Implementing a Simple Care Bundle Is Associated With Improved Outcomes in a National Cohort of Patients With Ischemic Stroke
Background and Purpose—Further research is needed to better identify the methods of evaluating processes and outcomes of stroke care. We investigated whether achieving 4 evidence-based components of a care bundle in a Scotland-wide population with ischemic stroke is associated with 30-day and 6-month outcomes.
Methods—Using national datasets, we looked at the effect of 4 standards (stroke unit entry on calendar day of admission [day 0] or day following [day 1], aspirin on day 0 or day 1, scan on day 0, and swallow screen recorded on day 0) on mortality and discharge to usual residence, at 30 days and 6 months. Data were corrected for the validated 6 simple variables, admission year, and hospital-level random effects.
Results—A total of 36 055 patients were included. Achieving stroke unit admission, swallow screen, and aspirin standards were associated with reduced 30-day mortality (adjusted odds ratio [95% confidence interval]: 0.82 [0.75–0.90], 0.88 [0.77–0.99], and 0.39 [0.35–0.43], respectively). Thirty-day all-cause mortality was higher when fewer standards were achieved, from 0 versus 4 (adjusted odds ratio [95% confidence interval], 2.95 [1.91–4.55]) to 3 versus 4 (adjusted odds ratio [95% confidence interval], 1.21 [1.09–1.34]). This effect persisted at 6 months. When less than the full care bundle was achieved, discharge to usual residence was less likely at 6 months (3 versus 4 standards; adjusted odds ratio [95% confidence interval], 0.91 [0.85–0.98]).
Conclusions—Achieving a care bundle for ischemic stroke is associated with reduced mortality at 30 days and 6 months and increased likelihood of discharge to usual residence at 6 months.
Healthcare bundles are a set of evidence-based practices (generally 3–5), which aim to help healthcare providers improve patient care and clinical outcomes.1 Tying these practices together into a bundle with audit of their implementation increases the likelihood that these interventions will be applied in a more consistent manner across and within different hospitals. Some studies suggest that publication of hospital performances toward achieving process standards may improve future patient care,2 although there are significant challenges in comparing practice in different settings. There is also understandable concerns about publishing data which are not robustly corrected for case mix.3
Although specialist stroke unit care is now part of most national guidelines for acute stroke management, further work to dissect out which specific elements of the care process improve outcomes, such as mortality or return home, is necessary. The Scottish Intercollegiate Guidelines Network produced evidence-based guidelines for stroke care in 1997 which in 2004 were incorporated into standards recommended for each patient with stroke admitted to hospital in Scotland. Guidelines were updated in 20084 and standards in 2009 (Appendix I for changes to standards in the online-only Data Supplement) and include admission to a stroke unit on calendar day of admission (day 0) or day following (day 1), swallow screen on calendar day 0, aspirin on calendar day 0 or day 1, and brain imaging on calendar day 0.5 These 4 standards are monitored nationally by the Scottish Stroke Care Audit. Cases are ascertained and data extracted from case notes locally in each hospital by trained audit staff and entered in a web-based database held centrally by Information Services Division. The Scottish Stroke Care Audit feeds back data to each hospital monthly and publishes an annual report on stroke care in each acute hospital in Scotland.6
Direct comparison of stroke care between institutions is complex because of the variability in data collection, stroke severity, comorbidities, and other variables which affect process and even more so outcomes.7 Measures of process may be valuable and reproducible, but they will only be clinically relevant if they translate into improvements in clinical outcomes. In addition, the factors that influence mortality may be different from those that influence functional recovery, and thus well-defined outcomes need to be evaluated. To increase the validity of data, case-mix adjustment is necessary. A recent scientific statement from the American Stroke Association highlights that further research is needed to better identify methods and metrics to evaluate outcomes of stroke care.3
The purpose of this study was to investigate whether the level of compliance with these 4 evidence-based components of a care bundle in a Scotland-wide population admitted with ischemic stroke was associated with better outcomes at 30 days and 6 months after admission, correcting for known outcome predictors using the six simple variable (SSV) model.8
We obtained data from Information Services Division of National Health Service National Services Scotland and the General Register Office for Scotland.
Information was obtained for all patients with stroke admitted between January 1, 2005, to September 15, 2011, at all 36 acute hospitals in Scotland.
The General Register Office records information relating to all deaths in Scotland. A unique patient identifier, the community health index number, allows records from Scottish Stroke Care Audit and the General Register Office death registry to be linked. Linkage was done by Information Services Division Scotland and then pseudoanonymized before data analysis.
We included all index ischemic stroke events, defined as stroke at final discharge diagnosis. We classified patients as either dead or alive by 30 days and 6 months after admission or after stroke occurrence if already hospitalized. Recorded discharge destination includes discharged home or to usual place of residence, to another acute hospital, care home, National Health Service continuing care, an over-riding diagnosis, death, rehabilitation, and others. Deprivation category is calculated according to postcode using the Scottish Index of Multiple Deprivation, where one is the least deprived quintile and 5 the most deprived.9
We investigated outcomes for patients achieving each of the standards individually and according to the number of standards achieved. Main outcome measures were all-cause mortality at 30 days and 6 months after admission, and the secondary outcomes were stroke mortality at 30 days and 6 months and discharge to home/usual place of residence at 30 days and 6 months.
Controlling for Bias
Early deaths after stroke may be nonmodifiable10 and may result in the patient dying before they can receive a component of the care bundle. To reduce this potential source of bias, patients who died on days 0 to 3 were removed from the data set before initial demographic analysis or models were fitted. Patients who were discharged on day 0 were also removed.
Prediction models for long-term outcome after stroke have been developed to adjust for important case-mix variables.8 We have used the SSV model, which includes age at admission, prestroke living arrangement, prestroke independence, arm power at admission, ability to walk at admission, and normal verbal component of the Glasgow Coma Scale at admission. These variables are included in the recommended admission data set and relatively simple for trained audit staff to extract from the case notes.11 This model performs as well as or better than other simple predictive systems for predicting the outcomes of being alive at 30 days and independent at 6 months and 1 year after stroke.8 We have previously shown that stroke unit admission is associated with better outcomes up to 1 year, using the SSV model for case-mix adjustment, with a receiver operating characteristic for mortality at 6 months of 0.82 (SE, 0.002).12 A systematic review of case-mix adjustment models for stroke confirms that the SSV model demonstrates statistical robustness, good discriminatory function in external validation studies and comprises variables that are clinically feasible to collect at ward levels by nonspecialist staff.13
Although stroke care in Scotland is generally similar in all acute National Health Service hospitals, there are differences in service organization and numbers of admissions, in addition to case mix. We therefore added hospital as an additional variable for adjustment.
The study was approved by Scotland A Research Ethics Committee, reference number=10/MRE00/76, and the Privacy Advisory Committee of Information Services Division, National Health Service Scotland, reference=76/11.
We performed data management and statistical analyses using SPSS version 22 and SAS version 9.2. Using standard descriptive statistics, characteristics for the study cohort were calculated as percentages for categorical variables and means/medians for continuous variables.
Using multilevel multivariable logistic regression models, we firstly investigated outcomes of patients achieving each of the standards individually and then estimated associations with 30-day and 6-month mortality. Patients who achieved the full bundle were used as the index group. Adjustment was made with the SSV and year of admission. Age was a continuous variable, whereas the others were categorical. The model was a 2-level multivariable logistic model using random intercepts for each hospital to account for the clustered nature of the data. The effect of the care bundle on outcome of discharge to home/usual place of residence at 30 days and 6 months was also investigated using logistic regression adjusting for the effects of the SSV, year of admission, and hospital.
Complete data were available for all outcomes measures. 13.1% had ≥1 of the case-mix adjustment variables missing, and exploratory analysis was performed to assess missing data patterns. Missing data were randomly distributed between hospitals but commoner in the earlier years of the audit.
To assess whether missing case-mix variables would affect the results, we performed missing data imputation using the Markov chain Monte Carlo method with 5 iterations. The adjusted odds ratios (ORs) with 95% confidence interval (CI) for outcomes restricted to the cases with complete case-mix information were more conservative than the results for all cases with imputation of missing data. All estimates were therefore focused on analyses of complete cases.
Patient demographics are presented in Table 1.
Data were available for 36 055 patients. The numbers and percentage receiving none, 1, 2, 3, and all 4 components of the bundle are shown in Table I in the online-only Data Supplement. A total of 2264 (6.3%) patients had a missing value for ≥1 of the bundle components. Between 2005 and 2011, there was a steady increase in the numbers achieving the full bundle (Figure). Further adjustments took year of index events into consideration.
Adjusted OR and 95% CI were obtained for complete cases (n=29 672). Table 2 shows the frequency of achieving each individual component of the bundle and adjusted OR for all-cause mortality 30 days and 6 months after admission. The most commonly attained standard was aspirin started on day 0 or 1 (84.4%), followed by swallow screen (77.4%) and brain scan on day 0 (77.1%). Admission to a stroke unit on day 0 or 1 was only achieved in 58.7%. Admission to a stroke unit on day 0 or 1, swallow screen on day 0, and aspirin on day 0 or 1 was associated with reduced mortality at 30 days.
Patients admitted to a stroke unit on day 0 or 1 were more likely to achieve the other 3 measured components of the bundle. If patients achieve the stroke unit standard, the OR for achieving the other components of the bundle is 2.57 (95% CI, 2.44–2.71) for scan on day 0, 3.42 (95% CI, 3.24–3.62) for swallow screen on day 0, and 1.46 (95% CI, 1.37–1.54) for aspirin on day 0 or 1.
Table 3 shows mortality according to the number of standards achieved for each patient. There was an incremental decrease in all-cause mortality with more standards achieved at both 30 days and 6 months after stroke. Where ≥1 of the bundle components was not recorded, patient mortality was comparable with the group who did not meet any of the standards.
Table 4 shows the adjusted OR for all-cause mortality at 30 days and 6 months using those meeting the full bundle as the index group. As the number of standards achieved increased, there was a significant reduction in mortality at both time points, compared with the index group.
Table II online-only Data Supplement shows mortality at 30 days and 6 months where stroke was the underlying cause of death on the death certificate, according to the number of standards achieved.
Table 5 shows the adjusted OR for the outcomes of discharge to usual residence at 30 days and 6 months, according to the number of components of the bundle achieved, with the index group being those achieving all components of the bundle. Although there was no significant relationship between compliance with standards and discharge to usual residence at 30 days, those who achieved only 1 component were less likely to return home compared with the index group. At 6 months, a relationship between numbers achieving the standards and discharge destination was seen, with those who achieved the full bundle more likely to have returned to usual place of residence.
Our national study has shown that implementation of a care bundle for ischemic stroke comprising 4 basic components of clinical care is associated with reduced mortality at 30 days and 6 months and with increased likelihood of discharge to usual place of residence at 6 months. Although not all standards individually predicted outcomes, the overall bundle contributes to improved patient outcomes of mortality and likelihood of successful discharge to usual place of residence.
Some, but not all, studies have shown that patients with stroke who achieve recommended standards of care are more likely to survive. A systematic review7 reports that 9 of 14 studies found an association between positive metric compliance and stroke outcomes, with considerable variation in size and population composition, risk adjustment methods, data capture, and time windows for measurement of outcomes. A national Danish study of mortality at 30 and 90 days after stroke found an inverse dose–response relationship between the number of quality standards met (early admission, early antiplatelets or anticoagulants, early scanning, early physiotherapist, occupational therapist, and nutrition assessment) and mortality at 30 and 90 days. This study did not include functional end points, such as discharge destination.14
A study of 36 179 hospital patients from English hospitals participating in the Stroke Improvement National Audit Program and the Sentinel Stroke National Audit Program also found a relationship between the process of care and mortality at 30 days.15 An organizational model and 3 process measures, (1) seen by a consultant or a specialty doctor within 24 hours of admission, (2) nutrition screening and formal swallow assessment within 72 hours, and (3) antiplatelet therapy and adequate fluid and nutrition, were associated with reduced mortality. Interestingly, they too found no association between early scanning and mortality. In addition, no association between admission to stroke unit within 4 hours of hospital admission and mortality was observed.15 In contrast, we found that achieving early admission to a stroke unit was associated with a reduction in all-cause mortality at 30 days and 6 months and results in an increased likelihood of other standards being implemented. The English study may have been subject to bias as it is a voluntary audit and lacks the independent data collection used in the Scottish Stroke Care Audit. Mortality data in our study were derived from independently collected and validated national data. We found that patients in whom stroke data were not recorded had mortality outcomes similar to those who do not achieve any of the standards. This supports evidence that voluntary reporting may result in not all patients who died being included in audit data.16,17
In addition to showing that there is a reduction in all-cause and stroke-specific mortality with a simple care bundle, we have also shown that discharge home at 30 days is more likely if >1 standard is achieved. There was a more striking dose–response relationship at 6 months. This suggests an on-going benefit on recovery from early evidence-based management. Taking hospital-level random effects into consideration is particularly important for this outcome measure, as the process of care may vary, with some urban centers able to provide early supported discharge, for example, which may affect the length of stay. Mortality at 30 days is the outcome recommended by the American Stroke Association.3,18 Examining functional outcomes, such as discharge later than 30 days may be appropriate particularly where healthcare models differ.
The improvement in numbers of patients achieving the standards over time may reflect the national publication of data at hospital levels and increasing awareness and training of staff at stroke unit levels. Data from the US Get with the Guidelines Stroke program has confirmed that routine collection and feedback of data are associated with marked improvements in the quality of care.19 This is supported by a Cochrane review of the effect of audit and feedback on healthcare outcomes at a local level,20 although there is less evidence of an effect of public release of performance data on changing professional or organizational behavior.21
The Danish Register research has shown a negative association between the process of care and medical complications after stroke; patients who had fewer medical complications (in particular pneumonia) have improved survival at 30 days and 1 year after stroke.22 Checking for impaired swallow on admission and modifying oral intake accordingly reduce the risk of pneumonia and improve survival in some studies.14,15 This standard was the one least likely to be achieved in our study and was also associated with mortality. Early antiplatelet therapy23 and stroke unit admission24 have been shown to reduce death and improve functional outcome, and these were confirmed in our study.
Strengths of this study include that this is a national data set from a health provider with standardized guidelines and audit data collection, including the SSV to correct for bias by age, stroke severity, and previous status. The SSV performs robustly in published studies and has been extensively validated. There is concern about publishing tables that compare stroke outcomes from different hospitals or health services and on-going discussion on how best to adjust data to take case-mix variation into consideration.3,18 In this study, adjustment of the raw data for the SSV along with year and hospital of admission enhanced the effect of achieving the full bundle on improving stroke outcomes. We removed patients with hemorrhage from analysis: thus >90% of the population would be eligible for the metrics recorded, and the size of the study population reduces the likelihood of random errors.
Potential weaknesses include the possibility that unmeasured variables are also likely to affect survival and outcome. Initial patient care may be influenced by perceived futility of stroke-specific intervention because of preexisting frailty, dementia, or stroke severity, and patient or relative choices on care pathways. We have gone some way to address this by removing from the analysis all patients who died within 3 days (Appendix II in the online-only Data Supplement shows the impact of this). We have not specifically measured dementia, but the SSV does take preexisting-independent living into account. We do not formally record functional outcomes at discharge with, for example, the modified Rankin scale, but discharge to usual residence is used as a surrogate measure.
Further limitations of this study may include the possibility that patients with more severe strokes may have been imaged more rapidly, whereas those with milder strokes are less likely to be scanned early.7 The SSV incorporates a measure of stroke severity, which reduces this potential bias; in addition, the SSV has been validated for 6- and 12-month outcomes. In keeping with other studies,7,14,15 we have corrected for the potential bias of clustering by hospitals by correcting for this in analysis. Missing data are also a potential limitation that we addressed by removing patients with missing case-mix variables from outcome analysis.
Implementation of evidence-based care standards requires time and effort in healthcare settings. There are a limited number of studies looking at public reporting of performance measures and subsequent improvements in quality indicators or patient outcomes after stroke.13 The data show that both data capture and the percentage of patients with stroke achieving the standards have improved over time. This may reflect the increased awareness of performance driving local service improvement, along with on-going staff education and increasing proportions of patients accessing the stroke unit.
In summary, we have confirmed that achieving a simple set of quality standards was associated with reduced mortality at 30 days and 6 months after stroke, when known predictors of outcome are taken into account. Achieving more components of the stroke bundle at admission is associated with an increased likelihood of discharge home and is a finding that is worthy of further exploration.
We acknowledge the support of all audit coordinators and clinicians who contribute to the Scottish Stroke Care Audit. Information Services Division, National Health Service Scotland (in particular Lindsey Waugh) supported data linkage with General Register Office.
Sources of Funding
This study was funded by Chest, Heart and Stroke Scotland (Grant no R11/A134). The Scottish Stroke Care Audit was funded by National Health Service Scotland. Neither funder had any role in the analysis.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.007608/-/DC1.
- Received October 1, 2014.
- Revision received January 16, 2015.
- Accepted January 21, 2015.
- © 2015 American Heart Association, Inc.
- Resar R,
- Griffin FA,
- Haraden C,
- Nolan TW
- Katzan IL,
- Spertus J,
- Bettger JP,
- Bravata DM,
- Reeves MJ,
- Smith EE,
- et al
- 4.↵Scottish Intercollegiate Guidelines Network (SIGN). Management of Patients With Stroke or TIA: Assessment, Investigation, Immediate Management and Secondary Prevention. Edinburgh: SIGN; 2008. Guideline No. 108, ISBN 978 1 905813 40 7. http://www.sign.ac.uk/guidelines/fulltext/108/. Accessed December 22, 2014.
- 5.↵Health Care Improvement Scotland. Clinical Standards for Stroke Services: Care of the patient in the acute setting–Update June 2009. http://www.healthcareimprovementscotland.org/previous_resources/standards/clinical_standards_for_stroke.aspx. Accessed October 1, 2014.
- 6.↵Scottish Stroke Care Audit. 2014 National Report: Stroke Services in Scottish Hospitals. http://www.strokeaudit.scot.nhs.uk/Reports/main.html. Accessed October 1, 2014.
- Parker C,
- Schwamm LH,
- Fonarow GC,
- Smith EE,
- Reeves MJ
- Counsell C,
- Dennis M,
- McDowall M
- 9.↵A Scottish Government National Statistics Publication. Scottish Index of Multiple Deprivation 2009: General Report. Edinburgh: Scottish Government; 2009.
- 11.↵Scottish Stroke Care Audit Inpatient Core Dataset Definitions and Values. SSCA Web site. http://www.strokeaudit.scot.nhs.uk/Forms/CoreDatasetDefinitions_INPATIENT_V42310714.pdf. Accessed January 8, 2015.
- Turner M,
- Barber M,
- Dodds H,
- Dennis M,
- Langhorne P,
- Macleod MJ
- Teale EA,
- Forster A,
- Munyombwe T,
- Young JB
- Bray BD,
- Ayis S,
- Campbell J,
- Hoffman A,
- Roughton M,
- Tyrrell PJ,
- et al
- Kelly A,
- Thompson JP,
- Tuttle D,
- Benesch C,
- Holloway RG
- Hollenbeak CS,
- Gorton CP,
- Tabak YP,
- Jones JL,
- Milstein A,
- Johannes RS
- Fonarow GC,
- Alberts MJ,
- Broderick JP,
- Jauch EC,
- Kleindorfer DO,
- Saver JL,
- et al
- Schwamm LH,
- Fonarow GC,
- Reeves MJ,
- Pan W,
- Frankel MR,
- Smith EE,
- et al
- Ketelaar NA,
- Faber MJ,
- Flottorp S,
- Rygh LH,
- Deane KH,
- Eccles MP
- Ingeman A,
- Andersen G,
- Hundborg HH,
- Svendsen ML,
- Johnsen SP
- Langhorne P