Objective To determine whether patients treated in hospital on the weekend report different experiences of care compared with those treated on weekdays.
Design This is a secondary analysis of the 2014 National Health Service (NHS) adult inpatient survey and accident and emergency (A&E) department surveys. Differences were tested using independent samples t-tests and multiple regression, adjusting for patient age group, sex, ethnicity, proxy response, NHS trust, route of admission (for the inpatient survey) and destination on discharge (for the A&E survey).
Setting The inpatient survey included 154 NHS hospital trusts providing overnight care; the A&E survey 142 trusts with major emergency departments.
Participants Three cohorts were analysed: patients attending A&E, admitted to hospital and discharged from hospital. From the inpatient survey’s 59 083 responses, 10 382 were admitted and 11 542 discharged on weekends or public holidays. The A&E survey received 39 320 responses, including 11 542 (29.4%) who attended on the weekend or on public holidays. Weekday and weekend attendees’ response rates were similar once demographic characteristics were accounted for.
Main outcome measures For the A&E survey, six composite dimensions covered waiting times, doctors and nurse, care and treatment, cleanliness, information on discharge, and overall experiences. For the inpatient survey, three questions covered admissions and two dimensions covered information about discharge and about medicines.
Results People attending A&E on weekends were significantly more favourable about ‘doctors and nurses’ and ‘care and treatment’. Inpatients admitted via A&E on a weekend were more positive about the information given to them in A&E than others. Other dimensions showed no differences between people treated on weekdays or on weekends.
Conclusions Patients attending emergency departments or admitted to or discharged from an inpatient episode on weekends and public holidays report similar or more positive experiences of care to other patients after adjusting for patient characteristics.
- health services research
- patient-centred care
- patient satisfaction
- standards of care
Statistics from Altmetric.com
There has been significant focus in recent years on the association between weekend hospital admissions and patient outcomes, particularly mortality. Many studies have reported an increased likelihood of mortality for people admitted on the weekend compared with those admitted on weekdays.1–6 This is sometimes called ‘the weekend effect’.7
A statistical association between mortality rates and hospital care on the weekend or on public holidays has been well demonstrated and is broadly accepted, although the statistical approaches to adjusting mortality rates remain contested.8 Nevertheless, the ‘weekend effect’ has been a controversial area of study, chiefly because of the difficulty in determining the causes of heightened weekend mortality rates. There may be a complex set of differences in the types of patients who are admitted on weekends, and why; in the care they receive in hospital and the number and training of the staff providing it; and in the other health and care services that patients may or may not find themselves able to access. In other words, people admitted to hospital on the weekend may have higher mortality because they are sicker in the first place, possibly in ways that are difficult to measure or adjust for1; because there are fewer staff members and particularly fewer senior doctors available to manage care on weekends2 9; because there may be less access to diagnostic support services (radiology, laboratories and others)10; and so on. Untangling these potentially interacting explanations is important if policy responses are to be effective and proportionate.
There is now a large volume of literature on the ‘weekend effect’. Since Bell and Redelmeier’s2 2001 paper showing higher mortality for people admitted to hospital during weekends compared with during the week in Ontario, Canada, more than 100 studies have investigated the issue and a weekend effect at hospital-wide level has proven near ubiquitous.11 The magnitude of the effect and its causes remain controversial, however. As Bray and Steventon8 highlight in a recent review, numerous different causes have been posited, with conflicting evidence.
Despite the controversy around the causes of weekend mortality, the ‘weekend effect’ has had a significant impact on healthcare policy in England, contributing to a push for ‘seven day services’: that is, for hospitals to provide ‘the same level of consultant assessment and review, diagnostic tests and consultant-led interventions every day of the week’.12 Associated changes to healthcare professionals’ contracts have been proposed to enable this,13 and concerns about the new contracts led to an unprecedented series of strikes by junior doctors.14
To date, investigation of the ‘weekend effect’ in peer-reviewed publications has focused on patients’ health outcomes only, and particularly on mortality. These constitute a measure of clinical effectiveness, but other elements commonly accepted as components of healthcare quality—patient safety and patient experience15—have not yet been investigated. The NHS Services, Seven Days a Week Forum reported a weekend effect in patient experience, but provided no evidence to support this claim.13 A preliminary study based on the National Health Service (NHS) inpatient survey 2014 found some evidence of differences in experiences by weekday, but this was not conclusive.16 The study looked at six questions from the 2014 adult inpatient survey, finding that people admitted or discharged at the weekend reported more favourable experiences for four of these items, but concluding that worse experience on Mondays ‘supports the view of an extended weekend effect’ (p7). The differences in experiences of patients admitted at the weekend versus those admitted on weekdays are a significant gap in the current evidence on the ‘weekend effect’. Identifying whether and what differences exist is important because, as Black17 notes, mortality alone is not a sufficiently sensitive quality measure to address the widely held and instrumental view that quality of care is poorer on weekends.
Data from England’s NHS accident and emergency (A&E) department and inpatient surveys offer further opportunity to investigate this important issue. Both surveys are part of the Care Quality Commission’s national patient survey programme and involve large numbers of patients. Together they cover two cohorts whose responses are particularly relevant to addressing the question of whether patients’ experiences are different when treated, admitted or discharged during the weekend versus during the week: first, people attending emergency departments, who may or may not be admitted at the end of their episode; and second, inpatients. In this study, data from both surveys are used to investigate whether experiences differ for patients receiving hospital care on weekends and on weekdays.
A&E department survey 2014
The 2014 A&E department patient survey was sent to 120 658 patients treated at 142 NHS acute hospital trusts in the first 3 months of 2014. Patients were eligible for the survey if they were aged 16 or over and had attended a major A&E department—that is, a major or consultant-led 24-hour service with full resuscitation facilities. Exclusions included patients who had subsequently died, who had attended due to pregnancy loss and those known to be current inpatients.18
Participating organisations were able to choose whether they used January, February or March as their sampling window. Each organisation then drew a systematic (or ‘random start, fixed interval’) sample of 850 patients from an age/sex sorted list of all eligible patients attending the department that month. Where a person attended more than once in the month, each episode was left in the list and any duplicate patients were removed and replaced after the sample was drawn—this means that frequent attendees had a greater probability of selection, ensuring the sample was representative of the A&E population. The majority of patients (63 713; 52.8%) attended in March 2014. Responses were received from 39 320 patients, representing a 33.7% response rate once patients who had died or whose questionnaires were returned undelivered were accounted for.
Adult inpatient survey 2014
The 2014 acute inpatient survey was sent to 130 077 adult patients treated at 154 NHS acute hospital trusts. Each trust was able to select a starting month of either June, July or August 2014 and worked backwards from the last day of their chosen month to reach a sample of 850 consecutively discharged patients. Patients were eligible for the survey if they were aged 16 years or over and had spent at least one night in hospital as an NHS inpatient. Patients were excluded if they had died, attended hospital for obstetrics or maternity services, had had a psychiatric admission or were known to be a current inpatient.19
The vast majority of patients (120 968; 93.0%) were discharged from hospital in July or August 2014. Responses were received from 59 083 people, representing a 47.2% response rate once patients who had died or whose questionnaires were returned undelivered were accounted for.
A complicating factor in analysing the inpatient survey results is that it covers patients with a wide range of lengths of stay. Only a minority of patients will have had a single night stay commencing on a Saturday; all others will have experienced care both on the weekend and on weekdays. This means that analysis of people’s experiences needs to focus on questions related to admission and discharge processes. Two analyses of the inpatient survey data were thus undertaken: the first comparing the experiences of hospital admissions for patients admitted during the week and during weekends, and the second focusing on the experience of the discharge process and comparing patients depending on whether they were discharged during the week or on the weekend.
Both of the surveys recorded a range of demographic and health activity variables for participants, including non-respondents. These variables included age, gender, ethnicity and date of episode (or, for inpatients, dates of admission and discharge). For the inpatient survey, a ‘route of admission’ variable was created to distinguish patients who had an elective admission and those admitted via an emergency department: this coding was based on both survey responses and administrative data about individual patients. In the A&E survey, data from one question (Q34: ‘What happened at the end of your visit to the A&E Department?’) allowed identification of patients who were admitted at the end of their episode.
For each survey, data were recoded and variables created to identify patients attending, admitted or discharged on a weekend day or public holiday. For ease of reading, these groups are henceforth referred to collectively as ‘weekend patients’.
Reviewing patient profiles
For each survey, descriptive statistics were produced to explore the numbers of patients in the sample and responding to the survey who had a care episode taking place, beginning or ending on the weekend or a public holiday. Adjusted response rates (ie, the proportion of responses received out of all patients sampled minus those known to have died or for whom questionnaires were returned undelivered) for patients experiencing care at the weekend and on public holidays were then compared with those from normal weekdays, first through independent samples t-tests and subsequently through multiple regression. For the A&E survey, the profile of respondents was compared with published activity data20 to test its coverage, and a logistic regression was undertaken to determine which patient variables were related to people being admitted at the end of an A&E episode.
The demographic profile of weekend and weekday respondents for each survey was then reviewed to identify any differences in case mix. Demographic characteristics including age, sex, elective or emergency route of admission,21 22 limiting long-term conditions,23 and proxy-assisted survey response24 are known to be related to survey responses. Whether or not people were admitted to hospital at the end of an emergency department visit was also hypothesised to be potentially important to people’s experiences, as were trust-level admission and discharge patterns.
For the A&E survey, responses were analysed by dimensions identified following principal components analysis in a previous study that had sought to identify unidimensional subscales of patient experience with good measurement properties.25 This included six dimensions, covering people’s feedback of waiting times; communication with and confidence in doctors and nurses; care and treatment (including information, involvement and privacy); hygiene; information before discharge (for patients who were discharged at the end of their visit); and overall experiences. Two dimensions (‘hygiene’ and ‘overall’) were modified to take account of changes to the content of the survey questionnaire between the 2008 version analysed by Bos et al and the 2014 version used in the present study. Specifically, a question on the cleanliness of toilets was removed in 2014 as it was viewed by stakeholders as being of comparatively low importance. Another item on overall patient satisfaction was replaced with a standard ‘overall experience’ item used across a number of surveys; the latter is included in the analysis of the ‘overall’ domain. A list of items included in each dimension is presented in online supplementary appendix A. To calculate dimension scores, responses for each constituent item were assigned a score of 0–100 on a linear scale between 0 (least positive response option) and 100 (most positive). The dimension score was then calculated as the average of constituent item scores.
Supplementary file 1
For the inpatient survey, it was not practical to use existing dimensions or subscales because these include questions about the whole hospital stay.23 26 Instead, questionnaire items were reviewed to identify those relevant to admission and discharge. Three questions evaluated admission, but covered superficially different areas: two questions were specifically for patients admitted via emergency departments and asked about how much information they received and whether they were given enough privacy in A&E; the third was applicable to all respondents and asked about waiting times from arrival at the hospital to getting to a bed on a ward. Combining these items resulted in a scale with poor internal reliability (Cronbach’s alpha=0.511), so they were analysed individually. For discharge, it was possible to identify two groups of related questions on ‘information at discharge’ and ‘information about medicines to take at home’. As in the A&E survey, composite scores were created by taking the case-level mean of constituent items within each group; item lists are included in online supplementary appendix B and both scales had high internal consistency (Cronbach’s alpha >0.8). To minimise the impact of item non-response, scores were calculated only for patients who gave an evaluative response to at least 50% of items within the dimension.
Results for each of the dimensions and admission items were analysed first for crude differences in weekend and weekday experiences using independent samples t-tests. Multiple regression analyses were then conducted with patient age group (16–35, 36–50, 51–65 and >65 years), sex (male or female), ethnicity (white, mixed, Asian or Asian British, black or black British, Arab, and any other ethnic category), use of proxy response (self-completed or supported) and NHS trust included as fixed factors. Survey-specific variables regarding patient routes were also added as fixed factors: route of admission (emergency or planned) in the inpatient survey, and destination postdischarge (admitted or discharged) in the A&E survey. Adjusted mean scores for patients treated on weekdays and on weekends were calculated using IBM SPSS V.23’s EMMEANS subcommand, which estimates population means of the dependent variable for each level of the factor (weekend/weekday treatment) by adjusting for covariates in the model. Regressions were also repeated using individual weekdays instead of grouped weekday/weekend variables to check in case there were substantial differences by weekday masked within the main analysis.
A&E department survey 2014
Of the 120 658 patients included in the survey, date of episode was recorded for 120 657. Of these, 35 899 (29.8%) attended on a Saturday or Sunday, and a further 937 (0.8%) attended on a public holiday (New Year’s Day 2014). The remaining 83 821 (69.5%) attended on a normal weekday. A total of 39 320 patients completed the survey, including 11 833 (30.1%) who attended on a weekend or public holiday (see table 1).
The profile of sample members and respondents was similar to that recorded from Hospital Episode Statistics in terms of the day on which people attended. There was only minor variation in response rate by date of attendance, with people attending on a Friday significantly more likely to respond than those attending on Sundays (p<0.001), Mondays (p=0.002) or Tuesdays (p=0.037) (see table 2). Overall, there was a slightly but significantly lower adjusted response rate among patients attending on weekends or on public holidays (33.9% vs 33.3%; t(68 143.5)=2.156, p=0.031). However, there was no significant main effect of weekend attendance on response rates when age group, sex and ethnicity were accounted for.
As expected, there were significant differences in the likelihood of admission to hospital by demographic group (see figure 1). Younger people were far less likely to be admitted following an A&E visit, as were people with no limiting long-term conditions and people who completed the questionnaire without assistance. Patients attending on weekdays had a higher likelihood of admission than those attending on weekends, although the difference was not statistically significant.
When crude scores were compared, people attending on the weekend or on public holidays were significantly more favourable than those admitted on weekdays for four of the six dimensions: ‘doctors and nurses’, ‘your care and treatment’, ‘information before discharge’ and ‘overall’ (see table 3). The differences for the ‘information before discharge’ and ‘overall’ dimensions disappeared when confounding variables were accounted for in regression analyses. However, there was a significant main effect of weekend attendance for the ‘doctors and nurses’ (F(1)=6.83, p=0.009) and ‘your care and treatment’ (F(1)=11.03, p=0.001) dimensions: in both cases, people attending on the weekend were more positive about their experiences. A very similar pattern of results was found when rerunning the analysis to test the effect of individual weekdays instead of weekends.
Acute inpatient survey 2014
Date of admission was recorded for 130 074 of the 130 077 patients in the sample. Of these, 25 115 (19.3%) were admitted on a Saturday or Sunday, and a further 2322 (1.8%) were admitted on a public holiday (in 96.6% of cases, on 25 August 2014). The remaining 102 637 (78.9%) had weekday admissions. Of the 25 115 admitted on a weekend, only 3885 (15.5%) were discharged on the same weekend. The distribution of discharge dates was very similar: 102 681 (78.9%) were on weekdays, 25 194 on weekends (19.4%) and 2199 (1.7%) on public holidays. A total of 59 083 patients completed the survey, including 10 382 (17.6%) and 11 525 (19.5%) patients with a weekend/public holiday admission or discharge, respectively (see table 1).
The survey includes both patients who had a planned admission and those admitted following an emergency attendance, and there was significant variation by day of the week in the ratio of emergency to elective patients. On weekdays, an average of 67.4% of patients were admitted following an emergency attendance compared with an average of 89.7% on weekends.
Overall, patients admitted on a weekend were less likely to respond to the survey than those admitted on a weekday (43.3% vs 48.2%; t(36 441.3)=−13.711, p<0.001). However, patients admitted on the weekend were more likely to be aged 16–50 than 51+ (31.2% vs 30.0%; t(37 760.1)=3.580, p<0.001) and were far more likely to have had an emergency admission (89.7% vs 67.4%; t(56 751.2)=92.547, p<0.001). Regression analysis of response rates accounting for age group, weekend vs weekday admission, elective vs emergency admission, and patient ethnicity found significant main effects of all variables except weekend admission, so the crude difference in observed response rate for all weekend attendees is attributable to other differences in the profile of people admitted on weekends.
Comparing crude scores, people admitted on the weekend or on public holidays were significantly more positive about the information and privacy they were given in A&E, but reported significantly longer waiting times to get to a bed on a ward (see table 3). When route of admission, age, gender, ethnicity, long-term conditions, proxy response and NHS trust were taken into account, there was a significant main effect of weekend admission on the item about information in A&E. There was no significant main effect of weekend admission on people’s reports about privacy in A&E or waiting times, although there was a statistically significant interaction between day of admission and limiting long-term conditions (F(2)=5.276, p=0.003). This is illustrated in figure 2, which shows that people with no limiting long-term conditions tended to report shorter waits when they were admitted at the weekend, but those with multiple long-term conditions tended to report longer waits when admitted on the weekend.
Patients discharged on a weekend or holiday gave significantly higher average reports for questions about information on medicines (7.54 vs 7.43; t(14 163.8)=−3.135, p=0.002), but were not significantly different from other patients when asked about general information on discharge. Neither difference was statistically significant when other patient characteristics were accounted for (see table 3).
Repeating the regression analysis by individual days of the week produced a very similar pattern of results. Again, there was a significant main effect of weekday on patients’ experiences of information in A&E, but no significant effects on other items or dimensions.
People attending emergency departments, admitted to hospital or discharged from hospital on the weekend or on public holidays have similar or better experiences of care to those treated on weekdays once differences in patients’ demographic characteristics are accounted for. Patients attending A&E on weekends were significantly more positive in their responses to questions on ‘doctors and nurses’ and ‘your care and treatment’, while inpatients admitted on the weekend were significantly more likely to say they were given enough information in A&E than those admitted on weekdays. People admitted to hospital as an inpatient on the weekend or on public holidays tended to report shorter waits to get to a bed on a ward, but the opposite was true for patients who self-reported as having more multiple limiting long-term conditions.
This study is the first to look in depth at how patients’ experiences of emergency department and inpatient care differ on weekends and on public holidays. The analysis is based on a large volume of data collected systematically from two very large national surveys, and results reported here demonstrate that both surveys are similarly representative of weekday and weekend episodes. The findings contrast with the only previous study to look at differences in patients’ experiences on the weekend.16 That study, which was also based on analysis of the NHS inpatient survey, reported evidence of an ‘extended weekend effect’ in patient experience. However, it accounted for a more limited set of confounding variables than the current analysis, weighting only for age, sex and emergency or elective route of admission—and the same study showed that people’s experiences on Saturdays and Sundays were generally more positive than on weekdays. The current analysis more fully adjusts for potential confounders by including patient ethnicity and use of proxy response; moreover it takes a broader view of patients’ experiences of hospital care on the weekend by considering two large surveys covering both inpatient and emergency department care.
Despite the strengths of the data set, there are some limitations to the analysis. First, the surveys necessarily exclude patients who have died or who are an inpatient at the time of mailing, so the experiences of people with worse outcomes may be under-represented: if the standard of clinical care on weekends is markedly lower, then this may lead to greater exclusion of patients whose care occurred on the weekend. Data on the proportions of sample members reported to have died during the survey fieldwork period suggest this may have a small impact; patients admitted to hospital on the weekend were significantly more likely to be reported as having died during the survey (2.69% vs 2.25%; t(40 614.1)=4.040, p<0.001). However, patients discharged on the weekend were significantly less likely to be reported as deceased (1.86% vs 2.47%; t(48 448.8)=−6.403, p<0.001) and there was no difference between weekend and weekday A&E attendees (p=0.231). Second, the analysis of the inpatient survey includes both patients who had an elective or an emergency admission to hospital. Route of admission is controlled for as a variable in the analysis, but it is a limitation that the analysis is not stratified to investigate the experiences of elective and emergency inpatients as separate cohorts. This reflects the design of the survey itself, and retaining this grouping serves to allow the analysis to remain concise and readily interpretable, but it does mean that further research is needed to understand whether there are different patterns of experiences for inpatients with elective or emergency admissions. This will be relevant because much of the debate on the ‘weekend effect’ has focused on emergency admissions. Third, it is difficult to isolate people’s reports about inpatient care to experiences occurring on the weekend. We have sought to address this by focusing only on questions explicitly about admission or discharge processes, but even this may be limiting—for example, discharge planning conversations for patients discharged on the weekend may take place earlier, on a weekday. This approach also limits the analysis of people’s experiences because it means that many items that are not directly associated with admission or discharge are not tested. Fourth, although we have sought to account for measured factors that may explain differences in people’s experiences, it remains possible that there are other unmeasured characteristics that may influence the results: for example, condition severity. This possibility cannot be eliminated but, while residual confounding may remain, the inclusion of a broad set of patient characteristics minimises the risk of confounding due to characteristics already known to be associated with patient experience. Future studies could usefully investigate how a wider set of patient characteristics shape their experiences of weekend care, particularly because findings here, including the longer waits reported on the weekend by patients with multiple long-term conditions, hint at the potential for further differences related to unmeasured characteristics such as condition severity or the complexity of individuals’ care needs. Finally, the A&E survey investigated here includes only patients treated in major emergency departments, which are consultant-led departments with full resuscitation facilities. These are by definition 24-hour services, but may or may not meet the definition of ‘seven day services’ as defined earlier (ie, providing equivalent levels of consultant-led care, interventions and diagnostic services 7 days a week).12 Single specialty emergency departments (eg, ophthalmology and dental), walk-in centres and other minor injury units are not included, and it is possible that the experiences of patients treated in these units may differ on weekends.
This research builds on previous studies investigating the ‘weekend effect’ by looking at a previously untested element of service quality—patients’ self-reported care experiences. It does not support a hypothesis that patients experience a lower quality service on weekends. As in other studies, though, it demonstrates the complex relationship between the case mix of people attending and admitted to hospital on the weekend and population-level differences. In this study, crude differences were observed for a number of variables, but differences were largely accounted for by other factors.
If people receiving hospital care on the weekend report similar experiences to those cared for on weekdays, then this suggests that NHS hospitals do not provide lower quality care at the weekend in terms of patients’ experiences. Patients’ experiences are one component of quality, and other measures, including those related to clinical effectiveness and patient safety, must also be considered to reach a rounded view. It remains possible that some aspects of quality, such as safety and clinical effectiveness, are lower at weekends. Equally, a focus on 7-day services may further improve people’s experiences of hospital care on the weekend, but the present findings show that there is not, as has sometimes been claimed, adverse variance in the self-reported experiences of patients cared for on the weekend.13 Policy makers should consider and acknowledge this lack of variation in patient experience as part of the overall evidence base for planning and delivering 7-day services. Health service managers and clinicians should be wary of assuming that patients treated on the weekend are likely to have poorer experiences, as this does not appear to be the case, but there may be merit in giving particular attention to arrangements for people presenting with complex conditions and multiple long-term conditions on weekends.
Contributors CG conceived of the article, undertook the analysis and prepared the manuscript.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. The surveys used as the data source for the secondary analysis were funded by the Care Quality Commission and by participating organisations.
Competing interests CG is employed by the Picker Institute, a charity dedicated to promoting person-centred care. The Picker Institute is commissioned by the Care Quality Commission to coordinate surveys in the NHS Patient Survey Programme, including the two surveys analysed in this study. The Picker Institute is also contracted by a number of NHS hospital providers to administer surveys and collect responses.
Ethics approval This study used secondary analysis of existing data from two national studies. Both original surveys were reviewed and received favourable ethical opinions from NHS research ethics committees (see www.nhssurveys.org). The secondary analysis did not require ethical review.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Abridged data sets for each of the surveys are available via the UK data service.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.