Relationship between volume and quality of health care: a review of the literature

OBJECTIVES:

To carry out a systematic review of the literature in order to examine the evidence for a relationship between the volume of activity of a hospital and patient outcomes in a range of procedures.

DATA SOURCES:

Search of MEDLINE (from 1985 to 1994) and of science Citation Index on the Bath Information and Data Service (BIDS). Both letters and editorials were excluded. The reference list of identified articles were also searched. Key relevant journals were also hand-searched: Medical Care from 1971 to 1994 and Health Policy from 1986 to 1994. Researchers in the field in the United Kingdom and the United States were consulted to identify published and unpublished evidence.

STUDY SELECTION:

Studies were included if they empirically assessed the relationship between the frequency with which procedures were carried out, or patients with a particular diagnosis were treated, and health outcomes such as mortality or morbidity. Individual studies were judged to be relevant by one reviewer. Data extraction was checked by another reviewer.

DATA EXTRACTION:

For studies of coronary artery bypass graft surgery (CABG) which examined the effect of volume on outcome, details of the source of data and the date of data collection, number of hospitals and patients, volume cut-off points used, statistical adjustements made in the analyses, together with the relationship found, were recorded.

DATA SYNTHESIS:

A narrative overview of most of the evidence was carried out. In the case of CABG surgery a meta-analysis by regression was used. the studies were pooled in order to assess the degree to which the estimates of the effect of volume on outcome were affected by the degree of adjustment for case-mix. Factors adjusted for included age, sex, severity of diagnoses and comorbidity. In addition the analysis investigated the relationship between the estimates of volume effect and year of data.

RESULTS:

There is a large literature, primarily from North America which has examined the relationship between volume and outcome for a wide range of (mainly surgical) procedures. Almost all of the research in this area has used an observational study design with hospital mortality rates as the measure of quality. The methods used in each study differed considerably and ranged from simple comparisons of outcomes by hospital volume to complex simultaneous-equation models. A wide range of levels of volume of activity have been used to categorise hospitals as high or low volume.
A positive relationship between volume and outcome (where higher volumes were associated with lower mortality rates) has been reported for most procedures: abdominal aortic aneurysm, vascular surgery, biliary tract surgery, cardiac catheterization and angiography, CABG surgery, total hip replacement, prostatectomy, intestinal operations, hysterectomy, acute myocardial infarction, perinatal ilness, hernia and trauma care. In contrast, studies of treatment for fracture of the femur, appendicectomy, stomach operations and burn care have been reported to have either no relationship or a negative relationship (where higher volumes were associated with higher mortality rates) between volume and outcome.
Few of the studies adjusted for the effects of case-mix differences between high and low volume hospitals. The 15 identified studies which examined CABG surgery differed with respect to the extent of adjustement for confounding. The analysis carried out suggests that the greater the adjustment for the effects fo case-mix, the smaller the size of the estimate of benefit associated with increased volumes of activity.
The available studies did not shed much light on the possible causes of any volume-outcome relationship and the level (e.g. hospital, ward, clinician) at which it might operate. Thus it was difficult to disentangle any direction of cause i.e. whether increased volume may have generated better outcomes of if better units attracted more patients.

CONCLUSIONS:

Whilst most studies report a positive relationship between hospital volume and outcome for several elective procedures, they may have biased estimates of the size of the effect of volume because of poor adjustment for the effect of differences in case-mix between high and low volume hospitals. There is also a paucity of information as to why high volume might be associated with better outcomes. For example, there is little evidence whether clinician-related elements such as better judgement about high and low risk patients may be related to volume or whether high volume clinicians keep up with the literature and develop an effective protocol of practice. If the latter were true then the adoption of clinical guidelines by other clinicians may improve quality of care and close the gap between smaller and larger units. Alternatively, there may be institution related elements such as the level of facilities which produce better outcomes. There is little evidence as to whether merging hospitals to create larger units will result in a change in outcomes over time. Because of the uncertainty in both the size and the interpretation of any effect of volume reported, caution should be exercised in using the research literature to justify policies of reorganisation of health care delivery. Therefore, the main recommendation is that policy-makers should be cautious when invoking the assumed improvements in outcome achieved by volume as a key argument for centralisation of services.