We used the 2007 register of the Kaunas University Hospital Department of Cardiology for medical care data for hospitalized patients. Kaunas University Hospital, a facility with approximately 2,000 beds, is the teaching hospital for the Kaunas University of Medicine. Lacking Lithuanian data, we used data from the United States to estimate service-delivery rates to ambulatory patients (4
). The MONICA research protocol was approved by the Kaunas Medical University institutional review board.
Our model divides the population into 3 prevalence pools: people with no apparent heart disease, people with symptomatic heart disease with a left ventricular ejection fraction (LVEF) greater than 35%, and people with symptomatic heart disease with an LVEF of 35% or less (4
). This division takes into account the marked differences in mortality among the pools and acknowledges that different types of interventions are most efficacious in each of the 3 pools.
We categorized acute cardiac events as out-of-hospital cardiac arrest, acute or emergent events, and disease discovered in the ambulatory setting. We subdivided acute/emergent events into acute myocardial infarction with ST-segment elevation (STEMI) on electrocardiogram (ECG), acute heart failure with an LVEF of 35% or less, acute myocardial infarction without ST segment elevation (nSTEMI) on ECG, and unstable angina or other acute cardiac events. The model can account for any intervention that would be directed at anyone in the population who is at risk for heart disease, has stable chronic heart disease, or is experiencing an acute event, because each person must belong to 1 of the 3 pools, and all types of acute events are subsumed under the 3 broad categories of acute events.
We selected the number of potentially postponable deaths (PPD) as the outcome of interest for this analysis. A similar outcome has been used to estimate the source of the change in death rates from heart disease in the United States and several other countries (5
). In this analysis, we calculated the number of deaths that can be prevented or postponed by improving risk factors or care as follows:
- PPD = (expected mortality reduction when the intervention is implemented) x (mortality rate) x (1 – current implementation rate) x (number in population).
The analysis used the cumulative-relative-benefit approach of Mant and Hicks to calculate the joint effect of simultaneous interventions (9
). This model has also been used to estimate the potential impact of improving care in the United States (4
In our analysis, we used mortality from all causes, for several reasons. Most intervention trials report outcomes in terms of total mortality. Reducing the burden of heart disease risk reduces total mortality and deaths from other chronic diseases, and using total mortality as the endpoint eliminates the possibility that an intervention simply results in death from a different cause rather than reducing the probability of death.
Prevalence and mortality data
We used the Kaunas Monitoring of Trends and Determinants in Cardiovascular Disease (MONICA) registry to estimate the prevalence of heart disease, and we used the Lithuanian death registry as the source of death rates for the subpopulation without heart disease (11
). We did not have access to accurate all-cause mortality rates for the subpopulations with symptomatic heart disease with an LVEF greater than 35% and symptomatic heart disease with an LVEF of 35% or less. Therefore, we used the mortality rate ratios from Olmsted County, Minnesota (12
). We estimated that the risk of death for people with heart disease and an LVEF greater than 35% is 2.84 times the risk of death for those without apparent heart disease, and the risk of death for people with heart disease and an LVEF of 35% or less is 11.02 times the risk of death for those without apparent heart disease.
We used published reports from clinical trials to estimate what the 1-year case-fatality rates for acute events would have been without the provision of modern treatments. On the basis of an epidemiologic observation (13
), we estimated that the LVEF is 35% or less in half the cases of heart failure.
Risk factor data
We used the 2001 Lithuanian MONICA registry risk factor data for the analysis (14
). At least 200 men and 200 women had been screened in every 10-year age group (35-44 y, 45-54 y, and 55-64 y). The response rate for the survey was 62.4%. The register contains data from 625 men and 778 women. The survey included physical measurements (blood pressure, height, body weight, and hip and waist circumference), blood samples for serum cholesterol levels, and face-to-face interviews by the research staff for information on smoking.
The MONICA smoking questionnaire included questions about smoking behavior (regular smoker, ex-smoker, never-smoker, occasional smoker), type of tobacco smoked (cigarettes, pipe, cigars), and number of cigarettes smoked per day. Participants who smoked at least 1 cigarette, cigar, or pipe per day were considered regular smokers.
MONICA uses standard mercury sphygmomanometers for blood pressure measurement. Blood pressure was measured from the right arm of the subject after 5 minutes of rest in a sitting position. The fifth phase of Korotkoff sounds was recorded as diastolic BP. The mean of 2 readings was used. Arterial hypertension was defined as a systolic blood pressure level greater than 140 mm Hg, a diastolic blood pressure level greater than 90 mm Hg, or both. Participants who had taken antihypertensive drugs in the last 2 weeks were classified as hypertensive regardless of their blood pressure level.
We used 95% confidence intervals, when available, to define a plausible range for the estimates of mortality reduction attributable to an intervention. Otherwise, we used ±20% of the expected value as the plausible range. For the plausible range of the current level of implementation, we used ±20% of the observed value. For estimates of the number of deaths prevented or postponed, we defined the lower bounds of the plausible range by the following product: the lower bounds of the estimates for the population size, expected mortality rate without intervention, and expected effect of the intervention and the upper bound of the current rate of intervention. We defined the upper bounds of the plausible range of deaths prevented or postponed by the following product: the upper bounds of the estimates for the population size, expected mortality rate without intervention, and expected effect of the intervention and the lower bound of the current rate of intervention.
Because we provided the plausible range for each of the values used in the calculations, the reader can estimate the impact of the achievable level of implementation. For example, the PPD associated with adequate physical activity is calculated to be 303.6 (). If the reader were to believe that the prevalence of physically active individuals could be increased by only 20 percentage points rather than 81 percentage points, the new PPD would be 303.6 × 20/81, or 75.0. This PPD can be compared with the PPD for any other intervention. For example, the maximum plausible PPD associated with increasing the rate of primary angioplasty for all patients with a STEMI is 11.6 ().
Estimated Impact of Interventions Before or Between Acute Cardiac Events, Lithuania
Estimated Impact of Interventions at the Time of an Acute Clinical Event, Lithuania