Variations in mortality rates among Canadian NICUs —and anonymous reporting

Perhaps the greatest contribution of the Canadian Neonatal Network's recent report1 is the debate it has initiated regarding institution identification in outcomes research. The accompanying commentary2 by CMAJ editors states that the decision by the authors of the report not to name the participating hospitals was based on 2 factors — a prior guarantee not to link results to specific institutions, and the opinion that identification of the institutions may be misleading.

The first consideration is not the obstacle it may initially appear. The situation is not analogous to obtaining consent for patient information under a guarantee to protect privacy. An initial agreement not to disclose institutional identity could be revised by the participants should they agree that the public interest is promoted by lifting the veil of anonymity.

Although the merits of publishing league tables of mortality rates in neonatal instensive care units (NICUs) are debatable,3 insisting on anonymity may have led to loss of additional valuable information. One landmark study has demonstrated a relationship between patient volume and outcomes for level 3 NICUs4 above the threshold of an average daily census of 15 patients. Other studies have found similar associations in pediatric ICUs,5 or have failed to replicate these results for infants with very low birth weights.6 Investigation of volume–outcome relationships in Canadian NICUs would be a valuable contribution. In the study published in CMAJ, partial information discerned from the figures points to an apparent moderate relationship between overall volumes and adjusted mortality rates.1 Obviously, the best and proper way to determine the existence and magnitude of any volume–outcome relationship is to analyze the data for specific subgroups of neonates and to directly test for the significance of a case-volume factor. This can be easily accomplished with the data available.

Reporting on a possible volume– outcome relationship may not have been pursued in the Neonatal Network's report in view of the possible unmasking of individual units' identities. If that is the case, the study participants have, in the interest of maintaining anonymity, forgone the opportunity to provide valuable information to inform policy and planning. This silence does not serve the public interest and is not tenable given the double-doses of public funding that support this research — first for patient care, and second to support the research network. Although it may not be advisable to report the units' results in league-table format, shying away from any disclosure may have led to the loss of valuable information. In the future, not only should journals be unwilling to publish anonymous results,2 but publicly accountable funders should also be unwilling to support research that favours researchers' interests over those of the public.

Antoni Basinski VP Research, Development and Quality Improvement THiiNC Information Management Inc. Toronto, Ont.