Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Understanding Health Deterioration and the Dynamic Relationship between Physical Ability and Cognition among a Cohort of Danish Nonagenarians

Understanding Health Deterioration and the Dynamic Relationship between Physical Ability and... Hindawi Journal of Aging Research Volume 2020, Article ID 4704305, 8 pages https://doi.org/10.1155/2020/4704305 Research Article Understanding Health Deterioration and the Dynamic Relationship between Physical Ability and Cognition among a Cohort of Danish Nonagenarians 1,2 1 2 Cosmo Strozza , Virginia Zarulli, and Viviana Egidi Interdisciplinary Centre on Population Dynamics, University of Southern Denmark, 5000 Odense, Denmark Department of Statistical Sciences, Sapienza University of Rome, 00185 Rome, Italy Correspondence should be addressed to Cosmo Strozza; cstrozza@health.sdu.dk Received 25 February 2020; Accepted 15 April 2020; Published 3 June 2020 Academic Editor: Elke Bromberg Copyright © 2020 Cosmo Strozza et al. &is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. &is study aims to determine how demographics, socioeconomic characteristics, and lifestyle affect physical and cognitive health transitions among nonagenarians, whether these transitions follow the same patterns, and how each dimension affects the transitions of the other. We applied a multistate model for panel data to 2262 individuals over a 2-year follow-up period from the 1905 Danish Cohort survey. Within two years from baseline, the transition probability from good to bad physical health—ability to stand up from a chair—was higher than dying directly (29% vs. 25%), while this was not observed for cognition (24% vs. 27%) evaluated with Mini-Mental State Examination—a score lower than 24 indicates poor cognitive health. Probability of dying either from bad physical or cognitive health condition was 50%. Health transitions were associated with sex, education, living alone, body mass index, and physical activity. Physical and cognitive indicators were associated with deterioration of cognitive and physical status, respectively, and with survivorship from a bad health condition. We conclude that physical and cognitive health deteriorated differently among nonagenarians, even if they were related to similar sociodemographic and lifestyle characteristics and resulted dynamically related with each other. found that physical functioning at baseline was associated 1. Introduction with longitudinal changes in cognition but the opposite &e proportion of the oldest-old has increased during the relationship was inconsistent. Physical mobility and func- last decades as a consequence of the decline in old-age tioning dynamically interact between healthy and unhealthy mortality [1–3]. &e share of nonagenarians in Denmark states [5, 9–11]. Cognitive health declines with age more increased from around 0.08% in 1950 to 0.82% in 2020 and is linearly, even though this decline can cover a more complex expected to further grow during the next years, reaching pattern [6, 12, 13]. It is therefore crucial to investigate 2.03% in 2050 [4]. &is phenomenon is taking place in most further how the physical and cognitive deteriorations evolve developed countries, fueling a growing interest on the health and whether they follow different patterns. &e relationship conditions of oldest-old [2]. Health transitions at older ages between the two dimensions of the health status has been are of particular interest as deterioration of both physical widely investigated. However, the literature lacks studies on and cognitive health conditions is very likely [5, 6]. the oldest-old and this is why in this study we focus on Physical and cognitive health decline have been inves- individuals aged 90+. tigated, in order to understand whether they can be partially With this analysis, we aim at investigating (1) how de- explained by other health characteristics [7]. A systematic mographic and socioeconomic characteristics and lifestyle review of the relationship between physical functioning and habits affect transitions in physical and cognitive health; (2) cognition was published by Clouston et al. in 2013 [8], which whether these transitions follow the same patterns; and (3) 2 Journal of Aging Research socioeconomic status of the participants. It was grouped how does each dimension (physical or cognitive) affect the transitions of the other dimension. To our knowledge, this is into three categories: (1) elementary school; (2) vocational education; and (3) higher education. Living condition was the first study that analyzes the relationship between physical and cognitive decline and the determinants of transitions in divided into people living (1) alone and (2) with someone. these two health dimensions (physical and cognitive) among &e loss of a close person, self-rated health, and depression nonagenarians. were used to assess the general health perception of the participant and the feelings related to it. &e loss of a close person was categorized into two classes: (1) lost someone 2. Methods (spouse, sons, and close friends) and (2) no people lost due 2.1. Study Population and Measures. &e study population to death within the last five years. Self-rated health was comes from the 1905 Danish Cohort survey, which contains assessed with the first question of Short-Form 12 (SF12) questionnaire [21]: ”How do you consider your health in many individual level information on the members of the cohort born in Denmark in 1905 interviewed and tested for general?”. It was grouped in three categories: (1) very poor or poor; (2) acceptable; and (3) good or excellent. Depression physical and mental health in their home by a survey agency. It is a longitudinal multiassessment survey conducted from was assessed using an adaptation of the depression section of 1998 to 2005 with four waves realized every 2-3 years. the Cambridge mental disorders of the elderly examination Detailed information about the study design are available in [22]. It uses a scale from 17 to 52 and it was grouped into three Nybo and colleagues [14]. In this work we use the first two equal-size categories: (1) 17–22; (2) 23–28; and (3) 29–52. waves of the Danish 1905 Cohort Survey, collected in 1998 Among the health behaviors, smoking habit was categorized and 2000, when the oldest-old were, respectively, 93 and 95 into (1) never smoked; (2) past smoker; and (3) current years old. &e initial population, corresponding to the study smoker. BMI was calculated on the basis of the reported population, was composed of 2626 individuals. &ey rep- height and weight at the interview and categorized into three groups: (1)< 22; (2) 22–28; and (3)> 28. Physical activity was resent 62.8% of the potential participants: individuals born in 1905 and living in Denmark. At the second data collection assessed by asking if they were performing light (light gar- dening, short walks, or bicycle rides) or heavy (heavy gar- in 2000, 874 were found to be dead (38.6%) reducing to 1388 individuals the number of potential participants to the dening, long walks or bicycle rides, sports, gymnastics, or second wave of the study. &e final population interviewed dancing) exercises at the time of the interview. It was grouped in 2000 was composed of 1086 individuals (78.2% of the into three categories: (1) never or not able; (2) light physical potential participants). activity; and (3) heavy physical activity. &e number of &e cognitive function was measured with the Mini- medications (daily intake) was coded according to the An- Mental State Examination (MMSE): the higher the score atomical &erapeutic Chemical classification system and it was grouped into three equal-size categories: (1) 0-1; (2) 2-3; (0–30), the better the cognitive status [15]. We grouped it into three standard categories, in order to distinguish people and (3) 4+. &e main reference for variable selection and classifi- with severe (0–17), mild (18–23), and no cognitive im- pairment (24–30) based on the most frequently used cate- cation is Appendix S1 of the article by &inggaard et al. (2016) that uses the same study population [17]. &e pro- gorization in literature [16, 17]. &e physical function was assessed by the Chair-Stand test: the elderly who can stand portion of dropouts is 13.4%. We performed a sensitivity up from a chair have better functional status than who need analysis in order to check whether dropout was associated to use hands or cannot do it. &is test was found to be a good with bad health and we did not find any significant asso- predictor of disability and mortality among the elderly other ciation with both health indicators. than a proper instrument to measure lower body strength [17–20]. 2.2. Statistical Analysis. We applied a multistate model for We dichotomized both health indicators, in order to panel data—with Markov chain assumption—[23, 24] to create two categories: healthy and unhealthy oldest-old. assess the association between the many potential drivers Regarding cognitive health, individuals were considered measured on the Danish nonagenarians and the probability cognitively impaired when reporting a MMSE score from 0 of transitioning from one health state to another (defined as to 23 and not cognitively impaired when the score was transition probability). &e possible transitions are from between 24 and 30. Regarding physical ability, individuals who were not able to stand up from the chair, even with aids, good to bad health status, from good health to death, and from bad health to death. were considered in bad physical health while individuals able to stand up from the chair, with and without use of aids, were &e multistate model we used is based on a stochastic multistate process (X(t), t ∈ T) with a finite state-space considered in good physical health. S � 1, ..., N , where T � [0, τ], τ<∞ represents the time Demographic and socioeconomic characteristics (sex, { } (discrete, for panel data). It is fully characterized through education, and living conditions), critical events (loss of a transition probabilities between states h and j: close relative or friend), health characteristics and be- haviors (self-rated health and depression, smoking habits, (s, t) � P(X(t) � j | X(s) � h). (1) body mass index (BMI), physical activity, and use of hj medications) were considered as confounders and con- for h, j ∈ S, t, s ∈ T or through transition intensities: trolled in the analysis. Education was used to measure the Journal of Aging Research 3 32.3% of women were not depressed) without declaring p (t, t + Δt) hj (2) α (t) � lim , hj better health conditions than women (12.5 % of men rated Δt⟶0 Δt their health as good or excellent while 14.2 % of women representing the instantaneous hazard of progression to did it). state j conditionally on occupying state h at the previous In terms of health behaviors, except for the higher share time. According to the Markov assumption, the probability of (past or current) smokers (78.8% of men vs. 32.4% of of the next transition depends only on the state occupied at women), men had higher BMI (73.1% of men vs. 55.3% of the time t. women had a BMI higher than 22) and performed more &e effect of the explanatory variables z on the tran- physical activity (43.8% of men vs. 28.9% of women perform it sition intensity for individual i at time t is modeled using some physical activity) than women. More details about proportional intensities, replacing α with baseline characteristics of the population are available in hj Table 1. 0 T α z 􏼁 � α exp􏼐β z 􏼑. (3) hj it hj hj it Men scored better in terms of cognitive (48.5% of men vs. 40.6% of women were not cognitively impaired) and We conducted the analysis separately for physical and physical (52.1% of men vs. 41.5% of women were able to cognitive health, in order to be able to include the baseline stand up from the chair without any aid) health compared to status of each dimension (cognitive or physical resp.) as women as reported in Table 2. potential driver in the model for the transitions related to the other dimension. 3.2. Multi-State Analysis Results. We analyzed physical and States have been defined according to the MMSE, when cognitive health deterioration in two different models in- assessing cognition, and according to the Chair-Stand test, cluding, respectively, cognitive and physical baseline health when the focus was on the physical status. Based on both status because the main aim of the study is to examine the classifications, we divided participants into two groups based dynamic relationship between these two health aspects and on their good or bad health condition. not because we considered them independent. &is implies Transitions between four states (good health, bad health, that part of the individuals in the different states of the two nonparticipant but alive, and nonparticipant because dead) analyzes are the same, resulting in similar transition have been estimated through transition probabilities. We probabilities and covariates associated with transition evaluated the effect of the covariates on the transition in- intensities. tensities only for the “worsening” transitions: from good to At baseline, 44.2% of the individuals were in good bad health condition, from good health condition to death, physical health while 42.7% were in good cognitive health. and from bad health condition to death. As expected, only After two years, 38.6% of the study population died while few people experienced “improving” transitions, as this is 13.4% dropped out from the study. unlikely at very old ages. &e probability of moving from a good to a bad physical Because of the relatively small number of individuals in health condition within two years was higher than dying analysis, we could only use the dichotomic classification of directly (29% vs. 25%). People in bad physical health con- MMSE and Chair-Stand test, as the sample size was too small dition have a 50% probability of dying from a bad physical to estimate the coefficients with a finer classification of the health status within two years. variables. We could not perform the analysis separately for When considering the cognitive health, the results men and women due to the small number of nonagenarian showed a different pattern. &e probability of worsening a men in the sample. good cognitive health condition was lower than experiencing We used methods of imputation with survey data [25] to death directly within two years (24% vs. 27%), while indi- deal with missing at random values. More information about viduals in a bad cognitive health have a 47% probability of the imputation method is available in Supplementary Text S1. dying from that condition in the next two years as shown in Statistical analysis was performed using R version 3.5.0 Figure 1. [26]. &e complete transition probabilities are available in Supplementary Tables S1 and S2. 3. Results &e effect of covariates on the transition intensities is reported in Figures 2 and 3 for the physical states and 3.1.DescriptiveResults. Of the 2262—93 years old—baseline cognitive states, respectively. participants of the study, one-fourth were men (25.8%) while Full details about the two models are available in Sup- the rest of the people were women (74.2%). plementary Tables S3 and S4. Men had, on average, a higher education level than women, especially in terms of vocational education (32.9% of men vs. 14.2% of women). Fewer men were living alone 3.2.1. Physical Health Transitions. Being women was asso- compared to women (50.5% of men vs. 64.4% of women). ciated with lower probability of dying for people in bad More men experienced the loss of a close person physical health (female vs. male HR = 0.66) as well as living (spouse, children, and close friends) due to death during alone (living alone vs. with someone HR = 0.60). Living the last five years (71.7% of men vs. 66.9% of women) but alone was also significantly associated with a lower proba- they reported lower rates of depression (39.0% of men vs. bility of transitioning from a good to a bad physical health 4 Journal of Aging Research Table 1: Characteristics of the study population in the first wave in (HR = 0.52). Having a BMI higher than 22 statistically de- 1998 when the individuals were 93 years old. creased the probability of dying, both from a good (BMI 22–28 vs.< 22: HR = 0.45) and a bad (BMI > 28 vs.< 22: Sex HR = 0.63) physical health. Performing physical activity Characteristics M F T p lowered the transition probability from good to bad physical n % n % n % health (heavy vs. no physical activity: HR = 0.35) and from Sample 584 25.8 1678 74.2 2262 100.0 bad physical health to death (light vs. no physical activity: Education <0.001 HR = 0.73). Finally, also being cognitively not impaired was Elementary 292 50.0 1254 74.7 1546 68.3 statistically associated with a lower probability of worsening Vocational 192 32.9 238 14.2 430 19.0 the physical health (HR = 0.47) and dying from a bad one Higher 100 17.1 186 11.1 286 12.6 (HR = 0.62). Living alone <0.001 No 289 49.5 598 35.6 887 39.2 Yes 295 50.5 1080 64.4 1375 60.8 3.2.2. Cognitive Health Transitions. Being a woman was Loss of a close person 0.033 associated with a lower probability of death (from good No 165 28.3 556 33.1 721 31.9 health: HR = 0.42; from bad health: HR = 0.65). Having Yes 419 71.7 1122 66.9 1541 68.1 higher level of education decreased the probability of Self-rated health 0.013 deteriorating the cognitive health (HR = 0.55) as well as Very poor or poor 307 52.6 886 52.8 1193 52.7 living alone (HR = 0.49), which was also a protective factor Acceptable 204 34.9 553 33.0 757 33.5 against transitioning from bad cognitive status to death Good or excellent 73 12.5 239 14.2 312 13.8 (HR = 0.59). BMI higher than 22 reduced the probability of Depression 0.008 dying from a good (BMI 22–28 vs.< 22: HR = 0.44) and a 29–52 184 31.5 591 35.2 775 34.3 bad (BMI > 28 vs.< 22: HR = 0.65) cognitive health. Doing 23–28 172 29.5 545 32.5 717 31.7 physical activity was significantly related to lower tran- 17–22 228 39.0 542 32.3 770 34.0 sition rates from bad to death (light vs. no physical activity: Smoke <0.001 HR = 0.65, heavy vs. no physical activity: HR = 0.52). As Current smoker 144 24.7 171 10.2 315 13.9 expected, using more than four medications per day was Past smoker 316 54.1 372 22.2 688 30.4 associated with higher probability of death when already in Never smoked 124 21.2 1135 67.6 1259 55.7 a bad cognitive health (HR = 1.27). Finally, being able to Body Mass Index <0.001 stand up from the chair without any aid was statistically < 22 157 26.9 750 44.7 907 40.1 associated with a lower probability of worsening the 22–28 348 59.6 785 46.8 1133 50.1 > 28 79 13.5 143 8.5 222 9.8 cognitive health (HR = 0.53) and dying from a bad one (HR = 0.61). Physical activity <0.001 None/irrelevant 328 56.2 1193 71.1 1521 67.2 Light 177 30.3 390 23.2 567 25.1 4. Discussion Heavy 79 13.5 95 5.7 174 7.7 Number of medications 0.057 &e increasing proportion of the oldest-old people in the last 4+ 228 39.0 714 42.6 942 41.6 decades increased the attention of researchers and policy 2-3 153 26.2 423 25.2 576 25.5 makers on this subgroup of individuals [2, 3]. As physical 0-1 203 34.8 541 32.2 744 32.9 and cognitive health are two dynamic processes and their Men versus women from Pearson χ test. deterioration is likely, especially at older ages, in the recent years it became a widely investigated topic [5, 6,9, 10, 12]. Finding the determinants of physical and cognitive health Table 2: Health conditions of the study population in the first wave changes and analyzing their longitudinal relationship are in 1998 when the individuals were 93 years old. considered, nowadays, two of the major public health challenges [27, 28]. However, only few studies analyzed such Sex deteriorations among the oldest-old [8, 29]. Studying the Characteristics M F T p determinants of physical and cognitive health transitions n % n % n % among very old people and analyzing the relationship be- Physical ability: <0.001 tween these two conditions will help to shed light on which Chair-Stand Test are the most vulnerable groups. Not able 70 12.0 293 17.5 363 16.0 &is study uses two waves of the 1905 Danish Cohort With use of hands 210 36.0 689 41.1 899 39.7 survey [14] to study the transitions in physical and cognitive Without use of hands 304 52.1 696 41.5 1000 44.2 health among individuals aged 93 at the baseline (1998) and Cognitive health: Mini-Mental <0.001 95 at the second wave (2000). Studies on this cohort showed State Examination 0–17 124 21.2 472 28.1 596 26.3 that high level of disability and poor cognitive and physical 18–23 177 30.3 524 31.2 701 31.0 performance are strong predictors of mortality in the oldest- 24–30 283 48.5 682 40.6 965 42.7 old [30, 31]. More precisely, &inggaad et al. (2016) found Men versus women from Pearson χ test. that being able to stand up from a chair and having a good Journal of Aging Research 5 G G 0.25 0.27 D D 0.29 0.02 0.24 0.07 0.50 0.47 B B (a) (b) Figure 1: Transition probabilities of the multistate model where states are defined according to (a) physical health and (b) cognitive health. Note. (G) good health status; (B) bad health status; (D) dead. Women vs. men Transitions Vocational vs. lower education G->B Higher vs. lower education G->D Living alone vs. with someone B->D Lost someone vs. no one Acceptable vs. (very)poor srh Good vs. (very)poor srh Mild vs. highly depressed Low vs. highly depressed Past vs. never smoker Current vs. never smoker BMI 22−28 vs. BMI < 22 BMI > 28 vs. BMI < 22 Light vs. no physical activity Heavy vs. no physical activity 2-3 vs. 0-1 medications 4+ vs. 0-1 medications Not vs. cognitively impaired 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Figure 2: Multivariate predictions (hazard ratios) of transitions in physical health. Note. Highlighted hazards ratios are significant; (G) good health status; (B) bad health status; (D) dead. level of cognition increased the probability of surviving to individuals in good and bad state for both physical and age of 100 for both women and men of the 1905 Danish cognitive health. Cohort Study [17]. Demographic and socioeconomic variables in both cases Our results partially confirm the trends shown in the resulted associated with health transitions. Not surprisingly, literature for both physical and cognitive health over the women had a lower probability of death [33–35]. However, by years among adults and younger elderly [7, 9, 11, 29, 32]. analyzing physical and cognitive dimension separately, we Even at very old ages, for individuals in good physical health were able to uncover interesting dynamics. Being a woman did conditions, the probability of dying directly was lower than not affect significantly the transition from good to bad health. the probability of first experiencing a health deterioration. However, it was instead associated with a lower probability of &is is what we called here a “one-step worsening pattern.” dying from both good and bad cognitive status but only However, this pattern was not observed for cognitive health lowered the probability of dying from a bad physical health in which the probability of deteriorating the level of cog- condition. As expected, having a higher level of education nition was lower than dying directly from a good cognitive decreased the probability of cognitive decline, confirming the status (24% vs. 27%). results found among younger adults [12, 32, 36, 37]. However, &e analysis of potential drivers of the health decline we found that the level of education did not affect the physical showed similar results for physical and cognitive health, status, contrary to what has been found for a similarly aged (8 years younger) cohort of Canadian elderly [11, 29]. Living showing that the two dimensions of the health status follow somewhat similar patterns. However, it is important to point alone is widely considered a predictor of physical [9, 10, 29] out that this might also partly be due to the overlap of but not for cognitive health transitions. In our study, instead, 6 Journal of Aging Research Women vs. men Transitions Vocational vs. lower education G->B Higher vs. lower education G->D Living alone vs. with someone B->D Lost someone vs. no one Acceptable vs. (very) poor srh Good vs. (very) poor srh Mild vs. highly depressed Low vs. highly depressed Past vs. never smoker Current vs. never smoker BMI 22−28 vs. BMI < 22 BMI > 28 vs. BMI < 22 Light vs. no physical activity Heavy vs. no physical activity 2-3 vs. 0-1 medications 4+ vs. 0-1 medications Able vs. not able to stand up 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Figure 3: Multivariate predictions (hazard ratios) of transitions in cognitive health. Note. Highlighted hazards ratios are significant; (G) good health status; (B) bad health status; (D) dead. we found that living alone affected both dimensions of the before among nonagenarians. However, we did not ob- health status by decreasing the probability of deterioration. serve the same pattern for cognition: individuals in good Anyway, it was not possible to disentangle the causal direction cognitive status at baseline are slightly more likely to die of the association (whether individuals in better health con- within two years compared to the first experience dete- ditions are able to live alone or whether living alone helps rioration of their cognition. &e strengths of this study are protecting the health condition). the sample size and the extensive information available, Surprisingly, emotional characteristics did not have which is rare to find given the age (93 years old) of the significant effect on any of the health transitions analyzed individuals under analysis. &is made it possible to here, despite the fact that other scholars found that self-rated control for many covariates. &e weakness of this study is health and depression have an active role in explaining that, even though the data set is longitudinal, it was not transitions in physical and cognitive health among old in- possible to clearly identify the causal relationship of some dividuals [7, 29, 38, 39]. of the associations. For both health conditions, having a BMI higher than Transitions in both health dimensions were related to similar sociodemographic and behavioral characteristics, 22 (both categories “22–28” and “>28”) resulted in lower probability of dying both from a good and a bad health with some interesting exceptions, but, surprisingly, not to status, confirming previous findings on younger adults emotional factors. &e two health dimensions resulted as- [40, 41] and in mortality research [42]. Light to moderate sociated with each other in terms of transitions: being in a exercise was significantly associated with lower probability better health condition according to one of the two health of dying from both bad physical and cognitive status, while measures lowered the probability of worsening the other engaging in heavy physical activity was associated with a health status or dying from a bad condition. &is confirms lower risk of deterioration of the physical health condition what have been discussed by the extensive literature review and a lower chance of dying when already in bad cognitive by Clouston and colleagues [8] about the role of the physical status. According to the instrument used by Nybo et al. condition at baseline on the transitions in cognitive health [14], the level of physical activity is related to the ability of and brings new evidence on the role of the cognitive status performing Activities of Daily Living (ADL). Other studies on the transitions in physical health for which the literature reported this association in terms of physical frailty so far has not found consistent evidence. [5, 9, 10] for disability transitions while only little is known about the association between physical exercise and cog- Data Availability nitive transitions [43]. As in the case of the living ar- rangement, it was not possible to distinguish the causal &e data that support the findings of this study are from 6e direction of the association between physical activity and 1905 Danish Cohort Study, but restrictions apply to the physical health. availability of these data, which were used under license for the current study, and so they are not publicly available. 5. Conclusions Our study sheds light on the dynamic relationship be- Conflicts of Interest tween physical and cognitive conditions among a cohort of nonagenarians, highlighting a “one-step worsening” &e authors declare that there are no conflicts of interest pattern in physical health, which has not been shown regarding the publication of this paper. Journal of Aging Research 7 estimates from CERAD,” Neurology, vol. 57, no. 6, pp. 957– Acknowledgments 964, 2001. [14] H. Nybo, D. Gaist, B. Jeune et al., “&e Danish 1905 cohort,” &e authors wish to thank James W. Vaupel, Kaare Journal of Aging and Health, vol. 13, no. 1, pp. 32–46, 2001. Christensen, and Mikael &inggaard for the helpful input [15] M. F. Folstein, S. E. Folstein, and P. R. McHugh, “Mini-mental and discussion during the conduction of this study. state,” Journal of Psychiatric Research, vol. 12, no. 3, pp. 189–198, 1975. Supplementary Materials [16] M. Folstein, J. C. Anthony, I. Parhad, B. Duffy, and E. M. Gruenberg, “&e meaning of cognitive impairment in Multistate analysis tables: S1–S4. Supplementary table S1 : the elderly,” Journal of the American Geriatrics Society, vol. 33, transition probabilities of the multistate model where states no. 4, pp. 228–235, 1985. are defined according to the physical health. Supplementary [17] M. &inggaard, M. McGue, B. Jeune, M. Osler, J. W. Vaupel, Table S2 : transition probabilities of the multistate model and K. Christensen, “Survival prognosis in very old adults,” where states are defined according to the cognitive health. Journal of the American Geriatrics Society, vol. 64, no. 1, Supplementary Table S3 : multivariate predictions (hazard pp. 81–88, 2016. [18] R. W. Bohannon, “Sit-to-Stand test for measuring perfor- ratios) of transitions in physical health. Supplementary Table mance of lower extremity muscles,” Perceptual and Motor S4 : multivariate predictions (hazard ratios) of transitions in Skills, vol. 80, no. 1, pp. 163–166, 1995. cognitive health. (Supplementary Materials) [19] J. M. Guralnik and L. Ferrucci, “Chapter 5. Demography and epidemiology,” in Hazzards Geriatr. Med. Gerontol, References J. B. Halter, J. G. Ouslander, M. E. Tinetti, S. Studenski, K. P. High, and S. Asthana, Eds., &e McGraw-Hill Com- [1] J. W. Vaupel, “Biodemographic trajectories of longevity,” panies, New York, NY, USA, 6th edition, 2009. Science, vol. 280, no. 5365, pp. 855–860, 1998. [20] J. M. Guralnik, L. Ferrucci, E. M. Simonsick, M. E. Salive, and [2] K. Christensen, G. Doblhammer, R. Rau, and J. W. Vaupel, R. B. Wallace, “Lower-extremity function in persons over the “Ageing populations: the challenges ahead,” 6e Lancet, age of 70 Years as a predictor of subsequent disability,” New vol. 374, no. 9696, pp. 1196–1208, 2009. England Journal of Medicine, vol. 332, no. 9, pp. 556–562, [3] J. W. Vaupel, “Biodemography of human ageing,” Nature, vol. 464, no. 7288, pp. 536–542, 2010. [21] J. E. Ware, M. Kosinski, and S. D. Keller, “A 12-item short- [4] World Population Prospects - Population Division - United form health survey,” Medical Care, vol. 34, no. 3, pp. 220–233, Nations n.d, 2018, https://esa.un.org/unpd/wpp/. [5] S. E. Hardy, J. A. Dubin, T. R. Holford, and T. M. Gill, [22] M. Roth, E. Tym, C. Q. Mountjoy et al., “CAMDEX: a “Transitions between states of disability and independence standardised instrument for the diagnosis of mental disorder among older persons,” American Journal of Epidemiology, in the elderly with special reference to the early detection of vol. 161, no. 6, pp. 575–584, 2005. dementia,” British Journal of Psychiatry, vol. 149, no. 6, [6] A. van den Hout and F. E. Matthews, “Multi-state analysis of pp. 698–709, 1986. cognitive ability data: a piecewise-constant model and a [23] P. Hougaard, “Multi-state models: a review,” Lifetime Data Weibull model,” Statistics in Medicine, vol. 27, no. 26, Analysis, vol. 5, no. 3, pp. 239–264, 1999. pp. 5440–5455, 2008. [24] C. H. Jackson, “Multi-state models for panel data: the msm [7] K. M. Mehta, K. Yaffe, and K. E. Covinsky, “Cognitive im- package for R,” Journal of Statistical Software, vol. 38, 2011. pairment, depressive symptoms, and functional decline in [25] J. Chen and J. Shao, “Nearest neighbor imputation for survey older people,” Journal of the American Geriatrics Society, data,” Journal of Official Statistics, vol. 16, pp. 113–131, 2000. vol. 50, no. 6, pp. 1045–1050, 2002. [26] R Core Team, “&e R project for statistical computing,” 2017, [8] S. A. P. Clouston, P. Brewster, D. Kuh et al., “&e dynamic https://www.r-project.org/. relationship between physical function and cognition in [27] F. Orfila, M. Ferrer, R. Lamarca, C. Tebe, A. Domingo-Sal- longitudinal aging cohorts,” Epidemiologic Reviews, vol. 35, vany, and J. Alonso, “Gender differences in health-related no. 1, pp. 33–50, 2013. quality of life among the elderly: the role of objective func- [9] T. M. Gill, H. G. Allore, E. A. Gahbauer, and T. E. Murphy, tional capacity and chronic conditions,” Social Science & “Change IN disability after hospitalization or restricted ac- Medicine, vol. 63, no. 9, pp. 2367–2380, 2006. tivity IN older persons,” Jama, vol. 304, no. 17, pp. 1919–1928, [28] C. Rosano, E. M. Simonsick, T. B. Harris et al., “Association between physical and cognitive function in healthy elderly: the [10] T. M. Gill, H. G. Allore, S. E. Hardy, and Z. Guo, “&e dy- health, aging and body composition study,” Neuro- namic nature of mobility disability in older persons,” Journal epidemiology, vol. 24, no. 1-2, pp. 8–14, 2005. of the American Geriatrics Society, vol. 54, no. 2, pp. 248–254, [29] A. H. van Houwelingen, I. D. Cameron, J. Gussekloo et al., “Disability transitions in the oldest old in the general pop- [11] R. Nikolova, L. Demers, F. Beland, ´ and F. Giroux, “Transitions in the functional status of disabled community-living older ulation. &e Leiden 85-plus study,” AGE, vol. 36, no. 1, pp. 483–493, 2014. adults over a 3-year follow-up period,” Archives of Geron- tology and Geriatrics, vol. 52, no. 1, pp. 12–17, 2011. [30] H. Nybo, D. Gaist, B. Jeune, M. McGue, J. W. Vaupel, and K. Christensen, “Functional status and self-rated health in [12] A. B. Mitnitski, N. Fallah, C. B. Dean, and K. Rockwood, “A multi-state model for the analysis of changes in cognitive 2,262 nonagenarians: the Danish 1905 cohort survey,” Journal of the American Geriatrics Society, vol. 49, no. 5, pp. 601–609, scores over a fixed time interval,” Statistical Methods in Medical Research, vol. 23, no. 3, pp. 244–256, 2014. 2001. [31] H. Nybo, H. C. Petersen, D. Gaist et al., “Predictors of [13] P. J. Neumann, S. S. Araki, A. Arcelus et al., “Measuring Alzheimer’s disease progression with transition probabilities: mortality in 2,249 Nonagenariansaˆ &e Danish 1905-cohort 8 Journal of Aging Research survey,” Journal of the American Geriatrics Society, vol. 51, no. 10, pp. 1365–1373, 2003. [32] H.-m. Yu, S.-s. Yang, J.-w. Gao, L.-y. Zhou, R.-f. Liang, and C.-y. Qu, “Multi-state Markov model in outcome of mild cognitive impairments among community elderly residents in Mainland China,” International Psychogeriatrics, vol. 25, no. 5, pp. 797–804, 2013. [33] S. N. Austad, “Why women live longer than men: sex dif- ferences in longevity,” Gender Medicine, vol. 3, no. 2, pp. 79–92, 2006. [34] A. Barford, D. Dorling, G. D. Smith, and M. Shaw, “Life expectancy: women now on top everywhere,” BMJ, vol. 332, no. 7545, p. 808, 2006. [35] V. Zarulli, K. Christensen, J. W. Vaupel, R. Lindahl-Jacobsen, K. Christensen, and J. W. Vaupel, “Reply to Delanghe et al.: iron status is not likely to play a key role in the gender survival gap under extreme conditions,” Proceedings of the National Academy of Sciences, vol. 115, no. 18, p. E4150, 2018. [36] R. E. Marioni, M. J. Valenzuela, A. van den Hout, C. Brayne, and F. E. Matthews, “MRC cognitive function and ageing study. Active cognitive lifestyle is associated with positive cognitive health transitions and compression of morbidity from age sixty-five,” PLoS ONE, vol. 7, Article ID e50940, [37] S. Wei, L. Xu, and R. J. Kryscio, “Markov transition model to dementia with death as a competing event,” Computational Statistics & Data Analysis, vol. 80, pp. 78–88, 2014. [38] A. E. Stuck, J. M. Walthert, T. Nikolaus, C. J. Bula, ¨ C. Hohmann, and J. C. Beck, “Risk factors for functional status decline in community-living elderly people: a sys- tematic literature review,” Social Science & Medicine, vol. 48, no. 4, pp. 445–469, 1999. [39] J. Chodosh, D. M. Kado, T. E. Seeman, and A. S. Karlamangla, “Depressive symptoms as a predictor of cognitive decline: MacArthur studies of successful aging,” 6e American Journal of Geriatric Psychiatry, vol. 15, no. 5, pp. 406–415, 2007. [40] F. C. Drumond Andrade, A. I. N. Mohd Nazan, M. L. Lebrão, and Y. A. d. Oliveira Duarte, “&e impact of body mass index and weight changes on disability transitions and mortality in Brazilian older adults,” Journal of Aging Research, vol. 2013, pp. 1–11, 2013. [41] B. B. Cronk, D. K. Johnson, and J. M. Burns, “Body mass index and cognitive decline in mild cognitive impairment,” Alz- heimer Disease & Associated Disorders, vol. 24, no. 2, pp. 126–130, 2010. [42] M. &inggaard, R. Jacobsen, B. Jeune, T. Martinussen, and K. Christensen, “Is the relationship between BMI and mor- tality increasingly U-shaped with advancing age? a 10-year follow-up of persons aged 70-95 years,” 6e Journals of Gerontology Series A: Biological Sciences and Medical Sciences, vol. 65A, no. 5, pp. 526–531, 2010. [43] D. Laurin, R. Verreault, J. Lindsay, K. MacPherson, and K. Rockwood, “Physical activity and risk of cognitive im- pairment and dementia in elderly persons,” Arch Neurol, vol. 58, 2001. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Aging Research Hindawi Publishing Corporation

Understanding Health Deterioration and the Dynamic Relationship between Physical Ability and Cognition among a Cohort of Danish Nonagenarians

Loading next page...
 
/lp/hindawi-publishing-corporation/understanding-health-deterioration-and-the-dynamic-relationship-5TPGSqjVv2
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2020 Cosmo Strozza et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
2090-2204
eISSN
2090-2212
DOI
10.1155/2020/4704305
Publisher site
See Article on Publisher Site

Abstract

Hindawi Journal of Aging Research Volume 2020, Article ID 4704305, 8 pages https://doi.org/10.1155/2020/4704305 Research Article Understanding Health Deterioration and the Dynamic Relationship between Physical Ability and Cognition among a Cohort of Danish Nonagenarians 1,2 1 2 Cosmo Strozza , Virginia Zarulli, and Viviana Egidi Interdisciplinary Centre on Population Dynamics, University of Southern Denmark, 5000 Odense, Denmark Department of Statistical Sciences, Sapienza University of Rome, 00185 Rome, Italy Correspondence should be addressed to Cosmo Strozza; cstrozza@health.sdu.dk Received 25 February 2020; Accepted 15 April 2020; Published 3 June 2020 Academic Editor: Elke Bromberg Copyright © 2020 Cosmo Strozza et al. &is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. &is study aims to determine how demographics, socioeconomic characteristics, and lifestyle affect physical and cognitive health transitions among nonagenarians, whether these transitions follow the same patterns, and how each dimension affects the transitions of the other. We applied a multistate model for panel data to 2262 individuals over a 2-year follow-up period from the 1905 Danish Cohort survey. Within two years from baseline, the transition probability from good to bad physical health—ability to stand up from a chair—was higher than dying directly (29% vs. 25%), while this was not observed for cognition (24% vs. 27%) evaluated with Mini-Mental State Examination—a score lower than 24 indicates poor cognitive health. Probability of dying either from bad physical or cognitive health condition was 50%. Health transitions were associated with sex, education, living alone, body mass index, and physical activity. Physical and cognitive indicators were associated with deterioration of cognitive and physical status, respectively, and with survivorship from a bad health condition. We conclude that physical and cognitive health deteriorated differently among nonagenarians, even if they were related to similar sociodemographic and lifestyle characteristics and resulted dynamically related with each other. found that physical functioning at baseline was associated 1. Introduction with longitudinal changes in cognition but the opposite &e proportion of the oldest-old has increased during the relationship was inconsistent. Physical mobility and func- last decades as a consequence of the decline in old-age tioning dynamically interact between healthy and unhealthy mortality [1–3]. &e share of nonagenarians in Denmark states [5, 9–11]. Cognitive health declines with age more increased from around 0.08% in 1950 to 0.82% in 2020 and is linearly, even though this decline can cover a more complex expected to further grow during the next years, reaching pattern [6, 12, 13]. It is therefore crucial to investigate 2.03% in 2050 [4]. &is phenomenon is taking place in most further how the physical and cognitive deteriorations evolve developed countries, fueling a growing interest on the health and whether they follow different patterns. &e relationship conditions of oldest-old [2]. Health transitions at older ages between the two dimensions of the health status has been are of particular interest as deterioration of both physical widely investigated. However, the literature lacks studies on and cognitive health conditions is very likely [5, 6]. the oldest-old and this is why in this study we focus on Physical and cognitive health decline have been inves- individuals aged 90+. tigated, in order to understand whether they can be partially With this analysis, we aim at investigating (1) how de- explained by other health characteristics [7]. A systematic mographic and socioeconomic characteristics and lifestyle review of the relationship between physical functioning and habits affect transitions in physical and cognitive health; (2) cognition was published by Clouston et al. in 2013 [8], which whether these transitions follow the same patterns; and (3) 2 Journal of Aging Research socioeconomic status of the participants. It was grouped how does each dimension (physical or cognitive) affect the transitions of the other dimension. To our knowledge, this is into three categories: (1) elementary school; (2) vocational education; and (3) higher education. Living condition was the first study that analyzes the relationship between physical and cognitive decline and the determinants of transitions in divided into people living (1) alone and (2) with someone. these two health dimensions (physical and cognitive) among &e loss of a close person, self-rated health, and depression nonagenarians. were used to assess the general health perception of the participant and the feelings related to it. &e loss of a close person was categorized into two classes: (1) lost someone 2. Methods (spouse, sons, and close friends) and (2) no people lost due 2.1. Study Population and Measures. &e study population to death within the last five years. Self-rated health was comes from the 1905 Danish Cohort survey, which contains assessed with the first question of Short-Form 12 (SF12) questionnaire [21]: ”How do you consider your health in many individual level information on the members of the cohort born in Denmark in 1905 interviewed and tested for general?”. It was grouped in three categories: (1) very poor or poor; (2) acceptable; and (3) good or excellent. Depression physical and mental health in their home by a survey agency. It is a longitudinal multiassessment survey conducted from was assessed using an adaptation of the depression section of 1998 to 2005 with four waves realized every 2-3 years. the Cambridge mental disorders of the elderly examination Detailed information about the study design are available in [22]. It uses a scale from 17 to 52 and it was grouped into three Nybo and colleagues [14]. In this work we use the first two equal-size categories: (1) 17–22; (2) 23–28; and (3) 29–52. waves of the Danish 1905 Cohort Survey, collected in 1998 Among the health behaviors, smoking habit was categorized and 2000, when the oldest-old were, respectively, 93 and 95 into (1) never smoked; (2) past smoker; and (3) current years old. &e initial population, corresponding to the study smoker. BMI was calculated on the basis of the reported population, was composed of 2626 individuals. &ey rep- height and weight at the interview and categorized into three groups: (1)< 22; (2) 22–28; and (3)> 28. Physical activity was resent 62.8% of the potential participants: individuals born in 1905 and living in Denmark. At the second data collection assessed by asking if they were performing light (light gar- dening, short walks, or bicycle rides) or heavy (heavy gar- in 2000, 874 were found to be dead (38.6%) reducing to 1388 individuals the number of potential participants to the dening, long walks or bicycle rides, sports, gymnastics, or second wave of the study. &e final population interviewed dancing) exercises at the time of the interview. It was grouped in 2000 was composed of 1086 individuals (78.2% of the into three categories: (1) never or not able; (2) light physical potential participants). activity; and (3) heavy physical activity. &e number of &e cognitive function was measured with the Mini- medications (daily intake) was coded according to the An- Mental State Examination (MMSE): the higher the score atomical &erapeutic Chemical classification system and it was grouped into three equal-size categories: (1) 0-1; (2) 2-3; (0–30), the better the cognitive status [15]. We grouped it into three standard categories, in order to distinguish people and (3) 4+. &e main reference for variable selection and classifi- with severe (0–17), mild (18–23), and no cognitive im- pairment (24–30) based on the most frequently used cate- cation is Appendix S1 of the article by &inggaard et al. (2016) that uses the same study population [17]. &e pro- gorization in literature [16, 17]. &e physical function was assessed by the Chair-Stand test: the elderly who can stand portion of dropouts is 13.4%. We performed a sensitivity up from a chair have better functional status than who need analysis in order to check whether dropout was associated to use hands or cannot do it. &is test was found to be a good with bad health and we did not find any significant asso- predictor of disability and mortality among the elderly other ciation with both health indicators. than a proper instrument to measure lower body strength [17–20]. 2.2. Statistical Analysis. We applied a multistate model for We dichotomized both health indicators, in order to panel data—with Markov chain assumption—[23, 24] to create two categories: healthy and unhealthy oldest-old. assess the association between the many potential drivers Regarding cognitive health, individuals were considered measured on the Danish nonagenarians and the probability cognitively impaired when reporting a MMSE score from 0 of transitioning from one health state to another (defined as to 23 and not cognitively impaired when the score was transition probability). &e possible transitions are from between 24 and 30. Regarding physical ability, individuals who were not able to stand up from the chair, even with aids, good to bad health status, from good health to death, and from bad health to death. were considered in bad physical health while individuals able to stand up from the chair, with and without use of aids, were &e multistate model we used is based on a stochastic multistate process (X(t), t ∈ T) with a finite state-space considered in good physical health. S � 1, ..., N , where T � [0, τ], τ<∞ represents the time Demographic and socioeconomic characteristics (sex, { } (discrete, for panel data). It is fully characterized through education, and living conditions), critical events (loss of a transition probabilities between states h and j: close relative or friend), health characteristics and be- haviors (self-rated health and depression, smoking habits, (s, t) � P(X(t) � j | X(s) � h). (1) body mass index (BMI), physical activity, and use of hj medications) were considered as confounders and con- for h, j ∈ S, t, s ∈ T or through transition intensities: trolled in the analysis. Education was used to measure the Journal of Aging Research 3 32.3% of women were not depressed) without declaring p (t, t + Δt) hj (2) α (t) � lim , hj better health conditions than women (12.5 % of men rated Δt⟶0 Δt their health as good or excellent while 14.2 % of women representing the instantaneous hazard of progression to did it). state j conditionally on occupying state h at the previous In terms of health behaviors, except for the higher share time. According to the Markov assumption, the probability of (past or current) smokers (78.8% of men vs. 32.4% of of the next transition depends only on the state occupied at women), men had higher BMI (73.1% of men vs. 55.3% of the time t. women had a BMI higher than 22) and performed more &e effect of the explanatory variables z on the tran- physical activity (43.8% of men vs. 28.9% of women perform it sition intensity for individual i at time t is modeled using some physical activity) than women. More details about proportional intensities, replacing α with baseline characteristics of the population are available in hj Table 1. 0 T α z 􏼁 � α exp􏼐β z 􏼑. (3) hj it hj hj it Men scored better in terms of cognitive (48.5% of men vs. 40.6% of women were not cognitively impaired) and We conducted the analysis separately for physical and physical (52.1% of men vs. 41.5% of women were able to cognitive health, in order to be able to include the baseline stand up from the chair without any aid) health compared to status of each dimension (cognitive or physical resp.) as women as reported in Table 2. potential driver in the model for the transitions related to the other dimension. 3.2. Multi-State Analysis Results. We analyzed physical and States have been defined according to the MMSE, when cognitive health deterioration in two different models in- assessing cognition, and according to the Chair-Stand test, cluding, respectively, cognitive and physical baseline health when the focus was on the physical status. Based on both status because the main aim of the study is to examine the classifications, we divided participants into two groups based dynamic relationship between these two health aspects and on their good or bad health condition. not because we considered them independent. &is implies Transitions between four states (good health, bad health, that part of the individuals in the different states of the two nonparticipant but alive, and nonparticipant because dead) analyzes are the same, resulting in similar transition have been estimated through transition probabilities. We probabilities and covariates associated with transition evaluated the effect of the covariates on the transition in- intensities. tensities only for the “worsening” transitions: from good to At baseline, 44.2% of the individuals were in good bad health condition, from good health condition to death, physical health while 42.7% were in good cognitive health. and from bad health condition to death. As expected, only After two years, 38.6% of the study population died while few people experienced “improving” transitions, as this is 13.4% dropped out from the study. unlikely at very old ages. &e probability of moving from a good to a bad physical Because of the relatively small number of individuals in health condition within two years was higher than dying analysis, we could only use the dichotomic classification of directly (29% vs. 25%). People in bad physical health con- MMSE and Chair-Stand test, as the sample size was too small dition have a 50% probability of dying from a bad physical to estimate the coefficients with a finer classification of the health status within two years. variables. We could not perform the analysis separately for When considering the cognitive health, the results men and women due to the small number of nonagenarian showed a different pattern. &e probability of worsening a men in the sample. good cognitive health condition was lower than experiencing We used methods of imputation with survey data [25] to death directly within two years (24% vs. 27%), while indi- deal with missing at random values. More information about viduals in a bad cognitive health have a 47% probability of the imputation method is available in Supplementary Text S1. dying from that condition in the next two years as shown in Statistical analysis was performed using R version 3.5.0 Figure 1. [26]. &e complete transition probabilities are available in Supplementary Tables S1 and S2. 3. Results &e effect of covariates on the transition intensities is reported in Figures 2 and 3 for the physical states and 3.1.DescriptiveResults. Of the 2262—93 years old—baseline cognitive states, respectively. participants of the study, one-fourth were men (25.8%) while Full details about the two models are available in Sup- the rest of the people were women (74.2%). plementary Tables S3 and S4. Men had, on average, a higher education level than women, especially in terms of vocational education (32.9% of men vs. 14.2% of women). Fewer men were living alone 3.2.1. Physical Health Transitions. Being women was asso- compared to women (50.5% of men vs. 64.4% of women). ciated with lower probability of dying for people in bad More men experienced the loss of a close person physical health (female vs. male HR = 0.66) as well as living (spouse, children, and close friends) due to death during alone (living alone vs. with someone HR = 0.60). Living the last five years (71.7% of men vs. 66.9% of women) but alone was also significantly associated with a lower proba- they reported lower rates of depression (39.0% of men vs. bility of transitioning from a good to a bad physical health 4 Journal of Aging Research Table 1: Characteristics of the study population in the first wave in (HR = 0.52). Having a BMI higher than 22 statistically de- 1998 when the individuals were 93 years old. creased the probability of dying, both from a good (BMI 22–28 vs.< 22: HR = 0.45) and a bad (BMI > 28 vs.< 22: Sex HR = 0.63) physical health. Performing physical activity Characteristics M F T p lowered the transition probability from good to bad physical n % n % n % health (heavy vs. no physical activity: HR = 0.35) and from Sample 584 25.8 1678 74.2 2262 100.0 bad physical health to death (light vs. no physical activity: Education <0.001 HR = 0.73). Finally, also being cognitively not impaired was Elementary 292 50.0 1254 74.7 1546 68.3 statistically associated with a lower probability of worsening Vocational 192 32.9 238 14.2 430 19.0 the physical health (HR = 0.47) and dying from a bad one Higher 100 17.1 186 11.1 286 12.6 (HR = 0.62). Living alone <0.001 No 289 49.5 598 35.6 887 39.2 Yes 295 50.5 1080 64.4 1375 60.8 3.2.2. Cognitive Health Transitions. Being a woman was Loss of a close person 0.033 associated with a lower probability of death (from good No 165 28.3 556 33.1 721 31.9 health: HR = 0.42; from bad health: HR = 0.65). Having Yes 419 71.7 1122 66.9 1541 68.1 higher level of education decreased the probability of Self-rated health 0.013 deteriorating the cognitive health (HR = 0.55) as well as Very poor or poor 307 52.6 886 52.8 1193 52.7 living alone (HR = 0.49), which was also a protective factor Acceptable 204 34.9 553 33.0 757 33.5 against transitioning from bad cognitive status to death Good or excellent 73 12.5 239 14.2 312 13.8 (HR = 0.59). BMI higher than 22 reduced the probability of Depression 0.008 dying from a good (BMI 22–28 vs.< 22: HR = 0.44) and a 29–52 184 31.5 591 35.2 775 34.3 bad (BMI > 28 vs.< 22: HR = 0.65) cognitive health. Doing 23–28 172 29.5 545 32.5 717 31.7 physical activity was significantly related to lower tran- 17–22 228 39.0 542 32.3 770 34.0 sition rates from bad to death (light vs. no physical activity: Smoke <0.001 HR = 0.65, heavy vs. no physical activity: HR = 0.52). As Current smoker 144 24.7 171 10.2 315 13.9 expected, using more than four medications per day was Past smoker 316 54.1 372 22.2 688 30.4 associated with higher probability of death when already in Never smoked 124 21.2 1135 67.6 1259 55.7 a bad cognitive health (HR = 1.27). Finally, being able to Body Mass Index <0.001 stand up from the chair without any aid was statistically < 22 157 26.9 750 44.7 907 40.1 associated with a lower probability of worsening the 22–28 348 59.6 785 46.8 1133 50.1 > 28 79 13.5 143 8.5 222 9.8 cognitive health (HR = 0.53) and dying from a bad one (HR = 0.61). Physical activity <0.001 None/irrelevant 328 56.2 1193 71.1 1521 67.2 Light 177 30.3 390 23.2 567 25.1 4. Discussion Heavy 79 13.5 95 5.7 174 7.7 Number of medications 0.057 &e increasing proportion of the oldest-old people in the last 4+ 228 39.0 714 42.6 942 41.6 decades increased the attention of researchers and policy 2-3 153 26.2 423 25.2 576 25.5 makers on this subgroup of individuals [2, 3]. As physical 0-1 203 34.8 541 32.2 744 32.9 and cognitive health are two dynamic processes and their Men versus women from Pearson χ test. deterioration is likely, especially at older ages, in the recent years it became a widely investigated topic [5, 6,9, 10, 12]. Finding the determinants of physical and cognitive health Table 2: Health conditions of the study population in the first wave changes and analyzing their longitudinal relationship are in 1998 when the individuals were 93 years old. considered, nowadays, two of the major public health challenges [27, 28]. However, only few studies analyzed such Sex deteriorations among the oldest-old [8, 29]. Studying the Characteristics M F T p determinants of physical and cognitive health transitions n % n % n % among very old people and analyzing the relationship be- Physical ability: <0.001 tween these two conditions will help to shed light on which Chair-Stand Test are the most vulnerable groups. Not able 70 12.0 293 17.5 363 16.0 &is study uses two waves of the 1905 Danish Cohort With use of hands 210 36.0 689 41.1 899 39.7 survey [14] to study the transitions in physical and cognitive Without use of hands 304 52.1 696 41.5 1000 44.2 health among individuals aged 93 at the baseline (1998) and Cognitive health: Mini-Mental <0.001 95 at the second wave (2000). Studies on this cohort showed State Examination 0–17 124 21.2 472 28.1 596 26.3 that high level of disability and poor cognitive and physical 18–23 177 30.3 524 31.2 701 31.0 performance are strong predictors of mortality in the oldest- 24–30 283 48.5 682 40.6 965 42.7 old [30, 31]. More precisely, &inggaad et al. (2016) found Men versus women from Pearson χ test. that being able to stand up from a chair and having a good Journal of Aging Research 5 G G 0.25 0.27 D D 0.29 0.02 0.24 0.07 0.50 0.47 B B (a) (b) Figure 1: Transition probabilities of the multistate model where states are defined according to (a) physical health and (b) cognitive health. Note. (G) good health status; (B) bad health status; (D) dead. Women vs. men Transitions Vocational vs. lower education G->B Higher vs. lower education G->D Living alone vs. with someone B->D Lost someone vs. no one Acceptable vs. (very)poor srh Good vs. (very)poor srh Mild vs. highly depressed Low vs. highly depressed Past vs. never smoker Current vs. never smoker BMI 22−28 vs. BMI < 22 BMI > 28 vs. BMI < 22 Light vs. no physical activity Heavy vs. no physical activity 2-3 vs. 0-1 medications 4+ vs. 0-1 medications Not vs. cognitively impaired 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Figure 2: Multivariate predictions (hazard ratios) of transitions in physical health. Note. Highlighted hazards ratios are significant; (G) good health status; (B) bad health status; (D) dead. level of cognition increased the probability of surviving to individuals in good and bad state for both physical and age of 100 for both women and men of the 1905 Danish cognitive health. Cohort Study [17]. Demographic and socioeconomic variables in both cases Our results partially confirm the trends shown in the resulted associated with health transitions. Not surprisingly, literature for both physical and cognitive health over the women had a lower probability of death [33–35]. However, by years among adults and younger elderly [7, 9, 11, 29, 32]. analyzing physical and cognitive dimension separately, we Even at very old ages, for individuals in good physical health were able to uncover interesting dynamics. Being a woman did conditions, the probability of dying directly was lower than not affect significantly the transition from good to bad health. the probability of first experiencing a health deterioration. However, it was instead associated with a lower probability of &is is what we called here a “one-step worsening pattern.” dying from both good and bad cognitive status but only However, this pattern was not observed for cognitive health lowered the probability of dying from a bad physical health in which the probability of deteriorating the level of cog- condition. As expected, having a higher level of education nition was lower than dying directly from a good cognitive decreased the probability of cognitive decline, confirming the status (24% vs. 27%). results found among younger adults [12, 32, 36, 37]. However, &e analysis of potential drivers of the health decline we found that the level of education did not affect the physical showed similar results for physical and cognitive health, status, contrary to what has been found for a similarly aged (8 years younger) cohort of Canadian elderly [11, 29]. Living showing that the two dimensions of the health status follow somewhat similar patterns. However, it is important to point alone is widely considered a predictor of physical [9, 10, 29] out that this might also partly be due to the overlap of but not for cognitive health transitions. In our study, instead, 6 Journal of Aging Research Women vs. men Transitions Vocational vs. lower education G->B Higher vs. lower education G->D Living alone vs. with someone B->D Lost someone vs. no one Acceptable vs. (very) poor srh Good vs. (very) poor srh Mild vs. highly depressed Low vs. highly depressed Past vs. never smoker Current vs. never smoker BMI 22−28 vs. BMI < 22 BMI > 28 vs. BMI < 22 Light vs. no physical activity Heavy vs. no physical activity 2-3 vs. 0-1 medications 4+ vs. 0-1 medications Able vs. not able to stand up 0.1 0.2 0.5 1.0 2.0 5.0 10.0 Figure 3: Multivariate predictions (hazard ratios) of transitions in cognitive health. Note. Highlighted hazards ratios are significant; (G) good health status; (B) bad health status; (D) dead. we found that living alone affected both dimensions of the before among nonagenarians. However, we did not ob- health status by decreasing the probability of deterioration. serve the same pattern for cognition: individuals in good Anyway, it was not possible to disentangle the causal direction cognitive status at baseline are slightly more likely to die of the association (whether individuals in better health con- within two years compared to the first experience dete- ditions are able to live alone or whether living alone helps rioration of their cognition. &e strengths of this study are protecting the health condition). the sample size and the extensive information available, Surprisingly, emotional characteristics did not have which is rare to find given the age (93 years old) of the significant effect on any of the health transitions analyzed individuals under analysis. &is made it possible to here, despite the fact that other scholars found that self-rated control for many covariates. &e weakness of this study is health and depression have an active role in explaining that, even though the data set is longitudinal, it was not transitions in physical and cognitive health among old in- possible to clearly identify the causal relationship of some dividuals [7, 29, 38, 39]. of the associations. For both health conditions, having a BMI higher than Transitions in both health dimensions were related to similar sociodemographic and behavioral characteristics, 22 (both categories “22–28” and “>28”) resulted in lower probability of dying both from a good and a bad health with some interesting exceptions, but, surprisingly, not to status, confirming previous findings on younger adults emotional factors. &e two health dimensions resulted as- [40, 41] and in mortality research [42]. Light to moderate sociated with each other in terms of transitions: being in a exercise was significantly associated with lower probability better health condition according to one of the two health of dying from both bad physical and cognitive status, while measures lowered the probability of worsening the other engaging in heavy physical activity was associated with a health status or dying from a bad condition. &is confirms lower risk of deterioration of the physical health condition what have been discussed by the extensive literature review and a lower chance of dying when already in bad cognitive by Clouston and colleagues [8] about the role of the physical status. According to the instrument used by Nybo et al. condition at baseline on the transitions in cognitive health [14], the level of physical activity is related to the ability of and brings new evidence on the role of the cognitive status performing Activities of Daily Living (ADL). Other studies on the transitions in physical health for which the literature reported this association in terms of physical frailty so far has not found consistent evidence. [5, 9, 10] for disability transitions while only little is known about the association between physical exercise and cog- Data Availability nitive transitions [43]. As in the case of the living ar- rangement, it was not possible to distinguish the causal &e data that support the findings of this study are from 6e direction of the association between physical activity and 1905 Danish Cohort Study, but restrictions apply to the physical health. availability of these data, which were used under license for the current study, and so they are not publicly available. 5. Conclusions Our study sheds light on the dynamic relationship be- Conflicts of Interest tween physical and cognitive conditions among a cohort of nonagenarians, highlighting a “one-step worsening” &e authors declare that there are no conflicts of interest pattern in physical health, which has not been shown regarding the publication of this paper. Journal of Aging Research 7 estimates from CERAD,” Neurology, vol. 57, no. 6, pp. 957– Acknowledgments 964, 2001. [14] H. Nybo, D. Gaist, B. Jeune et al., “&e Danish 1905 cohort,” &e authors wish to thank James W. Vaupel, Kaare Journal of Aging and Health, vol. 13, no. 1, pp. 32–46, 2001. Christensen, and Mikael &inggaard for the helpful input [15] M. F. Folstein, S. E. Folstein, and P. R. McHugh, “Mini-mental and discussion during the conduction of this study. state,” Journal of Psychiatric Research, vol. 12, no. 3, pp. 189–198, 1975. Supplementary Materials [16] M. Folstein, J. C. Anthony, I. Parhad, B. Duffy, and E. M. Gruenberg, “&e meaning of cognitive impairment in Multistate analysis tables: S1–S4. Supplementary table S1 : the elderly,” Journal of the American Geriatrics Society, vol. 33, transition probabilities of the multistate model where states no. 4, pp. 228–235, 1985. are defined according to the physical health. Supplementary [17] M. &inggaard, M. McGue, B. Jeune, M. Osler, J. W. Vaupel, Table S2 : transition probabilities of the multistate model and K. Christensen, “Survival prognosis in very old adults,” where states are defined according to the cognitive health. Journal of the American Geriatrics Society, vol. 64, no. 1, Supplementary Table S3 : multivariate predictions (hazard pp. 81–88, 2016. [18] R. W. Bohannon, “Sit-to-Stand test for measuring perfor- ratios) of transitions in physical health. Supplementary Table mance of lower extremity muscles,” Perceptual and Motor S4 : multivariate predictions (hazard ratios) of transitions in Skills, vol. 80, no. 1, pp. 163–166, 1995. cognitive health. (Supplementary Materials) [19] J. M. Guralnik and L. Ferrucci, “Chapter 5. Demography and epidemiology,” in Hazzards Geriatr. Med. Gerontol, References J. B. Halter, J. G. Ouslander, M. E. Tinetti, S. Studenski, K. P. High, and S. Asthana, Eds., &e McGraw-Hill Com- [1] J. W. Vaupel, “Biodemographic trajectories of longevity,” panies, New York, NY, USA, 6th edition, 2009. Science, vol. 280, no. 5365, pp. 855–860, 1998. [20] J. M. Guralnik, L. Ferrucci, E. M. Simonsick, M. E. Salive, and [2] K. Christensen, G. Doblhammer, R. Rau, and J. W. Vaupel, R. B. Wallace, “Lower-extremity function in persons over the “Ageing populations: the challenges ahead,” 6e Lancet, age of 70 Years as a predictor of subsequent disability,” New vol. 374, no. 9696, pp. 1196–1208, 2009. England Journal of Medicine, vol. 332, no. 9, pp. 556–562, [3] J. W. Vaupel, “Biodemography of human ageing,” Nature, vol. 464, no. 7288, pp. 536–542, 2010. [21] J. E. Ware, M. Kosinski, and S. D. Keller, “A 12-item short- [4] World Population Prospects - Population Division - United form health survey,” Medical Care, vol. 34, no. 3, pp. 220–233, Nations n.d, 2018, https://esa.un.org/unpd/wpp/. [5] S. E. Hardy, J. A. Dubin, T. R. Holford, and T. M. Gill, [22] M. Roth, E. Tym, C. Q. Mountjoy et al., “CAMDEX: a “Transitions between states of disability and independence standardised instrument for the diagnosis of mental disorder among older persons,” American Journal of Epidemiology, in the elderly with special reference to the early detection of vol. 161, no. 6, pp. 575–584, 2005. dementia,” British Journal of Psychiatry, vol. 149, no. 6, [6] A. van den Hout and F. E. Matthews, “Multi-state analysis of pp. 698–709, 1986. cognitive ability data: a piecewise-constant model and a [23] P. Hougaard, “Multi-state models: a review,” Lifetime Data Weibull model,” Statistics in Medicine, vol. 27, no. 26, Analysis, vol. 5, no. 3, pp. 239–264, 1999. pp. 5440–5455, 2008. [24] C. H. Jackson, “Multi-state models for panel data: the msm [7] K. M. Mehta, K. Yaffe, and K. E. Covinsky, “Cognitive im- package for R,” Journal of Statistical Software, vol. 38, 2011. pairment, depressive symptoms, and functional decline in [25] J. Chen and J. Shao, “Nearest neighbor imputation for survey older people,” Journal of the American Geriatrics Society, data,” Journal of Official Statistics, vol. 16, pp. 113–131, 2000. vol. 50, no. 6, pp. 1045–1050, 2002. [26] R Core Team, “&e R project for statistical computing,” 2017, [8] S. A. P. Clouston, P. Brewster, D. Kuh et al., “&e dynamic https://www.r-project.org/. relationship between physical function and cognition in [27] F. Orfila, M. Ferrer, R. Lamarca, C. Tebe, A. Domingo-Sal- longitudinal aging cohorts,” Epidemiologic Reviews, vol. 35, vany, and J. Alonso, “Gender differences in health-related no. 1, pp. 33–50, 2013. quality of life among the elderly: the role of objective func- [9] T. M. Gill, H. G. Allore, E. A. Gahbauer, and T. E. Murphy, tional capacity and chronic conditions,” Social Science & “Change IN disability after hospitalization or restricted ac- Medicine, vol. 63, no. 9, pp. 2367–2380, 2006. tivity IN older persons,” Jama, vol. 304, no. 17, pp. 1919–1928, [28] C. Rosano, E. M. Simonsick, T. B. Harris et al., “Association between physical and cognitive function in healthy elderly: the [10] T. M. Gill, H. G. Allore, S. E. Hardy, and Z. Guo, “&e dy- health, aging and body composition study,” Neuro- namic nature of mobility disability in older persons,” Journal epidemiology, vol. 24, no. 1-2, pp. 8–14, 2005. of the American Geriatrics Society, vol. 54, no. 2, pp. 248–254, [29] A. H. van Houwelingen, I. D. Cameron, J. Gussekloo et al., “Disability transitions in the oldest old in the general pop- [11] R. Nikolova, L. Demers, F. Beland, ´ and F. Giroux, “Transitions in the functional status of disabled community-living older ulation. &e Leiden 85-plus study,” AGE, vol. 36, no. 1, pp. 483–493, 2014. adults over a 3-year follow-up period,” Archives of Geron- tology and Geriatrics, vol. 52, no. 1, pp. 12–17, 2011. [30] H. Nybo, D. Gaist, B. Jeune, M. McGue, J. W. Vaupel, and K. Christensen, “Functional status and self-rated health in [12] A. B. Mitnitski, N. Fallah, C. B. Dean, and K. Rockwood, “A multi-state model for the analysis of changes in cognitive 2,262 nonagenarians: the Danish 1905 cohort survey,” Journal of the American Geriatrics Society, vol. 49, no. 5, pp. 601–609, scores over a fixed time interval,” Statistical Methods in Medical Research, vol. 23, no. 3, pp. 244–256, 2014. 2001. [31] H. Nybo, H. C. Petersen, D. Gaist et al., “Predictors of [13] P. J. Neumann, S. S. Araki, A. Arcelus et al., “Measuring Alzheimer’s disease progression with transition probabilities: mortality in 2,249 Nonagenariansaˆ &e Danish 1905-cohort 8 Journal of Aging Research survey,” Journal of the American Geriatrics Society, vol. 51, no. 10, pp. 1365–1373, 2003. [32] H.-m. Yu, S.-s. Yang, J.-w. Gao, L.-y. Zhou, R.-f. Liang, and C.-y. Qu, “Multi-state Markov model in outcome of mild cognitive impairments among community elderly residents in Mainland China,” International Psychogeriatrics, vol. 25, no. 5, pp. 797–804, 2013. [33] S. N. Austad, “Why women live longer than men: sex dif- ferences in longevity,” Gender Medicine, vol. 3, no. 2, pp. 79–92, 2006. [34] A. Barford, D. Dorling, G. D. Smith, and M. Shaw, “Life expectancy: women now on top everywhere,” BMJ, vol. 332, no. 7545, p. 808, 2006. [35] V. Zarulli, K. Christensen, J. W. Vaupel, R. Lindahl-Jacobsen, K. Christensen, and J. W. Vaupel, “Reply to Delanghe et al.: iron status is not likely to play a key role in the gender survival gap under extreme conditions,” Proceedings of the National Academy of Sciences, vol. 115, no. 18, p. E4150, 2018. [36] R. E. Marioni, M. J. Valenzuela, A. van den Hout, C. Brayne, and F. E. Matthews, “MRC cognitive function and ageing study. Active cognitive lifestyle is associated with positive cognitive health transitions and compression of morbidity from age sixty-five,” PLoS ONE, vol. 7, Article ID e50940, [37] S. Wei, L. Xu, and R. J. Kryscio, “Markov transition model to dementia with death as a competing event,” Computational Statistics & Data Analysis, vol. 80, pp. 78–88, 2014. [38] A. E. Stuck, J. M. Walthert, T. Nikolaus, C. J. Bula, ¨ C. Hohmann, and J. C. Beck, “Risk factors for functional status decline in community-living elderly people: a sys- tematic literature review,” Social Science & Medicine, vol. 48, no. 4, pp. 445–469, 1999. [39] J. Chodosh, D. M. Kado, T. E. Seeman, and A. S. Karlamangla, “Depressive symptoms as a predictor of cognitive decline: MacArthur studies of successful aging,” 6e American Journal of Geriatric Psychiatry, vol. 15, no. 5, pp. 406–415, 2007. [40] F. C. Drumond Andrade, A. I. N. Mohd Nazan, M. L. Lebrão, and Y. A. d. Oliveira Duarte, “&e impact of body mass index and weight changes on disability transitions and mortality in Brazilian older adults,” Journal of Aging Research, vol. 2013, pp. 1–11, 2013. [41] B. B. Cronk, D. K. Johnson, and J. M. Burns, “Body mass index and cognitive decline in mild cognitive impairment,” Alz- heimer Disease & Associated Disorders, vol. 24, no. 2, pp. 126–130, 2010. [42] M. &inggaard, R. Jacobsen, B. Jeune, T. Martinussen, and K. Christensen, “Is the relationship between BMI and mor- tality increasingly U-shaped with advancing age? a 10-year follow-up of persons aged 70-95 years,” 6e Journals of Gerontology Series A: Biological Sciences and Medical Sciences, vol. 65A, no. 5, pp. 526–531, 2010. [43] D. Laurin, R. Verreault, J. Lindsay, K. MacPherson, and K. Rockwood, “Physical activity and risk of cognitive im- pairment and dementia in elderly persons,” Arch Neurol, vol. 58, 2001.

Journal

Journal of Aging ResearchHindawi Publishing Corporation

Published: Jun 3, 2020

References