Journal of Public Health Advance Access originally published online on April 5, 2006
Journal of Public Health 2006 28(2):116-124; doi:10.1093/pubmed/fdl003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
The effects of surgical volumes and training centre status on outcomes following total joint replacement: analysis of the Hospital Episode Statistics for England
Andy Judge, Research Associate1
Jiri Chard, Research Fellow2
Ian Learmonth, Professor of Orthopaedic Surgery3
Paul Dieppe, Director1
1 MRC Health Services Research Collaboration, University of Bristol, Canynge Hall, Whiteladies Road, Bristol BS8 2PR, UK
2 National Collaborating Centre for Womens’ and Childrens’ Health, Royal College of Obstetrics and Gynaecology, 27 Sussex Place, Regent’s Park, London NW1 4RG, UK
3 Department of Orthopaedic Surgery, University of Bristol, Dolphin House, Bristol Royal Infirmary, Bristol BS2 8HW, UK
Address correspondence to Andrew Judge, E-mail: andrew.judge{at}bristol.ac.uk
Objective Previous work from other countries has shown a significant inverse relationship between the number of some surgical procedures undertaken in a hospital and in an adverse outcomes. In the light of the changing nature of the provision of joint replacements in the United Kingdom, we have examined the effects of surgical volumes and the presence/absence of training centre status, on outcomes following total joint replacement (TJR) in England.
Methods Analysis of the Hospital Episode Statistics (HES) on all hip/knee joint replacements in English National Health Service (NHS) trusts between financial years 1997 and 2002. Exposures explored were the volume of hip/knee replacements per annum in an NHS trust, training centre status and whether the admission was routine or emergency. Four surrogate measures of adverse outcome were assessed: 30-day in-hospital mortality, length of stay in hospital, readmission within a year and surgical revision within 5 years. Age and sex were controlled for as potential confounders.
Results Data from a total of 281 360 hip replacements and 211 099 knee replacements were examined. HES data show that the numbers of TJRs performed in low volume trusts are small and decreasing. Adverse outcomes were also uncommon. Nevertheless, significant associations between adverse outcomes and low volume units, and better outcomes in training centres, were detected. For example, the odds ratio (OR) for in-hospital death within 30 days of hip replacement in trusts doing <50 hip/replacements per annum is 1.98 [95% confidence interval (95% CI) = 1.13–3.47] compared with trusts doing 251–500 operations/annum. Similarly, surgery in non-training centres is more likely to result in mortality than that in training centres (OR = 1.25, 95% CI = 1.05–1.48). The examination of surgical revision indicated adverse outcomes in higher volume units; this may be due to case-mix.
Conclusion In England, there are fewer adverse events following TJR in high volume centres and in orthopaedic training centres. Standardization of procedures may account for this finding. The data have implications for private practice in the United Kingdom and for the current move to undertake TJRs in Independent Sector Treatment Centres.
Keywords: adverse outcomes, hip replacement, hospital admission, knee replacement, surgical volume
 |
Introduction |
|---|
 TOP Introduction MethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
Total hip replacement (THR) and total knee replacement (TKR) are among the most common elective procedures performed in England. They are effective interventions for patients suffering from severe hip and knee disease, making a substantial contribution to public health; patients undergoing these procedures experience the relief of pain, increased function and improvement in quality of life.1,2 Joint replacement is considered a highly cost-effective procedure.3,4 In English National Health Service (NHS) hospitals, rates of both primary joint replacement and revision joint replacement have risen substantially over the past decade.5 Although waiting lists are falling, there are still a substantial number of people waiting for surgery, and the population need for joint replacement is projected to continue to rise.5 About 80–85% of total joint replacements (TJRs) in England are performed within the NHS, the remainder being done in the private sector.6
To reduce waiting lists to meet targets in the NHS plan,7 the government needs to increase the capacity to meet the need for surgery. Capacity in the NHS is not yet growing fast enough, so extra capacity has been sought through independent sector-run treatment centres (ISTCs).8 Patients will soon be offered a choice of up to five hospitals and be allowed to choose the time and location of their treatment.9 The choice may be a fast route to surgery in an ISTC against a longer wait for a local NHS trust. This approach may not take account of the large body of research on the improvement of outcomes by the concentration of hospital services and standardization of surgical procedures and the need for and value of training centres.
Most previous research on surgical volume outcomes has been carried out in North America, and it has focused on coronary artery bypass grafting (CABG), where improved patient outcomes have been associated with higher hospital and/or surgeon volumes. This has lead to increases in the volume of surgery and improved training.10–12 CABG studies also indicate that standardization improves outcomes. A growing body of literature from the United States argues that the same may be true for joint replacement surgery.13,14
We have sought to examine the effects of unit volume and of being a training centre (a proxy for standardization of procedures) on outcomes following TJR in England through an analysis of Hospital Episode Statistics (HES).
|
Methods |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
The HES for England was used. The HES database holds information on patients who are admitted to NHS hospitals in England, either as day cases or as ordinary admissions. Each record in the database relates to one ‘finished consultant episode’. This is the period of time an individual spends under the care of one NHS consultant. Private procedures are mostly excluded from HES as there is no requirement for private hospitals to provide routine data. It has been estimated that around 20% of hip and knee joint operations are carried out in private institutions.6
The information held in the HES database includes the age and sex of the patient, area of usual residence, reason for admission (diagnosis) to hospital and procedure undertaken. Data for this analysis were derived from the database extract held at the University of Bristol, which currently contains data for financial years from 1 April 1991 to 31 March 2003.
THR and TKR operations, occurring between financial years 1997 and 2002, were extracted from the HES database. Episodes involving TJR surgery were identified as those with any of the following OPCS4 codes: W37, W38, W39 (hip joint) or W40, W41, W42 (knee joint), recorded in any of the four procedure fields. Revision operations were identified as episodes having any of the above procedure codes plus a primary diagnosis indicating complications because of internal prostheses, ICD10 codes T84 or T85.
Three main exposures have been explored.
- Volume of hip/knee replacement per annum in an NHS trust. A priori, we chose to split annual trust volume into clinically sensible categories for each intervention: 1–50, 51–100, 101–250, 251–500 and >500.
- Orthopaedic training centres. A list of the current orthopaedic training centres and their NHS trust bases was obtained from the British Orthopaedic Association (BOA) Website. This was coded as a dichotomous variable of whether the hospital trust was a training centre.
- Admission type. We identified whether patients were admitted to hospital as emergency or elective admissions, coding this as a dichotomous variable.
Statistical analysis
STATA 8.2 was used for all statistical analyses. All episodes where the speciality coding was not trauma and orthopaedics were excluded (1689 hip and 909 knee replacements). The degree of co-morbidity was classified for each patient by using the Charlson Co-morbidity Index.15 This is a weighted scale based on a variety of diseases that are thought to alter the risk of mortality. Diagnostic data from across all seven diagnostic fields were used to create a weighted score and an ordinal variable (0, ‘none’; 1, ‘mild’; 3, ‘moderate’; 3+, ‘severe’). Age and sex were controlled for as potential confounders. We have examined four main surrogate measures of patient outcomes: 30-day in-hospital mortality, length of stay, readmission within a year and surgical revision within 5 years.
Logistic regression analysis has been used to investigate odds of in-hospital death within 30 days of the operation, for all patients admitted between financial years 1997 and 2002. Univariate logistic regression looked at the association of the outcome with each exposure, and a multivariate regression analysis was carried out controlling for confounders. Likelihood ratio tests were used to assess whether the effect of volume can be described by a linear trend, by comparing a model with volume as a categorical variable to that with volume as a score. Likelihood ratio tests examined interactions. The mean and median length of stay were calculated for each exposure of interest in the 2002 financial year, and linear regression modelling was performed. As the distribution of length of stay was skewed, a log transformation was used to satisfy the assumption of normality.
Poisson regression modelling was used to calculate the rate of readmission within a year, for patients admitted in the 2001 financial year. The rate being defined as the number of admissions in 12 months, over the 12-month time period at risk. A likelihood ratio test was used to test whether the log rate of readmission changes linearly with volume. Likelihood ratio tests were used to examine interactions. A Cox proportional hazards model was fitted to investigate time till revision within 5 years after a primary joint operation for patients admitted in the 1997 financial year. Likelihood ratio tests assessed whether the log hazard of revision changes linearly with volume. The assumption of proportionality was checked through a visual check of a log–log plot and formal testing using Schoenfeld’s residuals. Where there was evidence of non-proportionality, separate time-varying hazard ratios were estimated.
|
Results |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
Although large numbers of joint replacements are recorded in HES, there are relatively few adverse events (Table 1), reducing the power of the study to detect effects of either training centres or hospital trust volume. A large number of THRs are undertaken as an emergency operation, largely because of hip fractures, but relatively few emergency TKRs are performed.
|
The number of operations being done in low volume trusts is small and decreasing over time (Table 2). There has clearly been a strong trend to more of these operations being carried out in high volume units. The number of joint operations has risen between 1997 and 2002, with TKR increasing more than THR.
|
Mortality
Relatively few patients died soon after their joint operation (Table 1). Analysis combining data for the years 1997–2002 shows good evidence that the odds of death decrease with increasing trust volume (Table 3) for both interventions. Repeating analyses for each financial year separately, the volume effect was most striking in 2002. For THR, there was weak evidence (P = 0.07) of extra variation in the odds of death beyond that of a linear trend but not for TKR (P = 0.83) where in the adjusted model the log odds decrease by –0.22, 95% confidence interval (95% CI) = –0.32 to –0.11, for a unit increase in volume group [odds ratio (OR) = 0.80, 95% CI = 0.72–0.89]. Training centres had a positive effect on mortality for THR, but there was no evidence of an effect for TKR after adjustment. Examining each financial year individually, weak evidence of a beneficial effect of training centres for TKR only became apparent in 2002.
|
Length of stay
For both interventions, length of stay is longer in lower volume trusts, with no observed differences for training centres (Table 4). Regression analysis provided good evidence of the relationship for volume. However, there was evidence that in training centres, length of stay was shorter for primary operations, whereas the opposite was true for revision. This may be because of revision operations in training centres being at the high end of complexity. However, these observed differences are small and of doubtful clinical importance.
|
Readmission to hospital within a year
For THR, crude analysis showed the rate was highest in low volume trusts. Adjusting for other variables, in particular type of admission, the effect in low volume trusts was reduced and no longer significant, but there was now some evidence of a higher rate in high volume trusts (Table 5). There was good evidence of non-linearity (P = 0.01), suggesting that there was more than just a linear trend. No relationship with volume was observed for TKR, and tests for trend showed good evidence of non-linearity (P < 0.001). There was good evidence that the rate of readmission was reduced in training centres for both THR and TKR.
|
Rate of revision surgery within 5 years
Evidence of a higher hazard of revision in high volume hospitals was observed for THR. Tests for trend found no evidence of non-linearity (P = 0.24). In the adjusted model, the log hazard of revision increases by 0.05 (95% CI = –0.01 to 0.12), for a unit increase in volume group (Hazard ratio = 1.05, 95% CI = 0.99–1.13). However, fitting volume as a quantitative variable provides no evidence that volume is associated with the hazard of revision surgery, and the trend could be in either direction. For TKR, there is some evidence for a higher hazard of revision in high volume hospitals up to 6 months after surgery (Table 6). No evidence of non-linearity was found (P = 0.33). In the adjusted model, the log hazard of revision increases by 0.06 (95% CI = –0.02 to 0.14), for a unit increase in volume group (OR = 1.06, 95% CI = 0.98–1.15), providing weak evidence that volume is associated with the hazard of revision surgery.
|
For THR, having the procedure in a training centre had no effect on revision but evidence of a beneficial effect was apparent for TKR up to 6 months after surgery.
|
Discussion |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
Main findings of this study
Analysis of the HES data showed that the numbers of TJRs being performed in low volume units in England are low and decreasing. However, there is a significant relationship between low volume units and more adverse outcomes. In addition, we have shown that training centre status is associated with better outcomes. The consistency of the trends detected for both THR and TKR reinforces the strength of these findings. The examination of revision within 5 years of surgery indicated a maximum threshold for surgery, but anecdotal evidence suggests that this is likely to be an issue of case-mix, as high volume hospitals are likely to take on more complex surgery. This is the focus of further research.
What is already known on this topic
The positive association that we have observed between higher volume units and improved outcomes is consistent with the orthopaedic literature from the United States13,14 and the CABG data.10–12 The literature highlights many factors that could explain these volume–outcome relationships.17 Surgeons need to maintain skills to achieve good outcomes, and as they gain experience are better able to select those most suitable for surgery. Enhanced education and training in the condition they are treating may lead to improved outcomes. Within hospitals, patient outcomes are related to the availability of resources, a high level of organizational standards, standardization of procedures and good teamwork and communications between professionals.18 We conclude that it is most likely that good teamwork and standardization of services in large volume orthopaedic centres are the likely explanations for our findings, but we have no data to support this. Other explanations are possible, including case-mix differences. However, we would expect the higher volume units to take on the more difficult cases, in which case the size of the real difference in outcomes between high and low volume units may be higher than that observed here. The same factors are also likely to explain the improved outcomes found in orthopaedic training centres.
If the use of a consistent approach to the procedure, including pre-, peri- and post-operative care, as well as prosthesis type and surgical methods does lead to improved patient outcomes, then similar approaches should be taken up in all hospitals performing joint replacement operations. The case for standardization of services has already been made within CABG studies, where researchers highlight the wide variation in outcomes within volume categories as evidence that the volume-effect is weak and outcome is more related to standards.19,20
What this study adds
This study is, we believe, one of the first of its type to be undertaken in the United Kingdom, using the HES. Its main strength is that HES use allowed a very large number of episodes to be analysed and that the effects of training centre status have been examined for the first time.
Within HES, we have already seen a considerable move towards operations being carried out in higher volume units for joint replacement. In 1997, 6.9% of hip operations and 21.0% of knee operations were done in trusts performing 100 or less operations/annum, whilst in 2002 the corresponding figure was 1.2% of hips and 1.4% of knees (Table 2). Together with the shift away from low volume hospitals, the evidence of better outcomes for training centres has been getting stronger over time and only truly become apparent for knee replacement in 2002. In the light of this, we believe that the findings on the relationship between volumes and outcomes will be of limited importance to the NHS in England. However, the training centre effect that we have recorded may be of more significance to service provision in the NHS. We interpret the better outcomes observed in training centres as being likely because of greater standardization and consistency of procedures and perhaps more routine use of evidence-based approaches and audit in relation to issues such as prevention of sepsis and deep vein thrombosis. This hypothesis emerges from the literature, but we have no data to refute or support it.
Whilst low volume units and lack of standardization may be diminishing problems within standard NHS provision of TJRs, our findings raise concerns about what might be going on within private practice and ISTCs, whose data are not in HES. We have not been able to find information on volumes within private hospitals but fear that in some cases they may be relatively low, and made up from large numbers of different surgeons doing a few cases each, and taking varying approaches to the procedures. Similar concerns were recently raised by the President of the Royal College of Surgeons about the standard of operations performed by surgeons from abroad working in ISTCs.21 We believe that both private hospitals and ISTCs should be required to make their data available to HES and to the recently formed hip replacement registry, to allow their activities to be examined and compared with NHS hospitals.
Another consideration relevant to changes in the provision of joint replacement surgery is the willingness of a patient to undergo the procedure in unfamiliar settings. Research by Losina et al.22 suggested that policies in the United States aimed at restricting hip replacement to high volume centres would differentially affect older, poor, less well-educated and rural patients, widening socio-economic disparities in the utilization of hip replacement. This is consistent with research in pancreatic cancer23 and breast cancer treatment.24 Using HES data, we have already shown that the most deprived in society receive fewer TJR operations.5 If the preference of the more deprived patients is to have their operation in a local hospital, it may be that the better off in society take advantage of ‘Patient Choice’,9 receiving faster access to treatment through greater use of major orthopaedic centres some distance from their homes, ISTCs or private hospitals. This could further increase inequity if it means the most deprived in society are faced with longer waiting times or would then be unwilling to undergo surgery. The government may succeed in its goal to reduce waiting lists but fail in its commitment to reducing health inequalities.7
Limitations of this study
There are, however, many limitations to the study. First, the choice of what outcomes to analyse was dependent on the information contained in the HES database. In joint replacement procedures, uncommon adverse events are arguably not the most important outcome measures, relief of pain and improvement in function are what matters most to the majority of patients and we would have preferred to use these measures for our outcomes. Second, it is likely that case-mix is an important issue, but this cannot be easily addressed within HES. We have attempted to adjust for this through the Charlson Co-morbidity Index, and the type of admission (emergency or routine: of little relevance to TKR). However, the issues of complexity of surgery and patient selection remain as a limitation to this study, particularly as outcomes of this sort of surgery, other than death, probably relate more to the pre-operative state of the patients bones and joints, rather than to co-morbidity. In prospective studies, that have included more detailed case-mix analysis, the size of the volume–outcome relationships are reduced compared with those that do not properly account for case-mix.16 Third, the HES database does not include private hospitals, many of which may be relatively low volume providers. No attempt has been made to analyse data according to individual surgeons, in part, because such coding may not reflect who actually does the surgery (trainee or consultant), even though the individual carrying out the operation may be important to outcomes.
|
Contributors |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
PD and IL had the original idea for the study. JC undertook the literature searching for the paper. AJ wrote the first draft and incorporated comments from all authors. AJ performed all statistical analyses. All authors contributed to the interpretation and writing of the paper. PD is the guarantor.
|
Conflict of interest statement |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
We declare that we have no conflict of interest.
|
Acknowledgements |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
This study was commissioned by and funded from the Department of Health.
The HES data were made available by the Department of Health. HES analyses conducted within the Department of Social Medicine are supported by the South West Public Health Observatory. Andrew Judge is supported by the Medical Research Council/Health Services Research Collaboration. The Department of Social Medicine is the lead Centre of the MRC Health Services Research Collaboration. The opinions expressed by the authors are theirs alone and do not represent the opinions of supporting organizations.
|
References |
|---|
|
TOPIntroductionMethodsResultsDiscussionContributorsConflict of interest statementAcknowledgementsReferences |
|---|
- McMurray A, Grant S, Griffiths S, Letford A. Health-related quality of life and health service use following total hip replacement surgery. J Adv Nurs 2002;40:663–72.[CrossRef][Medline]
- Hawker G, Wright J, Coyte P et al. Health-related quality of life after knee replacement. J Bone Joint Surg Am 1998;80(2):163–73.
[Abstract/Free Full Text] - NHS Centre for Reviews and Dissemination, University of York. Total hip replacement. Eff Health Care 1996;2:1–12.
[Free Full Text] - Chang RW, Pellisier JM, Hazen GB. A cost-effectiveness analysis of total hip arthroplasty for osteoarthritis of the hip. JAMA 1996;275:858–65.[Abstract]
- Dixon T, Shaw M, Ebrahim S, Dieppe P. Trends in hip and knee joint replacement: socioeconomic inequalities and projections of need. Ann Rheum Dis 2004;63:825–30.
[Abstract/Free Full Text] - Williams B, Whatmough P, McGill J, Rushton L. Private funding of elective hospital treatment in England and Wales, 1997–8: national survey. BMJ 2000;320:904–5.
[Free Full Text] - Department of Health. The NHS Plan. London: HMSO, 2000.
- Department of Health, Access Directorate, Secondary Care. Treatment Centres: Delivering Faster, Quality Care and Choice for NHS Patients. 2005.
- Department of Health. Choose & Book – Patient’s Choice of Hospital and Booked Appointment. 2004.
- Kalant N, Shrier I. Volume and outcome of coronary artery bypass graft surgery: are more and less the same? Can J Cardiol 2004;20(1):81–6.[ISI][Medline]
- Goodney PP, Lucas FL, Birkmeyer JD. Should volume standards for cardiovascular surgery focus only on high-risk patients? Circulation 2003;107(3):384–87.
[Abstract/Free Full Text] - Nallamothu BK, Saint S, Ramsey SD, Hofer TP, Vijan S, Eagle KA. The role of hospital volume in coronary artery bypass grafting: is more always better? J Am Coll Cardiol 2001;38(7):1923–30.
[Abstract/Free Full Text] - Katz JN, Barrett J, Mahomed NN, Baron JA, Wright RJ, Losina E. Association between hospital and surgeon procedure volume and the outcomes of total knee replacement. J Bone Joint Surg Am 2004;86-A(9):1909–16.
[Abstract/Free Full Text] - Hervey SL, Purves HR, Guller U, Toth AP, Vail TP, Pietrobon R. Provider volume of total knee arthroplasties and patient outcomes in the HCUP-nationwide inpatient sample. J Bone Joint Surg Am 2003;85-A(9):1775–83.
[Abstract/Free Full Text] - Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83.[CrossRef][ISI][Medline]
- NHS Centre for Reviews and Dissemination, University of York. Hospital volume and health care outcomes, costs and patient access. Eff Health Care 1996;2:1–16.
[Free Full Text] - Luft HS, Hunt SS, Maerki SC. The volume-outcome relationship: practice-makes-perfect or selective-referral patterns? Health Serv Res 1987;22:157–82.[ISI][Medline]
- Department of Health. Learning from Bristol: The Report of the Public Inquiry into Children’s Heart Surgery at the Bristol Royal Infirmary 1984–95. London: Department of Health, 2001.
- Halm EA, Lee C, Chassin MR. Is volume related to outcome in health care? A systematic review and methodologic critique of the literature. Ann Intern Med 2002;137:511–20.
[Abstract/Free Full Text] - Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349:2117–27.
[Abstract/Free Full Text] - Meikle J. Attack on Visiting Surgeons. http://society.guardian.co.uk/nhsperformance/story/0,8150,1415642,00.html (16 February 2005, date last accessed).
- Losina E, Barrett J, Baron JA, Levy M, Phillips CB, Katz JN. Utilization of low-volume hospitals for total hip replacement. Arthritis Rheum 2004;51:836–42.[CrossRef][ISI][Medline]
- Finlayson SR, Birkmeyer JD, Tosteson AN, Nease RF Jr. Patient preferences for location of care: implications for regionalization. Med Care 1999;37:204–9.[CrossRef][ISI][Medline]
- Palda VA, Llewellyn-Thomas HA, Mackenzie RG, Pritchard KI, Naylor CD. Breast cancer patients’ attitudes about rationing postlumpectomy radiation therapy: applicability of trade-off methods to policy-making. J Clin Oncol 1997;15:3192–200.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||