Health behaviour, health status and occupational prospects of apprentice nurses and kindergarten teachers in Germany: a cross-sectional study

A cross-sectional survey was carried out in Hamburg, Germany. Apprentice geriatric nurses, hospital nurses and kindergarten teachers/assistants were eligible for inclusion in the study. Apprentices in all years of vocational training, of both sexes and all ages were included at this stage. We calculated an estimate of the required sample size based on the prevalence of overweight, as defined by the body mass index (BMI). Overweight is an indicator for health behaviour and a factor influencing the health status. A mean prevalence of 30 % of overweight was assumed among apprentices [12, 13]. For a 95 % chance of our sample estimate being within five percentage points of this assumed prevalence (30 ± 5 %), a sample size of n = 322 was needed. We oversampled by 20 % in order to account for non-response and missing data. We therefore aimed to recruit approximately 403 apprentices.

Sixteen out of 20 vocational schools in Hamburg identified for the target professions were contacted at random. Eight of them gave their consent to participate in the study. The schools were responsible for selecting classes by convenience in which questionnaires could be distributed to apprentices during lessons. Apprentices were informed by a project member of the aim and content of the study, of measures to ensure confidentiality and of the voluntary nature of participation. By completing the questionnaires, participants provided their informed consent to take part in the study. Ethical approval was obtained from the Hamburg Medical Council Ethics Committee (# PV4649).

Data collection took place from January to March 2014. In total, 402 apprentices participated in the study (response rate: 98.8 %). The recruitment process is shown in Fig. 1. The age of participants varied greatly from 16 to 52 years. The health behaviour of younger apprentices differs considerably from that of older ones. For example, apprentices under the age of 30 years showed hazardous alcohol consumption (χ ² = 9.9 [df = 1], p <0.01) and unfavourable dietary patterns (χ ² = 11.4 [df = 2], p <0.01) significantly more often than their older counterparts. The study sample was therefore reduced to participants aged 16 to 30 years (n = 354). We set the cut-off at the age of 30 years to ascertain its comparability with the age group of 18 to 29-year-olds applied in the DEGS1 survey [6‐8].

A self-administered questionnaire was compiled from validated scales and single items which were applied in previous studies. An overview of the scales and items used in the analyses is given in Table 1. The internal consistencies of the self-efficacy, irritation and job satisfaction scale were good (Cronbach’s α = 0.81 to 0.88) and comparable to validation studies [22‐24]. The applicability and comprehensibility of the instrument was verified in a pre-test with one class of apprentice geriatric nurses (n = 15). This class was not included in the final study sample.

Information on the socio-demographic characteristics of participants included age, sex, height, weight, migration background, socioeconomic status (SES) and year of education. Apprentices were classified according to their BMI (kg/m²) as underweight (<18.5), normal weight (18.5 to 24.9), overweight (25 to 30) or obese (≥30). Migration background was assumed if apprentices themselves had immigrated to Germany and at least one of their parents was not born in Germany or was not of German nationality, or if both parents had immigrated or were not of German nationality [25]. The SES was determined from the highest educational level of apprentices and the current parental occupation. The following scores were administered to educational levels: 1 = comprehensive school qualification, 2 = general certificate of secondary education, 3 = technical college entrance qualification, 4 = university entrance qualification and 5 = technical college entrance qualification or university entrance qualification plus university degree. Parents’ occupations were coded according to the International Standard Classification of Occupations (ISCO-88) [26]. By using the International Socio-Economic Index of Occupational Status (ISEI), values from 16 to 90 were assigned to these codes [27]. Scoring categories were defined by quintiles. Scores for educational level were added to scores for the parental occupation with the higher ISEI value. The SES was classified as low (2 to 4), middle (5 to 7) or high (8 to 10) [28].

The health behaviour of apprentices was assessed by physical activity, dietary patterns, smoking habits and alcohol consumption. Regular physical activity was defined as a planned, structured and ongoing activity for increasing physical fitness. For the analysis, activity levels were classified as “no physical activity”, “up to 2 h/week” and “regularly, at least 2 h/week” [29]. A food frequency scale, ranging from never to daily, was applied to examine the consumption of fifteen food items. Each item had an individual scoring system from 0 = deviant to 2 = optimal. Scores were added together and classified according to the tertiles of the index [30]. A distinction was made between daily, occasional, former and never smokers. Apprentices smoking ≥20 cigarettes/day were classified as heavy smokers [7]. The Alcohol Use Disorder Identification Test (AUDIT-C) was performed to detect hazardous drinking behaviour. The cut-offs were >3 for women and >4 for men [31, 32].

The health status was rated on a five-point scale ranging from “excellent” to “poor”. Questions regarding diseases they had experienced during the previous twelve months were adapted from the Work Ability Index. Response categories were predefined as “no”, “yes, own opinion” or “yes, physician’s diagnosis” [33]. Two scales evaluated mental health. The General Self-Efficacy Scale identifies resources for coping with life’s challenges. It consists of ten items on a four-point Likert scale, with higher scores indicating better self-efficacy [24]. The Irritation scale measures emotional and cognitive psychological strain in occupational contexts. A higher score on a seven-point Likert scale reflects greater irritation [22].

Job satisfaction, motivation to stay in the profession, as well as occupational stresses and wishes were considered as prospects. Job satisfaction was measured by seven items on a four-point Likert scale. Response categories were transformed to 0 = very dissatisfied, 33 = dissatisfied, 66 = satisfied and 100 = very satisfied [23, 34]. Intentions to stay in the profession for the next five years were predefined as “yes”, “no” or “don’t know”. Two open questions asked about occupational stresses and wishes for the work situation. Out of the free texts, entries categories were formed, which were analysed quantitatively.

Frequency distributions and percentages as well as means with corresponding standard deviations (SD) were calculated. Differences between the three professional groups were analysed using the chi-square test for categorical variables. Fisher’s exact test was performed if the expected cell frequencies were less than five. Mean differences were evaluated using one-way ANOVA and Scheffé post-hoc tests. If assumptions of normal distribution and homogeneity of variance were not met, non-parametric tests were used. Associations between health-related factors and binary outcome variables (MSD and mental disorder) were examined by logistic regression. Firstly, all variables with a significance value of p <0.05 from univariate analyses were included in a hierarchical model. Secondly, variables with p-value <0.25 were added [35]. If the model did not improve significantly through the second step, a final model with all the variables from the first step was calculated using the “enter” method. Odds ratios with corresponding 95 % confidence intervals and explained variances (Nagelkerke’s Pseudo R²) were calculated from logistic regression analyses. Statistical significance was set at p <0.05. All statistical analyses were performed using SPSS version 20.

The use of heterogeneous instruments to measure health behaviour and status makes a comparison of literature difficult, and also limits the generalisability of the results of this study. Because of the cross-sectional design, no conclusions on causal relationships can be drawn. In addition, results are based on self-reports of participants. When validating the food frequency index, Winkler and Döring [30] pointed out that participants tend to overestimate intake of foods with a healthy image. Furthermore, frequencies of overweight in adolescents are underestimated by studies relying on self-report [44]. For that reason, unfavourable dietary patterns as well as overweight and obesity are likely to be even more prevalent than identified by this study. It was not feasible to draw a random sample of apprentices from the participating schools. The decisions of schools on which classes could take part in the study could not be influenced, as these depended on the upcoming practical periods and exams. Since a convenience sampling strategy had to be used, the study sample might not be representative. We calculated ORs to identify factors associated with MSD and mental disorders. However, these ORs should not be interpreted as estimates of relative risks, as the outcomes were frequent and because of the temporal ambiguity due to the cross-sectional design of our study (Table 4).

Limitations due to a non-response bias can be excluded because of the high response rate of 99 %. The number of missing values was under 5 % for nearly all variables. Extensive socio-demographic information on participants was collected in order to identify possible confounders.