Early identification of older individuals at risk of mobility decline with machine learning.

BACKGROUND: The early identification of individuals at risk of mobility decline can improve targeted strategies of prevention. AIMS: To evaluate the predictive performance of machine learning (ML) algorithms in identifying older individuals at risk of future mobility decline. METHODS: We used data from the SABE Study (Health, Well-being and Aging Study), a representative sample of people aged 60 years and more, living in the Municipality of São Paulo, Brazil. Mobility decline was assessed 5 years after admission in the study by self-reported difficulty to walk a block, climb steps, being able to stoop, crouch and kneel, or lifting or carrying weights greater than 5 kg. Popular machine learning algorithms were trained in 70% of the sample with 10-fold cross-validation, and predictive performance metrics were obtained from applying the trained algorithms to the other 30% (test set). RESULTS: Of the 1,615 individuals, 48% developed difficulty in at least one of the four tasks, 32% in stooping, crouching and kneeling, and 30% in carrying weights. The random forest algorithm had the best predictive performance for most outcomes. The tasks that the algorithm was able to predict with better performance were crouching and kneeling (AUC-ROC: 0.81[0.76-0.85]), and lifting or carrying weights (AUC-ROC: 0.80[0.75-0.84]). Age was the most important variable for the algorithms, followed by education and back pain, according to the SHAP (SHapley Additive exPlanations) values. CONCLUSION: Applications of ML algorithms are a promising tool to identify older patients at risk of mobility decline, with the potential of improving targeted preventive programs.

Cause-specific mortality prediction in older residents of São Paulo, Brazil: a machine learning approach.

BACKGROUND: Populational ageing has been increasing in a remarkable rate in developing countries. In this scenario, preventive strategies could help to decrease the burden of higher demands for healthcare services. Machine learning algorithms have been increasingly applied for identifying priority candidates for preventive actions, presenting a better predictive performance than traditional parsimonious models. METHODS: Data were collected from the Health, Well Being and Aging (SABE) Study, a representative sample of older residents of São Paulo, Brazil. Machine learning algorithms were applied to predict death by diseases of respiratory system (DRS), diseases of circulatory system (DCS), neoplasms and other specific causes within 5 years, using socioeconomic, demographic and health features. The algorithms were trained in a random sample of 70% of subjects, and then tested in the other 30% unseen data. RESULTS: The outcome with highest predictive performance was death by DRS (AUC-ROC = 0.89), followed by the other specific causes (AUC-ROC = 0.87), DCS (AUC-ROC = 0.67) and neoplasms (AUC-ROC = 0.52). Among only the 25% of individuals with the highest predicted risk of mortality from DRS were included 100% of the actual cases. The machine learning algorithms with the highest predictive performance were light gradient boosted machine and extreme gradient boosting. CONCLUSION: The algorithms had a high predictive performance for DRS, but lower for DCS and neoplasms. Mortality prediction with machine learning can improve clinical decisions especially regarding targeted preventive measures for older individuals.

Blurred lines: racial misclassification in death certificates in Brazil.

OBJECTIVES: To analyze the agreement between self-reported race and race reported on death certificates for older (≥ 60 years) residents of São Paulo, Brazil (from 2000 to 2016) and to estimate weights to correct mortality data by race. METHODS: We used data from the Health, Well-Being and Aging Study (SABE) and from Brazil's Mortality Information System. Misclassification was identified by comparing individual self-reported race with the corresponding race on the death certificate (n = 1012). Racial agreement was analyzed by performing sensitivity and Cohen's Kappa tests. Multinomial logistic regressions were adjusted to identify characteristics associated with misclassification. Correction weights were applied to race-specific mortality rates. RESULTS: Total racial misclassification was 17.3% (13.1% corresponded to whitening, and 4.2% to blackening). Racial misclassification was higher for self-reported pardos/mixed (63.5%), followed by blacks (42.6%). Official vital statistics suggest highest elderly mortality rates for whites, but after applying correction weights, black individuals had the highest rate (45.85/1000 population), followed by pardos/mixed (42.30/1000 population) and whites (37.91/1000 population). CONCLUSIONS: Official Brazilian data on race-specific mortality rates may be severely misclassified, resulting in biased estimates of racial inequalities.

Reproductive factors and age at natural menopause: A systematic review and meta-analysis.

OBJECTIVES: To systematically review the evidence on the association between age at natural menopause (NM) and reproductive factors such as age at menarche, parity and ever use of oral contraceptives. STUDY DESIGN: A literature search was carried out in PubMed, Scielo, Scopus and LILACS databases, without restriction of publication year until July 6, 2017. We excluded clinical trials, case-control studies, case reports and studies using statistical methods other than Cox proportional hazard models to assess the factors associated with age at NM. Cross-sectional studies evaluating women aged <50 years were also excluded. Random-effects models were used to pool the estimates. We registered the systematic review in the International Prospective Register of Systematic Review (PROSPERO) in August 2018, CRD42018099105. RESULTS: We identified 30 articles to include in the meta-analysis. We found that previous ever use of oral contraceptives (OC) (HR = 0.87, CI = 0.82, 0.93), age at menarche ≥13 years (HR = 0.90, CI = 0.84, 0.96), and having at least one live birth (HR = 0.79, CI = 0.74, 0.85) were associated with a later age of NM. CONCLUSIONS: Despite differences in results between countries and study design, our findings suggest that previous use of OC, age at menarche ≥13 and having at least one live birth are associated with later menopause. The results suggest that these factors could be markers of later ovarian aging.

Age at natural menopause and mortality: A survival analysis of elderly residents of São Paulo, Brazil.

OBJECTIVE: To conduct a survival analysis according to age at natural menopause (NM) in a representative sample of elderly women from the municipality of São Paulo, Brazil. STUDY DESIGN: We analyzed data from the Health, Well-Being and Aging study (SABE), a cohort that started in 2000. Mortality data up to September 2016 were obtained by linkage from the Program for Mortality Information of São Paulo (PRO-AIM). MAIN OUTCOME MEASURES: We used Cox regression to analyze all-cause and cause-specific mortality rates for cardiovascular diseases, respiratory diseases and cancer, according to age at menopause, categorized as <40, 41-44, 45-49, 50-54 (reference) and ≥55. RESULTS: After 16 years of follow-up, there were 444 deaths, of which 199 were from cardiovascular diseases, 73 from respiratory diseases and 65 from cancer. After adjustment for socioeconomic, reproductive and lifestyle factors, having an early menopause (at age 41-44) was associated with an increased risk of all-cause mortality (HR = 1.48, 95% IC: 1.03, 2.14) relative to NM at 50-54 years. Women aged 41-44 and 45-49 at NM had twice the risk of cancer mortality of the reference group. We did not find significant associations between age at NM and cause-specific mortality for respiratory and cardiovascular diseases. CONCLUSIONS: Our findings suggest that early menopause is associated with all-cause mortality in the largest city of Latin America. In addition, earlier age at NM was associated with cancer mortality. These results suggest that age at NM may be a biomarker for mortality, irrespective of country of residence.

Overachieving Municipalities in Public Health: A Machine-learning Approach.

BACKGROUND: Identifying successful public health ideas and practices is a difficult challenge towing to the presence of complex baseline characteristics that can affect health outcomes. We propose the use of machine learning algorithms to predict life expectancy at birth, and then compare health-related characteristics of the under- and overachievers (i.e., municipalities that have a worse and better outcome than predicted, respectively). METHODS: Our outcome was life expectancy at birth for Brazilian municipalities, and we used as predictors 60 local characteristics that are not directly controlled by public health officials (e.g., socioeconomic factors). RESULTS: The highest predictive performance was achieved by an ensemble of machine learning algorithms (cross-validated mean squared error of 0.168), including a 35% gain in comparison with standard decision trees. Overachievers presented better results regarding primary health care, such as higher coverage of the massive multidisciplinary program Family Health Strategy. On the other hand, underachievers performed more cesarean deliveries and mammographies and had more life-support health equipment. CONCLUSIONS: The findings suggest that analyzing the predicted value of a health outcome may bring insights about good public health practices.

Individual and contextual characteristics of indoor and outdoor falls in older residents of São Paulo, Brazil.

PURPOSE OF THE STUDY: To analyze a representative sample of older individuals of São Paulo, Brazil, according to outdoor fallers, indoor fallers and non-fallers, and to identify biological and socioeconomic (individual and contextual) factors associated with the occurrence and place of falls. MATERIALS AND METHODS: A cross-sectional study was conducted using data (n = 1345) from the 2010 wave of the Health, Wellbeing and Aging (SABE) Study, a representative sample of older residents (60 years and older) of São Paulo, Brazil. Multinomial logistic analysis was performed to identify individual factors associated with the occurrence and place of falls, and multilevel multinomial analysis to identify contextual effects (green areas, violence, presence of slums and income inequality). RESULTS: 29% had a fall in the last 12 months, with 59% occurring in indoor spaces. Individuals who had outdoor falls were overall not statistically different from non-fallers; on the other hand, those who had the last fall indoor had worse health status. Moderate homicide rate was a factor associated with increased presence of indoor falls, compared with non-fallers. IMPLICATIONS: Our results describe the importance of falls, a common problem in active and community-dwelling older adults of São Paulo, Brazil. Transforming outdoor spaces into walk-friendly areas is essential to allow socialization and autonomy with safety. Creating strategies that take into account the most vulnerable populations, as those who live in violent areas and the oldest older adults, will be a growing challenge among developing countries.