Prognostic Association of Major Frailty Domain Trajectories With 5-Year Mortality in Very Old Adults: Results From the PARTAGE Cohort Study

Prognostic Association of Major Frailty Domain Trajectories With 5-Year Mortality in Very Old... Abstract We aimed to identify trajectories of nutrition, cognitive function, and autonomy over time among very old adults and to assess their impact on mortality. A cohort of subjects aged ≥80 years (in 2007–2008) who were followed for 5 years in 72 Italian and French nursing homes was used for post hoc analyses. Body mass index (BMI; weight (kg)/height (m)2), Mini-Mental State Examination (MMSE) score, and Katz Index of Independence in Activities of Daily Living (ADL) score were assessed at 4 time points. Information on vital status was collected during follow-up. Latent trajectory and Cox models were used. In the 710 subjects included, the mean age at inclusion was 88.0 (standard deviation, 4.8) years, and 78.9% were female. We identified 7 composite trajectories based on BMI, MMSE, and ADL values. As compared with the reference group (trajectory 7—stable overweight; preserved cognitive function and autonomy), 2 trajectories presented increased hazards of dying: trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy (adjusted hazard ratio = 1.79, 95% confidence interval (CI): 1.26, 2.55)) and trajectory 6 (stable normal BMI; slight cognitive decline; and moderate, then degrading, loss of autonomy (adjusted hazard ratio = 1.67, 95% CI: 1.15, 2.44)). The C-index was 0.81 (95% CI: 0.72, 0.88). Repeated monitoring of BMI, MMSE score, and ADL in very old adults provides trajectories that produce better prognostic information than simple baseline assessment. aged, 80 and over, cognitive aging, frail elderly, holistic health, nutritional status, prognosis Frailty, defined by Campbell and Buchner in 1997 as “multi-system reduction in physiological capacity as a result of which an older person’s function may be severely compromised by minor environmental challenges” (1, p. 317), is associated with increased risk of falls, hospitalization, and impaired survival in older people (2). Despite growing evidence of its high prognostic value, frailty remains difficult to assess in clinical practice because of the lack of a reliable and simple instrument that requires little time to use (3). The 2 highly cited instruments (4)—the Fried Frailty Phenotype (5) and the Frailty Index (6)—are reliable, but they fail to meet the above 2 criteria. In assessing frailty, the most widely used instruments commonly explore domains of frailty such as physical function (sometimes including disability), cognition, weight loss, and physical activity (4). Loss of autonomy, cognitive function, and nutritional status can be approximately and rapidly assessed by means of the Katz Index of Independence in Activities of Daily Living (ADL) (7), the Mini-Mental State Examination (MMSE) (8), and body mass index (BMI; weight (kg)/height (m)2), respectively. The respective associations of these baseline assessments with mortality have been extensively studied (9–11). However, only a few studies have focused on the prognostic value of their trajectories (12–15) (i.e., their evolution over time, from baseline to the last follow-up visit), and when they have done so, investigators have preferred a 1-domain approach over a comprehensive multidomain approach to determine trajectories. The objectives of this investigation were to identify trajectories combining autonomy and 2 frailty domains—nutrition and cognitive function—over time in adults aged 80 years or more at baseline who were living in nursing homes, and to assess the association of these trajectories with long-term all-cause mortality. METHODS Design We used data from an observational longitudinal cohort study, the Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population (PARTAGE) Study, for a post hoc analysis. Setting The PARTAGE Study has been extensively described elsewhere (16, 17). Briefly, very old people (ages ≥80 years) living in 72 French and Italian nursing homes were recruited between January 2007 and June 2008 through 4 French (Dijon, Nancy, Paris, and Toulouse) and 2 Italian (Cesena and Verona) university hospitals. All participants were then followed annually for 2 years for collection of data on BMI, MMSE score, ADL, and all-cause mortality. Participants from the French nursing homes were then followed up at 5 years. Participants In the PARTAGE Study, inclusion criteria were age ≥80 years, living in a nursing home, and providing signed informed consent. Subjects who had severe functional impairment (i.e., ADL index <2) or severe dementia (i.e., MMSE score <12) or were under guardianship or some other measure of legal protection were excluded. The family and/or physician of the subject was informed of the study and gave approval for their relative’s or patient’s participation. All eligible subjects were informed about the study and were invited to participate during the recruitment period. The PARTAGE Study included 1,126 participants. For the present investigation, to allow for maximal follow-up, we restricted the analyses to all participants recruited in the 50 French nursing homes (n = 710) (16). Data collection At baseline, age, sex, conditions included in the Charlson Comorbidity Index (CCI) (18), weight and height (for calculation of BMI), MMSE score (8), and ADL index (7) were assessed and/or collected on a standardized form in the recruiting nursing homes by a trained medical research investigator at each university hospital involved. Then, weight and height were measured and MMSE score (8) and ADL index (7) were assessed annually in the recruiting nursing homes by the same trained medical research investigator. To assess participants’ weights (and heights) each time, the research investigator used the scales (and stadiometers) available in the nursing homes, after appropriate calibration as described by the supplier. The endpoint was all-cause mortality. Information on mortality was collected via regular (every 3 months) direct contact with the nurses and physicians of all nursing homes. Statistical analyses Participants’ baseline characteristics were determined as mean (standard deviation) for continuous variables and number (%) for categorical variables. Missing data were characterized and their mechanism assessed by searching for evidence of monotonicity, unit nonresponse, and file matching patterns. Semiparametric group-based trajectory models were used to capture unobserved heterogeneity in the BMI, MMSE, and ADL trajectories after inclusion. This model uses a multinomial mixture modeling strategy and identifies relatively homogeneous clusters of trajectories of change over time in the presence of repeated observations on analytical units. In other words, the model assumes that the population consists of a mixture of underlying trajectories. The “proc traj” package in SAS (SAS Institute, Inc., Cary, North Carolina) was used to fit the model. In the latent trajectory model, composite trajectories were determined using data for BMI, MMSE, and ADL. Models with an increasing number of trajectories (i.e., 1–8) based on quadratic polynomial order were assessed. Once the number of trajectories had been identified, we assessed models in descending polynomial order, starting from a quadratic polynomial order. The best model (i.e., number of components in a mixture and shape of trajectories) was selected by means of the Bayesian Information Criterion (i.e., the one with the lowest criterion value). This model resulted in classification of participants into a small number of trajectories, based on the pattern of their baseline and follow-up values for BMI, MMSE, and ADL. To assess potential imbalance between trajectory groups, we described participants’ baseline characteristics in each trajectory identified by the latent trajectory model and compared them using analysis of variance or the Kruskal-Wallis test (according to the variable distribution across groups) for continuous variables or the χ2 test for categorical variables. Finally, we assessed the crude association of composite trajectories with mortality using Kaplan-Meier survival estimates at 5 years and the log-rank test. Patients were followed up until death. For patients who remained alive, survival time was right-censored as of the date of the last clinical assessment. In addition, semiproportional hazards Cox models were used to estimate crude and adjusted hazard ratios with 95% confidence intervals. The proportional hazards assumption was assessed using Schoenfeld residuals. The final Cox model adjusted for the main available prognostic factors: baseline age, sex, and CCI. The C-index of the model was then computed and compared with that of a multivariate Cox model including baseline values for BMI, MMSE, ADL, age, sex, and CCI. Participants with missing values for BMI, MMSE, or ADL (i.e., longitudinal data) were included in the analyses as allowed when using mixed models without requiring imputation. No missing data were observed for age, sex, or CCI. The statistical significance level was set at 0.05. P values were 2-sided. SAS 9.3 was used for all analyses. Ethics The PARTAGE Study protocol was approved by national institutional review boards (the French Committee for Protection of Persons and the Italian Ethical Committee of the Vasta Romagna Area). The PARTAGE cohort is registered with the French National Institute of Health and Medical Research (cohort 2011-ASE11067MSA) and the US National Library of Medicine (https://clinicaltrials.gov/; identifier: NCT00901355). Informed consent was obtained from all individual participants included in the study. All procedures performed in studies involving human participants were carried out in accordance with the ethical standards of the relevant institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. RESULTS Participants Selection of participants for the current analysis is outlined in Web Figure 1 (available at https://academic.oup.com/aje). Baseline characteristics Baseline characteristics of participants are provided in Table 1. The mean age of participants was 88.0 (standard deviation (SD), 4.8) years, and 560 (78.9%) were female. The mean CCI score was 6.0 (SD, 1.9). The mean BMI, MMSE score, and ADL index at inclusion were 25.8 (SD, 4.8), 23.9 (SD, 4.9), and 5.0 (SD, 1.0), respectively. The lowest proportion of missing data at each time point was observed for ADL (Web Table 1). The highest proportion of missing data at each time point was observed for BMI (i.e., from 13.7% at baseline to 40.2% at 5 years). No structured missing data pattern, such as monotonicity, unit nonresponse, or file matching, was observed for BMI, MMSE, or ADL (data not shown). Accordingly, data were considered missing at random for BMI, MMSE, and ADL. Table 1. Baseline Characteristics of Nursing Home Residents Aged ≥80 Years in the PARTAGE Study (n = 710), France, 2007–2008 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; IQR, interquartile range; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; SD, standard deviation. a Weight (kg)/height (m)2. b See Web Table 1 for a description of missing data. c Katz Index of Independence in Activities of Daily Living (7). Table 1. Baseline Characteristics of Nursing Home Residents Aged ≥80 Years in the PARTAGE Study (n = 710), France, 2007–2008 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; IQR, interquartile range; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; SD, standard deviation. a Weight (kg)/height (m)2. b See Web Table 1 for a description of missing data. c Katz Index of Independence in Activities of Daily Living (7). Comprehensive BMI, MMSE, and ADL trajectories The best mixture model was shaped on degree 2 polynomial trajectories and retained 7 BMI, MMSE, and ADL components, with a Bayesian Information Criterion of −13,890. Data for the evolution of mean BMI, MMSE, and ADL from baseline to the last time point according to the retained 7 trajectories are plotted in Figure 1 and described in Web Table 2. Figure 1. View largeDownload slide Evolution of body mass index (BMI; weight (kg)/height (m)2) (A), Mini-Mental State Examination (MMSE) score (8) (B), and Katz Index of Independence in Activities of Daily Living (ADL) score (7) (C) over time among nursing home residents aged ≥80 years, according to 7 composite trajectories (T) of nutrition, cognitive function, and autonomy, France, 2007–2013. Latent class modeling produced the following 7 trajectories: T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Figure 1. View largeDownload slide Evolution of body mass index (BMI; weight (kg)/height (m)2) (A), Mini-Mental State Examination (MMSE) score (8) (B), and Katz Index of Independence in Activities of Daily Living (ADL) score (7) (C) over time among nursing home residents aged ≥80 years, according to 7 composite trajectories (T) of nutrition, cognitive function, and autonomy, France, 2007–2013. Latent class modeling produced the following 7 trajectories: T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Trajectory 1 involved adults with overweight that remained stable and with moderate cognitive impairment and moderate loss of autonomy, both greatly increased over time. Trajectory 2 involved an upper limit of normal BMI that remained stable over time and slightly impaired cognitive function and slight loss of autonomy, both markedly degrading over time. Trajectory 3 involved a normal stable BMI, slightly impaired cognitive function, and normal baseline autonomy that greatly declined over time. Trajectory 4 involved grade I obesity (BMI 30–34.9) and late decline of BMI, as well as slight cognitive impairment and slight loss of autonomy, both greatly degrading over time. Trajectory 5 involved normal BMI that remained stable, normal stable cognitive function, and preserved stable autonomy. Trajectory 6 involved an upper limit of normal BMI that remained stable over time, normal, slightly decreasing cognitive function, and moderate loss of autonomy, degrading over time. Trajectory 7 involved stable overweight, normal stable cognitive function, and preserved stable autonomy. Characteristic imbalance across trajectories is described in Web Table 2. Age (P < 0.001) but not sex was associated with trajectory groups, with the highest mean age being observed in trajectory 3 (stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy) and the lowest mean age being observed in trajectory 6 (stable normal BMI; slight cognitive decline; and moderate, then degrading, loss of autonomy). Comorbidity was also associated with trajectory groups (P < 0.001), with a high mean CCI being observed in trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy) and the lowest being observed in trajectory 7 (stable overweight; preserved cognitive function and autonomy). Follow-up and mortality A total of 395 deaths occurred during an overall median follow-up period of 3.77 years (range, 0.01–6.35). There were 65, 77, 16, 29, 94, 50, and 64 deaths among adults in the first, second, third, fourth, fifth, sixth, and seventh trajectories, respectively. The mean duration of time between study inclusion and follow-up visits was 1.00 (SD, 0.05) year at 1 year, 2.00 (SD, 0.07) years at 2 years, and 5.03 (SD, 0.18) years at 5 years. A total of 28 (3.9%), 31 (4.4%), and 60 (8.5%) very old individuals were lost to follow-up at 1 year, 2 years, and 5 years of follow-up, respectively. Kaplan-Meier survival curves according to the BMI, MMSE, and ADL trajectories are shown in Figure 2. The best survival rate was with trajectory 7 (stable overweight; preserved cognitive function and autonomy) and the worst was with trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy). Other trajectories had intermediate survival rates (P < 0.001). Figure 2. View largeDownload slide Long-term probability of survival according to trajectories of body mass index (weight (kg)/height (m)2), Mini-Mental State Examination score (8), and Katz Index of Independence in Activities of Daily Living score (7) among nursing home residents aged ≥80 years, France, 2007–2013. Survival was characterized using Kaplan-Meier estimates in each of 7 trajectories (T) resulting from latent class modeling. T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Appendix Table 1 displays the number of very old adults at risk at each survival time point. Log-rank test: P < 0.0001. Figure 2. View largeDownload slide Long-term probability of survival according to trajectories of body mass index (weight (kg)/height (m)2), Mini-Mental State Examination score (8), and Katz Index of Independence in Activities of Daily Living score (7) among nursing home residents aged ≥80 years, France, 2007–2013. Survival was characterized using Kaplan-Meier estimates in each of 7 trajectories (T) resulting from latent class modeling. T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Appendix Table 1 displays the number of very old adults at risk at each survival time point. Log-rank test: P < 0.0001. Associations of BMI, MMSE, and ADL trajectories with survival The proportional hazards assumption was not violated for models considering BMI, MMSE, and ADL composite trajectory as an independent variable (data not shown). The crude and adjusted associations of BMI, MMSE, and ADL trajectories with survival are shown in Table 2. With trajectory 7 (stable overweight; preserved cognitive function and autonomy) used as the referent, the trajectories associated with increased risk of death were (in increasing risk order): trajectory 3 (stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy (hazard ratio (HR) = 1.73, 95% confidence interval (CI): 1.00, 3.00)), trajectory 6 (stable normal BMI; slight cognitive decline; and moderate, then degrading, loss of autonomy (HR = 1.91, 95% CI: 1.32, 2.76)), and trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy (HR = 2.23, 95% CI: 1.58, 3.15)). Adjustment for age, sex, and CCI modified these results only slightly: As compared with trajectory 7, trajectory 3 remained associated with high risk of death, though not statistically significantly (adjusted HR = 1.46, 95% CI: 0.84, 2.53; P = 0.180), and risk of death remained increased for trajectories 6 (adjusted HR = 1.67, 95% CI: 1.15, 2.44; P = 0.008) and 1 (adjusted HR = 1.79, 95% CI: 1.26, 2.55; P = 0.001). Table 2. Associations of 7 Composite Trajectories of Nutrition, Cognitive Function, and Autonomya With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Abbreviations: BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Nutritional status, cognitive function, and loss of autonomy were assessed by means of BMI (weight (kg)/height (m)2), Mini-Mental State Examination (8) score, and Katz Index of Independence in Activities of Daily Living (7) score, respectively. b Cox model including the 7 trajectories of nutrition, cognitive function, and autonomy, sex, age, and CCI. c Wald χ2 test. d Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. e Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. f Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. g Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. h Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. i Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. j Trajectory involving stable overweight and preserved cognitive function and autonomy. Table 2. Associations of 7 Composite Trajectories of Nutrition, Cognitive Function, and Autonomya With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Abbreviations: BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Nutritional status, cognitive function, and loss of autonomy were assessed by means of BMI (weight (kg)/height (m)2), Mini-Mental State Examination (8) score, and Katz Index of Independence in Activities of Daily Living (7) score, respectively. b Cox model including the 7 trajectories of nutrition, cognitive function, and autonomy, sex, age, and CCI. c Wald χ2 test. d Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. e Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. f Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. g Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. h Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. i Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. j Trajectory involving stable overweight and preserved cognitive function and autonomy. The adjusted associations of baseline BMI, MMSE, and ADL with survival are presented in Table 3. The C-index was 0.81 (95% CI: 0.72, 0.88) for the multivariate Cox model based on the 7 composite BMI, MMSE, and ADL trajectories, and it was 0.70 (95% CI: 0.61, 0.79) for the model based on baseline values of BMI, MMSE, and ADL only. Table 3. Associations of Baseline Body Mass Index, Mini-Mental State Examination Score, and Activities of Daily Living Index With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population. a Cox model including baseline BMI, MMSE, ADL, sex, age, and CCI. b Weight (kg)/height (m)2. c Katz Index of Independence in Activities of Daily Living (7). Table 3. Associations of Baseline Body Mass Index, Mini-Mental State Examination Score, and Activities of Daily Living Index With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population. a Cox model including baseline BMI, MMSE, ADL, sex, age, and CCI. b Weight (kg)/height (m)2. c Katz Index of Independence in Activities of Daily Living (7). DISCUSSION As suggested by the high C-index for the model including frailty trajectories, longitudinal assessment of the evolution of BMI, MMSE, and ADL seems to be of greater prognostic interest than baseline levels of these factors. In other words, longitudinal assessment of nutrition, cognitive function, and autonomy tends to have a better prognostic value in very old adults than a one-shot assessment of these components of frailty. As compared with very old adults with stable overweight, preserved cognitive function, and autonomy (trajectory 7), very old adults with stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy (trajectory 6) were at greater risk of death, and very old adults with stable overweight but moderately impaired, then declining, cognitive function and autonomy (trajectory 1) were at even greater risk of death. Few studies have focused on the mortality prognosis of very old frail adults, and even fewer studies have employed long-term follow-up, such as that used in the PARTAGE cohort, for assessing prognostic factors. Actually, the PARTAGE Study focused on people living in nursing homes, a population that is increasing quickly in France and in most other developed societies. For example, during 2011–2014, the number of nursing home beds in France increased from 611,000 to 634,500, and in Italy the number increased from 220,000 to 234,000 (19). Development of epidemiologic and clinical research in this population can contribute to the improvement of care and preventive actions. Baseline BMI is of prognostic importance in very old adults. Older adults with overweight and grade I obesity (BMI 30–34.9) at baseline are at low risk of mortality, and those who are underweight and normal-weight at baseline are at high risk (11). We also found very old adults with baseline underweight to be at high risk and those who were overweight to be at low risk in comparison with normal-weight very old adults. Yet, baseline obesity was not associated with increased mortality in our sample. This finding might be due to the baseline BMI distribution’s not allowing us to split obesity into grade I, grade II (BMI 35–39.9), and grade III (BMI ≥40) obesity or to survival bias among very old obese people in our sample. In addition, normal-to-overweight stable BMI trajectories associated with stable MMSE and ADL and no major MMSE or ADL impairment (i.e., trajectories 5 and 7) had the highest observed survival rates. Descending BMI trajectories have been reported to entail a poor prognosis in the literature as compared with stable ones (13, 20). Our data could not support this evidence because they contained no clearly descending BMI trajectories, except perhaps trajectory 4, a later-obesity but slightly descending trajectory associated with impaired, then worsening, cognitive function and autonomy. As noted above, we observed a similar cognitive and autonomy trajectory but with a normal stable BMI (trajectory 5) showing a risk of death similar to that with stable overweight (trajectory 7). Accordingly, these results suggest that stable overweight might be necessary but not sufficient to achieve a better mortality prognosis in very old adults. In fact, when associated with impaired and declining cognitive function and autonomy (trajectory 1), stable overweight was actually associated with a poorer prognosis. Furthermore, observed survival rates tended to be different for very old adults with almost similar nutritional and cognitive trajectories (i.e., stable overweight (or upper-limit BMI)) and slightly (or moderately) impaired, then declining, cognitive function—namely, trajectories 1 and 2) but different autonomy trajectories. Indeed, persons with poor and declining autonomy (trajectory 1) were at increased risk of death, whereas for those with overall better autonomy (trajectory 2), the risk of death was similar to that for persons with stable optimal nutritional status, stable cognitive function, and preserved autonomy (trajectory 7), which suggests that loss of autonomy might be of greater prognostic value than nutritional status in very old adults. Loss of autonomy is associated with increased mortality in older adults (9). In our sample, preserved baseline autonomy was associated with better survival. In addition, as reported in the literature (12, 21, 22), trajectories with increased loss of autonomy (i.e., trajectories 1 and 6) were associated with poorer prognosis as compared with a normal stable composite trajectory. However, late increased loss of autonomy showed no clear prognostic value in very old adults with preserved baseline autonomy (trajectory 2). Impaired baseline cognitive function is associated with a poor prognosis in older adults (10). We did not find low baseline MMSE to be associated with increased long-term mortality. However, decline in cognitive function, such as that observed in trajectories 1, 3, and 4, was associated with impaired survival, though not always statistically significantly, in comparison with normal stable cognitive function. Declining cognitive function had already been reported to be associated with impaired prognosis in older adults (14). However, using a composite approach, we also found severe decline in cognitive function, when combined with both better baseline autonomy and delayed decline in autonomy (trajectory 2), to not be clearly associated with an impaired prognosis. Our investigation had some limitations. First, we did not identify clearly descending BMI trajectories. The annual delay between 2 time points might be too large to capture severe and short-term BMI decline leading to death. For instance, a decline in BMI occurring over a few months and leading to death might occur without being able to capture it over 2 annual time points. Another explanation for our not observing descending BMI trajectories may relate to the selection process used for the PARTAGE cohort: The investigators recruited persons living in nursing homes, where staffs are used to tracking and preventing malnutrition, thus preventing trajectories involving malnutrition from being observed in our sample. Second, these results are exploratory and were derived from post hoc analyses of the PARTAGE cohort. They need to be confirmed in other settings. Third, participants were recruited from 50 nursing homes, and data from participants recruited in the same nursing homes might correlate with each other. Unfortunately, we could not assess a possible nursing home effect by using mixed models, due to a lack of statistical power. Finally, observational studies are prone to residual confounding related to unknown factors. One should interpret our results with caution. To our knowledge, no researchers have previously used a composite approach to study the evolution of essential functions and their long-term prognostic impact in very old adults. In addition to baseline values and longitudinal assessment, a composite approach to essential functions and disability might help identify very old adults requiring special attention. Reliable and simple measures reflecting nutrition, cognitive frailty, and disability that require little time (such as weight and height measurement, MMSE, and ADL), if assessed repeatedly and comprehensively, may hold prognostic information that could help clinicians adjust health-care efforts in the very old. ACKNOWLEDGMENTS Author affiliations: Clinical Investigation Center–Clinical Epidemiology 1433, Regional University Hospital Center of Nancy (CHRU Nancy), National Institute of Health and Medical Research (INSERM), University of Lorraine, Nancy, France (Nelly Agrinier, Francis Guillemin, Marie-Line Erpelding); Adaptation, Measure and Evaluation in Health (APEMAC) Unit, University of Lorraine, Nancy, France (Nelly Agrinier, Francis Guillemin); Chronic and Acute Cardiovascular Deficiency Unit, INSERM, Nancy, France (Carlos Labat, Athanase Benetos); Department of Geriatrics, CHRU Nancy, Nancy, France (Sylvie Gautier); and Department of Geriatrics, University of Lorraine, CHRU Nancy, Nancy, France (Athanase Benetos). The PARTAGE Study was funded by the French Ministry of Health as part of the National Clinical Research Hospital Program in 2006 (grant PHRC 2006-A00042-49). We thank everyone at the 4 academic centers in France who participated in recruitment and investigation within the framework of the PARTAGE Study, especially Prof. P. Manckoundia (Dijon), Dr. A. Zervoudaki (Nancy), Dr. A. Kearney-Schwartz (Nancy), Dr. S. Buatois (Nancy), Dr. D. Dubail (Paris), Prof. O. Hanon (Paris), Prof. Y. Rolland (Toulouse), and Dr. O. Toulza (Toulouse). We also thank the directors, physicians, and personnel of the 50 French nursing homes in PARTAGE for contributing to this study. We thank Laura Smales for editing the manuscript. Conflict of interest: none declared. Abbreviations ADL activities of daily living BMI body mass index CCI Charlson Comorbidity Index CI confidence interval HR hazard ratio MMSE Mini-Mental State Examination PARTAGE Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population SD standard deviation REFERENCES 1 Campbell AJ , Buchner DM . Unstable disability and the fluctuations of frailty . Age Ageing . 1997 ; 26 ( 4 ): 315 – 318 . Google Scholar CrossRef Search ADS PubMed 2 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 ( 4 ): 382 – 389 . Google Scholar CrossRef Search ADS PubMed 3 Anaya DA , Johanning J , Spector SA , et al. . Summary of the panel session at the 38th Annual Surgical Symposium of the Association of VA Surgeons: what is the big deal about frailty? JAMA Surg . 2014 ; 149 ( 11 ): 1191 – 1197 . Google Scholar CrossRef Search ADS PubMed 4 Buta BJ , Walston JD , Godino JG , et al. . Frailty assessment instruments: systematic characterization of the uses and contexts of highly-cited instruments . Ageing Res Rev . 2016 ; 26 : 53 – 61 . Google Scholar CrossRef Search ADS PubMed 5 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 ( 3 ): M146 – M156 . Google Scholar CrossRef Search ADS PubMed 6 Mitnitski AB , Mogilner AJ , Rockwood K . Accumulation of deficits as a proxy measure of aging . ScientificWorldJournal . 2001 ; 1 : 323 – 336 . Google Scholar CrossRef Search ADS PubMed 7 Katz S . Assessing self-maintenance: activities of daily living, mobility, and instrumental activities of daily living . J Am Geriatr Soc . 1983 ; 31 ( 12 ): 721 – 727 . Google Scholar CrossRef Search ADS PubMed 8 Folstein MF , Folstein SE , McHugh PR . “Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician . J Psychiatr Res . 1975 ; 12 ( 3 ): 189 – 198 . Google Scholar CrossRef Search ADS PubMed 9 Wu LW , Chen WL , Peng TC , et al. . All-cause mortality risk in elderly individuals with disabilities: a retrospective observational study . BMJ Open . 2016 ; 6 ( 9 ): e011164 . Google Scholar CrossRef Search ADS PubMed 10 An R , Liu GG . Cognitive impairment and mortality among the oldest-old Chinese . Int J Geriatr Psychiatry . 2016 ; 31 ( 12 ): 1345 – 1353 . Google Scholar CrossRef Search ADS PubMed 11 Al Snih S , Ottenbacher KJ , Markides KS , et al. . The effect of obesity on disability vs mortality in older Americans . Arch Intern Med . 2007 ; 167 ( 8 ): 774 – 780 . Google Scholar CrossRef Search ADS PubMed 12 Zimmer Z , Martin LG , Jones BL , et al. . Examining late-life functional limitation trajectories and their associations with underlying onset, recovery, and mortality . J Gerontol B Psychol Sci Soc Sci . 2014 ; 69 ( 2 ): 275 – 286 . Google Scholar CrossRef Search ADS PubMed 13 Zajacova A , Ailshire J . Body mass trajectories and mortality among older adults: a joint growth mixture-discrete-time survival analysis . Gerontologist . 2014 ; 54 ( 2 ): 221 – 231 . Google Scholar CrossRef Search ADS PubMed 14 Dodge HH , Wang CN , Chang CC , et al. . Terminal decline and practice effects in older adults without dementia: the MoVIES Project . Neurology . 2011 ; 77 ( 8 ): 722 – 730 . Google Scholar CrossRef Search ADS PubMed 15 Lunney JR , Lynn J , Foley DJ , et al. . Patterns of functional decline at the end of life . JAMA . 2003 ; 289 ( 18 ): 2387 – 2392 . Google Scholar CrossRef Search ADS PubMed 16 Benetos A , Buatois S , Salvi P , et al. . Blood pressure and pulse wave velocity values in the institutionalized elderly aged 80 and over: baseline of the PARTAGE Study . J Hypertens . 2010 ; 28 ( 1 ): 41 – 50 . Google Scholar CrossRef Search ADS PubMed 17 Benetos A , Gautier S , Labat C , et al. . Mortality and cardiovascular events are best predicted by low central/peripheral pulse pressure amplification but not by high blood pressure levels in elderly nursing home subjects: the PARTAGE (Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population) Study . J Am Coll Cardiol . 2012 ; 60 ( 16 ): 1503 – 1511 . Google Scholar CrossRef Search ADS PubMed 18 Charlson ME , Pompei P , Ales KL , et al. . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation . J Chronic Dis . 1987 ; 40 ( 5 ): 373 – 383 . Google Scholar CrossRef Search ADS PubMed 19 Eurostat . Healthcare resource statistics—beds. http://ec.europa.eu/eurostat/statistics-explained/index.php?title=Healthcare_resource_statistics_-_beds&oldid=314834. Modified November 6, 2017. Accessed April 5, 2017. 20 Murayama H , Liang J , Bennett JM , et al. . Trajectories of body mass index and their associations with mortality among older Japanese: do they differ from those of Western populations? Am J Epidemiol . 2015 ; 182 ( 7 ): 597 – 605 . Google Scholar CrossRef Search ADS PubMed 21 Covinsky KE , Palmer RM , Fortinsky RH , et al. . Loss of independence in activities of daily living in older adults hospitalized with medical illnesses: increased vulnerability with age . J Am Geriatr Soc . 2003 ; 51 ( 4 ): 451 – 458 . Google Scholar CrossRef Search ADS PubMed 22 Wakefield BJ , Holman JE . Functional trajectories associated with hospitalization in older adults . West J Nurs Res . 2007 ; 29 ( 2 ): 161 – 177 . Google Scholar CrossRef Search ADS PubMed Appendix Table 1. Number of Nursing Home Residents Aged ≥80 Years at Risk of Mortality at Each Time Point During Follow-up, PARTAGE Study, France, 2007–2013 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Abbreviations: BMI, body mass index; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. b Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. c Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. d Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. e Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. f Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. g Trajectory involving stable overweight and preserved cognitive function and autonomy. View Large Appendix Table 1. Number of Nursing Home Residents Aged ≥80 Years at Risk of Mortality at Each Time Point During Follow-up, PARTAGE Study, France, 2007–2013 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Abbreviations: BMI, body mass index; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. b Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. c Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. d Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. e Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. f Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. g Trajectory involving stable overweight and preserved cognitive function and autonomy. View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png American Journal of Epidemiology Oxford University Press

Prognostic Association of Major Frailty Domain Trajectories With 5-Year Mortality in Very Old Adults: Results From the PARTAGE Cohort Study

Loading next page...
 
/lp/ou_press/prognostic-impact-of-main-frailty-domain-trajectories-on-5-year-atxQs1nc9k
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0002-9262
eISSN
1476-6256
D.O.I.
10.1093/aje/kwy050
Publisher site
See Article on Publisher Site

Abstract

Abstract We aimed to identify trajectories of nutrition, cognitive function, and autonomy over time among very old adults and to assess their impact on mortality. A cohort of subjects aged ≥80 years (in 2007–2008) who were followed for 5 years in 72 Italian and French nursing homes was used for post hoc analyses. Body mass index (BMI; weight (kg)/height (m)2), Mini-Mental State Examination (MMSE) score, and Katz Index of Independence in Activities of Daily Living (ADL) score were assessed at 4 time points. Information on vital status was collected during follow-up. Latent trajectory and Cox models were used. In the 710 subjects included, the mean age at inclusion was 88.0 (standard deviation, 4.8) years, and 78.9% were female. We identified 7 composite trajectories based on BMI, MMSE, and ADL values. As compared with the reference group (trajectory 7—stable overweight; preserved cognitive function and autonomy), 2 trajectories presented increased hazards of dying: trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy (adjusted hazard ratio = 1.79, 95% confidence interval (CI): 1.26, 2.55)) and trajectory 6 (stable normal BMI; slight cognitive decline; and moderate, then degrading, loss of autonomy (adjusted hazard ratio = 1.67, 95% CI: 1.15, 2.44)). The C-index was 0.81 (95% CI: 0.72, 0.88). Repeated monitoring of BMI, MMSE score, and ADL in very old adults provides trajectories that produce better prognostic information than simple baseline assessment. aged, 80 and over, cognitive aging, frail elderly, holistic health, nutritional status, prognosis Frailty, defined by Campbell and Buchner in 1997 as “multi-system reduction in physiological capacity as a result of which an older person’s function may be severely compromised by minor environmental challenges” (1, p. 317), is associated with increased risk of falls, hospitalization, and impaired survival in older people (2). Despite growing evidence of its high prognostic value, frailty remains difficult to assess in clinical practice because of the lack of a reliable and simple instrument that requires little time to use (3). The 2 highly cited instruments (4)—the Fried Frailty Phenotype (5) and the Frailty Index (6)—are reliable, but they fail to meet the above 2 criteria. In assessing frailty, the most widely used instruments commonly explore domains of frailty such as physical function (sometimes including disability), cognition, weight loss, and physical activity (4). Loss of autonomy, cognitive function, and nutritional status can be approximately and rapidly assessed by means of the Katz Index of Independence in Activities of Daily Living (ADL) (7), the Mini-Mental State Examination (MMSE) (8), and body mass index (BMI; weight (kg)/height (m)2), respectively. The respective associations of these baseline assessments with mortality have been extensively studied (9–11). However, only a few studies have focused on the prognostic value of their trajectories (12–15) (i.e., their evolution over time, from baseline to the last follow-up visit), and when they have done so, investigators have preferred a 1-domain approach over a comprehensive multidomain approach to determine trajectories. The objectives of this investigation were to identify trajectories combining autonomy and 2 frailty domains—nutrition and cognitive function—over time in adults aged 80 years or more at baseline who were living in nursing homes, and to assess the association of these trajectories with long-term all-cause mortality. METHODS Design We used data from an observational longitudinal cohort study, the Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population (PARTAGE) Study, for a post hoc analysis. Setting The PARTAGE Study has been extensively described elsewhere (16, 17). Briefly, very old people (ages ≥80 years) living in 72 French and Italian nursing homes were recruited between January 2007 and June 2008 through 4 French (Dijon, Nancy, Paris, and Toulouse) and 2 Italian (Cesena and Verona) university hospitals. All participants were then followed annually for 2 years for collection of data on BMI, MMSE score, ADL, and all-cause mortality. Participants from the French nursing homes were then followed up at 5 years. Participants In the PARTAGE Study, inclusion criteria were age ≥80 years, living in a nursing home, and providing signed informed consent. Subjects who had severe functional impairment (i.e., ADL index <2) or severe dementia (i.e., MMSE score <12) or were under guardianship or some other measure of legal protection were excluded. The family and/or physician of the subject was informed of the study and gave approval for their relative’s or patient’s participation. All eligible subjects were informed about the study and were invited to participate during the recruitment period. The PARTAGE Study included 1,126 participants. For the present investigation, to allow for maximal follow-up, we restricted the analyses to all participants recruited in the 50 French nursing homes (n = 710) (16). Data collection At baseline, age, sex, conditions included in the Charlson Comorbidity Index (CCI) (18), weight and height (for calculation of BMI), MMSE score (8), and ADL index (7) were assessed and/or collected on a standardized form in the recruiting nursing homes by a trained medical research investigator at each university hospital involved. Then, weight and height were measured and MMSE score (8) and ADL index (7) were assessed annually in the recruiting nursing homes by the same trained medical research investigator. To assess participants’ weights (and heights) each time, the research investigator used the scales (and stadiometers) available in the nursing homes, after appropriate calibration as described by the supplier. The endpoint was all-cause mortality. Information on mortality was collected via regular (every 3 months) direct contact with the nurses and physicians of all nursing homes. Statistical analyses Participants’ baseline characteristics were determined as mean (standard deviation) for continuous variables and number (%) for categorical variables. Missing data were characterized and their mechanism assessed by searching for evidence of monotonicity, unit nonresponse, and file matching patterns. Semiparametric group-based trajectory models were used to capture unobserved heterogeneity in the BMI, MMSE, and ADL trajectories after inclusion. This model uses a multinomial mixture modeling strategy and identifies relatively homogeneous clusters of trajectories of change over time in the presence of repeated observations on analytical units. In other words, the model assumes that the population consists of a mixture of underlying trajectories. The “proc traj” package in SAS (SAS Institute, Inc., Cary, North Carolina) was used to fit the model. In the latent trajectory model, composite trajectories were determined using data for BMI, MMSE, and ADL. Models with an increasing number of trajectories (i.e., 1–8) based on quadratic polynomial order were assessed. Once the number of trajectories had been identified, we assessed models in descending polynomial order, starting from a quadratic polynomial order. The best model (i.e., number of components in a mixture and shape of trajectories) was selected by means of the Bayesian Information Criterion (i.e., the one with the lowest criterion value). This model resulted in classification of participants into a small number of trajectories, based on the pattern of their baseline and follow-up values for BMI, MMSE, and ADL. To assess potential imbalance between trajectory groups, we described participants’ baseline characteristics in each trajectory identified by the latent trajectory model and compared them using analysis of variance or the Kruskal-Wallis test (according to the variable distribution across groups) for continuous variables or the χ2 test for categorical variables. Finally, we assessed the crude association of composite trajectories with mortality using Kaplan-Meier survival estimates at 5 years and the log-rank test. Patients were followed up until death. For patients who remained alive, survival time was right-censored as of the date of the last clinical assessment. In addition, semiproportional hazards Cox models were used to estimate crude and adjusted hazard ratios with 95% confidence intervals. The proportional hazards assumption was assessed using Schoenfeld residuals. The final Cox model adjusted for the main available prognostic factors: baseline age, sex, and CCI. The C-index of the model was then computed and compared with that of a multivariate Cox model including baseline values for BMI, MMSE, ADL, age, sex, and CCI. Participants with missing values for BMI, MMSE, or ADL (i.e., longitudinal data) were included in the analyses as allowed when using mixed models without requiring imputation. No missing data were observed for age, sex, or CCI. The statistical significance level was set at 0.05. P values were 2-sided. SAS 9.3 was used for all analyses. Ethics The PARTAGE Study protocol was approved by national institutional review boards (the French Committee for Protection of Persons and the Italian Ethical Committee of the Vasta Romagna Area). The PARTAGE cohort is registered with the French National Institute of Health and Medical Research (cohort 2011-ASE11067MSA) and the US National Library of Medicine (https://clinicaltrials.gov/; identifier: NCT00901355). Informed consent was obtained from all individual participants included in the study. All procedures performed in studies involving human participants were carried out in accordance with the ethical standards of the relevant institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. RESULTS Participants Selection of participants for the current analysis is outlined in Web Figure 1 (available at https://academic.oup.com/aje). Baseline characteristics Baseline characteristics of participants are provided in Table 1. The mean age of participants was 88.0 (standard deviation (SD), 4.8) years, and 560 (78.9%) were female. The mean CCI score was 6.0 (SD, 1.9). The mean BMI, MMSE score, and ADL index at inclusion were 25.8 (SD, 4.8), 23.9 (SD, 4.9), and 5.0 (SD, 1.0), respectively. The lowest proportion of missing data at each time point was observed for ADL (Web Table 1). The highest proportion of missing data at each time point was observed for BMI (i.e., from 13.7% at baseline to 40.2% at 5 years). No structured missing data pattern, such as monotonicity, unit nonresponse, or file matching, was observed for BMI, MMSE, or ADL (data not shown). Accordingly, data were considered missing at random for BMI, MMSE, and ADL. Table 1. Baseline Characteristics of Nursing Home Residents Aged ≥80 Years in the PARTAGE Study (n = 710), France, 2007–2008 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; IQR, interquartile range; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; SD, standard deviation. a Weight (kg)/height (m)2. b See Web Table 1 for a description of missing data. c Katz Index of Independence in Activities of Daily Living (7). Table 1. Baseline Characteristics of Nursing Home Residents Aged ≥80 Years in the PARTAGE Study (n = 710), France, 2007–2008 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Characteristic No. of Persons % Mean (SD) Median (IQR) Minimum Maximum Sex  Male 150 21.1  Female 560 78.9 Age, years 710 88.0 (4.8) 87.2 (84.5–91.4) 78.8 104.6 CCI 710 6.0 (1.9) 6.0 (5.0–7.0) 3.0 19.0 BMIa,b  Baseline 613 25.8 (4.8) 25.1 (22.4–28.7) 15.2 48.0  1 year 369 26.0 (4.9) 25.7 (22.8–28.7) 14.4 41.8  2 years 449 25.8 (5.1) 25.5 (22.2–29.0) 10.1 45.4  5 years 155 25.3 (4.9) 24.8 (22.1–28.3) 13.3 39.1 MMSE scoreb  Baseline 710 23.9 (4.9) 25.0 (20.0–28.0) 12.0 30.0  1 year 595 22.1 (5.8) 23.0 (18.0–27.0) 0.0 30.0  2 years 508 20.4 (6.9) 22.0 (16.0–26.0) 0.0 30.0  5 years 243 17.9 (8.8) 20.0 (11.0–25.0) 0.0 30.0 ADL indexb,c  Baseline 710 5.0 (1.0) 5.5 (4.5–6.0) 2.0 6.0  1 year 613 4.8 (1.4) 5.5 (4.0–6.0) 0.0 6.0  2 years 529 4.2 (1.7) 5.0 (3.0–5.5) 0.0 6.0  5 years 255 3.3 (2.1) 3.5 (1.0–5.5) 0.0 6.0 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; IQR, interquartile range; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; SD, standard deviation. a Weight (kg)/height (m)2. b See Web Table 1 for a description of missing data. c Katz Index of Independence in Activities of Daily Living (7). Comprehensive BMI, MMSE, and ADL trajectories The best mixture model was shaped on degree 2 polynomial trajectories and retained 7 BMI, MMSE, and ADL components, with a Bayesian Information Criterion of −13,890. Data for the evolution of mean BMI, MMSE, and ADL from baseline to the last time point according to the retained 7 trajectories are plotted in Figure 1 and described in Web Table 2. Figure 1. View largeDownload slide Evolution of body mass index (BMI; weight (kg)/height (m)2) (A), Mini-Mental State Examination (MMSE) score (8) (B), and Katz Index of Independence in Activities of Daily Living (ADL) score (7) (C) over time among nursing home residents aged ≥80 years, according to 7 composite trajectories (T) of nutrition, cognitive function, and autonomy, France, 2007–2013. Latent class modeling produced the following 7 trajectories: T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Figure 1. View largeDownload slide Evolution of body mass index (BMI; weight (kg)/height (m)2) (A), Mini-Mental State Examination (MMSE) score (8) (B), and Katz Index of Independence in Activities of Daily Living (ADL) score (7) (C) over time among nursing home residents aged ≥80 years, according to 7 composite trajectories (T) of nutrition, cognitive function, and autonomy, France, 2007–2013. Latent class modeling produced the following 7 trajectories: T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Trajectory 1 involved adults with overweight that remained stable and with moderate cognitive impairment and moderate loss of autonomy, both greatly increased over time. Trajectory 2 involved an upper limit of normal BMI that remained stable over time and slightly impaired cognitive function and slight loss of autonomy, both markedly degrading over time. Trajectory 3 involved a normal stable BMI, slightly impaired cognitive function, and normal baseline autonomy that greatly declined over time. Trajectory 4 involved grade I obesity (BMI 30–34.9) and late decline of BMI, as well as slight cognitive impairment and slight loss of autonomy, both greatly degrading over time. Trajectory 5 involved normal BMI that remained stable, normal stable cognitive function, and preserved stable autonomy. Trajectory 6 involved an upper limit of normal BMI that remained stable over time, normal, slightly decreasing cognitive function, and moderate loss of autonomy, degrading over time. Trajectory 7 involved stable overweight, normal stable cognitive function, and preserved stable autonomy. Characteristic imbalance across trajectories is described in Web Table 2. Age (P < 0.001) but not sex was associated with trajectory groups, with the highest mean age being observed in trajectory 3 (stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy) and the lowest mean age being observed in trajectory 6 (stable normal BMI; slight cognitive decline; and moderate, then degrading, loss of autonomy). Comorbidity was also associated with trajectory groups (P < 0.001), with a high mean CCI being observed in trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy) and the lowest being observed in trajectory 7 (stable overweight; preserved cognitive function and autonomy). Follow-up and mortality A total of 395 deaths occurred during an overall median follow-up period of 3.77 years (range, 0.01–6.35). There were 65, 77, 16, 29, 94, 50, and 64 deaths among adults in the first, second, third, fourth, fifth, sixth, and seventh trajectories, respectively. The mean duration of time between study inclusion and follow-up visits was 1.00 (SD, 0.05) year at 1 year, 2.00 (SD, 0.07) years at 2 years, and 5.03 (SD, 0.18) years at 5 years. A total of 28 (3.9%), 31 (4.4%), and 60 (8.5%) very old individuals were lost to follow-up at 1 year, 2 years, and 5 years of follow-up, respectively. Kaplan-Meier survival curves according to the BMI, MMSE, and ADL trajectories are shown in Figure 2. The best survival rate was with trajectory 7 (stable overweight; preserved cognitive function and autonomy) and the worst was with trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy). Other trajectories had intermediate survival rates (P < 0.001). Figure 2. View largeDownload slide Long-term probability of survival according to trajectories of body mass index (weight (kg)/height (m)2), Mini-Mental State Examination score (8), and Katz Index of Independence in Activities of Daily Living score (7) among nursing home residents aged ≥80 years, France, 2007–2013. Survival was characterized using Kaplan-Meier estimates in each of 7 trajectories (T) resulting from latent class modeling. T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Appendix Table 1 displays the number of very old adults at risk at each survival time point. Log-rank test: P < 0.0001. Figure 2. View largeDownload slide Long-term probability of survival according to trajectories of body mass index (weight (kg)/height (m)2), Mini-Mental State Examination score (8), and Katz Index of Independence in Activities of Daily Living score (7) among nursing home residents aged ≥80 years, France, 2007–2013. Survival was characterized using Kaplan-Meier estimates in each of 7 trajectories (T) resulting from latent class modeling. T1, trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy; T2, trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy; T3, trajectory involving stable normal BMI, slightly impaired, then declining, cognitive function, and normal baseline, then declining, autonomy; T4, trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy; T5, trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy; T6, trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy; T7, trajectory involving stable overweight and preserved cognitive function and autonomy. Appendix Table 1 displays the number of very old adults at risk at each survival time point. Log-rank test: P < 0.0001. Associations of BMI, MMSE, and ADL trajectories with survival The proportional hazards assumption was not violated for models considering BMI, MMSE, and ADL composite trajectory as an independent variable (data not shown). The crude and adjusted associations of BMI, MMSE, and ADL trajectories with survival are shown in Table 2. With trajectory 7 (stable overweight; preserved cognitive function and autonomy) used as the referent, the trajectories associated with increased risk of death were (in increasing risk order): trajectory 3 (stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy (hazard ratio (HR) = 1.73, 95% confidence interval (CI): 1.00, 3.00)), trajectory 6 (stable normal BMI; slight cognitive decline; and moderate, then degrading, loss of autonomy (HR = 1.91, 95% CI: 1.32, 2.76)), and trajectory 1 (stable overweight; moderately impaired, then declining, cognitive function and autonomy (HR = 2.23, 95% CI: 1.58, 3.15)). Adjustment for age, sex, and CCI modified these results only slightly: As compared with trajectory 7, trajectory 3 remained associated with high risk of death, though not statistically significantly (adjusted HR = 1.46, 95% CI: 0.84, 2.53; P = 0.180), and risk of death remained increased for trajectories 6 (adjusted HR = 1.67, 95% CI: 1.15, 2.44; P = 0.008) and 1 (adjusted HR = 1.79, 95% CI: 1.26, 2.55; P = 0.001). Table 2. Associations of 7 Composite Trajectories of Nutrition, Cognitive Function, and Autonomya With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Abbreviations: BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Nutritional status, cognitive function, and loss of autonomy were assessed by means of BMI (weight (kg)/height (m)2), Mini-Mental State Examination (8) score, and Katz Index of Independence in Activities of Daily Living (7) score, respectively. b Cox model including the 7 trajectories of nutrition, cognitive function, and autonomy, sex, age, and CCI. c Wald χ2 test. d Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. e Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. f Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. g Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. h Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. i Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. j Trajectory involving stable overweight and preserved cognitive function and autonomy. Table 2. Associations of 7 Composite Trajectories of Nutrition, Cognitive Function, and Autonomya With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Variable Crude HR 95% CI Adjusted HRb 95% CI P Valuec Male sex 1.82 1.46, 2.28 1.78 1.42, 2.23 <0.001 Age, years 1.06 1.04, 1.08 1.05 1.03, 1.07 <0.001 CCI 1.17 1.12, 1.22 1.14 1.09, 1.20 <0.001 Trajectory 0.004  T1d 2.23 1.58, 3.15 1.79 1.26, 2.55 0.001  T2e 1.24 0.89, 1.73 1.05 0.75, 1.47 0.769  T3f 1.73 1.00, 3.00 1.46 0.84, 2.53 0.180  T4g 1.45 0.94, 2.25 1.40 0.90, 2.18 0.141  T5h 1.13 0.82, 1.55 1.14 0.83, 1.57 0.424  T6i 1.91 1.32, 2.76 1.67 1.15, 2.44 0.008  T7j 1.00 Referent 1.00 Referent Abbreviations: BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Nutritional status, cognitive function, and loss of autonomy were assessed by means of BMI (weight (kg)/height (m)2), Mini-Mental State Examination (8) score, and Katz Index of Independence in Activities of Daily Living (7) score, respectively. b Cox model including the 7 trajectories of nutrition, cognitive function, and autonomy, sex, age, and CCI. c Wald χ2 test. d Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. e Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. f Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. g Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. h Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. i Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. j Trajectory involving stable overweight and preserved cognitive function and autonomy. The adjusted associations of baseline BMI, MMSE, and ADL with survival are presented in Table 3. The C-index was 0.81 (95% CI: 0.72, 0.88) for the multivariate Cox model based on the 7 composite BMI, MMSE, and ADL trajectories, and it was 0.70 (95% CI: 0.61, 0.79) for the model based on baseline values of BMI, MMSE, and ADL only. Table 3. Associations of Baseline Body Mass Index, Mini-Mental State Examination Score, and Activities of Daily Living Index With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population. a Cox model including baseline BMI, MMSE, ADL, sex, age, and CCI. b Weight (kg)/height (m)2. c Katz Index of Independence in Activities of Daily Living (7). Table 3. Associations of Baseline Body Mass Index, Mini-Mental State Examination Score, and Activities of Daily Living Index With Long-Term Survival Among Nursing Home Residents Aged ≥80 Years, PARTAGE Study, France, 2007–2013 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Variable Adjusted HRa 95% CI P Value Male sex 1.83 1.43, 2.33 <0.001 Age, years 1.06 1.04, 1.08 <0.001 CCI 1.15 1.09, 1.20 <0.001 BMIb 0.041  <18 (underweight) 1.31 0.80, 2.15 0.284  18–24.9 (normal-weight) 1.00 Referent  25–29.9 (overweight) 0.75 0.58, 0.96 0.022  ≥30 (obese) 1.02 0.76, 1.38 0.890 MMSE score 1.00 0.98, 1.02 0.838 ADL indexc 0.84 0.75, 0.94 0.002 Abbreviations: ADL, activities of daily living; BMI, body mass index; CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; MMSE, Mini-Mental State Examination; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population. a Cox model including baseline BMI, MMSE, ADL, sex, age, and CCI. b Weight (kg)/height (m)2. c Katz Index of Independence in Activities of Daily Living (7). DISCUSSION As suggested by the high C-index for the model including frailty trajectories, longitudinal assessment of the evolution of BMI, MMSE, and ADL seems to be of greater prognostic interest than baseline levels of these factors. In other words, longitudinal assessment of nutrition, cognitive function, and autonomy tends to have a better prognostic value in very old adults than a one-shot assessment of these components of frailty. As compared with very old adults with stable overweight, preserved cognitive function, and autonomy (trajectory 7), very old adults with stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy (trajectory 6) were at greater risk of death, and very old adults with stable overweight but moderately impaired, then declining, cognitive function and autonomy (trajectory 1) were at even greater risk of death. Few studies have focused on the mortality prognosis of very old frail adults, and even fewer studies have employed long-term follow-up, such as that used in the PARTAGE cohort, for assessing prognostic factors. Actually, the PARTAGE Study focused on people living in nursing homes, a population that is increasing quickly in France and in most other developed societies. For example, during 2011–2014, the number of nursing home beds in France increased from 611,000 to 634,500, and in Italy the number increased from 220,000 to 234,000 (19). Development of epidemiologic and clinical research in this population can contribute to the improvement of care and preventive actions. Baseline BMI is of prognostic importance in very old adults. Older adults with overweight and grade I obesity (BMI 30–34.9) at baseline are at low risk of mortality, and those who are underweight and normal-weight at baseline are at high risk (11). We also found very old adults with baseline underweight to be at high risk and those who were overweight to be at low risk in comparison with normal-weight very old adults. Yet, baseline obesity was not associated with increased mortality in our sample. This finding might be due to the baseline BMI distribution’s not allowing us to split obesity into grade I, grade II (BMI 35–39.9), and grade III (BMI ≥40) obesity or to survival bias among very old obese people in our sample. In addition, normal-to-overweight stable BMI trajectories associated with stable MMSE and ADL and no major MMSE or ADL impairment (i.e., trajectories 5 and 7) had the highest observed survival rates. Descending BMI trajectories have been reported to entail a poor prognosis in the literature as compared with stable ones (13, 20). Our data could not support this evidence because they contained no clearly descending BMI trajectories, except perhaps trajectory 4, a later-obesity but slightly descending trajectory associated with impaired, then worsening, cognitive function and autonomy. As noted above, we observed a similar cognitive and autonomy trajectory but with a normal stable BMI (trajectory 5) showing a risk of death similar to that with stable overweight (trajectory 7). Accordingly, these results suggest that stable overweight might be necessary but not sufficient to achieve a better mortality prognosis in very old adults. In fact, when associated with impaired and declining cognitive function and autonomy (trajectory 1), stable overweight was actually associated with a poorer prognosis. Furthermore, observed survival rates tended to be different for very old adults with almost similar nutritional and cognitive trajectories (i.e., stable overweight (or upper-limit BMI)) and slightly (or moderately) impaired, then declining, cognitive function—namely, trajectories 1 and 2) but different autonomy trajectories. Indeed, persons with poor and declining autonomy (trajectory 1) were at increased risk of death, whereas for those with overall better autonomy (trajectory 2), the risk of death was similar to that for persons with stable optimal nutritional status, stable cognitive function, and preserved autonomy (trajectory 7), which suggests that loss of autonomy might be of greater prognostic value than nutritional status in very old adults. Loss of autonomy is associated with increased mortality in older adults (9). In our sample, preserved baseline autonomy was associated with better survival. In addition, as reported in the literature (12, 21, 22), trajectories with increased loss of autonomy (i.e., trajectories 1 and 6) were associated with poorer prognosis as compared with a normal stable composite trajectory. However, late increased loss of autonomy showed no clear prognostic value in very old adults with preserved baseline autonomy (trajectory 2). Impaired baseline cognitive function is associated with a poor prognosis in older adults (10). We did not find low baseline MMSE to be associated with increased long-term mortality. However, decline in cognitive function, such as that observed in trajectories 1, 3, and 4, was associated with impaired survival, though not always statistically significantly, in comparison with normal stable cognitive function. Declining cognitive function had already been reported to be associated with impaired prognosis in older adults (14). However, using a composite approach, we also found severe decline in cognitive function, when combined with both better baseline autonomy and delayed decline in autonomy (trajectory 2), to not be clearly associated with an impaired prognosis. Our investigation had some limitations. First, we did not identify clearly descending BMI trajectories. The annual delay between 2 time points might be too large to capture severe and short-term BMI decline leading to death. For instance, a decline in BMI occurring over a few months and leading to death might occur without being able to capture it over 2 annual time points. Another explanation for our not observing descending BMI trajectories may relate to the selection process used for the PARTAGE cohort: The investigators recruited persons living in nursing homes, where staffs are used to tracking and preventing malnutrition, thus preventing trajectories involving malnutrition from being observed in our sample. Second, these results are exploratory and were derived from post hoc analyses of the PARTAGE cohort. They need to be confirmed in other settings. Third, participants were recruited from 50 nursing homes, and data from participants recruited in the same nursing homes might correlate with each other. Unfortunately, we could not assess a possible nursing home effect by using mixed models, due to a lack of statistical power. Finally, observational studies are prone to residual confounding related to unknown factors. One should interpret our results with caution. To our knowledge, no researchers have previously used a composite approach to study the evolution of essential functions and their long-term prognostic impact in very old adults. In addition to baseline values and longitudinal assessment, a composite approach to essential functions and disability might help identify very old adults requiring special attention. Reliable and simple measures reflecting nutrition, cognitive frailty, and disability that require little time (such as weight and height measurement, MMSE, and ADL), if assessed repeatedly and comprehensively, may hold prognostic information that could help clinicians adjust health-care efforts in the very old. ACKNOWLEDGMENTS Author affiliations: Clinical Investigation Center–Clinical Epidemiology 1433, Regional University Hospital Center of Nancy (CHRU Nancy), National Institute of Health and Medical Research (INSERM), University of Lorraine, Nancy, France (Nelly Agrinier, Francis Guillemin, Marie-Line Erpelding); Adaptation, Measure and Evaluation in Health (APEMAC) Unit, University of Lorraine, Nancy, France (Nelly Agrinier, Francis Guillemin); Chronic and Acute Cardiovascular Deficiency Unit, INSERM, Nancy, France (Carlos Labat, Athanase Benetos); Department of Geriatrics, CHRU Nancy, Nancy, France (Sylvie Gautier); and Department of Geriatrics, University of Lorraine, CHRU Nancy, Nancy, France (Athanase Benetos). The PARTAGE Study was funded by the French Ministry of Health as part of the National Clinical Research Hospital Program in 2006 (grant PHRC 2006-A00042-49). We thank everyone at the 4 academic centers in France who participated in recruitment and investigation within the framework of the PARTAGE Study, especially Prof. P. Manckoundia (Dijon), Dr. A. Zervoudaki (Nancy), Dr. A. Kearney-Schwartz (Nancy), Dr. S. Buatois (Nancy), Dr. D. Dubail (Paris), Prof. O. Hanon (Paris), Prof. Y. Rolland (Toulouse), and Dr. O. Toulza (Toulouse). We also thank the directors, physicians, and personnel of the 50 French nursing homes in PARTAGE for contributing to this study. We thank Laura Smales for editing the manuscript. Conflict of interest: none declared. Abbreviations ADL activities of daily living BMI body mass index CCI Charlson Comorbidity Index CI confidence interval HR hazard ratio MMSE Mini-Mental State Examination PARTAGE Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population SD standard deviation REFERENCES 1 Campbell AJ , Buchner DM . Unstable disability and the fluctuations of frailty . Age Ageing . 1997 ; 26 ( 4 ): 315 – 318 . Google Scholar CrossRef Search ADS PubMed 2 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 ( 4 ): 382 – 389 . Google Scholar CrossRef Search ADS PubMed 3 Anaya DA , Johanning J , Spector SA , et al. . Summary of the panel session at the 38th Annual Surgical Symposium of the Association of VA Surgeons: what is the big deal about frailty? JAMA Surg . 2014 ; 149 ( 11 ): 1191 – 1197 . Google Scholar CrossRef Search ADS PubMed 4 Buta BJ , Walston JD , Godino JG , et al. . Frailty assessment instruments: systematic characterization of the uses and contexts of highly-cited instruments . Ageing Res Rev . 2016 ; 26 : 53 – 61 . Google Scholar CrossRef Search ADS PubMed 5 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 ( 3 ): M146 – M156 . Google Scholar CrossRef Search ADS PubMed 6 Mitnitski AB , Mogilner AJ , Rockwood K . Accumulation of deficits as a proxy measure of aging . ScientificWorldJournal . 2001 ; 1 : 323 – 336 . Google Scholar CrossRef Search ADS PubMed 7 Katz S . Assessing self-maintenance: activities of daily living, mobility, and instrumental activities of daily living . J Am Geriatr Soc . 1983 ; 31 ( 12 ): 721 – 727 . Google Scholar CrossRef Search ADS PubMed 8 Folstein MF , Folstein SE , McHugh PR . “Mini-mental state”: a practical method for grading the cognitive state of patients for the clinician . J Psychiatr Res . 1975 ; 12 ( 3 ): 189 – 198 . Google Scholar CrossRef Search ADS PubMed 9 Wu LW , Chen WL , Peng TC , et al. . All-cause mortality risk in elderly individuals with disabilities: a retrospective observational study . BMJ Open . 2016 ; 6 ( 9 ): e011164 . Google Scholar CrossRef Search ADS PubMed 10 An R , Liu GG . Cognitive impairment and mortality among the oldest-old Chinese . Int J Geriatr Psychiatry . 2016 ; 31 ( 12 ): 1345 – 1353 . Google Scholar CrossRef Search ADS PubMed 11 Al Snih S , Ottenbacher KJ , Markides KS , et al. . The effect of obesity on disability vs mortality in older Americans . Arch Intern Med . 2007 ; 167 ( 8 ): 774 – 780 . Google Scholar CrossRef Search ADS PubMed 12 Zimmer Z , Martin LG , Jones BL , et al. . Examining late-life functional limitation trajectories and their associations with underlying onset, recovery, and mortality . J Gerontol B Psychol Sci Soc Sci . 2014 ; 69 ( 2 ): 275 – 286 . Google Scholar CrossRef Search ADS PubMed 13 Zajacova A , Ailshire J . Body mass trajectories and mortality among older adults: a joint growth mixture-discrete-time survival analysis . Gerontologist . 2014 ; 54 ( 2 ): 221 – 231 . Google Scholar CrossRef Search ADS PubMed 14 Dodge HH , Wang CN , Chang CC , et al. . Terminal decline and practice effects in older adults without dementia: the MoVIES Project . Neurology . 2011 ; 77 ( 8 ): 722 – 730 . Google Scholar CrossRef Search ADS PubMed 15 Lunney JR , Lynn J , Foley DJ , et al. . Patterns of functional decline at the end of life . JAMA . 2003 ; 289 ( 18 ): 2387 – 2392 . Google Scholar CrossRef Search ADS PubMed 16 Benetos A , Buatois S , Salvi P , et al. . Blood pressure and pulse wave velocity values in the institutionalized elderly aged 80 and over: baseline of the PARTAGE Study . J Hypertens . 2010 ; 28 ( 1 ): 41 – 50 . Google Scholar CrossRef Search ADS PubMed 17 Benetos A , Gautier S , Labat C , et al. . Mortality and cardiovascular events are best predicted by low central/peripheral pulse pressure amplification but not by high blood pressure levels in elderly nursing home subjects: the PARTAGE (Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population) Study . J Am Coll Cardiol . 2012 ; 60 ( 16 ): 1503 – 1511 . Google Scholar CrossRef Search ADS PubMed 18 Charlson ME , Pompei P , Ales KL , et al. . A new method of classifying prognostic comorbidity in longitudinal studies: development and validation . J Chronic Dis . 1987 ; 40 ( 5 ): 373 – 383 . Google Scholar CrossRef Search ADS PubMed 19 Eurostat . Healthcare resource statistics—beds. http://ec.europa.eu/eurostat/statistics-explained/index.php?title=Healthcare_resource_statistics_-_beds&oldid=314834. Modified November 6, 2017. Accessed April 5, 2017. 20 Murayama H , Liang J , Bennett JM , et al. . Trajectories of body mass index and their associations with mortality among older Japanese: do they differ from those of Western populations? Am J Epidemiol . 2015 ; 182 ( 7 ): 597 – 605 . Google Scholar CrossRef Search ADS PubMed 21 Covinsky KE , Palmer RM , Fortinsky RH , et al. . Loss of independence in activities of daily living in older adults hospitalized with medical illnesses: increased vulnerability with age . J Am Geriatr Soc . 2003 ; 51 ( 4 ): 451 – 458 . Google Scholar CrossRef Search ADS PubMed 22 Wakefield BJ , Holman JE . Functional trajectories associated with hospitalization in older adults . West J Nurs Res . 2007 ; 29 ( 2 ): 161 – 177 . Google Scholar CrossRef Search ADS PubMed Appendix Table 1. Number of Nursing Home Residents Aged ≥80 Years at Risk of Mortality at Each Time Point During Follow-up, PARTAGE Study, France, 2007–2013 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Abbreviations: BMI, body mass index; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. b Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. c Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. d Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. e Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. f Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. g Trajectory involving stable overweight and preserved cognitive function and autonomy. View Large Appendix Table 1. Number of Nursing Home Residents Aged ≥80 Years at Risk of Mortality at Each Time Point During Follow-up, PARTAGE Study, France, 2007–2013 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Trajectory Time Point, years 0 1 2 3 4 5 6 T1a 95 78 60 36 27 14 0 T2b 144 128 113 89 74 41 0 T3c 25 23 20 12 9 3 0 T4d 52 48 37 29 21 12 0 T5e 187 168 147 125 102 48 0 T6f 70 62 51 34 27 17 3 T7g 137 124 116 96 79 47 0 Abbreviations: BMI, body mass index; PARTAGE, Predictive Values of Blood Pressure and Arterial Stiffness in Institutionalized Very Aged Population; T, trajectory. a Trajectory involving stable overweight and moderately impaired, then declining, cognitive function and autonomy. b Trajectory involving stable upper normal BMI and slightly impaired, then declining, cognitive function and autonomy. c Trajectory involving stable normal BMI; slightly impaired, then declining, cognitive function; and normal baseline autonomy, then declining autonomy. d Trajectory involving baseline obesity with late nutritional decline and declines in cognitive function and autonomy. e Trajectory involving stable normal nutrition, stable cognitive function, and preserved autonomy. f Trajectory involving stable normal BMI, slight cognitive decline, and moderate, then degrading, loss of autonomy. g Trajectory involving stable overweight and preserved cognitive function and autonomy. View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

American Journal of EpidemiologyOxford University Press

Published: Aug 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off