We conducted a cross-sectional study to determine, by chart review, how consistently one FHT was implementing elements of PAC with patients in their last year of life. We chose the last year of life to align with literature showing sharp increases in levels of disability in the last year of life, the frequent reference to the last year of life in recommendations to initiate palliative care, and recognizing that a palliative approach to care should be incorporated from the time of diagnosis of a progressive life-limiting illness.
The study was conducted in an FHT made up of two multi-physician family medicine clinics in one city, serving approximately 40,000 patients. There are 40 family physician practices, diverse allied health professionals who work between practices and 80 family medicine residents. One hundred patient charts were randomly selected among 192 patients who died from January 1st to December 31st 2017. With inter-rater reliability being previously established, charts were equally divided between 2 reviewers for data extraction. Inclusion criteria were patients who had been a member of the practice for at least 1 year prior to their death and had a chronic, progressive life-limiting illness. Exclusion criteria included: orphan patients (admitted to the practice at end of life for the sole purpose of palliative care and not having a previously identified family doctor); sudden and unexpected deaths unrelated to a chronic progressive life-limiting illness; and patients residing in a long-term care facility for over 6 months of the last year of their life.
We created a chart review tool to capture the elements of PAC in primary care (Supplementary file 1). To develop the tool, we combined three conceptual domains of a palliative approach, and three core elements of published definitions of what constitutes palliative care from another systematic review. The three domains (early upstream approach, adaptation of palliative care knowledge and expertise to other settings, integration into systems) and three elements of the definition (whole person care, quality of life focus, mortality acknowledgement) were linked to the operational items/elements in the chart review form to create a tool for evaluating PAC in primary care. Categories in the final chart audit form included advance care planning topics documented, symptoms documented, type of clinicians involved in care, frequency of telephone contacts and home visits, involvement of formal home care services and attention to caregiver involvement and well-being. For descriptive purposes we also collected information on demographics, health conditions and illnesses most responsible for death.
Since there is no current operational definition of a ‘palliative approach,’ the chart reviewers provided an overall assessment of their judgement of whether the care given by the family practice in the last year of life aligned with PAC. To make this assessment of having received PAC, patients required clinical documentation of elements mapping to all 3 domains. For practicality and to ensure the focus was on the efforts of the primary care team, documentation considered included the chart’s running commentary and the cumulative patient profile only. Consultant notes, messages and external documents were not considered because we wanted to capture care provided by the family practice, and because it was not practicable to review the volume of information. The timeframe for the chart review and assessment of a palliative approach was one year prior to the patient’s death.
For reasons of feasibility, we randomly selected 100 charts to review. Patient characteristics and outcomes of interest were described using mean (standard deviation) for continuous variables and count (percent) for categorical variables. For comparisons of outcomes between patients who did and did not receive a PAC, independent sample t-tests (for continuous data) and Pearson’s Chi-squared tests (for categorical data) or Fisher’s exact tests (for categorical data with expected cell sizes <5) were used. Haldane’s correction was applied to contingency tables with observed zero values. Statistical significance of comparisons was indicated by a p-value <0.05 (two-tailed) for the test. For comparisons of proportions, an absolute difference of 30% between groups could be detected with 80% power, and there was greater than 80% power to detect a mean difference of 1.0 between groups for continuous variables.
Following data analyses, results were presented to clinicians within the FHT (N=58) in a group meeting. A pre- and post- questionnaire was administered immediately before and then after the presentation, to determine perceptions around how consistently clinicians perceived they delivered the elements of a palliative approach. The questionnaire was created by the authors (Supplementary File 1). The survey included 8 questions and used 7-point Likert scale responses (1=infrequently, 7=regularly). The difference in mean scores for each question were compared by the paired t-test. Analyses were conducted using SAS© software, version 9.4 for Windows.
The STROBE checklist for cross-sectional studies was used. Ethics approval for the chart review was received by the Hamilton Integrated Research Ethics Board for this project (File number 2017-4204-C). We did not seek ethics approval for the anonymous staff questionnaire as this was undertaken to inform potential quality improvement efforts locally. We explained the purpose of the questionnaire and informed staff that completion was voluntary.