Since the comparison of Billroth's and Halsted's recurrence rates after mastectomy, we have known that a relation exists between the volume of procedures and the outcome of treatment. This relation still holds major promise for improved safety of patients. The NHS needs a more systematic approach to identify volume thresholds and to ensure that they are met. This will provide a firmer foundation for evidence based assessment of service mergers, capital developments, and for informed choice by patients.
In 1996, the NHS Centre for Reviews and Dissemination published a systematic review to determine for which procedures such a relation existed.1 The list included coronary artery bypass surgery, paediatric heart surgery, acute myocardial infarction, coronary angioplasty, aortic aneurysm, amputation of the lower limb, gastric surgery, cholecystectomy, intestinal operations, knee replacement, and neonatal intensive care. This review was not a meta-analysis—the studies included different groups of patients, outcome measures, and methods of categorising volumes. Moreover for some procedures the better the adjustment for case mix the weaker was the relation between volume and outcome. Consequently debate about the relationship has continued.
The NHS performance indicators include league tables of death rates and other outcomes, but they do not consider any relation of volume to outcome. Although the relation has found expression in health policy in the United Kingdom—for example, in the NHS guidance on breast cancer, the guidance on colorectal cancer states that evidence of a volume effect is not found in most studies.2 This contrasts with a review by the US National Academy of Sciences, which found that colonic resection for cancer was associated with lower death rates in hospitals that did more procedures.3
Recent research has shed more light on this issue. In the United States, four new reports have been published: a systematic review in 2000,4 which received wide coverage in the media, and three other studies in 2002.5–7 The study by Hannan et al took the analysis of volume and outcome a step further—to the individual surgeon performing cancer surgery in a hospital.5 Using large population based samples, the investigators showed a statistically significant association between “high volume” surgeons in “high volume” hospitals and lower death rates in hospital, and vice versa.
Another study using national data examined the mortality associated with six cardiovascular procedures and eight major cancer resections.6 Mortality decreased as volume increased for all 14 procedures, but the relative importance of volume varied markedly according to the procedure. Absolute differences in adjusted mortality ranged from over 12% to only 0.2%. The authors concluded that in the absence of better information patients can reduce their risks by selecting a high volume hospital.
The study of radical prostatectomy showed significant trends in the relation between volume and outcome with respect to postoperative and late complications.7 Postoperative morbidity was 27% in very high volume hospitals and 32% in low volume hospitals, and was lower when prostatectomy was done by very high volume surgeons. The rates of late complications followed a similar pattern.
A national workshop held by the US Institute of Medicine in 2000 to discuss the systematic review agreed that action should occur if criteria of plausibility, effect size, consistency, and reproducibility were met. The United States' aggressive healthcare purchasers are already responding. The Leapfrog Group (www.leapfrog.org), a coalition of employers and healthcare plans covering 31 million US citizens, will soon require hospitals to meet volume standards for five procedures. Leapfrog has estimated that nationwide implementation of these standards would save 2581 lives annually. Leapfrog is also moving to reward hospitals if they can show evidence of implementing other safety standards.
In the United Kingdom, a report on paediatric cardiac surgery showed a strong and consistent inverse association between mortality and volume of cases (excluding data from Bristol).8 Yet, despite easy access to data and an integrated national healthcare system, little research has been done using a consistent methodology across a range of procedures. No policy is in place to ensure that positive findings are implemented nationally. The National Patient Safety Agency could lead such work.
Some questions need to be answered if such information is to be used for selective referral or service mergers. Should rural areas be excluded? Distance decay in intervention rates is still a feature of the NHS. As the York review queried, is there evidence that previous service mergers have improved outcomes?1 Since the relation is often linear, how should volume thresholds be set? How should the effects of surgeon and hospital volume be combined? Are research findings from North America applicable to Europe? Can intensive training and observation of surgeons through clinical networks compensate for low volumes?9
Case mix adjustment is another problem. Administrative data sources such as hospital episode statistics do not support the stratification of outcomes based on risk, although both the Department of Health and the Dr Foster website are attempting to overcome this in their league tables. More clinical information systems such as the national cardiac surgical database will be needed to do this well. The 1999-2000 annual report from the database (www.ctsnet.org/doc/853) provides an exemplary explanation of the statistical issues involved in risk stratification and control charts—issues that will concern more clinicians as more clinical performance information is published.
In the United States, the publication of performance data has been associated with an improvement in health outcomes.10 It was hospitals themselves, rather than users or commissioners, that seemed to be most responsive to the data. Will this occur in the United Kingdom? Ongoing evaluation of the impact of the indicators of NHS performance would be helpful.
Organisational resistance to service mergers aimed at meeting volume thresholds and sheer logistics may be barriers.11 So other interventions could be needed to ensure implementation. Commissioning by primary care trusts should provide one stimulus; another could be the new Commission for Healthcare Audit and Inspection. The patient choice initiative, through which providers will earn extra resources for additional patients that move to them, could also support implementation—but paradoxically reduce choice.12 Finally, if patients are to follow the government's exhortations to exercise choice they will need more information than they currently have and the education to use it.