Physical frailty prediction model for the oldest old

ABSTRACT Objective: to present a physical frailty prediction model for oldest old users of primary health care, according to clinical variables. Method: cross-sectional study with proportional stratified sample of 243 oldest old subjects. Data were collected through a structured clinical questionnaire, handgrip strength test, walking speed, weight loss, fatigue/exhaustion, and physical activity level. For the analysis of the data, univariate and multivariate analysis by logistic regression were used (p<0.05), which resulted in prediction models. The odds ratios (95% Confidence Interval) of the models were calculated. Each model was evaluated by deviance analysis, likelihood ratios, specificity and sensitivity, considering the most adequate. All ethical and legal precepts were followed. Results: the prediction model elected was composed of metabolic diseases, dyslipidemias and hospitalization in the last 12 months. Conclusion: clinical variables interfere in the development of the physical frailty syndrome in oldest old users of basic health unit. The choice of a physical frailty regression model is the first step in the elaboration of clinical methods to evaluate the oldest old in primary care.


Introduction
Senescence is characterized by inevitable structural, physiological, and functional changes in the organism.
For some people, these changes are accentuated and lead to increased risk of morbidity and mortality, while others remain robust, even in old age. Given the heterogeneity of the aging process, the concept of frailty has been increasingly discussed.
Physical frailty is a multicausal medical condition with several associated factors. It is characterized by a decrease in strength and endurance and an increase in the individual's vulnerability for developing increased dependency and/or mortality (1) . This syndrome is an important marker of an individual's physiological reserve and an indicator of the risk of negative outcomes to the health of the oldest-old (2)(3) .
Aiming to construct a phenotype of frailty, international authors developed a model based on the markers decrease in handgrip strength, self-reported exhaustion or fatigue, diminished walking speed, unintentional weight loss and low level of physical activity (4) . Older adults without any of the markers are considered non-frail, those with one or two markers are called pre-frail and the presence of three or more markers characterizes frail older adults.
The oldest-old are characterized as a group that should be screened, even without evidence of disability (1,(5)(6) . The high prevalence of physical frailty and the increase in the demand for health services among the oldest-old has stimulated discussions for the definition of predictors to better evaluate, characterize and monitor this age group (7) .
Among the factors related to the development and worsening of the frailty syndrome, the most prominent are clinical factors. An international cross-sectional study with 115 participants aged 65 and over in the Singapore University Hospital highlighted the association between the syndrome and recurrent hospital admissions, polypharmacy, and falls (8) . Another international longitudinal study conducted with 2,925 Italian older adults with a mean age of 74.4 years showed that clinical variables, such as polypharmacy, chronic diseases and obesity, may worsen the frailty state (9) . Similar results were obtained in a national cross-sectional study carried out with 385 independent older adults in the city of Ribeirão Preto, São Paulo, which found that frail older adults had a greater chance of having had a hospitalization in the prior 12 months, had more medical visits, and had more cases of cerebrovascular events, diabetes, urinary and fecal incontinence, osteoporosis and neoplasms (10) .
The identification of clinical factors associated with adverse outcomes for the health of older adults and the careful evaluation of the markers of physical frailty are essential for an adequate management of the syndrome, with the elaboration of effective interventions in the care of older adults.
One of the possible strategies for screening for physical frailty among older adults is the use of prediction models. International authors point out that this is a simple and clinically relevant tool that allows the use of routinely collected data in a systematic manner, optimizing data quality and reliability (11) . For nurses in primary care, strategies like this can increase the speed and effectiveness of the care provided to the older adult.
The present study aimed to present a physical frailty prediction model for oldest-old patients of primary health care according to clinical variables.

Method
Cross-sectional study conducted in households in the area covered by three Basic Health Units (BHU) of the city of Curitiba, Paraná. The criteria for choosing the BHU were: having users belonging to the income classes C, D and E (12) , since the classes A and B are not included in the BHU care; and having a significant number of older adults registered. The study population consisted of older adults aged 80 years or over and registered in these BHU.
Proportional stratified sampling was adopted considering that none of the BHU was overestimated or underestimated. The sample calculation considered a beta power of 80% (1-ß), a 5% significance level(α=0.05) and a minimum significant difference of The following inclusion criteria were established for the participants: (a)being 80 years old or older; (b)being registered in one of the BHU of the research; (c)scoring higher than the cut-off in the cognitive test of the Mini-Mental State Examination (MMSE) (13) considering 13 points as illiterate, 18 as low (1 to 4 incomplete years) and average (4 to 8 incomplete years) education level and 26 as high education level (8 years or more) (14) . Older adults undergoing chemotherapy or with previous diagnosis of serious mental illness or deficits that prevented participation in the study were excluded.
In the case of older adults with no cognitive conditions to answer the research questions (n =36) at this stage, the family caregiver was invited to participate, for which the following inclusion criteria were adopted: a) being 18 years or older; b) being a family caregiver; c) be living with the older adult for at least three months. The markers of the syndrome were evaluated based on the phenotype of frailty (4) , with some adaptations.
Handgrip strength (HGS) was measured using a Jamar® hydraulic dynamometer. Three measurements in kilogram/force (Kgf) were taken with the dominant hand, with one-minute intervals to regain strength and the highest reading was recorded (16) . Values were adjusted according to gender and body mass index (BMI, in Kg/m 2 ), considering the values in the lowest quintile as markers of physical frailty ( Figure 1).  considered. An international literature review study evaluating walking speed tests, pointed out that sixmeter courses have been widely used with older adults and that 4 to 6-meter courses can be used, according to the purpose of the study (17) .
After adjusting for gender and height, values equal or higher than the cutoff points were considered frailty markers ( Figure 2).

Discussion
The prevalence of frailty among the oldest-old found in this study is slightly different from the results obtained in an international systematic review, which investigated the same index among older adults aged 60 and over who lived in communities in Latin American and Caribbean countries (19.6% frail) (20) . Another international review that assessed the prevalence of the syndrome in developing countries found a variation of 17% to 31% in Brazilian studies with similar samples (21) .
When considering the distribution of physical frailty by In the present study, the group of drugs that was significantly associated with the development of the syndrome was the antidiabetics. The mechanisms of the association between diabetes mellitus (DM) and frailty are still uncertain (23) ; however, there is evidence that DM is a potential risk factor for the development of the syndrome.
An international prospective study with 1750 older adults in Spain found an increased risk (OR 2.18, 95% CI, 1.42-3.37) of frailty in participants with diabetes.
In addition, it pointed out that the use of antidiabetic medication reduced the risk to 1.01 (95% CI, 0.46-2.20) (23) . The use of medications of this class by the oldest old may contribute to the maintenance of lean mass, muscular strength and functional capacity (24) . Therefore, the control of glycemic indexes is a fundamental goal in the management of physical frailty in the oldest old.
In the final regression model, the participants who were more likely to become frail had had a hospitalization in the last 12 months (OR=2.50), dyslipidemia (OR=0.32) and metabolic disease (OR=1.99).
The association of the syndrome with hospitalization in the last 12 months was highlighted in national (10) and international (8,(25)(26) authors. A systematic review evaluated 31 international articles and found that frailty increases the risk of hospitalization from 1.2 to 1.8 times (25) . This finding is similar to another cross-sectional study carried out with 993 older adults aged 70 years or older residing in Albacete, Spain, which found a 1.7 times increased risk of hospitalization (26) . Physical frailty generates a greater demand for care due to the reduced capacity of response to several stressors and the decrease in the of homeostasis, which causes negative health outcomes, such as hospitalization.
The high chances of hospitalization in the present study are possibly related to the age range of the sample.
There is a scarcity of national and international studies that exclusively address the oldest old. This approach is necessary due to the peculiarities of this age group, which are different from those of younger adults, especially due to higher rates of negative health outcomes.
Regarding the variable "dyslipidemia", which was associated with greater probability of physical frailty in this study, international authors (23,(27)(28) highlighted the relationship between this factor, sarcopenia and other morbidities, especially Diabetes Mellitus and cardiovascular diseases. Dyslipidemia associated with other chronic diseases favors the occurrence of neuromuscular changes and, consequently, leads to changes in walking speed, balance and to the physical frailty syndrome (28)(29) .
Regarding the influence of the variable "metabolic disease" in the predictive model, it is possibly related to neuroendocrine dysregulation, one of the factors that leads to the development of physical frailty (30) .
Hormonal alterations (31) and hypovitaminosis (32) have been identified as important disorders associated with the syndrome.
Vitamin D can be highlighted for its role in the musculoskeletal health of older adults and its consequent relationship with the sarcopenic process.
A prospective international study with 727 older adults aged 65 years and over in the Augsburg region of Germany found that participants with low vitamin D levels had significantly higher odds of developing the syndrome (OR=2.53) when compared to those with normal levels (32) . In this sense, orientation and encouragement regarding exposure to the sun, intake of food rich in vitamin D and practice of physical exercises is considered a nursing role.
For gerontological nursing, the elaboration of a physical frailty prediction model contributes to a greater objectivity in the screening of the oldest old (33) . This is

Conclusion
The present study proposed a Physical Frailty Prediction Model for the oldest old according to clinical variables, which included "metabolic disease", "dyslipidemia" and "hospitalization in the last 12 months". In the univariate analysis of the data, the clinical variables "hospitalization in the last 12 months" and "antidiabetics" were associated with the development of the physical frailty syndrome.
Regarding the management of physical frailty in primary care, the nurse must provide an assistance