HomeLarge Type Edition
HOME ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
PubMed
Right arrow PubMed Citation
The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 58:M954-M959 (2003)
© 2003 The Gerontological Society of America

Is Social Integration Associated With the Risk of Falling in Older Community-Dwelling Women?

Kimberly A. Faulkner1, Jane A. Cauley1, Joseph M. Zmuda1, Joan M. Griffin2 and Michael C. Nevitt3

1 Department of Epidemiology, University of Pittsburgh, Pennsylvania.
2 Center for Chronic Disease Outcomes Research, Minneapolis Veterans Affairs Medical Center, Minnesota.
3 Department of Epidemiology and Biostatistics, University of California, San Francisco.


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 REFERENCES
 
Background. Social integration may lead to social support and influence that may in turn protect older community-dwelling adults from falls.

Methods. We examined incident falls over 3 years across quartiles of social integration scores in 6692 Caucasian women enrolled in the Study of Osteoporotic Fractures (mean age = 77 ± 5 years). Social integration was assessed using family networks, friendship networks, and interdependence scores. Higher scores correspond to greater integration. Data were analyzed using Poisson regression with generalized estimating equations. Multivariate analyses were used to adjust for other risk factors and potential confounders.

Results. Women reported 11,863 falls, averaging 0.60 falls per person annually, 95% confidence interval (CI) (0.57, 0.63), or 600 falls per 1000 women. In age-adjusted analysis, the average incidence rate of falls correlated inversely with family networks, interdependence, and composite integration scores (p <.05). In multivariate analysis, increasing family networks were inversely associated with fall risk, p(trend) =.02. Compared to the lowest quartile, the relative risk of falls (95% CI) associated with family network scores in the second, third, and fourth quartiles were 0.90 (0.79–1.03), 0.86 (0.74–1.00), and 0.84 (0.71–0.99), respectively.

Conclusions. Strong family networks may protect against the risk of falls in older community-dwelling adults.

APPROXIMATELY 1 in 3 older adults falls at least once each year (1–3), contributing to the majority of fractures (4–6) and accidental deaths (7,8) in this population. Research has linked social integration to physical and mental health and longevity (9–15), but its relation to fall risk is unclear. Previous fall research has identified many risk factors for falls (1,2,16–19), spanning a number of medical, functional, and extrinsic factors, but little is know about the potential benefits from social relationships. Possible protective mechanisms include encouraging attention to hazards in the environment and facilitating adaptations to changes in physical function as with environmental designs and/or assistance with routine or rare tasks such as taking out the garbage or reaching an object on a high shelf (20,21).

We identified 2 cross-sectional studies and 2 prospective studies that analyzed the relationship of social integration (10,21), social support (22), or both (23) on the risk of falling in older community-dwelling adults. One study reported that family and neighbor support, and separately, not living alone, a single-item indicator of social integration, were associated with a positive history of falls in men and women, respectively (23). A second study reported that frequency of social interaction, a single-item indicator of social integration, was associated with reduced fall risk after adjusting for age, sex, and mobility (21). However, data were cross-sectional and thus, the sequence of falling and social indicators are unclear (23). Results from prospective studies were mixed (10,22). One study reported no difference in fall risk by level of emotional support (22) and the second reported social networks were positively associated with reduced fall risk in crude analyses (10).

Previous research is consistent with the concept that social integration may protect against falls. They also suggest that social integration may provide pathways, other than social support, that protect against falls, such as social influence and opportunity for assistance. Only one of the previous studies of falls included an assessment of social networks, a measure of social integration, described as the web of social relationships surrounding an individual that provide opportunities for social support, influence, engagement, contact with others, and access to resources (24). Previous falls studies were limited by use of single-item indicators of social integration, cross-sectional design, or by their lack of adjustment for key confounders including depression and other medical and physical factors. The present study will analyze the possible relationship of social integration and falls in a large prospective study of older community-dwelling women, and improve on previous research by assessing social networks and simultaneously controlling for a wide range of potential confounding factors.


   METHODS
Top
 Abstract
 Methods
Results
 Discussion
 REFERENCES
 
Study Population
We performed a 3-year prospective analysis of fall rates among women enrolled in the Study of Osteoporotic Fractures (SOF). Detailed methods of SOF have been reported elsewhere (25). In brief, 9704 nonblack women were recruited from population-based lists (such as voter registrations) in 4 clinical centers—Baltimore, Pittsburgh (Monongahela Valley), Minneapolis, and Portland from 1986 to 1988 inclusive. To qualify for the SOF, women had to be at least 65 years old, ambulatory, live in the community, and not have had bilateral hip replacement. Eligible women were invited to attend an initial clinic examination from 1986–1988 (visit 1) and regular follow-up clinic examinations in 1989 (visit 2), 1991 (visit 3), and 1992 (visit 4). During each clinic visit, data were collected on risk factors and health outcomes using self-administered questionnaires, interviews, and physical examinations. The study sample for the present analysis consisted of the 6692 women who completed the Lubben Social Network Scale (LSNS) at the fourth clinic visit approximately 5 years after baseline (80% of participants). The mean age of women was 77 years (range, 69–101 years), and 98% were residing in the community (including private homes, senior complexes, and retirement homes). All participants provided written informed consent. The study protocol was approved by Institutional Review Boards representing each SOF clinic.

Outcomes
Data were prospectively collected on number of falls using a series of postcards mailed every 4 months for 3 years immediately following the fourth clinic visit. On each postcard, participants were asked if they had fallen in the past 4 months and, if so, how many times. Participants were contacted by phone when postcards were not returned. Follow-up on postcard mailings were 95% complete.

Social Integration
Social integration was assessed using the 10-item LSNS (26). The LSNS measures family and friendships social networks and also interdependence, an indicator of social support exchange. Items related to family and friendship networks included number of contacts with whom respondent felt close, number of those contacts that were seen at least once a month, and frequency of contact with the one seen the most. Items related to interdependence included having and being a confidant, helping others, and living situation. Each item was weighted equally (score range, 0–5). Scores were calculated by summing over individual items. Higher scores correspond to more social integration. SOF investigators used a slightly modified questionnaire in which responses to 2 of the interdependence items, having and being a confidant, were simplified. Instead of 6 responses, the modified version included 4 responses: "never," "seldom," "sometimes," and "very often/always," in which "often" was omitted and "very often" and "always" were combined and scored 0, 1, 2, and 5 points, respectively.

Other Potential Risk Factors
All measures were collected at the fourth clinic visit by trained clinic staff except where noted. Demographic information, including date of birth, and education, was collected by self-report (clinic visit 1). Body weight was measured using a balance beam scale. Medical symptoms and disease (self-report) included the following: trouble with dizziness in the past year, urinary incontinence (involuntary loss of urine that the participant considered to be a problem), and physician diagnoses for common chronic conditions.

Physical functioning was assessed, including grip strength measured as the average of right and left using an adjustable hand grip dynameter (Preston Grip Dynameter; Takei Kiki Kogyo, Tokyo, Japan), quadriceps strength as the average of right and left (BodyMasters; Lafayette Instruments Co., Lafayette, IN), walking speed (seconds/6 meters), and time to complete 5 chair-stands without use of arms. Health impairment data included cognitive impairment assessed as scores on a modified Mini-Mental Status Examination (27,28) that corresponded to the lowest decile (score < 23), hearing impairment as the failure to hear tones at both 1000 Hz and 2000 Hz in the better ear based on a 40-decibel screen (29), functional difficulty assessed as difficulty in 1 or more of the 13 Instrumental Activities of Daily Living due to a health problem using the modified version of the Stanford Health Assessment Questionnaire (30), depression as 6 or more depressive symptoms on the 15-item Geriatric Depression Scale Shortened (12,31,32), and also self-reported fear of falling (clinic visit 1).

Prior to clinic visit 4, data related to health impairment and physical functioning were collected: distance-depth perception using the Howard-Dolman (33), scored as the standard deviation of 4 trials and corrected visual acuity (34) by averaging the scores for low and high spatial frequencies (clinic visit 1), hand-reaction time as the average response (clinic visit 2), and static balance as the ability to hold a standing tandem position for 10 seconds with eyes open (clinic visit 3). Because measures from earlier clinic visits were correlated (p <.05) and associated (p <.11) with those same measures taken on only approximately 25% of the women at visit 4, we used the data collected on the entire sample from the earlier visits.

Data on the following health behaviors were also collected (self-report): current smoking status, recent alcohol use (past 30 days), and walking for exercise (at least 1 block without stopping). Women were instructed to list all medications used in the past 4 weeks on a form, and subsequently 1 physician and doctor of pharmacy categorized all medications using the Veterans Affairs Medication Classification System (35). Medication data were coded as number of medications currently used excluding vitamins and minerals and use of central nervous system (CNS)-active medications (e.g., benzodiazepines, anticonvulsants, narcotics, and antidepressants).

Statistics
Women were divided into 4 roughly equal groups according to quartiles of social integration scores obtained for the composite and each type of component measure. For each social integration score, the relative risk (RR) was computed using Poisson regression models with generalized estimating equations (GEE) to adjust standard errors for correlated data points at 4-month intervals (36). A Huber-White Sandwich estimator of variance was used with GEE to construct valid standard errors (36). The RR was computed as the fall rate (average number of falls per 4 months) in a specific quartile of social integration score divided by that in the group with the lowest social integration score with adjustment for age. In multivariate social integration models, we simultaneously analyzed social integration and other potential risk factors. A forward-stepwise selection procedure was used to add and remove potential covariates from the multivariate regression models (entry criteria: p <.05 and exit criteria: >=.10). Models were also used to compute overall and age-adjusted fall rates and 95% confidence intervals (CI) in relation to quartiles of social integration scores. Baseline characteristics and potential risk factors for falls were analyzed across quartiles of the composite social integration score using analyses of covariance and chi square tests. We tested for both main effects of social integration as well as for second-order interaction effects with age, functional impairment, and alcohol consumption. All statistical analyses were performed using STATA version 7 (StataCorp, College Station, TX) (38).


   RESULTS
Top
 Abstract
 Methods
 Results
Discussion
 REFERENCES
 
During the 3-year follow-up period, women reported 11,863 falls averaging 0.60 falls annually (95% CI: 0.57–0.63), or 600 falls per 1000 women annually. Subscale scores of social integrations were positively, although weakly correlated with one another (Table 1). In age-adjusted analyses, social integration was inversely correlated with age, medical symptoms, disease (except Parkinson's disease), health impairment, and use of medications, and positively correlated with physical function and consumption of alcohol (Table 2).


View this table:
[in this window]
[in a new window]
  		Table 1. Mean Social Integration Score and Correlation of Specific Types of Social Integration at Baseline.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Age-Adjusted Baseline Characteristics and Risk Factors for Falls According to Quartiles of the Composite Social Integration Score (N = 6692).

 
Age-adjusted trends in annual fall rates are reported for the composite social integration score and separately by type (Figure 1), with the greatest evidence of a trend seen for the family network scores (p <.001). Lower composite scores were significantly associated with increased fall risk in age-adjusted but not multivariate models (Table 3). The age-adjusted relationship disappeared in multivariate analysis after adjustment for depth perception, grip strength, and time to complete 5 chair-stands, but the largest confounding effect was from depression. Adjustment for medication use and medical symptoms/disease variables had no further effect on risk estimates.



View larger version (47K):
[in this window]
[in a new window]
  		Figure 1. Age-adjusted annual fall rates by quartiles of social integration scores. ***p(trend) =.001; **p(trend) =.01; *p (trend) <=.05

 

View this table:
[in this window]
[in a new window]
  		Table 3. Rate Ratios of Falls According to Quartiles of Social Integration Score.

 
Family networks and interdependence were each associated with falls in age-adjusted analyses. After multivariate adjustment, the interdependence association virtually disappeared and the family network association was attenuated slightly, with increasing quartiles of family network scores associated with decreasing fall risk (p <.02). The loss of the interdependence association was because of confounding from depression. The attenuation of the family network association was primarily from depression and also grip strength followed by chair-stand time. Friendship networks, which lacked any association in the age-adjusted analysis, strengthened somewhat in multivariate analysis with adjustment for depression and also chair-stand time but remained nonsignificant, p(trend) =.16. Analyses of second order interactions with age, functional difficulty, and alcohol consumption were all nonsignificant, p >.15.

Because family and friendship network scores were correlated with each other and somewhat with falls, we attempted to isolate the main effect of each social integration measure by additionally adjusting for quartiles of family and friendship networks scores in subsequent analyses. We did this by adding each component(s) to the multivariate models, which include age, arthritis, Parkinson's disease, incontinence, low distance depth perception, depression, dizziness, use of CNS-active medications, use of >=4 medications, grip strength, and chair-stand time. After adjustment for friendship networks, the family network association became significant, p(trend) =.02, with greater family networks associated with decreased fall risk. After adjustment for family networks, friendship networks, surprisingly, showed a tendency to be positively correlated with fall risk, p(trend) =.06. In order to be able to provide a single assessment of the protection associated with stronger family networks, we dichotomized the variable across its threshold. Stronger family networks (second, third, and fourth quartiles vs first) were associated with a 13% decreased risk of falls (RR = 0.87; 95% CI: 0.78, 0.98). The estimated protective effect of stronger family networks, while seemingly modest, were actually similar to most factors (results not shown)—a 5-second decreased time to perform 5 chair-stands, a 5 kg increase in grip strength, 5-year-younger age, taking 3 or fewer medications, having better distance-depth perception, not reporting a history of arthritis, and not using CNS-active medications, which were associated with 13, 13, 10, 15, 12, 13, and 23% fewer falls, respectively.


   DISCUSSION
Top
 Abstract
 Methods
 Results
 Discussion
REFERENCES
 
In this large prospective study of older community-dwelling Caucasian women aged 70 years and older, we evaluated the relationship between social integration and fall risk. We found stronger family networks were associated with a decreased risk of falls and, conversely, that weaker friendship networks showed a tendency for decreased risk of falls. Because of the opposite effects of family and friendship networks on fall rates, the composite measure of social integration was not significantly associated with fall risk.

These findings are consistent with theories that relatively strong family networks may lead to a variety of support (physical assistance and access to resources), support-related (opportunities for support in times of need), and influential mechanisms (expressed concerns about hazardous activities and/or environmental factors), which, in turn, may protect against falls. The tendency for stronger friendship networks to be associated with increased fall risk suggests that friendships may keep one socially active, but may not necessarily lead to the same benefits that may be relevant to reducing fall risk in older community-dwelling women.

To account for the possible confounding effects of other risk factors on fall risk, we adjusted for a multitude of both medical and physical factors. The slight weakening of the results of family network in age-adjusted analysis to multivariate adjusted analysis suggests that medical and physical factors have important confounding effects. The association of friendship networks with falls from age-adjusted to multivariate analysis was reversed and actually strengthened, suggesting even more important confounding effects. Finally, comparisons of interdependence in age-adjusted and multivariate analysis indicate that medical and physical factors contribute to most of the association between interdependence and falls. Adjusting multivariate models for the different types of network measures strengthened the associations for family and friendship networks, suggesting that these variables are also important confounders.

Our findings are consistent with other studies of social integration and falls (10,21), suggesting that stronger integration may protect against falls. However, our study is the first study to analyze integration among family members separately from that with friends. Studies focusing on social support (22,23) did not suggest a protective relationship with falls, possibly because social integration among family may lead to protective factors other than support that are important for preventing falls.

Strengths of this study include its assessment of falls three times annually, large population-based sample size, inclusion of a comprehensive array of potential risk factors, and assessment of social integration using an assessment tool specifically designed and validated for use in older populations. Nevertheless, there are some limitations. Falls were never defined for participants, and it is possible that they were over-reported. It is possible that we underestimated risk factors for falls among women who fell multiple times, since we collected data on risk factors only once. Findings have limited generalizability, since the study population consisted of a survivor cohort of Caucasian women. Future research should also consider the potential effects of social integration on fall risk in other ethnic groups and in men.

Summary
Our results suggest that stronger social ties with family members may reduce fall risk among older adults. Given the average annual incident fall rate reported in this study (600 falls per 1000 women), findings suggest that strong family networks are associated with 780, 1560, and 2340 fewer falls over 1, 2, and 3 years, respectively, in a population of 10,000 women of otherwise average risk. These findings, while seemingly modest to an individual, are important to public health and suggest that older adults should try to strengthen or at least maintain strong relations with family.


   Acknowledgments
 
Address correspondence to Kimberly Faulkner, MPH, GSPH, Department of Epidemiology, 130 Desoto Street, Parran 509, Pittsburgh, PA 15261. E-mail: kaf24{at}pitt.edu

Received September 19, 2003

Accepted January 3, 2003


   REFERENCES
Top
 Abstract
 Methods
 Results
 Discussion
 REFERENCES
 

  1. O'Loughlin JL, Robitaille Y, Boivin J, Suissa S. Incidence of and risk factors for falls and injurious falls among the community dwelling elderly. Am J Epidemiol.. 1993;137:342-354.[Abstract/Free Full Text]
  2. Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community-based prospective study of people 70 years and older. J Gerontol Biol Sci Med Sci.. 1989;44A:M112-M117.
  3. Prudham D, Evans JG. Factors associated with falls in the elderly: a community study. Age Aging.. 1981;10:141-146.
  4. Sterling DA, O'Connor JA, Bonadies J. Geriatric falls: Injury severity is high and disproportionate to mechanism. J Trauma Injury Infect Crit Care.. 2001;50:116-119.
  5. De Laet CE, Pols HA. Fractures in the elderly: epidemiology and demography. Best Pract Res Clin Endocrinol Metabol.. 2000;14:171-179.
  6. Luukinen H, Herala M, Koski K, Kivela SL, Honkanen R. Rapid increase of fall-related severe head injuries with age among older people: a population-based study. J Am Geriatr Soc.. 1999;47:1451-1452.[Medline]
  7. Rivara FP, Grossman DC, Cummings P. Injury prevention. Second of two parts. N Engl J Med.. 1997;337:613-618.[Free Full Text]
  8. Nyberg L, Gustafson Y, Berggren D, Brannstrom B, Bucht G. Falls leading to femoral neck fractures in lucid older people. J Am Geriatr Soc.. 1996;44:156-160.[Medline]
  9. Cerhan JR, Wallace RB. Change in social ties and subsequent mortality in rural elders. Epidemiol.. 1997;8:475-481.[Medline]
  10. Horsten M, Mittleman MA, Wamala SP, Schenck-Gustafsson K, Orth-Gomer K. Social relations and the metabolic syndrome in middle-aged Swedish women. J Cardiovasc Risk.. 1999;6:391-397.[Medline]
  11. Cohen S, Doyle WJ, Skoner DP, Rabin BS, Gwaltney JM, Jr. Social ties and susceptibility to the common cold. JAMA.. 1997;277:1940-1944.[Abstract/Free Full Text]
  12. Levenstein S, Smith MW, Kaplan GA. Psychosocial predictors of hypertension in men and women. Arch Intern Med.. 2001;161:1341-1346.[Abstract/Free Full Text]
  13. Ell K. Social networks, social support, and health status: a review. Soc Serv Rev.. 1984;58:133-149.
  14. Eng PM, Rimm EB, Fitzmaurice G, Kawachi I. Social ties and change in social ties in relation to subsequent total and cause-specific mortality and coronary heart disease incidence in men. Am J Epidemiol.. 2002;155:700-709.[Abstract/Free Full Text]
  15. Vogt TM, Mullooly JP, Ernst D, Pope CR, Hollis HF. Social networks as predictors of ischemic heart disease, cancer, stroke and hypertension: incidence, survival and mortality. J Clin Epidemiol.. 1992;45:659-666.[Medline]
  16. Tinetti ME, Speechly M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med.. 1988;319:1701-1707.[Abstract]
  17. Nevitt MC, Cumming SR, Kidd S, Black D. Risk factors for recurrent non-syncopal falls. JAMA.. 1989;261:2663-2668.[Abstract/Free Full Text]
  18. Fletcher PC, Hirdes JP. Risk Factors for falling among community-based seniors using home care services. J Gerontol Med Sci.. 2002;57A:M504-M510.[Abstract/Free Full Text]
  19. Whooley MA, Kip KE, Cauley JA, Ensrud KE, Nevitt MC, Browner WS. Depression, falls, and fracture. Arch Intern Med.. 1999;159:484-490.[Abstract/Free Full Text]
  20. King MB, Tinetti ME. Falls in community-dwelling older persons. J Am Geriatr Soc.. 1995;43:1146-1154.[Medline]
  21. Cwikel J. Falls among elderly people living at home: medical and social factors in a national sample. Isr J Med Sci.. 1992;28:446-453.[Medline]
  22. Studentski S, Duncan PW, Chandler J, et al. Predicting falls: the role of mobility and nonphysical factors. J Am Geriatr Soc.. 1994;42:297-302.[Medline]
  23. Campbell AJ, Reinken J, Allan BC, Martinez GS. Falls in old age: a study of frequency and related clinical factors. Age Aging.. 1981;10:264-270.[Abstract/Free Full Text]
  24. Berkman LF, Glass T, Brissette I, Seeman TE. From social integration to health: Durkheim in the new millennium. Soc Sci Med.. 2000;51:843-857.
  25. Cummings SR, Browner WS, Bauer DC, et al. Risk factors for hip fracture in white women. N Engl J Med.. 1995;332:767-773.[Abstract/Free Full Text]
  26. Lubben JE. Assessing social networks among elderly populations. Family Commun Health.. 1988;11:42-52.
  27. Pfieffer E. A short portable mental status questionnaire for the assessment of organic brain deficits in elderly patients. J Am Geriatr Soc.. 1975;23:433-441.[Medline]
  28. Folstein MF, Folstein SE. Mini-Mental State: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res.. 1975;12:189-198.[Medline]
  29. Lichtenstien MJ, Bess FH, Logan SA. Validation of screening tools for identifying hearing-impaired elderly in primary care. JAMA.. 1988;259:2875-2878.[Abstract/Free Full Text]
  30. Pincus T, Summey JA, Soraci S, Jr, Wallston KA, Hummon NP. Patient satisfaction in Activities of Daily Living using a modified Stanford Health Assessment Questionnaire. Arthrit Rheum.. 1983;26:1346-1353.[Medline]
  31. Yesavage JA. Geriatric depression scale. Psychopharmacol Bull.. 1988;24:709-711.[Medline]
  32. Lyness JM, Noel TK, Cox C, King DA, Conwell Y, Caine ED. Screening for depression in elderly primary care patients. Arch Intern Med.. 1997;157:449-454.[Abstract/Free Full Text]
  33. Gibson JJ. The Perception of the Visual World. Boston: Houghton Mifflin; 1950.
  34. Bailey IL, Lovie JE. New design principles for visual acuity letter charts. Am J Optom Physiol Opt.. 1976;53:740-745.[Medline]
  35. United States Pharmacopoeia Dispensing Information (USPDI).. Vol. 1. Appendix VII. 22nd Ed. Greenwood Village, CO: Micromedix Thompson Healthcare;. .; 2002:3033-3050.
  36. Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika.. 1986;73:13-22.[Abstract/Free Full Text]
  37. StataCorp. Stata Statistical Software: Release 7.0. College Station, TX: Stata Corporation; 2001.



This article has been cited by other articles:


Home page
 Am J EpidemiolHome page
B. K. S. Chan, L. M. Marshall, K. M. Winters, K. A. Faulkner, A. V. Schwartz, and E. S. Orwoll
Incident Fall Risk and Physical Activity and Physical Performance among Older Men: The Osteoporotic Fractures in Men Study
Am. J. Epidemiol., March 15, 2007; 165(6): 696 - 703.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
PubMed
Right arrow PubMed Citation

HOME ARCHIVE SEARCH TABLE OF CONTENTS