Hospital discharge abstract databases are increasingly being used in health services research. AMI diagnostic codes in Canadian administrative databases have been validated using the hospital discharge data and chart audits [15
]. For example, Austin et al.
found a sensitivity of 89% and a specificity of 93% for patients admitted to coronary care units [17
]. Administrative data have been used in surveillance of cardiovascular surveillance [1
] and other conditions [22
] with success.
The decreasing hospitalization rate for AMI in Canada has been described by others previously. In particular, in 2009, Tu and his colleagues reported a substantial decrease in hospitalization rates (standardized to the 1991 population) for AMI and other cardiovascular disorders from 1994 to 2004 using CIHI data for the whole of Canada [1
]. Our study resulted in a rate for 2004 are quite similar to that of Tu et al.
, despite not having data for Quebec.
Similarly, the increasing use of PCI and the declining use of CABG in AMI patients have also been previously reported. Ko et al.
showed that, between 2001-03 and 2004-06, the rate of PCI use per 100,000 population in Ontario rose by 20%, while the rate of CABG use fell by 10% [23
]. In the CIHI data, the rate of PCI in Canada (excluding Quebec) increased between 2002 and 2009 by 125% and the rate of CABG declined by over 60%. Thus, there has been a significant change in the type of care (PCI v. CABG) over recent years in Canada with PCI, which has been described as having a direct impact on patient survival. PCI remains a central therapy for patients with symptomatic coronary artery disease, particularly those with AMI, and has generated tremendous attention in the last decade with issues such as the risks and benefits of drug-eluting stents and anti-thrombotic therapies [24
]. There are few representative data describing the contemporary patterns of care and outcome trends in patient with AMI and/or undergoing PCI. This is of particular importance because the process of updating clinical practice guidelines and quality metrics for AMI and PCI has accelerated.
The present analysis extends the previous work of Tu et al.
] to a more recent year (2009) and attempts to project not only the rates, which our analysis suggests will continue to decline in those aged ≥ 65 years and remain relatively stable in those aged < 65 years, but also the numbers of AMI hospitalizations to 2020. Despite projected decreasing or stable rates of AMI hospitalization, the number of hospitalizations is expected to increase substantially as a result of the aging of the Canadian population.
Our model indicates an increase in Canada (excluding Quebec) of 4667 AMI hospitalizations by 2020, with a sensitivity range of 3691 to 5650. Quebec represents approximately 25% of the Canadian population, which suggests an increase of about 6200 for the whole of Canada (sensitivity range, 4900 to 7500). Using the CIHI estimate for the average cost of a myocardial infarction/shock/arrest without cardiac catheter in Canada in 2008/09 of $7412 [14
], the additional cost for this increase in AMI hospitalizations would be approximately $46 million (sensitivity range, $36-$55 million). If one assumes that the majority of AMI hospitalizations (80%) will include a cardiac catheter for which the average cost in Canada in 2008/09 was $8984 [14
], the additional cost for the projected increase in the number of AMI hospitalizations to 2020 would be approximately $54 million (sensitivity range, $42-$65 million). These figures do not take account of economic inflation, changes in AMI risk due to the epidemic in diabetes and obesity, or the fact that seniors generally have a longer LOS.
Attempting to incorporate changes in the use of PCI or CABG and in LOS with the projected trend in AMI hospitalizations was problematic. If one extrapolates the linear increase in the proportion of AMI hospitalizations with a PCI between 2002 and 2009, by 2020 almost all AMI hospitalizations would have a PCI. Projecting the proportion of AMI hospitalizations with a CABG would suggest that almost no AMI hospitalizations in 2020 would have a CABG. These scenarios are unrealistic. While the proportion of AMI hospitalizations with a PCI are likely to continue to increase and the proportion with a CABG may continue to decrease, it is anticipated that they will plateau at some point, but our data do not allow us to estimate where this could be expected to occur, although it seems likely that it will be before 2020. Similarly, the trend of shortening average LOS may continue for some years but eventually a minimum must be reached. Thus, we have not provided projections for these variables.
Our analysis has limitations. The CIHI data record only hospitalizations and not individual episodes of patient care and it was not possible to include Quebec in the analysis. The coding of hospitalizations changed from ICD-9 to ICD-10 between 2001/02 and 2002/03 in most provinces and territories but did not occur until 2003/04 in New Brunswick and 2004/05 in Manitoba. The change in coding system included a reduction in the eight-week period after an initial AMI hospitalization in which readmission for AMI-related problems would be counted as continuing care in ICD-9 to four weeks in ICD-10. This change in coding does not explain the decrease in 2006; we are unable to ascertain a reason for this reduction, but it should be noted that, if data from only 2002/03 to 2006/07 had been used, the results of the projection in numbers of AMI hospitalizations would likely have been quite different. We considered it appropriate to use all the data available rather than selectively use them. Fiscal year was used as a linear variable, which is a traditional approach, because it was considered to be the most reliable based on the limited number of years available and the structure of the data (Figures and ).
As previously mentioned, the projection in AMI hospitalization rates did not take account of the increasing prevalence of diabetes and obesity in the Canadian population (especially in young adults) or changes in cardiovascular drug therapy, such as increasing use of statins and anti-hypertensives. It also assumed that the treatment of cardiovascular disease and AMI in particular will remain largely the same over the next 10 years, although new clinical practices, in which physicians are able to predict the prognosis for an AMI patient with greater accuracy, could emerge that would lead to a reduction in the likelihood of hospitalization for AMI or the length of stay [25
]. These limitations point to the need for a more comprehensive and refined national disease surveillance system in Canada with which more in depth studies would be performed.