Background: Citizens have conveyed to professionals that care at the end of life is less than
optimal. Efforts to improve matters have tended to work in piecemeal fashion, on tangible
more than personal aspects of care, and without the benefit of documented perspectives of
those who face dying. Policy initiatives and clinical interventions need guidance from a broad
framework that is validated by patients' perspectives.
Purpose: Our goals were to: (1) assess the construct validity and stability over time of the
portions of a conceptual framework that concern patients' subjective experiences; (2) develop
a foundation for measurement of these personally meaningful factors; and (3) examine these
factors' associations for potential clinical or policy significance.
Population: Patients were from six diverse geographical areas whose physicians judged
their survival prognosis to be 6 months or less. Physicians who referred the patients were randomly
selected from state and specialty association lists.
Methods: We used in-person survey methodology and multivariate analysis of patient responses.
The analysis fell into two parts. Using exploratory factor analysis, we looked for evidence
of discrete dimensions of experiences. Using regression analysis, we examined associations
among them.
Results: Of 1,131 eligible patients, 988 were interviewed (87.4% response rate). Of 682 patients
who survived to follow-up interview 4-6 months later, 650 were interviewed (95.3% response
rate). Exploratory factor analysis identified 12 discrete factors (accounting for 55% of
variance; maximum Spearman's p = 0.24), 8 of which met criteria for representing measurable
dimensions (accounting for 46% of variance). These 8 were: patient-clinician relationship; social
connectedness; caregiving needs; psychological distress; spirituality/religiousness; personal
acceptance; sense of purpose; and clinician communication. Eigenvalues ranged from
1.45 to 6.30 and Cronbach's α from 0.63 to 0.85.
The concordance between these dimensions and those in the proposed framework indicated
that two dimensions required minor modifications and six were confirmed, providing
evidence of good construct validity for this portion of the framework. The same dimensions
were also evident at follow-up except that the first two above-listed loaded on one combinedfactor, clinician interaction (eigenvalues 1.83-7.92; Cronbach's α from 0.64 to 0.86). This provides
evidence of the construct's stability over time.
Clinical communication and patient-clinician relationship were associated (odds ratio [OR]
2.79, 2.31-3.36). Better clinician communication correlated with somewhat better personal acceptance
(OR 1.10,1.02-1.19), and a better patient-clinician relationship correlated with less
psychological distress (OR 0.84, 0.75-0.95).
Conclusions: We conclude that: (1) Personally meaningful aspects of patients' experience of
terminal illness can be represented in valid, multidimensional constructs that are stable over
time; (2) They are measurable; and (3) Aspects of the therapeutic relationship appear to correlate
with patients' experience of the dying process.
