The ‘Ü45-Check’ was conducted as a prospective cross-sectional study to assess the need for prevention and rehabilitation based on preventive health examinations compared to a questionnaire survey. The preventive health examinations were offered to adults aged 45 to 59 years in the federal city of Berlin and the federal state of Brandenburg, Germany. In June 2021, the German Pension Fund (GPF) identified and categorized all individuals insured by them, falling within the age range of 45 to 59 years, and residing in either Berlin or Brandenburg. These individuals were systematically enrolled in the GPF registers and classified into one of three risk groups based on their potential for work disability, as determined by the RI-DP (low risk, moderate risk, and high risk). Participants were randomly selected from each of the three groups and invited by the GPF to participate in the ‘Ü45-Check’ in the following four consecutive calendar years (2021 to 2024). Out of 500,000 people in the register, a total of 9,800 people were invited to the preventive health examinations ‘Ü45-Check’, out of whom 1,241 were interested in participating. A total of 1,040 subjects were finally included in the study (Figure 1). The study design has already been published (17).

The ‘Ü45-Check’ was a preventive health examination performed in the outpatient clinic of the Department of Sports Medicine, Charité - Universitätsmedizin Berlin/Humboldt-Universität zu Berlin, Germany. The check-up included a clinical examination and a questionnaire survey. Participants were eligible if they were between 45 and 59 years old, lived in Berlin or Brandenburg, Germany, and were insured with the GPF. Because of the predefined exclusion criteria, individuals with insufficient skills in German or English language, individuals who were already in early retirement, and individuals who have recently had rehabilitation treatment were excluded from the study.

Immediately after the Ü45-Check, the participant received the results of their personal health status based on the examination and discussed their results with the physician.

In our research project, the GPF collected and considered the RI-DP during the recruitment process. The RI-DP seemed not to have an impact on whether individuals chose to participate or not in the study. Neither the physicians nor the study participants were provided with any information regarding the specific index values of the subjects; only the project management team had access to this data.

The findings offered by this index can potentially improve the quality of rehabilitation programs as they can be developed with a more targeted approach. The RI-DP, which assesses the risk associated with disability pensions, can be determined by analyzing secondary data from the three years leading up to the reference year. An individual has a low risk if the RI-DP is < 50 points, a moderate risk if the RI-DP is 50 to < 60 points, and a high risk of disability pension if the RI-DP is ≥ 60 points (13, 14).

Before the health screening, individuals were asked to complete a web-based questionnaire, which takes approximately 10 min to complete. This two-page questionnaire is an assessment to identify existing prevention and rehabilitation needs. It was developed by another research group as part of the ‘Ü45-Check’ (18, 19) and can be found in the Supplementary Appendix Table S1.

The questionnaire is evaluated by a scoring system where varying values are assigned based on respondents’ answers to each question. Each of the five dimensions (work ability, mental health, functional ability, coping behavior, and sports and exercise behavior) within the questionnaire is assessed separately and weighed differently. The questionnaire had previously been validated and the three dimensions most relevant for the score (work ability, mental health, functional ability) had Cronbach’s Alpha values in the range of 0.79–0.90.

The questionnaire consists of 7 main questions comprising a total of 24 items, distributed across five dimensions as follows: work ability (3 items), mental health (4 items), functional ability (4 items), coping behavior (4 items), sports and exercise behavior (4 items). Participants responded to the questions using different response formats tailored to the content of each dimension. Most items use a Likert-type scale, offering 4 to 5 ordered response options (e.g., “not at all” to “almost every day,” or “without any problems” to “impossible”). One item in the work ability section uses a forced-choice single-response format, and another allows multiple responses to reflect the complexity of the condition. The sports and exercise behavior items use a frequency-based Likert scale ranging from “not at all” to “2 h or more.” These varied response formats were specifically designed to capture the multifaceted nature of each dimension adequately. Three domains are distinguished in the evaluation of the questionnaire: no action needed, prevention program suggested, and rehabilitation program suggested (18, 19).

The preventive health examination lasted a total of approximately 120 min and included anthropometric measures [height, weight, hip, and waist circumference, Bioelectrical Impedance Analysis (BIA)], handgrip strength, twelve-lead resting electrocardiogram (ECG), systolic and diastolic blood pressures, anamnesis, cardiovascular examinations, and blood samples. The detailed presentation of all parameters is published in the study protocol (17). The physical examination mainly focused on the cardiovascular system, the lungs, and the abdomen. Blood analyses were performed under fasting conditions.

In the subsequent consultation with the physician, a focus was paid on the relevant medical history. Medical needs and current health problems were also addressed, and advice for further treatment and evaluation was given. In addition, suggestions on regular exercise, health habits, and a healthy diet were provided.

Following the preventive health examination, physicians evaluated the individual’s health using a traffic light system. The assessment was conducted by the physicians in accordance with the dual control principle. Green indicated no additional action was needed, yellow suggested a prevention program, and red indicated a rehabilitation program. These categories represented an increasing health risk, with “green” indicating the lowest risk and “red” indicating the highest or the need for rehabilitation. The study physician evaluated the subject’s health status based on their medical history and examination findings, independent of the Ü45-Screening questionnaire results. Participants received detailed feedback on each measurement and overall health evaluation. Additional examinations were recommended if necessary, with specific guidance tailored to the individual’s risk profile.

Initially, demographic and anthropometric characteristics are described descriptively stratified by gender presenting, mean and standard deviation (sd) for continuous variables and absolute and relative frequencies for categorical variables (cf. Table 1). As none of the variables had a skewed distribution, we refrained from reporting the median and range. The same table, but stratified according to the results of the preventive health examination and the questionnaire, can be found in the supplement S2. To illustrate the differences between the results of the preventive health examination and the questionnaire, a cross table was created, showing the absolute number of overlapping and non-overlapping results and the corresponding percentage of the total of 1,039 cases (cf. Table 2). In addition, a Bowker test was applied, and Cohens’g estimated it (20, 21). Given the exploratory nature of this study, the p-values should only be interpreted descriptively.

We also asked the doctors involved in the study which measurements they particularly consider when making their decision. To analyze the association of these variables on the final decision taking all levels of the outcome into account, we estimated a multivariable cumulative ordinal regression model. As this is not a confirmatory study, the model falls within the scope of descriptive models. Multivariable models were estimated in order to capture the entire dependency structure of the variables, and univariable models were refrained from. Thereby, the variables Body-Mass-Index (BMI) [kg/m²], Waist-to-Hip-Ratio (WHR) [per 0.1 cm/cm], gender [female/male], Systolic Blood Pressure (SBP) [per 10 mmHg], Low-density Lipoprotein (LDL) cholesterol [per 10 mg/dL], HbA1c [%], hand grip strength [kg], currently smoking [yes/no], depressive moods [yes/no] and physical activity [0 h/1-2 h/3-5 h/ ≥ 6 h] were analyzed. All variables were continuous, except the binary variable gender and the categorical variable physical activity. The same variables were also used to analyze whether these variables also had an indirect influence on the results of the questionnaire in a second multivariable cumulative regression model. All continuous variables included in the model had a Pearson correlation of lower than 0.5 (Figure S3). Graphical inspections of the variables did not show any outliers. Since only 33 (3.2%) of the data was missing, a complete case analysis was done. Results are reported as odds ratios (OR) and their corresponding 95%-confidence intervals (CI). Furthermore, the predictions for differing BMI, depressive moods, and physical activity and the predictions for differing SBP, current smoking, and physical activity for both models are visualized in Figures 2, 3, respectively. For this illustration, all other variables are set to the mean value of the population if it is a continuous variable or to the most frequent category if it is a categorical variable. For a more detailed explanation of the plots, we refer to the results section.

Similarities between the RI-DP and the questionnaire were also analyzed using a cross table (Table 3).

For the analysis and figures, the software R (version: 4.3.1) was used, and for the ordinal regression, the function polr() (R-package MASS version: 7.3) was used.

Ethics approval was obtained by the “Ethics Committee of the Faculty of Culture, Social and Educational Sciences, Humboldt-Universität zu Berlin” on August 20, 2020 (reference number: HU-KSBF-EK_2020_0010). The work described was performed in accordance with the Declaration of Helsinki for experiments on humans. All subjects were informed by the study staff about the study procedure, subsequent data storage, and confidentiality and pseudonymity of the data. Written informed consent was obtained from all subjects for the study center to use the data for research analyses and publication of the data.

The questionnaire, as well as the physicians during the preventive health examination, determine the prevention or rehabilitation needs of the participants independently of each other. According to the questionnaire survey, the prevention or rehabilitation needs of the 1,039 participants are as follows: no additional action needed (green) n = 733 (70%), prevention program suggested (yellow) n = 215 (21%), rehabilitation program suggested (red) n = 91 (9%). As part of the preventive health examination, the physicians assessed the prevention or rehabilitation needs of the 1,039 participants as follows: no additional action needed (green) n = 141 (14%), prevention program suggested (yellow) n = 717 (69%), rehabilitation program suggested (red) n = 181 (17%). Table 2 displays the absolute and relative frequencies of the variables between the two assessments. The table shows that the physicians’ assessments differ from the results of the questionnaire. It is noteworthy that the results of the questionnaire and the clinical examination matched only 29% green, n = 123 (12%), yellow, n = 127 (12%), and red, n = 52 (5%). The Bowker test indicated an asymmetry and, thereby, differences between the questionnaire and preventive health examination (p-value = 2.2 * 10–16, Cohen’s g = 0.425). Due to the large sample size, the test results should be interpreted with caution, and more attention should be paid to the interpretation of the effects.

In particular, physicians often suggested a prevention program whereas, the questionnaire suggested that no additional measures were needed [n = 553 (53%)]. According to the questionnaire, the majority of the participants did not need an intervention. In contrast to this, the physicians found, based on clinical parameters, that at least a preventive action should be taken. This means that the content of the questionnaire and the self-assessment differ from the physicians’ assessment. The overlap in the rehabilitation program category suggests that both assessments consistently identify individuals who need a higher level of intervention.

Supplementary Appendix Table S2 presents a stratified descriptive analysis regarding demographic and anthropometric characteristics of the 1,039 individuals based on the results of preventive health examinations (green, n = 141; yellow, n = 717; red, n = 181) and the results from the questionnaire (green, n = 733; yellow, n = 216; red, n = 91).

The gender distribution across the groups is relatively balanced, with a slight predominance of men in all groups, particularly in the “yellow” group. Age increases slightly from the “green” to the “red” group, suggesting that older individuals may have higher rehabilitation needs. BMI, PBF, and VFA increase from the “green” to the “red” group, indicating a higher prevalence of overweight and obesity among individuals with greater rehabilitation needs. Both screening tools indicate similar trends across the groups: an increase in age, BMI, PBF, and VFA, with a decline in employment status from the “green” to the “red” group. These trends suggest that a greater need for rehabilitation is associated with older age, higher obesity rates, and unemployment.

Multivariable cumulative ordinal regression models were used to identify potential predictors from the preventive health examination and questionnaire to determine the need for prevention and rehabilitation. The parameters examined included BMI, WHR, gender, SBP, LDL cholesterol, HbA1c, handgrip strength, smoking status, depressive mood, and physical activity.

Table 4 presents the OR and their 95%-CI for all predictors. The multivariable cumulative ordinal regression models show that individual factors may influence prevention or rehabilitation needs within the traffic light system based on the preventive health examination and the questionnaire.

In summary, both the preventive health examination and the questionnaire identified several potential predictors influencing the chance of falling into a higher category. BMI, LDL cholesterol, handgrip strength, depressive mood, and certain levels of physical activity showed similar associations in both assessments. Specifically, an increase in BMI was associated with higher odds of being in a higher category in both the preventive health examination [OR: 1.135 95%-CI (1.094; 1.178)] and the questionnaire [OR: 1.012 (1.087)], although the effect size was smaller for the questionnaire. LDL cholesterol showed a decrease in odds for a higher category with increasing levels (10 mg/dL) in both assessments [preventive health examination OR: 0.937 (0.900; 0.975); questionnaire OR: 0.924 (0.885; 0.964)]. Depressive mood strongly increased the odds of being in a higher category in both assessments, with a stronger association in the questionnaire [preventive health examination OR: 2.254 (1.563; 3.254); questionnaire OR: 4.506 (3.216; 6.322)]. Increased physical activity was associated with lower odds in both assessments, although the effect sizes varied [preventive health examination OR for 3–5 h/week: 0.307 (0.204; 0.460) and ≥6 h/week: 0.361 (0.218; 0.598); questionnaire OR for 3–5 h/week: 0.541 (0.353; 0.822) and ≥6 h/week: 0.812 (0.477; 1.358)] compared to no physical activity. All results are to be understood descriptively and do not imply a causal relation.

However, there were variables that showed different results between the two assessments. Gender had contrasting ORs, with men having lower odds of being in a higher category than females in the preventive health examination [OR: 0.679 (0.406; 1.131)] but higher odds in the questionnaire [OR: 1.790 (1.057; 3.046)] than women. Current smoking status increased the odds in the preventive health examination [OR: 1.691 (1.212; 2.366)] but had a smaller association in the questionnaire [OR: 1.331 (0.947; 1.862)].

In summary, in the preventive health examination, predictors influencing the chance of falling into a higher category included BMI, WHR, SBP, LDL cholesterol, handgrip strength, current smoking status, depressive mood, and physical activity. Gender and HbA1c levels showed no association.

Additionally, predictors influencing a higher category in the questionnaire included BMI, WHR, gender, LDL cholesterol, handgrip strength, depressed moods, and a certain level of physical activity. Blood pressure, HbA1c level, and current smoking status showed no association in our models.

Figure 2 displays prediction plots illustrating the probability of individuals falling into one of three categories based on BMI, depressive moods, and physical activity: (1) no action needed, (2) prevention program suggested, and (3) rehabilitation program suggested. Recall that we are in the setting of an ordinal regression model. Hence, certain standard measures, such as marginal effects, cannot be applied in the same way as in standard logistic regression models. Therefore, we chose to use a graphical display of the predicted probabilities. This should address a similar perspective and should ease interpretation in the setting of ordinal regression modeling.

For “green” (no action needed), both graphs show higher probabilities for individuals with lower BMI, no depressive moods, and higher levels of physical activity. An example shows that a person who does not do physical activities has no depressive moods, and has a BMI of 20 would most likely end up in “yellow” according to the preventive health examination and in “green” according to the Questionnaire.

The preventive health examination offers clearer and more distinct trends (e.g., overlap of the “yellow” and “red” curves with increasing BMI and depressive moods), whereas the questionnaire exhibits greater variability and less distinct patterns. The figure indicates that the probability of falling into the “yellow” category, according to the questionnaire, is relatively constant, regardless of the level of BMI. However, it is noteworthy that depressive moods exert a greater association with the questionnaire results, followed by a notable decline in the probability of falling into the “green” category. For “red” (rehabilitation program suggested), both diagrams indicate the highest probabilities for individuals with higher BMI, depressive moods, and lower levels of physical activity.

These prediction plots highlight the varying probability of required interventions (prevention or rehabilitation needs) based on the assessed health parameter. These similarities and differences underline the complementary nature of both approaches. While preventive health examinations provide detailed health assessments, questionnaires offer a practical method for initial screening and risk stratification.

Figure 3 displays prediction plots illustrating the probability of individuals falling into one of three categories based on blood pressure (BP), smoking status, and physical activity, while all other variables remain constant.

In the context of a preventive health examination, the probability of requiring rehabilitation services is elevated among participants who engage in tobacco use and present with elevated BP. In contrast, both BP and smoking show a minimal relation to the questionnaire results. The number of hours spent engaging in physical activity also exerts only a slight association with the probability of falling into a higher category on the questionnaire. In the preventive health examination, a more pronounced trend emerges, indicating that the probability of requiring rehabilitation is highest among individuals who engage in no physical activity and have high BP compared to those who engage in other levels of physical activity.

This indicates that the questionnaire is inadequate for assessing the parameters of physical activity, smoking, and BP, given that these factors are only addressed indirectly.

The RI-DP is divided into three levels, depending on the level of risk of early retirement [low risk n = 540 (52%), moderate risk n = 307 (30%), and high risk n = 193 (19%)]. Table 3 compares the results of the questionnaire survey with those from preventive health examinations concerning the RI-DP and recommended measures for three different risk categories: low risk, moderate risk, and high risk.

The results show that n = 425 (79%) with a low RI-DP, according to the questionnaire survey, do not require any additional action compared to the results of the preventive health examinations [n = 93 (17%)]. In contrast, people with high RI-DP are more likely to need a rehabilitation program according to preventive health examinations n = 69 (36%) than according to the questionnaire survey n = 32 (17%). This indicates that a greater need for intervention was recognized in the sample during the preventive health examinations than in the questionnaire surveys, irrespective of the RI-DP.

One of the primary strengths of the study is the large sample size with a very small amount of missing data, which enhances the reliability and generalizability of the findings. With 1,040 participants, the study is well-powered to detect associations and provides a robust dataset for analysis. Additionally, the early detection of risk factors is crucial for achieving better health outcomes and potentially preventing the onset of associated diseases or secondary complications (40, 41, 63). This proactive approach may not only help reduce the financial burden on the healthcare system but also indicate that comprehensive check-ups can offer significant health benefits to individuals (42).

The specific examinations within the Ü45-Check have been carefully selected to provide meaningful results and to ensure their economic feasibility for use in various settings, such as primary care practices (17). Notable strengths include the use of standardized methods for assessing body composition and the application of established measurement techniques, such as ECG, and handgrip strength diagnostics (64). The study also highlights the importance of combining questionnaire surveys with preventive health examinations to ensure a comprehensive assessment of an individual’s health needs. This integrated approach might increase the effectiveness of health interventions by ensuring that individuals receive appropriate recommendations based on a combination of self-reported data and clinical assessments.

Further strengths of the study include the use of diverse assessment tools, which provide a comprehensive evaluation of health by capturing both objective clinical data and subjective self-reported information. The standardized protocols and measurement techniques ensure consistency and accuracy in data collection, enhancing the validity of the results. By focusing on individuals aged 45 to 59 years, a critical age group for the onset of many chronic conditions, the study allows for early intervention and timely preventive measures.

Moreover, the health status evaluations by physicians were conducted independently of the questionnaire results, reducing bias and providing an objective assessment of health needs. These assessments were further strengthened by the involvement of several specialists, particularly in internal medicine and cardiology, to ensure that assessments were made from different expert perspectives. In addition, the implementation of the “four-eye principle,” whereby a chief physician reviewed each assessment, added an extra layer of scrutiny and reliability to the assessments. Participants received detailed feedback on their health status and personalized recommendations, which could encourage proactive health management and improve individual health outcomes.

Self-selection is closely related to health consciousness, which represents a state of heightened awareness and voluntary engagement in health-promoting activities, behaviors, and lifestyle choices. A potential source of bias lies in the fact that individuals who feel unwell may be more inclined to accept the invitation to participate in the Ü45-Check. On the other hand, individuals with a higher risk profile may choose not to participate in preventive medical check-ups. The RI-DP could serve as an active strategy to encourage greater participation in these check-ups.

The study design was cross-sectional, which means that we cannot definitively determine whether participants followed the recommendations for prevention or rehabilitation. There is also the question of which tool determines the real need, which we cannot answer with our cross-sectional survey because we cannot look into the participants’ future. However, it is important to note that follow-up is both planned and requested as part of the informed consent process. Furthermore, no causal effects can be estimated within this study.