Exploring the iceberg of morbidity: a comparison of different survey methods for assessing the occurrence of everyday illness.

This paper examines problems in measuring the occurrence of acute symptoms of ill health. Health interview surveys and health diaries often lead to different results. Two key hypotheses assume that: 1, interviews using checklists are more sensitive to the respondent's psychological distress than are the open-ended questions of health diaries; and 2, health diaries demand high levels of compliance leading to underreporting of symptoms. An additional 3rd hypothesis assumes that the effect of psychological distress on response patterns is strong for reporting psychological symptoms but insignificant for musculoskeletal symptoms. The hypotheses were tested and explored with data from the Dutch Survey of General Practice, a nationwide study among 161 GPs. A random sample of 100 patients per GP was approached for a health interview and asked to keep a structured health diary during three weeks. Symptoms were recorded during the interview with a checklist and queried in the health diary with open-ended questions. The occurrence of symptoms was modelled with logistic regression. High levels of psychological distress increase the likelihood of recording symptoms for both instruments, but the increase is greater for the interviews. Respondents who have only received limited education, heavy smokers and those who suffer from chronic conditions have a significantly lower likelihood of recording symptoms in the diary as compared to the questionnaire. There was no significant effect of taking an interest in health matters, gender, and work and domestic role obligations. Taking the nature of symptoms into account, it was found that psychological distress had indeed a great effect on the response pattern for psychological symptoms, but not for musculoskeletal symptoms. The criticism that symptom checklists are sensitive to psychological distress rather than to physical illness alone, is confirmed in this study. Open-ended questions prevent biased responses, but result in fewer symptoms being recorded. Health diaries with open-ended questions 'produce' more symptoms but take more effort to complete, requiring sufficiently motivated respondents. It is recommended that a less biased specific list for the assessment of acute symptoms be developed.