Premium
Design of a clustered observational study to predict emergency admissions in the elderly: statistical reasoning in clinical practice
Author(s) -
Lancaster Gillian A.,
Chellaswamy Hannah,
Taylor Steve,
Lyon David,
Dowrick Chris
Publication year - 2007
Publication title -
journal of evaluation in clinical practice
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.737
H-Index - 73
eISSN - 1365-2753
pISSN - 1356-1294
DOI - 10.1111/j.1365-2753.2006.00663.x
Subject(s) - observational study , sample size determination , confidence interval , cluster sampling , confounding , sample (material) , cluster (spacecraft) , research design , class (philosophy) , medicine , stratified sampling , cluster analysis , sampling (signal processing) , health care , clinical study design , psychology , statistics , computer science , environmental health , mathematics , artificial intelligence , clinical trial , population , chemistry , filter (signal processing) , chromatography , pathology , economics , computer vision , programming language , economic growth
Objective To describe the statistical design issues and practical considerations that had to be addressed in setting up a clustered observational study of emergency admission to hospital of elderly people. Study design and setting Clustered observational study (sample survey) of elderly people registered with 18 general practices in Halton Primary Care Trust in the north‐west of England. Results The statistical design features that warranted particular attention were sample size determination, intra‐class correlation, sampling and recruitment, bias and confounding. Pragmatic decisions based on derived scenarios of different design effects are discussed. A pilot study was carried out in one practice. From the remaining practices, a total of 4000 people were sampled, stratified by gender. The average cluster size was 200 and the intra‐class correlation coefficient for the emergency admission outcome was 0.00034, 95% confidence interval (0–0.008). Conclusion Studies that involve sampling from clusters of people are common in a wide range of healthcare settings. The clustering adds an extra level of complexity to the study design. This study provides an empirical illustration of the importance of statistical as well as clinical reasoning in study design in clinical practice.