PURPOSE

We use longitudinal data to test and extend a structural equation model documenting changes in the causal connections among symptoms experienced in the final weeks of life. Our central thesis is that the relief of suffering and the promotion of quality end of life care require tailoring interventions to reflect the shifting causal foundations of symptoms.

METHODS

Symptom information on pain, anxiety, nausea, shortness of breath, drowsiness, loss of appetite, tiredness, depression, and well-being was extracted from a palliative care database. For each of the 82 study participants, symptom scores measured at 4 full weeks and 1 full week prior to death were used to test a structural equation model of the causal structures underlying symptom clusters.

RESULTS

This investigation confirms the reasonableness of our previously developed model. Tiredness, depression, and well-being were sufficiently labile that the observations at one week before death were not significantly dependent on the corresponding observations 3 weeks earlier. Patients' assessments of pain, anxiety, nausea, shortness of breath, drowsiness, and appetite were only moderately stable over this same period.

CONCLUSIONS

The stability in some, and instability in other, symptoms meshed convincingly with the changes in symptom causal structures previously derived from cross-sectional data. Investigations assessing temporal shifts in palliative symptom coordination over longer periods of time and for specific medical conditions and social contexts seem warranted.