Background

The American Heart Association (AHA) is a scientific society that supports evidence development through research and publishes evidence-based guidelines for healthcare providers to improve the prevention and treatment of cardiovascular disease and stroke. However, the publication of evidence-based guidelines alone is not enough to improve clinical practice. Barriers to adherence included lack of familiarity or awareness of the guidelines, lack of motivation and outcome expectancy, lack of time and resources, organisational constraints and perceived malpractice liability.1 2 The AHA has several mechanisms for measuring processes of care and outcomes through which it promotes improved quality of care. In collaboration with the American College of Cardiology, it has a formal mechanism for the development and promulgation of performance measures that are evidence-based recommendations supported by the highest level of evidence and suitable for public reporting by federal agencies.3 Within the AHA suite of quality programmes, the Get With the Guidelines (GWTG) programmes include achievement and quality measures that have been developed by expert review and consensus from available evidence and that have sufficient evidence that failure to provide the recommended care is likely to result in poor patient outcomes.4 Performance measures are formally evaluated for validity, feasibility and impact on outcomes and are usually submitted to the National Quality Forum for formal review and endorsement by an independent agency. For the purposes of this review, we will use the term ‘performance measures’ more generally to encompass both formal performance measures and other measures of quality used within the GWTG programmes.

In the late 1990s, stroke care in the USA was fragmented, and rates of stroke incidence and mortality were high.5 Despite its approval by the US Food and Drug Administration in 1996, intravenous tissue plasminogen activator (IV tPA) was underutilised in the acute treatment of stroke.6 The AHA established the American Stroke Association (ASA) in 1998, and one of its first missions was to help bridge the gap between evidence and practice. The ASA formed the Metro Stroke Task Force to increase stroke awareness and improve the stroke system of care with an emphasis on hospital access and emergency response to stroke. The programme evolved into Operation Stroke, a community-based public awareness programme for stroke prevention and treatment. Operation Stroke ran through 2004 and was associated with improvements in stroke screenings and hospital infrastructure. In 2001, Senator Edward M. Kennedy worked with Massachusetts ASA volunteer Dr Lee Schwamm to craft a bill to fund the development of a stroke system of care. Together with Senator Bill Frist, MD, he introduced Senate Bill 1274 ‘The Stroke Treatment and Ongoing Prevention (STOP Stroke) Act of 2001’, which appropriated $40 million for stroke prevention, treatment and rehabilitation.7 Though the bill passed the Senate in 2001 and was reintroduced for several subsequent years, it was never successfully passed by both houses of Congress and into law. However, it was clear that hospital-level change was required to effect a reduction in stroke risk and improvement in patient outcomes, and efforts were initiated in parallel to the legislative actions to impact stroke care delivery.

In 2001, the Centers for Disease Control and Prevention (CDC) funded four Paul Coverdell National Acute Stroke Registry (PCNASR) prototypes to monitor stroke care and guide quality improvement.8 The registry, named for Senator Paul Coverdell who suffered a fatal stroke due to cortical venous sinus thrombosis, served the dual purpose of disease registry and quality improvement initiative. Representatives from the CDC, Brain Attack Coalition (BAC), National Institute for Neurological Disorders and Stroke, Centers for Medicare & Medicaid Services (CMS) and Veterans Association developed an initial set of proposed data elements that would be collected from prehospital transport through postdischarge follow-up. These data elements were reviewed and refined collaboratively by the PCNASR prototype investigators from four states to develop a working pilot registry in the 2001–2005 cycle.

The Massachusetts prototype funded the development of the GWTG-Stroke alpha pilot. Patients admitted for new onset acute stroke or transient ischaemic attack (TIA) with symptoms present on hospital arrival were entered in the registry. The infrastructure of the programme was modelled on the GWTG-Coronary Artery Disease (GWTG-CAD, now ACTION Registry-GWTG) pilot launched in 2000. Data collection included measures from the acute treatment phase through hospital-based secondary prevention measures at discharge. The Research Triangle Institute, independent auditor to the CDC, evaluated the prototypes.9 Analysis of the 6867 admissions from 2001 to 2002 found low adherence to established treatment guidelines, reaffirming the need for hospital-level interventions in stroke care. In 2003, GWTG-Stroke launched nationally and became available to all US hospitals on a voluntary basis.

Organisation and implementation

The GWTG programme was developed as a systems-based rather than practitioner-focused intervention to address the gap between knowledge of guidelines and translation to clinical practice.10 GWTG was based in part on similar efforts including the Cardiac Hospitalisation Atherosclerosis Management Program.11 Guidelines were integrated in a Patient Management Tool (PMT) to maintain adherence, with hospitals able to run reports comparing themselves to other peer hospitals. Results from 24 Massachusetts hospitals participating in the 1-year GWTG-CAD pilot showed clinically and statistically significant increases in adherence to guidelines and provided proof of concept for active interventions to improve patient outcomes.12 Based on these promising results, the AHA was emboldened to publicly announce its intention of reducing coronary heart disease, stroke and risk by 25% by 2010, with the GWTG suite of programmes at the heart of its efforts.13

GWTG-Stroke operates at the national and local levels and involves a series of quality improvement cycles and collaborative workshops to refine and develop hospital protocols and processes. At the core of GWTG-Stroke are the stroke performance measures, which were developed through the PCNASR process and subsequently harmonised with other similar efforts by other societies in a 2-day consensus conference attended by representatives of the AHA, CDC and the Joint Commission (JC). At the national level, Steering and Quality Improvement Committees recruit experts in registries, data acquisition and quality improvement to oversee the programme, review new evidence, recommend changes to the programme and disseminate guidelines to practitioners. The committees hold organisational stakeholder and opinion leader meetings, recruit and recognise hospitals for high performance, host collaborative learning sessions, develop hospital tool kits and drive legislative change.2

The PMT (Quintiles, Cambridge, MA) is crucial to the operation of GWTG. The PMT is an electronic case report form that serves the dual purpose of quality assurance and data collection. The PMT is integrated with electronic medical records and allows users to enter individual patient data at the point of care or during retrospective chart abstraction. Drop-down reminders ensure assessments and interventions are completed, and real-time data checks identify inconsistent entries or values that are out of range, eliminating delays in performance feedback on the individual level. Referral notes, patient letters and patient education materials are built into the tool for use at discharge if desired. Quintiles serves as the data collection and coordination centre, and the Duke Clinical Research Institute serves as the data analysis centre. The AHA works with government agencies to ensure that hospitals can report their data to State Health Departments when required for state-based stroke centre certification programmes and that hospital emergency departments have protocols for acute stroke treatment.

Participation in GWTG is voluntary. In the early stages, hospitals were recruited by AHA staff and volunteers based on interest, senior leadership commitment and geographic, ethnic and socioeconomic diversity of sites. An initial regional learning session was held to bring together stakeholders from all the participating hospitals, and 30 records were entered in the PMT at hospital enrolment to assess baseline performance. At the local level, hospitals define focused goals for improvement in adherence to achievement and quality measures. Reports generated by the PMT provide instant feedback to allow sites to problem-solve barriers to adherence and change protocols and order sets in their site if needed. Multidisciplinary teams in each hospital convene to review the data and develop strategies to further refine protocols in Plan Do Study Act (PDSA) cycles. The rapid cycles afforded by the continuous quality improvement (CQI) framework of the PDSA system allow for testing on a small scale and encourage rapid, innovative changes (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

GWTG quality improvement framework. Foundation of the GWTG quality improvement programme. Adapted from Current Treatment Options in Cardiovascular Medicine. 2005;7:253–258. GWTG, Get With the Guidelines; WS, workshop.

At initial implementation, collaborative learning sessions are held every quarter and bring together multidisciplinary teams from different hospitals to address barriers to care. Hospital teams report on the success of different tools they have implemented, fostering a sense of community and accelerating improvement. Findings from clinical trials are presented, and guidelines for acute care and secondary prevention are disseminated. In between learning sessions, collaboration is continued with monthly conference calls, webinars, online discussion groups and email exchanges. Over time, as the majority of large US hospitals within communities have joined GWTG-Stroke, new learning sessions are not held in person but ongoing continuing stroke education and problem solving are shared via national webinars featuring senior volunteers.

Hospital recognition awards encourage progress and provide publicity. The Performance Achievement Award (PAA) recognises hospitals with a multidisciplinary team, a physician champion, orders or protocols that include GWTG measures, submission of data for 1 year of stroke discharges, commitment to ongoing data collection and CQI and adherence to the seven achievement measures in 85% of all eligible patients (ie, those without any documented contraindication to treatment).

There are considerable financial and human resource costs associated with the development and implementation of GWTG. A sustainable national registry that serves a CQI function ideally requires a single set of data elements and performance measures, funding for data collection, improved electronic data collection, access to patient information while hospitalised and a commitment to ongoing improvement.14 Additional costs include education, system re-engineering, executive-level sponsorship and local staffing. Recommended hospital staff include a stroke coordinator who manages the site and does data abstraction, a physician champion, actively engaged physicians and nurses and an effective multidisciplinary team of health professionals committed to achieving the programme goals.

Achievement measures

Seven GWTG achievement measures, including four acute and three discharge measures, were developed from a consensus of stroke experts (table 1).15 Composite and defect-free measures are included to assess hospital performance in providing all the appropriate interventions for each patient. Hospitals report the proportion of eligible patients receiving the measure divided by the total number of eligible patients without contraindications.

Additional measures of quality that are not yet supported at the highest levels of evidence include dysphagia screening before any oral intake, door-to-CT time ≤25 min in patients presenting with stroke symptoms <3 hours duration, stroke education at discharge and assessment for stroke rehabilitation services. Additional quality measures were subsequently added to include IV tPA in patients arriving within 3.5 hours of symptom onset and treated within 4.5 hours of symptom onset, time to IV tPA ≤60 min, National Institutes of Health Stroke Scale (NIHSS) documented, LDL documented and intensive statin therapy for ischaemic stroke patients with evidence of atherosclerosis prescribed at discharge.16 Reporting measures include patient demographics and treatment time intervals throughout the hospitalisation.

Enrolment and reporting

In 2003, GWTG-Stroke was launched in eight additional states with a national launch later that year. There are currently over 4.5 million patients enrolled in GWTG-Stroke in over 2000 hospitals across the country, admitting ~50% of index stroke hospitalisations annually (figure 2).17 GWTG hospitals comprise a mix of JC-certified stroke centres, PCNASR hospitals and small and large hospitals in urban and rural settings across the USA and Puerto Rico. Comparison with statistics from the 2000 US Census shows the population of patients enrolled in GWTG is similar in age and racial makeup to the US population.18 Medicare beneficiaries linked to the GWTG registry are similar in demographics, comorbidities and inhospital outcomes compared with Medicare beneficiaries who are not linked.19 GWTG hospitals are more likely to be larger academic institutions located in urban areas in the Northwest and South, which is also where the majority of stroke patients in the USA are admitted.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Enrolment in GWTG-Stroke from 2003 to 2016. Yearly and cumulative enrolment in GWTG-Stroke from 2003 to 2016. Used with permission from the American Heart Association. GWTG, Get With the Guidelines.

The GWTG Steering Committee conducted a national data validation audit in 2012 that showed high accuracy and reliability across sites.20 An improvement in quality of care may reflect a greater number of treated patients (increase in the numerator) or a greater number of patients excluded from the target population (decrease in the denominator). An analysis was performed to test the assumption that the improvement in performance measure compliance was an indication of better patient care.21 The size of the target population did not change over time, and the improved performance reflected a higher proportion of patients receiving guideline-based treatment rather than a reduction in size of the target population or increased documentation of contraindications.

Outcomes

Early analyses of GWTG data saw marked improvement in adherence to measures.22 23 Quality measures improved after 6 months of programme initiation, not immediately, suggesting that the increased compliance with measures was the result of effective hospital interventions and not changes to data documentation. Analysis of 790 GWTG hospitals enrolled from April 2003 to July 2007 found clinically meaningful and statistically significant improvement in the seven achievement measures and a composite measure after 5 years of programme implementation.18 Results showed a 30.8% increase in IV tPA use for patients arriving within 2 hours of symptom onset, 15.8% increase in deep venous thrombosis (DVT) prophylaxis, 14.7% increase in lipid-lowering drugs for elevated LDL and 28.4% increase in smoking cessation, while smaller but significant gains were seen in measures with high baseline performance. After controlling for secular trends, GWTG was associated with a 1.18-fold annual increase in the odds of receiving guideline-recommended care. This effect was observed across different sizes and geographic distribution of hospitals, though larger hospitals and teaching hospitals saw the greatest improvement.

Analysis of the first 1 million GWTG-Stroke patients provided further evidence of improvements in achievement measures and patient outcomes.15 Admissions from 1419 hospitals from April 2003 to August 2009 showed 4.3% increase in discharge antithrombotics, 41.9% increase in IV tPA use in eligible patients, 51.0% increase in smoking cessation education, 20.8% increase in the composite score and 40.3% increase in the defect-free care measure. There was a 9.4-fold increase in odds of receiving guideline-recommended all-or-none care that was independent of patient and hospital characteristics. Temporal trends showed the proportion of patients discharged home increased, while hospital length of stay and inhospital mortality decreased.

Subsequent studies show continued compliance across achievement measures and improved patient outcomes.24–30Compared with control hospitals matched for teaching status, region, ischaemic stroke volume and mortality rates, GWTG Medicare beneficiaries showed an increased proportion of patients discharged home as well as decreased 30-day and 1-year mortality rates.31

Limitations

Quality improvement measures focus on reducing long-term disability and secondary prevention. A major limitation of the registry is the lack of postdischarge outcomes, without which the effect on long-term outcomes is challenging to measure. However, the resources required to collect long-term outcomes could prove a barrier to participation in and sustainability of the programme. In addition, linkage to large claims databases such as the Medicare Fee for Service dataset have allowed for longitudinal follow-up of outcomes such as death, rehospitalisation and time spent free of institutional living.32 Several clinical trials or observational studies have been performed within the GWTG-Stroke hospital cohort. The Patient-Centered Research Into Outcomes Stroke Patients Prefer and Effectiveness Research study builds on the GWTG programme to collect patient feedback on their hospitalisation and postdischarge quality of life and outcomes.33 Multiple analyses have linked cases with the Medicare Fee for Service database to analyse data on postdischarge resource utilisation and outcomes. Prospective studies such as the Adherence Evaluation After Ischemic Stroke Longitudinal registry analysed medication adherence and functional outcomes among GWTG patients after discharge.34 35 Access to longitudinal data has allowed for comparison of rehospitalisation and mortality rates in TIA and stroke patients, correlation between home-time and Modified Rankin Scale (mRS) score, outcomes among patients with atrial fibrillation treated with warfarin, incidence of depression and identification of predictors of discharge medication compliance 1-year post stroke.36–41

Milestones in stroke quality care

In 2005 the ASA endorsed the model of coordinated stroke systems of care.42 At the time, stroke remained the third leading cause of death and a significant source of long-term disability in the USA. The Institute of Medicine (IOM) of the National Academy of Science concluded that the fragmentation of healthcare delivery resulted in failure to provide effective stroke care. Comprehensive, coordinated stroke systems of care were needed. Stroke systems of care are longitudinal systems that address all aspects of stroke care delivery, including primordial and primary prevention, community education, notification and response of emergency medical services, acute stroke treatment, subacute stroke treatment and secondary prevention, rehabilitation and CQI activities.

The ASA established the Task Force on the Development of Stroke Systems to define and provide recommendations for stroke care systems. The Task Force defined criteria for stroke systems: effective interaction and collaboration among agencies involved in patient care; a standardised approach to care in each facility; identification of performance measures for evaluation of effectiveness; tools and coordination of resources for stroke prevention, treatment and rehabilitation; prioritisation of patient-centred protocols; identification of obstacles to implementation, including political, legal and economic concerns; and customisation for optimal stroke care by each state or region. The Task Force determined CQI strategies are a critical function of stroke systems to optimise effectiveness and recommended ongoing evaluation of overall patient outcomes, linkages among system components and with other entities and obstacles and potential treatment gaps.

By 2006, many hospitals were participating in GWTG, PCNASR and JC Primary Stroke Center (PSC) programmes, each with different but overlapping performance measures. In May 2006, the ASA, CDC and JC harmonised 10 key performance measures that included (1) DVT prophylaxis for non-ambulatory patients by the end of hospital day 2, (2) antithrombotic therapy at discharge, (3) anticoagulation at discharge for patients with atrial fibrillation, (4) thrombolytic therapy administered within 3 hours of time last known well for patients with acute ischaemic stroke who arrive at the hospital within 2 hours of last known well, (5) antithrombotic therapy by the end of hospital day 2, (6) discharge on cholesterol-reducing medication for patients with LDL >100, or LDL not measured or on cholesterol-reducer before admission, (7) dysphagia screening, (8) stroke education, (9) smoking cessation and (10) assessment for rehabilitation.4 The 10 performance measures were submitted formally by the JC in January 2008, and the National Quality Forum endorsed eight of the measures. This single, standardised set of performance measures facilitates quality improvement across hospitals and reduces costs in implementing registries. Following this endorsement and substantial advocacy efforts led by AHA/ASA, CMS required the eight measures to be reported in the Medicare Reporting Hospital Quality Data for Annual Payment Update system, making these data available from almost all US hospitals. CMS also added a structural measure asking hospitals to report participation in a Systematic Clinical Database Registry for Stroke Care and endorsed measures for use in the Physician Quality Reporting Initiative programme.

The shift from fragmented to comprehensive stroke systems of care was a critical step in national stroke care delivery. Stroke systems can be improved through participation in stroke centre certification programmes or CQI programmes like GWTG.43 Many states require GWTG as part of stroke centre designation, and the majority of PCNASR states use GWTG. Currently, 18 states and Washington, D.C. have statewide standards for the formal recognition of stroke facility designations and development of transport protocols, and 12 states and Washington, D.C. have standards for the development and utilisation of stroke registries.44 GWTG serves as the data collection platform for many hospitals to transmit information to JC, CDC, CMS and state health departments or emergency medical services (EMS) agencies for analysis of performance.

Comparison of adherence to performance measures among GWTG PAA hospitals and JC PSC-certified hospitals showed that conformity with each performance measure was highest in PAA hospitals regardless of PSC certification.45 PSC certification does not require evidence of achieving a certain level of performance to maintain recognition as the PAA does, suggesting a continuous quality monitoring system that requires explicit performance thresholds that could improve stroke care delivery. The BAC and ASA recommend that PSCs participate in CQI, which may account for the high uptake of GWTG.17 Most primary stroke centres now participate in GWTG, and GWTG data have been used to compare PSC certification programmes by the JC, state based and other organisations.46

Transforming care

GWTG has facilitated the dissemination of new findings in stroke research and has led to the rapid translation of new findings into clinical practice (table 2). After the 2008 European Cooperative Acute Stroke Study (ECASS) III demonstrated that IV tPA administration 3–4.5 hours after symptom onset improves outcome, the ASA released a science advisory to reflect the results of the trial.47 Updated guidelines were disseminated to GWTG hospitals via learning sessions and national webinars, and this was associated with a rapid adoption of the expanded tPA window.48 Similarly, after publication of the Management of Atherothrombosis With Clopidogrel in High-Risk Patients trial results, GWTG hospitals saw a rapid reduction in prescription of dual antiplatelet therapy.49

GWTG studies have also led to new interventions. Hospitals had suboptimal rates of tPA administration within 60 min and had variation in timeliness of tPA initiation.50–59 In 2010, the ASA launched Target: Stroke, a campaign to reduce door-to-needle (DTN) time modelled after similar successful efforts to reduce door-to-balloon time in primary percutaneous coronary interventions for acute myocardial infarction. Protocols and processes were analysed to establish the reasons for delayed DTN times.60–63 GWTG toolkits were revised to include interventions such as the Stroke Rapid-Treatment Readiness Tool.64 After the launch of Target: Stroke, median DTN time decreased from 74 to 59 min, and the percentage of patients treated within 60 min increased from 29.6% to 53.3% (figure 3).65 Inpatient mortality and long-term disability were reduced in patients treated within 60 min. Absolute rates of tPA use within 3 hours of symptom onset among all ischaemic stroke patients admitted nearly doubled from 4.0% in 2003–2005 to 7.0% in 2010–2011 and expanded to include more patients who were older, non-white race/ethnicity and presented with mild deficits.66 Despite this progress, delays and disparities still exist in administration of tPA, necessitating further iterations of interventions.67–73

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Impact of the Target: Stroke quality improvement initiative. Time trend in the proportion of patients with door-to-needle times for tPA ≤60 min during the preintervention and postintervention periods of Target: Stroke. Reproduced with permission from JAMA. 2014; 311(16):1632–1640. Copyright ©2014 American Medical Association. All rights reserved. tPA, tissue plasminogen activator.

GWTG studies have influenced expansion of the patient population eligible for tPA by studying the rates of adverse events in patients older than 80 or with other exclusions from the 3–4.5 hour treatment window recommendations based on the ECASS III trial. The additional tPA exclusion criteria of age >80 years, history of stroke and diabetes mellitus, oral anticoagulant treatment (regardless of international normalised ratio (INR)) and NIHSS >25 were analysed in GWTG-Stroke to detect any signals of lack of safety or efficacy.74 Among the 31.5% of patients given tPA beyond 3 hours who met at least one exclusion criterion, no increased risk of symptomatic haemorrhage or of worsening outcomes was observed, suggesting that expansion of the inclusion criteria could be considered. Other studies have examined the risks and benefits of tPA in specific patient populations including patients taking novel oral anticoagulants or with hyperglycaemia, malignancy, leukoaraiosis, dementia, sickle cell disease or mild symptoms at presentation.75–82 Given its prominent role in the patterns of care delivery with IV tPA, the ‘drip and ship’ method of tPA administration (ie, the treatment with tPA at an initial hospital followed by transfer to a stroke centre of higher capability for admission and further care) has been evaluated in GWTG and has been shown to be safe and efficacious while further increasing the proportion of patients who can receive tPA.83–86

GWTG studies led to the development of novel validated risk scores and mortality models for both ischaemic and haemorrhagic stroke, which have aided in prognostication and better understanding of case fatality rates.87–91 The NIHSS was determined to be a be a strong discriminator of 30-day mortality risk and was instrumental in helping to revise a CMS stroke mortality measure that was lacking a measure of stroke severity.91 Other studies have identified opportunities for improvement in hospital prenotification and EMS diagnosis,92–97 rates and risks of procedures and inpatient complications98–103 and discharge processes.104–106

A priority in national healthcare is the reduction of racial/ethnic and socioeconomic disparities. Quality of care improved across black, white and Hispanic patients after GWTG implementation, although black patients are still less likely than white or Hispanic patients to receive evidence-based care.107 Several studies have examined persistent disparities in race/ethnicity, age, sex and socioeconomic status that must be addressed.108–116 Studies have also found variation in care according to hospital region, time of presentation and stroke subtype.117–122 A recent study found that despite the regional variability of healthcare resources available for acute stroke treatment, quality of care and inhospital outcomes in GWTG hospitals did not differ by regional resource availability.123