Estimation of the risk of coronary artery disease using logistic regression models for electrocardiogram parameters
| Author | Affiliation | |||
|---|---|---|---|---|
LT | ||||
| Date | Volume | Issue | Start Page | End Page |
|---|---|---|---|---|
2008 | 15 | 1 | 15 | 22 |
Tyrimo tikslas – įvertinti EKG parametrų ryšį su vainikinių arterijų (VA) spindžio susiaurėjimais (> 50 proc. ir/arba >70 proc.) ligoniams, kuriems buvo nustatyta stabilioji krūtinės ir nerasta EKG išeminių pokyčių. 171 ligoniui atlikta koronarografija bei registruota skaitmeninė EKG, įvertinti depoliarizacijos ir repoliarizacijos parametrai. Penkių EKG parametrų kompleksas (QTcmin, JTcd, αQRS-T, T index ir QRSmin) atskyrė ligonius, kuriems nustatyti VA susiaurėjimai ≥ 50 proc. ir < 50 proc. Rizikos balų sistema, pritaikius logistinės regresijos metodą, įvertino parametrus QTcmin>370 ms, JTcd>56 ms, αQRS-T>39, T index >1,56 ir QRSmin>84 ms, atitinkamai – 1, 4, 2, 1 ir 5 balais. Apskaičiuotas ROC kreivės plotas buvo 0,764. Jautrumas ir specifiškumas prognozuojant VA susiaurėjimus ≥ 50 proc. buvo atitinkamai – 80 proc. ir 62,9 proc., kai rizikos balas didesnis už 3. Parametrų kompleksas JTd apex≥44 ms, T plotas > 257, αQRS-T>39, QRSmin>84 ms ir amžiaus grupė parodė stiprų ryšį, prognozuojant VA susiaurėjimus ≥ 70 proc., ir įvertino šiuos parametrus atitinkamai – 4, 4, 3, 4 ir 1 balais. Apskaičiuotas ROC kreivės plotas buvo 0,865. Jautrumas ir specifiškumas prognozuojant VA susiaurėjimus ≥ 70 proc. buvo 84,3 proc. ir 58,28 proc., kai rizikos balas didesnis už 7. EKG parametrų kompleksas gali padėti nustatyti VA susiaurėjimus ≥ 50 proc. arba ≥ 70 proc. ligoniams, sergantiems išemine širdies liga.
The objectives of this study were to investigate the relation between the changes of electrocardiogram (ECG) parameters and the presence of angiographically significant coronary artery disease (CAD) (> or =50%, and > or =70% lumen stenosis) in patients previously hospitalized for CAD event without ischaemic ECG changes. Digital ECG was recorded (discretisation parameters 12 bit, 2 kHz, recording interval of 10 s, 12 standard leads) and coronary angiography were performed in 171 patients. The mean age was 56.7 ± 8.4 years. Angiographically nonsignificant CAD (stenosis ≤30% or stenosis≤70%) were verified in 66 and 85 of the patients; significant CAD (stenosis >50% or >70%) were found in 105 and 86 patients, repectively. Multivariate logistic regression analysis was used to estimate probability of CA stenosis and the accuracy of this logistic model was estimated by ROC curve. Applying 5 binary ECG-parameters (QTcmin, JTcd, αQRS-T, T index and QRSmin) a logistic regression formula can be derived which separates the two groups of patients with and without CA lumen ≥ 50 %. Based on the odds ratio of the logistic regression, QTcmin>370 ms, JTcd>56 ms, αQRS-T>39, T index >1.56 and QRSmin>84 ms were given scores of 1, 4, 2, 1, and 5 respectively. The accuracy of this risk score was estimated by ROC analysis and the area under ROC curve was 0.764. Sensitivity and specificity of CA lumen ≥ 50% was 80 % and 62.9% when risk score was higher than 3. Multivariate analysis identified the combination of JTd apex≥44 ms, T area > 257, αQRS-T>39, QRSmin>84 ms and range of age showed the strong association with coronary artery lumen ≥ 70 %, and were given scores 4, 4, 3, 4, and 1 respectively. The area under ROC curve was ROC analysis yielded area under curve of 0.865, indicating excellent discrimination power of this prognostic score for suspicion of coronary artery lumen ≥ 70 %. Sensitivity and specificity for prognosis of CA lesions ≥ 70% was 84.3% and 58.