Meacock

3

Higher mortality rates amongst emergency patients admitted to hospital at weekends reflect a lower probability of admission

Meacock. J Health Serv Res Policy, published online first May 6, 2016 doi:10.1177/1355819616649630

Clinical Question

  • Patients admitted as an emergency at the weekend have higher mortality rates compared to patients admitted on weekdays. Does the admission of fewer and/or sicker patients at the weekend compared with a weekday, account for this mortality rate difference?

Design

  • Retrospective observational study
  • Patients followed up for 30 days
  • Data collected from Hospital Episode Statistics
  • Statistical analysis
    • T-test used to compare attendances and deaths by day of week
    • Logistic regression used to estimate – 30 day mortality
    • In patients admitted via Accident and Emergency (A&E), model adjusted for: age, sex, ethnicity, diagnosis, arrival mode, first or follow-up visit, incident location, accident type, referral source, deprivation quintile, month, hospital attended
    • In patients admitted without going through A&E, model adjusted for: age, sex, ethnicity, diagnosis, co-morbidity, admission method, admission source, deprivation quintile, month, hospital
    • C statistic used to summarise the goodness of fit of the models
    • Each day compared with Wednesday

Setting

  • All 140 non-specialist acute hospital trusts in England
  • April 2013 – February 2014

Population

  • Inclusion criteria: All admissions
  • Exclusion criteria: None
  • 12,670,788 A&E attendances; 4,656,586 emergency admissions; 772 records excluded as risk-adjusted variables missing

Cases

  • Patients admitted at the Weekend

Control

  • Patients admitted on a Wednesday

Outcome

  • A&E attendances:
    • Average number of people attending A&E is highest on Monday and lowest of Friday (41,416 vs. 36,426)
    • Number of people that attend on a weekend day vs. a weekday – no significant difference (38,254 vs. 37,812)
    • 30 day Crude Mortality following A&E attendance is significantly lower in patients who present at the weekend vs. a weekday (0.99% vs. 1.03%)
    • Risk adjusted 30 day mortality
      • No significant difference in patients that attend on a weekday vs. weekend (Odds ratio 1.02, 95% C.I. 0.997-1.022)
      • Model strongly predictive of mortality – C statistic 0.92
      • Comparing individual days, patients attending on a Sunday or a Monday had a significantly increased mortality compared with a Wednesday
        • With an extra 1 in every 2941 patients dying who presented on a Sunday, and an extra 1 in 2703 patients dying who presented on a Monday
  • Admissions via A&E departments:
    • The proportion of patients attending A&E that are admitted is significantly lower in patients presented at the weekend vs. weekday (27.5% vs. 30%)
    • The crude mortality at 30 days is significantly higher in patients admitted at the weekend vs. weekday (3.59% vs. 3.42%)
    • The risk adjusted mortality was significantly higher in patients admitted at the weekend vs. weekday (Odds ratio 1.054, 95% C.I. 1.040-1.069)
    • Risk adjusted model was strongly predictive of mortality with a C-statistic of 0.91
  • Direct admissions:
    • The number of admissions on a weekday was dramatically higher than on a weekend day (3402 vs. 1317)
    • Comparing patients admitted on a weekend vs. weekday
      • Significantly more likely to be a child, young adult or age >90
      • Most common primary diagnosis was influenza, abdominal pain, intestinal infections, vs. abdominal pain, influenza and headache
      • Significantly less likely to have most of the co-morbidities listed
    • Crude mortality was significantly higher in patients admitted at the weekend vs. weekday (2.72% vs. 2.37%)
    • Risk adjusted mortality
      • Significantly higher in patients admitted at the weekend vs. weekday (Odds ratio 1.212, 95% C.I. 1.162-1.263)
      • Model to predict mortality had a C-statistic of 0.92
      • An extra 1 in every 204 patients died who were directly admitted at the weekend vs. a weekday

Screen Shot 2016-05-20 at 09.09.59

Screen Shot 2016-05-20 at 09.10.26

Authors’ Conclusions

  • There are fewer deaths following hospital admissions at weekends. Higher mortality rates are found only among the subset of patients who are admitted

Strengths

  • Large, multi-centre study
  • Appropriate statistical analysis
  • A priori specified subgroup analyses
  • Risk adjustment models with high accuracy based on C statistics
  • Very few patients excluded

Weaknesses

  • Like other papers (Freemantle et al) this paper cannot demonstrate causality- it shows associations only
  • Roughly 30% of patients were admitted, which means roughly 70% of them were discharged. With no data on out of hospital deaths in this study, this can be seen as 70% of patients being lost to follow up
  • There is no mention of controlling for readmission. Were they counted as separate attendances or as a single attendance?
  • Risk model did not directly include level of patient sickness, relying on parameters such as age, sex and coded diagnosis- as opposed to physiological or biochemical parameters such as would be included in APACHE II, NEWS, or SOFA
  • ~47% of patients attending A&E had a diagnosis “not classifiable” or “missing” This will introduce bias and limit the ability of the paper to determine if there are any differences between populations
  • A proprietary risk-adjustment model is used. While their published C-statistics are impressive, this model has only been validated by the authors. Using an internationally validated model such as APACHE or SOFA, or conversely making their model open-access, would be helpful in providing external validity

The Bottom Line

  • In a large population of UK patients, attending A&E at the weekend is not associated with a significant difference in the risk adjusted 30 day in-hospital mortality compared with attending during the week.
  • Being admitted to hospital at the weekend is associated with an increased risk adjusted in-hospital mortality compared with being admitted on a weekday. This association was more pronounced in patients who were admitted directly to the hospital rather than via A&E
  • There are differences in the age, diagnoses and co-morbidities of patients who are directly admitted at the weekend compared with those who are admitted during the week. These differences are likely to result in differences in outcomes for patients admitted at the weekend vs. a weekday. It is unclear how much of the difference in outcomes reported is due to the difference in populations that present at a weekend compared with a weekday and how much is due to other factors

External Links

Metadata

Summary author: Segun Olusanya
Summary date: 12th May 2016
Peer-review editor: Dave Slessor and Steve Mathieu

2 comments

  • Malcolm WG Gordon

    It doesn’t surprise me that mortality rate is higher at weekends especially if not admitted via ED because most EDs have recognised the need for a consultant presence in ED at the weekend which is not true of all wards where patients are admitted to.

  • Pingback: LITFL Review 232 – FOAM Ed

Leave a Reply