LETTERS TO THE EDITOR Response to an Article in the October 2005 Issue of Medical Care To the Editor: W e are responding to the interesting report from Rasmussen et al. 1 where the authors reported a lower mortality af- ter first acute myocardial infarction (AMI) in tertiary cardiac care centers compared with local hospitals by analyzing dis- charge records of a wide calendar period (1995–2002). Notably, new diagnostic cri- teria for AMI were introduced in 2000, 2 but the authors were confident that possi- ble distortions in the estimated performance of hospitals could be only marginal. We think that the potential bias in- duced by new diagnostic criteria could be more than moderate for 2 reasons: cases identified by new diagnostic codes present a markedly lower mortality and the adop- tion of new diagnostic codes probably is different in different types of hospitals. If we assume that, in Denmark between 2000 and 2002, there were 5000 additional cases of AMI identified by new diagnostic criteria 3 and that the short-term mortality of these cases is 7%, without differences between hospi- tal types, 4 we can compute different odds ratios for the short-term mortality of local versus tertiary hospitals assum- ing different proportions in the detection of new cases (Table 1). We think that tertiary hospitals are moderately more disposed to the adop- tion of new criteria than local hospitals, with an important potential impact, mainly on the statistical significance of the association. For single hospitals, the impact may be even greater, with impor- tant changes in the ranking of hospitals. A way to take into account for this factor is the introduction in the statistical model of another variable: type of AMI. We think that it could feasible using the fourth digit of ICD-10 codes, given that in our experience ICD-9-CM codes allow a clear identification of subtypes of AMI. Stefano Brocco, MD Ugo Fedeli, MD Paolo Spolaore, MD Epidemiological Department–Veneto Region–Italy Castelfranco Veneto (TV), Italy REFERENCES 1. Rasmussen S, Zwisler ADO, Abildstrom SZ, et al. Hospital variation in mortality after first acute myocardial infarction in Denmark from 1995 to 2002. Med Care. 2005;43: 970 –978. 2. The Joint European Society of Cardiology/ American College of Cardiology Committee. Myocardial infarction redefined–A consensus document of The Joint European Society of Cardiology/American College of Cardiology Commettee for the Redefinition of Myocardial infarction. Eur Heart J. 2000;21:1502–1513. 3. Abildstrom SZ, Rasmussen S, Madsen M. Changes in hospitalization rate and mortality after acute myocardial infarction in Denmark after diagnos- tic criteria and methods changed. Eur Heart J. 2005;26:990 –995. 4. Brocco S, Fedeli U, Schievano E, et al. Effect of the new diagnostic criteria for ST-elevation and non-ST-elevation acute myocardial infarc- tion on 4-year hospitalizaton: an analysis of hospital discharge records in the Veneto Re- gion. J Cardiovasc Med. 2006;7:45–50. Response to Letter From Dr. Stefano Brocco et al To the Editor: W e appreciate the interest and are thankful for this opportunity to respond to the concern, from Brocco et al, that the introduction of the new di- agnostic criteria in 2000 1 for acute myo- cardial infarction (AMI) could misrep- resent the comparison between type of hospitals and their relation to mortality after AMI in our study. 2 This under the assumption that tertiary hospitals have a greater propensity to use these criteria and that AMI diagnosed with these cri- teria had a lower mortality. First of all, we think that it still remains unclear whether the new criteria actually identify a group of AMI pa- tients with a lower mortality in popula- tion-based studies. On the contrary, Pell et al 3 and Salomaa et al 4 report a higher mortality for this group of patients, even after adjusting for other important risk factors. Also, Abildstrom et al 5 report that mortality after AMI seems to be insensitive to introduction of the new diagnostic criteria. As suggested in the letter from Brocco et al regarding the use of the international classification of disease (ICD) to identify different types of AMI, there is not yet a clear identification of ST-segment elevation AMI (STEMI) and non-STEMI in the Danish version of ICD-10. The Danish National Board of Health and the Danish Society of Cardiology have developed a new clas- sification scheme with the opportunity to classify AMI as STEMI and non- STEMI, but it is not yet implemented in the Danish administrative databases. However, when appropriately imple- mented and used it will certainly prepare the ground for interesting analyses. In the Rasmussen et al, 2 we ad- dressed the introduction of the use of biochemical markers (elevated troponin T and I concentrations) in the time pe- riod between 1995 and 2001 by a ques- tionnaire distributed to hospitals in Den- mark with acute care for AMI. The proportion of local hospitals (n  36), tertiary hospitals (n  5), and regional hospitals (n  32) that had introduced the diagnostic criteria for AMI before 2002 were 91%, 80%, and 74%, respec- tively, such that there was no association between type of hospital and the initia- tion time of the new and sensitive bio- chemical markers (P  0.31). There- fore, the hypothesis in the letter of Brocco et al that tertiary hospitals are more likely to adopt the new criteria for AMI than local hospitals cannot be con- firmed in a Danish setting. Furthermore, a test for interaction, using the statistical model defined in Rasmussen et al 2 for TABLE 1. Potential Impact of Different Proportions in the Detection of New Cases of AMI on the Mortality Odds Ratio (OR) of Local Hospitals Versus Tertiary Hospitals Propensity in the Adoption of New Diagnostic Criteria for AMI in Different Types of Hospitals % of New AMIs Mortality OR Local vs. Tertiary Hospitals Local Hospitals Tertiary Hospitals Same propensity 23.3 16.7 1.00 Moderate difference in propensity 10 40 1.08 Extreme difference in propensity 0 80 1.20 Medical Care • Volume 44, Number 7, July 2006 701