Coronary care units have now been in use for 40 years, and it is generally acknowledged that they have helped to improve prognosis and reduce hospital stay among patients with acute myocardial infarction. This was confirmed in our AMI cohort where we observed a strong protective effect of CCU admission on 30 days mortality and a significantly shorter hospital stay for patient admitted to CCU.
We observed that in most Rome hospitals the proportion of AMI patients directly admitted to CCU is lower than it should be according to international recommendations,  and lower than that observed in other developed countries [22,23]. Moreover, we found wide differences in rates of CCU admission among hospitals. It is beyond the scope of this paper to investigate which structural and organizational characteristics at the hospital level are associated to high proportion of non-admission to CCU, however admission rates do not increase in hospitals where the number of AMI patients is low in comparison to available CCU beds. While available data, and the results of a previous study  suggest a less than optimal use of CCU resources. In fact, we found that, among the 11243 patients who passed through the 112 CCU beds available in the 11 Rome hospitals in the year 2000, for an overall length of stay of 62622 days, only 40% had a diagnosis of AMI, while 46% had principal diagnosis of other acute cardiac disease and 14% had other diagnoses. In summary, variable, and incongruous admission and discharge policies as well as actual shortage of beds could have affected the CCU admission rate of AMI patients, whatever the reasons CCU is apparently a scarce resource in Lazio hospital which should be used unbiasedly.
On the contrary, our results showed that age, severity of illness, and SES are important determinants of the probability that a patient with AMI who reaches qualified Rome hospital is directly admitted to CCU. Previous studies have documented restricted access to CCU and invasive procedures, and under use of well-established therapies such as aspirin, reperfusion and beta-blockers among elderly , female , and poorer AMI patients . A recent systematic review suggests that patients who are perceived not to benefit from critical care are more often refused intensive care unit admission .
The age-related admission policy to CCU we observed has been documented previously [27,28], as well as the lower probability of being accepted in CCU for patients with higher severity . The factors influencing admission decisions are likely to exclude large numbers of patients who could benefit from advanced diagnostic and therapeutic options .
We used discharge abstract data, coded according to the International Classification of Diseases IX revision, so it was impossible for us to distinguish between ST-segment elevation and non ST-segment elevation MI. Even though ST-elevation may (and should) influence the physician referral decision, we think that, if the percentages of non-ST segment MI in the groups under study are the same, our results should not strongly be biased.
We used a small area-based SES index, because direct individual data on social class were not available. This index has been shown to be a strong predictor of differences in mortality,  and associated to inequalities in access to important health interventions [32,33] and medical management  in Rome. Small-area data have been widely used to impute individual socio-economic status, and despite some criticism  inferences based on this method appear to be valid [36,37].
Age and admitting hospitals were the variables responsible for a negative confounding effect on the association of socio-economic status with direct CCU admission. Patients with low SES levels are younger than patients with high SES levels and tend to be admitted to hospital with higher provision of CCU care.