Journal of Occupational Health
Online ISSN : 1348-9585
Print ISSN : 1341-9145
ISSN-L : 1341-9145
Originals
A Prospective Study of Factors Associated with Risk of Turnover among Care Workers in Group Homes for Elderly Individuals with Dementia
Miwa Suzumura Yasuhiro FushikiKota KobayashiAsae OuraShigeo SuzumuraMasafumi YamashitaMitsuru Mori
Author information
JOURNAL FREE ACCESS FULL-TEXT HTML

2013 Volume 55 Issue 6 Pages 487-494

Details
Abstract

Objectives: A prospective study was conducted to assess factors associated with risk of turnover among care workers in group homes (GHs) for elderly individuals with dementia. Methods: In January, 2010, 51 out of 238 GHs in Sapporo City responded to our request for participation in a survey. During February and March of 2010, 438 out of 700 care workers (62.6%) in the GHs returned a completed questionnaire to us. They were followed up with until March of 2012, and 395 subjects (90.1%) responded to our survey, which was conducted twice. Over the course of 2 years, 91 subjects left their jobs. Cox's proportional hazards model was used for analysis, adjusting for gender and age. Results: Less provision of social support by supervisors, colleagues, family or friends was significantly associated with increased risk of turnover. Financial aids for off-the-job training was marginally significantly associated with reduced risk of turnover. Longer duration of working for frail elderly care was significantly associated with decreased risk of turnover. Habitual smoking was significantly associated with increased risk of turnover. Conclusions: Social support in the workplace may be important to reduce the risk of turnover in care workers at GHs. Further prospective studies are required in other areas in Japan.

(J Occup Health 2013; 55: 487–494)

Introduction

A higher staff turnover rate of care workers in nursing facilities for the elderly was reported (21.9%) in 20081) as compared with the rate for general workers in 20082) in Japan (13.8%). A high turnover rate for staffs caring for people with dementia was also revealed in the United Kingdom in private facilities (38%) and NHS facilities (16%)3). Moreover, the 1-year turnover rates were reported to be 56.4, 39.7 and 35.8% for certified nurse aides, licensed practical nurses, and registered nurses, respectively, in the USA4). These higher turnover rates might be caused by relatively lower wages5), increased job stress6), reduced job satisfaction7) and less provision of social support6). Lack of employee continuity may contribute to recurring increased staff training costs and decreased quality of care provided to care receivers6). The significance of understanding turnover in the aged care setting is especially important when considering an aging population such as in Japan, in which, 31.6 and 38.8% of the Japanese population will be older than 65 in 2030 and 2055, respectively8).

Group homes for elderly individuals with dementia, abbreviated as GHs in this article, have spread widely throughout Japan since the 1990s9), and now, GHs are considered a type of community service, not a type of institutional service, in the National Care Insurance System in Japan. Because one unit in a GH consists of 5 to 9 inhabitants, and the allowed maximum number of units in a GH is 2, only a small number of care workers work in a unit in a GH. Accordingly, care workers at GHs experience carry problems such as increased stress or anxiety. Furthermore, the environment in a GH sometimes tends to be too close, causing problems between inhabitants and care workers. In contrast to the stressful work environment in the GH in Japan, according to te Boekhorst et al.10),job satisfaction was higher and burnout was lower in professional caregivers at GHs than in traditional nursing homes in the Netherlands.

Few prospective studies3) have been conducted to assess risk factors associated with staff turnover in care workers, to our knowledge. Consequently, we conducted a prospective study of care workers at GHs to assess factors associated with risk of their turnover.

Subjects and Methods

The Ethics Committee at Sapporo Medical University approved this study in 2009. In January, 2010, 51 out of 238 GHs (21.4%) in Sapporo City responded to our request for participation in a survey. During February and March of 2010, 438 out of 700 care workers (62.6%) returned, by post, a completed questionnaire of a baseline survey directly to the Department of Public Health, Sapporo Medical University School of Medicine. The questionnaire included such things as demographic characteristics, experience in care giving, duration of working as a care staff number, average number of days and nights working in a month, wage, social support10) from supervisors, colleagues, family or friends, depressive status according to the Center of Epidemiological Studies-Depression Scale (CES-D)12) and Ozeki's coping style for stress13).

We tried to conduct follow-up surveys for turnover twice, in March of 2011 and March of 2012, for the 438 study subjects. If a subject quit a job, his or her exact date and reason for quitting were asked. We sent the questionnaires for the follow-up surveys to the study subjects at their home addresses by post. Thereafter, 395 subjects (90.1%) returned, by post, the completed questionnaires for the follow-up surveys directly to the Department. Data for the 3 surveys were connected based on the subjects’ personal information including their full name. Eventually, 91 subjects left their jobs during the 2-year period.

The number of observational days was calculated as the date of turnover minus the date of the baseline survey for workers that left their jobs or as the date of the last follow-up survey minus the date of the baseline survey for workers continuing in a job. The turnover rate per person-year (%) was calculated as the number of turnovers multiplied by both of 365.25 and 100 divided by the number of observational days. Cox's proportional hazards model14) was used to find the association of potential factors with risk of turnover, adjusting for age and sex. SPSS 16.0 (Japanese version) was used for every analysis. The significance level was set at the 5% level.

Results

Table 1 shows hazard ratios (HRs) and their 95% confidence intervals (95%CIs) for turnover in the study subjects according to basic characteristics. As shown in Table 1, the turnover rate for all study subjects was 14.6% per person-year, and that of the female subjects (13.8% per person-year) was lower than that of the male subjects (19.7% per person-year), although the difference was not significant. Furthermore, older subjects had a lower risk of turnover than younger subjects, although the difference was also not significant. Analyses adjusted for age and sex revealed that, having a spouse was not associated with risk of turnover and that experience caring for a frail elderly family member was also not associated with risk of turnover.

Table 1. Age- and sex-adjusted hazard ratio (HR) and its 95% confidence interval (95%CI) for turnover in workers at group homes for elderly individuals with dementia according to basic characteristics
Variables Contents Number of subjects Observational days Number of turnover subjects Turnover rate per person-year (%) HR 95%CI p value
Sex Male 57 31,467 17 19.7 1.00
Female 338 196,410 74 13.8 0.78 0.46–1.34 0.367
Total 395 227,877 91 14.6
Age 20–29 71 39,190 21 19.6 1.00
30–39 82 46,298 19 15.0 0.76 0.41–1.42 0.395
40–49 104 60,167 26 15.8 0.85 0.48–1.51 0.577
50–59 97 57,760 17 10.8 0.58 0.31–1.11 0.101
60< 41 24,462 8 11.9 0.65 0.29–1.48 0.309
Total 395 227,877 91 14.6 p for trend=0.075
Having spouse Yes 181 105,324 39 13.5 1.00
No 214 122,553 52 15.5 0.93 0.69–1.69 0.748
Total 395 227,877 91 14.6
Experience of Yes 123 72,694 22 11.0 1.00
care for a fail No 272 155,183 69 16.2 1.29 0.78–2.13 0.329
family member Total 395 227,877 91 14.6

Table 2 shows age-and sex-adjusted HRs and their 95%CIs for turnover in the study subjects according to working status. Risk of turnover was slightly higher in the subjects with a duration of working in frail elderly care of between 3 and 5 years than in those with a duration shorter than 3 years. However, risk of turnover was significantly lower in the subjects with a duration of working in frail elder care of 10 years or more (p for trend=0.029).

Wage was not associated with the risk of turnover. Moreover, since wage expectation was thought to be different between full-time workers and parttime workers, full-time workers and part-time workers were separately analyzed in terms of the association between wage and the risk of turnover. The results showed that wage was not associated with the risk of turnover either in full-time workers or part-time workers (data not shown). None of the subjects left their job because they had reached the end of their contract.

Other variables such as being proud of one's job, duration of working at present GH, job status, average working days per month, average working nights per month, or being concerned about persons in the group home when not at work were not associated with risk of turnover.

Table 3 shows age-and sex-adjusted HRs and their 95%CIs for turnover in the study subjects according to support from others. Less provision of support by a supervisor when consulted by a subject was significantly associated with an increased risk of turnover (p for trend<0.001). Similarly, less provision of support by colleagues when consulted by a subject was significantly associated with an increased risk of turnover (p for trend, p=0.006). Furthermore, less provision of support by family or friends when consulted by a subject was also significantly associated with an increased risk of turnover(p for trend, p=0.010).

Table 4 shows age-and sex-adjusted HRs and their 95%CIs for turnover in the study subjects according to off-the-job training for care working. The results showed that financial aid for off-the-job training was marginally significantly associated with reduced risk of turnover (HR=0.64, 95%CI 0.40, 1.01, p=0.05). However, presence of primary or midcourse off-the-job training in GHs was not associated with risk of turnover.

Table 5 shows age-and sex-adjusted HRs and their 95%CIs for turnover in the study according to the CES-D score and Ozeki's coping style. The results showed that the CES-D score, problem-focused coping, emotional-focused coping and avoidance coping were not associated with risk of turnover.

Table 6 shows age-and sex-adjusted HRs and their 95%CIs for turnover in the study subjects according to lifestyle habits. Habitual smoking was significantly associated with increased risk of turnover (p for trend, p=0.006). However, intention to continue present job, satisfaction of sleeping, usual effect of drinking and leisure time activities were not associated with risk of turnover.

Discussion

We found that less provision of support by a supervisor, colleagues, family or friends when consulted was strongly associated with risk of turnover in care workers in GHs, and these are the first findings obtained from a longitudinal study in Japan to our knowledge. Brannonet al.15) showed that a care worker's assessment of the quality of supervision they received was inversely associated with the odds ratio for intent to quit a job. They15) suggested that understanding of the caregiving role by a supervisor and colleague encouraged care workers in their respective jobs and their intention to retain their job. Karantzas et al.6) also suggested that supervisor support demonstrated numerous indirect inverse associations with job quitting intentions. Onoderaet al.16) showed that factor analysis of stressors in care workers for the elderly revealed four large factor structures including conflict with supervisor and conflict with colleagues. Soet al.17) also indicated with multiple linear regression analysis that competence and growth of care workers in special nursing homes was significantly positively associated with support from a supervisor or colleague. These findings might suggest that personal relationships with a supervisor or colleagues strongly affect risk of turnover in care workers. Anderson et al.18) also explained that adequate staffing and longer nursing tenure of the nursing director were important predictors of reduced turnover risk.

Although we anticipated that there would be a positive relationship between stressful work environments and the turnover rate, we could not detect this relationship. Karantzaset al.6) showed that stressors directly influenced intentions to quit work based on

Table 2. Age- and sex-adjusted hazard ratio (HR) and its 95% confidence interval (95%CI) for turnover in workers at group homes for elderly individuals with dementia according to working status
Variables Contents Number of subjects Observational days Number of turnover subjects Turnover rate per person-year (%) HR 95%CI p value
Total duration of working in frail elderly care <3 years 92 52,079 23 16.1 1.00
3–5 years 101 53,517 31 21.1 1.47 0.83–2.60 0.187
6–9 years 116 68,927 27 14.3 1.06 0.59–1.92 0.846
>10 years 86 53,354 10 6.8 0.56 0.25–1.22 0.141
Total 395 227,877 91 14.6 p for trend=0.029
Duration of working at present group home <1 year 59 31,917 14 16.0 1.00
1–3 years 177 101,020 42 15.2 0.97 0.53–1.79 0.931
4–7 years 142 83,472 34 14.9 0.95 0.51–1.77 0.870
>8 years 17 11,468 1 3.20 0.23 0.03–1.75 0.155
Total 395 227,877 91 14.6 p for trend=0.510
Job status Full-time, regular work 280 160,934 68 15.4 1.00
Full-time, contractual work 67 39,096 14 13.1 1.04 0.57–1.90 0.903
Part-time, regular work 33 20,343 3 5.4 0.42 0.13–1.34 0.141
Part-tme, contractual work 5 2,274 3 48.2 2.94 0.91–9.54 0.073
Others 9 4,529 3 20.9 1.98 0.61–6.39 0.254
Total 394# 227,176 91 14.6
Average working days per month <15 days 52 31,715 7 8.10 1.00
16–21 days 138 77,714 36 16.9 1.78 0.78–4.05 0.172
>22 days 205 118,448 48 14.8 1.53 0.68–3.45 0.303
Total 395 227,877 91 14.6 p for trend=0.241
Average working nights per month None 77 46,497 12 9.4 1.00
1–4 times 120 70,146 28 14.6 1.54 0.76–3.10 0.230
5–9 times 193 108,050 50 16.9 1.58 0.83–2.99 0.163
>10 times 5 3,184 1 11.5 1.08 0.14–8.36 0.940
Total 395 227,877 91 14.6 p for trend=0.265
Wages per month <100,000 yen 52 29,894 10 12.2 1.00
100,00–149,999 yen 163 92,374 40 15.8 1.12 0.55–2.28 0.755
150,000–199,999 yen 138 79,389 34 15.6 0.98 0.47–2.06 0.957
200,000–249,999 yen 29 17,446 6 12.6 0.80 0.28–2.28 0.681
>250,000 yen 11 7,516 1 4.9 0.35 0.04–2.78 0.322
Total 393# 226,619 91 14.7 p for trend=0.305
Being concerned for persons in the group home when not at work Always 91 50,444 25 18.1 1.00
Sometimes 248 148,482 46 11.3 0.57 0.35–0.94 0.027
Scarcely 46 24,440 15 22.4 1.15 0.60–2.18 0.677
None 10 4,511 5 40.5 2.13 0.81–5.60 0.126
Total 395 227,877 91 14.6 p for trend=0.347
Being proud of one's job Yes 306 180,177 65 13.2 1.00
Scarcely 78 42,156 22 19.0 1.32 0.81–2.16 0.269
No 11 5,544 4 26.3 2.00 0.73–5.49 0.178
Total 395 227,877 91 14.6 p for trend=0.107
#  The number is lower due to missing values.

their survey measuring four aspects of stress, including role ambiguity, role conflict, work overload, and work-family conflict. Therefore, it is necessary for us to use more refined variables or scales to identify stressful work environments in a future study.

Table 3. Age- and sex-adjusted hazard ratio (HR) and its 95% confidence interval (95%CI) for turnover in workers at group homes for elderly individuals with dementia according to support by others when consulted
Variables Contents Number of subjects Observational days Number of turnover subjects Turnover rate per person-year (%) HR 95%CI p value
Support by Yes 184 111,260 33 10.8 1.00
supervisors So-so 133 76,670 30 14.3 1.30 0.79–2.13 0.300
when consulted Little 55 30,551 16 19.1 1.82 1.00–3.31 0.049
No 22 8,692 12 50.4 4.89 2.51–9.52 <0.001
Total 394# 227,173 91 14.6 p for trend<0.001
Support by Yes 237 140,451 48 12.5 1.00
colleagues when So-so 139 78,607 35 16.3 1.26 0.81–1.96 0.297
consulted Little 11 5,270 4 27.7 2.62 0.93–7.38 0.069
No 7 2,845 4 51.3 4.18 1.50–11.69 0.006
Total 394# 227,173 91 14.6 p for trend=0.006
Support by Yes 242 143,321 47 12.0 1.00
family or friends So-so 126 71,273 34 17.4 1.46 0.94–2.27 0.097
when consulted Little 21 10,442 8 28.0 2.33 1.10–4.96 0.028
No 6 2,841 2 25.7 2.24 0.54–9.23 0.266
Total 395 227,877 91 14.6 p for trend=0.010
#  The number is lower due to missing values.

Table 4. Age- and sex-adjusted hazard ratio (HR) and its 95% confidence interval (95%CI) for turnover in workers at group homes for elderly individuals with dementia according to off-the-job training for care work
Variables Contents Number of subjects Observational days Number of turnover subjects Turnover rate per person-year (%) HR 95%CI p value
Primary off-the-job training for care working Present 216 128,275 44 12.5 1.00
Absent 178 98,898 47 17.4 1.40 0.92–2.11 0.113
Total 394# 227,173 88# 13.7
Midcourse off-thejob training for care working Present 323 189,081 72 13.9 1.00
Absent 71 38,088 19 18.2 1.49 0.89–2.50 0.130
Total 394# 227,169 91 14.6
Financial aid for offthe-job training No 90 48,370 26 19.6 1.00
Yes 302 177,416 65 13.4 0.64 0.40–1.01 0.055
Total 392# 214,786 91 15.5
#  The number is lower due to missing values.

Longer duration of working in frail elderly care was not associated with reduced risk of turnover in our study. Soet al.17) reported that length of time in care work was significantly related to preparation for problem solving. Another possible explanation of this finding may be related to the fact that workers with maladjustment to the work environment in the GH are apt to leave there at an earlier time.

Financial aid for off-the-job training was marginally significantly associated with reduced risk of turnover in our study. Financial support for off-the-job train ing may help care workers to improve their ability to deal with care receivers, because Nitta et al.19) as well as Miura and Kato20) showed a significantly good effect of training in terms of care for the elderly with dementia. However, Banaszak-Holl and Hines21) indicated that total hours of training were not related to turnover rate of nursing aides in nursing homes.

Table 5. Age- and sex-adjusted hazard ratio (HR) and its 95% confidence interval (95%CI) for turnover in workers at group homes for elderly individuals with dementia according to CES-D score and Ozeki's coping style
Variables Contents Number of subjects Observational days Number of turnover subjects Turnover rate per person-year (%) HR 95%CI p value
CES-D score 16< 358 207,417 81 14.3 1.00
16≥ 36 19,868 10 18.4 1.21 0.62–2.34 0.581
Total 394# 227,285 91 14.6
Problem-focused 0–4 87 49,182 21 15.6 1.00
coping 5–7 133 74,756 36 17.6 1.09 0.64–1.87 0.749
8–15 167 98,444 34 12.6 0.78 0.45–1.35 0.379
Total 387# 222,382 91 14.8 p for trend=0.101
Emotional-focused 0–3 82 47,116 20 15.5 1.00
coping 4–6 185 107,028 45 15.3 1.04 0.61–1.76 0.892
7–9 121 68,911 26 13.8 0.97 0.54–1.74 0.907
Total 388# 223,055 91 14.9 p for trend=0.639
Avoidance coping 0–6 101 60,719 20 12.0 1.00
7–12 224 127,134 57 16.4 1.43 0.86–2.38 0.172
13–18 62 34,529 14 14.8 1.39 0.70–2.77 0.354
Total 387# 222,382 91 14.9 p for trend=0.460
#  The number is lower due to missing values.

Wage was not associated with the risk of turnover either in full-time workers or part-time workers. Although we compared reasons for turnover between full-time workers and part-time workers, the proportions of reasons such as worsened relationship with inhabitants or co-workers, low wage, anxiety for future life and health problem, were not significantly different between the 2 groups. According to the result of the Danish Nurse Cohort Study by Friis et al.22), nurses who had relatively low gross incomes showed an increased probability of early retirement. Hanaoka5) also reported that lower wages had a significant effect on increased risk of turnover, especially, during a year after employment.

We found that being proud of one's job was not associated with risk of turnover. Several previous reports indicated that job satisfaction was associated with lower risk of turnover6, 23, 24). Ogisoet al.23) reported that higher professional pride was related to lower risk of turnover. Karantzaset al.6) suggested that lower job satisfaction directly influenced intention to quit a job among care staff in the aged care sector. Castleet al.24) also indicated that high overall job satisfaction was associated with lower risk of turnover in nursing aides in nursing homes.

Smoking habits were significantly associated with risk of turnover in our study. Although we could not find any article indicating a relationship between smoking habits and turnover in care workers, it is thought that stress may be associated with both habitual smoking and staff turnover. However, relationships of smoking habits with the CES-D score or coping style were not statistically significant in our study (data were not shown). In addition, it is possible that workers with smoking habits would be subjects that might quit their job because of a smok-ing-related health problem.

The yearly turnover rate of our study subjects (14.6%) was shown to be a little higher, but not by much, than that of general workers in Japan (13.5%)2). Selection bias may exist in our study and might have resulted in the relatively lower turnover rate in GH workers, because only 21.4% of GHs and 62.6% of care workers at them participated in the study. Furthermore, because the work environment of care workers in Sapporo City may be different from other areas in Japan, our findings should not be generalized to the entirety of Japan. In particular, because the lowest legal wage in Sapporo City is lower than that in other areas of Japan, it is possibly difficult to find a significant association between lower wages and an increased risk of turnover in Sapporo City compared with in other areas of Japan.

Table 6. Age- and sex-adjusted hazard ratio (HR) and its 95% confidence interval (95%CI) for turnover in workers at group homes for elderly individuals with dementia according to lifestyle habits
Variables Contents Number of subjects Observational days Number of turnover subjects Turnover rate per person-year (%) HR 95%CI p value
Intention to continue present job Strongly want to continue 128 74,773 29 14.2 1.00
Slightly want to continue 201 118,921 40 12.3 0.92 0.57–1.50 0.748
Slightly want to quit 54 28,054 19 24.7 1.67 0.93–2.98 0.087
Strongly want to quit 10 4,858 3 17.9 1.56 0.47–5.12 0.466
Total 393# 226,606 91 14.7 p for trend=0.542
Satisfaction of sleeping Very satisfied 115 65,826 28 15.5 1.00
Slightly satisfied 159 91,629 37 14.7 0.89 0.54–1.46 0.652
Slightly unsatisfied 93 54,691 19 12.7 0.77 0.43–1.39 0.390
Very unsatisfied 28 15,731 7 16.2 1.06 0.46–2.44 0.899
Total 395 227,877 91 14.6 p for trend=0.646
Smoking Never 171 102,352 30 10.7 1.00
Quitting 82 48,879 14 10.5 1.01 0.53–1.91 0.989
Sometimes 21 10,501 8 27.8 2.39 1.09–5.25 0.030
Daily 121 66,145 39 21.5 1.84 1.14–2.98 0.013
Total 395 227,877 91 14.6 p for trend=0.006
Usual effect of drinking Little effect 153 88,289 36 14.5 1.00
Slightly drunk 173 101,373 34 12.3 0.71 0.44–1.15 0.167
Somewhat drunk 69 38,215 21 20.1 1.21 0.70–2.09 0.498
Total 395 227,877 91 14.6 p for trend=0.665
Leisure time activities Yes 166 96,538 36 13.6 1.00
No 229 131,339 53 14.7 1.19 0.77–1.82 0.438
Total 395 227,877 89# 14.3
#  The number is lower due to missing values.

In conclusion, less provision of social support by supervisors, colleagues, family or friends was significantly associated with increased risk of turnover. Social support in the workplace may be important to reduce risk of turnover in care workers at GHs. Further prospective studies are required with a longer duration in other areas of Japan.

Acknowledgments: This study was supported with a grant from Hokkaido Kosyu Eisei Kyokai.

References
 
2013 by the Japan Society for Occupational Health
feedback
Top