Geographic Influences on Appendicectomy Outcomes

The aim of this study is to explore the possibility that outcome from appendicectomy may vary according to:

1. the urban-rural nature of the patient's home location and
2. travel time from hospital.

This is a retrospective study of all appendicectomies performed in Scotland during the period from 1st January 2001 - 31st December 2010. It will use routinely collected administrative data from the Information Services Division (ISD) of NHS National Services Scotland.

Potentially significant confounding variables such as age, gender, and co-morbidity will be studied for their predictive value in a univariate model and included in a multivariate model if they remain significant.

There is no single appendicectomy outcome measure which is a literature standard, so all of the following will be evaluated as primary outcome measures: risk-adjusted 30 day/inpatient mortality, 30 day readmission rate, 30 day re-operation rate, length of stay and negative appendicectomy rates.

The registry which will supply the data for this study is the Scottish Morbidity Record 01 (SMR01), the full title of which is the "General / Acute Inpatient and Day Case dataset" (see http://www.adls.ac.uk/nhs-scotland/general-acute-inpatient-day-case-smr01/?detail). SMR01 is collated and administered by ISD, and data submission is mandatory for all Scottish NHS providers of in-patient or day-case care. Approximately 1.4 million records are added each year. Diagnoses are coded according to International Classification of Diseases (ICD)-10 standards and procedures are coded according to the United Kingdom's Office of Population Census Statistics (OPCS) standards, the most recent of which is version 4.5.

The data quality in SMR01 is high and is assured by regular internal audits. In the 2010 audit of accuracy, Main Condition was recorded with an accuracy of 88% and Main Procedure was recorded with an accuracy of 94%. Where data inconsistencies are identified in the extract supplied for this study, further clarification will be obtained where possible with ISD's data retrieval support team. Data completeness is very high in SMR01. However, where significant volumes of data are missing or unusable, the need for data imputation will be explored.

The study period was decided on pragmatically by a desire to provide an assessment of current practice, fully within the era of widely practised laparoscopic surgery. A power calculation also suggested that this would provide an adequate sample size to demonstrate differences.

Our power calculation- specific to length of stay- was based on Faiz O, Clark J, Brown T, Bottle A, Antoniou A, Farrands P, et al. Traditional and Laparoscopic Appendectomy in Adults. Ann Surg. 2008 Nov;248(5):800-6. In their cohort of 259,735 appendicectomies performed from 1996-2006, the geometric mean length of stay was 3.52, with SD 1.8. We decided that a difference of 0.5 days would be 'clinically significant'. We specified alpha 0.05 and Power 0.9, and an allocation ratio of 2 (allowing comparison of one tertile to two others). A two-sided t-test of difference between two independent means was performed in G*Power 3.1.7. This demonstrated that a total N of 616 was required to demonstrate this difference. According to ISD figures, there were 3,712 appendicectomy procedures performed in 2010/11, so with numbers available we will be able to detect a clinically significant difference in length of stay.

The investigators will use the Scottish government's own 8-fold urban-rural classification system (see http://www.scotland.gov.uk/Topics/Statistics/About/Methodology/UrbanRuralClassification) to investigate the possible relationship between patient geographical location and outcome. Urban-rural categories may be grouped for analysis, depending on initial exploratory work.

The investigators will further calculate the approximate travelling time from patient home location (using postcode) to the treating hospital, by constructing isochrones at time/distance intervals from hospitals. Travel time will be evaluated as a continuous and discrete variable.