Assessment on patient outcomes of primary hip replacement: an interrupted time series analysis from ‘The National Joint Registry of England and Wales’

Objectives Effects of the UK Department of Health’s national Enhanced Recovery After Surgery (ERAS) Programme on outcomes after primary hip replacement. Design Natural experimental study using interrupted time series to assess the changes in trends before, during and after ERAS implementation (April 2009 to March 2011). Setting Surgeries in the UK National Joint Registry were linked with Hospital Episode Statistics containing inpatient episodes from National Health Service trusts in England and patient reported outcome measures. Participants Patients aged ≥18 years from 2008 to 2016. Main outcome measures Regression coefficients of monthly means of length of hospital stay, bed day cost, change in Oxford Hip Scores (OHS) 6 months post-surgery, complications 6 months post-surgery and revision rates 5 years post-surgery. Results 438 921 primary hip replacements were identified. Hospital stays shortened from 5.6 days in April 2008 to 3.6 in December 2016. There were also improvements in bed day costs (£7573 in April 2008 to £5239 in December 2016), positive change in self-reported OHS from baseline to 6 months post-surgery (17.7 points in April 2008 to 22.9 points in December 2016), complication rates (4.1% in April 2008 to 1.7% March 2016) and 5 year revision rates (5.9 per 1000 implant-years (95% CI 4.8 to 7.2) in April 2008 to 2.9 (95% CI 2.2 to 3.9) in December 2011). The positive trends in all outcomes started before ERAS was implemented and continued during and after the programme. Conclusions Patient outcomes after hip replacement have improved over the last decade. A national ERAS programme maintained this improvement but did not alter the existing rate of change.

2) Please offer a more detailed description of ERAS for those of us that do not work in the NHS. Include reasons for its implementation, as additional justification for the outcomes you have chosen. Could the ERAS have strengthened the changes that were already underway, or was it a non-entity?
3) Can you offer an explanation for why you excluded patients with an LOS >15 days? Why not 5 days? Was this an arbitrary cut-off? 4) Can you expand on additional factors that may have changed over the study period that may have affected outcomes? Changes in surgical approach or patient demographics, etc?

REVIEWER
THOMAS WAINWRIGHT Bournemouth University UK REVIEW RETURNED 25-Jul-2019

GENERAL COMMENTS
Thank you for inviting me to review this work. The authors have done an excellent job and I think that this is an important research question which the authors have answered thoroughly. I have one point that I feel is important to make clear and to be addressed, in both the methods, and in the discussion on outcomes. During the time period studied, there has been an increasing amount of NHS hip replacements performed in non-NHS settings as either waiting list initiative work, or through patient choice and the AQP route. Whilst these pts may still be registered on the NJR they will not to my knowledge appear in HES statistics. Please can the authors be explicit on how this has been handled within the data, and also consider the effects on outcomes that this may of had in the discussion. If this point is addressed I would recommend for publication. 2) Please offer a more detailed description of ERAS for those of us that do not work in the NHS. Include reasons for its implementation, as additional justification for the outcomes you have chosen. Could the ERAS have strengthened the changes that were already underway, or was it a non-entity? Author response: ERAS is a new approach to the preoperative, intraoperative and postoperative care of patients undergoing surgery. Originally pioneered in Denmark 2 it is now being introduced in England by a growing number of surgeons, anaesthetists, nurses, allied health professionals and NHS managers. The ERAS protocols have a series of evidence-based care elements that all support recovery by reducing the bodily stress reactions caused by injury. These reductions in the stress responses are of particular importance for the vulnerable patient with co-morbidities, who is often also frail and elderly. 3 To inform the list of important outcomes for this study, we conducted a forum with the University of Bristol's Musculoskeletal Research Unit's (MRU) patient involvement group: the 'Patient Experience Partnership in Research' (PEP-R). PEP-R comprises twelve patients with musculoskeletal conditions, most of whom have had joint replacement, all of whom have had experience of long-term pain. The group were given a list of outcomes for consideration and discussion that were available in the routine datasets: length of stay, readmission, reoperation, revision surgery, complications, mortality, Oxford hip and knee scores on pain and function. Patients considered pain and function to be the most important outcome, followed by complications of surgery. Length of stay was considered a mid-ranking outcome, with the group agreeing that it was important but very dependent upon the level of support at home. Revision surgery was also a mid-ranking outcome for the group. Mortality was ranked low by the group in respect of its importance to them, and hence has not been included in the analysis for this paper.

VERSION 1 -AUTHOR RESPONSE
In respect of the reviewers comment regarding the inclusion of reasons for ERAS implementation, as additional justification for the outcomes, enhanced recovery improves quality of care by helping patients to get better sooner after major surgery, that in turn reduces length of stay with benefits to the NHS. 4 Length of stay is a proxy for efficiency of health care provision in the absence of data on clinical quality. [5][6][7] In addition, length of stay is the outcome used for the UK Department of Health to report results of ERAS implementation. 8 Complications are a measure of efficacy and safety of ERAS implementation. Revision surgery was the main method for assessing outcomes of hip replacement surgery until patient reported outcome measures were introduced in 2008.
Regarding the reviewers comments as to whether ERAS could have strengthened the changes that were already underway, we would suggest that an important consideration of the findings of this study are that the secular trends of improved outcomes have continued following the introduction of ERAS, rather than stopping. These improvements have occurred whilst other changes have been occurring, including reductions in the numbers of available beds/wards/operating theatres, in addition to increasing absolute numbers of patients undergoing THR year on year. Further the case mix of patients has also been changing over the study time period. In our dataset we find a higher percentage of sicker patients in the intervention period (April 2009-March 2011) and even that is higher again in the post intervention period (April 2011-December 2016) than preintervention period (April 2008-March 2009). In the discussion section we have attempted to comment on these "external influencing factors" (pages 19-20, lines 454-474).
Author action: we have added to the text further explanation of ERAS. Page 5, lines 100-103: "ERAS has a series of evidence-based care elements that all support recovery by reducing the bodily stress reactions caused by injury during surgery. These reductions in the stress responses are of particular importance for the vulnerable patient with co-morbidities, who is often also frail and elderly." We also provide further detail about ERAS pathway. Page 6, lines 114-120: "In primary care haemoglobin levels and pre-existing co-morbidities like diabetes are assessed. At the hospital nurses test cardiopulmonary exercise and appropriate anaesthetic for covering surgery is evaluated. In addition, there is informed decision making after offering to the patient information and managing her/his expectations. At the admission, hospitals arrange to admit patients the same day of surgery, fluid hydration is optimised using oral complex carbohydrates to reduce patient anxiety, reduce the body's resistance to insulin and inflammatory response." Page 6, lines 121-124: "It includes: minimally invasive surgery if possible, individualised fluid therapy, avoid crystalloid overload, use of regional/spinal and local anaesthetic with sedation, and hypothermia prevention." Page 6, lines 125-129: "It includes: no routine use of wound drains and/or nasogastric tubes, active, planned mobilisation within 24 hours, early oral hydration and nutrition, intravenous therapy stopped early, catheters removed early, oral analgesia avoiding systemic opiates where possible. Follow-up covers: discharge on planned day or when criteria met, therapy support (stoma, physiotherapy, dietitian…) and 24 hour telephone follow-up if appropriate." 3) Can you offer an explanation for why you excluded patients with an LOS >15 days? Why not 5 days? Was this an arbitrary cut-off? Author response: There were very few patients with a LOS > 15 days (around 1.5% of patients stayed longer than 15 days). We made a decision to exclude them, as we were concerned that leaving those patients in the analysis could create statistical "noise" in the observed trends. year. Further the case mix of patients has also been changing over the study time period. In our dataset we find an increased percentage of older patients (>80 years old) undergoing hip replacement (see Response Figure R1 below). We also observe an increasing percentage of sicker patients (see ascending trends or percentage of patients with mild, moderate and severe Charlson co-morbidity index in Response Figure R2). Similar increasing trends are observed: in obese patients (Response Figure R3); in patients with a pre-operative incapacitating disease or life-threatening disease (Response Figure R4) and patients undergoing a surgical posterior approach (Response Figure R6). Percentages of IMD categories (Response Figure R5) are steady across the study period except for those living in the most deprived areas with a decreasing percentage of surgeries. Response Figure R1.  (6). It has been estimated that 97,516 total hip replacements will take place in 2035(44)." Author action: We have added the following text to the discussion (page 20/lines 471-474): "ERAS has kept the improvements happening when other changes were occurring which may have caused deterioration, e.g.: older, sicker, and more obese patients. However, changes in the case-mix of patients have not altered improving trends in outcomes of surgery" We observe an increasing trend in the monthly percentage of hip replacements in ISTCs until 2010.
Author action: We have added the following text to the limitations of the study (page 21, lines 495-501) "Although the NJR registry captures all primary hip replacements including those undertaken in the private sector, linkage to English HES data means that we only have access to information on patients receiving NHS funded operations including public and private hospitals. Therefore, this study does not include private funded operations undertaken by the independent sector. It is estimated 13.7% to 19.7% of all hip replacements were carried out by the independent sector in 2012-2013 and 2016-2017, respectively (Source: Hospital Episode Statistics, NHS Digital.)" In the discussion section on external influencing factors, we also now include the following (page 20, lines 463-471): "During the period of our study we observed an increasing trend in the proportion of NHS funded primary hip replacements being carried out in independent hospitals (increasing from around 10% in 2008 to 27% in 2016) and a small increase in those within Independent Sector Treatment Centres (ISTC) (from 3.5% in 2008 to 5% in 2016). These changes will have supported an increase in capacity for surgery (although such centres typically treat healthier and less complex patients than nearby public hospitals, with a worsening case-mix of those patients treated in public hospitals 9 . Such changes in the sorting of routine and complex patients between public and private hospital settings over time could also influence observed changes in outcomes of surgery over time."