Table 2

Definition and methodological approach for secondary outcomes

Secondary outcome	Definition	Approach
Time from admission to death	Binary indicator for death (yes/no). Time to event defined as index admission date until the earliest of: death end of follow up Deaths will primarily be identified through linkage to ONS deaths registration data. However, a supplementary analysis will use HES data (where the method of discharge field is coded as ‘dead’ (4)) in addition to ONS deaths data. This is because HES records may better ascertain information on recent deaths where there is a delay in death registration (eg, because a coroner’s report is required).	Cox proportional hazards model
Duration of index hospital admission	(Date of discharge)—(date of admission)	Poisson/Zero-inflated Poisson model
Time from admission to avoidable deaths	Binary indicator for avoidable death (yes/no). Time to event defined as index admission date until the earliest of: avoidable death14 end of follow up Supplementary analyses will investigate amenable and preventable deaths (see online supplementary file 1), which are a subset of avoidable deaths.	Cox proportional hazards model
Time from discharge to ACS condition admission	Binary indicator for admission with ACS (yes/no). Time to event defined as index discharge date until the earliest of: ACS condition admission end of follow up ACS admissions are flagged within HES 14
Time from discharge to next elective admission	Binary indicator for elective readmission (yes/no). Time to event defined as index discharge date until the earliest of: elective readmission end of follow up Elective admissions are those where the admission method is 11, 12 or 13 in addition, a sensitivity analysis investigating planned admissions only (admission method 13) will be undertaken.14
Overall readmission rates	Number of readmissions divided by the total time under follow up between admissions (ie, where the patient was not already hospitalised). Calculated as (number of admissions occurring in the time from index discharge date to the earliest of death or November 2016) divided by (number of days from index discharge date to the earliest of death or November 2016 minus the number of days in the same time period where the individual was admitted to hospital).	Poisson model with the log of follow-up time as an offset
Unscheduled readmission rates	As for overall readmission rates (above) but excluding (from the numerator only) admissions where the admission method was elective (ie, 11–13).	Poisson model with the log of follow-up time as an offset
All-cause mortality expressed as a standardised mortality ratio	Deaths will primarily be identified through linkage to ONS deaths registration data, but also through HES (where the method of discharge field is coded as ‘dead’ (4)) as the latter method may better ascertain information on recent deaths where there is a delay in death registration (eg, because a coroner’s report is required).	Calculation of SMR using Office of National Statistics death data by age and gender.
ICD-10 chapter specific SMR	As for all-cause mortality (above), but examining deaths by ICD-10 chapter for primary cause of death
In-patient costs using HRG	Each entry will be assigned a unit cost based on its HRG. A total cost for each patient calculated as the sum of costs across all entries during the period.	A discounting rate of 3.5% will be applied and GLM modelling willbe undertaken with Gamma specification.

ACS, ambulatory care sensitive; GLM, generalised linear model; HES, Hospital Episode Statistics; HRG. Health Resource Group; ICD, International Classification of Diseases 10th Revision; ONS, Office for National Statistics; SMR, standardised mortality ratio.