Introduction
Most care home residents live with dementia (Livingston et al., Reference Livingston, Barber and Marston2017; Luppa et al., Reference Luppa2010) and are cared for by staff who may experience high physical and psychological workloads (Fjelltun et al., Reference Fjelltun, Henriksen, Norberg, Gilje and Normann2009; VonDras et al., Reference VonDras, Flittner, Malcore and Pouliot2009), which may lead to burnout and delivery of lower-quality care (Woodhead et al., Reference Woodhead, Northrop and Edelstein2016).
Though often conceptualized within the same framework, stress and burnout differ (Malach-Pines, Reference Malach-Pines2008). Burnout is a work-related syndrome, resulting from prolonged job stressors (Maslach et al., Reference Maslach2001). It is usually defined as comprising three components: emotional exhaustion (EE) which is a feeling of emotional depletion; depersonalization (DP), also labelled as “cynicism,” which describes the development of negative, cynical attitudes towards care recipients; and lastly, staff negative perceptions about their own professional accomplishment (PA) (Maslach and Jackson, Reference Maslach and Jackson1981). In contrast, stress is a broader description of a state of mental or emotional strain resulting from adverse or demanding circumstances, which does not necessarily lead to staff burnout (Pines, Reference Pines2002). High levels of emotional exhaustion and stress are linked, while depersonalization may protect from stress, possibly as it leads to staff being less concerned about residents (McManus et al., Reference McManus, Winder and Gordon2002).
In the 2008 House of Commons report on improving dementia services in England, urgent concerns were raised about high staff turnover and vacancies in dementia care (House of Commons Committee of Public Accounts, 2008). More recently it has been suggested that “job satisfaction, stress and burnout have a significant correlation with [nurses’] intention to leave and the UK has one of the highest rates of nurses reporting burnout across Europe” (Health Education England, 2014). However, it is unclear how common and to what degree burnout affects care home staff in England or other countries.
High staff turnover in care homes is associated with lower quality of care (Castle and Engberg, Reference Castle and Engberg2005). Increased job stress and self-perceived delivery of poor care (Schmidt et al., Reference Schmidt, Dichter, Bartholomeyczik and Hasselhorn2014) have been linked with high staff turnover and absenteeism (Chiu et al., Reference Chiu, Chung, Wu and Ho2009; Larrabee et al., Reference Larrabee2010; Schaefer and Moos, Reference Schaefer and Moos1996), although social support may be protective (Woodhead et al., Reference Woodhead, Northrop and Edelstein2016). Furthermore high job strain in mid-life significantly increases the risk of onset of mental illness (Harvey et al., Reference Harvey2018). Despite the many challenges staff caring for people with dementia in 24-hour-care-settings facility (Zimmerman et al., Reference Zimmerman2005), an earlier systematic review found only five studies evaluating stress or burnout in long-term care staff (Pitfield et al., Reference Pitfield, Shahriyarmolki and Livingston2011). The authors concluded that most staff who remain working in homes do not report high levels of stress or burnout.
We aimed to update the evidence in this important area. Our objectives were: (1) to synthesize and meta-analyze the evidence regarding the prevalence and level of burnout and psychological stress in staff caring for people living with dementia in long-term care and (2) to examine putative socio-demographic, employment (nurse, care assistant, day or night or both, professional experience, education), and other risk factors.
Method
The protocol was registered on PROSPERO (number CRD42017074762).
We searched MEDLINE (1950–), PsycINFO (1872–), CINAHL (1961–) and Web of Science (1945–) databases from 2009 to August 10, 2017. Our search terms were: care, nursing, residential, old age, part III/three/3, elderly mentally ill/EMI, 24-hour care or old people’s homes, combined with staff, carers, workers, care workers, nurses, nursing assistants, employees, or healthcare assistants, combined with burden, burnout, stress, distress, anxiety, depression, and strain. We included primary, quantitative research studies that used a valid and reliable instrument to measure psychological distress or burnout in staff directly caring for people with dementia living in a 24-hour, long-term care settings.
Data extraction and quality assessment
We extracted data from all studies and two authors (HC, SW) independently rated the papers’ quality according to eight criteria, with one point for each positive answer (Boyle, Reference Boyle1998). These criteria were also used in an earlier systematic review of burnout in this population (Pitfield et al., Reference Pitfield, Shahriyarmolki and Livingston2011). We only included studies that (1) had a valid measure of burnout or psychological distress and (2) used standardized data collection for inclusion, and therefore all studies scored a minimum of two points. Higher-quality studies were defined as those meeting six criteria and above (i.e., validity score of 6 or more).
The criteria were:
(1) Was the target population clearly defined by clear inclusion and exclusion criteria?
(2) Was probability sampling used to identify potential respondents (or the whole population approached)?
(3) Did characteristics of respondents match the target population, i.e., was the response rate ≥80% or appropriate analysis included comparing responders and non-responders?
(4) Were data collection methods standardized?
(5) Was the burnout/psychological distress measure reliable? (If the original measure was valid but it was translated or adapted without reliability of changed measure being reported, we allocated 0.5 points.)
(6) Was the burnout/psychological distress measure valid? (If the original measure was valid but it was translated or adapted without validity of changed measure being reported, we allocated 0.5 points.)
(7) Were features of sampling accounted for in the analysis, through appropriate weighting of the data, or the whole population approached?
(8) Did the reports include confidence intervals for statistical estimates or, if not, did they provide sufficient data to allow for confidence intervals to be calculated?
Analysis
If studies were longitudinal or trials of interventions with multiple data collection points, we examined baseline data. We calculated 95% Confidence Intervals (CIs) for the prevalence or mean values of distress or burnout where the authors had provided sufficient information but had not given CIs. The criterion for conducting meta-analysis was if three or more studies with comparable burnout measures were identified. We used random effects models in StatsDirect software (version 3.1.14) to meta-analyze pooled mean effect scores from studies, as it is suitable for combining studies where heterogeneous populations are reported as it accounts for between-study variance. We rated heterogeneity using I2 indexes. We categorized results using burnout cut-off scores based on American normative population data (Maslach and Jackson, Reference Maslach and Jackson1981; Schaufeli and Vandierendonck, Reference Schaufeli and Vandierendonck1995).
Results
We identified 2,854 papers, excluded 2,681 of these by title and retrieved the remaining 173 full papers; we included 17 of these in the final systematic review (see PRISMA diagram in Figure 1).
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_fig1g.gif?pub-status=live)
Figure 1. PRISMA flow diagram of systematic study selection.
Validity criteria score (see Table 1)
The majority (14/17) of studies used probability sampling to identify respondents or approached the whole population. 8/17 studies fulfilled 6 criteria or more during quality assessment and were rated higher quality. Only 4/17 studies accounted for non-responders or had a response rate over 80%. Two studies from the same author used a measurement tool with a low reported reliability coefficient score (Cronbach’s alpha = 0.49) (Edvardsson et al., Reference Edvardsson, Sandman, Nay and Karlsson2009; Edvardsson et al., Reference Edvardsson, Sandman and Borell2014).
Table 1. Demographics and Validity Criteria scores of all included studies
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_tab1.gif?pub-status=live)
Swedish Demand-Control-Support Questionnaire (SDCS), 36-Item Short Form Survey (SF-36), 12 Item Short Form Survey (SF-12), Beck Depression Inventory (BDI), Beck Anxiety Inventory (BAI), Zarit Caregiver Burden Interview (ZBI), Hopkins Symptom Check List, 10 items (HSCL-10), Perceived Stress Scale (PSS), Health Professions Stress Inventory (HPSI), Maslach Burnout Inventory (MBI), Utrecht Burnout Scale (UBOS), Personal Burnout’ scale of the Copenhagen Psychosocial Questionnaire (COPSOQ). (N) Nurses, (NA) Nurse aides, (–) Not stated.
Studies reporting care staff burnout (Table 2)
Levels of Burnout
Nine studies reported mean burnout scores in care staff working in either nursing or residential care home facilities in eight different countries. Seven used versions of the MBI; four employed the complete 22-item MBI (Barbosa et al., Reference Barbosa2015; Duffy et al., Reference Duffy, Oyebode and Allen2009; Furumura and Ishitake, Reference Furumura and Ishitake2014; Juthberg et al., Reference Juthberg, Eriksson, Norberg and Sundin2010), one a 9-item version (Chamberlain et al., Reference Chamberlain2017). Two studies used a 20-item Dutch translation of the MBI, called the Utrecht Burnout Scale (UBOS) (de Rooij et al., Reference de Rooij, Luijkx, Declercq, Emmerink and Schols2012; Willemse et al., Reference Willemse2015).
Four studies found low levels and three studies moderate levels of emotional exhaustion (see Table 2). The highest reported emotional exhaustion score was found in a small sample (n = 61) of staff working in seven UK continuing care homes for residents with dementia (Duffy et al., Reference Duffy, Oyebode and Allen2009). All six studies reporting the depersonalization subscale score found moderate or low levels of depersonalization (see Table 2). Seven studies found moderate or high scores of personal accomplishment. The only study to report high burnout in any subscale was of 333 care staff working in dementia nursing homes in Japan that found low levels of personal accomplishment (Furumura and Ishitake, Reference Furumura and Ishitake2014).
Table 2. Studies reporting mean burnout scores
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_tab2.gif?pub-status=live)
Maslach Burnout Inventory (MBI) - 22 items, EE (9 items), DP (5 items), PA (8 items) answered on a 7-point scale, ranging from 0 (never) to 6 (every day). Utrecht Burnout Scale (UBOS) - Dutch translation derived MBI. 20 items EE (8 items), DP (5 items), PA (7 items). MBI (shortened) - scoring cut offs provided by personal communication with author (Leiter) of MBI. Personal Burnout’ scale of the Copenhagen Psychosocial Questionnaire (COPSOQ). Not stated (–)
Meta-analysis of the four studies that used the complete 22-item MBI found moderate mean emotional exhaustion levels (mean 18.34, 95% CI 14.59–22.10), low depersonalization (6.29, 2.39–10.19), and moderate personal accomplishment (33.29, 20.13–46.46) (Figures 2–4). Heterogeneity between studies in the meta-analysis was high, with I2 indexes of 95.1%, 98.8%, and 99.8% for emotional exhaustion, depersonalization and professional accomplishment respectively.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_fig2g.gif?pub-status=live)
Figure 2. Meta-analysis of emotional exhaustion burnout score.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_fig3g.gif?pub-status=live)
Figure 3. Meta-analysis of depersonalisation burnout score.
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_fig4g.gif?pub-status=live)
Figure 4. Meta-analysis of personal accomplishment burnout score.
Prevalence of Burnout
Three studies reported prevalence of burnout (Chamberlain et al., Reference Chamberlain2017; Duffy et al., Reference Duffy, Oyebode and Allen2009; Juthberg et al., Reference Juthberg, Eriksson, Norberg and Sundin2010). All used different cut-off points, only one of which was validated for the population (Juthberg et al., Reference Juthberg, Eriksson, Norberg and Sundin2010). The percentage of staff reporting high levels of emotional exhaustion ranging from 22.1% to 68.6%, depersonalization ranging from 9.2% to 46% and low levels of personal accomplishment, ranging from 4% to 24.5% (Chamberlain et al., Reference Chamberlain2017; Duffy et al., Reference Duffy, Oyebode and Allen2009; Juthberg et al., Reference Juthberg, Eriksson, Norberg and Sundin2010).
Two studies used the four-item “Personal Burnout” scale of the Copenhagen Psychosocial Questionnaire (COPSOQ) in 731 and 305 care staff in German nursing homes respectively (Schmidt et al., Reference Schmidt, Dichter, Palm and Hasselhorn2012; Schmidt et al., Reference Schmidt, Dichter, Bartholomeyczik and Hasselhorn2014). Burnout scores were significantly higher relative to the comparative normative data available from Denmark (mean = 43.9 SD 27.0 vs mean = 34.1 SD 18.2) (Pejtersen et al., Reference Pejtersen, Kristensen, Borg and Bjorner2010; Schmidt et al., Reference Schmidt, Dichter, Palm and Hasselhorn2012). The second study (Schmidt et al., Reference Schmidt, Dichter, Bartholomeyczik and Hasselhorn2014) found similar results (2007: 42.7).
Studies using measures of care staff psychological distress (see Table 3)
Three studies used versions of the Short Form-36 (SF-36), to measure mental health-related quality of life score in staff working in care homes in Australia, Brazil, and the United Kingdom respectively (Gao et al., Reference Gao, Newcombe, Tilse, Wilson and Tuckett2014; Islam et al., Reference Islam, Baker, Huxley, Russell and Dennis2017; Lucchetti et al., Reference Lucchetti2014). All found lower mental health quality-of-life scores than the equivalent age and sex matched normative populations. One of these studies additionally reported that of the 105 nurse aides that participated in the study, 5.8% were depressed, 23.2% had significant anxiety and 38% experienced significant stress, using the Beck Anxiety Inventory (BAI) and Beck Depression Inventory (BDI), though the definition of “significant stress” and diagnostic cut offs were not reported (Lucchetti et al., Reference Lucchetti2014).
A study of 198 staff working in care homes in Norway measured symptoms of anxiety and depression using the 10-item Hopkins Symptom Check List (HSCL-10) (Testad et al., Reference Testad, Mikkelsen, Ballard and Aarsland2010). The mean item score (1.33) was lower than the cut point for psychological distress (>1.85) (Testad et al., Reference Testad, Mikkelsen, Ballard and Aarsland2010). The two other studies reporting mean stress score used the Health Professions Stress Inventory (HPSI) and Swedish Demand-Control-Support Questionnaire (SDCS), which are valid measures of job strain and stress but do not generate a diagnosis nor have population normative values (Edvardsson et al., Reference Edvardsson, Sandman and Borell2014; Vogel et al., Reference Vogel, De Geest, Fierz, Beckmann and Zuniga2017).
Table 3. Studies reporting mean stress scores
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20190830105301503-0465:S1041610218001606:S1041610218001606_tab3.gif?pub-status=live)
Swedish Demand-Control-Support Questionnaire (SDCS), 36-Item Short Form Survey (SF-36), 12 Item Short Form Survey (SF-12), Beck Depression Inventory (BDI), Beck Anxiety Inventory (BAI), Zarit Caregiver Burden Interview (ZBI), Hopkins Symptom Check List, 10 items (HSCL-10), Perceived Stress Scale (PSS), Health Professions Stress Inventory (HPSI).
Summary
A meta-analysis of four studies found that staff had moderate levels of emotional exhaustion and personal accomplishment and low levels of depersonalization relative to USA normative data.
None of the seven studies measuring mean burnout with the MBI found high levels of emotional exhaustion or depersonalization. One large study reported low levels of personal accomplishment.
All three studies examining mental health-related quality-of-life scores reported lower levels in care staff than the equivalent age and sex matched normative populations. In one study comparing care staff mental health to normative population data, levels of psychological distress were not elevated.
Factors associated with staff burnout or psychological distress
Demographic factors
The two largest studies to measure the association between age and burnout (n = 1194) (Chamberlain et al., Reference Chamberlain2017), or age and stress (n = 344) (Edvardsson et al., Reference Edvardsson, Sandman, Nay and Karlsson2009), found no significant association. This was consistent with a study of 212 UK care staff which reported no association between length of time working in the care sector (which has a relationship to age) and mental-health-related quality of life (Islam et al., Reference Islam, Baker, Huxley, Russell and Dennis2017). However, 3/6 smaller studies found that younger age was associated with higher emotional exhaustion (Duffy et al., Reference Duffy, Oyebode and Allen2009), higher depersonalization (Furumura and Ishitake, Reference Furumura and Ishitake2014) and higher stress (Lucchetti et al., Reference Lucchetti2014). Furthermore, a study of 198 care staff in nursing homes in Norway reported higher age was correlated with perceived stress (Testad et al., Reference Testad, Mikkelsen, Ballard and Aarsland2010).
Most care staff in the studies were female. Only one of three studies reporting the relationship between burnout or psychological stress and care staff gender found a significant association: that male staff had higher depersonalization scores (Furumura and Ishitake, Reference Furumura and Ishitake2014).
One large Canadian study of care aides working in nursing homes examined the association between staff’s first language and burnout. Of 1,194 care aides included, 48.3% did not speak English as a first language and this was associated with higher levels of emotional exhaustion (p = 0.008) and depersonalization (p = 0.002) (Chamberlain et al., Reference Chamberlain2017).
Summary
Six studies (n = 2116) examined age of care staff with burnout and stress with no association found in most (n = 1617).
Three studies (n = 1739) examined the association of burnout or stress with sex of care staff with no clear association with no association found in most (n = 1406).
The only study (n = 1194) to consider first language reported significantly higher levels of emotional exhaustion and depersonalization in nursing home staff who did not speak English as a first language.
Staff role, education, experience and shift patterns
Two studies assessed whether stress and burnout levels varied with staff role (Islam et al., Reference Islam, Baker, Huxley, Russell and Dennis2017; Juthberg et al., Reference Juthberg, Eriksson, Norberg and Sundin2010). No significant difference in burnout levels was found between 50 nurses and 96 nurse aides working in care homes in Sweden (Juthberg et al., Reference Juthberg, Eriksson, Norberg and Sundin2010). In contrast a UK study in 72 care homes found that nursing staff experienced less job satisfaction and lower mental-health-related quality of life than staff without a nursing qualification (Islam et al., Reference Islam, Baker, Huxley, Russell and Dennis2017). This was in line with a study of 333 staff in Japanese nursing homes that found significantly lower levels of personal accomplishment in carers with formal care qualifications (Furumura and Ishitake, Reference Furumura and Ishitake2014). However, a study of 344 care staff in Sweden found education levels to be significantly lower in those reporting higher job strain (Edvardsson et al., Reference Edvardsson, Sandman, Nay and Karlsson2009).
The only study to consider turnover, found shift work was negatively associated with turnover in 239 care staff (b = 0.22, p = 0.02) (Gao et al., Reference Gao, Newcombe, Tilse, Wilson and Tuckett2014). Greater care aide job satisfaction was associated with lower burnout on all three burnout subscales of the MBI (Chamberlain et al., Reference Chamberlain2017). Similarly, nurses that reported high levels of satisfaction with quality of care for residents with dementia had lower levels of burnout and higher general health (Schmidt et al., Reference Schmidt, Dichter, Bartholomeyczik and Hasselhorn2014).
Summary
We found conflicting evidence regarding whether carer role and qualifications was associated with burnout or stress levels.
One study of 239 care staff found that turnover was lower in staff who did shift work.
Both studies (n = 1499) exploring job satisfaction found that higher satisfaction was associated with lower burnout.
Staffing and care facility-related factors
Three studies (n = 5204), examined staffing levels. Higher perceived staffing and resource adequacy on the Practice Environment Scale-Nursing Work Index (PES –NWI) (Lake, Reference Lake2007) was significantly associated with lower stress on all three dimensions of stress of the HPSI (p < 0.001) in 3,922 care staff in Switzerland (Vogel et al., Reference Vogel, De Geest, Fierz, Beckmann and Zuniga2017). A study of 1,194 care staff in Canadian nursing homes found perception of fewer staffing resources were associated with increased emotional exhaustion (Chamberlain et al., Reference Chamberlain2017). Similarly, in a Swedish study of 88 care staff, those with insufficient time to accomplish their work reported more job strain (Edvardsson et al., Reference Edvardsson, Sandman, Nay and Karlsson2009).
Four studies (n = 5408) assessed care home factors. A study of 80 care staff, comparing burnout between “small-scale” and “traditional” long-term care settings in Belgium and the Netherlands concluded that staff experienced significantly more emotional strain in small scale (size undefined) compared to traditional settings (de Rooij et al., Reference de Rooij, Luijkx, Declercq, Emmerink and Schols2012). By contrast, in Canadian nursing homes (n = 1194) significantly higher emotional exhaustion was found in medium (80–120 bed) compared to small (35–79 bed) facilities but no difference in burnout level was found between small and large (>120 bed) facilities (Chamberlain et al., Reference Chamberlain2017). A UK study found that nursing home staff reported worse mental health and wellbeing on the SF-12 than staff working in residential homes (Islam et al., Reference Islam, Baker, Huxley, Russell and Dennis2017). Finally, a study of 3,922 Swiss care staff found dementia specialist care unit staff reported higher stress levels than those on non-specialized units (Vogel et al., Reference Vogel, De Geest, Fierz, Beckmann and Zuniga2017).
Summary
All three studies (n = 5204) examining perceived low staffing levels found it was associated with increased stress, job strain, and emotional exhaustion.
There was contradictory findings from the two studies (n = 1274) to explore relationships of stress or burnout to care home size.
In two individual studies, stress levels were higher in UK nursing home staff (compared to residential homes) (n = 212) and in Swiss dementia specialist care unit staff compared to non-specialist units (n = 3922).
Work environment
All seven studies (n = 7186) that evaluated the relationship of the work environment to burnout or stress in care staff reported a relationship between a poor environment and burnout or stress. A study of 1,194 care staff found a perception of insufficient space to discuss care needs was associated with increased emotional exhaustion (Chamberlain et al., Reference Chamberlain2017).
Higher burnout scores were found in 1,093 care staff in the Netherlands reporting higher job demands, greater authority to make their own decisions at work and less supervisor or co-worker support (Willemse et al., Reference Willemse, de Jonge, Smit, Depla and Pot2012). The same study found supervisor support protected against emotional exhaustion related to job demands in staff with low decision authority; however, co-worker support was associated with lower personal accomplishment scores in high job demands settings (Willemse et al., Reference Willemse, de Jonge, Smit, Depla and Pot2012). Similarly, a study of 3,922 care staff in Switzerland found those that perceived more effective leadership was associated with significantly less work stress (Vogel et al., Reference Vogel, De Geest, Fierz, Beckmann and Zuniga2017).
A study of 239 Australian nurses found those with lower reported job demands and higher coping resources reported significantly higher levels of emotional wellbeing as measured on the SF-36 and were less like to leave their role (Gao et al., Reference Gao, Newcombe, Tilse, Wilson and Tuckett2014).
Perceived reciprocity with colleagues protected from emotional exhaustion, while self-efficacy was significantly negatively associated with burnout in 61 UK care staff (Duffy et al., Reference Duffy, Oyebode and Allen2009). Similarly, a study of 333 care staff in Japan found those who felt they lacked workplace support and reported conflict between staff had significantly higher levels of depersonalization and lower personal accomplishment (Furumura and Ishitake, Reference Furumura and Ishitake2014). This finding of workplace support being a factor in staff burnout and stress was consistent with results in 344 care staff in Swedish nursing homes that found “possibilities to have discussions” about difficult situations with colleagues at work was significantly associated with lower job strain (Edvardsson et al., Reference Edvardsson, Sandman, Nay and Karlsson2009). The same study also found a “perceived caring climate” in the unit, as determined using a visual analogue scale ranging from “very bad” to “very good” was significantly associated with lower job strain (Edvardsson et al., Reference Edvardsson, Sandman, Nay and Karlsson2009).
Summary
All seven studies (n = 7186) that evaluated the relationship of the work environment to burnout or stress in care staff found a positive association between a poor environment and burnout or stress. Four studies (n = 1575) found perceived support from colleagues protected against burnout and stress.
Two studies (n = 5015) found more effective perceived leadership protected against burnout and stress.
Two studies (n = 1332) found lower job demands was significantly associated with lower burnout and better emotional wellbeing.
One study (n = 61) examined “self-efficacy” and found it was significantly associated with lower burnout.
One study (n = 344) found a “perceived caring climate in the unit” was significantly associated with lower job strain.
Resident factors
Three out of four studies examining the role of resident behavior and agitation found a significant association with levels of burnout and stress in care staff. A study of 731 care staff in German nursing homes reported significantly higher levels of staff distress and burnout when there was more challenging resident behavior, with highest distress related to resident aggression (Schmidt et al., Reference Schmidt, Dichter, Palm and Hasselhorn2012). Similarly, a study in 156 Swiss nursing homes found that staff who had experienced or observed verbal or sexual aggression against other residents had significantly higher stress scores compared to those who had not (Vogel et al., Reference Vogel, De Geest, Fierz, Beckmann and Zuniga2017). Care aides in Canadian nursing homes who experienced more dementia-related agitated behaviors reported significantly higher levels of emotional exhaustion and depersonalization (Chamberlain et al., Reference Chamberlain2017).
No significant association between agitation, as measured on the Cohen-Mansfield Agitation Inventory, and stress was found in 198 care staff in Norway (Testad et al., Reference Testad, Mikkelsen, Ballard and Aarsland2010).
Summary
Three out of four studies (n = 5847) found higher burnout or stress in care staff exposed to agitated or aggressive resident behavior.
Discussion
To our knowledge this is the largest systematic review of burnout and stress in staff caring for people with dementia in 24-hour care settings and the first meta-analysis. We found in meta-analysis, that staff in general report moderate levels of emotional exhaustion, low depersonalization and moderate personal accomplishment. The only study to report high levels of burnout on any subscale was a study of 333 nurses working in a dementia care focused nursing home in Japan that reported low levels of personal accomplishment. The high heterogeneity that we found is in keeping with similar findings (I2 indexes >90%) in other meta-analyses of burnout using the MBI measure, with language, country, and sampling method being moderator variables of heterogeneity (Aguayo et al., Reference Aguayo2011). The possible stressors and protective factors are likely to vary between countries and the differences suggests that burnout can be avoided or alleviated. We used normative data about burnout from America to classify burnout levels. However, there were differences in scoring cut points of burnout across countries (Maslach and Jackson, Reference Maslach and Jackson1981). The cut off points determined in the American normative sample are significantly higher than those for a Dutch normative sample (Schaufeli and Vandierendonck, Reference Schaufeli and Vandierendonck1995).
Studies examining burnout or stress related factors suggested those caring for residents exhibiting abusive or aggressive behaviors, those who work in environments where they feel unsupported, had lower job satisfaction (which is unsurprising as it may be the opposite of personal accomplishment)—and a perceived lack of staff and resources, and one large study found higher levels in those who do not speak English (the language of the country in which the study took place) as a first language. Staffing levels can be higher because a home is better resourced or because the residents have more needs. Therefore perception of adequacy of staffing may be a better measure. However, perception of inadequate staffing it may also reflect feelings of emotional and physical exhaustion and be hard to disentangle. These factors may explain the wide range of reported prevalence of emotional exhaustion (22.1%–68.6%) in three studies and suggests there are at risk groups of carers with higher burnout levels.
We only examined burnout and stress in long-term care settings and it would be unsurprising if this varied between settings. For example, two studies in Japan in intermediate care and psychiatric hospitals reported high levels of burnout and stress in care staff (Tanaka et al., Reference Tanaka2015; Yada et al., Reference Yada2014). This may be because of higher care needs and more challenging behavioral symptoms of patients which are associated with burnout level of staff (Agoub et al., Reference Agoub, Elyazaji and Battas2000). It may also reflect sociocultural factors in Japan, which has the lowest proportion of residential care for people ≥65 years old, and the highest proportion of people aged ≥65 in hospitals (Ikegami and Campbell, Reference Ikegami and Campbell1995; Ribbe et al., Reference Ribbe1997). An understanding of the country specific provision of care and care environments for people living with dementia is necessary to evaluate the pressures on staff working in these environments.
If burnout is not high then it is unlikely to account for high staff turnover. It is, however, possible that care staff with high burnout levels were not identified by these studies. Though many of the studies had high response rates, none considered whether responders differed from non-responders. This could result in a systematic bias. In addition, if people with high burnout leave more rapidly they will be less represented in a prevalence study.
There was a lack of longitudinal follow up of staff and only one study analyzed staff turnover by sending follow up questionnaires to 239 Australian care staff after two years (Gao et al., Reference Gao, Newcombe, Tilse, Wilson and Tuckett2014). This found that higher job demands and lower coping resources were directly associated with turnover. The only other study to follow up care staff showed an increase in burnout levels over a two year period (Schmidt et al., Reference Schmidt, Dichter, Bartholomeyczik and Hasselhorn2014). The group of staff reported as “still satisfied” with the quality of care of residents with dementia, had the highest overall ratings of work ability (a measure of motivation and capability at work, assessed by the work ability index) and lowest burnout, which would support the idea that those who were dissatisfied and thus more likely to leave would be more burnt out.
Conclusion
We conclude that current evidence suggests that most staff caring for people living with dementia in long-term care do not have a high level of burnout or psychological stress. It is likely that there are at risk groups of care staff who are more susceptible to burnout and stress or who are in environments which are particularly likely to lead to burnout. Preliminary evidence in small studies suggest these may include: factors relating to the difficulties of caring such as those who care for residents with aggressive behavior; factors relating to working in a poor environment such as not having sufficient space; factors related to the care staff; such as not speaking English as a first language, which may make people feel culturally isolated and have less local family support as well; and feeling unsupported with poor leadership within the care setting. It is now important to have a large prospective study of burnout and find out if these factors are independently associated with higher burnout. Longitudinal follow-up will also enable the effect on staff turnover to be evaluated. This will enable understanding of what are logical interventions to target support of staff and improve resident care.
Conflict of interest
None.
Description of author’s roles
Harry Costello–—conducted systematic review and meta-analysis, wrote the paper. Sebastian Walsh—second reviewer of studies identified by systematic review. Claudia Cooper—supervised study design and statistical analysis and assisted with writing the paper. Gill Livingston—designed study, assisted with analysis and writing the paper.
Acknowledgments
GL is supported by North Thames NIHR CLAHRC. CC and GL are supported by UCL NIHR BRU. HC is supported by UCL NIHR.