| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
RESEARCH ARTICLE |
1 National Institute for Public Health and the Environment, Bilthoven, The Netherlands.
2 Julius Centre for Health Sciences and Primary Care, University Medical Centre Utrecht, The Netherlands.
3 National Institute of Health, Rome, Italy.
4 National Public Health Institute, Helsinki, Finland.
5 Division of Human Nutrition, Wageningen University, The Netherlands.
Address correspondence to B. M. van Gelder, MSc, National Institute for Public Health and the Environment, Centre for Prevention and Health Services Research, PO Box 1, Internal Postal Code 101, 3720 BA, Bilthoven, the Netherlands. E-mail: Boukje.van.Gelder{at}rivm.nl
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
PREVIOUS studies have shown that marital status, living situation, and preservation of a social network are related to health status and mortality (House, Landis, & Umberson, 1988
; Iwasaki et al., 2002; Johnson, Backlund, Sorlie, & Loveless, 2000; Seeman, 1996; Waldron, Hughes, & Brooks, 1996). One such study showed that participation in social and leisure activities was associated with a lower risk of dementia (Fabrigoule et al., 1995), which suggests that this lower risk may be due to cognitive stimulation. Furthermore, other studies showed that having few social contacts and a poor social network is associated with a higher risk for cognitive decline and dementia (Bassuk, Glass, & Berkman, 1999; Fratiglioni, Wang, Ericsson, Maytan, & Winblad, 2000; Wang, Karp, Winblad, & Fratiglioni, 2002; Zunzunegui, Alvarado, Del Ser, & Otero, 2003).
Research is now focusing on marital status as well as living situation in relation to cognitive functioning. Two longitudinal studies focusing on marital status showed that men and women who were married were at decreased risk for dementia and Alzheimer's disease compared with widowed, divorced, separated, or never married men and women (Fratiglioni et al., 2000; Helmer et al., 1999). Two other studies focusing on living situation showed that men and women who lived alone were at an increased risk for dementia compared with persons who lived with others (with spouse, children, others, or in a nursing home; Fratiglioni et al.; Sibley et al., 2002).
Some possible explanations for the protective effect of being married or living with someone in relation to cognitive decline could be proposed. For example, the cognitive stimulation of a partner or other person may protect the brain from deterioration. Furthermore, the loss of a partner could cause changes in lifestyle (such as changes in smoking, drinking, and dietary habits) or even stress and depression. Especially among men, the loss of a partner is associated with more distress and a higher risk for depression (Lee, DeMaris, Bavin, & Sullivan, 2001), which could cause adverse health effects, including cognitive decline. Therefore, we selected men for the present study.
A drawback of previous studies on social status and cognitive functioning is that marital status or living situation was measured only once in time and not repeatedly during a longer period. Especially in old age, this is important, because social situations change frequently. Furthermore, most previous studies focused on dementia and Alzheimer's disease and not on cognitive decline.
We investigated the associations between types of marital status (married vs unmarried) as well as types of living situation (lived with others vs lived alone) over a period of 5 years in relation to subsequent 10-year cognitive decline among 1,042 elderly men in Finland, Italy and the Netherlands.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). In 1985, at baseline, 2,285 men between 65 and 84 years of age participated (response rate varied between 74% and 94%). In 1990, there were 1,734 men still alive, of whom 1,416 participated in the 1990 survey (response rates in the three countries varied between 77% and 88%). Detailed information on this study population can be found elsewhere (Bijnen et al., 1996). Approval of the Medical Ethics Committee in the different countries was obtained for each participating center, and participants gave their informed consent.
Health surveys were carried out once every 5 years since 1985. Information on cognitive functioning was available for 1,363 of the 1,416 men examined in 1990. Information about marital status and living situation was collected in both 1985 and 1990 and was available for respectively 1,310 and 1,270 of the 1,363 men. Because low cognitive functioning at baseline may cause a change in marital status or living situation instead of being a consequence, we excluded men who lived in an institution at baseline (because of their poor health status) and those who were severely cognitively impaired (men with a Mini-Mental State Examination [MMSE] score 
18; see Murden, McRae, Kaner, & Bucknam, 1991). Complete information on cognition, possible confounding factors in 1990, and marital status was available for 1,042 men and for living situation for 1,014 men.
Transition in Marital Status and Living Situation
We obtained information on marital status and living situation by using standardized questionnaires. We categorized marital status into two categories: being married and being unmarried, which also included being separated, divorced, and widowed. To analyze transition in marital status, we classified participants according to their marital status in respectively 1985 and 1990: married–married (n = 782), married–unmarried (n = 98), unmarried–unmarried (n = 150), and unmarried–married (n = 12). Because of the small number of men in the last category, we excluded these men from the analyses.
We divided living situation into living alone and living with others, that is, with spouse, with spouse and children, with family, or with others. We defined transition in living situation according to the living situation in 1985 and 1990: living with others–living with others (n = 822), living with others–living alone (n = 94), living alone–living alone (n = 88), and living alone–living with others (n = 10). Because of the small number of men, we excluded the last category from the analyses.
Cognitive Functioning
Trained researchers assessed overall cognitive functioning during the 1990, 1995, and 2000 surveys with the MMSE (Folstein, Folstein, & McHugh, 1975). However, information on cognitive functioning was not collected in Finland during the 1995 survey. The MMSE includes questions on orientation to time and place, registration, attention and calculation, recall, language, and visual construction. The MMSE is developed as a screening instrument to assess the severity of cognitive impairment and cognitive changes over time (Tombaugh & McIntyre, 1992). Originally, this test was created for clinical use (Folstein et al.), but it is now used extensively in epidemiologic studies (Launer, 1992). Furthermore, the MMSE has good test–retest reliability; because of its high sensitivity and specificity, it is a valid indicator for normal and impaired cognitive functioning (Siu, 1991; Tombaugh & McIntyre). Although the MMSE is a measure of global cognitive functioning and does not assess different cognitive domains in detail, it is sensitive enough to detect "clinically significant" global cognitive decline (Lyketsos, Garrett, Liang, & Anthony, 1999).
The maximum score on the MMSE is 30 points, and a higher score indicates better cognitive functioning. If a subject did not answer 4 or more individual items (of a total of 20), we considered the total MMSE score to be missing (n = 9). If fewer than 4 items were lacking, we rated these missing items as zero, and we still calculated a total MMSE score (Fillenbaum, George, & Blazer, 1988).
Other Variables
During all examinations, researchers collected information on demographic, lifestyle, and other variables by using standardized questionnaires (Keys et al., 1966); physicians and nurses obtained medical information during physical examinations. Researchers assessed education as the number of years of formal education, and they categorized smoking status into never, former, and current smoking and alcohol consumption into consumers and nonconsumers. At the end of the physical examination, a physician measured systolic and diastolic blood pressure with a random-zero sphygmomanometer with the participant in a supine position. Researchers obtained information about the use of antihypertensive drugs with a questionnaire. Because of this elderly population, we defined hypertension as having a systolic blood pressure of 160 mm Hg or greater, having a diastolic blood pressure of 95 mm Hg or greater, or using antihypertensive medication (Guidelines Sub-committee, 1993). Researchers obtained information on the prevalence of myocardial infarction, stroke, diabetes, and cancer by questionnaire, and they checked this with information from hospital registries or general practitioners (yes or no). They assessed depressive symptoms by using the Self-Rating Depression Scale developed by Zung (1965). The score range on the Self-Rating Depression Scale is from 25 to 100, and a higher score indicates having more depressive symptoms. Researchers used a value of at least 50 to indicate the presence of depressive symptoms (yes or no). They measured functional status with an activities of daily living questionnaire and categorized participants as not disabled, disabled in instrumental activities of daily living only, disabled in instrumental activities and mobility, and disabled in all domains (Hoeymans, Feskens, van den Bos, & Kromhout, 1996). They assessed physical activity with a validated self-administered questionnaire, which was originally designed for the retired man, and divided respondents into four categories: low active, medium-low active, medium-high active, and high active (Caspersen, Bloemberg, Saris, Merritt, & Kromhout, 1991).
Statistical Analyses
To test whether characteristics in 1985 of nonparticipants or of men with missing values in 1990 differed from the participating men in 1990, we used the Wilcoxon test. We also calculated the Spearman correlation coefficient between the variables of marital status and living situation in 1990.
Marital Status, Living Situation, and Cognitive Functioning
We tested differences in characteristics between the three categories of marital status and living situation in 1990 with the Kruskal–Wallis and chi-square tests. We performed multiple linear regression analyses to obtain adjusted mean 1990 MMSE scores per marital status or living situation category. In all analyses, we made adjustments for potential confounding factors. We describe these factors at the end of statistical analysis section.
To investigate the independent effect of marital status as well as living situation on subsequent 10-year cognitive decline, we used mixed longitudinal random coefficient models (SAS Proc Mixed procedure). These models take into account the intracorrelation of measurements (in 1990, 1995, and 2000) performed by the same person and does not exclude persons with incomplete data at follow-up. To investigate whether 10-year cognitive decline differed between the marital status or living situation categories, we included the product of these categories with time in the model. We gave cognitive decline (in number of points with 95% confidence interval, or CI) for the reference group married–married (or living with others–living with others) and additional cognitive decline (compared with the decline of the reference group) for the other categories as output.
To investigate whether marital status or living situation is a stronger predictor for cognitive decline, we performed stratified analyses. Because the number of participants in some stratified groups was very small, we did only two stratified analyses. We used a general linear model to investigate whether 10-year cognitive decline differed among men who were married and men who were not married, but who both lived with others. In the same way, we investigated whether cognitive functioning differed among men who lived with others or who lived alone, but who both were unmarried.
Potential Confounding Factors
In all analyses, we made adjustments for the following well-known confounding factors: age (continuous), education (continuous), country (categorical), smoking status (categorical), and alcohol consumption (categorical; Hofman et al., 1991; Kalmijn, van Boxtel, Verschuren, Jolles, & Launer, 2002; Launer et al., 1999). Because disease status could influence cognitive functioning, we also adjusted for the prevalence of myocardial infarction (categorical), stroke (categorical), diabetes (categorical), and cancer (categorical). To investigate whether the effect of marital status was independent of the living situation of the men, we adjusted for living situation (categorical) in the analyses regarding marital status. We did similar analyses for living situation. In the longitudinal analyses, we also adjusted for baseline cognitive functioning (continuous), because baseline cognitive functioning may influence cognitive decline. Additionally, we adjusted for hypertension (categorical) as a risk factor for cognitive decline, and possible intermediates such as physical activity (categorical), depression (categorical), and functional status (categorical). Finally, we excluded participants who, in 1990, had a MMSE score below 24, which indicates mild impaired cognition (n = 234; Siu, 1991). We considered values of p .05, two sided, to be statistically significant.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Marital Status, Living Situation, and Cognitive Functioning
Men who were married in both 1985 and 1990 were the youngest (75.7 years old), and men who lost a partner were the oldest (77.5 years; p <.001; see Table 1). Married men had the highest average baseline cognitive test score (25.8 points), and men who were unmarried the lowest (25.2 points; p =.01). The percentage of men with depressive symptoms was the highest among men who lost a partner (18%) compared with married and unmarried men (8% and 6%, respectively; p =.001). Married men and men who lost a partner both spent 717 min per week on physical activities, and unmarried men spent 570 min per week (p =.04). The percentage of men with a history of myocardial infarction was highest for unmarried and married men (respectively 13% and 14%) and lowest for men who lost a partner (5%; p =.06). Married and unmarried men were most often of Dutch origin, and men who lost a partner were most often of Italian origin (p =.01).
|
|
|
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
A major strength of this study is its prospective design, with a 15-year follow-up and repeated measurements. In addition, adjustment could be made for numerous important confounding factors. The study has also limitations. Selection bias caused by nonresponse or missing values could have influenced our results. Overall, nonparticipants were older, had fewer years of education, had a higher prevalence of stroke, and were more often unmarried than were men who participated in the present study. Therefore, the strength of the relationships observed in the present study may have been underestimated. Furthermore, response rates in 1985, 1990, 1995, and 2000 were high and varied between 65% and 94%. To deal with the possible selection caused by the dropout, we used a mixed longitudinal random coefficient model, which does not exclude participants with incomplete data on cognitive functioning at follow-up. We also repeated the analyses among survivors with complete data (until 2000). These analyses confirmed the results found among all men. Therefore, selection bias caused by incomplete data on cognitive functioning at follow-up has not influenced our results.
The MMSE was used to assess global cognitive functioning. Although this is a screening test, the MMSE is a reliable and valid indicator of cognitive impairment, with a good test–retest reliability; it is often used in epidemiologic studies (Launer, 1992; Siu, 1991; Tombaugh & McIntyre, 1992). A limitation of the MMSE is that it measures global cognitive functioning and does not assess specific cognitive domains in detail. Therefore, we recommend that future studies include a more extensive cognitive test battery and focus on specific cognitive domains, such as memory, concentration, attention, learning, language, and visual construction.
Because of differences in the percentages of Finnish, Dutch, and Italian men in each category of marital status and living situation, it is of interest to know if the observed relationships differ from country to country. However, the number of participants in the different categories for each country was too low for meaningful analyses; therefore, we could not perform analyses stratified by country. However, an adjustment for country in the multivariate analyses did not change our results, suggesting that the findings were the same for each country.
To our knowledge, no other longitudinal study has investigated the effect of marital status and living situation over a 5-year period on subsequent cognitive decline. Most studies have focused on an association between baseline marital status and living situation with cognitive decline. The PAQUID study found that initially nondemented elderly persons who were never married were at a higher risk of dementia or Alzheimer's disease than were married persons (Helmer et al., 1999). Another study showed that participants who lived alone were diagnosed with dementia at an earlier stage than were participants who did not live alone (Sibley et al., 2002). The Kungsholmen Project showed that unmarried participants who lived alone and who had no friends were at a higher risk of developing dementia (Fratiglioni et al., 2000). This study also showed that socially or mentally stimulating activities may protect against dementia (Wang et al., 2002). Furthermore, our finding that marital status may be a stronger predictor than living situation for cognitive decline confirms the results of the Kungsholmen Project, which also suggests that being single (not married) is associated with a higher risk for dementia than is living alone (Berkman, 2000; Fratiglioni et al.).
Several possible underlying mechanisms could explain our findings. Unmarried men or men who lived alone in 1985 and 1990 could have had less social support and cognitive stimulation than married men or men with a housemate. According to the "use it or lose it" hypothesis, this may have resulted in a stronger cognitive decline (Coyle, 2003). This hypothesis states that participation in mentally stimulating activities may increase neuronal growth and maintenance and thereby protect the brain from neuronal degeneration and subsequent cognitive decline. Although this hypothesis is speculative, animal research supports it (Lucassen, van Someren, & Swaab, 1998). One study showed that cognitive training resulted in less cognitive decline in older persons without dementia (Ball et al., 2002). Other epidemiological studies have also shown that participation in mentally stimulating activities or leisure activities lowers the risk of dementia and Alzheimer's disease (Verghese et al., 2003; Wilson et al., 2002). However, a Swedish study showed that participation in leisure activities was not protective against dementia and Alzheimer's disease among men (Crowe, Andel, Pedersen, Johansson, & Gatz, 2003). Because marital status may be a stronger predictor than living situation for cognitive decline, cognitive stimulation or social support from a partner could be more protective against cognitive decline than interaction with or social support from other persons. It may also be the satisfaction of the affiliation with a partner that protects against cognitive decline.
Furthermore, losing a partner, being unmarried, starting to live alone, and living alone may cause adverse health effects such as stress and depressive symptoms, which lead to an increase in cortisol production. High cortisol levels are implicated in hippocampal damage, the part of the brain where memory is located, and may thereby result in memory deficits (Berkman, Glass, Brissette, & Seeman, 2000; Kalmijn et al., 1998; McEwen & Sapolsky, 1995). Adjustment for depressive symptoms did not affect our results, though. Finally, after the loss of a partner or after a person started to live alone, the person's lifestyle could change, such as a reduction in physical activity or an alteration in alcohol drinking and smoking habits. These lifestyle changes may have an adverse effect on cognitive functioning (Kalmijn et al., 2002; Van Gelder et al., 2004). However, adjustments for these lifestyle factors did not alter our results.
The Kungsholmen Project already suggested that an extensive social network protects against dementia (Fratiglioni et al., 2000). Probably, an extensive social network is related to marital status and living situation. Therefore, the results of the present study could also be a result of an extensive social network of the participants instead of their marital status or living situation. Unfortunately, in this study we had no information about social networks available, so we were unable to investigate this possibility.
In old age, transitions in marital status and living situation as well as cognitive decline are very common. The results of the present study support the hypothesis that having a partner or living together with others is associated with a smaller cognitive decline. This knowledge may have important implications for public health programs aimed at healthy aging. We should stimulate elderly men to be around other people and not to be alone. Furthermore, caretakers should be aware of the fact that elderly men who are unmarried or who live alone carry a higher risk of cognitive decline.
|
Acknowledgments |
|---|
Co-author Marja Tijhuis passed away after this article was accepted for publication.
|
Footnotes |
|---|
Received for publication March 4, 2005. Accepted for publication November 13, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
