Public Health Wales / The weekend effect
The weekend effect:a rapid review of the literature
Author:Dr Mary Webb, Public Health Specialist
Date:5 July2011 / Version:0b
Publication/ Distribution:
  • Public Health Wales
  • Unscheduled Care Programme Board
  • 1,000+ Lives Campaign

Review Date:N/A
Purpose and Summary of Document:Admission to hospital at weekends is reported to be associated with poor outcomes (usually excess mortality) and is often termed the weekend (W/E)effect. The evidence from the literature on outcomes fromW/E admissions was mainly observational and contradictory. Most studies on emergency admissions demonstrated a W/E effect, but excess mortality was dependent on the clinical condition studied.No convincing evidence was found to demonstrate improved outcomes at a weekend. Staffing issues, communication/handover problems and delays in diagnostic and therapeutic procedures have been proposed to explain the W/E effect. There was a lack of evidence on effective interventions to combat the poor outcomes with weekend admissions. The evidence is therefore suggestive of a W/E effect, but there is clearly a need for well planned studies to demonstrate whereit exists. Work is also required on methods known to improve patient safety that may combat such an effect.The expertise on patient safety methods within the 1000 Lives Plus programme in Wales should be utilised.
Work Plan reference: HS48

Contents

Page
Key messages / 3
1. Introduction / 4
2.Aim / 4
3. Methods / 4
4. RESULTS / 5
4.1 Emergency admissions
4.2 Condition specific
4.3 Interventions to ameliorate the weekend effect / 5
7
10
5. references / 12
Table1
Table 2
Table 3 / 18
30
32
Appendix 1 / 36

2011 Public Health Wales NHS Trust.

Material contained in this document may be reproduced without prior permission provided it is done so accurately and is not used in a misleading context.

Acknowledgement to Public Health Wales NHS Trust to be stated

Key messages

  • It has been reported for several decades that admission to hospital at weekends is associated with poor patient outcomes and is often referred to as the weekend (W/E) effect.
  • Increased mortality has been reported for W/E admissions for specific medical conditions such as myocardial infarction, stroke and upper gastrointestinal haemorrhage.
  • The evidence for the W/E effect is mainly observational and not all research studies have reported poorer outcomes at the W/E.
  • The rapid review confirmed the lack of good quality Level 1 and Level 2 evidence and the inconsistency for a weekend effect for hospital admissions and treatment. Problems interpreting the evidence were found due to the heterogeneity of the hospital design and healthcare systemsin different countries.There is clearly a need for well planned international studies to elucidate the causative factors of the weekend effect.
  • Where the studies concerned all emergency admissions, the W/E effect was usually dependent on specific clinical conditions.
  • The clinical conditions with the most consistent evidence were births, stroke, upper gastrointestinal haemorrhage, renal disease and ST elevated myocardial infarction. The evidence came however from a small number of papers and was mainly observational in type i.e. Level 3/4, or where Level 2, the potential for significant bias was present.
  • Factors that have been proposed to explain the W/E effect are staffing issues, communication/handover problems and delays in diagnostic and therapeutic procedures.
  • There was a lack of evidence on the effectiveness of interventions to improve W/E outcomes through addressing the suggested associated factors. There were observational studies on the use of methods such as early warning systems, tools to aid communication and facilitate handovers, but none had been used to study directly the W/E effect.
  • Expert opinion has stressed the need for well planned trials to obtain good quality evidenceon the effectiveness of interventions to improve patient outcomes at weekends.

  1. Introduction

Admission to hospital at weekends (W/Es) has been associated with adverse outcomes and is often referred to as the “weekend effect”. Increased mortalityhas been reported for weekend (W/E) admissions with specific medical conditions such as myocardial infarction,[1]heart failure,[2] upper gastrointestinal bleeding,[3][4]pulmonary embolism,[5]stroke[6] and intracerebral haemorrhage. [7]

Worse outcomes have also been reported to be more common for ‘night-time’ (6pm–8am) emergency medical admissions in a UK pilot study of quality of care issues[8]and for admissions to someintensive care units.[9] This difference in outcomes may be due to variations in case mix of the patients presenting to hospitals at different time periods, variations in processes of care on W/Es/out of hours, such as decreased levels of staffing, availability of experienced staff and restricted availability of tests and procedures.[10]

Not all research studies have reported a W/E effect. A single-centre UK study of six specific medical conditions failed to demonstrate this effect[11] and other studies have also reported a lack of W/E effect. [12][13]There is therefore a need to review the evidence for the W/E effect on outcomes for emergency admissions and to investigate reports of interventions aimed at reducing the adverse effects of W/E admissions.

  1. Aims

Rapid review of the evidence on mortality at the W/E for hospital emergency admissions, the so called “weekend effect”. Information was also sought on interventions that have been implemented to try and improve the outcomes of W/E admissions.

  1. Methods

As per the protocol contained in the Public Health Wales Guide to Searching [14]a scoping search was initially performed to identify major papers on published evidence and refine the final search strategy. For the present overview, search terms contained in the search strategies were used from published reviews and they were kept broad to maximise retrieval of references. The type of literature on the “weekend effect”necessitated the use of a pragmatic approach to searching for evidence and it should be emphasised that the review is not a systematic review of primary studies.

For critical appraisal, the tables recommended for use in the National Institute for Health and Clinical Excellence Guideline Development Methods manual[15]were modified to accept the type of studies identified forthe “weekend effect”. The quality of the evidence was graded using the NICE hierarchy of evidence and the quality checklists. Evidence was rejected if graded as poor quality, apart from where it was of Level 1 type (see Appendix 1 for explanation of evidence grading system) and was highly relevant to the questions. Evidence levels are given after each reference in the reference section.

Due to practical limitations a single reviewer performed the final selection, critical appraisal and data extraction. Every effort was made to minimise reviewer bias.

Inclusion Criteria

Search period January 2000 – February 2010

Papers in English, German, French or Spanish

Papers relating to the W/E effect

Randomised controlled trials

Systematic reviews

Meta-analyses

Guidelines

Observational studies (where higher quality evidence was not available)

  1. Results

There were a large number of references revealed by the search. Full details of the search strategies and results are available from the author. Not surprisingly, considering the topic area there was a lack of Level 1 RCT evidence. An annotated bibliography from 1978 to 2010 from the Interdisciplinary Nursing Quality Research Institute provided a useful source of references.[16]

The results are split into evidence related to outcomes for all emergency admissions,specific specialty/condition admissions and interventions used to combat the W/E effect. Where relevant, papers dealing with “out of hours” treatment of patients were also included as factors associated with the W/E effect are also present at such times.

4.1Emergency admissions

Table 1 indicates the included major references and whether a W/E effect was demonstrated. Of the 12 studies included,10 studies [17][18][19][20][21][22][23][24][25][26]indicated an increase in mortality at W/Es, but in most of these studies excess mortality was dependent on the clinical condition. In 2 studies a W/E effect was not seen. [27][28]

Statistical information from the Healthcare Cost and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS) in the United States (US) indicated that of the 39.5 million community hospital stays in 2007, 19% (7.7 million) stays began on a W/E.On the day of admission, W/E-admitted patients received only 36% of major procedures that they would receive during their stays, compared with 65% for patients admitted on weekdays. On the first day after admission, 64% of W/E-admitted patients with heart attacks received major cardiac procedures, compared with 76% for weekday-admitted patients. Similarly, 44% of W/E-admitted patients with gastrointestinal (GI) bleeds received endoscopy, compared with 58% of weekday-admitted patients. See Table 2 for results comparing the results for timing of procedures at W/Es vs weekdays) Patient characteristics such asaverage age, gender and median household income, were comparable for W/E and weekday hospital admissions. The distribution of discharges across hospital characteristics such as bed size, teaching status, urban/rural location, and region were also comparable for W/Evs weekday admissions. 20

A recently published study of routinely collected hospital data from the United Kingdom (UK) contained results on W/E mortality for 4,317,866 emergency admissions. Area level socioeconomic deprivation scores were assigned to each patient using their postcode of residence and a Charlston comorbidity score was also used. An overall crude mortality rate of 5.0% (5.2% for all W/E admissions and 4.9% for all weekday admissions) was demonstrated. The overall adjusted odds of death for all emergency admissions was 10% higher (odds ratio [OR] 1.10, 95% confidence interval [CI] 1.08-1.11) in those patients admitted at the W/E compared with patients admitted during a weekday (p<0.001). 18

The paper also split the results into condition specific mortality. A total of 28 clinical groups had significantly higher odds of mortality at the W/E when compared with the weekday (p<0.05); 17 were significant at p<0.001. Three conditions had significantly lower odds of mortality at the W/E with p<0.05, but no conditions were significant at p<0.001. For the 32 medical groups, 8 had significantly (p<0.001) higher odds of death for patients admitted at the W/E. Of the 7 surgical groups, one had significantly (p<0.001) higher odds of death for patients admitted at the W/E. The overall odds of death for all emergency admissions, adjusted by the Clinical

Classification System (CCS) diagnostic group, age, sex, deprivation quintile and comorbidity, was 10% higher (OR 1.10, 95% CI 1.08 to 1.11) in those patients admitted at the W/E compared with patients admitted during a weekday (p<0.001). The authors estimated a possible excess of 3369 deaths (95% CI 2.921 to 3.820) occurring at the W/E for 2005/2006, which is equivalent to a 7% higher risk of death. (Table 3)

A study in Scotland of acute medical admissions reported that there was no increase in mortality for patients admitted at W/Es. 27This study was small and therefore limited in power to detect mortality differences. Another UK study that did not demonstrate a W/E effect found that total mortality was increased for Monday admissions, at night and out of hours. Factors such as reduced staffing levels, less senior and experienced staff and hand over procedures also apply at night and out of hours. The authors discuss the possibility that the W/E effect is due to both patient and pre-hospital factors i.e. how ill the patients are before arriving in hospital. The major limitations of the study are the lack of adjustment for comorbidities and the small numbers of admissions.28

4.2Condition specific

As mentioned in Section 4.1, mortality at W/Es is only present for certain medical conditions. Section 4.2 describes the evidence for a W/E effect for several medical conditions and specialties.

Intensive care admissions

Twelve papers were included and in 6/12 a W/E effect for mortality was demonstrated. [29][30][31][32][33][34]and in the other 6 no effect was seen.9 1213[35][36][37]The problem with the evidence for intensive care admissions (ICU) was that the studies were very heterogeneous and the structure and staffing of the units differed, as did the correction for confounding factors.

Paediatric ICU

A retrospective cohort study of the morbidity records of more than 1 million birthsinScotlandreported that there was an increased risk of neonatal death at term with out of hours deliveries. [38]The other 3 included studies did not find a W/E effect.[39][40][41] Again the problem with drawing conclusions is the heterogeneity of the structure of the units.

Neonatal/births

Four out of 5 studies demonstrated a W/E effect, but the effect size varied. [42][43][44][45] A large study of over 1 million live births did not find an effect. [46]The authors of one study indicated that the likelihood of delivering at the W/E increases with certain socio-demographic factors. Researchers have discussed how women receiving prenatal care often plan their deliveries for a weekday and more of those who do not receive prenatal care deliver at W/Es.44

Traumaand orthopaedics

Interpretation of the evidence is complicated by the existence of trauma units in some countries, particularly in the US. Two/5 papers found a W/E effect for trauma admissions. [47][48] Of the three papers in which a W/E effect was not demonstrated,two were from hospitals with highly organised trauma units. [49][50] After subset analysis for confounding factors, the authors of 1 large trauma series did not find a W/E effect. [51]The survey by the National Audit Office in England confirmed the W/E effect and the authors considered that the lack of 24/7 dedicated trauma consultants contributes to this effect. In 2010, only 1 of the hospitals surveyed had a 24/7 trauma consultant. 47

Stroke

Eight studies were included for stroke and 7/8indicated a W/E effect even after correction for confounding factors such as case mix.[52][53][54][55][56][57][58]A

retrospective case series from the US did not find a W/E effect for patients with acute ischaemic stroke (AIS) and found that AIS patients were more likely to receive tissue plasminogen activator on W/Es than on weekdays. [59]

Respiratory/pulmonary embolism

Two papers were included. One paper dealt with all respiratory deaths using data supplied by the Office of National Statistics for all hospitals in the South East of England for 1989 to 2001. The results indicated that mortality was not increased within the 2 day period from the W/E.[60] Another paper studied W/E versus weekday admission and mortality after acute pulmonary embolism (PE) and found that patients with PE admitted at the W/E have higher short term mortality than weekday admissions. 5

Upper gastrointestinal bleeding

Sevenstudies were included, 5/7 demonstrated a W/E effect for gastrointestinal haemorrhage.34[61][62]64 A study published in 2011from Wales reported on a record linkage of hospital in-patient and mortality data between 1999 and 2007. Of 24,421 hospitalisations for upper gastrointestinal bleeding, case fatality was 13% higher for W/E and 41% higher on public holidays. The higher death rates could not be explained by case mix and the authors suggested that they could be associated with reduced staffing levels or delays in investigative procedures such as endoscopy. The authors found little variation in mortality with social deprivation, hospital size or distance from hospital. The evidence on delays to endoscopy as a factor in causing poor survival at W/E was however inconsistent.61 Two studies after adjusting for timing of endoscopy found that W/E admission remained an independent predictor of increased mortality.462 Expert opinion suggests that the optimal timing of endoscopy

requires further investigation, but that the available evidence indicates that early endoscopy (≤24h) results in shorter and more economical hospital stays. [63] A study of outcomes for oesophageal variceal haemorrhage and peptic ulcer did not demonstrate a W/E effect. [64][65]

Renal disease

Canadian data demonstrated that patients with renal failure have a 34% higher risk of adjusted in–hospital mortality when admitted on a W/E compared with a weekday. [66]A later large scale study (963,730 admissions

with a diagnosis of acute kidney injury (AKI) between 2003 and 2006)looked at W/E deaths in patients with acute kidney disease. 214,962 admissions (22%) designated AKI asthe primary reason for admission (45,203 on a W/E and 169,759 on a weekday). Compared with admission on a weekday, patients admitted with a primary diagnosisof AKI on a W/E had a higher odds of death [adjusted OR 1.07, 95% CI 1.02 to 1.12]. The risk for death with admission on a W/E for AKI was more pronounced insmaller hospitals (adjusted OR 1.17, 95% CI 1.03 to 1.33) compared with larger hospitals (adjusted OR 1.07, 95% CI 1.01 to 1.13). 10

Patients with end stage renal disease (ESRD) who are admitted at the W/E were also shown to be at increased risk of death in the preliminary results of a study of 800,000 admissions for ESRD; 20% were admitted over the W/E. With adjustment for confounding factors, such as comorbidities and hospital characteristics patients admitted on W/E were 17% more likely to die compared with patients admitted during the week. Patients admitted at W/E also experienced delays in starting dialysis.[67]

Neurological

Two papers were included. No relationship was found in a retrospective cohort study between short term mortality risk and W/E admission in patients hospitalised for subarachnoid haemorrhage. [68]A paper from the same authors found a 12% increase in W/E mortality in patients admitted with intracerebral haemorrhage (ICH),compared with admissions during the rest of the week.[69] This is in contrast with the lack of W/E effect for ICH found by Clarke et al.19

Cardiology

Again the results for increased W/E mortality were inconsistent. For acute myocardial infarction (AMI) 2/4 papers demonstrated an increased mortality1 [70] and 2/4 did not find an association with time of admission and mortality.[71][72]A W/E effect was found in 2/2 papers reporting on ST elevated myocardial infarction (STEMI).[73][74] For heart failure 1/2 papers found a W/E effect. 2 The paper on heart failure that did not demonstrate a

W/E effect concerned patients entered into a dedicated programme to improve the treatment of patients with heart failure; no follow up data was available after the programme was completed. [75]

4.3Interventionsto ameliorate the weekend effect

Despite the extensive literature demonstrating poor outcomes for admissions to hospital at the W/E and various suggestions as to factors associated with this W/E effect, there was a lack of literature on interventions to improveW/E outcomes. Most of the research has involved large data sets that illustrate the decrease in survival rates, but not what causes it. Nurse researchers in the US have explored the risks to off peak admissions. The results of a focus group study of nurses indicated that there was much less direct supervision and problems obtaining physician backup for emergencies at W/Es. The nurses made no real distinction between night shift and W/E environments. [76]