Using care bundles to reduce in-hospital mortality: quantitative surveyBMJ 2010; 340 doi: https://doi.org/10.1136/bmj.c1234 (Published 01 April 2010) Cite this as: BMJ 2010;340:c1234
- Elizabeth Robb, director of nursing1, visiting professor 2,
- Brian Jarman, emeritus professor3, part-time senior fellow 4,
- Ganesh Suntharalingam, clinical director and consultant in intensive care medicine1,
- Clare Higgens, clinical director, director of medical education, and consultant physician 1,
- Rachel Tennant, consultant physician1,
- Karen Elcock, director of practice and workbased learning 5
- 1North West London Hospitals NHS Trust, Harrow HA1 3UJ
- 2Nursing, London South Bank University, London SE1 0AA
- 3Imperial College, London EC1A 9LA
- 4Institute for Healthcare Improvement, Cambridge, MA 02138, USA
- 5Thames Valley University, Slough SL1 1YG
- Correspondence to: B Jarman
- Accepted 8 February 2010
Problem To reduce hospital inpatient mortality and thus increase public confidence in the quality of patient care in an urban acute hospital trust after adverse media coverage.
Design Eight care bundles of treatments known to be effective in reducing in-hospital mortality were used in the intervention year; adjusted mortality (from hospital episode statistics) was compared to the preceding year for the 13 diagnoses targeted by the intervention care bundles, 43 non-targeted diagnoses, and overall mortality for the 56 hospital standardised mortality ratio (HSMR) diagnoses covering 80% of hospital deaths.
Setting Acute hospital trust in north west London.
Strategies for change Use of clinical guidelines in care bundles in eight targeted clinical areas.
Interventions Use of care bundles in treatment areas for the diagnoses leading to most deaths in the trust in 2006-7.
Key measures for improvement Change in adjusted mortality in targeted and non-targeted diagnostic groups; hospital standardised mortality ratio (HSMR) during the intervention year compared with the preceding year.
Effect of the change The standardised mortality ratio (SMR) of the targeted diagnoses and the HSMR both showed significant reductions, and the non-targeted diagnoses showed a slight reduction. Cumulative sum charts showed significant reductions of SMRs in 11 of the 13 diagnoses targeted in the year of the quality improvements, compared with the preceding year The HSMR of the trust fell from 89.6 in 2006-7 to 71.1 in 2007-8 to become the lowest among acute trusts in England. 255 fewer deaths occurred in the trust (174 of these in the targeted diagnoses) in 2007-8 for the HSMR diagnoses than if the 2006-7 HSMR had been applicable. From 2006-7 to 2007-8 there was a 5.7% increase in admissions, 7.9% increase in expected deaths, and 14.5% decrease in actual deaths.
Lessons learnt Implementing care bundles can lead to reductions in death rates in the clinical diagnostic areas targeted and in the overall hospital mortality rate.
About 7% of hospital admissions are associated with adverse events, of which about 8% result in death, and about half of the adverse events are avoidable.1 2 The UK Bristol Royal Infirmary Inquiry, the US Institute of Medicine (IOM) reports, and the UK Department of Health review High Quality Care for All have all emphasised the importance of objective measures of clinical outcomes and healthcare quality improvement.3 4 5 6 The hospital standardised mortality ratio (HSMR), a comparative measure of a hospital’s overall mortality, focuses on a group of diagnoses that account for 80% of all deaths in hospitals nationally7; it provides a tool for analysing hospital outcomes over time.8 9 A figure of 100 is the national reference value and hospitals with higher or lower adjusted mortality have values above or below 100.
The North West London Hospitals NHS Trust serves a population of about 500 000. Within it, Northwick Park Hospital site had about 62% of admissions and 65% of deaths in 2007-08; Central Middlesex Hospital had 22% and 32%; and St Mark’s Hospital had 16% and 3%. The trust received adverse media coverage in 2005 and 2006 following the Healthcare Commission’s report on its maternity services in July 200510 and in March and April 200611 regarding a drug trial in which six men developed organ failure. Both stories had an impact on staff morale and on patients’ perceptions of Northwick Park Hospital, even though the drug trial took place in an independent research unit. The HSMR for the trust for the financial year (April to March) 2006-7 was 84.6, 15.4 points below the national value using the England death rates in 2006-7 as the baseline for comparison, and was 89.6 (84.5 to 95.0) for 2006-7 using the England death rates in 2007-8 as the baseline. These were lower than the national rate but were higher than the best performing trust in the north west sector of London (72.2 in 2006-7 when using the 2006-7 baseline, or 73.1 if compared with the best non-specialist acute trust.)12 13
In 2006, one of the authors (ER) studied the “100,000 Lives” campaign run by the Institute for Healthcare Improvement in Cambridge, Massachusetts,14 15 as part of her scholarship from the Florence Nightingale Foundation. In collaboration with key clinical colleagues we agreed to initiate a similar scheme aimed at reducing mortality rates and thus increasing public confidence in the quality of patient care.
The aim was to reduce mortality in the year April 2007 to March 2008, compared with 2006-7. In 2006-7 there were 1142 deaths in North West London Hospitals NHS Trust within the diagnoses used to calculate the HSMR. A reduction of the trust’s HSMR from 89.7 to 73.1, the value for the best non-specialist acute trust without any change in the number of admissions or their case mix, would have implied a reduction of approximately 210 deaths, but it was considered reasonable to attempt the goal of 110 reduction in 2007-8, equivalent to an HSMR of 81.0. The reduction in mortality was the only improvement that the assessment was designed to evaluate.
We decided to target the diagnoses responsible for the largest number of deaths for which evidence based methods to reduce death rates were available. Two of these areas (central line infections and ventilator associated pneumonia) are covered by the Institute for Healthcare Improvement’s “care bundles.”16 Care bundles are checklists of accepted clinical guidelines printed on forms that are made conveniently available to all clinicians in clinical areas. The institute defines a care bundle as “a collection of processes needed to effectively and safely care for patients undergoing particular treatments with inherent risks. Several interventions are ‘bundled’ together and, when combined, significantly improve patient care outcomes.” They provide an easily available reminder of elements of care vital to the improvement of clinical outcomes. Compliance with all elements is critical to optimum treatment.
For the six other treatment areas, the clinical elements of the bundles were agreed by consensus with clinicians, based on UK clinical guidelines (eGuidelines.co.uk), developed by the National Institute for Health and Clinical Excellence and the Department of Health,17 18 and US guidelines developed by the Agency for Healthcare Research and Quality and the Centers for Disease Control and Prevention.19 20 Two of the trust’s clinical directors (GS and CH) designed the format of the bundles. Each bundle had a medical notes component, a flag to mark the patients involved, and a tracer backing form. Each bundle was printed on a single sheet of paper, with a peel-off sticker that was stuck in the patient’s medical notes on the day treatment was started. On the front of the patient’s case notes, stickers of different colours for each bundle indicated that a care bundle had been started (see web figure 1 for an example). Clinicians were trained and encouraged to use the bundles and write on the backing sheets the date and time the bundle was started, and patient’s details.
The intervention was the use of the easily available clinical guidelines given in the care bundles in each of the eight targeted clinical areas. The main expenditure was the small cost of printing these checklist sheets (about £300 for each batch of 100 sheets).
We considered that the proposed quality improvement work was an extension of normal clinical care, part of providers’ professional responsibility, rather than “research”; this meant that an independent review board review was neither necessary nor appropriate. However, the study was approved by the trust’s clinical audit and effectiveness committee, part of the normal clinical governance structure, endorsed by executive team and agreed with the medical director and clinical directors. As an extension of normal clinical care, use of the care bundles did not lead to additional concerns for patients’ privacy or protection of physical wellbeing.
The top 25 diagnoses in the Clinical Classification System,21 used by Imperial College and Dr Foster, that in 2006-7 led to the largest number of deaths in the trust (web table 1) were targeted using the eight care bundles (box 1). The 13 diagnoses that were monitored as best covering the eight care bundles—the targeted diagnoses—are shown in box 2.
Box 1 Care bundles used to reduce in-hospital mortality
Central venous catheter/line asepsis
Diarrhoea and vomiting
Ventilator acquired pneumonia
Meticillin resistant Staphylococcus aureus infections
Surgical site infections
Chronic obstructive pulmonary disease
Box 2 Targeted diagnoses from Clinical Classification System21
Peritonitis and intestinal abscess
Senility and organic mental disorders
Pleurisy pneumothorax pulmonary collapse
Aspiration pneumonitis food/vomitus
Skin and subcutaneous tissue infections
Urinary tract infections
Acute cerebrovascular disease
Other gastrointestinal disorders
Septicaemia (except in labour)
Chronic obstructive pulmonary disease and bronchiectasis
Congestive heart failure non-hypertensive
The quality improvement project formally started on 1 April 2007. The quantitative measure used for evaluation was the adjusted mortality for each of the 13 targeted clinical diagnostic areas, as well as the change in the trust’s overall HSMR. Adjusted mortality was determined from monthly mortality analyses of routinely recorded hospital episode statistics data by Imperial College and Dr Foster Intelligence.22 23 24 25 These covered all admissions to the trust and had been used since 1 August 2005 at the individual patient and aggregated levels; the analyses were available on the web on a monthly basis. Questions regarding the confidentiality of patient identifiable data were supervised by the Patient Information Advisory Group.26
Cumulative sum (cusum) charts were used to track the changes of mortality for each diagnosis. The probability of death is calculated for each patient admitted using the national death rates by age group, sex, elective or emergency admission, pre-existing comorbidities, subdiagnosis or procedure, socioeconomic group, number of previous admissions, palliative care, year of discharge, and month of admission (for respiratory conditions). A function (log odds ratio) of the weighted difference between the actual deaths and the expected deaths over a series of patients was plotted cumulatively on the cusum charts.27 28 Once the chart reaches a preset level (we mostly used the 99% control limit for this evaluation) an alert is registered, indicating double, or half, the odds of death compared with the national death rate. We used death rates for England for the year 2007-8 as the reference baseline for all calculations of expected deaths.
About 1200 care bundle sheets were used, 1000 at Northwick Park Hospital and 200 at the other two hospital sites—a higher usage rate, as a proportion of admissions, at Northwick Park than at the other hospitals.
An assessment of the quality of the hospital episode statistics records found that for April 2007 to March 2008, 0.026% of records had missing or unusable values of date of birth; 0.019% of sex, and 0.207% of primary diagnosis (50, 37, and 405, respectively of 195 988 records).
The 95% confidence intervals29 were calculated using Byar’s approximation for the 56 standardised mortality ratios (SMRs) that constitute the HSMR and for the targeted and non-targeted diagnostic groups (see web table 2) for 2006-7 and 2007-8. The SMR for the targeted diagnostic groups and the HSMR showed a significant reduction, but the SMR of the non-targeted group was not significantly reduced (fig 1⇓).
Figure 2⇓ compares the trust’s HSMRs from 1996-7 to 2008-9 with HSMRs for England (see web figure 3 for a comparison using running year comparator). Figure 3⇓, a funnel plot of HSMRs of all acute hospital trusts in England in 2007-8, also shows that the trust’s HSMR reduced from 89.6 in 2006-7 to 71.1 in 2007-8, the lowest in England.
We calculated (web table 3) the number of deaths that would have occurred in 2007-8 if the relevant 2006-7 mortality ratio had applied, and compared this with observed deaths in 2007-8. There were 83 fewer deaths in the non-targeted diagnoses and 174 for the targeted diagnoses (255 for the HSMR).
In 2007-8 the crude death rate for the non-targeted diagnoses dropped to 2.7% from 3.4% in 2006-7, to 5.7% from 6.9% for the targeted diagnoses, to 3.7% and from 4.6% for all diagnoses. This is a rough indication of changes in mortality because no adjustment was made for any change in patients’ characteristics between the comparison years.
Figure 4⇓ shows the change from April 2006 to March 2008 of the HSMR cusum statistic (log of the odds ratio) with time. The cusum charts for the individual targeted diagnoses (web figures 2a to 2m) show that the charts crossed the 99% alert level for seven of the 13 targeted diagnoses, and for the HSMR (web figure 4) indicating a halving of the risk of death compared with the national value. For several diagnoses it crossed more than once, the maximum being five times for the diagnosis acute cerebrovascular disease. For a further four diagnoses the cusum statistic line crossed the 95% alert level (not shown.
Figure 5⇓ shows the HSMR at the three hospital sites that form the North West London Hospitals trust (see web table 4). HSMR was significantly reduced at Northwick Park Hospital, the hospital with the highest usage of care bundles, but not at the other two hospitals.
This paper reports the results of implementing eight care bundles (guidelines for effective and safe patient care) in the North West London Hospitals NHS Trust during the financial year April 2007 to March 2008, targeted on reducing mortality in several diagnostic areas with high numbers of deaths. For the 13 Clinical Classification System diagnoses that most closely covered the eight care bundles used, and also for all 56 diagnoses used to calculate the overall hospital standardised mortality rate, the adjusted death rate was significantly lower in 2007-8 than in 2006-7. In the intervention period 174 fewer deaths occurred in the targeted diagnoses, and 255 fewer deaths in all diagnoses, than if the 2006-7 adjusted death rates applied in 2007-8.
The trust’s HSMR fell from 89.6 in 2006-7 to 71.1 in 2007-8 to become the lowest of the English acute trusts that year (fig 3⇑). This reduction occurred during the year of intervention targeted at selected high risk patients: it became significant (at the 99% control limit) about one month after the start of the quality improvement initiative. HSMRs have been reducing in England since 2001-2, and from 2004-5 to 2007-8 the reduction in our trust (45.7) was greater than the England reduction (19.2) and proportionately greater during 2007-8, when the reduction for England was 6.2 and for our trust was 18.7 (fig 2⇑).
Northwick Park Hospital used five sixths of the total care bundles used in the trust. It was the only one of the three hospitals in the trust with a significant reduction of HSMR from 2006-7 to 2007-8.
A limitation of this type of study is the difficulty establishing whether there is a causal relation between introducing the targeted care bundles and the reduced mortality. A randomised trial design was not practical in this clinical situation. However, the significant reduction in mortality occurred only at the site where the care bundles were predominantly used, rather than at the other two hospitals in the trust; it occurred in the year in which the care bundles were introduced, starting in the month of introduction, and not in the preceding year, during which the mortality for the targeted diagnoses did not change significantly; and it occurred for the targeted diagnoses and not for the not-targeted diagnoses—indicating that the introduction of the care bundles is among possible causes of the reduction in death rates.
Other explanations for the mortality reduction include the influence of a new acute assessment unit employing four extra consultants doctors at the Northwick Park site and a new medical rota providing better continuity of care at this site introduced in December 2007. This revised medical rota may have resulted in a greater uptake of the care bundles, but it began eight months after the start of the intervention year, and the significant reduction of mortality first occurred about one month from the start date. This service only covered the assessment unit and not the whole hospital. A new chief executive of the trust was appointed in April 2007, but her responsibilities extended to all three sites. There could have been a Hawthorne effect related to the participants in the project knowing that they were being monitored.29 There could also have been a contamination effect because the clinicians would have treated patients with both targeted and non-targeted diagnoses and their knowledge of the guidelines in the care bundles may have influenced their management of the non-targeted diagnoses, but this would have reduced the difference between the two groups.
Two earlier UK studies using HSMRs, both with a wide range of initiatives designed to reduce hospital death rates, have reported similar mortality reductions to ours.8 9 In the United States the “100,000 Lives” campaign recruited more than 3000 hospitals to a mortality reduction project that made some use of care bundles and HSMRs.30 There have been similar initiatives in Canada31 32 33 and Wales.34 More focused studies have included action to reduce death rates from fractured neck of femur35 (based on evidence of the effect of reducing the time from admission to operation36) and introducing surgical check lists37; both reduced inpatient deaths.
The methods described in this study, of determining the main causes of death in a hospital trust and then using methods of treatment (care bundles) developed from evidence of their effectiveness, to reduce mortality in the targeted diagnostic areas could be generally applicable. Dr Foster Intelligence’s monthly analyses of adjusted death rates are routinely used by about 67% of acute hospitals in England and make monitoring changes in mortality relatively simple. These methods could be used to monitor whether mortality reduction continues over time and whether other care bundles are effective.
Cite this as: BMJ 2010;340:c1234
We thank the many excellent clinicians at North West London Hospitals NHS Trust for their collaboration. Derek Bell was external adviser in the early stages of developing the acute assessment unit and supported the care bundle concept and Theresa Murphy was the project leader. Dr Foster Intelligence provided access to adjusted death rates data and the Institute for Healthcare Improvement gave advice on the use of care bundlesFrank Davidoff, Paul Batalden, Paul Aylin, Alex Bottle, Gareth Parry, Joanne Zaborowski, and Steve Middleton read and commented on the paper. We found the SQUIRE guidelines (www.squire-statement.org) particularly helpful.
Contributors: ER provided much of the background information about the project. ER, GS, CH, RT, and KE planned, conducted, and reported the work. BJ wrote the paper and did the analyses. All authors read and commented on the manuscript. BJ is guarantor.
Funding: No separate funding.
Competing interests: ER, GS, CH, and RT work at the North West London Hospitals NHS Trust. BJ is director of the Imperial College Dr Foster unit, advises Dr Foster Intelligence, and is part time senior fellow at the Institute for Healthcare Improvement.
Provenance and peer review: Not commissioned; externally peer reviewed.