Study and participants

Data from the fourth wave of the German lidA (leben in der Arbeit) cohort study in 2022/23 was analyzed. The lidA-study is a longitudinal research project that explores the interplay between work, age, health, and labor market participation among individuals of three birth cohorts, born 1959, 1965 or 1971. A total of 8,884 employees were interviewed (Ruiz Marcos et al. 2023). The sample is considered representative of socially insured employees from these birth cohorts (Ruiz Marcos et al. 2023). The lidA-study conducted computer-assisted personal interviews (CAPI) with the participants. Further study information can be found in previous study publications (Ruiz Marcos et al. 2023; Hasselhorn et al. 2014). For the analysis, only data of participants who were employed full-time, part-time or marginal at the time of the survey wave was included (N = 7,514). Data from sworn civil servants, self-employed individuals, and those already retired was excluded. 418 respondents were identified as employed nurses.

Variables Preferred retirement age

Participants were asked until what age (in years) they would like to continue working (free answer option).

Reasons for an early preferred retirement age

Respondents who named a preferred retirement age under 65 years were asked for reasons. The cut off was chosen because a retirement age of 65 years was the legal retirement age in Germany until 2011. A response ‘before the age of 65 years clearly’ indicates the preference for an early exit, while 65plus reflects standard norms. The participants were asked whether twelve reasons played a major, a minor or no role in their preference to leave employment early (Fig. 1). The reason options were inspired from the representative German Survey ´Transitions and Old Age Potential (TOP)´ (GESIS Leipnitz-Institut für Sozialwissenschaften 2021).

Based on earlier research findings, this study investigates the impact of the following individual and work-related factors on the preferred retirement age.

Sociodemographic and individual socioeconomic factors

Participants were born in 1959, 1965 or 1971 (birth cohorts). At the time of participation, they were about 63, 57 or 51 years old. The interviewer recorded participant’s sex (male or female) at the beginning of the assessment. Employment status was classified by the participant’s answer to their working hours specified in their employment contract (full-time/part-time). As decribed in the German Working Hours Act, a full-time job comprizes of 36 to 40 hours per week, with a legal limit of 48 hours per week (German Working Hours Act, ArbZG). Financial resources were investigated by the equivalent household net income related to median income at time of measurement (Lampert and Kroll 2009).

Professional status

The categorization is based on participants actual performed job setting and job position. The authors’ expertise in nursing science and practical experience, combined with their knowledge of the German nursing sector, guided the methodological approach. This foundation allowed for accurate interpretation of free-text responses and workplace descriptions, leading to a clear definition of professional nursing positions and their respective settings. The coding of the variables is illustrated in Online Resource 1. In Germany, the nursing sector can be divided in three job settings: acute care, longterm care and ambulant care (Breinbauer 2020; Lützerath et al. 2022). The job position distinguishes between management position, specialized nurse, registered nurse and nursing assistant. The categorization led to a total sample of 418 employed nurses.

Working conditions

Physical strain was measured by the estimation on how participants rate their current work capacity related to their physical job demands on a five-point scale, from very good (low strain) to very poor (high strain). Physical strain is part of the work ability construct (item of the work ability index, dimension 2; Hasselhorn and Freude 2007; Ebener and Hasselhorn 2019). Psychosocial work stress was examined by the effort-reward imbalance ratio (ERI-R) by Siegrist et al. (2009). Employees were asked about potential efforts (3 items related to work interruptions, quantitative demands, time pressure) and rewards (7 items, e. g. opportunities for professional development, receiving deserved appreciation by supervisor). In lidA, the ERI-R ranges from 0.25 (low stress) to 7 (high stress). Work time control was investigated by a modified version (Albrecht et al. 2016) of the work time control scale by Ala-Mursula et al. (2005). The mean scale includes how much participants were able to influence the starting and finishing times of a workday, break times and time for dealing with private matters. The sumscore ranges from 0 (low control) to 100 (high control). Regarding their working hours, participants were asked to indicate whether they work shifts or not (shift work yes/no).

Health. Self-rated general health was measured by SF-12 single item (Ware et al. 1995), asking participants to rate their general health status on a five-point scale from poor to excellent.

Motivational factors

Job involvement was measured by a mean scale from three items of the Job Diagnostic Survey (JDS, German version by Schmidt and Kleinbeck 1999) - the meaning of work focusing on one’s present work situation (impact of performed work, experience of inner contentment if work is performed effectively, experienced responsibility for personal work performance), ranging from 1 (not involved) to 7 (highly involved). Job satisfaction was measured by the rating of the SOEP (German Socio-Economic Panel) single item on a ten point scale from completely unsatisfying to completely satisfying (Fietze 2011).

Handling of missing data

Of the 418 employed nurses, nine participants had missing data for the outcome variable preferred retirement age. There were 22 participants with missing data for the variable household net income. Since these were not missing at random, a ‘not specified’ category was added to include the participants in the analyses. Another three participants had missing data for the independent variables psychosocial work stress and job satisfaction. To ensure comparability of the results, the three participants were excluded. The final sample consists the data of 406 employed nurses.

Analysis

The analysis was conducted in three steps. First, the preferred retirement age of the study sample was estimated. Second, proportions including 95 %-confidence interval (CI) of the twelve reasons for a preferred retirement age under 65 years were presented. Third, the interplay of the individual and work-related factors, as well as the job setting and job position on the preferred retirement age was tested using multiple linear regression models. Because of different response ranges, the variables, physical strain, psychosocial stress, health, job involvement and job satisfaction were z-transformed for the multiple regression analysis enabling better comparison of the effects of these metric predictors on the outcome. Bivariate regression models were previously calculated. Only variables that showed a statistically significant effect in the bivariate models were included in the multiple model. To prevent multicollinearity intercorrelations, tolerance and VIF were checked.

The analysis was performed with IBM SPSS Statistics Version 28.0. (Armonk, NY, USA: IBM Corp.).