Context

Context and rationale for the research

The ‘weekend effect’ and 7-day working policies

In 2001, Bell and Redelmeier1 published a landmark study showing that mortality was higher among people admitted to hospital during weekends or ‘out of hours’ than among those admitted during normal hours on weekdays, based on the health records of 3.8 million Canadian patients. This effect has been repeatedly demonstrated across a range of acute conditions, patient populations and health systems,1–3 such as myocardial infarction4 and stroke,5–9 patients undergoing elective surgery10 and admissions to emergency departments (EDs)11 or intensive care units.12 However, a number of studies report equivalent or better outcomes in some patients treated outside normal working hours.10,12–14 Although contested, such studies indicate that high-quality out-of-hours care is possible and that changes in organisation of services might address the current performance gap.

The Secretary of State for Health made this issue a priority in 2015 by suggesting that ‘someone is 15% more likely to die if admitted on a Sunday than on a Wednesday because we do not have as many doctors in our hospitals at the weekends as we have mid-week’ (House of Commons, 2015. Contains Parliamentary information licensed under the Open Parliament Licence v3.0. See https://www.parliament.uk/site-information/copyright-parliament/open-parliament-licence/).15 This led to controversial changes to doctors’ work contracts, provoking large-scale strikes16 and an NHS England policy for a 7-day NHS by 2020.17 The NHS Long Term Plan18 has identified, in particular, that all stroke units will, over the next 5 years, meet the NHS 7-day standards for stroke care and the national clinical guidelines for stroke over the next 5 years.

Several possible causes for an ‘out-of-hours effect’ are proposed, but many relevant factors are likely to be associated with out-of-hours working, including differences in patient case mix affecting outcomes among patients admitted at different times of the week;19 reduced staffing levels;2,11,20 reduced access to supporting investigations;2,11 lower seniority and clinical experience of available staff;1,2 less familiarity of staff with patients;1 less familiarity with unit-specific processes and policies, and insufficient training to address this;2 and fewer supervisory staff and less supervisor familiarity with the staff they supervise.1 Systems by which such issues may be identified and addressed, such as clinical governance, and cultures that support staff engagement in these systems, may play an important role in how the out-of-hours effect manifests. Ward and organisational leadership may also represent significant factors in supporting service development, learning and engagement.21

The controversy raised a number of relevant issues for research: first, that exposure of differences in outcomes across the week was having an impact politically, and that it was necessary to answer the question of whether this was a true effect, or an artefact of the patient mix or the quality of the data; second, that it was important to measure other outcomes and care processes, as well as mortality; and, finally, to change the approach towards acknowledging patterns of temporal variation.22–25 This movement is characterised by the idea that certain times of the day, or days of the week, are associated with poorer outcomes for the patient, but that this is not necessarily limited to the weekend or night-time.22 As a result, the concept of ‘out-of-hours’ working or the so-called ‘weekend effect’, which once dominated the discourse of the literature on temporal variation,22,26–28 has been re-evaluated.

It is now recognised that temporal variation results, at least in part, from delivery models of care and the cultural values of the health-care sector, enshrined in everyday practice, funding models, staffing practices and wider policies.29 Recent research, such as the High-Intensity Specialist-Led Acute Care study,30,31 have taken this approach to examine the intensity of specialist-led care in emergency medical admissions, focusing particularly on the organisation and delivery of care, and medical staffing.30,31

In the case of urgent stroke care, recent studies have shown that temporal variation in mortality is now decreasing,32,33 including in London’s hyperacute stroke unit (HASU) model.34 However, there has been no large-scale, organisational research to understand what creates temporal in consistency, particularly in services that aim to operate consistently over 7 days, such as the London stroke services. The current study adds a wealth of information to this gap.

Reconfiguration of stroke services and the aspiration of 7-day working

In 2010, acute stroke services in London (population 8.2 million) were centralised. Hospitals were selected to become sites for specialist stroke care (over the first hours following stroke) in a hub-and-spoke network.35 In London, prior to the reorganisation, 30 hospitals provided acute stroke care. After the reorganisation, initial specialist stroke care was routinely provided to all patients with suspected acute stroke at eight designated HASUs, with all patients being assessed shortly after arrival at the designated hospital by specialised stroke assessment teams able to provide access to immediate brain imaging in all patients and delivering immediate intravenous thrombolysis, when appropriate (Figure 1). The HASU model involved the London Ambulance Service taking all patients with suspected stroke within 48 hours of onset to one of the eight HASUs.36 Acute stroke patients seen at other medical facilities were similarly transferred as an emergency to a HASU. The aim of the HASUs was to provide specialised care for all acute stroke patients during the first 72 hours after onset of stroke. Twenty-four acute stroke units in London, linked to HASUs, were designated to provide specialist acute stroke care beyond 72 hours for those patients still requiring inpatient treatment. Eight of these were in the same hospital trust as a HASU.37 On average, across all patients, the quality of acute stroke care in London was better than in the rest of England and increased as a result of the centralisation,38 with a decrease in mortality.35

FIGURE 1 Summary of London’s stroke service model before and after reconfiguration. A&E, accident and emergency; ASU, acute stroke unit; MAU, medical admission unit; SU, stroke unit. (Figure not reproduced here; see the original article.)

Performance standards for HASUs, linked to payments, were initially set by Healthcare for London39,40 and subsequently the London Strategic Clinical Networks to maintain high quality of care across the HASU stay. Standards include staffing quotas, required infrastructure and delivery of care. Some standards were set to provide rapid access to time-critical ‘front-door’ measures [e.g. dysphagia screen within 4 hours of admission, computerised tomography (CT) scan within 1 hour of admission and thrombolysis administration within 60 minutes]. Other standards were set with less stringent time constraints (e.g. stroke specialist consultant physician assessment within 24 hours and physiotherapist assessment within 72 hours). An evaluation of London’s stroke centralisation found that mortality and length of stay (LOS) were reduced, if all stroke patients receive hyperacute care.35 It has also been shown that these changes are sustainable over time.41

The London model has been shown to improve clinical outcomes and care provision,35,37,38 and is cost-effective.42 Beyond London, many regions in the NHS are currently seeking to centralise stroke service organisation in different ways.43,44 The development of mechanical thrombectomy (surgical removal of a blood clot from a blood vessel) services is an important new development for a subset of stroke patients. As of May 2020, further reorganisation of the London model is currently in progress with the designation of a subset of HASUs that can provide thrombectomy. At the time of our study (2014–15), no HASU was providing a formal thrombectomy service for stroke associated with large vessel occlusion. Therefore, 24/7 availability of scans and assessments by specialists will be vital to enable the correct patients to be transferred to a thrombectomy centre.

Aims and research objectives

Research aims

This study aimed to identify the extent to which London stroke services were providing care 24 hours a day, 7 days a week (24/7), and, if they were, how this was being achieved. The research provides an opportunity to examine how and if a range of organisational factors influence 24/7 provision of stroke care, and to assess the extent to which temporal patterns of provision are reflected in patient outcomes, including mortality, LOS and functional independence. The research also aimed to explore service developments related to temporal consistency. These lessons are of potential value to people who commission, organise and run stroke services, both in London and across any area of the English NHS that seeks to develop or provide 24/7 hyperacute stroke care. They may also have relevance to research in, and provision of, 24/7 care in other health-care settings, such as emergency and intensive care, treatment of myocardial infarction and elective surgery.

Research objectives

Objective 1

To investigate the extent to which evidence-based care received by stroke patients differs between in-hours and out-of-hours services in London HASUs.

Objective 2

To investigate the extent to which clinical outcomes differ between in-hours and out-of-hours services in the London HASUs.

Objective 3

To investigate how in-hours and out-of-hours services are organised and provided in London HASUs.

Objective 4

To investigate how these services are perceived by stakeholders (including stroke clinicians, patients and carers).

Objective 5

To investigate the costs of acute stroke care in London HASUs on different days and at different times of admission.

Overview of the research project

This project was funded by the National Institute for Health Research Health Services and Delivery Research programme, from January 2014, to study the development, provision, effects and cost of 24/7 stroke care in the London HASU system as part of an open call. The project was extended from its original end date of December 2016 to April 2019 due to delays and the eventual inability to obtain the Sentinel Stroke National Audit Programme (SSNAP) data linked to mortality data from the Office for National Statistics (ONS). The reason for this delay was difficulties in formalising data-sharing agreements between the Health and Social Care Information Centre (now NHS Digital, data controller for ONS data) and Healthcare Quality Improvement Partnership (data controller for SSNAP data). It was agreed with the funder that we would publish the results based on the SSNAP data we did have, rather than continue to delay the publication of the study. At an early stage of our investigation, we recognised the pivotal roles undertaken by the HASU lead nurses. Understanding these activities and their impact on the functioning of the HASUs required the additional capture of the nursing rotas for each HASU for the period of the study. Our analysis of the impact of nursing numbers and seniority in relationship to temporal variation and clinical outcomes is presented here (see Chapters 3 and 5), and strongly supports our initial observations on the central importance of nurse-led activities to our work. This enabled us to undertake new and originally unplanned analyses to look at the impact of nursing numbers and seniority (see Chapter 5) in relation to temporal variation and clinical outcomes (see Chapter 3).

Structure of the report

The report is structured as follows.

See Chapter 2 for:

• an overview of the study sites
• data and analysis methods.

(Detailed methods are provided within each findings chapter.)

See Chapter 3 for:

• costs of acute stroke care in London HASUs on different days and at different times of admission.

See Chapter 4 for factors that influence, temporally:

• consistent care
• inconsistent care.

See Chapter 5 for:

• nursing roles on the HASU
• temporal variation in nursing provision
• associated processes and outcomes.

See Chapter 6 for:

• temporal structures acting on the HASU patient pathway
• how this can have an impact on measurement of temporal variation.

See Chapter 7 for:

• discussion and summary of our findings in relation to our research questions
• implications for health services, research and 7-day working policies.

Changes to the protocol

We made some changes to the protocol because of difficulties in obtaining access to all our data sources, especially mortality data, as noted in Overview of the research project. Initially, we experienced difficulties gaining access to the HASUs, and later on it proved difficult to recruit patients and carers to our interviews. This is partly because of the long delay periods between recruitment on the wards and patients returning to their homes or care facilities when they would be ready to complete the interview. In addition, there was a significant dropout rate between patients expressing interest in taking part and completing interviews: 80 patients expressed initial interest, but either declined later on or would not respond to communications. Patient recruitment improved with registration of the study on the Clinical Research Network portfolio, which allowed access to the Clinical Research Network research practitioner teams in each HASU and a more systematic approach to patient identification. This released our team to discuss the study with patients who had already expressed interest in taking part.

Lack of access to mortality data or long-term dependency data

The most substantial change made to our protocol is in the absence of 3-month mortality and 6-month modified Rankin Scale (mRS) data. The basis of our data set is the SSNAP. The SSNAP is one of several ongoing mandatory national clinical audits and has contributed significantly to the debates around quality of stroke care at the weekends, by providing a detailed data set to several studies. We had hoped to link this data set to Hospital Episode Statistics and ONS data on mortality after discharge. Similar analyses have been achieved by other research teams (e.g. Bray et al.22). We estimate that we waited for these linked data for 24 months before we decided to drop these analyses from our study. The delay was due to an ongoing disagreement between NHS Digital and Healthcare Quality Improvement Partnership about formalising data-sharing arrangements for our study and similar studies. Follow-up data after discharge from HASU or hospital were collected in the SSNAP for a very small number of patients and it was not feasible to run this analysis.

Without these data our final outputs are weaker, but still important. We are still taking a unique approach to this problem by including within-hospital mortality data, mRS data and nursing rota data, by looking at patterns of temporal variation, rather than binary weekday or weekend patterns, and by incorporating interview and observational data.

Lack of access to carotid surgery data

The consultations about carotid surgery data were made with National Vascular Registry managers. According to the information provided, there was a lag of approximately 18 months between recording of carotid surgery data and their release (as of the end of January 2015 the carotid surgery data were available for linkage up to September 2013). In addition, the SSNAP data requested did not contain patient identifiers and it was not possible to assure the quality of linkage between the SSNAP and the National Vascular Registry data (Table 1).