In this cohort study, we adopted an integrated PPPM approach using multidimensional data to develop and validate a personalized risk model predicting 1-, 3-, 5-, and 8-year all-cause mortality for community-dwelling older adults. Nine risk factors were identified, i.e., increased age, male, alcohol status, higher daily liquor consumption, history of cancer, elevated fasting glucose, lower hemoglobin, higher heart rate, and the occurrence of heart block. The AUC and C-index show high precision of the nomogram model. The lower IBS and calibration curves suggest that the nomogram prediction model exhibits good calibration. Moreover, DCA suggests that the nomogram has a positive net benefit. Our model provides accurate predictions within diverse community-based elderly populations, aiding in the identification of individuals at high risk of mortality and facilitating early interventions for mortality prevention. This model holds significant potential for influencing healthcare and policy formulation for the elderly, fostering the development of personalized preventive and interventional strategies. Our estimations serve as valuable tools in clinical discussions, assisting in informed decision-making, with an emphasis on considering life expectancy in treatment decisions [23]. Through this research, our aim is to propel the shift of healthcare models from traditional treatments towards a more proactive and preventive approach, ultimately enhancing the quality of life for the elderly and reducing all-cause mortality.

This model holds tremendous potential in the realms of prediction, prevention, and personalized medicine for the elderly, particularly as the global elderly population continues to grow, posing challenges to healthcare resources and economic burdens. Prior research mainly focuses on exploring the progress and prognosis of the disease to reduce mortality [24, 25]. However, few studies have predicted all-cause mortality in community-dwelling older adults. For these hospitalized elderly patients who have significant symptoms, the current healthcare outcomes are considered inadequate. Compared with previous reactive medicine, PPPM is a new integrative paradigm that focuses on predicting and preventing disease before the onset of symptoms, as well as providing personalized treatment and focuses on proactive preventative measures [7]. All-cause mortality prediction, targeted prevention, and intervention before the onset of symptoms are important and could improve the quality of life and extend life expectancy. Previous all-cause mortality prediction models for community populations have been developed in Europe, but their prediction indicators are limited in demographic variables, comorbid conditions, and lifestyle behaviors [26,27,28,29]. The participants of these studies were not limited to community-dwelling older adults, and also included young and middle-aged adults, and none of these models underwent external validation [26,27,28,29]. There is especially a lack of studies to predict all-cause mortality for community-dwelling older adults in China. To solve this problem, we developed an all-cause mortality risk prediction model in community-dwelling older adults to identify individuals on the basis of individual heterogeneity following the concept of PPPM. To the best of our knowledge, this is the first study using clinical multidimensional variables to develop and validate a nomogram prediction model, with external validation, for 1-, 3-, 5-, and 8-year all-cause mortality among community-dwelling older adults. Our study fills a gap, in the framework of PPPM, for all-cause mortality prediction with multidimensional data and external validation to provide targets for proactive prevention and individual management in personalized medicine of community-dwelling older adults. This easy-to-use prediction model can easily identify elderly people at high risk of all-cause mortality, guide-targeted prevention, provide personalized interventions for high-risk individuals, and promote a paradigm shift from delayed reactive medicine to proactive medicine. Our risk-prediction model can help generate a tailored, targeted PPPM approach that benefits both individuals and healthcare systems.

Our prediction model helps identify people at high mortality risk for all-cause mortality. However, early prediction may have little effect if effective interventions are not implemented. In this direction, we identified nine risk factors from clinical multidimensional data, including three nonmodifiable (increased age, male, and history of cancer) and six modifiable (alcohol status, daily liquor consumption, elevated fasting glucose, lower hemoglobin, higher heart rate, and the occurrence of heart block) risk factors. The identified risk factors are common and easily obtained in clinical work at low cost, which makes it convenient to use in early personal prevention and interventions for high-risk individuals.

Although the three nonmodifiable risk factors cannot be used for prevention and intervention, there is an important role in prediction for the identification of people at high risk of all-cause mortality. Our study found that increased age was an independent risk factor for all-cause mortality, which is consistent with the previous models [28, 29]. Advanced age may lead to multi-system functional decline, including the immune system, and cardiovascular, metabolic, autoimmune, and neurodegenerative diseases [30], thereby increasing the risk of all-cause mortality. Healthcare professionals should prioritize assessing all-cause mortality risk in elderly patients and implement interventions, such as health assessments, chronic disease management, and promotion of healthy lifestyles, to enhance longevity and quality of life. In our study, male gender had a higher risk of all-cause mortality, which is consistent with the previous models [28, 29]. This gender difference may be attributable in part to differences in lifestyle and behavioral risk factors (e.g. inadequate diet, physical inactivity, tobacco use, and excessive alcohol use), and psychosocial and environmental exposures [31]. Identifying male gender as a risk factor for all-cause mortality helps healthcare professionals to better identify high-risk individuals. It also provides a basis for developing gender-specific interventions to reduce all-cause mortality. We also found that a history of cancer is an independent risk factor for all-cause mortality, which is consistent with the previous studies [28]. Psychosocial stress of cancer survivors may lead to dysregulated immune function and induce chronic inflammation [32]. In addition, some cancer treatments may be cardiotoxic, resulting in cardiovascular disease being the second leading cause of morbidity and mortality in cancer survivors, after recurrent malignancies [33]. Healthcare professionals should pay more attention to all-cause mortality risks for the elderly with a history of cancer. The health and psychosocial stress of cancer survivors should be regularly assessed, and interventions should be made at an early stage to improve their quality of life and extend their life expectancy.

The six modifiable risk factors are beneficial for providing targets for prediction, prevention, and intervention at early stages in high-risk individuals. Heavy alcohol consumption is associated with the risk of fall [34] and cardiovascular disease [35], potentially increasing the cost of unnecessary hospitalization. In our study, we found that high, daily liquor consumption and alcohol status were predictors of all-cause mortality. However, the relationship between alcohol consumption and all-cause mortality may be complex. Continued heavy alcohol use can cause substantial morbidity and mortality from all causes, cancer, and accidents [36]. Light and moderate alcohol consumption have been inversely associated with mortality from all causes, cardiovascular disease, chronic lower respiratory tract diseases, Alzheimer’s disease, and influenza and pneumonia [37]. Clarifying the relationship between alcohol status and higher daily liquor consumption and all-cause mortality can help healthcare professionals better advise their patients. Strict control of daily liquor consumption is a good behavior that will avoid unnecessary hospitalizations and long-term care.

Elevated fasting glucose, whether incident, persistent, or due to diabetes, confers a higher risk of mortality [38]. For patients with diabetes mellitus, a good control of glucose facilitates the reduction of short- and long-term complications, reduces the burden of treatment, and improves quality of life [39]. Our findings emphasize the potential impact of high fasting glucose levels to increase all-cause mortality in older adults. Hyperglycemic states may trigger several physiological and metabolic changes, such as inflammation, oxidative stress, and vascular dysfunction, which may contribute to the onset and progression of multiple chronic diseases [40]. Healthcare professionals should closely monitor blood glucose levels in older patients and intervene as necessary to reduce long-term complications and mortality. This may include maintaining normal blood glucose levels through an improved diet, increased physical activity, medication, or other interventions.

Low hemoglobin, an indicator of anemia, is very common in the older population and is associated with reduced quality of life and increased mortality [41]. It is also strongly associated with more frequent hospitalization and longer hospital stays [42]. Even mild anemia may substantially affect physical and cognitive capacities and quality of life [42]. In our study, a low hemoglobin level was clearly identified as an independent factor associated with all-cause mortality, which is consistent with the previous studies [43]. Low hemoglobin levels may influence the risk of all-cause mortality through mechanisms that reduce oxygen supply, increase cardiovascular burden, and impair immune function. Clinicians should pay special attention to hemoglobin levels in elderly patients and take the necessary steps to correct the low hemoglobin state, including complementary therapies (iron, vitamin B12 or folic acid supplementation, use of erythropoietic drugs) [42] or elimination of the underlying cause to correct the anemia, thereby improving oxygen supply and reducing the risk of all-cause mortality.

Higher heart rates have been reported to be associated with an increased risk of dementia and accelerated rates of cognitive decline in the general elderly population, which can affect the quality of life in older adults [44]. Our findings found that high heart rate is an important risk factor in the increased risk of all-cause mortality in older adults. Previous studies also indicated that accelerated heart rate increases the risk of sudden cardiac death, heart failure, and all-cause mortality [45]. Higher heart rates in patients with stable coronary heart disease raise the risk of complications, even when other factors are under control [46]. Keeping the heart rate of the elderly in the normal range contributes to improved quality of life and survival rates. Clinicians should closely monitor heart rate levels in older patients and take the necessary measures to control high heart rates. This may include heart rhythm medication, physical activity, and measures to improve cardiovascular health.

Our findings show that heart block is associated with a risk of all-cause mortality in older adults. In this study, the heart block includes an atrioventricular block, right bundle branch block, and left bundle branch block. An atrioventricular block is associated with heart failure, atrial fibrillation, and all-cause mortality [47]. Patients with right bundle branch block are more likely to have congestive heart failure, cardiogenic shock, hypertension, and ventricular tachyarrhythmias [48]. The left bundle branch block reduces left ventricular ejection fraction, which is an unfavorable prognostic parameter in patients with congestive heart failure [49]. Heart block may decrease the heart’s perfusion, systolic and diastolic function, and hemodynamics [50]. These lead to an increased risk of cardiac events and thus an increased risk of all-cause mortality. Clinicians should closely monitor the function of the cardiac conduction system in older patients and take the necessary measures to treat heart blocks. This may include medication, pacemaker implantation, and other cardiac therapeutic interventions.

Therefore, the application of the nine risk factors in predicting all-cause mortality among community-dwelling older adults seems to be well-founded. From the perspective of PPPM, modifiable risk factors should be monitored when developing targeted prevention and intervention measures to decrease mortality for community-dwelling older adults. We recommend increased monitoring of alcohol status, daily alcohol consumption, fasting glucose, hemoglobin, and electrocardiograms among older people in the community. Conducting person-specific preventative and treatment measures (e.g. strict control of daily liquor consumption, lowering fasting glucose, raising hemoglobin, controlling heart rate, and treatment of heart block) for high-risk elderly may help improve the quality of life and survival rates [23]. Eliminating these risk factors may reduce the frequency of hospitalization and care, reducing the length of hospital stay and saving healthcare costs. This easy-to-use predictive model may also be able to improve adherence/willingness to health care for the elderly [23]. Moreover, our findings aid in guiding physicians’ decisions on treatment and management, avoiding over- and under-treatment.

The strengths of the study need to be emphasized: Firstly, in this two-center study, we developed and externally validated a nomogram prediction model, and showed that the model has good discrimination and calibration capabilities. Secondly, this is a clinical multidimensional variable cohort study with a large sample size. The 8 years of follow-up comprised a total of 16,432 person years among 5085 participants. We screened predictors from 95 clinical multidimensional variables, including demographics, lifestyle behaviors, symptoms, medical history, family history, physical examination, laboratory examination, and ECG data. Finally, the prediction models help to predict all-cause mortality in older adults and to identify high-risk individuals. The identified risk factors help the high-risk individuals to take personalized preventive and intervention measures at an early stage. The results of this study have significant clinical value in improving the health status, quality of life, and survival of older people.

There are also several limitations in this study. Firstly, this study lacks certain social factors, for example, health domains, housing status, transportation, years of retirement, or even activities of daily living, as no relevant data was available in the questionnaire. Secondly, we also did not obtain molecular omics data, such as genomics, proteomics, and transcriptome data. Molecular omics testing is mostly used in scientific research and is rarely performed for health checkups. Increasing these social factors and molecular omics data has the potential to improve the accuracy of the model. Finally, the demographic variables of this study were self-reported, which may have resulted in subjective bias, and may cast doubt on model effectiveness. However, the questionnaire was collected by trained medical personnel, which may lower this bias. The final prediction model did not include subjective variables.