养老院护理人员工作环境与遗漏和仓促护理任务之间的关系

Importance  In Canada, approximately 81% of residents of nursing homes live with mild to severe cognitive impairment. Care needs of this population are increasingly complex, but resources, such as staffing, for nursing homes continue to be limited. Staff risk missing or rushing care tasks and interfering with quality of care and life.

Objective  To assess the association of work environment with missing and rushing essential care tasks in nursing homes.

Design, Setting, and Participants  This cross-sectional study used survey data collected from a random sample of 93 urban nursing homes in Western Canada, stratified by health region, owner-operator model, and facility size, between May and December 2017. All 5411 eligible care aides were invited to participate, and 4016 care aides agreed and completed structured, computer-assisted interviews in person. Analyses were conducted from July 4, 2018, to February 27, 2019.

Main Outcomes and Measures  Self-reported number of essential care tasks missed (range, 0-8) or rushed (range, 0-7) in the most recent shift. Two-level random intercept hurdle regressions controlled for care aide, care unit, and nursing home characteristics.

Results  Of 4016 care aides, 2757 (68.7%) were 40 years or older, 3574 (89.1%) were women, and 2663 (66.3%) spoke English as an additional language. For their most recent shift, 2306 care aides (57.4%) reported missing at least 1 essential care task and 2628 care aides (65.4%) reported rushing at least 1 essential care task. Care aides on units with more favorable work environments (eg, more effective leadership, better work culture, higher levels of buffering resources) were less likely to miss any care tasks (odds ratio, 1.59; 95% CI, 1.34-1.90; P < .001) and less likely to rush any care task (odds ratio, 1.66; 95% CI, 1.38-1.99; P < .001).

Conclusions and Relevance  This study found that rates of missed and rushed essential care in Canadian nursing homes were high and were higher in units with less favorable work environments. This finding suggests that work environment should be added to the list of modifiable factors associated with improving nursing home care, as it may be an important pathway for improving quality of care. Further research is needed to understand associations of missed and rushed care and of improving work environments with outcomes among residents of nursing homes.

Nursing homes care for people with complex medical and social care needs. More than 80% of direct care is performed by care aides (also called nursing assistants).¹ Despite increasingly complex care needs of residents, nursing home funding remains limited,² while workloads of care aides increase and become more complex. Care aides may rush and miss care tasks to meet tight schedules and task lists,³ although those tasks are essential to both quality of care and quality of life. Emerging research suggests that care aides frequently leave essential care undone^4,5 and rush essential care.⁶

Both quality of care and quality of life for residents likely diminish with missed and rushed care, but it is not fully understood why missed and rushed care happens. That knowledge is critical for interventions to reduce it. Kaplan et al⁷ identified optimized organizational context (ie, work environment) as important to successful interventions for quality improvement. In acute care, better organizational context is associated with less missed care.^8-11 However, to our knowledge, there are no data for nursing homes.

We examined how modifiable elements of organizational context are associated with missed and rushed care by care aides in nursing homes to inform the design of interventions that could reduce missed and rushed care. We hypothesized that care aides working on care units with more favorable organizational context would report fewer missed and rushed essential care tasks than those working on care units with less favorable context.

This study used cross-sectional survey data from the Translating Research in Elder Care (TREC) research program,¹² which is a multilevel, longitudinal program of applied health services research to improve quality of care and quality of life for residents of nursing home and quality of work life for the staff who care for them. Since 2007, TREC has collected data from a cohort of 93 nursing homes across Western Canada. Variables used in this study were initially developed in 2012, piloted during data collection from 2014 to 2015, and first comprehensively collected from May to December 2017. This study is reported following Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Canada’s health regions oversee most aspects of nursing homes, including quality of care and nursing home policies, with exceptions in some regions for for-profit operators.¹³ Health regions vary in oversight.¹³ Study facilities for TREC were randomly selected from lists stratified by health region (ie, British Columbia Fraser Health Authority, British Columbia Interior Health Authority, Alberta Health Edmonton Zone, Alberta Health Calgary Zone, or Winnipeg Regional Health Authority), owner-operator model (ie, public nonprofit, voluntary nonprofit, or private for-profit), and facility size (ie, small, defined as <80 beds; medium, 80-120 beds; or large, >120 beds). Specific sampling and data collection approaches are described in detail elsewhere.¹² Care units (also known as clinical microsystems^14,15) in this study were physical locations in nursing homes with a unit manager, a nurse overseeing shift-by-shift care, and dedicated teams of regular staff providing direct care. These clinical microsystems are essential foci for improvement programs.

This study was approved by the University of Alberta research ethics board. For original TREC survey data collection, participating organizations gave operational approvals. Using multiple strategies, we informed potential survey participants about this study. Written informed consent was obtained from participants.

Care aides were eligible to participate if they had worked in a study facility for longer than 3 months, could identify a unit where they worked for at least 50% of their time during the data collection period, and worked on that unit for 6 or more shifts in the past month. Trained TREC data collectors went to each participating nursing home and invited all eligible care aides to participate.¹² Care aides who agreed to participate completed computer-assisted, structured interviews in person¹⁶ between May and December 2017.

Only care units with responses from at least 8 care aides were included in this study. This criterion reflects our finding that stable estimates of organizational context at the unit level are achieved when 8 or more individual responses are aggregated.¹⁵

Our independent variable was organizational context at the unit level, measured by the Alberta Context Tool (ACT). This validated instrument measures 10 modifiable elements of organizational context: leadership, culture, evaluation, formal interactions, informal interactions, structural and electronic resources, social capital, organizational slack in use of time, organizational slack in use of staff, and organizational slack in use of space.¹⁷ eTable 1 in the Supplement summarizes the psychometric properties of ACT. Each element has 2 to 9 items. Each element was aggregated within respondents by the mean or the count of items (Table 1). We used care aide responses to create unit-level scores for context because care aides are the only sufficiently large workforce group in nursing homes to support aggregate scores. Care aides also provide 80% or more of direct resident care¹ and are most familiar with residents and conditions of work and resident life. They are well positioned to provide context scores that are closest to resident experiences. We aggregated to the unit level the 10 ACT elements from each care aide surveyed. Then, drawing on Milligan and Cooper’s clustering analysis approach¹⁸ (n of predetermined clusters = 2), we determined whether a unit had a more or less favorable organizational context.¹⁹

Our dependent variables were missed care and rushed care at the individual care aide level. Missed care and rushed care items were developed iteratively by engaging care aides in developing previous TREC surveys.⁴ Items followed the form, on your last shift, did you leave mouth care for residents undone because you did not have enough time? Care aides were asked yes or no for each missed and rushed task item. Yes responses were counted and summed separately for missed care and rushed care.

For multivariable analysis, we adjusted for covariates related to care aide, unit, and nursing home characteristics (Table 1). Inclusion of these variables in final models was informed by previous research on missed care in acute settings and nursing homes.^4,6,8,20 Care aides self-reported their demographic and work characteristics.¹² Unit and facility characteristics were collected in short structured interviews with unit and facility managers. For missing data, we used listwise deletion. Less than 0.2% of data were missing for all variables, completely at random (Little missing completely at random test,²¹P = .84). eTable 2 in the Supplement shows characteristics of care aides with complete vs incomplete data.

Analyses were conducted from July 4, 2018, to February 27, 2019, using SAS statistical software version 9.4 (SAS Institute). For descriptive analysis, variable categories were care aide demographic characteristics, care aide work characteristics, unit and facility characteristics, and health region. We calculated means and SDs for continuous variables and frequency counts and proportions for categorical and binary variables. We summed the number of times that care aides answered yes to missing or rushing each care task and calculated percentage of occurrence in the whole sample. We then ranked missed and rushed care tasks from highest to lowest percentage of occurrence in the whole sample. To inform inclusion of variables in subsequent multivariable analysis, we drew on previous literature on missed and rushed care in short- and long-term care settings. Multicollinearity was assessed, and 2 variables (ie, hours worked in 2 weeks and time worked as a care aide) were removed before multivariable analysis.

For regression analyses, we ran 2 sets of analyses, one using ACT as a binary variable and the other with the 10 ACT elements individually. Hurdle Poisson regression models were used for both analyses. They account for excessive numbers of 0s using 2 components: logistic and Poisson regression models. The logistic model recodes the dependent variable (ie, missed care or rushed care) as a binary variable: 0 or 1. It models probability when the dependent variable is 0. The Poisson regression model only uses data with a non-0 dependent variable: 1 to 8 for missed care and 1 to 7 for rushed care. Odds ratios (ORs) with 95% CIs were generated for the logistic model. Relative rates (RRs) with 95% CIs were generated for the Poisson model. To control for the clustering effect of individual care aides within units, we added a random intercept parameter to each component. P values were 2-tailed, and statistical significance was set at P < .01.

Of 5411 eligible care aides in 312 care units in 93 nursing homes, 4016 care aides (74.2%) responded to the survey invitation. Data to assess nonresponder characteristics were not available because data on nursing home care aides in Canada are not kept systematically.¹³ Characteristics of included care aides are presented in Table 2. Care aides were predominantly women (3574 care aides [89.1%]) 40 years or older (2757 care aides [68.7%]) who spoke English as an additional language (2663 care aides [66.3%]) (Table 2). Characteristics of care units and nursing homes are presented in Table 3.

Missing and rushing care tasks were common in nursing homes. For their most recent shift, 2306 care aides (57.4%) reported missing at least 1 care task and 2628 care aides (65.4%) reported rushing at least 1 care task (Table 4). The most frequently missed task was taking residents for a walk, which 1492 care aides (37.2%) reported missing. The most frequently rushed task was talking with residents, which 1977 care aides (49.2%) reported rushing. Performing mouth care was missed by 567 care aides (14.1%) and rushed by 1580 care aides (39.3%). Other care tasks, such as toileting, preparing residents for sleep, bathing residents, feeding residents, and dressing residents, were each missed by less than 10% of care aides but were each rushed by more than 30% of care aides (Table 4).

Compared with care aides who worked on units with less favorable organizational context, care aides who worked on units with more favorable organizational context were 59% less likely to miss any care task (OR, 1.59; 95% CI, 1.34-1.90; P < .001) and 66% less likely to rush any care task (OR, 1.66; 95% CI, 1.38-1.99; P < .001) (Table 5). Among care aides who reported rushing at least 1 task, those working on units with more favorable organizational context rushed 7% fewer care tasks (RR, 0.93; 95% CI, 0.88-0.98; P = .007) than care aides on units with less favorable organizational context (eTable 3 and eTable 4 in the Supplement).

To understand which specific elements of organizational context were associated with missed and rushed care in our sample, we conducted further analyses with the 10 ACT elements. Missed care was associated with 4 elements: culture, organizational slack in use of staffing or use of time, and social capital (eTable 5 in the Supplement). With a 1-unit increase in care unit culture, care aides were 62% more likely to miss any care task (OR, 0.38; 95% CI, 0.19-0.73; P = .004). However, with a 1-unit increase in organizational slack in use of staff, care aides were 65% less likely to miss any care task (OR, 1.65; 95% CI, 1.32-2.05; P < .001). For organizational slack in use of time, care aides were 103% less likely to miss any care tasks (OR, 2.03; 95% CI, 1.44-2.86; P < .001). Among care aides who reported missing at least 1 care task, a 1-unit increase in social capital was associated with missing 49% fewer care tasks (RR, 0.51; 95% CI, 0.36-0.70; P < .001), and a 1-unit increase of organizational slack in use of staff was associated with missing 20% fewer care tasks (RR, 0.80; 95% CI, 0.71-0.90; P < .001) (eTable 5 in the Supplement).

Rushed care was associated with 2 elements: organizational slack in use of staff and culture. With a 1-unit increase of organizational slack in use of staff on care units, care aides were 91% less likely to rush any care task (OR, 1.91; 95% CI, 1.51-2.42; P < .001). However, care aides who reported rushing at least 1 task rushed 34% more care tasks with a 1-unit increase in culture (RR, 1.34, 95% CI, 1.09-1.65; P = .006) (eTable 6 in the Supplement).

In this cross-sectional study, high proportions of nursing home care aides reported missing and rushing essential care tasks—tasks that are important to quality of care and quality of resident life. Excessive staff busyness has been identified as a significant daily challenge in nursing homes in Europe,²⁰ Canada,⁴ Japan,²² and Australia.²³ Care aides’ work is frequently interrupted, with only 1- to 3-minute uninterrupted intervals provided for approximately half of care tasks, including feeding and bathing.⁶ In Switzerland, nurses and care aides reported missing 46% of activities of daily living at least once in the previous 7 days.²⁰ In Japan, care workers have described only having time for minimum care (eg, bathing, changing diapers) and none for extra care, such as taking residents for a walk.²²

Research from acute settings supports that missing or rushing essential care interferes with quality and safety.^8,24 Missed care explained 40% of variation in quality ratings of US-based acute care hospitals²⁵ and 9.2% of variance in patient falls.⁹ In US and European acute care studies, missed care was associated with increased nosocomial infections, pressure ulcers, patient dissatisfaction, medication errors, readmission to hospitals, critical incidents, compromised patient safety, and increased mortality.^8,9,26 A 2018 study across 9 European countries²⁷ found that every 10% increase in missed care by nurses was associated with a 16% increase in odds of 30-day postoperative mortality. That study²⁷ also found that missed care was associated with mediating the association of nurse staffing level with postoperative mortality rate. From these findings, Ball and Griffiths²⁸ concluded that missed nursing care (ie, errors of omission) should be a key patient safety measure in hospitals alongside errors of commission. These reports suggest value in measuring and addressing missed care in nursing homes.

Almost no research is available on rushed care, to our knowledge. However, evidence is emerging on the benefits of so-called slow care, the opposite of rushed care, for residents of nursing homes, especially those living with dementia.²⁹ Staff who can give residents the time they need facilitate a sense of coherence and foster dignity.^29,30 Unrushed care by staff has potential to decrease responsive behaviors of residents.³¹ These studies strongly suggest that there are potential harms for residents when staff rush care tasks.

The 2 most frequently missed and rushed care tasks for nursing home care aides in our sample were walking and talking with residents. Talking is directly associated with preventing loneliness and boredom and with encouraging social engagement and creation of meaning. Walking is directly associated with mobility, a serious challenge in nursing homes.³² Immobility is associated with multiple adverse health and quality of life outcomes (eg, fecal incontinence, pressure ulcers, skin tears).^32-34 Both talking and walking with residents are associated with care quality and quality of life.

In our study, care aides working on care units with more favorable organizational context were less likely to miss or rush care tasks. Research has documented that modifiable features of organizational context (eg, leadership, culture, team communication) are associated with resident outcomes^35-39 and that these features are interrelated.^40-42 Each element of organizational context is modifiable and offers intervention possibilities. However, we propose that nursing home managers and researchers use organizational context as an omnibus construct, modifying multiple elements simultaneously instead of targeting single elements as they develop quality improvement interventions for resident outcomes. We adopted a validated measure of organizational context that draws on a conceptual framework⁴³ and is operationalized through a rigorous process of cluster analysis¹⁸ with the 10 modifiable ACT elements. Our findings may provide potential directions for nursing home managers to reduce missed and rushed care by care aides through improved local organizational context.

We found that units for which care aides perceived more organizational slack in use of staff and use of time had lower likelihoods or numbers of missed and rushed essential care tasks. Research has reported mixed evidence on the association of nursing staffing and resident outcomes, such as 2 systematic reviews in nursing homes that concluded that the evidence was inconsistent—higher staffing levels were associated with both better and worse resident outcomes.^44,45 However, staff perceptions of staffing level may differ from actual levels. Perceptions are affected by care unit composition, staff composition, and resident composition.⁴⁶ In our study, we identified variation in care aide perceptions of staffing and time across care units, although staffing levels for nursing homes (eg, number of care hours per resident day) are essentially constant across Western Canadian jurisdictions.^2,47,48 While funding for nursing homes continues to be limited, our finding suggests opportunities for nursing home managers to improve care aides’ perceptions of staffing. Focus on organizational context and its various elements may reduce essential care tasks missed or rushed by care aides.

We found that care aides on units with better work environment culture had higher likelihood or number of missed and rushed essential care tasks, opposite to our hypothesis. We do not yet understand why, but we urge further research on this association.

We found that care aides on units with higher levels of social capital (eg, active connections through information sharing) missed fewer care tasks. Existing evidence is mixed on the association of work environment social capital with resident outcomes. Leonard et al⁴⁹ have argued that effective communication is key to safe care; however, a randomized clinical trial by Colón-Emeric et al⁵⁰ teaching nursing home staff to improve connections with coworkers reported improved staff communication but not an increased number of fall risk reduction activities. Our finding offers evidence for the association of social capital with less frequently missed care, providing a potential mechanism for the association of social capital in care units with resident outcomes.

Our study has some strengths, including that we controlled for the clustering effect of care aides within care units in our analyses¹⁴ and we used a large stratified random sample of nursing homes.¹² Data collection used a rigorous in-person structured interview process with real-time data quality assessments. We included robust data and findings on both missed and rushed care, generating a more complete portrait of these phenomena in nursing homes than previous research has generated by studying only missed or rushed care.^6,20

Our study also has some limitations. We used survey data, which may be susceptible to self-report biases, although recall bias was reduced by asking aides to report on their most recent shift. The potential for misreporting, such as underreporting of missed and rushed care tasks, was reduced by our interview structure and data quality assessments. The study may have bias from omitted variables because we did not control for resident characteristics (eg, cognitive status, responsive behaviors) associated with missed and rushed care.⁵ We reduced bias by controlling for multiple proxy variables for these characteristics, such as unit type (eg, general long-term care, dementia care units) and the experiences of care aides of residents’ responsive behaviors. Also, the continuous part of the Hurdle Poisson regression models assumes that categories are equidistant. In addition, we did not include sampling weights in our models, as we were not interested in exact estimates of true population values but were interested in possible associations. For this purpose, sampling weights were less relevant.⁵¹ Furthermore, data were obtained from a stratified random sample of 93 of 524 facilities in the 3 provinces in Western Canada; thus, caution should be taken with wider generalizations.

This study’s findings suggest that rates of missed and rushed essential care are high, which may put residents of nursing homes at risk of adverse health outcomes and decreased quality of life. Researchers, policy makers, and health care system and nursing home managers should consider adding work environment to a list of modifiable factors to improve care and offer new intervention pathways for improving care quality.

Accepted for Publication: December 3, 2019.

Published: January 29, 2020. doi:10.1001/jamanetworkopen.2019.20092

Correction: This article was corrected on April 3, 2020, to correct an error in the Abstract and in the Results. The number of care aides who spoke English as an additional language should have been 2663.

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Song Y et al. JAMA Network Open.

Corresponding Author: Yuting Song, PhD, Faculty of Nursing, University of Alberta, 11405 87 Ave, 5-007D Edmonton Clinic Health Academy, Edmonton, AB T6G 1C9, Canada (yuting.song@ualberta.ca).

Author Contributions: Dr Song had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: All authors.

Acquisition, analysis, or interpretation of data: Song, Norton, Estabrooks.

Drafting of the manuscript: Song, Hoben, Estabrooks.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Song.

Obtained funding: Estabrooks.

Supervision: Norton, Estabrooks.

Conflict of Interest Disclosures: None reported.

Funding/Support: Funding was provided by the Canadian Institutes of Health Research and partners in the Ministries of Health in British Columbia, Alberta, and Manitoba, Canada, as well as regional health authorities in participating British Columbia and Alberta regions.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Meeting Presentation: This article was presented at the Canadian Association for Health Services and Policy Research Conference; May 30, 2019; Halifax, Nova Scotia, Canada.

Additional Contributions: The Translating Research in Elder Care 2.0 team contributed data to this study. The data were provided without compensation. Jayna Holroyd-Leduc, MD (Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta, Canada), and Sube Banerjee, MD (University of Plymouth, Plymouth, United Kingdom), provided constructive feedback on the manuscript. They did not receive compensation for the work. Cathy McPhalen, PhD (thINK Editing Inc), provided editorial support, which was funded by Carole Estabrooks’ Canada Research Chair, Ottawa, Ontario, Canada, in accordance with Good Publication Practice guidelines.