Original article
The validity of the diagnosis of acute myocardial infarction in routine statistics: A comparison of mortality and hospital discharge data with the Danish MONICA registry

https://doi.org/10.1016/S0895-4356(02)00591-7Get rights and content

Abstract

We analyzed the validity of the diagnosis of acute myocardial infarction (MI) in national registries by individual record linkage with the Danish MONICA registry. MI events were identified by record linkage between The National Hospital Registry and The National Death Registry. The sensitivity and specificity were estimated based on three different definitions of an MI event. For events with MI as the primary diagnosis or underlying cause of death the predictive value was 93.6% and sensitivity 77.6% when compared to DANMONICA definite or possible MI. Compared to DANMONICA, definite MI the predictive value was 75.0% and the sensitivity 92.8%. When secondary diagnosis and contributory cause of death were included, the sensitivity increased and the predictive value decreased only slightly. The combination of the national registries was found to be a valid and powerful tool for monitoring the population incidence of MI.

Introduction

The decline in coronary heart disease mortality in most industrialized countries during recent decades is of major significance for public health [1]. The mortality rate depends on the number of new cases and on the survival among the patients with the disease. The decline in mortality may thus either be caused by a decline in the incidence of coronary heart disease because of healthier lifestyles or improved social conditions in the population or from better survival as a result of better treatment or a changed spontaneous course of the disease [2].

Most countries have two main sources of routinely collected information on trends in heart disease: mortality statistics based on death certificates, and morbidity statistics based on hospital discharge data. Both data sources are based on classification and coding of the clinical diagnosis leading to death or hospital admission according to the International Classification of Diseases (ICD). These data are now increasingly used to compare incidence and mortality over time and among countries 1, 3, 4, 5, 6.

In Denmark, the Danish Registry of Causes of Death include all deaths since 1943 [7] and the National Patient Registry include all hospitalizations since 1978 [8]. These two nationwide registries now allow the analysis of mortality trends for more than 50 years and the morbidity (measured by hospitalization) for more than 20 years.

Most cases of symptomatic acute myocardial infarction (MI) lead to hospital admission or sudden death outside the hospital. Information about hospital discharges and deaths may be combined to identify events of MI by application of appropriate algorithm. This record linkage can be used to estimate the incidence of MI in the population if the registries are carefully validated.

The WHO MONICA (Monitoring Trends and Determinants in Cardiovascular Disease) Project provides a unique opportunity to validate routine mortality and morbidity statistics. The MONICA Project was a multicenter study monitoring the incidence of MI, levels of risk factors, and patterns of therapy over 10 years [2]. The study included 38 well-defined populations in 21 different countries, and all cases of MI and stroke over 10 years were registered according to standardized diagnostic criteria. A drawback of this study is that none of the centers had national coverage, and the degree to which the results can be generalized to the national level in each country is not obvious. In many countries, the routine statistics are nationwide but do not standardize the diagnostic criteria to the same degree as in the MONICA Project. It is, therefore, of great importance to utilize the MONICA data for evaluating the national registries.

Validation studies of routine statistics have been carried out over the years with somewhat heterogeneous results due to differences in methodology or reflecting true differences in the validity of the routinely collected data between countries. Some studies have been carried out comparing community registries with national statistics 9, 10. Other studies have been based on a comparison of routine mortality statistics and hospital statistics and data from the MONICA Project 11, 12. In general, the hospital statistics seem to be fairly valid and the mortality statistics somewhat less valid 4, 13, 14. These findings stress the importance of validating the national registries to determine whether and how they can be used to reflect “true” incidence and mortality.

Thus, this study analyzed the validity of the diagnosis of acute MI in the National Patient Registry and the Danish Registry of Causes of Death by individual-based comparison of the national registries and the registrations in the Danish MONICA study (DANMONICA). We also compared the estimated age-specific incidence of first MI based on routine statistics and on the MONICA registry.

Section snippets

The Danish Registry of Causes of Death

The Danish Registry of Causes of Death includes individual records of all deaths in Denmark since 1943. The record contains the name, personal identification (the unique CPR number assigned to all Danish citizens and other residents of Denmark since 1968), and information from the death certificate including birthplace, date, and place of death and classification of the underlying and contributory causes of death. During the period of the Danish MONICA (DANMONICA) study, the causes of deaths

Results

From 1982 to 1991, 5,539 MI admissions were registered in the National Patient Registry. Most cases (4,768 or 86%) had MI as the primary diagnosis. Record linkage to the DANMONICA registry identified 5,075 (92%) of the admissions with an attack date within 28 days before or after the admission date in the National Patient Registry, and a further 5% were found in the registry with another attack date. The admission could not be found in 3% of the cases.

In the same period, 2,378 deaths were

Discussion

This study showed an overall high agreement between the national routine statistics and the DANMONICA registry. Both the predictive value and the sensitivity were higher for admissions compared with mortality data. The use of MI as either the primary or secondary diagnosis or underlying or contributory cause of death led to a reasonable balance between the predictive value and sensitivity for both hospital admissions and fatal events. Extending the definition by including all deaths with other

Conclusion

In conclusion, we find the combination of the national registries to be a valid and powerful tool for monitoring cardiovascular risk in the general population.

References (17)

There are more references available in the full text version of this article.

Cited by (440)

View all citing articles on Scopus
View full text