S Hunskaar1, A Seim, T Freeman. 1. Department of Public Health and Primary Health Care, University of Bergen, Norway.
Abstract
BACKGROUND: The selection process of patients from community to hospitals may introduce bias into research and hamper the generalization of hospital-based research back to general practice. OBJECTIVE: The objective of this study was to use female urinary incontinence as a model in an attempt to provide empirical support for selection bias. METHOD: The analyses are based on three populations of incontinent women: community level (epidemiological survey, 535 women), primary care level (general practice, prospective clinical study, 105 women), and secondary care level (university hospital, prospective clinical study, 228 women). RESULTS: The general practice patients were older and the hospital patients younger than those in the community. From community via general practice to hospital, there was an increase in duration, frequency of leakage, amount of leakage, severity and perceived impact of incontinence. Help-seeking at the primary care level was associated with increasing age and severity, and with urge symptoms and substantial impact. Referral from general practice to hospital level was only associated with age and urge symptoms. CONCLUSION: The study provides empirical evidence to support the existence of selection bias. This phenomenon must not be overlooked when recommendations developed at the consultant level are presented at a level with a significantly different clinical picture of a condition.
BACKGROUND: The selection process of patients from community to hospitals may introduce bias into research and hamper the generalization of hospital-based research back to general practice. OBJECTIVE: The objective of this study was to use female urinary incontinence as a model in an attempt to provide empirical support for selection bias. METHOD: The analyses are based on three populations of incontinent women: community level (epidemiological survey, 535 women), primary care level (general practice, prospective clinical study, 105 women), and secondary care level (university hospital, prospective clinical study, 228 women). RESULTS: The general practice patients were older and the hospital patients younger than those in the community. From community via general practice to hospital, there was an increase in duration, frequency of leakage, amount of leakage, severity and perceived impact of incontinence. Help-seeking at the primary care level was associated with increasing age and severity, and with urge symptoms and substantial impact. Referral from general practice to hospital level was only associated with age and urge symptoms. CONCLUSION: The study provides empirical evidence to support the existence of selection bias. This phenomenon must not be overlooked when recommendations developed at the consultant level are presented at a level with a significantly different clinical picture of a condition.