Polyscore of Non-invasive Cardiac Risk Factors

Abstract

Non-invasive risk stratification of cardiac patients has been the subject of numerous studies. Most of these investigations either researched unique risk predictors or compared the predictive power of different predictors. Fewer studies suggested a combination of a small number of non-invasive indices to increase the accuracy of high-risk group selection. To advance non-invasive risk assessment of cardiac patients, we propose a combination score (termed the Polyscore) of seven different cardiac risk stratifiers that predominantly quantify autonomic cardiovascular control and regulation, namely the slope of heart rate turbulence, deceleration capacity of heart rate, non-invasively assessed baroreflex sensitivity, resting respiration frequency, expiration triggered sinus arrhythmia, post-ectopic potentiation of systolic blood pressure, and frequency of supraventricular and ventricular ectopic beats. These risk stratification tests have previously been researched and their dichotomies defining abnormal results have been derived from previous reports. The Polyscore combination was defined as the number of positive tests among these seven risk predictors, giving a numerical scale which ranges from 0 (all tests normal) to 7 (all tests abnormal). The Polyscore was tested in a population of 941 contemporarily treated survivors of acute myocardial infarction (median age 61 years, 182 females) of whom 72 (7.65%) died during a 5-year follow-up. In these patients, all the risk predictors combined in the Polyscore were assessed during in-hospital 30-min simultaneous non-invasive recordings of high-frequency orthogonal electrocardiogram, continuous blood pressure and respiration. Compared to Polyscore 0 stratum, the hazard ratios of mortality during follow-up increased almost exponentially in strata 1 through 7 (vs. stratus 0, the hazard ratios were 1.37, 1.96, 7.03, 15.0, 35.7, 48.2, and 114, in strata 1 to 7, respectively; p < 0.0001). This allowed selecting low-risk (Polyscore ≤ 2), intermediate risk (Polyscore 3 or 4) and high-risk (Polyscore ≥ 5) sub-groups of the population that differed greatly in the Kaplan-Meier probabilities of mortality during follow-up. Since the Polyscore was derived from recordings of only 30-min duration, it can be reasonably applied in different clinical situations including population-wide screening. We can therefore conclude that the Polyscore is a reasonable method for cardiac risk stratification that is ready for prospective validation in future independent studies.

Keywords: all-cause mortality; blood pressure monitoring; combination of risk factors; electrocardiogram; low-risk and high-risk group separation; non-invasive autonomic testing; resting respiration; survivors of myocardial infarction.

Figure 1

All-cause mortality in the study population. The Kaplan–Meier curve of probability of death is shown together with its 95% confidence intervals. Numbers of patients at risk are shown below the time axis.

Figure 2

Contribution of individual elements of Polyscore to different Polyscore strata. For each stratum of Polyscore and for each individual risk factor, the graph shows the percentage of patients of the given Polyscore strata in whom this risk factor was positive. Note that the different contribution profiles allow approximate distinction of three groups of risk factors with BRS, respiration frequency, and ETA being more sensitive; TS and PESP being more specific; and DC and ectopic frequency in-between (see the text for further details). TS, turbulence slope; DC, deceleration capacity; BRS, baroreflex sensitivity; Respiration, average respiration frequency; ETA, expiration triggered sinus arrhythmia; PEST, post-ectopic systolic blood pressure potentiation; Ectopics, frequency of ventricular or supraventricular ectopic beats.

Figure 3

Hazard ratios of individual Polyscore strata. For each stratum 1, 2, …, 7, the figure shows the result of a univariable Cox regression model comparing the stratum with the lowest-risk stratum of Polyscore 0. The Hazard ratios are shown together with their 95% confidence intervals. Note the logarithmic scale of the vertical axis. Note also that the development of the hazard ratios allows defining a low-risk group of Polyscore ≤ 2 (shown in green), an intermediate risk group of Polyscore 3 or 4 (shown in blue), and a high-risk group of Polyscore ≥ 5 (shown in red). CI, confidence interval.

Figure 4

Comparison of Kaplan–Meier probabilities of death in the population sub-groups defined by Polyscore ≤ 2 (green), Polyscore 3 or 4 (blue) and Polyscore ≥ 5 (red). Albeit positively biased (see the text for details) the difference between the probabilities of death in these groups was statistically significant (p < 0.0001, χ² = 220). Numbers of patients at risk in the individual sub-groups are shown below the time axis.

Figure 5

The same comparison of Kaplan–Meier probabilities of death as shown in Figure 4 repeated separately for patients with the diagnosis of diabetes mellitus (top panel) and for patients without the diagnosis of diabetes mellitus (bottom panel). In both cases, the differences between the probabilities of death in these groups was statistically significant (χ² = 59 and χ² = 123, respectively, both p < 0.0001). Numbers of patients at risk in the individual sub-groups are shown below the time axes.