Primary care providers (PCPs) in our study wanted to care for their dying patients and largely felt competent to provide end-of-life care, but many faced obstacles related to five structural elements of the system in which they practiced: continuity of care, scheduling flexibility and time with patients, coordination of care, information-sharing, and authority to act. We found that the presence of these structures fostered patient trust in the physician and physician knowledge of the patient, which, in turn, were necessary for PCPs to address both the physical and psychosocial needs at the end of life. Interestingly, the elements physicians needed for providing good end-of-life care are elements germane to providing good primary care in general; i.e., no structural elements unique to the provision of palliative care were identified.
This study has its limitations. Because this was a qualitative study, our findings are not meant to be generalized to all primary care practices, but rather to provide insight into phenomena that cannot be studied quantitatively. There is the possibility of sample bias, as those individuals who volunteered for the focus groups could have been particularly sympathetic or antagonistic to the topic of inquiry. Moreover, primary care practices in southeast Michigan may not be illustrative of primary care practices elsewhere.
Additionally, the small size of some of our focus groups makes it more likely that one strong opinion could sway the tenor of discussion. There is also the possibility that focus group participants provided answers that were socially desirable or that their responses were influenced by cognitive dissonance; however, the spectrum of responses suggests that this was not common. Lastly, we did not directly observe providers in practice or collect data from patients, and so could not confirm providers’ self-reports.
Nevertheless, our findings resonate with the existing literature. Studies from Europe have shown that general practitioners value coordination, time, authority, and continuity.33–36
Studies in the US have shown that “role ambiguity” is a barrier to end-of-life care when multiple providers attend to the same patient,27
and that communication between primary care and subspecialists is inadequate.37,38
Additionally, the connections between continuity and trust,39
as well as continuity and physician knowledge40
have been demonstrated before.
Our data provide some clues as to how practices can overcome some of the barriers to providing end-of-life care. These include personal actions by PCPs, such as personally calling subspecialists to ask for prognostic information, or making structural changes to the practice, such as hiring a social worker. Still, the specific structures and processes that are put into place to address structural barriers will need to vary based on organizational context41
and practice type. A future study with an even larger purposive sample stratified by organizational and practice characteristics would be needed to produce a valid description of how these characteristics affect provision of end-of-life care.
Some might posit that advance directives could be used to delegate authority over end-of-life care to the PCP; however, it is important to remember that advance directives only come into play once patients have lost decisional capacity, while delegation of authority is an issue the entire time a patient is being cared for within a complex medical system.
Many of the factors we identified as crucial to providing end-of-life care are consistent with elements of the Patient Centered Medical Home (PCMH),23
which emphasizes patient access to care, care coordination, and smooth information transfer, both within the practice and with outside providers. Whether or not the PCMH model will allow PCPs to provide robust end-of-life care is a hypothesis worth future examination.
Having the means to provide robust end-of-life care does not guarantee that PCPs will provide the care. Across our panel of subjects, we saw variation in the sense of ownership of end-of-life care. How physician attitudes determine practice change or use of available supports is an important relationship that was not explored in this study. Moreover, while the PCPs in this study felt confident in their skills for providing end-of-life care, other studies have demonstrated that many lack those skills.42–44
Future qualitative work is necessary to understand how PCP knowledge and attitudes affect the willingness to put in place and use structures that facilitate good end-of-life care.
It is important to underscore that our results do not suggest that all five practice elements are necessary or that any one is sufficient for PCPs to provide end-of-life care. Moreover, the factors themselves are not binary in nature, but they can be present within a practice in varying degrees depending upon the characteristics of the practice. In-depth ethnographic work could shed more light on how these factors work together in various contexts to affect the quality and delivery of end-of-life care. Future research into how to measure these factors may provide tools necessary for practice evaluation and improvement, as well as statistical testing of relationships.