This study was based on administrative data from the Danish National Patient Registry, which covers all admissions to Danish hospitals since 1978.11
Briefly, the registry contains basic information about the patient (age and gender), dates of admission and discharge, and all diagnoses during hospital stay. Data for the registry are processed at each hospital discharge and include a primary discharge diagnosis coded according to the International Classification of Diseases (ICD) by the discharging physician. We identified patients aged 30 years or older with a first admission for an AMI during the period 1 January 1985 to 31 December 2002. The population was subsequently tracked for first hospitalisation for recurrent AMI occurring within the first year after discharge. The rationale for the follow‐up duration of 1 year was a clustering of events during the first 6 months and far fewer events after this. Information on survival status was obtained from the Civil Registration System which registers all deaths in Denmark within 2 weeks. One‐year mortality status was available for all patients.
Definitions of a first AMI and a recurrent AMI
A first AMI was defined as first hospital admission recorded in the National Patient Registry on condition that no previous AMI had been recorded in the preceding 7 years. This arbitrary limit was selected because this was the furthest back we were able to trace patients admitted in 1985, and this length of time was applied to all patients in our cohort. To identify the hospitalisations for a first AMI or a first recurrent AMI we used the ICD‐8 code 410 (from 1978 to 1994) and ICD‐10 code I21–22 (from 1994 to 2002). Only primary discharge diagnoses were used. A recurrent AMI was defined as any new admission for AMI within 1 year after discharge following the initial AMI, provided that the interval between the day of admission for first AMI and the day of admission for the recurrent AMI was at least 5 days to avoid misclassification. If a patient experienced more than one reinfarction, only the first was used in analyses.
Validation of the diagnosis of a first and a recurrent AMI
The diagnosis of AMI in the National Hospital Registry has been validated against the Danish Monitoring Trends and Determinants of Cardiovascular Disease (MONICA) Registry where standardised WHO definitions of AMI are used. The positive predictive value was 93.6% and sensitivity was 78% for definite or possible AMI (93% for definite AMI).12
To study the validity of a recurrent AMI diagnosis, we randomly validated 48 diagnoses in two hospitals by a review of the hospital records. To be classified as a definite recurrent AMI, two out of three criteria had to be fulfilled: clinical symptoms together with either significant enzyme/troponin rise or ST‐segment changes. We only chose recent cases between 2000 and 2001 so that we could obtain a complete sample of the records. Of these 48 patients, 40 (83%) were retrospectively found to have a definite recurrent AMI and an additional three had possible reinfarction, leaving 90% with a diagnosis of definite or possible reinfarction. In patients where the diagnosis of a recurrent AMI could not be confirmed, two had angina, one had congestive heart failure and two were incorrectly coded as AMI during revascularisation procedures.
The calendar years from 1985 to 2002 were categorised into four pre‐selected periods: 1985–1989, 1990–1994, 1995–1999 and 2000–2002. We chose a shorter interval for the latest period to make a clear cut at year 2000, when the definition of AMI changed according to international guidelines. The remaining periods were divided into equal intervals of 4 years. Due to the frequent use of early reinfarction as the endpoint in recent trials, we investigated trends in mortality from both early and late recurring AMI. A recurrent AMI was categorised as early if it occurred within the first 30 days of the initial AMI (counting from the day of admission) or late if it occurred from day 31 to 365.
Trends in continuous baseline variables were tested by one‐way variance analysis and categorical variables by χ2 test. One‐year case‐fatality plots, stratified by period, were generated separately for the whole AMI population, that is, all patients with first time AMI, and for patients with a recurrent AMI by means of the Kaplan–Meier method and compared by the log‐rank test. Lifetime was calculated as time from admission for either first AMI or recurrent AMI.
A Cox proportional hazard model was used to estimate hazard rates of case‐fatality from first AMI or recurrent AMI during periods with the earliest period (1985–1989) as reference. The model was adjusted for sex and age (10‐year age groups). To compare risk of death within the first year between patients with a recurrent AMI and patients without a recurrent AMI, an extended Cox model was used, in which recurrent AMI served as a time‐dependent variable. In brief, this model switches a patient with recurrent AMI from the initial value of 0 (at the time of the index admission) to a value of 1 at the time of the reinfarction. In this way, the period that patients had survived to the moment of the recurrent AMI was calculated as a survival period without exposure. Patients entered the model from the discharge date of their first AMI (delayed entry) to adjust for any differences in length of hospital stay during periods. Thus, only hospital survivors of the initial admission were included in these analyses.
Hazard ratios were reported as relative risks with 95% confidence intervals (CIs), obtained from the likelihood ratio test. Calculations were made with Statistical Analysis System software, version 8.2 (SAS Institute, Cary, NC).
According to Danish law, retrospective register studies do not require ethical approval. The study was approved by the Danish Data Protection Agency.