Related Experiment Videos
Regional differences in outcome from subarachnoid haemorrhage
P Mitchell1, B A Gregson, T Hope
1Department of Neurosurgery, University of Newcastle upon Tyne, Newcastle General Hospital, Newcastle upon Tyne, UK. Patrick.Mitchell@ncl.ac.uk
Acta Neurochirurgica
|August 5, 2005
Summary
Statistical analysis of surgical outcomes can be misleading. Initial data suggested differences between centers, but after accounting for patient factors, these disparities vanished, highlighting the need for comprehensive data before drawing conclusions.
Area of Science:
- Neurosurgery
- Clinical Outcomes Research
Background:
- Surgeons face pressure to publish results and adopt practice changes based on others' data.
- Concerns exist regarding the misinterpretation of clinical outcome data.
- This study illustrates potential pitfalls in interpreting surgical performance metrics.
Purpose of the Study:
- To investigate apparent differences in outcomes between two surgical centers.
- To examine the impact of confounding variables on initial statistical findings.
- To highlight the importance of comprehensive data collection in evaluating surgical performance.
Main Methods:
- Prospective data collection on subarachnoid hemorrhage treatment outcomes from 1993-1998.
- Comparison of outcomes between two centers (Newcastle and Nottingham).
- Detailed analysis incorporating confounding variables to re-evaluate initial findings.
Main Results:
- Initial analysis showed a statistically significant difference favoring Nottingham (odds ratio 2.3, p<0.00001).
- After adjusting for confounding factors, the performance difference between the centers disappeared.
- Differences in admissions policies, influenced by bed availability, accounted for the initial disparity.
Conclusions:
- Applying industrial statistical methods to complex medical data can be dangerous.
- Accurate assessment of surgical unit performance requires comprehensive data on all influencing factors.
- Judgments on apparent statistical differences should only be made after thorough consideration of all relevant variables.