Armstrong P; AcademyHealth. Meeting (2005 : Boston, Mass.).
Abstr AcademyHealth Meet. 2005; 22: abstract no. 3664.
University of East London, Health & Bioscience, Romford Road, London, E15 4LZ Tel. (0208) 223-4520 Fax (0208) 223-4965
RESEARCH OBJECTIVE: To compare two commonly used methods for calculating waiting times (an event-based approach and a census-based approach) with a third method, an enrollment-based approach, recently outlined in Statistics in Medicine. STUDY DESIGN: Secondary analysis of the same retrospective data using three different estimates of the percentage waiting 0-3 months. The event-based estimate counts all admissions from the waiting list within a specified calendar period and reports the proportion that was admitted within 0-3 months. (This approach was modified slightly to allow us to compare the same individuals regardless of how we calculate waiting times.) The census-based estimate counts everyone still awaiting admission on a particular date and reports the proportion found to have been waiting 0-3 months. The enrollment-based estimate reports admissions as a proportion of the total enrolled rather than as a proportion of the total admitted. POPULATION STUDIED: EyeNet provided information on all the cataract extractions reported to the National Cataract Registry for Sweden, 1992-2003 inclusive. Analysis was restricted to the 452,279 extractions performed within two years of enrollment. PRINCIPAL FINDINGS: The three methods yielded very different results. The event-based approach often (6 times out of 38) differed from the enrollment-based approach by ten percentage points, or more. The result was 12 percentage points too small in one instance and 14 percentage points too big in another. The census-based approach often (17 times out of 38) differed from the enrollment-based approach by ten percentage points, or more. Three months being a relatively short wait, the census-based approach generally gave an estimate that was too big. In one instance, the result for the National waiting list for cataract surgery was 24 percentage points too big.The enrollment-based estimate shows that the promptness of admission fell every summer, accompanied by a net increase in the size of the waiting list. It also shows that the promptness of admission rose every autumn, accompanied by a net decrease in the size of the waiting list. The two commonly used methods showed no evidence of seasonal variation. CONCLUSIONS: The two commonly used approaches give inaccurate estimates and invalid patterns. They hide changes in the length of waiting times. IMPLICATIONS FOR POLICY, DELIVERY OR PRACTICE: Researchers should beware of studies that collect retrospective data on the waiting times of a sample of consecutive admissions: these studies do not allow us to report the distribution of waiting times correctly. Policy-makers should review the evidence that suggests that elective activity has little effect on the size of a waiting list or on the length of waiting times: the commonly-used approaches are not designed to take account of the size of a waiting list or the amount of surgical activity. Health Service Managers should not expect census-based estimates or event-based estimates to show the effect of a waiting list initiative. They should use an enrollment-based approach if they want to examine changes in the promptness of admission.
Publication Types:
Keywords:
- Cataract Extraction
- Censuses
- Registries
- Sweden
- Waiting Lists
- methods
- statistics & numerical data
- surgery
- hsrmtgs
UI: 103623127
From Meeting Abstracts