| Acute myocardial infarction (AMI) is a critical illness in the emergency departmen and has a higher early mortality. It is a huge challenge for emergency and cardiovascular physicians to predict, prevent and reduce the mortality at the acute stage. Malignant ventricular arrhythmia (MVA) and cardiogenic shock are the main causes of death at the early stage of AMI, ventricular fibrillation (VF) and ventricular tachycardia (VT) are the most common reasons to induce sudden cardiac death (SCD), thus they are often life-threatening and portend a poor prognosis. Exploring the risk factors of MVA, particularly high risk factors that could be corrected, can help physicians in the emergency department to identify patients with high risk of MVA and prevent it, then improve the prognosis. Potassium is a major determinant of the electrophysiologic properties of the heart, and it can affect the threshold of ventricular fibrillation, the depolarization and repolarization of myocardial membrane. It has been found that hypokalemia is frequently observed at the early stage of AMI and often coexists with MVA, however, it needs to be further studied whether hypokalemia is the cause of MVA and could increase inhospital mortality or not. Although it has been recognized that hypokalemia often coexists with AMI, but the threshold for the diagnosis of hypokalemia varied widely which resulted in the difference of reported incidence of hypokalemia. The clinical features of hypokalemia at the early stage of AMI and the optimal cutoff value of serum potassium for predicting clinical outcome remain unclear.Although several multivariable models have been developed for risk stratification in AMI patients, but there are limitations to the current models, which include: 1) the requirement for detailed information which may not be readily available at the time of emergency admission, such as the GRACE and CADILAC score. 2) Exclusion of new or emerging prognostic variable or laboratory parameters, such as TIMI risk score. Moreover, studies have found that these scores were prediction of mortality but not of MVA. It is important to look for a simple bedside risk score which is suitable for the application in the emergency department. This study was designed to explore the clinical characteristics and possible influencing factors of hypokalemia at the early stage of AMI and to assess the value of hypokalemia in predicting MVA .We also set out to assess whether hypokalemia could improve the predictive value of TIMI risk score, and to develop a new simple model for predicting inhospital outcome of AMI in the emergency department.The study was divided into five parts:Part I Study on the relationship between hypokalemia and malignant ventricular arrhythmia at the early stage of acute myocardial infarctionObjective: To explore the incidence, phase process and influencing factors of hypokalemia and the relationship between hypokalemia and malignant ventricular arrhythmia at the early stage of acute myocardial infarction.Methods: This is a retrospective cohort study focusing on MVA occurrence for patients with first attack of AMI. A total of 758 patients who were admitted to the emergency department of our hospital from January, 2002 to January, 2008 were eligible for inclusion. Baseline clinical characteristics and serum potassium concentration were collected by a retrospective review of medical records for each patient. Patients were categorized into hypokalemia (LK) or normokalemic (NK) group according to the threshold of serum potassium less than 3.5mmol/L. Explore the incidence of hypokalemia and MVA during 24h post AMI. Baseline clinical characteristics and MVA occurrence were compared between group LK and NK in order to explore the influencing factors, phase process of hypokalemia and its relationship with MVA attack. High risk factors that were independent predictors of MVA were identified by binary logistic regression model. The prognostic value of hypokalemia in predicting MVA was also evaluated.Results: During the 6-year period of the study, 758 patients were eligible for the inclusion. The incidence of hypokalemia was 9.5% (n=72). Earlier presentation to the emergency department, anterior wall myocardial infarction, higher serum glucose level and peak CKMB were associated with the incidence of hypokalemia. The incidence of MVA was 9.9% (n=75) and the incidence of MVA was significantly greater in patients with hypokalemia compared with those classified as normokalemia (29.2% vs.7.9%, P<0.001).The incidence of hypokalemia in Group A (patients within 3 hours from onset to admission) was 22.3% (25/112), with 11.3% (29/256) in Group B (within 3 to 6 hours) and 4.6% (18/390) in Group C (within over 6 hours), which had a statistical difference among these three groups (x2=32.45, p<0.001). In Group A, the incidence of hypokalemia was the highest and the level of serum potassium was the lowest (3.82±0.59 vs. 4.07±0.36 vs. 4.24±0.43, p<0.001). The incidences of MVA were 24.1% (27/112), 12.1% (31/256) and 4.4% (17/390) respectively among these three groups.Patients were categorized into 1 of 2 study groups, MVA group and non-MVA group. The serum potassium level of MVA group was lower than that of non-MVA group (3.8mmol/L vs. 4.2mmol/L, P<0.001). Compared with non-MVA group, the ratio of anterior wall myocardial infarction (57.3% vs.27.7%, P<0.001) and KillipⅡ-Ⅳ(44% vs.10.2%, P<0.001)were higher in MVA group; presentation to the emergency department was earlier (6h vs.7h, P=0.008), the systolic blood pressure was lower(118mmHg vs. 124mmHg, P=0.031), the heart rate was faster (78bpm vs. 71bpm, P=0.01), serum glucose level was higher (7.5mmol/L vs. 6.8mmol/L, P=0.019), the white blood cell count was higher (9.66×109/L vs. 8×109/L, P=0.004) in MVA group.The binary logistical regression model was performed to explore the risk factors of MVA attack during 24h after AMI, and the independent variables included in analysis were as follows: hypokalemia (OR=4.33, 95%CI=2.17-8.66, P<0.001), presentation to the emergency department less than 6h (OR=2.06, 95%CI=1.16-3.66, P=0.013), heart rate more than 100bpm (OR=2.76, 95%CI=1.11-6.41, P=0.018), the systolic blood pressure less than 100mmHg (OR=2.60, 95%CI=1.2-5.64, P=0.015), the ratio of KillipII-IV (OR=2.43, 95%CI=1.80-3.29, P<0.001), the ratio of anterior wall myocardial infarction (OR=2.28, 95%CI=1.31-3.97, P=0.003), serum glucose level (OR=1.69, 95%CI=1.09-2.59, P=0.018) and the white blood cell count more than 10×109/L (OR=2.06, 95%CI=1.19-3.57, P=0.01).Conclusions: AMI patients are susceptible to acute hypokalemia at the early stage of AMI. The phase process of hypokalemia is characterized by the highest incidence of hypokalemia within 3 hours from onset to admission with the corresponding increase of MVA attack. Hypokalemia is an independent risk factor of MVA attack during 24 hours from onset to admission.Part II Comparative study on different grouping methods of serum potassium levels in the prediction of MVA at early stage of AMIObjective: To evaluate the predictive value of different levels of serum potassium classified by different methods on MVA attack during 24 hours from onset to admission.Methods: Based on the same database of Part l, we observed the predictive value on MVA of different levels of serum potassium classified by different methods. First, the enrolled patients were categorized into 5 continuous groups: Group A, SK <3.0mmol/L; Group B, 3.0mmol/L≤SK <3.5mmol/L; Group C: 3.5mmol/L≤SK<4.0mmol; Group D, 4.0mmol/L≤SK<4.5mmol/L; Group E, SK≥4.5mmol/L; and the MVA occurrence during 24h post AMI were compared among groups. Then, the relation between quartile groups and MVA occurrence was analyzed. Comparing nested models for MVA attack during 24h post AMI were constructed to evaluate the prognostic value of the 2 classification methods for admission serum potassium. The receiver operating characteristic (ROC) curve was used to evaluate predictive value of admission serum potassium concentration on MVA occurrence and to identify the best cut-off point of serum potassium. The prognostic value of serum potassium for MVA was assessed again according to the best cut-off.Results: when the patients were classified into 5 groups, there was a significant difference of MVA occurrence among the groups (P<0.001), 75% in Group A, 23.4% in Group B, 13.6% in Group C, 8.5% in Group D and 0.5% in Group E. The binary logistical regrssion model was performed to explore the risk factors of MVA occurrence during 24h post AMI and the independent variables included in analysis were as follows: hypokalemia, the systolic blood pressure less than 100mmHg, anterior wall myocardial infarction, serum glucose level, white blood cell count, the time presentation to the emergency department, the peak CKMB and Killip II-IV. Compared with Group E, the risk for MVA attack in Goup A (OR=646.06, 95%CI=60.98-6844.7, P<0.001, Group B (OR=37.80, 95%CI=7.17-199.13, P<0.001), Group C (OR=77.71, 95%CI=14.60-407.30, P<0.001) and Group D (OR=25.94, 95%CI=5.33-126.16, P<0.001) were all increased. Combined the Group A, B, C and D into one group-Group IA-D, and serum potassim was analyzed as a 2 categorical variables, then we found that SK≥4.5mmol/L was the protective factor for MVA attack (OR=0.01, 95% CI: 0.001-0.09, P<0.001).Patients were stratified into quartiles groups (Ql to Q4) by admission serum potassium concentrations of Ql (SK≤3.8mmol/L), Q2 (3.8mmol/L4.5mmol/L). There was a significant difference of MVA attack among the different groups (P<0.001), 19.5% in Group Q1, 12.3% in Group Q2, 6.6% in Group Q3, 0.5% in Group Q4.The binary logistical regrssion model was performed to explore the risk factors of MVA occurrence during 24h post AMI and the independent variables included in analysis were as follows: hypokalemia, Killip ll-IV, the systolic blood pressure less than 100mmHg, serum glucose level, white blood cell count, the time presentation to the emergency department, the peak CKMB and anterior wall myocardial infarction. Compared with Group Q4, the risk for MVA attack in Goup Q1 (OR=83.02, 95% CI=16.79-410.45, P<0.001), Group Q2 (OR=97.46, 95% CI=17.92-530.04, P<0.001) and Group Q3 (OR=14.83, 95% CI=2.90-75.74, P=0.001) were all increased. Combined the Group Q1, Q2, and Q3 into one group-Group Q1-3, and serum potassium was analyzed as a 2 categorical variables, then we found that SK>4.5mmol/L was a protective factor for MVA attack (OR=0.01, 95% CI=0.002-0.10, P<0.001).Nested models were compared to determine whether logistic regression models that included the method of five groups provided a significantly better fit than did logistic regression models included quartile groups. It showed that both classification methods can be used to analyze the relation of SK and MVA attack, there was no significant difference between the 2 classification methods.The area under the ROC curve (AUC) of serum potassium for predicting MVA occurrence during 24h post AMI was 0.74(95% CI=0.71-0.74, P<0.001), the best cutoff value was 4.14mmol/L, the sensitivity was 82.7% and specificity was 53.6%; the sensitivity was 28% and specificity was 92.5% when SK≤3.49 (<3.5) mmol/L; the sensitivity was 100% and specificity was 25.8% when SK≤4.5mmol/L. The MVA occurrence was higher in patients with SK≤4.14mmol/L than that in patients with SK>4.14mmol/L (16.4% vs.3.4%, P<0.001).According to the best cutoff value, serum potassium was analyzed as a 2 categorical variables, then we found that SK>4. 14mmol/L was a protective factor for MVA attack ((OR=0.06, 95%CI=0.03-0.13, P<0.001).Conclusions: Hypokalemia was a predictor of incidence of MVA attack regardless of the classification method for serum potassium. There was also higher risk for MVA attack when the serum potassium was at a lower nromal level. It was a protective factor for MVA attack when the serum potassium level was higher than 4.14mmol/L or 4.5mmol/L according to the different classifications of serum potassium.Part III Study on the clinical features of hypokalemia and preventive effect of intravenous potassium therapy for MVA at the early stage of acute myocardial infarctionObjectives: To explore whether the serum potassium level changed acutely after the AMI attack by comparing the differences of serum potassium levels among patients with AMI, healthy people and patients with chronic coronary heart disease. To explore the clinical characteristics and possible influencing factors of hypokalemia coexisted with early AMI by comparing with hypokalemia caused by other reasons. And to explore the preventive effect of intravenous potassium therapy for MVA at the early stage of acute myocardial infarction via observing the effect of different protocols of potassium therapy on the incidence of MVA. Methods: A total of 537 patients who were admitted to the emergency department of our hospital from February, 2008 to October, 2010 were candidates for the study. According to the time from onset to admission, all the patients with AMI were grouped as A (≤3h), B (3-6h) and C (>6h). The study also included 50 healthy people (group D), 50 patients with chronic coronary heart disease (group E) and 147 patients with hypokalemia caused by other reasons (group F). The incidence of MVA was compared between different groups. Patients with hypokalemia and AMI were classified as group G. Baseline clinical characteristics, muscle strength, serum potassium concentration, duration and amount of potassium supplement, incidence of MVA were compared between group F and G.All the patients with AMI were randomly divided into group K1 and K2. Potassium supplement therapy was given when SK<3.5mmol/L in group K1 and when SK<4.5mmol/L in group K2. Incidence of MVA was compared between the 2 groups. We also analyze those of the subgroup in which the serum potassium level was 3.5-4.5mmol/L.Results: The incidence of hypokalemia was 14.2% in the total patients with AMI, the incidence of hypokalemia in Group E was 4%, and there was no hypokalemia in Group D. The incidence of hypokalemia was 29.2% in Group A, 14.2% in Group B and 8.8% in Group C respectively. The incidence of hypokalemia was higher in Group A compared with that in Group E (χ2=12.76, P<0.001), Group B (χ2=8.69, P=0.003) or Group C (χ2=24.20, P<0.001).The serum potassium level of patients with AMI was lower than that of patients in Group D (4.09±0.36mmol/L vs. 4.29±0.31mmol/L,P=0.02), the serum potassium level was lower in Group A than that in Group D (3.92±0.57mmol/L vs. 4.29±0.31 mmol/L,P<0.01) and E (3.92±0.57mmol/L vs. 4.17±0.38mmol/L, P<0.05). The serum potassium level was lower in Group B than that in Group D (4.04±0.55mmol/L vs. 4.29±0.31mmol/L,P<0.01) and E (4.04±0.55mmol/L vs. 4.17±0.38mmol/L,P<0.05). The serum potassium levels in Group A (3.92±0.57mmol/L vs. 4.19±0.53mmol/L) and Group B (4.04±0.55mmol/L vs. 4.19±0.53mmol/L) were lower than that in Group C. There was no significant difference between Group E and D (4.17±0.38mmol/L vs. 4.29±0.31mmol/L, P>0.05), Group C and D (4.19±0.53mmol/L vs. 4.29±0.31mmol/L, P>0.05) or E (4.19±0.53mmol/L vs. 4.17±0.38mmol/L, P>0.05).The incidence of MVA in patients with AMI was 8.9% (n=48). There was significant difference in incidence of MVA among Group A (16.7%), Group B (8.3%) and Group C (6.6%) (P=0.011).Seventy-six patients with AMI were diagnosed as hypokalemia (group G), among them 17 patients (22.4%) was moderate hypokalemia, 59 patients (77.6%) were mild hypokalemia and none was severe hypokalemia. 72 (49%) of the 147 patients in Group F were severe hypokalemia, 36 (24.5%) were moderate hypokalemia, 39 (26.5%) were mild hypokalemia. There was significant difference between the patients with AMI and without AMI (P<0.001).A total of 75 patients with mild to moderate hypokalemia and without AMI were defined as group H. Compared Group H with Group G, we found the follows: the incidence of MVA was lower in Group H than that in Group G (5.3% vs. 21.1%, P=0.004). the duration of potassium supplement was significantly shorter in Group G compared with that in Group H (6.08±2.28h vs. 13.37±5.45h, P<0.001) and the amount of supplementary potassium in Group G was significantly less than that in Group H (4.17±1.84g vs. 9.73±2.74g, P<0.001). Most patients in Group G had normal muscle strength except for 4 patients whose muscle strength were grade IV, while the decreased muscle strength was more apparent in Group H, grade I, 2.7%, grade II, 2.7%, grade III, 26.7%, grade IV, 46.7%, grade V, 21.3%, there was significant difference between the 2 groups (P<0.001). PVCs, atrial premature, atrial fibrillation, QT interval prolongation, changes of T wave and U wave were also seen in ECG of patients in Group F, but these typical changes of ECG were not seen in patients of Group G.There were 245 patients in Group K1 and 292 patients in Group K2 enrolled. The incidence of MVA was 10.6% in Group K1 and 7.5% in Group K2, there was no significant difference between the 2 groups (P=0.213). And subgroup analysis showed that there was significant difference of MVA occurrence between Group K1sub (7.7%) and Group K2sub (3.2%) (P=0.028). Conclusions: The average serum potassium concentration was lower in patients with AMI compared with that in healthy people and in patients with chronic coronary heart disease. The highest incidence of MVA was seen within 3 hours post AMI. The hypokalemia occurring at the early stage of AMI had the following features: mainly mild to moderate hypokalemia; rarely severe hypokalemia; higher incidence of MVA and no obvious decline of muscle strength; without typical ECG of hypokalemia such as QT interval prolongation and u waves. The hypokalemia was easy to be corrected which needs less potassium volume and less time than hypokalemia caused by other reasons. All of these indicate that the most possible mechanism of hypokalemia at the early stage of AMI was a stress-induced intracellular-serum potassium shift. The mortality induced by MVA may be dropped by active potassium supplement. The incidence of MVA may be decreased by potassium supplement when the serum potassium level was 3.5-4.5mmol/L. Intravenous metoprolol can be effective in treating ventricular tachycardia storm, and to some extent can help correct hypokalemia. Potassium therapy has a preventive effect on MVA at the early stage of AMI and can reduce the incidence and mortality of MVA.Part IV Auxiliary diagnostic value of serum potassium in addition to TIMI risk score in the prediction of malignant ventricular arrhythmia at the early stage of acute myocardial infarctionObjective: To evaluate if the admission serum potassium level could improve the prognostic value of TIMI risk score in predicting MVA attack during 24h post AMI.Methods: we assessed the predictive value of TIMI risk score for MVA occurrence during 24h post AMI by ROC curve analysis.In order to evaluate if serum potassium could add prognostic value to TIMI risk score, we add the information of serum potassium into the TIMI risk score and calculate serum potassium-modified TIMI risk score. To calculate the weight of serum potassium in the new score, Killip II-Ⅳwas taken as a reference. The ratio between the OR of serum potassium and KillipⅡ-Ⅳwas multiplied by 2 points in order to determine the score of the serum potassium in the new score. The accuracy of this new score was compared with the traditional TIMI-risk score by comparing the AUC of the 2 scores.Results: The AUC of TIMI risk score for predicting MVA attack during 24h post AMI was 0.57 (95% CI=0.49-0.65, P=0.108), the sensitivity was 68.7%, and the specificity was 45.4%. The AUC of TIMI risk score for predicting inhospital mortality was 0.74 (95% CI=0.66-0.81, P=0.039), the sensitivity was 80% and the specificity was 70.8%, the best cutoff was TIMI>3. According to the OR in the logistic regression, the different scores of serum potassium according to the different threshold were determined and incorporated into the TIMI risk score to create different modified TIMI risk scores: SK-TIMI1 (SK<3.5mmol/L), SK-TIMI2 (SK≤4.14mmol/L), SK-TIMI3 (SK<4.5mmol/L). The AUC of SK-TIMI1 for predicting MVA attack during 24h post AMI was 0.72 (95%CI=0.65-0.80), P=0.04, sensitivity was 77.1%, specificity was 78.9%; The AUC of SK-TIMI2 for predicting MVA attack during 24h post AMI was 0.62 (95%CI=0.55-0.70), P=0.039, sensitivity was 93.7%, specificity was 55.8%; The AUC of SK-TIMI3 for predicting MVA attack during 24h post AMI was 0.59 (95%CI=0.51-0.68), P=0.042, sensitivity was 52.1%, specificity was 60.9%; there was significant difference between the AUC of TIMI risk score and that of SK-TIMI1 (Z=4.61, P<0.001) or SK-TIMI2 (Z=2.592, P=0.01) in relation to MVA attack, but there was not significant difference between the AUC of TIMI risk score and that of SK-TIMI3 (Z=1.497, P=0.134).Conclusions: TIMI risk score has a good predictive value for inhospital mortality, but a poor predictive value for MVA attack during 24h after AMI. Admission serum potassium level could improve the predictive ability of TIMI risk score for MVA attack, and the best threshold for this value is <3.5mmol/L.Part V Role of modified risk score in the prognosis of patients with acute myocardial infarctionObjective: To develop a new model for prediction of MVA attack of patients with AMI based on the logistic regression.Methods: A predictive risk score was developed for MVA attack based on the logistic regression using parameters available on admission to the ED. The model was developed on patients from the first stage (January, 2002 to January, 2008) and validated on patients in the second stage (February, 2008 to October, 2010). The points of the variables were determined according to the OR in the logistic regression: 1 point for 2≥OR≥1.5; 2 points for 2.5≥OR≥2 and 3 points for OR≥2.5. The predictive ability of the risk score for MVA attack was assessed by ROC curve.Results: The variables of the risk score for predicting MVA attack during 24h after AMI included hypokalemia, anterior wall MI, Killip II-IV, heart rate, systolic blood pressure, blood glucose, white blood cell count, the time from onset to admission, the total points of the score is 19, we called it Risk-score1.In the first stage, the AUC of Risk-score1 for predicting MVA attack was 0.82 (95%CI=0.77-0.88, P<0.001), the best cutoff was >4 points, and the sensitivity was 74%, specificity was 86.7%. The AUC of TIMI risk score for predicting MVA attack was 0.59 (95%CI=0.49-0.68, P=0.0676), the best cutoff was >3 points, and the sensitivity was 52.3%, specificity was 61.3%. There was significant difference between Risk-score1 and TIMI risk score (Z=2.066, P=0.002).We tested the modified risk scores on the validation set during February, 2008 to October, 2010. the AUC of Risk-score1 for predicting MVA attack was 0.78 (95%CI=0.71-0.86, P<0.001), the best cutoff was >4 points, and the sensitivity was 78.7%, specificity was 82%, the AUC of TIMI risk score for predicting MVA attack was 0.57 (95%CI=0.49-0.65, P=0.108). There was significant difference between the 2 scores (Z=3.645, P=0.001). The incidence of MVA increased as Risk-score1 increased: 3.6% when Risk-score1≤4, 23% when Risk-score1=5-7 and 38.2% when Risk-score1≥8.Conclusions: The modified risk score based on the logistic regression of our hospital data has a better predictive value for MVA attack during 24h after AMI than that of TIMI risk score.
|
PDF Full Text Request