Evaluation of the Association Between Day and Time of Admission to Critical Care and Acute Hospital Outcome

1. Objective To conduct an analysis of the Case Mix Programme database to evaluate the association between day and time of admission to critical care and acute hospital outcome for unplanned admissions to National Health Service adult, general critical care units in England, Wales and Northern Ireland.
2. Exposure The primary exposure variable will be the day and time of admission to the critical care unit, dividing the week into 14 time periods: Monday 08:00-17:59; Monday 18:00-Tuesday 07:59; Tuesday 08:00-17:59, etc. For summary tables, these will be collapsed into the following categories:

• Weekday daytime: Monday-Friday 08:00-17:59
• Weekend daytime: Saturday-Sunday 08:00-17:59
• Weekday night: Monday-Friday 18:00-07:59 the next day
• Weekend night: Saturday-Sunday 18:00-07:59 the next day

1. Outcome The primary outcome will be acute hospital mortality, defined as death before ultimate discharge from acute hospital.
2. Key confounders The key potential confounders to be adjusted for in the analysis are:

• Age
• Severe conditions in the past medical history
• Prior dependency
• Number of days from hospital admission to critical care unit admission
• Location prior to admission
• Cardiopulmonary resuscitation (CPR) within 24 hours prior to admission to critical care
• Primary reason for admission to critical care
• Acute severity of illness - Intensive Care National Audit \& Research Centre (ICNARC) Physiology Score Age will be defined as the age in whole years at the time of admission to the critical care unit and will be included in the model as a linear term. Severe conditions in the past medical history will be defined according to the Acute Physiology And Chronic Health Evaluation (APACHE) II methodology and must have been evident and documented in the six months prior to admission to the critical care unit. Conditions will be categorised into seven binary covariates:
• Severe cardiovascular disease - fatigue, claudication, dyspnoea or angina at rest (New York Heart Association functional class IV)
• Severe respiratory disease - shortness of breath with light activity due to pulmonary disease or home ventilation
• End-stage renal failure - requirement for chronic renal replacement therapy for irreversible renal disease
• Severe liver disease - biopsy proven cirrhosis, portal hypertension or hepatic encephalopathy
• Haematological malignancy - acute or chronic myelogenous/lymphocytic leukaemia, multiple myeloma or lymphoma
• Metastatic disease - distant metastases documented by surgery, imaging or biopsy
• Immunocompromise - radiotherapy, chemotherapy, daily high-dose steroids, HIV/AIDS or congenital immuno-humoral or cellular immune deficiency state Prior dependency will be assessed based on the ability to perform usual daily activities (e.g. bathing, dressing, going to the toilet, moving in/out of bed/chair, continence and eating) during the two weeks prior to admission to acute hospital and prior to the onset of the acute illness, categorised as:
• Able to live without assistance in daily activities
• Some (minor/major) assistance with daily activities
• Total assistance with all daily activities The number of days from acute hospital admission to critical care unit admission will be categorised as 0 days (i.e. admission to the critical care unit on the same calendar day as admission to hospital), 1 day, 2 days, 3-7 days or 8 or more days. Location prior to admission to the critical care unit will be categorised as:
• Emergency department/not in hospital
• Theatre (emergency/urgent surgery)
• Ward, obstetric area or intermediate care area Patients admitted from a transient area (e.g. imaging, specialist treatment area, recovery) will be assigned to a category according to their previous location. CPR within 24 hours prior to admission to critical care will be defined as receipt of internal or external cardiac massage (with or without defibrillation) and categorised as either in-hospital (administered by an in-hospital resuscitation team or equivalent), out-of-hospital or no CPR. The primary reason for admission to the critical care unit will be recorded using the ICNARC Coding Method, a hierarchical coding method specifically designed for recording reasons for admission to critical care. For descriptive tables, primary reason for admission will be categorised according to the body system (respiratory, cardiovascular, gastrointestinal, neurological, genitourinary, endocrine or musculoskeletal/dermatological). For modelling, primary reason for admission will be categorised according to body system and pathological/physiological process (e.g. respiratory infection), with categories containing fewer than 100 admissions combined at the body system level. Acute severity of illness will be assessed using the ICNARC Physiology Score based on the highest/lowest values of 12 physiological parameters during the first 24 hours following admission to the critical care unit. The APACHE II Score will also be reported as a descriptor of severity of illness but not included in modelling due to collinearity with age, past medical history and physiology. Delayed admission to the critical care unit (defined as a delay of at least one hour between the documented decision to admit a patient to the critical care unit and the actual admission, and further subcategorised as a delay of more than 4 hours versus up to 4 hours) will be explored as a potential mediator of the effect of day/time of admission on outcome.

1. Handling of missing data From previous analyses of this high-quality clinical database, levels of missing data are anticipated to be very low. Patients missing the primary outcome or key confounders (age, location prior to admission, primary reason for admission) will therefore be excluded from the analysis. Patients missing other binary/categorical confounders/mediators will be assumed to be in the lowest risk category (i.e. no severe conditions in past medical history, no assistance with daily activities, no CPR, no delay). Following standard approaches, missing physiological parameters will be assumed to be normal and assigned zero points in the ICNARC Physiology Score (except for patients missing all physiology, who will be excluded).
2. Analysis

Flow diagram of study inclusion

Number of admissions by day/time of admission (1-hour categories)

Number of admissions by day/time of admission (14 categories used in analysis)

Characteristics of included patients by broad categories of day/time of admission:

• Age mean (SD)
• Sex n (%)
• Severe conditions in the past medical history (cardiovascular, respiratory, renal, liver, haematological malignancy, metastatic disease, immunosuppression) n (%)
• Prior dependency (no/some/total assistance with daily activities) n (%)
• Number of days from hospital admission to critical care unit admission (0, 1, 2, 3-7, >7) n (%)
• Location prior to admission to critical care (ED/not in hospital, theatre, ward/intermediate care) n (%)
• Delayed admission to critical care (no delay, up to 4 hours, >4 hours) n (%)
• Duration of delay (hours) median (IQR)
• CPR within 24 hours prior to admission to critical care (no, in-hospital, out-of-hospital) n (%)
• Primary reason for admission to critical care (body system) n (%)
• ICNARC Physiology Score mean (SD)
• APACHE II Score mean (SD)
• Critical care unit outcome n (%)
• Critical care unit length of stay stratified by outcome median (IQR)
• Acute hospital outcome n (%)
• Acute hospital length of stay stratified by outcome median (IQR)

Logistic regression modelling, reported as odds ratios with 95% confidence intervals. Three multilevel logistic regression models will be fitted on the outcome of acute hospital mortality (each including a random effect of critical care unit):

1. Unadjusted model with a single covariate of day/time of admission (8-hour categories)
2. Adjusted for the following additional covariates:

• Age (linear)
• Severe conditions in past medical history (binary)
• Prior dependency (categorical)
• Number of days from hospital admission to critical care unit admission (categorical)
• Location prior to admission (categorical)
• CPR prior to admission (categorical)
• Primary reason for admission (categorical: body system and pathological/ physiological process)
• ICNARC Physiology Score (linear)

1. Including an additional covariate of delayed admission (no delay, up to 4 hour delay, >4 hour delay) to evaluate mediation of the effect of day/time of admission

The following hypothesis tests will be conducted on each model:

• The global effect of day/time of admission (test of homogeneity)
• A comparison of weekend (Sat 08:00-Mon 07:59) versus weekday (Mon 08:00-Sat 07:59)
• A comparison of night (18:00-07:59) versus daytime (08:00-17:59)
• A comparison of weekend daytime (Sat-Sun 08:00-17:59) versus weekday daytime (Mon-Fri 08:00-17:59)
• A comparison of weekend night (Sat-Sun 18:00-07:59 the next day) versus weekday night (Mon-Fri 18:00-07:59 the next day)

Post-estimation marginal predicted mortality holding all other covariates at the values observed in the dataset (and zero random effect) will be reported by day/time of admission (14 categories)