Studying symptoms: sampling and measurement issues

Abstract

Physical symptoms account for more than half of all outpatient visits, but the specific causes and treatments of many common symptoms are frequently not well established. Because clinical research has primarily targeted specific diseases, symptoms-based research is still an emerging field of scientific inquiry, and optimal methods are still being developed. This paper considers eight important issues in sampling and measurement. First, recognition of the exceptionally high base rates of many symptoms is an essential primary step in selecting appropriate case-patients and controls. Second, the type of population studied (community, primary care clinic, subspecialty clinic, or symptom clinic) can greatly influence the severity and other characteristics of the symptom being evaluated. Third, symptom detection relies on three types of symptoms: documented, elicited, and volunteered. Fourth, temporal factors include recency of onset, episodicity, and duration. Fifth, severity can be assessed with single- or multiple-item symptom-specific scales as well as evaluation of generic functional status and health-related quality-of- life measures. Sixth, recognition of a symptom's cause can be strengthened by explicit criteria, uniform assessment protocols, and multiple raters. Seventh, comorbidity that should be assessed includes medical disorders, psychiatric disorders, and other symptoms. And eighth, since death and serious illness are uncommon, alternative outcomes that might be measured include symptom alleviation, health-related quality of life, patient and provider satisfaction, and health care costs.