Data were collected online using web-based surveys. Participants could choose not to answer any item, resulting in missing values for some questions. Sum scores or mean values were not calculated for any scales that had missing items for that participant.

Participants provided basic sociodemographic information, including gender (male or female) and age. They were then sorted into 4 age groups (≤39 years, 40-49 years, 50-59 years, and ≥60 years) for further analyses.

The participants were asked to indicate the type of school they currently teach at. To ensure adequate statistical power given the federal state-specific nature of the German educational system, these responses were dichotomized into 2 groups for analysis: primary school teachers and those teaching at all other school types.

In addition, participants were asked to indicate for how long they have been working as a teacher since finishing their professional education (German: Referendariat): 1-9 years, 10-19 years, or ≥20 years. Lastly, they reported the number of lessons they teach in an average week, that is, their teaching load (German: Deputat), with the following response options: <15 lessons, 15-19 lessons, 20-24 lessons, 25-27 lessons, and ≥28 lessons.

The German version of the PERMA Profiler questionnaire was used to assess overall well-being and health [37,38]. The PERMA Profiler instrument assesses well-being according to the PERMA model as a multidimensional construct based on 5 core components (Positive Emotions, Engagement, Relationships, Meaning, and Accomplishments). The questionnaire consists of 23 items. There are 3 items for each of the 5 PERMA components. Two additional 3-item scales refer to Negative Emotions and Health. Moreover, there is 1 item referring to Happiness and 1 item referring to Loneliness. The 8 additional items beyond the core concept of PERMA well-being act as filler items to disrupt response tendencies and provide additional information to provide a more comprehensive insight into the participants’ mental health. Of the 23 items, 10 items are rated on 11-point Likert scale ranging from “never” (0) to “always” (10), 11 items are rated on a 11-point Likert scale ranging from “not at all” (0) to “completely” (10), and 2 items are rated on an 11-point Likert scale ranging from “terrible” (0) to “excellent” (10). For this investigation, an overall score of PERMA well-being is calculated as the mean of all 15 PERMA core items (ie, all items referring to Positive Emotions, Engagement, Relationships, Meaning, and Accomplishments). In addition, the mean value for the health subscale was calculated as the average value across the 3 health items. Higher values in both the well-being score and the health score denote higher expression of the construct. As the PERMA Profiler was primarily developed to assess well-being [38], not health, we added a brief analysis of convergent validity for the health scale. Specifically, we computed the Pearson correlation between the mean PERMA health scale value and an additional single item assessing physical health (“Overall, how would you rate your physical health at present?”). Responses of the single item were recorded on a scale ranging from “very bad” (0) to “very good” (10).

The German translation of the 16-item version of the European Health Literacy Survey Questionnaire (HLS-EU-Q16 [39-41]) was used to assess levels of general health literacy of the participants. This questionnaire asks the participants to indicate their subjective difficulty in accessing, understanding, appraising, and applying health information concerning health care, disease prevention, and health promotion. Each of the 16 items is answered on a 4-point Likert scale (“very easy,” “fairly easy,” “fairly difficult,” and “very difficult”). To obtain an overall score, the individual item responses are dichotomized in a first step (assigning a value of 1 to “fairly easy” and “very easy” responses and a value of 0 to “fairly difficult” and “very difficult” responses). Then, a sum score is calculated by adding up these values across the dichotomized items. The resulting sum score has a range of 0-16, where higher values refer to higher levels of general health literacy. Cutoff scores were established to classify individuals according to their individual sum scores into the following groups: inadequate and problematic health literacy (scores 0-12), and adequate health literacy (scores 13-16 [42]).

An adapted form of the German version of the Mental Health Literacy Tool for the Workplace (MHL-W-G [43,44]) questionnaire was used to assess mental health literacy. The instructions were slightly modified by adding the notion that participants should imagine they were currently working, if that was not actually the case. However, as all participants in this sample were currently working as teachers, this addition was not relevant to them. The adaptation was only made to ensure that results were comparable to samples that also include unemployed participants [34]. The questionnaire consists of 4 sets of 4 questions. Each set of questions is presented alongside a case vignette. In each vignette, a short portrayal of a coworker who experiences mental health issues is given, including a description of the person’s current mental state, symptoms, and life circumstances. Two case vignettes refer to female coworkers, and 2 case vignettes refer to male coworkers. In addition, the vignettes differ in the nature of the mental health problem without explicitly labeling it. For each vignette, the participants answer the same set of 4 items. These refer to the ability to recognize specific mental disorders, knowledge and beliefs about risk factors and prevention, knowledge and attitudes to facilitate help-seeking, and knowledge and beliefs about mental health interventions. Specifically, the participants were asked to rate their knowledge about what might be happening, how they could prevent the situation from worsening, what to say or do in the situation, and which resources and services might be helpful on a 5-point Likert scale with the poles “disagree” (1) and “agree” (5) in the German translation. The ratings were combined into a sum score across all items and vignettes. This sum score ranges from 16 to 80, with higher values indicating higher mental health literacy. So far, no cutoffs to categorize individuals with low and high levels of mental health literacy according to this questionnaire have been established, which is why the sum score is analyzed as a continuous variable.

Detailed descriptive statistics for all groups described below can be found in Table 4. The PERMA well-being score correlated positively with the PERMA health subscale (r=0.54, 95% CI 0.49-0.58; P<.001).

Women did not significantly differ from men concerning their overall well-being (t₁₀₀₀=–0.70; P=.48; d=–0.05) or health according to the PERMA model (t₁₀₀₃=–0.80; P=.42; d=–0.06).

There were significant differences between the 4 age groups (≤39 years, 40-49 years, 50-59 years, and ≥60 years) regarding overall well-being (F_{3, 998}=4.25; P=.005; η²=0.013). On a descriptive level, well-being was lowest among the 40- to 49-year-olds, followed by the 50- to 59-year-olds, the ≤39-year-olds, and the ≥60-year-olds. Post hoc Bonferroni-corrected t tests (corrected α=.008) revealed significant differences between the youngest age group and the 40- to 49-year-olds (t_510.91=3.38; P<.001; d=0.28), but not between the youngest age group and the 50- to 59-year-olds (t₆₂₅=1.05; P=.30; d=0.09) and the ≥60-year-olds (t₄₇₆=–0.05; P=.96; d=–0.01), respectively. With the corrected α-thresholds, the 40- to 49-year-olds did not significantly differ from their peers in the 2 older age groups: 50- to 59-year-olds (t₅₂₂=–2.05; P=.04; d=–0.18) and the ≥60-year-olds (t₃₇₃=–2.31; P=.02; d=–0.26), respectively.

There were no significant differences regarding health when comparing the 4 age groups (F_{3, 1001}=1.84; P=.14; η²=0.005).

There were no significant differences in well-being between primary school teachers and teachers who work at other types of schools (t₁₀₀₀=–1.22; P=.22; d=–0.08), but primary school teachers reported significantly lower health (according to the PERMA model) than teachers who work at other types of schools (t_516.55=–2.27; P=.02; d=–0.16).

There were no significant differences in well-being between teachers with different teaching loads (ie, number of lessons per week), F_{4, 996}=1.79; P=.13; η²=0.007. Similarly, teaching load was not associated with health (according to the PERMA model), F_{4, 999}=0.99; P=.41; η²=0.004.

There was a significant effect of the number of years teaching professionally on well-being (F_{2, 993}=6.22; P=.002; η²=0.012). Descriptively, well-being was lowest among those teachers with 10-19 years of professional teaching when compared to those with more or less teaching experience. Post hoc Bonferroni-corrected t tests (corrected α=.017) revealed significantly higher well-being among those with 1-9 years of experience than among those with 10-19 years of experience (t₆₅₂=2.93; P=.004; d=0.23). Similarly, well-being was significantly higher among those with 20 or more years of experience than among those with 10-19 years of experience (t₆₇₈=–3.12; P=.002; d=–0.24). There were no significant differences between those with 1-9 years of experience and those with 20 or more years of experience (t₆₅₇=–0.25; P=.80; d=–0.02).

Health (according to the PERMA model) was also significantly associated with the number of years teaching professionally (F_2, 996=3.98; P=.02; η²=0.008). Descriptively, health was lowest in those teachers with 10-19 years of professional teaching when compared to those with more and less teaching experience. Post hoc pairwise t tests with a Bonferroni-corrected α=.017 partly supported this pattern of results by showing significant differences between the teachers with an intermediate level of teaching experience when compared to those with less experience: 1-9 years of experience vs 10-19 years of experience (t₆₅₄=2.68; P=.008; d=0.21). However, there were no differences when comparing those with 20 or more years of experience vs 10-19 years of experience (t₆₈₁=–2.17; P=.03; d=–0.17), and teachers with 1-9 years of experience vs teachers with 20 or more years of experience (t₆₅₈=0.54; P=.59; d=0.04).

When interpreting general health literacy as a dichotomous categorical variable, teachers with inadequate or problematic general health literacy had lower well-being (t_409.68=–7.14; P<.001; d=–0.55) and lower health (according to the PERMA model; t_405.52=–6.24; P<.001; d=–0.49) than those with adequate general health literacy.

When looking at health literacy as a continuous measure, there were positive correlations between general health literacy and well-being (r=0.26, 95% CI 0.20-0.32; P<.001), and general health literacy and health according to the PERMA model (r=0.25, 95% CI 0.20-0.31; P<.001). Similarly, mental health literacy correlated positively with well-being (r=0.20, 95% CI 0.14-0.26; P<.001) and health (r=0.10, 95% CI 0.04-0.16; P=.002), but correlation coefficients were smaller than for general health literacy. All correlation coefficients for the associations of general and mental health literacy with well-being and health were small but statistically significant.

Preliminary analyses confirmed that relevant assumptions for multiple linear regression were met. Collinearity diagnostics showed no multicollinearity issues in both models, with all variance inflation factors below standard thresholds (<5). Additionally, visual inspection of residual plots supported the assumptions of normality and homoscedasticity, and maximum Cook distances (<1) confirmed the absence of influential outliers.

In the multivariate regressions predicting PERMA well-being and health, both overall models were statistically significant. The well-being model (F_{13, 977}=10.66; P<.001) explained approximately 11.3% of the variance in the PERMA well-being score (adjusted R²=0.113). Details for the individual predictors are in Table 5. General health literacy and mental health literacy emerged as significant positive predictors of well-being. Furthermore, the analysis revealed an age- and experience-specific pattern: teachers aged 40-49 years and 50-59 years, as well as those with 10-19 years of teaching experience, reported significantly lower well-being compared to their younger or early-career counterparts. School type, teaching load, and gender did not retain statistical significance in the well-being model. The health model (F_{13, 980}=8.34; P<.001) accounted for 8.8% of the variance in the PERMA health outcome (adjusted R²=0.088). Details for the individual predictors are in Table 6. While general health literacy was a significant positive predictor, mental health literacy did not reach statistical significance for the health outcome. Health scores were significantly lower for teachers in the midcareer phase (10-19 years) and in both groups aged 50 years and older compared to their younger colleagues (≤39 years). In addition, teaching at nonprimary schools and having a high teaching load (≥28 hours) were positively associated with health.

Our findings add to the literature by showing, in recent data from Germany, that teachers’ well-being, as measured within the PERMA framework, reached moderate to high scores when interpreted relative to the range of the questionnaire. This pattern of results is in line with prior research demonstrating that teachers’ well-being is not generally at risk [10]. In this context, however, it is essential to acknowledge that no validated and widely acknowledged thresholds have been established to categorize individuals into different levels of well-being for this and similar questionnaires, making the interpretation of the absolute scores more difficult [10,37,38]. Compared to the initial validation study of the German-language PERMA-Profiler [37]—which used the same instrument and web-based administration format—our sample reported slightly higher well-being and comparable health. However, because this reference group is a cross-national convenience sample (Austria, Germany, and Switzerland) and data collection took place more than a decade ago, it provides only an approximate benchmark. To offer a more precise, population-specific context, we also cross-referenced our data with a recent study explicitly focusing on the core elements of PERMA well-being among teachers in Germany [15], with which our descriptive findings largely converge.

While no significant associations with gender could be observed in our sample, the results demonstrate that there seems to be a particularly vulnerable period regarding well-being in terms of age: middle-aged teachers reported the lowest well-being. This dip in well-being during midlife is not unique to the group of teachers in Germany, as it has also been found in general population samples from other high-income countries [49-51]. One possible explanation for this finding is that this period is characterized by heightened sensitivity to the negative psychological impact of unrealized life expectations and aspirations [52], which might also be relevant to teachers. In addition, midlife entails unique challenges such as dual caretaking for older parents and adolescent children, being confronted with the onset of physical and cognitive aging, and the pressure to maintain economic stability [53].

When looking at the associations of well-being and health in terms of professional factors, in our sample, years of teaching experience emerged as an important factor for well-being: teachers with an intermediate level of teaching experience (ie, 10-19 years) reported the lowest well-being. A similar pattern was observed for the health outcome, though not fully supported by the corrected pairwise comparisons. Importantly, for the interpretation of these findings, it has to be noted that work experience, measured as the number of years of teaching, is typically correlated with age. Consequently, the underlying mechanisms driving these findings may partly overlap with those associated with age. However, in the multivariate analyses for both outcomes, (intermediate) experience and (middle) age were significant predictors in the absence of multicollinearity (all variance inflation factors <5), which supports the unique contribution of both factors.

Taken together, these findings regarding midlife and midcareer vulnerabilities highlight that teachers in these cohorts might be a priority target group for workplace health and well-being interventions.

Interestingly, workload, operationalized as the teaching load per week, was not significantly associated with either well-being or health in the bivariate analyses. In the multiple regression analysis for the PERMA health outcome, a higher teaching load unexpectedly emerged as a positive predictor of health when controlling for other factors. This is in contrast to prior research that shows associations between working conditions, such as workload and well-being or job satisfaction, which was not mirrored in our findings on the association of well-being and workload [54,55]. This discrepancy might be explained by the fact that the number of lessons taught per week is only an approximation of the actual workload. For example, teaching some subjects might require more work (eg, preparation and grading) outside of the classroom than others. Furthermore, teachers with a lower teaching load may carry significant responsibilities outside of traditional employment, such as caregiving, affecting their well-being and health status. Finally, well-being may be influenced less by workload per se and more by individual psychological and behavioral responses to it. For instance, factors such as effective coping strategies, self-efficacy, and resilience might lower the subjective workload burden for some individuals [11].

Primary school teachers reported significantly lower health (according to the PERMA model) than teachers from other types of schools, both in the bivariate analysis and when all other factors were controlled for in the multiple regression. To date, health-related differences between primary school and other teachers have been observed inconsistently [3,56-58]. A possible explanation for the lower health status in primary school teachers might be that they are confronted with distinct job demands that affect their physical health without necessarily impacting their general well-being, such as voice problems [58].

On a descriptive level, health literacy levels of teachers were higher than in the general population, with fewer individuals categorized as having inadequate or problematic health literacy (26.4%) than in recent data from a general population sample (approximately 40% [59]). Similarly, the mean values for the general health literacy questionnaire were higher in teachers than in the general population [34,59]. This might be partially explained by the higher level of education in teachers [60,61]. Importantly, however, it is alarming that about 1 in 4 teachers do not have adequate health literacy, given their important function as facilitators of (health) literacy education in schools [62].

As there are no established thresholds to classify individuals into categories of mental health literacy levels with the instrument used here, we can only compare the descriptive findings to those from other publications. Relative to a general population sample from Germany [34], a similar mean value was observed among teachers in our sample, suggesting that the differences between samples observed for general health literacy do not generalize to mental health literacy. These findings underscore that teachers might benefit from mental health literacy interventions just as much as individuals from the general population. Of note, some interventions have already been developed and evaluated regarding their efficacy in improving aspects of mental health literacy in teachers with positive results [27,63]. For example, a recent randomized controlled trial showed that short educational videos are effective tools for the promotion of mental health literacy among both teachers and students [28].

Consistent with findings from other countries [64,65], we observed significant bivariate associations between lower levels of general and mental health literacy and poorer well-being and health (according to the PERMA model). However, the association between mental health literacy and health did not remain statistically significant when all other factors were controlled for in the regression analysis. Crucially, while the absolute effect sizes were small to moderate [46], they were among the largest observed in this study. Furthermore, in the multivariate analyses, general health literacy emerged as the strongest predictor (yielding the highest beta weights) for both outcomes, followed by mental health literacy for the well-being outcome. This pattern of results underscores the unique associations between health literacies and the 2 outcomes studied here that should be further investigated regarding their practical significance. For example, these findings suggest that health literacy could be a relevant focus for future interventions aimed at supporting teachers’ health and well-being.