Overview
In this 5-year programme of research, we used quantitative and qualitative methodologies to investigate why the risk of death seems to be higher for patients admitted to hospitals at weekends than for those admitted during the week. The answer turns out to be complex. Contrary to our initial hypothesis, we find no evidence that this situation is related to deficiencies in hospital medical staffing. We do find circumstantial evidence that the causes may lie in the community, upstream of hospital admission. Importantly, we find that, although the overall quality of hospital care in general is not worse at weekends, and improved between 2013/14 and 2017/18, the limited measures available to us suggest that community care is worse at weekends and seems to have deteriorated with time.
The 7-day services policy was – and is – a commendable package of measures intended to improve the overall quality of care for acutely ill patients and EAs to hospital, but the use of the weekend effect as an argument to effect change was unhelpful. As the weekend effect has been reported for 20 years from most health systems worldwide, the cause or causes needed to be identified to inform policy, rather than merely attributing it to deficiencies in hospital medical staffing.
Determining causation in health services usually relies on interpreting associations, strengthened by the use of instrumental variables, where these can be identified. We took the opportunity to use the introduction of 7-day services as an (uncontrolled) intervention to test the hypothesis that increasing the availability of hospital specialists (consultants, attendings) at weekends would mitigate the weekend effect. As this was, in effect, a ‘natural experiment’, we wished to strengthen the interpretation of the findings using triangulation: looking for similar signals from different and independent sources. We also wanted to ensure that we had adequate representation from those on the front line, namely patients and staff. We therefore used qualitative methods in addition to quantitative measures. We also examined changes over time. By comparing secular changes in weekend and weekday differences within individual trusts, we were able to reduce confounding from between-trust comparisons, for example in case mix, or service delivery and organisation.
Summary of main findings
Here we summarise our main findings and then consider their implications.
Emergency admissions and post-hospital mortality rates are increasing (2007–18)
Hospital specialist availability is not the cause of the weekend effect
Consultants are key clinical decision-makers from admission to discharge. Effective consultants contribute diagnostic accuracy and therapeutic specificity, provide visible on-site leadership for and support to junior doctors and other members of the multidisciplinary team, and promote timely hospital discharge.
The intensity of consultant input into the care of EAs at weekends across acute trusts in England is, on average, half that of weekdays.
The weekend-to-weekday specialist intensity difference has diminished (i.e. improved) in the last 2 years. This appears to be attributable to a modest increase in specialist hours throughout the 7 days, which has been masked by the proportionately greater increase in EAs, particularly on weekdays.
The difference in weekend-to-weekday specialist input is not associated with the difference in weekend-to-weekday admission mortality risk; the weekend effect is not linked to the availability of consultants at weekends.
There has not been a parallel increase in the numbers of non-consultant grade (junior) doctors to support the growing emergency workload.
Contextual factors influence local adoption of the 7-day services policy
Organisational slack: those trusts that had more resources and fewer infrastructure challenges exhibited greater flexibility and responsiveness to policy imperatives.
Trust ‘culture’ matters: visible and clear leadership, combined with a collaborative approach to listening to staff, responding to concerns and involving them in decision-making, promoted engagement with the policy.
‘Clan’ cultures inhibited clinician engagement with the policy in the absence of shared interpretation of the data used to identify the problem to which 7-day services was the proposed solution.
Community services that were poorly integrated with secondary care acted as a barrier to clinical and managerial staff trying to introduce the policy.
Increasing specialist intensity at weekends may be cost-effective by promoting timely patient discharge from hospital
Health economics modelling suggests that 7-day services would be cost-effective if weekend specialist intensity were to achieve parity with that currently provided on weekdays; however, the mechanism of benefit is reducing the length of hospital stay by promoting earlier discharge, not by influencing the care quality provided to EAs at weekends. Community services would need to be engaged for this effect to be fully realised.
Care quality of emergency admissions in hospital has improved over time, but may be deteriorating in the community
Patients and staff could identify deficiencies in hospital weekend care processes and quality for already-admitted patients and convalescent patients awaiting discharge back into the community, but considered that newly admitted emergency patients could receive more timely care at the weekend than those admitted on weekdays.
These perceptions are supported by evidence that hospital care processes for emergency admissions are more reliable at weekends than on weekdays. It is likely that, on weekdays, elective admissions (e.g. from clinics or operating theatres) compete with EAs for scarce resources.
Case record reviewer judgements of global care quality aggregated at trust level concurred with the independent observations of trust weekend care quality that qualitative researchers made on site. This form of triangulation strengthens the validity of our observations.
Error rates, error-related adverse event rates and global care quality are similar for weekend and weekday admissions; all have improved further with time. By contrast, indicators of community care quality (sicker patients, more patients have chronic disease, more patients require palliative care, fewer GP referrals preceding admission) are worse at weekends and have deteriorated further with time.
The causal pathway for the weekend effect includes community health care preceding hospital admission
Being admitted to hospital at a weekend is consistently associated with a 16% higher mortality rate relative to weekday admission in both UK and international literature.
The case mix of patients admitted to hospital at weekends differs from that of patients admitted on weekdays: patients admitted at the weekend are more severely ill, which accounts for most of the surplus mortality of weekend admission. They also have more comorbid conditions, are more likely to be candidates for palliative care, and are less likely to be discharged into the community before midnight on the day of their admission. These adverse features of weekend case mix had become more frequent by 2016/17.
Although the same numbers of patients are presenting to EDs at weekends and on weekdays, fewer patients are admitted at weekends. This contributes to the weekend effect by reducing the denominator of the weekend mortality rate.
The reduction in admissions is attributable partly to a reduction of two-thirds in the proportion of patients referred directly to hospital at weekends by their family doctor (GP). This reduction in GP referrals at weekends has become more marked over time.
Implications for policy and practice
The main implication of our research is that efforts to improve the care of patients who require EAs to hospital must be system-wide and occur across all days of the week. The focus on mortality associated with hospital admission has distracted attention from deficits in community health services at weekends; moreover, mortality is not a good indicator of care quality.129,164 Our findings provide evidence to support many of the recommendations in National Institute for Health and Care Excellence guideline 94 entitled Emergency and Acute Medical Care in Over 16s: Service Delivery and Organisation on emergency and acute medical care135 and are consistent with a recent systematic review of models of integrated care.165 They also provide further justification for the promotion of the development of integrated care systems in the NHS149 through Sustainability and Transformation Partnerships166 and acute and emergency care collaborations,167 such as the one in London that incorporates the development of seamless partnerships across social and primary and secondary health care.168 New research being commissioned by the NIHR Applied Research Collaboration West Midlands169 has a focus on integrated care and a subtheme of acute care. The emergency care pathways for frail elderly people and those with multimorbidity are an important target for further research.
Pre hospital
The adequacy and availability of community care at weekends need to be examined. The earlier application of more sophisticated interventions for acute illness in the community might prevent a proportion of hospital admissions. As proposed in National Institute for Health and Care Excellence guideline 94 entitled Emergency and Acute Medical Care in Over 16s: Service Delivery and Organisation,135 these interventions include paramedics who have enhanced competencies, acute hospital-at-home services, and enhanced advanced care-planning and palliative care provision to ensure that patients’ wishes are known and respected. This will involve access to data, as well as sharing information and patient care responsibilities 24/7 among the various agencies involved in community health care: general practice, community nursing, community palliative care, the emergency services and hospital-based community outreach services.
In hospital
Improvements in hospital care processes should be focused on all 7 days, not just the weekends; key decision-makers (usually consultants) need to be visible, and supported by managers and an effective multidisciplinary team using standardised and structured systems of care to optimise safety and efficiency. An expansion in consultant numbers alone is unlikely to solve a system-wide problem. The ease with which new ways of working are adopted is enhanced by a collaborative local culture, which is set by the attitudes and behaviours of senior leaders (trust executives, managers, senior clinical staff).
Post hospital
The timely discharge of a patient from hospital involves a (multidisciplinary) judgement about the impact that the transition from a supported to a less supported or unsupported environment will have on the patient’s current health status. Rehabilitation starting in hospital may need to be continued after the patient’s transfer back into the community to ensure that efforts to promote timely discharge are invested effectively. Hospital-at-home services and physiotherapy have the potential to bridge the gap between secondary and community care, but the effectiveness of these interventions will only be realised with the integration of information and care processes throughout the patient pathway.
Recommendations for future research
The global burden of emergency medical diseases has been estimated to cause 28.3 million deaths and to contribute to 50.7% of all deaths worldwide.170 The need for a system-wide integrated approach to improving outcomes through research is well recognised in the UK.169,171 The HiSLAC study demonstrates that signals from performance indicators conventionally attributed to hospital practice may originate in the community, both before and after discharge from hospital. Our research recommendations, therefore, take this wider view of exploring quality of care throughout the patient pathway.
How does decision-making by senior doctors differ from less experienced or non-physician staff to promote effective and timely patient-centred care?
The HiSLAC study has shown that hospital specialists are regarded as key decision-makers from admission to discharge, influencing not just the quality of care but also patient ‘flow’, and that they are regarded as more effective than junior doctors or other professionals. The health economics model indicates that increasing specialist input at weekends may be cost-effective, primarily by promoting the timely discharge of patients from hospital. It might, therefore, have been expected that specialists would have a shared mental model when evaluating the quality of care of individual patients from case records; this is true in aggregate, but not at the level of individual patients, as we have shown in the case record review. Therefore, we would propose studies that examine medical decision-making in relation to experience, professional status, values and beliefs, combining qualitative research observations of practice with ‘think-aloud’ experiments.172 This should involve not just hospital specialists, but, crucially, also GPs. The outputs of such research would help to inform the development of both physician and allied health professional training programmes, and promote the efficient transit of patients through the health system.
Is there a causative association between the lower rates of general practitioner emergency referrals to hospital at weekends and the weekend effect?
The reduction in the proportion of GP referrals at weekends has become more marked over time. Possible causes include the introduction of the 111 urgent call service, although this should have had the same influence on weekdays as at weekends; the progressive withdrawal of GPs from providing emergency care services; GPs instructing patients who contact them urgently to call the emergency number (999) for an ambulance if they themselves are overloaded with emergency calls; or GPs failing to act as ‘filters’ of EAs to hospital specifically at weekends, thereby permitting an increase in the proportion of patients admitted who have palliative care decisions or multimorbidity. Exploring these possibilities requires the evaluation and linkage of primary care and HES data sets, direct observation of emergency practice at weekends and on weekdays, and qualitative research involving practitioners and patients.
What is the cost-effectiveness of integrated care initiatives focused on preventing acute deterioration in the community?
The HiSLAC study has shown that, compared with those admitted on weekdays, patients admitted at weekends are more severely ill, have more comorbid diseases, have a higher likelihood of receiving palliative care, are less likely to be discharged home on the day of admission and are more likely to be transferred directly to intensive care following hospital admission. This raises the question of whether or not opportunities to intervene earlier in the community are being missed. A number of initiatives are already being implemented in England that partially address this issue; some of the 50 vanguard sites173 are focused on improving acute and emergency care, with the aim of reducing avoidable ED attendances and admissions, but evaluation of these has so far been incomplete174 and at least one intervention, a new hospital specifically for emergency care, has resulted in an increase in ED attendances.175 NHS England has funded seven ‘accelerator’ sites since April 2020 to establish rapid response teams intended to respond within 2 hours to acute deterioration in the health of older people and those with complex care needs to enable them to remain at home instead of being transferred to hospital.176 There are no details at present on how these patients might be identified and assessed; acute physiology (vital signs, NEWS) monitoring by health-care professionals (whether directly through more frequent visits or using telemedicine) is feasible but the impact is uncertain.177 This important initiative provides an opportunity for a collaborative mixed-methods evaluation. The work would be complemented by international comparisons of other health systems to determine how social and primary care structures and processes are engaged in responding to the deteriorating patient in the community.
Is the increase in post-discharge mortality rates since 2007 linked to the increase in emergency admission rates of frail elderly people and those with multimorbidity?
Are these avoidable deaths, part of planned palliative care, or appropriate but unexpected deaths? Could the care of these patients be improved by more intensive rehabilitation before discharge or by more intensive support in the community following discharge? Are the admissions preventable in the first place, through new ways of delivering care in the community or by better sharing of information about end-of-life care preferences, or does active medical care need to be expedited and more timely during ‘off hours’ such as weekends? During the patient journey, before admission or after discharge, are there gaps or deficiencies in data access, discussions about risks and benefits of treatment, detection of acute deterioration when this occurs, decision-making by informed and competent individuals, discharge decisions or destination following discharge, and, if so, how might these gaps or deficiencies be addressed?
Conclusions and lessons learned
The HiSLAC study has shown that the weekend effect is most likely to be a system-wide pathway problem that has its origins in community services. Contributory factors in the community include budgetary constraints during the period of austerity that followed the financial crisis of 2008, and the progressive increase in health service utilisation, accompanied by persistent demands for greater efficiencies and reductions in hospital length of stay. The improvements in hospital safety and care quality between 2012/13 and 2016/17 contrast with the deterioration in community care metrics. Although direct comparisons of secondary and primary care were not possible in this study, the evidence we have presented here is consistent with the hypothesis that the closer integration of social, primary and secondary care will provide benefits to patients.178,179
The marked reduction in hospital errors and adverse events, and the improvement in global quality identified between epochs by the HiSLAC case record reviewers has parallels with the MERIT study,180 the Safer Patients’ Initiative116,125 and the Matching Michigan study;181 it has been described as the ‘rising tide’ phenomenon.124 The reduction in errors and adverse events demonstrated in HiSLAC could be a manifestation of a non-specific long-term secular trend, but it is also consistent with being driven by the salience of the 7-day services policy, in particular the six standards that require increased consultant involvement in the acute-care pathway. If this were the case, then the 7-day services policy should be regarded as having benefited patients across all days of the week.
Lessons learned
We focused our attention primarily on in-hospital care because this was the main focus of the 7-day services policy. However, an inspection of , which shows the possible mechanisms for generating the weekend effect, draws attention to the potential contribution of variation in community health service provision. It would have been helpful to have included community services in the HiSLAC project from the start.
Negotiations to obtain standard access to HES through NHS Digital were extremely prolonged and could have compromised the timely conclusion of the entire project. We hope that new processes at NHS Digital will accelerate access for research and permit better data linkage between secondary and primary health-care data sets.
The 20-hospital case record review (see Chapter 6) required each trust to obtain, copy, redact and digitally convert 200 case records. The time it took for different trusts to complete this task varied considerably, although funding was available for this purpose and site visits were undertaken with all those involved in facilitating access to case records for research purposes. The difference between trusts seemed to be related to local leadership, staff engagement and competing priorities. Encouraging local teams to provide the case records consumed a considerable amount of the project team’s time and resources, and delayed completion of this phase. The gradual transition in the NHS from paper to electronic case records should facilitate research, but, at present, the variation in local clinical information systems does not necessarily make the task of collation easier.
To conduct a research programme across the NHS in England for 5 years, we had to gain the support of senior leadership centrally, locally in each trust, and at the level of individual clinicians. We were fortunate to have the support of the Academy of Medical Royal Colleges and individual Royal Colleges and faculties, and of NHS England and NHS Confederation. The majority of trust chief executives and medical directors were supportive, and we were grateful to them and to each local project lead for the time they gave to the project, as well as to each consultant who completed the annual survey. We maintained contact with this diverse group through a newsletter and social media, but we were well aware of the competing demands on their time, not least in implementing the 7-day services policy.
