VA is committed to providing high quality care in its Community Living Centers (CLCs). One way in which care quality is being improved is through a cultural transformation aimed at making CLCs more person-centered and less institutional by providing CLC residents with more choice and autonomy. A cultural transformation of this type necessitates changes to staff roles and responsibilities and has implications for direct-care workers. Previous research has shown that the ability of direct-care workers to perform in these new and more person-centered ways may be dependent on changes to the supervisory styles of nurse managers and other supervisors (Tyler & Parker, 2011; Lopez, 2006). Thus, while the cultural transformation taking place in VA CLCs will require direct-care workers to acquire new skills and knowledge, their ability to apply these new skills and knowledge may be reliant on the management practices of their supervisors.
This study utilized a multi-level approach to test training interventions for direct-care workers and their supervisors in VA CLCs. Our specific aims and hypotheses are as follows:
Aim 1: Determine the effectiveness of an education intervention for DCWs designed to improve the response to challenging resident behaviors, communication skills, and clinical problem-solving skills of CLC direct-care workers.
Aim 2: Determine the effectiveness of a multi-level education intervention to both increase DCW knowledge and skills and improve supervisor management and communication skills.
Aim 3: Describe the link between CLC resident health outcomes (i.e. quality indicators) and both direct-care worker knowledge, skills, and communication and management practices.
Aim 4: Determine what factors are implicated in the effectiveness of the intervention implementation and what differentiates CLCs that achieve better results from those that do not.
Twenty CLCs participated in this study: Five served as a control group; five received training for nurse managers only; five received training for DCWs only; and five received both training interventions. Using a pre-test and post-test design with a control group, our analyses examine the extent to which the combinations of interventions led changes in practice and resident outcomes. We measured changes in practice using the Care Coordination Survey. We also utilized the Resident Assessment Instrument Minimum Data Set (MDS) data to examine health outcomes for residents. Finally, telephone interviews were conducted with staff at six select CLCs to explain differences in intervention implementation and success.
Aim 1: Effectiveness of DCW Training
DCWs at intervention sites received two trainings. The first focused on improving staff communication and the second on improving skills related to the care of residents with dementia and/or problem behaviors. After collection of our pre-intervention surveys, our consultants in VACO requested a change to our planned dementia training. This necessitated a change in our survey questions related to dementia knowledge and skills and resulted in our being unable to measure change in these skills using pre-test post-test methods. Instead we could only analyze cross-sectional (post-survey) differences among the participating CLCs. Our ability to do this was further inhibited by VACOs requirement, issued during our intervention phase, that all VA CLC staff be educated using the same dementia training that was utilized in our study. This resulted in staff at several control sites receiving this part of the intervention. Not surprisingly, we detected no significant differences between our intervention and control sites on our measures of dementia skills and knowledge.
We did find statistically significant differences between intervention sites on our measures of supervisory support and relational coordination. Three of the intervention sites showed significant improvements on one or both of these scores while three showed significant declines on these measures.
Aim 2: Effectiveness of Multi-Level Intervention
We used hierarchical linear modeling (HLM) to examine the effect of our DCW and nurse supervisor (coaching) training interventions on our survey based measures of communication, decision making, staffing, human resource management, supervisor support and relational coordination. We found no significant differences between intervention groups in communication, decision making, staffing or human resource management. We found that the coaching only group had significantly higher supervisory support scores, but significantly and unexpectedly lower relational coordination scores than the other groups.
The purpose of Aim 3 was to describe the link between CLC resident health outcomes (e.g., antipsychotic use, pressure ulcers, urinary tract infections, behavioral symptoms, and functional decline) and direct-care worker knowledge, skills, and communication and management practices. We used hierarchical linear modeling (HLM) to examine the effect of our survey based measures on our 5 resident outcomes including random effects for facility and resident identification number in the model. We found no statistically significant effects of the interventions, in part because due to low variation on some of these outcomes and in part due to the mixed results of the interventions overall.
The most promising model was for the behavioral symptoms outcome where there were marginally significant findings for supervisory support (p=0.060) and supportive organizational context scales (p=0.083) when models were run separately. These findings suggest that behavioral symptoms could be improved by changing supervisory support and supportive organizational context, as the interventions had been intended to do.
In order to determine what factors are implicated in the effectiveness of the intervention implementation and what differentiates CLCs that achieve better results from those that do not we conducted telephone interviews with staff at six CLCs. We used our measures of supervisory support and relational coordination to identify the six CLCs. Three of the intervention sites were found to have significant improvements on one or both of these scores (high performers) and three sites were found to have significant declines on these measures (low performers). Twenty-four interviews were conducted across the six CLCs, including at least three interviews at each CLC. A mix of staff including nurses, nursing assistants and CLC leadership were interviewed.
One challenge reported by all six CLCs was leadership turnover. Key leaders at all six of the CLCs had left the CLC during the course of this two year project. The primary difference between the high performers and low performers was the timing of this leadership turnover. In the high performer sites, leaders did not leave the facility until after the training intervention had been implemented, while in the low performer sites leaders left the CLC before the training was implemented.
Among the low performers, the intervention was not fully implemented, and few if any staff received the training. Among the high performers, nearly all targeted staff received the training intervention. This was often accomplished by securing support of middle managers, especially nurse supervisors, who as a result were more willing to release their staff from duties on their unit in order to receive the training.
The interventions did not have the hoped for effect on care. However, the study findings suggest that the impetus-to target supervisory support and organizational support in bolstering the effects of DCW training-focused in the right direction. Key findings about the importance of leadership consistency to see through an intervention and buy in from middle management may be helpful in other interventions.
- Sullivan JL, Weinburg DB, Gidmark S, Engle RL, Parker VA, Tyler DA. Collaborative capacity and patient-centered care in the Veterans' Health Administration Community Living Centers. International Journal of Care Coordination. 2019 Jun 1; 22(2):90-9.