently married men, whereas men ending a marriage had a higher risk of weight loss (26). While changes in physical activity could have been influencing the reported changes in body weight, physical activity was not directly measured in this study. In addition, marital status and weight change were analyzed only twice during the ten years. A subsequent analysis of U.S. women from this data set found a similar risk of weight gain during the ten-year follow-up for women entering marriage during the follow-up period (27).
The current study sought to expand this literature by evaluating the relationship between marital transitions and physical activity reported on five occasions across a ten-year period in a population-based sample. The extended time period coupled with multiple measurement points in each marital state provided the potential for additional power in evaluating these relationships. The primary question of interest concerned the effect of making a marital transition on reported physical activity levels relative to remaining in one marital state (either single or married) across the decade. We were particularly interested in evaluating whether changes in marital status could set the stage for increases or decreases in physical activity levels over a specified time period. METHODS Subjects
* = Dependent variable of interest: Physical activity slope from second period minus physical activity slope from first period. “~ = First and second periods defined by two- or three-point physical activity slopes, based on the proportion of subjects with two- versus three-point slopes for each time period in each of the two marital transition groups (i.e. married-to-single and single-to-married groups). w = “Inactive” defined as a rating of 1, 2, or 3 on the 7-point global physical activity item (“How do you rate the physical activity you are now getting compared to others of your same sex and age? Think about both your leisure and work activities.”).
Subjects had participated as part of the cohort survey sample of the Stanford Five-City Project (28). The design and methods of the Five-City Project have been described in detail elsewhere (28,29) and will be only briefly described here. Households in four central California cities (two intervention and two control) were randomly selected to participate in a series of health surveys across the ten-year study period using commercial household directories. All individuals 12 to 75 years of age in the selected households were eligible to participate in the original Five-City Project and were invited to local community clinics, where measures were obtained by trained nurses and interviewers.
Repeated surveys of those participants who participated at baseline were conducted at 17, 39, 60, and 110 months. Individuals aged 25 years and older who participated in all five cohort surveys were included in the current analyses. The multiple risk factor educational intervention spanned six years and incorporated the use of multimedia (i.e. television, radio, and print materials) with direct education. Approximately 8% of the total educational exposure episodes delivered to the intervention cities were focused on physical activity promotion (30). Overall, the intervention resulted in few discernable effects on reported physical activity levels in the cohort sample (30).
Measures For the analyses reported here, physical activity was measured with an interviewer-administered item, “How do you rate the physical activity you are now getting compared to others of your same sex and age? Think about both your leisure and work activities.” Subjects indicated their response on a seven-point Likert scale ranging from 1 (extremely inactive) to 7 (extremely active). This question was used because it had been collected consistently across all five data collection points over the ten-year period and has been validated previously in this study population using daily energy expenditure as assessed from the Stanford seven-day physical activity recall (31). For instance, comparisons were made across the two measures in the sample of 1,077 men and 1,206 women who participated in the Five-City Project community health survey in 1979-1980.
There was a significant linear relationship (p values < 0.0001) found in both sexes between the seven-level global physical activity item and increases in energy expenditure as measured on the seven-day physical activity recall for kilocalories/kilogram expended and total kilocalories expended in all activity as well as in hard and very hard activity (31). As part of the Five-City Project survey, participants were also asked about their current marital status (i.e. single, married, divorced/separated, widowed) at each time point.
Statistical Analyses The overall goal of the analyses was to compare the change in physical activity level of individuals who changed marital state with the change in physical activity level of individuals who remained in one marital state throughout the ten-year period. Two analyses were conducted–one analysis comparing persons who went from single to married throughout the ten-year period with persons who remained single throughout that period and one analysis comparing persons who went from married to single during the ten-year period with persons who remained married during that period. The two comparisons of interest are summarized in Table 1. Subjects were first categorized according to their change (if any) in marital state across the ten-year period as follows: remained married, remained single, went from married to single, or
King et al.
(40%) to the initial marital state (i.e. single) was first calculated. Then a similar proportion of two- versus three-point slopes was calculated through random selection techniques for the remainedsingle group. The difference scores calculated from these slopes were compared between the single-to-married and remained-single groups using an independent-sample t-test. The mean baseline age for the single-to-married group was 29.3 -+ 10.3 years, while the mean baseline age for the remainedsingle group was 41.5 + 17.8 years ( p < 0.02). Subjects in the remained-single comparison group were therefore matched by age (to within three years) to the subjects in the single-to-married transition group. The matching procedure resulted in 36 subjects being dropped from the remained-single comparison group (resulting N = 105; mean age for the age-matched remained-single group = 31.7 +– 12.2 years; p value comparing mean age for the two groups > 0.20).
went from single to married. All subjects who had never married or who were divorced, separated, or widowed were labeled as single. Subjects who changed marital state two or more times during the decade (N = 23) were excluded from the current investigation. For those making a marital transition, subjects were included who contributed at least two data points in each marital state so that a slope fitted to the data points in the two marital states could be created. The slope was used as the primary analysis statistic to obtain a potentially more sensitive estimate of physical activity change over time than a single-point estimate (e.g. the mean). The mean and other single-point estimates offer a static summary of a behavior over a given time period and thus may obscure situations where a behavior could be naturally increasing or decreasing over that time period. Such single-point estimates have not been found to be sensitive to transition effects in previous research of this type focusing on physical activity (20). In contrast, the slope provides more dynamic information on a given health behavior and how it might be changing over time. We also have included information on the mean physical activity levels by time period for each defined marital group (Table 3).
Of the 2,394 potentially eligible subjects who had participated in the baseline Five-City Project survey, 594 had complete data across the ten-year period with respect to the major questions of interest for the current investigation (i.e. marital status, physical activity rating). After deleting the 36 subjects to obtain an age-matched sample with which to compare the single-to-married and remained-single groups as described above, the final study sample consisted of 302 women and 256 men between the ages of 25 and 74 years (mean age = 44.0 _+ 15.1 years). A majority of the sample was White (92.1%), the remainder being Hispanic (3.8%) or of another ethnicity (4.1%). The mean educational level was 13.4 _+ 2.8 years for women and 14.3 -+ 3.4 years for men.
When a comparison was made of the subjects from the original cohort sample who had missing data across the ten-year period (N = 1,800) and those with complete data who are the focus of the current investigation (N = 558), the subsample with complete data was found to be better educated (13.9 -.+ 3.0 versus 12.2 _+ 3.5 years, respectively; t-test [ d f = 1059.8] = 10.5, p < 0.0001) and was older (43.5 4- 14.2 versus 34.0 _-+ 16.8 years, respectively; t-test [df = 1087.5] = 13.1, p < 0.0001). The two subsamples were similar with respect to percent women (subsample with complete data = 54% women; subsample without complete data = 51% women) and mean baseline physical activity ratings (subsample with complete data = 4 . 7 – 1.4 subsample without complete data = 4.6 – 1.5) ( p values > 0.12). Fifty-eight adults (women = 34; men = 24) changed their marital state once sometime during the ten-year period.
Thirty-five (women = 16; men = 19) reported changing from single to married during this time period. Thirty-seven percent of these had been either divorced or widowed originally, with the rest having been never married. Twenty-three reported changing from married to single (women = 18; men = 5). Of those who became single, 69.6% had gotten divorced or separated and 30.4% were widowed. Among those who remained single throughout the ten-year period, 40% had never been married, 45% were divorced or separated, and 15% were widowed. There were no significant differences among the four groups (i.e. the two marital transition groups and the remained-married and remained-single groups) on the percentage of men versus women in each group (approximately 59% of persons in each group were women) or on the percentage of persons from intervention versus control cities as defined in the original trial
Evaluation of the Effect of Transitioning from Married to Single: For those subjects who went from married to single (N = 23), two slopes were created for each subject—one slope representing his or her physical activity levels in the married state and the other slope representing the physical activity levels in the single state. The physical activity level slope during the initial marital state (married) was subtracted from the physical activity level slope during the second marital state (single) to obtain a difference score. This difference score served as the dependent variable in evaluating changes in physical activity between these two groups during the study period. The mean difference score calculated for the married-to-single transition group was compared with a mean difference score calculated for persons who had remained married throughout the ten-year period (N = 395).
The mean baseline age for these two groups was similar (43.3 +_ 14.2 years and 46.3 – 13.4 years, respectively; p > 0.25). In creating the two slopes for the marriedto-single transition group, it was noted that a proportion of people (26%) contributed a two-point slope (i.e. made their marital transition by the third survey point) to the first marital period and a three-point slope (i.e. made their marital transition by the fourth survey point) to the second marital period, while the remaining proportion (74%) had the reverse pattern. In order to create a similar difference score using physical activity slopes for the remained-married group, a similar proportion of two- versus three-point slopes was calculated through random selection techniques for the remained-married group.
The slopes representing the initial portion of the ten-year period for this group were next subtracted from the slopes representing the second portion of the ten-year period. The difference scores calculated by subtracting the initial physical activity slope from the second physical activity slope were compared between the married-to-single and remainedmarried groups using an independent-sample t-test. Two-tailed tests were used and alpha was set at 0.05.
Evaluation of the Effect of Transitioning from Single to Married: The second analysis compared the single-to-married transition group (N = 35) with the group who had remained single throughout the ten-year period (N = 141) in the same manner as described for the first analysis. Similar to the first analysis, the percentage of people in the single-to-married transition group who had contributed a two-point slope (60%) versus a three-point slope
Marital Transitions and Exercise
(approximately 51% of persons in each group were from an intervention city) (p values > 0.10). To evaluate whether sex or community status could be associated with the dependent variable of interest (i.e. the difference score described above), analysis of variance procedures (32) were initially conducted for each of the two comparisons of interest (the married-to-single versus remained-married comparison and the single-to-married versus remained-single comparison) whereby first sex and then community status were entered as main effects in addition to marital transition status (i.e. made a transition or not). None of these analyses resulted in a significant effect for either sex or community status. Results from the simpler t-test analyses are therefore presented below.
General Description of Physical Activity Ratings There were no significant baseline differences in physical activity rating between the married-to-single transition group and the remained-married comparison group or between the single-tomarried transition group and the remained-single comparison group. Neither were there differences in the proportion of persons in each group who rated themselves as inactive (defined as a rating of 1, 2, or 3 on the 7-point physical activity item) at baseline (see Table 1). All four groups showed a small, non-significant decline in physical activity levels from the beginning to the end of the ten-year study period (p > 0.1). Ratings of global physical activity at the end of the ten-year period declined across all four groups by approximately .33 of a point from their baseline global physical activity rating (mean physical activity rating for the entire sample at the ten-year point = 4.57 – 1.49).
There were no significant differences across the four groups on rated physical activity level at the final measurement point ( p > 0.1). The proportion of persons who rated themselves as inactive at the final measurement point (defined as a rating of 1, 2, or 3 on the physical activity item) was 25% for the remained-married group, 13% for the married-tosingle group, 21% for the remained-single group, and 22% for the single-to-married group (group differences non-significant). Effect of Becoming Single on Self-Reported Physical Activity Levels The t-test analysis indicated no significant differences in physical activity slopes between persons who went from married to single and those remaining married throughout the ten-year period ( p > .80). The mean difference score between slopes was found to be 0.10 for persons transitioning from married to single and 0.06 for those remaining in the married state.
The calculated slopes for each time period for the two groups are presented in Table 2. The means by time period for each of the above two marital groups are presented in Table 3. There were no between-group differences observed when this single-point estimate of physical activity level was used (p > 0.90). Effect of Marriage on Self-Reported Physical Activity Levels The t-test analysis indicated a significant difference in physi-cal activity slopes between persons who went from single to married and those remaining single throughout the ten-year period, t(df = 138) = 1.95, p < 0.05. The calculated slopes for each time period for the two groups are presented in Table 2. The mean difference score between slopes for persons who went from single to married was 0.40 compared to a difference score of 0.05 for persons remaining single. This indicates that persons who went from single to married showed a greater increase
* = Defined as the period prior to when a marital transition occurred for those groups making a marital transition. For each of the two comparisons, the group remaining in one marital state was matched with the group making a marital transition in terms of the proportion of people in each group contributing a two- versus three-point slope during the first and second time periods. t” = Independent-sample t-test comparing difference score between the single-to-married and remained single groups significant atp < 0.05.
TABLE 3 Means (Standard Deviations) for Global Physical Activity Rating Comparing Persons Who Made A Marital Transition with Those Remaining in One Marital State By Time Period Marital Status Remained Married Married-to-Single Remained Single Single-to-Married First Period* 4.41 (1.24) 4.95 (1.18) 4.84 (1.3) 4.74 (1.19) Second Period 4.39 4.93 4.70 4.52 (1.25) (1.10) (1.29) (1.06) 2nd – 1st Period Difference1″ -0.02 -0.02 -0.14 -0.22
* = Defined as the period prior to when a marital transition occurred for those groups making a marital transition. For each of the two comparisons, the group remaining in one marital state was matched with the group making a marital transition in terms of the proportion of people in each group contributing a two- versus three-point slope during the first and second time periods. I” = Independent-sample t-tests comparing difference scores between each of the above two sets of marital groupings (i.e. comparison of remained married and married-to-single and comparison of remained single and single-to-married) non-significant (p values > 0.35). over time in their reported overall physical activity levels during the second period relative to the first period compared to the remained-single group.
Of the four marital groups shown in Table 2, only the single-to-married group evidenced a positive physical activity slope during the second time period. While the single-to-married transition group appeared to have a somewhat steeper negative slope during the initial period (see Table 2), a comparison of the initial slopes for this group and the remained-single group did not reach statistical significance (p > 0.1). However, the initial period slope for the single-tomarried group was found to be significantly different from zero ( p < 0.03).
To explore whether the drop in physical activity in the initial pretransition period for this group could potentially be due to stress or time pressures that might occur in preparing to make a transition to a married state, we analyzed additional items that had been developed specifically for the Five-City Project survey. Six items focusing on perceived stress (rated on a six-point Likert scale and anchored with “strongly agree” to “strongly disagree”) and one item (rated on a five-point Likert scale anchored with “almost never” to “very frequently”) addressing perceived time pressures experienced from daily tasks (e.g. work, homemaker activities) were available. Examples of the perceived stress items were “Finding time to relax is practically impossible for me” and “The King et al. tion, the relatively small number of persons who actually reported a marital transition during the study period in combination with a reasonably large proportion of subjects from the original cohort with incomplete data across the ten-year period limits both our power to detect significant differences as well as the strength of the conclusions that can be drawn.
In particular, the subsample with complete data that was the focus of the current investigation was significantly older and better educated than the subsample with missing data, although the two subsamples did not differ significantly at baseline on the major variable of interest (i.e. reported physical activity level). Clearly, the identification of samples that would provide a greater number of cases of marital transition across a specified time period is an important goal for future research in this area. Second, we were limited in our use of physical activity questions to a single global item that was assessed regularly across the ten-year period which, similar to the sample size issue, may have made it more difficult to detect significant effects via analysis of means, as opposed to the potentially more powerful slope analyses (33). The potential limitations of single-item measures notwithstanding, the item used in the current investigation has been found to be significantly associated in the study population with daily energy expenditure as assessed from the Stanford 7-day physical activity recall (31).
However, future studies in this area could benefit from more comprehensive evaluations of physical activity across time, particularly with respect to specific type and amount (e.g. frequency, duration, intensity) of physical activity reported. In particular, the results of the slope analyses suggest that it would be useful to ascertain what forms of physical activity might be especially affected when a marital transition occurs as a first step toward formulating interventions that could take advantage of such natural propensities for change. While parental status represents an additional source of social integration that may affect health behavior in addition to marital status (19,20,34), parental status information was not consistently obtained across the ten-year study period and thus could not be included in the current investigation.
The investigation of the parental status-marital status relationship is recommended in future studies in light of the potentially important interaction between these two social integration variables (20). Similarly, while some studies have suggested a potential interaction between marital status and employment status on health trends in at least some populations (35), employment status information was not consistently available across the study period. Given the significant length of time between measurement points during the ten-year observation period, we were unable to evaluate the acute effects of a marital transition, including changes occurring during the transition itself, on physical activity behavior. It is possible that the increases in stress that often accompany recent changes in marital status could have particularly powerful effects on health behaviors such as cigarette use, alcohol intake, sleep, and physical activity (20).
Such effects remain underexplored in the physical activity arena. The participants living in the four cities targeted by this investigation were largely White and well-educated. Additional studies are needed to determine whether the relationships observed in this study generalize to other populations with different demographic profiles. We conclude that, given the inconsistent pattern of results found in the current study, further research is required to more clearly determine the potential effects of changes in marital status on physical activity levels. Within this context, the current findings way I handle stress is a serious problem for me.” The perceived time pressure item was phrased as “How often have you felt pressed for time?” Using paired-comparison t-tests, there were no significant pretransition to posttransition differences found on any of these items for the single-to-married group (p values > .20).
Within-group comparisons of slopes using paired-comparison t-tests indicated that while there were no significant differences between the first and second portions of the ten-year period for the remained-single group, there was a significant difference between the first and second portions of the ten-year period for persons who went from single to married (paired-comparison t statistic = 2.55 [df = 34], p < 0.015). In the single-to-married group, positive change in reported physical activity was found to be greater in the married state relative It the single state. The means by time period for each of the above two marital groups are presented in Table 3. There were no between-group differences when this single-point estimate of physical activity level was used (p > 0.30). DISCUSSION The current study evaluated changes in physical activity levels associated with changes in marital status using two different types of summary statistics–means and slopes.
No significant differences among marital groups were found when the physical activity means of persons making a marital transition were compared with the means of those remaining in one marital state across the ten-year period, providing little support for the supposition that changes in marital status affect overall levels of physical activity in either a positive or a negative direction. In contrast, the slope analyses suggest that, within the targeted time period, the pattern of physical activity change over time may differ based upon whether a marital transition occurs or not. Specifically, individuals who became married during the ten-year period showed a significantly different pattern of physical activity change (i.e. a decline in the premarriage period followed by a relative leveling off in the postmarriage period) compared to individuals who remained single throughout the study period. The comparatively sharp decline in physical activity levels experienced during the premarriage period in this group is also reflected by the fact that the single-to-married group had the largest mean difference from baseline levels during the initial period (baseline minus first period mean = .36) relative to the other three marital groups (range for baseline minus first period mean = .05 to .29).
Although it did not reach statistical significance in the between-group analysis, the somewhat steeper negative slope occurring during the initial period for the single-to-married group is intriguing. It suggests the possibility that preparing for a marital transition could have a negative impact on physical activity behaviors that may be potentially ameliorated, at least to some degree, following the marital transition. Unfortunately, there was only a small amount of information available in the current data set with which to examine the potential effects of perceived stress or time issues on this pretransition decrement in physical activity.
While there were no significant differences detected in the stress and time items examined pre- to posttransition in this group, this issue deserves further exploration using larger sample sizes and more comprehensive measurement tooIs. Clearly, the inconsistent results stemming from the analyses of means versus slopes prompts a number of questions in this area that merit further investigation. First, although population-based survey techniques were utilized in identifying the cohort that served as the sample of interest for this first-generation investiga- suggest the possibility that at least some types of marital transitions may potentially set the stage for changes in physical activity, but may not be sufficient alone to influence overall physical activity levels. Such transitions, however, may provide a window of opportunity for natural changes in physical activity that could be capitalized on through appropriate intervention.
(1) Bouchard C, Shephard RJ, Stephens T (eds): Physical Activity, Fitness, and Health: International Proceedings and Consensus Statement. Champaign, IL: Human Kinetics, 1994. (2) U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion: Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: 1996. (3) King AC, Blair SN, Bild DE, et al: Determinants of physical activity and intervention in adults. Medicine and Science in Sports and Exercise. 1992, 24:$221-$236. (4) Prochaska JO, DiClemente CC: States and processes of self-change in smoking: Toward an integrative model of change. Journal of Consulting and Clinical Psychology. 1983, 51:390-395. (5) Marcus BH, Simkin LR: The transtheoretical model: Applications to exercise behavior. Medicine and Science in Sports and Exercise. 1994, 26:1400-1404. (6) Felner RD, Farber SS, Primavera J: Transitions and stressful life events: A model for primary prevention. In Felner RD, Jason LA, Moritsugu JN, Farber SS (eds), Preventive Psychology: Theory, Research, and Practice. New York: Pergamon Press, 1983, 42-67. (7) King AC: Community intervention for promotion of physical activity and fitness. Exercise and Sport Sciences Reviews. 1991, 19:211-259. (8) Winett RA, King AC, Altman DG: Health Psychology and Public Health: An Integrative Approach. Elmsford, NY: Pergamon Press, 1989. (9) Bloom BL: Community Mental Health: A General Introduction (2nd Ed.). Belmont, CA: Brooks/Cole, 1984. (10) Klorman R, Hilpert PL, Michael R, LaGana C, Sveen OB: Effects of coping and mastery modeling on experienced and unexperienced pedodonticpatients’ disruptiveness.Behavior Therapy. 1980,11:156168. (11) Owens JF, Matthews KA, Wing RR, Kuller LH: Can physical activity mitigate the effects of aging in middle-aged women? Circulation. 1992, 85:1265-1270. (12) Gove W: Sex, marital status, and mortality. American Journal of Sociology. 1973, 79:45-67. (13) House J, Landis K, Umberson D: Social relations and health. Science. 1988, 241:540-545. (14) Goldman N, Korenman S, Weinstein R: Marital
status and health among the elderly. Social Science and Medicine. 1995, 40:17171730. (15) Hu YR, Goldman N: Mortality differentials by marital status: An international comparison. Demography. 1990, 27:233-250. (16) Reissman C, Gerstel N: Marital dissolution and health: Do males or females have greater risk? Social Science and Medicine. 1985, 20:627-635. (17) Stroebe M, Stroebe W: Who suffers more? Sex differences in health risks of the widowed. PsychologicalBulletin. 1983, 93:279-301.
(18) Wingard D: The sex differential in morbidity, mortality, and lifestyle. Annual Review of Public Health. 1984, 5:433–458. (19) Umberson D: Family status and health behaviors: Social control as a dimension of social integration. Journal of Health and Social Behavior 1987, 28:306-319. (20) Umberson D: Gender, marital status, and the social control of health behavior. Social Science and Medicine. 1992, 34:907-917. (21) Joung IM, Strortks K, van de Mheen H, Mackenbach JP: Health behaviours explain part of the differences in self-reported health associated with partner/marital status in The Netherlands. Journal of EpidemioIogy and Community Health. 1995, 49:482-488. (22) Case RB, Moss AJ, Case N, McDermott M, Eberly S: Living alone after myocardial infarction: Impact on prognosis. Journal of the American Medical Association. 1992, 267:515-519. (23) Ford ES, Jones DH: Cardiovascular health knowledge in the United States: Findings from the National Health Interview Survey, 1985. Preventive Medicine. 1991,20:725-736. (24) Temple MT, Fillmore KM, Hartka E, et al: A meta-analysis of change in marital and employment status as predictors of alcohol consumption on a typical occasion. British Journal of Addiction. 1991, 86:1269-1281. (25) Wickrama K, Conger RD, Lorenz FO: Work, marriage, lifestyle, and changes in men’s physical health. Journal of Behavioral Medicine. 1995, 18:97-111. (26) Kahn HS, Williamson DF: The contributions of income, education, and changing marital status to weight change among U.S. men. International Journal of Obesity. 1990, 14:1057-1068. (27) Kahn HS, Williamsou DF, Stevens JA: Race and weight change in U.S. women: The roles of socioeconomic and marital status. American Journal of Public Health. 1991, 81:319-323. (28) Farquhar JW, Fortmann SP, Maccoby N, et al: The Stanford Five-City Project: Design and methods. Journal of the American Medical Association. 1985, 122:323-334. (29) Farquhar JW, Fortmann SE Flora
JA, et al: The Stanford Five-City Project: Effects of community-wide education on cardiovascular disease risk factors. Journal of the American Medical Association. 1990, 264:359-365. (30) Young DR, Haskell WL, Taylor CB, Fortmann SP: Effect of community health education on physical activity knowledge, attitudes, and behavior: The Stanford Five-City Project. American Journal of Epidemiology. 1996, 144:264-274. (31) Blair SN, Haskell WL, Ho E et al: Assessment of habitual physical activity by a seven-day recall in a community survey and controlled experiments. American Journal of Epidemiology. 1986, 122:794804. (32) Spector PC, Goodnight JH, Sail JP, Sarle WS: The GLM procedure. In SAS User’s Guide: Statistics, Version 5 Edition. Cary, NC: SAS Institute Inc., 1985, 433-506. (33) Kraemer HC, Thiemann SA: A strategy to use “soft” data effectively in randomized clinical trials. Journal of Consulting and Clinical Psychology. 1989, 57:148-154. (34) Verhoef MJ, Love EJ: Women’s exercise participation: The relevance of social roles compared to non-role-related determinants. Canadian Journal of Public Health. 1992, 83(5):367-370. (35) Waldron I, Hughes ME, Brooks TL: Marriage protection and marriage selection–Prospective evidence for reciprocal effects of marital status and health. Social Science and Medicine. 1996, 43:113-123.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.