Article Text
PDF
Abstract
Background and Objective Several studies report poorer quality healthcare for patients presenting at weekends. Our objective was to examine how timely surgery for patients with hip fracture varies with day and time of their presentation.
Methods This population-based cohort study used 2017 data from the National Hip Fracture Database, which recorded all patients aged 60 years and over who presented with a hip fracture at a hospital in England, Wales and Northern Ireland. Provision of prompt surgery (surgery within 36 hours of presentation) was examined, using multivariable logistic regression with generalised estimating equations to derive adjusted risk ratios (RRs). Time was categorised into three 8-hour intervals (day: 08:00–15:59, evening: 16:00–23:59 and night: 00:00–07:59) for each day of the week. The model accounted for clustering by hospital and was adjusted by sex, age, fracture type, operation type, American Society of Anesthesiologists grade, preinjury mobility and location.
Results We studied 68 977 patients from 177 hospitals. The average patient presenting during the day on Friday or Saturday was significantly less likely to undergo prompt surgery (Friday during 08:00–15:59, RR=0.93, 95% CI 0.91 to 0.96; Saturday during 08:00–15:59, RR=0.91, 95% CI 0.88 to 0.94) than patients in the comparative category (Thursday, during the day). Patients presenting during the evening (16:00–23:59) were consistently significantly less likely to undergo prompt surgery, and the effect was more marked on Fridays and Saturdays (Friday during 16:00-23:59, RR=0.83, 95% CI 0.80 to 0.85; Saturday during 16:00–23:59, RR=0.81, 95% CI 0.78 to 0.85). Patients presenting overnight (00:00–07:59), except on Saturdays, were significantly more likely to undergo surgery within 36 hours (RR>1.07).
Conclusion The provision of prompt hip fracture surgery was complex, with evidence of both an ‘evening’ and a ‘night’ effect. Investigation of weekly variation in hip fracture care is required to help implement strategies to reduce the variation in timely surgery throughout the entire week.
- surgery
- adverse events, epidemiology and detection
- healthcare quality improvement
- hospital medicine
Data availability statement
Data may be obtained from a third party and are not publicly available. Data were obtained from the National Hip Fracture Database, and permission to use their data must be sought (www.nhfd.co.uk).
Statistics from Altmetric.com
Introduction
Numerous studies of the ‘weekend effect’ have suggested that patients receive poorer quality care and may have worse outcomes if they are admitted at weekends and/or outside routine working hours.1–5 These studies have led to substantial changes in how weekend healthcare is delivered,6 7 though this may be premature, given the lack of understanding about why weekend admission might have poorer outcomes.8
Work describing a nationwide registry study of acute stroke patients observed complex variation in the quality of care provided across the week, suggesting that the 'weekend effect' is an oversimplification of the true problem.9 A key issue is that for many conditions, the patients who present as emergencies at the weekend may be very different from those who are able to wait or who are admitted semielectively during the week.
Patients presenting with hip fracture are all classified as emergencies and should not differ by day and time of presentation, unlike other medical conditions. Thus, this is an ideal condition with which to examine the 'weekend effect'. It is common, an incidence of around one in a thousand means that 70 000 people present in the UK each year, at a cost to the National Health Service and social care of over £1 billion.10 11 Patients require emergency hospital admission for trauma; almost all undergo surgery within a few days; and mortality is substantial within a frail and elderly group of people.
The national clinical audit of hip fractures was established in England and Wales in 2007 with the aim of improving outcomes. The programme included a National Hip Fracture Database (NHFD) in England, Wales and Northern Ireland, which promotes high-quality care as defined in national guidance.12 In 2010, the NHFD was the basis for a pay-for-performance initiative, called the ‘Best Practice Tariff’ (BPT). The BPT scheme pays hospitals a supplement of £1335 per patient whose care satisfied six clinical standards, such as surgery within 36 hours of presentation, as this is known to improve outcome.12–15 The success of this means that this national clinical audit has shown its value by being used to document a reduction in 30-day mortality for patients with hip fracture from 10.9%, when it was established in 2007, to just 6.1% in 2016.10 16 17 This is despite patient case mix getting progressively worse over time.
We used NHFD data to define how patients’ chances of receiving prompt hip fracture surgery varied across the hours of the day and days of the week.
Methods
Study design and data sources
We performed an analysis of prospectively collected NHFD data. The NHFD was established in 2007 and captured more than 97% of all hip fractures in people aged 60 years and above in England, Wales and Northern Ireland.10 Data were collected on patient characteristics, including the type of fracture, the surgery performed, details of the care they receive and outcomes, which included 30-day mortality. These data were collected by the staff who provided care to patients with hip fracture locally and were used to assess each hospital’s performance against established evidence-based quality performance indicators, including time from first presentation to operation (surgery within 36 hours).
Participants and procedures
We studied an anonymised extract of data for all 70 573 patients who presented at one of 177 hospitals in England, Wales and Northern Ireland between 1 January 2017 and 31 December 2017. The NHFD collects precise data on their operation. Contralateral hip fractures in the same patient were considered independent events. Less than 3% of patients were excluded as their date and time of their presentation (n=6) or of their operation were not available (n=1590).
The primary exposure of interest is the day and time of patient presentation, meaning admission to hospital. A number of key performance indicators have been defined for hip fracture,11 but in this study, the outcome of interest was whether a patient received prompt hip fracture surgery (an operation starting within 36 hours of presentation). This is an important indicator of care quality derived from national guidelines, against which hospitals are formally assessed.12 18
Covariates chosen a priori to control for as potential confounding factors19 included patient age; sex; location from which the patient was admitted (‘own home or sheltered housing’ or ‘not their own home’); preinjury mobility (‘freely mobile without aids’, ‘mobile outdoors with one aid or two aids or one frame’, and ‘some indoor mobility or no functional mobility’); fracture type (intracapsular, intertrochanteric and subtrochanteric); type of surgery performed; and American Society of Anesthesiologists (ASA) grade. The ASA grade ranges from 1 (healthy patient) to 5 (moribund patient not expected to live for more than 24 hours with or without surgery).20
The type of hip fracture that a patient has will influence the type of surgery that they receive, such as a partial or total hip replacement, or fixation with plates, screws and rods. This may influence time to surgery based on availability of suitably experienced hip surgeons to perform different types of operations. Although fracture type and type of surgery are associated with each other, the association is not strong enough to lead to collinearity. There is variation between surgeons and units such that all types of operations have been used for all types of fractures, even if some of these are in smaller numbers.
Statistical analysis
Descriptive statistics were used to summarise the epidemiology of hip fractures by day and time of presentation, and relevant patient and surgical factors.
Analyses were performed according to the day and time of patient presentation. The effect of time on the outcome of ‘receiving surgery within 36 hours’ was found to be non-linear. Hence, we were not able to treat time as a continuous variable in the model. Time was categorised into three 8-hour intervals (day: 08:00–15:59, evening: 16:00–23:59 and night: 00:00–07:59) for each day of the week. This provided 21 different categories over the whole week. These time periods were chosen based on clinical relevance relating to when patients present and undergo surgery.
A generalised linear model was used to assess the effect of the 21 8-hour periods of the week, and the other covariates (described earlier) on the risk of patients undergoing surgery within 36 hours of admission. All predictors were assessed in univariable models, and then an analysis was conducted using a generalised linear model with a binomial error structure and log link function (log-logistic model) in order to estimate the relative risk ratio (RR) and risk difference (RD) using an identity link function. The model accounted for clustering by hospital with generalised estimating equations using a population average approach.21 We were interested in an average effect for the entire population rather than an effect to be communicated to patients, which would require a random effects model.
Given that the outcome of interest is time to surgery, we thought that the competence of the surgeon would be an unlikely confounder, but that surgeon availability might be (eg, to undertake a hip replacement). Clustering by hospital was chosen because when a surgeon is not available, either the hospital will wait for a surgeon to become available or the hospital may transfer the patient. This will add differential delay between hospitals, with hospitals with lower availability of replacement surgeons fairing worse for the outcome of interest.
All analyses were conducted using STATA V.14.2.
Patient and public involvement
Patients and members of the public were not directly involved in the design of this research project, but a patient panel managed by the Royal Osteoporosis Society advises on the approach taken by the NHFD and other projects within the Falls and Fragility Fracture Audit Programme at the Royal College of Physicians.
Results
There were 68 977 patients with hip fracture eligible for inclusion (table 1). The mean age was 82.7 (range 60–109) years, and 48 798 (71%) were women. The percentage of patients presenting each day nationally was similar (14% or 15% each day) during Monday–Saturday, but slightly fewer were admitted on a Sunday (13%). Most presented during the ‘day’ (42% during 08:00–15:59 and 41% during the ‘evening’ 16:00–23:59), with the remainder admitted at ‘night’ (16% during 00:00–07:59).
Baseline characteristics of patients with hip fracture treated in England, Wales and Northern Ireland during 2017, as recorded in the National Hip Fracture Database
The mean time between presentation and surgery was 34.0 (SD=41.6) hours (table 1). The median was 24.4 hours; the 25th centile was 18.6, the 75th centile was 40.3, the 90th centile was 60 and the 95th centile was 160.1. Overall, 70% (n=48 441) of all patients underwent surgery within 36 hours. The overall percentage of patients undergoing prompt surgery nationally ranged from 60% to 80%, depending on the specific day and time of the week. Fewer patients underwent prompt surgery if they presented on a Friday (67%) or Saturday (65%) compared with other days of the week (72%). Fewer patients underwent prompt surgery if they presented in the evening between 16:00 and 23:59 (66%) compared with being admitted during the day between 08:00 and 15:59 (71%) or overnight between 00:00 and 07:59 (78%).
A generalised linear model, which accounted for clustering by hospital, involving all covariates based on 67 048 hip fractures with complete data available, demonstrated that both day and time of presentation affected the time to surgery for the average patient presenting at a certain day or time (table 2). The average patient presenting during the day on Friday or Saturday was significantly less likely to undergo prompt surgery than patients in the comparative category (Thursday, during the day) (Friday, during 08:00–15:59, RR=0.93, 95% CI 0.91 to 0.96; Saturday during 08:00-15:59, RR=0.91, 95% CI 0.88 to 0.94). The average patient presenting during the evening (16:00–23:59) was consistently significantly less likely to undergo prompt surgery, and the effect was more marked on Fridays and Saturdays (Friday during 16:00-23:59, RR=0.83, 95% CI 0.80 to 0.85; Saturday during 16:00-23:59, RR=0.81, 95% CI 0.78 to 0.85). In absolute terms, this is equivalent to a RD of approximately 13%–14% compared with Thursday daytime (Friday during 16:00–23:59, RD=−0.132, 95% CI −0.15 to −0.11; Saturday during 16:00-23:59, RD=−0.141, 95% CI −0.17 to −0.11). The number needed to treat is eight, which means that one additional person does not undergo prompt surgery for every eight patients who present on Friday or Saturday evening. The average patient presenting overnight during 00:00–07:59, except on Saturdays, was significantly more likely to undergo surgery within 36 hours (RR>1.07).
Variation in patients receiving surgery within 36 hours of presentation by day and time of the week with a hip fracture during 2017 in England, Wales and Northern Ireland, as recorded in the National Hip Fracture Database
Supplemental material
Factors associated with a lower likelihood of prompt surgery were male sex, age under 80 years, ASA grades 4 and 5, patients who were not freely mobile without aids, and those undergoing total hip replacement (online supplementary table 1).
Discussion
There were complex patterns in timely surgery across the week, with evidence of a day effect (08:00-15:59) on Fridays and Saturdays, a consistent ‘evening effect’ (16:00–23:59) that was more marked on Fridays and Saturdays, and a ‘night effect’ (except on Saturdays) on performance.
The reasons for this temporal variation are likely to be multifactorial, and our findings suggest that current attempts to increase weekend services and staffing will not fully address variation in prompt surgery for patients with hip fracture.
Our confirmation of this temporal variation in timely surgery at a national level would suggest that care quality in other acute medical and surgical conditions should be investigated in a similar fashion.
We also observed other factors which influenced timely surgery. Patients were less likely to undergo hip fracture surgery within 36 hours of presentation if they were sicker (ASA grades 4 and 5) and frailer (not freely mobile without aids). This is unsurprising as such patients often need more preoperative care and optimisation before they are considered suitable to undergo major surgery. Total hip replacement is generally performed in younger patients with intracapsular fractures and is offered as an alternative to hemiarthroplasty.22 However, this is more technically demanding to perform than the other types of hip fracture surgery, and as such, there have been substantial variations in its use across the country in eligible patients.23 When total hip replacement is offered, the requirement of appropriate surgical expertise can result in delays before performing surgery. Therefore, this may explain our findings that both age under 80 years and patients undergoing total hip replacement were less likely to undergo timely surgery.
Strengths and weaknesses
This study includes nearly every patient presenting with hip fracture in England, Wales and Northern Ireland. We have used analytical methods to model temporal variation in timely surgery in patients with hip fracture by the day and time of the week that accounted for clustering by hospital. Since the NHFD was designed to capture relevant covariates for the hip fracture population, we were able to make appropriate adjustments in these analyses.
Time to theatre is a key indicator of care quality for hip fracture surgery, but individual hospitals may perform differently for other indicators of care quality, such as prompt postoperative mobilisation. Future work will explore these other indicators. Missing data for some variables may have influenced the findings, but the proportion of missing data were small (2% or less). Our statistical models were adjusted for important covariates relevant to the hip fracture population, but we were unable to adjust for other potentially important variables (like specific medical comorbidities) or unknown confounders.
We do not have data on time of fracture, prehospital delay or the total delay from time of fracture to surgery. It is possible that prehospital delay may vary by time of presentation; for example, patients experiencing a fracture at night waiting longer before they present at the hospital. A strength of the NHFD is that unlike other data sources, day and time of admission are accurately recorded as required by the BPT standards.
Comparison with other studies
Many studies have suggested that patients admitted at weekends and/or outside of routine working hours receive poorer quality healthcare.1–5 24–27 However, these analyses have often simplified the problem by directly comparing weekday with weekend care, or normal working hours with out-of-hours. This makes comparison and interpretation of the literature difficult. Studies with more sophisticated methodology have demonstrated temporal patterns of care quality across both the day of the week and time in acute stroke patients,9 all hospital inpatients28 and in obstetrics.29
Prompt surgery for hip fracture is a better test of the weekend effect since similar patients will present across the week, unlike other conditions for which patients may be admitted semielectively or for investigation during the working week, and only as an emergency at the weekend.
Studies have shown that, compared with weekday admissions, patients presenting with hip fracture at weekends have higher mortality30 and longer delays to surgery.31 Furthermore, there is evidence of increased mortality in patients with hip fracture having surgery on a Sunday.14 A study from the Netherlands found no weekend effect in outcomes when simply comparing patients presenting during the week with those presenting at the weekend.32 However, to our knowledge, there are no studies assessing the temporal variation in this care quality indicator across both the day and time of the week.
The 'evening effect' and 'night effect' on timely surgery observed in our study was similar to that previously reported in the acute stroke population, with patients admitted overnight and at the weekend less likely to receive timely thrombolysis.9 Like thrombolysis, receiving hip fracture surgery in a timely fashion is a complex process relying on many steps in the care pathway to occur swiftly and effectively. These include initial assessment, diagnostic imaging, preoperative optimisation, and operating theatre capacity and staff availability. Poorer provision of prompt surgery for patients who present in the evening or at weekends is likely to be related to problems with at least one of these milestones. Individual ‘heat maps’ for each hospital have been published by NHFD on its website (www.nhfd.co.uk), which will allow local teams to understand where they are failing to provide a prompt and coordinated response. The presence of key decision makers, may have the greatest potential for correcting weaknesses at certain times during the week.9
Given the complex nature of patients with hip fracture, it is recommended their management is coordinated by senior experts using a multidisciplinary approach, including orthopaedic surgeons, orthogeriatricians and anaesthetists.10 12 Availability of key professionals may be reduced at weekends and overnight, and may contribute to poorer quality healthcare for patients with hip fracture during these times.
Potential explanations for delays in patients undergoing hip fracture surgery also relate to theatre capacity and efficiency. The NHFD audits reasons for surgical delays, and in 2017, they reported that although surgical delays for clinical reasons (eg, needing medical investigation and optimisation) remained stable compared with previous years, delays due to lack of capacity on theatre lists and list over-runs have increased from 13.2% in 2016 to 14.4% in 2017.10 This is compounded by frequent reports of theatre lists routinely starting late, avoidable cancellation of individual patients and 61% of hospitals not having a dedicated hip fracture operating list, meaning such cases would need to be rationed with other trauma operations and/or other surgical emergencies.10 23 Financial incentives may also influence time to theatre, as hospitals do not receive the BPT if surgery is performed after 36 hours. This includes all hospitals in Wales and Northern Ireland who are not offered this financial incentive. Therefore, new hip fracture admissions may be prioritised for theatre over patients who have already waited more than 36 hours.
The NHFD annual report 2019 has shown there is substantial variation observed in time to hip fracture surgery between hospitals, ranging from 14% and 95%.33 Our study and the work of the NHFD clearly demonstrate that many hospitals are delivering prompt hip fracture surgery throughout the week. Important lessons can be learnt from better performing hospitals.
The entire pathway from admission to surgery should be scrutinised and include emergency department care, preoperative assessment and optimisation, and logistics around operating theatre capacity, as well as understanding the number and expertise of the multi-disciplinary healthcare team involved in the care of patients throughout the entire week.
Aside from the direct performance financial incentive to perform timely surgery described earlier, it is clear theatre capacity and efficiency that is becoming increasingly problematic.10 Suggestions to improve this include dedicated daily hip fracture lists, extra daytime lists and/or lists with extended hours and formally categorising hip fracture as an emergency when competing with other specialties for space on emergency theatre lists. It is currently recommended that hip fracture surgery occurs during normal working hours by senior experienced staff.34 However, the evidence of higher mortality with delayed surgery,13–15 coupled with the increasing problems with patient demand versus theatre capacity, means that hospitals need to examine their dedicated trauma theatre capacity and, if required, increase sessional availability into evenings and across weekends.
Meaning of the study
There are complex temporal variations in the provision of timely surgery to patients with hip fracture on a national level. The variation highlights there are many opportunities available at all levels (local, regional and national) to substantially improve the delivery and quality of care for patients with hip fracture.
Unanswered questions and future research
We recommend urgent investigation of this variation in timely hip fracture surgery to help implement strategies to reduce the variation through the entire week. We suggest that focussing only on improving time to surgery at the weekend will not fully address the problems during the entire week. We recommend that similar analyses are performed to assess the temporal variation in care quality for other acute medical and surgical conditions.
What is already known on this topic
In patients with hip fracture, many studies have assessed and demonstrated that prompt surgery (within 36 hours or less from presentation) significantly reduces patient mortality.
The evidence regarding a true ‘weekend effect’ in mortality following hip fracture surgery is contentious. Temporal variation has been investigated in this manner in other patient cohorts, including obstetrics and all hospital inpatients.
A recent study in patients with acute stroke observed that care quality varied across the whole week and suggested that the weekend effect’ was only one of several patterns of variation in quality that occur in clinical practice.
What this study adds
We have demonstrated that the patterns in variation of timely surgery for patients with hip fracture were complex throughout the week, with evidence of both an ‘evening effect’ and a ‘night effect’, with unusual patterns on Fridays and Saturdays.
Data availability statement
Data may be obtained from a third party and are not publicly available. Data were obtained from the National Hip Fracture Database, and permission to use their data must be sought (www.nhfd.co.uk).
Ethics statements
Ethics approval
Research ethics committee approval was not required for secondary analysis of administrative data in line with Governance Arrangements for Research Ethics Committee guidance.
References
Footnotes
Contributors AJo and AJu conceived and designed the study. AS cleaned the data, did the statistical analyses, interpreted the findings and drafted the paper. GSM interpreted the findings, added clinical commentary, performed the literature review and drafted the paper. All authors, including DI and EF, contributed to the interpretation of the data and findings, and revised the manuscript. All authors approved the final manuscript submitted.
Funding This work was performed to provide individual hospitals participating in the NHFD with a picture of variation in their performance across the week, as part of a Royal College of Physicians audit programme commissioned by the HealthCare Quality Improvement Partnership (HQIP). The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the NHS or the Department of Health. The funding source had no role in the design and conduct of the study, in the collection, analysis and interpretation of the data, or in the preparation, review or approval of the manuscript. The corresponding author had full access to all the study data and had final responsibility for the decision to submit for publication.
Competing interests GSM has received financial support for other research work from Arthritis Research UK, the Orthopaedics Trust, Royal College of Surgeons of England and Royal Orthopaedic Hospital Hip Research and Education Charitable Fund. GSM has also received personal fees for undertaking medicolegal work for Leigh Day. AJu has received consultancy fees from Freshfields, Bruckhaus, Deringer and Anthera Pharmaceuticals Inc.
Provenance and peer review Not commissioned; externally peer reviewed.