- Email: [email protected]
Social Frailty and Functional Disability: Findings From the Singapore Longitudinal Ageing Studies
JAMDA 18 (2017) 637.e13e637.e19
JAMDA journal homepage: www.jamda.com
Original Study
Social Frailty and Functional Disability: Findings From the Singapore Longitudinal Ageing Studies Nigel Teo BSocSci (Hons) a, Qi Gao PhD b, Ma Shwe Zin Nyunt PhD b, Shiou Liang Wee PhD a, c, Tze-Pin Ng MD a, b, * a
Geriatric Education and Research Institute, Singapore Gerontology Research Programme, Department of Psychological Medicine, National University of Singapore, Singapore c Health and Social Sciences Cluster, Singapore Institute of Technology, Singapore b
a b s t r a c t Keywords: Frailty disability long-term care social activity social relationship
Background/Objective: To examine the association between the social frailty (SF) phenotype and functional disability, independently of the physical frailty (PF) phenotype, and compare the abilities of the PF, SF, and combined social and physical (PSF) indexes for predicting functional disability. Method: Cross-sectional and longitudinal analyses of a population-based cohort (Singapore Longitudinal Ageing Study, SLAS-1) of 2406 community-dwelling older adults with 3 years of follow-up (N ¼ 1254 and N ¼ 1557 for instrumental activity of daily living (IADL) disability and severe disability (3 basic ADL) respectively). Measurements: Seven-item social frailty index (living arrangements, education, socioeconomic status, and social network and support, 0 ¼ nil SF, 1 ¼ low, 2e7 ¼ high), PF phenotype (Fried criteria), and instrumental activities of daily living (IADLs) disability and severe disability (3 basic ADLs). Results: Compared to nil SF, low and high SF were significantly associated with 1.3 to 2.4 fold increased prevalence and incidence of IADL disability, and 6.3 fold increase in severe disability. Frail individuals with and without SF stood out with 5-11 fold increased prevalence and incidence of IADL disability and 21-25 fold increased prevalence and incidence of severe disability, compared to robust individuals without SF. A combined PSF index more accurately identified individuals with increased risk of functional disability (ROC ¼ 64%) and severe disability (ROC ¼ 81%) than either the SF or the PF indexes alone (55% to 68%). Conclusion: The SF index alone or in combination with the PF index has clinical relevance and utility for identifying and stratifying older people at risk of disability. The mental frailty construct is closely related to SF and should be further investigated in future studies. Ó 2017 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Frailty is a state of increased vulnerability due to adverse health outcomes related to aging.1 It is recognized as a multidimensional construct comprising physical, psychological, and social domains.2,3 The physical frailty (PF) phenotype is most widely described and has been shown in many studies to predict adverse health outcomes, such as disability, hospitalization, and mortality.4,5 Social frailty (SF), on the
The authors declare no conflicts of interest. This study was funded by research grants from the Agency for Science Technology and Research (A*STAR) Biomedical Research Council (BMRC) (Grant 08/1/21/ 19/567) and from the National Medical Research Council (Grant NMRC/1108/2007). The sponsors had no role in the conduct of the study or preparation of this manuscript. * Address correspondence to Tze-Pin Ng, MD, Gerontology Research Programme, Department of Psychological Medicine, National University of Singapore, NUHS Tower Block, 9th Floor, 1E Kent Ridge Road, Singapore 119228, Singapore. E-mail address: [email protected] (T.-P. Ng). http://dx.doi.org/10.1016/j.jamda.2017.04.015 1525-8610/Ó 2017 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
other hand, is the least explored and understood. The relevance, validity, and utility of the SF construct has not been made clear. Some studies have narrowly defined SF in terms of the lack of participation in social networks and perceived lack of contacts and support. However, a variety of facets of social vulnerability have been used to define the concept.6 Thus, a multifaceted concept proposed by Bunt et al1 is that SF is a continuum of being at risk of losing, or having lost general or social resources, social behaviors and activities, and self-management abilities that are important for fulfilling basic social need(s). The SF construct has salient relevance to the development of functional disability. The process of disablement postulates the interactions among biomedical, behavioral, and social-environmental factors in producing functional disability as a terminal outcome.7 SF may thus have utility on its own or within a multidomain context for assessing the risk of disability and the need for nursing home care,
637.e14
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
thus helping in preemptive interventions to relieve the burden of care in aging societies and promote healthy aging. However, few studies have determined the predictive value of the SF phenotype vis-à-vis the PF phenotype, and their combined ability to predict adverse outcomes. A study of community-dwelling seniors in Italy reported that the SF phenotype using a deficit accumulation model significantly predicted mortality risks differentially from the PF phenotype.8 In another study among Japanese community-dwelling seniors, a 7-item SF index was shown to predict disability onset, independent from the PF index.9 The Canadian Study of Health and Aging showed that an index of social vulnerability operationalized by the deficit accumulation approach was associated with increased mortality.6 It was related to but was also distinct from the cumulated deficit multidomain frailty index. Our previous studies have shown that the PF phenotype strongly predicted functional disability and other adverse health outcomes.10,11 In this study, we developed a 7-item SF index based on multiple facets of general or social resources, and social behaviors and activities. Using cross-sectional and longitudinal data of community-dwelling seniors in the Singapore Longitudinal Ageing Study, we examined the association between the SF phenotype and functional disability, independently of the PF phenotype; we compared its predictive value for functional disability with the PF phenotype, and determined whether the combined social and physical phenotype substantially increased the ability to predict functional disability, and finally assessed its clinical utility using data in the population-based sample. Methods Participants We used data collected from the Singapore Longitudinal Ageing Studies Wave 1 (SLAS-1) cohort, a population-based longitudinal study of aging and health of community-dwelling older Singaporeans aged 55 and older, excluding individuals with severe physical or mental disabilities. As detailed previously,12 participants were first recruited in 2003 or 2005 and have completed 2 approximately 3yearly interval follow-ups, up to December 31, 2009. Baseline data collected includes demographic, medical, behavioral, biological, psychosocial, and neurocognitive characteristics via questionnaires, interviews, and physical or cognitive assessments. The study was approved by the National University of Singapore Institutional Review Board, and written informed consent was obtained from all participants. From the total of 2804 older adults who were recruited at baseline, we conducted analysis on the data of 2406 Chinese participants with complete baseline data on the designated variables. Longitudinal analysis was performed on 1254 participants who were free of instrumental activities of daily living (IADL) or severe disability (3 or more activities of daily living [ADL] dependencies) at baseline (n ¼ 1577) and had complete follow-up data on IADLs and severe disability. Baseline Measurements SF at baseline was operationalized and assessed through sociodemographic variables and self-reported survey questionnaires and based on the following criteria: (1) Living alone: assessed through the question: “Who do you live with?” (Alone or with others). (2) No education: assessed through the question: “What is your education level?” (Nil, Primary, Secondary or Institute of Technical Education, Pre-University or Polytechnic and University).
(3) Absence of a confidant: assessed through the question: “Do you have someone to confide in?” (Yes or No). (4) Infrequent contact: assessed through 3 questions enquiring the frequency of visits or calls by the individual’s family, friends, or loved ones, and perceived extent of help that can be obtained in their time of need: the presence of any one or more of none or no more than once a year visits from family, friends, or loved ones; none or no more than once a year calls from family, friends, or loved ones; or none to a very little extent of help when they require it. (5) Infrequent social activities: assessed through a series of questions on the number and frequencies of usual participation on 6 categories of social activities using a 3-point Likert scale. Participants who have indicated that they rarely or do not at all participate in all categories of social activities are considered to have this risk indicator. (6) Financial difficulty: assessed through the question: “Are you limited by your financial resources to pay for needed medical service?” with a 3-point Likert scale. This SF indicator was deemed to be present for participants who indicate that they were limited “to a great extent.” (7) Socioeconomic deprivation: assessed by proxy of the participant’s housing type, which has been previously validated to be reliable in predicting readmission risk and increased utilization of hospital services in Singapore.13 Participants who lived in “1to 2-room flats” were deemed to have this SF indicator. Scores were assigned to each SF indicator (1 ¼ present, 0 ¼ absent), and the summed scores were used to categorize individuals as having high score (2e7), low score (1), and nil score (0 point) on the SF index. Physical frailty at baseline was assessed based on the criteria used in the Cardiovascular Health Study, with operational modifications as detailed in previous publications.10,11 Scores were assigned to each of the 5 frailty components (1 ¼ present, 0 ¼ absent), and the summed scores were used to categorize subjects as frail (score ¼ 3e5), prefrail (score ¼ 1 or 2), and robust (score ¼ 0 point). These modified categorical criteria have been shown in previous studies to predict IADLADL disability, depression, hospitalization, and poor quality of life.10,11 Functional Disability Inability to perform IADLs or basic ADLs was assessed by self-report measures of IADL and ADL.14,15 The presence of functional disability was indicated by the requirement for help on 1 or more IADL or ADL items. Severe disability was denoted by dependency on 3 or more ADL items, which in Singapore often necessitates formal help in nursing home care placement and qualifies for disability insurance payout. Other Variables Sociodemographic data included age and gender. Medical comorbidity was determined through self-reported responses to a checklist of whether participants did or did not have a doctor’s diagnosis and treatment for each of 16 specified or other medical condition(s) in the past year, and estimating the total number of medical conditions. Seniors with 3 or more medical conditions were considered to have medical comorbidity. Lifestyle variables included self-reports of current or history of smoking and daily alcohol drinking. Depressive symptoms were determined by the Geriatric Depression Scale (GDS), which has been validated for use on Singaporean Chinese, Malay, and Indian individuals.16 The presence of depressive symptoms was operationalized as having a GDS score of 5 or more. Cognitive function was determined using scores of the Chinese version of the Mini-Mental State Examination (CMMSE), with total scores ranging from 0 to 30 (higher scores indicating better cognition). This test has been validated for local use in Singaporean older adults.17 Cognitive impairment was defined as having a MMSE score equivalent to or lower than 23. Hospitalization and
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
637.e15
Table 1 Baseline Characteristics of Chinese Older Adults by Social Frailty (SF) Indicators (n ¼ 2406) Variables
Total N ¼ 2406 (%)
Nil SF (0) N ¼ 1107 (%)
Low SF (1) N ¼ 856 (%)
High SF (2-7)* N ¼ 443 (%)
Live Alone No Education No confidant Infrequent contact (visits, calls or request for help) Infrequent social activities Financial Difficulty (To a great extent) Social economic deprivation (1-2 room flats & others) Social Frailty Index Count (SD)
175 458 119 490 331 235 151 0.81
0 0 0 0 0 0 0 0
61 217 27 246 176 98 31 1.00
114 241 92 244 155 137 120 2.49
(7.3) (19.0) (4.9) (20.4) (13.8) (9.8) (6.3) (0.98)
(0) (0) (0) (0) (0) (0) (0) (0)
(7.1) (25.3) (3.2) (28.7) (20.6) (11.5) (3.6) (0.00)
(25.7) (54.4) (20.8) (55.1) (35.0) (30.9) (27.1) (0.81)
(1) Infrequent contact refers to participants who have either indicated that they have (a) none or once a year visits from family, friends or loved ones; (b) none or once a year calls from family, friends or loved ones or (c) none to a very little extent of help when they require it. If the participants indicate that they have any one of the three, they qualify for this criterion. (2) Infrequent social activities refers to participants who have indicated that they rarely or do not at all participate in all of the six social activities stated in the social activities questionnaire found in the SLAS-1 Questionnaire. These six activities are: (1) Attendance at any religious service; (2) visits to cinemas, restaurants or sports events; (3) day or excursion trips; (4) if they play cards, games, bingo, mahjong; (5) if they attend senior citizen club activities or (6) if they attend social group activities. *The range of scores observed for SF statuses within this sample was from 0 e 6. No one obtained the full 7-point SF score.
physician visits were determined by participants’ self-reports of new hospitalizations and physician visits for each of the 16 listed major and other chronic medical conditions over the previous year. Polypharmacy was defined as taking 6 or more medications and determined through participants’ self-reports of the medication they took. Poor self-rated health status was assessed via the question, “In general, would you say your health is excellent, very good, good, fair, poor?” This question has been validated in numerous studies to be highly predictive of health, functional, and mortality outcomes. Hearing impairment was assessed using self-report and the standard whisper test, whereas visual impairment was defined as having corrected binocular vision worse than 20/40, as used in other studies.18 Statistical Analysis Pearson c2 and 1-way analysis of variance (ANOVA) tests were used to test the differences in baseline characteristics between
participants among the 3 different SF categories. The odds ratio (OR) of the association between SF and IADL and severe disability was estimated using a series of hierarchical logistic regression models. The first model (model 1) was adjusted for age, sex, and marital status. The second model (model 2) was adjusted for age, sex, and marital status and PF status. The third and final model (model 3) adjusted for age, sex, and marital status, PF, and baseline CMMSE score, depressive symptoms, and poor self-rated health. Longitudinal analysis was based on participants without IADL dependency and severe disability at baseline, wherever appropriate. We compared the abilities of SF index, PF index, and the combined social and PF index in predicting IADL and severe disability. We estimated the ORs of association and their 95% confidence intervals (95% CI) for 6 different categories of PF and SF combinations: (1) robust without SF, (2) robust with SF, (3) prefrail without SF, (4) prefrail with SF, (5) frail without SF, and (6) frail with SF. The presence of SF was indicated by SF score of 2 or more. Estimated ORs were adjusted as
Table 2 Baseline Characteristics of SLAS-1 Chinese Older Adults (n ¼ 2406) Variables
Total N ¼ 2406 (%)
Nil SF (0) N ¼ 1107 (%)
Low SF (1) N ¼ 856 (%)
High SF (2-7) N ¼ 443 (%)
P value
Age, mean (SD) y* Male Single, Divorced Or Widowed Functional Dependency (ADL/IADL Dependency) ADL Dependency (1 ADL) Severe Dependency (3 ADL) IADL Dependency (1 IADL) Cognitive impairment (MMSE<23) (N ¼ 2405) Depressive symptoms (GDS5) MMSE score ,mean (SD) (N ¼ 2405)y Poor Self-Reported Health Status (N ¼ 2397) Co-morbidity (3 or more) Hospitalization for any condition in past year Poly-pharmacy (taking 6 or more drugs) Physician Visit for any condition in the past year Arthritis Stroke Hip Fracture Mental Illness (N ¼ 2401) Hearing Impairment Visual Impairment Current smoking (versus others) Current and past smoking (N ¼ 2404) Alcohol drinking Obese (BMI 27.5) (N ¼ 2404)
66.1 882 610 597 147 31 537 284 328 27.0 59 1025 96 255 4.06 325 89 20 42 66 815 154 405 259 316
64.9 464 167 225 48 7 198 40 105 28.0 15 447 36 105 3.90 129 33 5 17 27 359 63 168 144 139
66.0 282 223 215 49 7 191 103 126 27.0 23 379 46 104 4.20 134 30 8 9 14 289 50 130 79 117
69.2 136 220 157 50 17 148 141 97 24.5 21 199 14 46 4.21 62 26 7 16 25 167 41 107 36 60
<.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 .123 .036 .162 .389 .034 .023 .080 .003 <.001 .140 .025 <.001 .004 .732
(7.63) (36.7) (25.4) (24.8) (6.1) (1.3) (22.3) (11.8) (13.6) (3.47) (2.5) (42.6) (4.0) (10.6) (5.38) (13.5) (3.7) (0.8) (1.7) (2.7) (33.9) (6.4) (16.8) (10.8) (13.1)
(6.94) (41.9) (15.1) (20.3) (4.3) (0.6) (17.9) (3.6) (9.5) (2.18) (1.4) (40.4) (3.3) (9.5) (4.49) (11.7) (3.0) (0.5) (1.5) (2.4) (32.4) (5.7) (15.2) (13.0) (12.6)
Significance Tests: One-Way ANOVA for continuous variables, Chi-Square for Categorical Variables. *,yPost-hoc tests show significant difference between Nil, Low and High SF Groups at P <.05 level.
(7.77) (32.9) (26.1) (25.1) (5.7) (0.8) (22.3) (12.0) (14.7) (3.18) (2.7) (44.3) (5.4) (12.2) (5.04) (15.7) (3.5) (0.9) (1.1) (1.6) (33.8) (5.9) (15.2) (9.2) (13.7)
(8.10) (30.7) (49.7) (35.4) (11.3) (3.8) (33.4) (31.9) (21.9) (5.04) (4.8) (44.9) (3.2) (10.4) (7.61) (14.0) (5.9) (1.6) (3.6) (5.6) (37.7) (9.3) (24.2) (8.1) (13.6)
637.e16
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
Table 3 Baseline Physical Frailty Status by Social Frailty Status Nil SF Low SF High SF Notes N ¼ 1107 (%) N ¼ 856 (%) N ¼ 443 (%) Sample N Physically Robust 643 (N ¼ 1240) Physically Pre-frail 446 (N ¼ 1084) Physically Frail (N ¼ 82) 18 Sample proportions (%) Physically Robust 26.7 Physically Pre-frail 18.5 Physically Frail 0.7 Prevalence of physical frailty (column %) Physically Robust 58.1 Physically Pre-frail 40.3 Physically Frail 1.6
433
164
398
240
25
39
18.0 16.5 1.0
6.8 10.0 1.6
50.6 46.5 2.9
37.0 54.2 8.8
P <.001
appropriate for age, sex, education, medical comorbidity, current smoking, alcohol drinking, depressive symptoms, baseline CMMSE scores. Receiver operating characteristics (ROC) analyses were used to compare the abilities among PF, SF, and combined PF-SF phenotypes in predicting IADL and severe disability at baseline and at a 2-year follow-up, in cross-sectional and longitudinal analyses. A 2-sided P value of .05 was considered as statistically significant in the study. All the analysis was performed using StataCorp, Stata, version 14.0 (StataCorp, College Station, TX). Results The mean age of the study sample was 66.1 (7.63 SD), 63.3% were women, and 25.4% were single, divorced, or widowed at baseline.
Overall, 18.4% in the study sample were deemed to have a high SF, 35.6% low, and 46.0% had nil SF, whereas 3.4% were physically frail, 45.1% were prefrail, and 51.5% were robust overall. The baseline characteristics and follow-up outcomes for older adults with nil, low, and high SFs are shown in Tables 1e3. Compared with participants having nil SF, those with low and high SFs were more likely to report poor health; to be women; to be single, divorced, or widowed; more hospitalizations; and to be physically frail; and also to have a higher proportion of current and past smokers but a lower proportion of current drinkers. A greater prevalence of stroke, arthritis, mental illness, cognitive, hearing and depressive symptoms at baseline was also observed among those who had low and high SFs. However, there were no significant associations with comorbidity, physician visits, and polypharmacy.
SF and IADL and Severe Disabilities The associations between SF and IADL and severe disabilities in cross-sectional and longitudinal analyses are shown in Table 4. In cross-sectional analyses, significant trends in ORs of association with prevalent IADL disability were observed for low SF (OR ¼ 1.32) and high SFIs (OR ¼ 2.30) compared to nil SF; and with severe disability for low SF (OR ¼ 1.30) and high SF (OR ¼ 6.27). After adjustment for physical frailty and other covariates, the same trend of associations were observed, and high SF remained significantly associated with IADL dependency. However, further adjustment for baseline MMSE, depression, poor self-reported health greatly reduced the ORs of all associations. In longitudinal analyses, the same trend of associations were observed, and high SF groups were significant in predicting for incident IADL dependency. Further adjustments of physical frailty and other covariates greatly reduced this association.
Table 4 Cross-sectional Analyses of Associations of Baseline Social Frailty Status With Prevalent and Incident Functional Dependency IADL Dependency
Cross-sectional Analysis (N ¼ 2406) Unadjusted Nil Low High Linear Trend Model 1 Low High Linear Trend Model 2 Low High Linear Trend
Severe Dependency (3 BADL)
N
%
OR (95%CI)
198/1107 191/856 148/443
17.9 22.3 33.4
1.00 1.32 (1.06-1.65) 2.30 (1.79-2.96)
P
.015 <.001 <.001
Model Model Model Model
38/629 23/439 25/186
%
OR (95%CI)
7/1107 7/856 17/443
0.6 0.8 3.8
1.00 1.30 (0.45-3.71) 6.27 (2.58-15.23)
P
.629 <.001 <.001
1.24 (0.97-1.58) 1.61 (1.07-1.10)
.090 .002 .002
0.93 (0.31-2.79) 2.37 (0.87-6.50)
.898 .093 .081
1.11 (0.86-1.42) 1.18 (0.86-1.61)
.428 .316 .286
0.81 (0.24-2.72) 1.12 (0.33-3.75)
.734 .857 .840
1254 Longitudinal analysis Unadjusted Nil Low High Linear Trend Model 1a Low High Linear Trend Model 2a Low High Linear Trend
N
1557
6.0 5.2 13.4
1.00 0.86 (0.50-1.46) 2.41 (1.42-4.12)
.579 .001 .008
1/761 3/543 2/253
0.1 0.6 0.8
1.00 4.22 (0.44-40.70) 6.06 (0.55-67.07)
.213 .142 .118
0.70 (0.40-1.23) 1.56 (0.87-2.80)
.215 .137 .291
2.98 (0.29-30.59) 1.31 (0.09-19.28)
.358 .846 .845
0.63 (0.36-1.13) 1.32 (0.70-2.50)
.120 .386 .653
3.30 (0.27-40.15) 2.15 (0.12-38.21)
.350 .602 .600
1: [Age, sex, Marital Status] þ Physical Frailty. 2: [Age, sex, Marital Status, Physical Frailty] þ Poor Self-reported Health, Baseline MMSE Score and Depression. 1a: [Age, sex, Marital Status, IADL Baseline measure] þ Physical Frailty. 2a: [Age, sex, Marital Status, IADL Baseline measure, Physical Frailty] þ Poor Self-reported Health, Baseline MMSE Score and Depression.
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
637.e17
Table 5 Associations of Baseline Combined SF and PF Phenotype With Prevalent and Follow up Functional Disability IADL Dependency N Cross-sectional Analysis Robust without SF Robust with SF Pre-frail without SF Pre-frail with SF Frail without SF Frail with SF Linear Trend Robust (with or without SF) Pre-frail or Frail without SF Pre-frail or frail with SF Linear Trend Longitudinal Analysis N Robust without SF Robust with SF Pre-frail without SF Pre-frail with SF Frail without SF Frail with SF Linear Trend Robust (with or without SF) Pre-frail or Frail without SF Pre-frail or frail with SF Linear Trend
Severe Dependency (3 BADL)
%
OR (95%CI)
2403 158/1076 23/164 202/844 91/240 29/43 34/39
14.7 14.0 23.9 37.9 67.4 87.2
1.00 0.83 1.43 1.60 4.57 11.15
181/1240 231/887 125/279
14.6 26.0 44.8
1.00 1.56 (1.23-1.99) 1.96 (1.39-2.76)
(0.50-1.37) (1.11-1.84) (1.10-2.31) (2.19-9.51) (3.98-31.2)
P
.465 .006 .013 <.001 <.001 <.001 <.001 <.001 <.001
1253
N
%
OR (95%CI)
2403 2/1076 0/164 2/844 9/240 10/43 8/39
0.2 0.0 0.2 3.8 23.3 20.5
1.00 0.00 1.06 5.42 69.91 25.20
2/1240 12/887 17/279
0.2 1.4 6.1
1.00 5.33 (1.14-24.79) 5.80 (1.13-29.80)
1/731 0/105 1/551 1/137 2/21 1/11
0.1 0.0 0.2 0.7 9.5 9.1
1.00 0.00 0.77 2.78 27.60 21.13
1/837 3/572 2/148
0.1 0.5 1.4
1.00 2.05 (0.19-21.67) 3.20 (0.21-47.87)
(0.00) (0.14-7.65) (0.97-30.30) (12.57-388.9) (3.77-168.5)
P
1.00 .957 .054 <.001 .001 <.001 .033 .035 .042
1556
24/627 9/89 34/432 15/95 3/9 1/2
3.8 10.1 7.9 15.8 33.3 50.0
1.00 2.43 1.79 2.55 5.65 5.71
33/716 37/441 16/97
4.6 8.4 16.5
1.00 1.57 (0.94-2.61) 2.07 (1.01-4.23)
(1.04-5.67) (1.02-3.14) (1.19-5.48) (1.21-26.5) (0.22-147.5)
.040 .043 .016 .028 .293 .003 .084 .046 .024
(0.00) (0.04-13.08) (0.13-58.06) (1.44-525.80) (0.66-679.03)
.00 .85 .46 .027 .085 .029 .551 .399 .393
SF ¼ 2 or more positive Social Frailty indicators. Adjusted for age, gender, medical comorbidity, current smoking, alcohol drinking, depressive symptom as well as baseline MMSE scores.
Relationships Between SF and PF Phenotypes Table 3 shows that although the SF index was strongly associated with the PF index (P < .001), there was no complete overlap in the categories of the SF index with the PF index (kappa ¼ 0.076). Among participants with high SF, for example, 63% were prefrail or frail, and 37% were robust. Among those with no SF, 42% were either prefrail or frail, and only 58% were physically robust.
Combined PF and SF Phenotype and IADL and Severe Dependencies Table 5 shows the results of cross-sectional and longitudinal analyses of the associations between prefrailty and frailty with and without high SF (2 SF indicators), and prevalent and incident IADL or severe disability, using physically robust participants without SF as the reference group. IADL Disability In cross-sectional analyses, the OR estimates were increased among the prefrail and frail without SF, but were greater in the presence of SF; prefrailty or frail without SF: 1.56 (1.23e1.99), prefrail or frail with SF: 1.96 (1.39e2.76) for prevalent disability. In Table 6 Receiver Operator Characteristics of PFI, SFI and PSFI in Predicting IADL and Severe Disabilities
longitudinal analyses, the estimates for pre-frail or frail with SF were significantly elevated for incident disability (OR ¼ 2.07 (1.01e4.23)). Even much higher ORs for IADL disability were found among those who were both physically and socially frail: OR ¼ 11.15 (3.98e31.20) for prevalent IADL disability, and OR ¼ 5.71 (0.22e147.50) for incident IADL disability. Severe Disability Similar trends of associations for severe disability were found in cross-sectional and longitudinal analyses. In cross-sectional analyses, relative to the robust without SF, the presence of SF increased the prevalence and OR of association with severe disability from 0.2% to 3.8%, (OR 1.06 to 5.42) among the pre-frail; physically frail participants with and without SF had especially high likelihood (over 20%) of prevalent severe disability. In longitudinal analyses, the risks of incident severe disability, compared to robust without SF (0.1%) were clearly highest (over 9%) in the groups who were frail with and without SF. Receiver Characteristic Analyses The abilities of PF, SF, and PSF indexes to predict prevalent and incident IADL and severe disability are shown in Table 6, Figures 1 and 2. Satisfactory areas under the ROC were observed for PF index and SF index alone, especially for severe disability, but PSF index stood out with the highest area under the ROC (81% to 91%) for predicting prevalent and incident IADL and severe disability.
Areas Under ROC Cross-sectional Analysis
Social Frailty Index (SFI) Physical Frailty Index (PFI) Physical and Social Frailty Index (PSFI)
Longitudinal Analysis
Discussion Our study shows that the SF is a valid construct, and is able to predict older individuals who are at risk of becoming functionally disabled. We found that socially frail older adults compared with their counterparts were more likely to present with or develop functional disability, independently of PF. Our results, however, indicate that the association of SF with functional disability was not independent of
IADL
Severe ADL
IADL
Severe ADL
0.580 0.637 0.651
0.696 0.855 0.901
0.565 0.608 0.642
0.679 0.794 0.810
ROC ¼ Receiver Operator Characteristics.
637.e18
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
Fig. 1. (A) Cross-sectional analysis, IADL disability. (B) Cross-sectional analysis, severe 3 basic BADL disabilities.
mental health variables such as MMSE, depressive symptoms, and poor self-reported health. This suggests that cognitive and psychological factors closely associated with SF are another set of factors representing mental frailty that have an impact on functional disability. This should be further investigated in future studies. SF was also closely related to PF, but they do not completely overlap. SF shared many risk factors in common with PF, but it was different in not being associated with factors such as comorbidity, physician visits, and polypharmacy. It is already well established that physically frail older persons are more likely to be found disabled, or to become disabled in time. This study shows that those who were both socially and physically frail were even more likely to be disabled or become disabled. Thus, although the SF index and the PF index independently and strongly predict IADL and severe disability, a combined physical and social index was able to identify individuals with much higher risk of functional and severe disability than either the SF index or the PF index alone. The results suggest that conceptually a multidomain phenotype of frailty is better able than a single-domain physical phenotype to
predict disability risk.8 Brief frailty scales, such as the Tilburg Frailty Index and the Edmonton Scale,19e21 based on a multidimensional construct, may thus have this advantage of increased global prediction for relevant specific outcomes such as disability. On the other hand, it is important to note that single-domain indices such as the PF phenotype or the SF phenotype may be specifically useful in targeting unique sets of risk factors and desired outcomes of interventions. Clinical Utility The screening and assessment of frailty among older persons with measurement tools that are predictive of future risks of functional disability is an important interventional approach to reduce the negative impact and burden of functional disability on health and social services. In this population sample, approximately 18.4% in the sample had high SF scores (positive on 2 to 7 social frailty indicators) (Table 3). As shown in Table 4, this group of individuals had a 33.4% absolute likelihood of being functionally disabled and 13.4% absolute likelihood of becoming functionally disabled, 2 times more than those
Fig. 2. (A) Longitudinal analysis, IADL disability. (B) Longitudinal analysis, severe 3 basic BADL disabilities.
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
without SF. Approximately 12% in the sample were both socially and physically frail (SFI2 and PFI1) (Table 3). They had a 16% to 44.8% absolute likelihood of being disabled or becoming disabled, 2 times more than those who were physically and socially robust. Furthermore, they had a 1.4% e 6% absolute likelihood of being or becoming severely disabled on 3 or more basic ADLs, 3 e 6 times more than those who were physically and socially robust. Strengths and Limitations The study has strengths and limitations. There was good content and construct validity in the SF index from embedding it in a welldefined theoretical framework for including salient social factors that are relevant to social situations and health outcomes. However, SF factors were based on self-report rather than objectively measured. Perceived social support was assessed rather than instrumental social support, and may have a differing impact on adverse health outcomes.22 The type of social support or engagement the older adult received did not differentiate family, friends, or formal service providers. Further research identifying the effects on the differing types of relationships, as well as instrumental social support and their effects on adverse health outcomes should be considered. The number of incident severe disability cases was small, thus limiting the statistical precision of the estimates of associations in longitudinal analyses. Nonresponders and dropouts from follow-up in the study were more likely to be socially or physically frail and therefore more likely to have higher mortality and poor functional outcomes. Thus, the OR estimates of association between the frailty indexes and functional disability are likely be underestimated and should be viewed as conservative. Conclusion SF is associated with increased likelihood of being and becoming disabled on daily living activities, independent of PF. Although individually, PF and SF are highly predictive of future risks of disability, their combined index has the highest predictive value for functional disability, especially severe disability. The SF index with or without the PF index has relevance and utility in health care and social service settings for identifying and stratifying the population at risk of disabilities, and implementing early interventions to reduce the burden of care and promoting healthy aging in the elderly population. Further research should be conducted to examine the mental frailty construct that is closely related to SF. The interrelationships among social, mental, and physical domains in a multidimensional frailty construct and its clinical relevance and utility should be further investigated. Acknowledgments We thank the following voluntary welfare organizations for their support: Geylang East Home for the Aged, Presbyterian Community
637.e19
Services, St Luke’s Eldercare Services, Thye Hua Kwan Moral Society (Moral Neighbourhood Links), Yuhua Neighbourhood Link, Henderson Senior Citizens’ Home, NTUC Eldercare Co-op Ltd, Thong Kheng Seniors Activity Centre (Queenstown Centre) and Redhill Moral Seniors Activity Centre. References 1. Bunt S, Steverink N, Olthof J, et al. Social frailty in older adults: A scoping review. Eur J Ageing. In press. 2. Rockwood K. What would make a definition of frailty successful? Age Ageing 2005;34:432e434. 3. Gobbens RJ, Luijkx KG, Wijnen-Sponselee MT, Schols JM. Toward a conceptual definition of frail community dwelling older people. Nurs Outlook 2010;58: 76e86. 4. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:146e157. 5. Ensrud KE, Ewing SK, Taylor BC, et al. Comparison of 2 frailty indexes for prediction of falls, disability, fractures, and death in older women. Arch Intern Med 2008;168:382e389. 6. Andrew MK, Mitnitski AB, Rockwood K. Social vulnerability, frailty and mortality in elderly people. PLoS One 2008;3:1e8. 7. Verbrugge LM, Jette AM. The disablement process. Soc Sci Med 1994;38:1e14. 8. Garre-Olmo J, Calvó-Perxas L, López-Pousa S, et al. Prevalence of frailty phenotypes and risk of mortality in a community-dwelling elderly cohort. Age Ageing 2013;42:46e51. 9. Makizako H, Shimada H, Tsutsumimoto K, et al. Social frailty in communitydwelling older adults as a risk factor for disability. J Am Med Dir Assoc 2015; 16:1003.e7e1003.e11. 10. Ng TP, Feng L, Nyunt MSZ, et al. Frailty in older persons: Multisystem risk factors and the Frailty Risk Index (FRI). J Am Med Dir Assoc 2014;15: 635e642. 11. Feng L, Zin Nyunt MS, Gao Q, et al. Cognitive frailty and adverse health outcomes: Findings from the Singapore Longitudinal Ageing Studies (SLAS). J Am Med Dir Assoc 2016;18:252e258. 12. Niti M, Yap KB, Kua EH, et al. Physical, social and productive leisure activities, cognitive decline and interaction with APOE-epsilon 4 genotype in Chinese older adults. Int Psychogeriatr 2008;20:237e251. 13. Low LL, Wah W, Ng MJ, et al. Housing as a social determinant of health in Singapore and its association with readmission risk and increased utilization of hospital services. Front Public Health 2016;4:109. 14. Mahoney FI, Barthel DW. Functional evaluation: The Barthel Index. Md State Med J 1965;14:61e65. 15. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:179e186. 16. Nyunt MS, Fones C, Niti M, Ng TP. Criterion-based validity and reliability of the Geriatric Depression Screening Scale (GDS-15) in a large validation sample of community-living Asian older adults. Aging Ment Health 2009;13: 376e382. 17. Feng L, Chong MS, Lim WS, Ng TP. The modified mini-mental state examination test: Normative data for Singapore Chinese older adults and its performance in detecting early cognitive impairment. Singapore Med J 2012;53:458e462. 18. Tielsch JM, Sommer A, Witt K, et al. the B.E.S.R. Group. Blindness and visual impairment in an American urban population. Arch Ophthalmol 1990;108: 286e290. 19. Gobbens RJ, van Assen MA, Luijkx KG, et al. The Tilburg frailty indicator: Psychometric properties. J Am Med Dir Assoc 2010;11:344e355. 20. Bergman H, Ferrucci L, Guralnik J, et al. Frailty: An emerging research and clinical paradigmdissues and controversies. J Gerontol A Biol Sci Med Sci 2007; 62:731e737. 21. Rolfson DB, Majumdar SR, Tsuyuki RT, et al. Validity and reliability of the Edmonton Frail Scale. Age Ageing 2006;35:526e529. 22. Malhotra R, Ang SJ. Social support for older adultsda bane or a boon for their health? Ann Acad Med Singapore 2016;45:172e173.
JAMDA journal homepage: www.jamda.com
Original Study
Social Frailty and Functional Disability: Findings From the Singapore Longitudinal Ageing Studies Nigel Teo BSocSci (Hons) a, Qi Gao PhD b, Ma Shwe Zin Nyunt PhD b, Shiou Liang Wee PhD a, c, Tze-Pin Ng MD a, b, * a
Geriatric Education and Research Institute, Singapore Gerontology Research Programme, Department of Psychological Medicine, National University of Singapore, Singapore c Health and Social Sciences Cluster, Singapore Institute of Technology, Singapore b
a b s t r a c t Keywords: Frailty disability long-term care social activity social relationship
Background/Objective: To examine the association between the social frailty (SF) phenotype and functional disability, independently of the physical frailty (PF) phenotype, and compare the abilities of the PF, SF, and combined social and physical (PSF) indexes for predicting functional disability. Method: Cross-sectional and longitudinal analyses of a population-based cohort (Singapore Longitudinal Ageing Study, SLAS-1) of 2406 community-dwelling older adults with 3 years of follow-up (N ¼ 1254 and N ¼ 1557 for instrumental activity of daily living (IADL) disability and severe disability (3 basic ADL) respectively). Measurements: Seven-item social frailty index (living arrangements, education, socioeconomic status, and social network and support, 0 ¼ nil SF, 1 ¼ low, 2e7 ¼ high), PF phenotype (Fried criteria), and instrumental activities of daily living (IADLs) disability and severe disability (3 basic ADLs). Results: Compared to nil SF, low and high SF were significantly associated with 1.3 to 2.4 fold increased prevalence and incidence of IADL disability, and 6.3 fold increase in severe disability. Frail individuals with and without SF stood out with 5-11 fold increased prevalence and incidence of IADL disability and 21-25 fold increased prevalence and incidence of severe disability, compared to robust individuals without SF. A combined PSF index more accurately identified individuals with increased risk of functional disability (ROC ¼ 64%) and severe disability (ROC ¼ 81%) than either the SF or the PF indexes alone (55% to 68%). Conclusion: The SF index alone or in combination with the PF index has clinical relevance and utility for identifying and stratifying older people at risk of disability. The mental frailty construct is closely related to SF and should be further investigated in future studies. Ó 2017 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
Frailty is a state of increased vulnerability due to adverse health outcomes related to aging.1 It is recognized as a multidimensional construct comprising physical, psychological, and social domains.2,3 The physical frailty (PF) phenotype is most widely described and has been shown in many studies to predict adverse health outcomes, such as disability, hospitalization, and mortality.4,5 Social frailty (SF), on the
The authors declare no conflicts of interest. This study was funded by research grants from the Agency for Science Technology and Research (A*STAR) Biomedical Research Council (BMRC) (Grant 08/1/21/ 19/567) and from the National Medical Research Council (Grant NMRC/1108/2007). The sponsors had no role in the conduct of the study or preparation of this manuscript. * Address correspondence to Tze-Pin Ng, MD, Gerontology Research Programme, Department of Psychological Medicine, National University of Singapore, NUHS Tower Block, 9th Floor, 1E Kent Ridge Road, Singapore 119228, Singapore. E-mail address: [email protected] (T.-P. Ng). http://dx.doi.org/10.1016/j.jamda.2017.04.015 1525-8610/Ó 2017 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
other hand, is the least explored and understood. The relevance, validity, and utility of the SF construct has not been made clear. Some studies have narrowly defined SF in terms of the lack of participation in social networks and perceived lack of contacts and support. However, a variety of facets of social vulnerability have been used to define the concept.6 Thus, a multifaceted concept proposed by Bunt et al1 is that SF is a continuum of being at risk of losing, or having lost general or social resources, social behaviors and activities, and self-management abilities that are important for fulfilling basic social need(s). The SF construct has salient relevance to the development of functional disability. The process of disablement postulates the interactions among biomedical, behavioral, and social-environmental factors in producing functional disability as a terminal outcome.7 SF may thus have utility on its own or within a multidomain context for assessing the risk of disability and the need for nursing home care,
637.e14
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
thus helping in preemptive interventions to relieve the burden of care in aging societies and promote healthy aging. However, few studies have determined the predictive value of the SF phenotype vis-à-vis the PF phenotype, and their combined ability to predict adverse outcomes. A study of community-dwelling seniors in Italy reported that the SF phenotype using a deficit accumulation model significantly predicted mortality risks differentially from the PF phenotype.8 In another study among Japanese community-dwelling seniors, a 7-item SF index was shown to predict disability onset, independent from the PF index.9 The Canadian Study of Health and Aging showed that an index of social vulnerability operationalized by the deficit accumulation approach was associated with increased mortality.6 It was related to but was also distinct from the cumulated deficit multidomain frailty index. Our previous studies have shown that the PF phenotype strongly predicted functional disability and other adverse health outcomes.10,11 In this study, we developed a 7-item SF index based on multiple facets of general or social resources, and social behaviors and activities. Using cross-sectional and longitudinal data of community-dwelling seniors in the Singapore Longitudinal Ageing Study, we examined the association between the SF phenotype and functional disability, independently of the PF phenotype; we compared its predictive value for functional disability with the PF phenotype, and determined whether the combined social and physical phenotype substantially increased the ability to predict functional disability, and finally assessed its clinical utility using data in the population-based sample. Methods Participants We used data collected from the Singapore Longitudinal Ageing Studies Wave 1 (SLAS-1) cohort, a population-based longitudinal study of aging and health of community-dwelling older Singaporeans aged 55 and older, excluding individuals with severe physical or mental disabilities. As detailed previously,12 participants were first recruited in 2003 or 2005 and have completed 2 approximately 3yearly interval follow-ups, up to December 31, 2009. Baseline data collected includes demographic, medical, behavioral, biological, psychosocial, and neurocognitive characteristics via questionnaires, interviews, and physical or cognitive assessments. The study was approved by the National University of Singapore Institutional Review Board, and written informed consent was obtained from all participants. From the total of 2804 older adults who were recruited at baseline, we conducted analysis on the data of 2406 Chinese participants with complete baseline data on the designated variables. Longitudinal analysis was performed on 1254 participants who were free of instrumental activities of daily living (IADL) or severe disability (3 or more activities of daily living [ADL] dependencies) at baseline (n ¼ 1577) and had complete follow-up data on IADLs and severe disability. Baseline Measurements SF at baseline was operationalized and assessed through sociodemographic variables and self-reported survey questionnaires and based on the following criteria: (1) Living alone: assessed through the question: “Who do you live with?” (Alone or with others). (2) No education: assessed through the question: “What is your education level?” (Nil, Primary, Secondary or Institute of Technical Education, Pre-University or Polytechnic and University).
(3) Absence of a confidant: assessed through the question: “Do you have someone to confide in?” (Yes or No). (4) Infrequent contact: assessed through 3 questions enquiring the frequency of visits or calls by the individual’s family, friends, or loved ones, and perceived extent of help that can be obtained in their time of need: the presence of any one or more of none or no more than once a year visits from family, friends, or loved ones; none or no more than once a year calls from family, friends, or loved ones; or none to a very little extent of help when they require it. (5) Infrequent social activities: assessed through a series of questions on the number and frequencies of usual participation on 6 categories of social activities using a 3-point Likert scale. Participants who have indicated that they rarely or do not at all participate in all categories of social activities are considered to have this risk indicator. (6) Financial difficulty: assessed through the question: “Are you limited by your financial resources to pay for needed medical service?” with a 3-point Likert scale. This SF indicator was deemed to be present for participants who indicate that they were limited “to a great extent.” (7) Socioeconomic deprivation: assessed by proxy of the participant’s housing type, which has been previously validated to be reliable in predicting readmission risk and increased utilization of hospital services in Singapore.13 Participants who lived in “1to 2-room flats” were deemed to have this SF indicator. Scores were assigned to each SF indicator (1 ¼ present, 0 ¼ absent), and the summed scores were used to categorize individuals as having high score (2e7), low score (1), and nil score (0 point) on the SF index. Physical frailty at baseline was assessed based on the criteria used in the Cardiovascular Health Study, with operational modifications as detailed in previous publications.10,11 Scores were assigned to each of the 5 frailty components (1 ¼ present, 0 ¼ absent), and the summed scores were used to categorize subjects as frail (score ¼ 3e5), prefrail (score ¼ 1 or 2), and robust (score ¼ 0 point). These modified categorical criteria have been shown in previous studies to predict IADLADL disability, depression, hospitalization, and poor quality of life.10,11 Functional Disability Inability to perform IADLs or basic ADLs was assessed by self-report measures of IADL and ADL.14,15 The presence of functional disability was indicated by the requirement for help on 1 or more IADL or ADL items. Severe disability was denoted by dependency on 3 or more ADL items, which in Singapore often necessitates formal help in nursing home care placement and qualifies for disability insurance payout. Other Variables Sociodemographic data included age and gender. Medical comorbidity was determined through self-reported responses to a checklist of whether participants did or did not have a doctor’s diagnosis and treatment for each of 16 specified or other medical condition(s) in the past year, and estimating the total number of medical conditions. Seniors with 3 or more medical conditions were considered to have medical comorbidity. Lifestyle variables included self-reports of current or history of smoking and daily alcohol drinking. Depressive symptoms were determined by the Geriatric Depression Scale (GDS), which has been validated for use on Singaporean Chinese, Malay, and Indian individuals.16 The presence of depressive symptoms was operationalized as having a GDS score of 5 or more. Cognitive function was determined using scores of the Chinese version of the Mini-Mental State Examination (CMMSE), with total scores ranging from 0 to 30 (higher scores indicating better cognition). This test has been validated for local use in Singaporean older adults.17 Cognitive impairment was defined as having a MMSE score equivalent to or lower than 23. Hospitalization and
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
637.e15
Table 1 Baseline Characteristics of Chinese Older Adults by Social Frailty (SF) Indicators (n ¼ 2406) Variables
Total N ¼ 2406 (%)
Nil SF (0) N ¼ 1107 (%)
Low SF (1) N ¼ 856 (%)
High SF (2-7)* N ¼ 443 (%)
Live Alone No Education No confidant Infrequent contact (visits, calls or request for help) Infrequent social activities Financial Difficulty (To a great extent) Social economic deprivation (1-2 room flats & others) Social Frailty Index Count (SD)
175 458 119 490 331 235 151 0.81
0 0 0 0 0 0 0 0
61 217 27 246 176 98 31 1.00
114 241 92 244 155 137 120 2.49
(7.3) (19.0) (4.9) (20.4) (13.8) (9.8) (6.3) (0.98)
(0) (0) (0) (0) (0) (0) (0) (0)
(7.1) (25.3) (3.2) (28.7) (20.6) (11.5) (3.6) (0.00)
(25.7) (54.4) (20.8) (55.1) (35.0) (30.9) (27.1) (0.81)
(1) Infrequent contact refers to participants who have either indicated that they have (a) none or once a year visits from family, friends or loved ones; (b) none or once a year calls from family, friends or loved ones or (c) none to a very little extent of help when they require it. If the participants indicate that they have any one of the three, they qualify for this criterion. (2) Infrequent social activities refers to participants who have indicated that they rarely or do not at all participate in all of the six social activities stated in the social activities questionnaire found in the SLAS-1 Questionnaire. These six activities are: (1) Attendance at any religious service; (2) visits to cinemas, restaurants or sports events; (3) day or excursion trips; (4) if they play cards, games, bingo, mahjong; (5) if they attend senior citizen club activities or (6) if they attend social group activities. *The range of scores observed for SF statuses within this sample was from 0 e 6. No one obtained the full 7-point SF score.
physician visits were determined by participants’ self-reports of new hospitalizations and physician visits for each of the 16 listed major and other chronic medical conditions over the previous year. Polypharmacy was defined as taking 6 or more medications and determined through participants’ self-reports of the medication they took. Poor self-rated health status was assessed via the question, “In general, would you say your health is excellent, very good, good, fair, poor?” This question has been validated in numerous studies to be highly predictive of health, functional, and mortality outcomes. Hearing impairment was assessed using self-report and the standard whisper test, whereas visual impairment was defined as having corrected binocular vision worse than 20/40, as used in other studies.18 Statistical Analysis Pearson c2 and 1-way analysis of variance (ANOVA) tests were used to test the differences in baseline characteristics between
participants among the 3 different SF categories. The odds ratio (OR) of the association between SF and IADL and severe disability was estimated using a series of hierarchical logistic regression models. The first model (model 1) was adjusted for age, sex, and marital status. The second model (model 2) was adjusted for age, sex, and marital status and PF status. The third and final model (model 3) adjusted for age, sex, and marital status, PF, and baseline CMMSE score, depressive symptoms, and poor self-rated health. Longitudinal analysis was based on participants without IADL dependency and severe disability at baseline, wherever appropriate. We compared the abilities of SF index, PF index, and the combined social and PF index in predicting IADL and severe disability. We estimated the ORs of association and their 95% confidence intervals (95% CI) for 6 different categories of PF and SF combinations: (1) robust without SF, (2) robust with SF, (3) prefrail without SF, (4) prefrail with SF, (5) frail without SF, and (6) frail with SF. The presence of SF was indicated by SF score of 2 or more. Estimated ORs were adjusted as
Table 2 Baseline Characteristics of SLAS-1 Chinese Older Adults (n ¼ 2406) Variables
Total N ¼ 2406 (%)
Nil SF (0) N ¼ 1107 (%)
Low SF (1) N ¼ 856 (%)
High SF (2-7) N ¼ 443 (%)
P value
Age, mean (SD) y* Male Single, Divorced Or Widowed Functional Dependency (ADL/IADL Dependency) ADL Dependency (1 ADL) Severe Dependency (3 ADL) IADL Dependency (1 IADL) Cognitive impairment (MMSE<23) (N ¼ 2405) Depressive symptoms (GDS5) MMSE score ,mean (SD) (N ¼ 2405)y Poor Self-Reported Health Status (N ¼ 2397) Co-morbidity (3 or more) Hospitalization for any condition in past year Poly-pharmacy (taking 6 or more drugs) Physician Visit for any condition in the past year Arthritis Stroke Hip Fracture Mental Illness (N ¼ 2401) Hearing Impairment Visual Impairment Current smoking (versus others) Current and past smoking (N ¼ 2404) Alcohol drinking Obese (BMI 27.5) (N ¼ 2404)
66.1 882 610 597 147 31 537 284 328 27.0 59 1025 96 255 4.06 325 89 20 42 66 815 154 405 259 316
64.9 464 167 225 48 7 198 40 105 28.0 15 447 36 105 3.90 129 33 5 17 27 359 63 168 144 139
66.0 282 223 215 49 7 191 103 126 27.0 23 379 46 104 4.20 134 30 8 9 14 289 50 130 79 117
69.2 136 220 157 50 17 148 141 97 24.5 21 199 14 46 4.21 62 26 7 16 25 167 41 107 36 60
<.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 <.001 .123 .036 .162 .389 .034 .023 .080 .003 <.001 .140 .025 <.001 .004 .732
(7.63) (36.7) (25.4) (24.8) (6.1) (1.3) (22.3) (11.8) (13.6) (3.47) (2.5) (42.6) (4.0) (10.6) (5.38) (13.5) (3.7) (0.8) (1.7) (2.7) (33.9) (6.4) (16.8) (10.8) (13.1)
(6.94) (41.9) (15.1) (20.3) (4.3) (0.6) (17.9) (3.6) (9.5) (2.18) (1.4) (40.4) (3.3) (9.5) (4.49) (11.7) (3.0) (0.5) (1.5) (2.4) (32.4) (5.7) (15.2) (13.0) (12.6)
Significance Tests: One-Way ANOVA for continuous variables, Chi-Square for Categorical Variables. *,yPost-hoc tests show significant difference between Nil, Low and High SF Groups at P <.05 level.
(7.77) (32.9) (26.1) (25.1) (5.7) (0.8) (22.3) (12.0) (14.7) (3.18) (2.7) (44.3) (5.4) (12.2) (5.04) (15.7) (3.5) (0.9) (1.1) (1.6) (33.8) (5.9) (15.2) (9.2) (13.7)
(8.10) (30.7) (49.7) (35.4) (11.3) (3.8) (33.4) (31.9) (21.9) (5.04) (4.8) (44.9) (3.2) (10.4) (7.61) (14.0) (5.9) (1.6) (3.6) (5.6) (37.7) (9.3) (24.2) (8.1) (13.6)
637.e16
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
Table 3 Baseline Physical Frailty Status by Social Frailty Status Nil SF Low SF High SF Notes N ¼ 1107 (%) N ¼ 856 (%) N ¼ 443 (%) Sample N Physically Robust 643 (N ¼ 1240) Physically Pre-frail 446 (N ¼ 1084) Physically Frail (N ¼ 82) 18 Sample proportions (%) Physically Robust 26.7 Physically Pre-frail 18.5 Physically Frail 0.7 Prevalence of physical frailty (column %) Physically Robust 58.1 Physically Pre-frail 40.3 Physically Frail 1.6
433
164
398
240
25
39
18.0 16.5 1.0
6.8 10.0 1.6
50.6 46.5 2.9
37.0 54.2 8.8
P <.001
appropriate for age, sex, education, medical comorbidity, current smoking, alcohol drinking, depressive symptoms, baseline CMMSE scores. Receiver operating characteristics (ROC) analyses were used to compare the abilities among PF, SF, and combined PF-SF phenotypes in predicting IADL and severe disability at baseline and at a 2-year follow-up, in cross-sectional and longitudinal analyses. A 2-sided P value of .05 was considered as statistically significant in the study. All the analysis was performed using StataCorp, Stata, version 14.0 (StataCorp, College Station, TX). Results The mean age of the study sample was 66.1 (7.63 SD), 63.3% were women, and 25.4% were single, divorced, or widowed at baseline.
Overall, 18.4% in the study sample were deemed to have a high SF, 35.6% low, and 46.0% had nil SF, whereas 3.4% were physically frail, 45.1% were prefrail, and 51.5% were robust overall. The baseline characteristics and follow-up outcomes for older adults with nil, low, and high SFs are shown in Tables 1e3. Compared with participants having nil SF, those with low and high SFs were more likely to report poor health; to be women; to be single, divorced, or widowed; more hospitalizations; and to be physically frail; and also to have a higher proportion of current and past smokers but a lower proportion of current drinkers. A greater prevalence of stroke, arthritis, mental illness, cognitive, hearing and depressive symptoms at baseline was also observed among those who had low and high SFs. However, there were no significant associations with comorbidity, physician visits, and polypharmacy.
SF and IADL and Severe Disabilities The associations between SF and IADL and severe disabilities in cross-sectional and longitudinal analyses are shown in Table 4. In cross-sectional analyses, significant trends in ORs of association with prevalent IADL disability were observed for low SF (OR ¼ 1.32) and high SFIs (OR ¼ 2.30) compared to nil SF; and with severe disability for low SF (OR ¼ 1.30) and high SF (OR ¼ 6.27). After adjustment for physical frailty and other covariates, the same trend of associations were observed, and high SF remained significantly associated with IADL dependency. However, further adjustment for baseline MMSE, depression, poor self-reported health greatly reduced the ORs of all associations. In longitudinal analyses, the same trend of associations were observed, and high SF groups were significant in predicting for incident IADL dependency. Further adjustments of physical frailty and other covariates greatly reduced this association.
Table 4 Cross-sectional Analyses of Associations of Baseline Social Frailty Status With Prevalent and Incident Functional Dependency IADL Dependency
Cross-sectional Analysis (N ¼ 2406) Unadjusted Nil Low High Linear Trend Model 1 Low High Linear Trend Model 2 Low High Linear Trend
Severe Dependency (3 BADL)
N
%
OR (95%CI)
198/1107 191/856 148/443
17.9 22.3 33.4
1.00 1.32 (1.06-1.65) 2.30 (1.79-2.96)
P
.015 <.001 <.001
Model Model Model Model
38/629 23/439 25/186
%
OR (95%CI)
7/1107 7/856 17/443
0.6 0.8 3.8
1.00 1.30 (0.45-3.71) 6.27 (2.58-15.23)
P
.629 <.001 <.001
1.24 (0.97-1.58) 1.61 (1.07-1.10)
.090 .002 .002
0.93 (0.31-2.79) 2.37 (0.87-6.50)
.898 .093 .081
1.11 (0.86-1.42) 1.18 (0.86-1.61)
.428 .316 .286
0.81 (0.24-2.72) 1.12 (0.33-3.75)
.734 .857 .840
1254 Longitudinal analysis Unadjusted Nil Low High Linear Trend Model 1a Low High Linear Trend Model 2a Low High Linear Trend
N
1557
6.0 5.2 13.4
1.00 0.86 (0.50-1.46) 2.41 (1.42-4.12)
.579 .001 .008
1/761 3/543 2/253
0.1 0.6 0.8
1.00 4.22 (0.44-40.70) 6.06 (0.55-67.07)
.213 .142 .118
0.70 (0.40-1.23) 1.56 (0.87-2.80)
.215 .137 .291
2.98 (0.29-30.59) 1.31 (0.09-19.28)
.358 .846 .845
0.63 (0.36-1.13) 1.32 (0.70-2.50)
.120 .386 .653
3.30 (0.27-40.15) 2.15 (0.12-38.21)
.350 .602 .600
1: [Age, sex, Marital Status] þ Physical Frailty. 2: [Age, sex, Marital Status, Physical Frailty] þ Poor Self-reported Health, Baseline MMSE Score and Depression. 1a: [Age, sex, Marital Status, IADL Baseline measure] þ Physical Frailty. 2a: [Age, sex, Marital Status, IADL Baseline measure, Physical Frailty] þ Poor Self-reported Health, Baseline MMSE Score and Depression.
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
637.e17
Table 5 Associations of Baseline Combined SF and PF Phenotype With Prevalent and Follow up Functional Disability IADL Dependency N Cross-sectional Analysis Robust without SF Robust with SF Pre-frail without SF Pre-frail with SF Frail without SF Frail with SF Linear Trend Robust (with or without SF) Pre-frail or Frail without SF Pre-frail or frail with SF Linear Trend Longitudinal Analysis N Robust without SF Robust with SF Pre-frail without SF Pre-frail with SF Frail without SF Frail with SF Linear Trend Robust (with or without SF) Pre-frail or Frail without SF Pre-frail or frail with SF Linear Trend
Severe Dependency (3 BADL)
%
OR (95%CI)
2403 158/1076 23/164 202/844 91/240 29/43 34/39
14.7 14.0 23.9 37.9 67.4 87.2
1.00 0.83 1.43 1.60 4.57 11.15
181/1240 231/887 125/279
14.6 26.0 44.8
1.00 1.56 (1.23-1.99) 1.96 (1.39-2.76)
(0.50-1.37) (1.11-1.84) (1.10-2.31) (2.19-9.51) (3.98-31.2)
P
.465 .006 .013 <.001 <.001 <.001 <.001 <.001 <.001
1253
N
%
OR (95%CI)
2403 2/1076 0/164 2/844 9/240 10/43 8/39
0.2 0.0 0.2 3.8 23.3 20.5
1.00 0.00 1.06 5.42 69.91 25.20
2/1240 12/887 17/279
0.2 1.4 6.1
1.00 5.33 (1.14-24.79) 5.80 (1.13-29.80)
1/731 0/105 1/551 1/137 2/21 1/11
0.1 0.0 0.2 0.7 9.5 9.1
1.00 0.00 0.77 2.78 27.60 21.13
1/837 3/572 2/148
0.1 0.5 1.4
1.00 2.05 (0.19-21.67) 3.20 (0.21-47.87)
(0.00) (0.14-7.65) (0.97-30.30) (12.57-388.9) (3.77-168.5)
P
1.00 .957 .054 <.001 .001 <.001 .033 .035 .042
1556
24/627 9/89 34/432 15/95 3/9 1/2
3.8 10.1 7.9 15.8 33.3 50.0
1.00 2.43 1.79 2.55 5.65 5.71
33/716 37/441 16/97
4.6 8.4 16.5
1.00 1.57 (0.94-2.61) 2.07 (1.01-4.23)
(1.04-5.67) (1.02-3.14) (1.19-5.48) (1.21-26.5) (0.22-147.5)
.040 .043 .016 .028 .293 .003 .084 .046 .024
(0.00) (0.04-13.08) (0.13-58.06) (1.44-525.80) (0.66-679.03)
.00 .85 .46 .027 .085 .029 .551 .399 .393
SF ¼ 2 or more positive Social Frailty indicators. Adjusted for age, gender, medical comorbidity, current smoking, alcohol drinking, depressive symptom as well as baseline MMSE scores.
Relationships Between SF and PF Phenotypes Table 3 shows that although the SF index was strongly associated with the PF index (P < .001), there was no complete overlap in the categories of the SF index with the PF index (kappa ¼ 0.076). Among participants with high SF, for example, 63% were prefrail or frail, and 37% were robust. Among those with no SF, 42% were either prefrail or frail, and only 58% were physically robust.
Combined PF and SF Phenotype and IADL and Severe Dependencies Table 5 shows the results of cross-sectional and longitudinal analyses of the associations between prefrailty and frailty with and without high SF (2 SF indicators), and prevalent and incident IADL or severe disability, using physically robust participants without SF as the reference group. IADL Disability In cross-sectional analyses, the OR estimates were increased among the prefrail and frail without SF, but were greater in the presence of SF; prefrailty or frail without SF: 1.56 (1.23e1.99), prefrail or frail with SF: 1.96 (1.39e2.76) for prevalent disability. In Table 6 Receiver Operator Characteristics of PFI, SFI and PSFI in Predicting IADL and Severe Disabilities
longitudinal analyses, the estimates for pre-frail or frail with SF were significantly elevated for incident disability (OR ¼ 2.07 (1.01e4.23)). Even much higher ORs for IADL disability were found among those who were both physically and socially frail: OR ¼ 11.15 (3.98e31.20) for prevalent IADL disability, and OR ¼ 5.71 (0.22e147.50) for incident IADL disability. Severe Disability Similar trends of associations for severe disability were found in cross-sectional and longitudinal analyses. In cross-sectional analyses, relative to the robust without SF, the presence of SF increased the prevalence and OR of association with severe disability from 0.2% to 3.8%, (OR 1.06 to 5.42) among the pre-frail; physically frail participants with and without SF had especially high likelihood (over 20%) of prevalent severe disability. In longitudinal analyses, the risks of incident severe disability, compared to robust without SF (0.1%) were clearly highest (over 9%) in the groups who were frail with and without SF. Receiver Characteristic Analyses The abilities of PF, SF, and PSF indexes to predict prevalent and incident IADL and severe disability are shown in Table 6, Figures 1 and 2. Satisfactory areas under the ROC were observed for PF index and SF index alone, especially for severe disability, but PSF index stood out with the highest area under the ROC (81% to 91%) for predicting prevalent and incident IADL and severe disability.
Areas Under ROC Cross-sectional Analysis
Social Frailty Index (SFI) Physical Frailty Index (PFI) Physical and Social Frailty Index (PSFI)
Longitudinal Analysis
Discussion Our study shows that the SF is a valid construct, and is able to predict older individuals who are at risk of becoming functionally disabled. We found that socially frail older adults compared with their counterparts were more likely to present with or develop functional disability, independently of PF. Our results, however, indicate that the association of SF with functional disability was not independent of
IADL
Severe ADL
IADL
Severe ADL
0.580 0.637 0.651
0.696 0.855 0.901
0.565 0.608 0.642
0.679 0.794 0.810
ROC ¼ Receiver Operator Characteristics.
637.e18
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
Fig. 1. (A) Cross-sectional analysis, IADL disability. (B) Cross-sectional analysis, severe 3 basic BADL disabilities.
mental health variables such as MMSE, depressive symptoms, and poor self-reported health. This suggests that cognitive and psychological factors closely associated with SF are another set of factors representing mental frailty that have an impact on functional disability. This should be further investigated in future studies. SF was also closely related to PF, but they do not completely overlap. SF shared many risk factors in common with PF, but it was different in not being associated with factors such as comorbidity, physician visits, and polypharmacy. It is already well established that physically frail older persons are more likely to be found disabled, or to become disabled in time. This study shows that those who were both socially and physically frail were even more likely to be disabled or become disabled. Thus, although the SF index and the PF index independently and strongly predict IADL and severe disability, a combined physical and social index was able to identify individuals with much higher risk of functional and severe disability than either the SF index or the PF index alone. The results suggest that conceptually a multidomain phenotype of frailty is better able than a single-domain physical phenotype to
predict disability risk.8 Brief frailty scales, such as the Tilburg Frailty Index and the Edmonton Scale,19e21 based on a multidimensional construct, may thus have this advantage of increased global prediction for relevant specific outcomes such as disability. On the other hand, it is important to note that single-domain indices such as the PF phenotype or the SF phenotype may be specifically useful in targeting unique sets of risk factors and desired outcomes of interventions. Clinical Utility The screening and assessment of frailty among older persons with measurement tools that are predictive of future risks of functional disability is an important interventional approach to reduce the negative impact and burden of functional disability on health and social services. In this population sample, approximately 18.4% in the sample had high SF scores (positive on 2 to 7 social frailty indicators) (Table 3). As shown in Table 4, this group of individuals had a 33.4% absolute likelihood of being functionally disabled and 13.4% absolute likelihood of becoming functionally disabled, 2 times more than those
Fig. 2. (A) Longitudinal analysis, IADL disability. (B) Longitudinal analysis, severe 3 basic BADL disabilities.
N. Teo et al. / JAMDA 18 (2017) 637.e13e637.e19
without SF. Approximately 12% in the sample were both socially and physically frail (SFI2 and PFI1) (Table 3). They had a 16% to 44.8% absolute likelihood of being disabled or becoming disabled, 2 times more than those who were physically and socially robust. Furthermore, they had a 1.4% e 6% absolute likelihood of being or becoming severely disabled on 3 or more basic ADLs, 3 e 6 times more than those who were physically and socially robust. Strengths and Limitations The study has strengths and limitations. There was good content and construct validity in the SF index from embedding it in a welldefined theoretical framework for including salient social factors that are relevant to social situations and health outcomes. However, SF factors were based on self-report rather than objectively measured. Perceived social support was assessed rather than instrumental social support, and may have a differing impact on adverse health outcomes.22 The type of social support or engagement the older adult received did not differentiate family, friends, or formal service providers. Further research identifying the effects on the differing types of relationships, as well as instrumental social support and their effects on adverse health outcomes should be considered. The number of incident severe disability cases was small, thus limiting the statistical precision of the estimates of associations in longitudinal analyses. Nonresponders and dropouts from follow-up in the study were more likely to be socially or physically frail and therefore more likely to have higher mortality and poor functional outcomes. Thus, the OR estimates of association between the frailty indexes and functional disability are likely be underestimated and should be viewed as conservative. Conclusion SF is associated with increased likelihood of being and becoming disabled on daily living activities, independent of PF. Although individually, PF and SF are highly predictive of future risks of disability, their combined index has the highest predictive value for functional disability, especially severe disability. The SF index with or without the PF index has relevance and utility in health care and social service settings for identifying and stratifying the population at risk of disabilities, and implementing early interventions to reduce the burden of care and promoting healthy aging in the elderly population. Further research should be conducted to examine the mental frailty construct that is closely related to SF. The interrelationships among social, mental, and physical domains in a multidimensional frailty construct and its clinical relevance and utility should be further investigated. Acknowledgments We thank the following voluntary welfare organizations for their support: Geylang East Home for the Aged, Presbyterian Community
637.e19
Services, St Luke’s Eldercare Services, Thye Hua Kwan Moral Society (Moral Neighbourhood Links), Yuhua Neighbourhood Link, Henderson Senior Citizens’ Home, NTUC Eldercare Co-op Ltd, Thong Kheng Seniors Activity Centre (Queenstown Centre) and Redhill Moral Seniors Activity Centre. References 1. Bunt S, Steverink N, Olthof J, et al. Social frailty in older adults: A scoping review. Eur J Ageing. In press. 2. Rockwood K. What would make a definition of frailty successful? Age Ageing 2005;34:432e434. 3. Gobbens RJ, Luijkx KG, Wijnen-Sponselee MT, Schols JM. Toward a conceptual definition of frail community dwelling older people. Nurs Outlook 2010;58: 76e86. 4. Fried LP, Tangen CM, Walston J, et al. Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:146e157. 5. Ensrud KE, Ewing SK, Taylor BC, et al. Comparison of 2 frailty indexes for prediction of falls, disability, fractures, and death in older women. Arch Intern Med 2008;168:382e389. 6. Andrew MK, Mitnitski AB, Rockwood K. Social vulnerability, frailty and mortality in elderly people. PLoS One 2008;3:1e8. 7. Verbrugge LM, Jette AM. The disablement process. Soc Sci Med 1994;38:1e14. 8. Garre-Olmo J, Calvó-Perxas L, López-Pousa S, et al. Prevalence of frailty phenotypes and risk of mortality in a community-dwelling elderly cohort. Age Ageing 2013;42:46e51. 9. Makizako H, Shimada H, Tsutsumimoto K, et al. Social frailty in communitydwelling older adults as a risk factor for disability. J Am Med Dir Assoc 2015; 16:1003.e7e1003.e11. 10. Ng TP, Feng L, Nyunt MSZ, et al. Frailty in older persons: Multisystem risk factors and the Frailty Risk Index (FRI). J Am Med Dir Assoc 2014;15: 635e642. 11. Feng L, Zin Nyunt MS, Gao Q, et al. Cognitive frailty and adverse health outcomes: Findings from the Singapore Longitudinal Ageing Studies (SLAS). J Am Med Dir Assoc 2016;18:252e258. 12. Niti M, Yap KB, Kua EH, et al. Physical, social and productive leisure activities, cognitive decline and interaction with APOE-epsilon 4 genotype in Chinese older adults. Int Psychogeriatr 2008;20:237e251. 13. Low LL, Wah W, Ng MJ, et al. Housing as a social determinant of health in Singapore and its association with readmission risk and increased utilization of hospital services. Front Public Health 2016;4:109. 14. Mahoney FI, Barthel DW. Functional evaluation: The Barthel Index. Md State Med J 1965;14:61e65. 15. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:179e186. 16. Nyunt MS, Fones C, Niti M, Ng TP. Criterion-based validity and reliability of the Geriatric Depression Screening Scale (GDS-15) in a large validation sample of community-living Asian older adults. Aging Ment Health 2009;13: 376e382. 17. Feng L, Chong MS, Lim WS, Ng TP. The modified mini-mental state examination test: Normative data for Singapore Chinese older adults and its performance in detecting early cognitive impairment. Singapore Med J 2012;53:458e462. 18. Tielsch JM, Sommer A, Witt K, et al. the B.E.S.R. Group. Blindness and visual impairment in an American urban population. Arch Ophthalmol 1990;108: 286e290. 19. Gobbens RJ, van Assen MA, Luijkx KG, et al. The Tilburg frailty indicator: Psychometric properties. J Am Med Dir Assoc 2010;11:344e355. 20. Bergman H, Ferrucci L, Guralnik J, et al. Frailty: An emerging research and clinical paradigmdissues and controversies. J Gerontol A Biol Sci Med Sci 2007; 62:731e737. 21. Rolfson DB, Majumdar SR, Tsuyuki RT, et al. Validity and reliability of the Edmonton Frail Scale. Age Ageing 2006;35:526e529. 22. Malhotra R, Ang SJ. Social support for older adultsda bane or a boon for their health? Ann Acad Med Singapore 2016;45:172e173.