Covariates
- Age in years
- Sex
- Financial year of admission
- Hospital Frailty Risk Score (HFRS)9
- Complex ESS: Complex surgery was defined using any occurrence of any
of the OPCS-4 codes E143, E147, E148, E151, E152 (seeSupplementary material Table S1 for descriptions).
- Diagnosis during index admission of obesity (ICD-10 code: E66-)
- Diagnosis of any of the 17 conditions that make up the Charlson
comorbidity index10
- Likely complications of surgery identified during the index admission.
Complications were defined as described above for the outcome
measures. This covariate was chosen in recognition of the fact that a
complication identified during the index procedure would, in many
cases, preclude same-day discharge and be associated with poorer
post-discharge outcomes.
Data management and
statistical analyses
Data were analysed using standard statistical software: Microsoft Excel
(Microsoft Corp, Redmond, WA, USA), Stata (Stata Corp LLC, College
Station, TX, USA) and Alteryx (Alteryx Inc, Irvine, CA, USA).
Age data were broadly normally distributed on visual inspection and
summarised using the mean and standard deviation. All other data were
categorical and were summarised using frequency and percentage.
Multilevel (hierarchical) logistic regression models were constructed.
All variables were treated as categorical in model building except age,
which was modelled as a continuous variable using restricted cubic
splines; knots (at the 10th, 50thand 90th percentile) were found to give optimal model
fit for the primary outcome based on Akaike’s Information
Criterion.11 Adjusted outcomes were calculated based
on fixed-effects within a conditional framework. Confidence intervals
(CIs) were used to draw inference, with a 95% CI for an odds ratio (OR)
not including the value 1 taken to indicate significance.