ORIGINAL ARTICLE Year : 2009  Volume : 4  Issue : 3  Page : 124127 A computerbased matrix for rapid calculation of pulmonary hemodynamic parameters in congenital heart disease Antonio Augusto Lopes, Rogerio dos Anjos Miranda, Rilvani Cavalcante Goncalves, Ana Maria Thomaz Department of Pediatric Cardiology and Adult Congenital Heart Disease, Heart Institute (InCor), School of Medicine, University of Sao Paulo, Sao Paulo, Brazil Correspondence Address: Background : In patients with congenital heart disease undergoing cardiac catheterization for hemodynamic purposes, parameter estimation by the indirect Fick method using a single predicted value of oxygen consumption has been a matter of criticism. Objective : We developed a computerbased routine for rapid estimation of replicate hemodynamic parameters using multiple predicted values of oxygen consumption. Materials and Methods : Using Microsoft^{®} Excel facilities, we constructed a matrix containing 5 models (equations) for prediction of oxygen consumption, and all additional formulas needed to obtain replicate estimates of hemodynamic parameters. Results : By entering data from 65 patients with ventricular septal defects, aged 1 month to 8 years, it was possible to obtain multiple predictions for oxygen consumption, with clear betweenage groups ( P <.001) and betweenmethods ( P <.001) differences. Using these predictions in the individual patient, it was possible to obtain the upper and lower limits of a likely range for any given parameter, which made estimation more realistic. Conclusion : The organized matrix allows for rapid obtainment of replicate parameter estimates, without error due to exhaustive calculations.
Materials and Methods The first step consisted of preparing a computerbased matrix with all of the formulas required for calculations of oxygen consumption according to the 5 methods. [2],[3],[4],[5],[6] Except for the methods proposed by Bergstra et al. [5] and Wessel et al. , [2] where one single regression model (equation) is applicable to all patients, the tendency of other authors is towards using more than one equation. [3],[4],[6] Thus, our first task was to use 0 and 1 code columns to transform 2 or 3 equations (as is the case in the methods proposed by LaFarge and Miettinen, [3] Lindahl [4] and Lundell et al. [6] ) into a single one. For example, while estimating oxygen consumption in a 2year old child by the method proposed by Lundell et al. , [6] "codes 0" was used to eliminate the equations developed by the authors for males and females above the age of 3 years. Thus, the major work consisted of developing "singleequation models." The second step consisted of estimating oxygen consumption in 65 infants and children with ventricular septal defects, aged up to 8 years. For this purpose, demographic data were collected from unoperated patients without comorbidities or metabolic disturbances, in the absence of febrile or inflammatory states, and not using any medications that could potentially alter the heart rate. Patients were divided into 4 age groups, and oxygen consumption was estimated using the 5 predictive models for each age group. Statistical methods were applied to investigate withingroup and betweengroup differences. The third step consisted of extending the matrix to contain all the equations necessary for calculations of pulmonary and systemic arteriovenous differences in oxygen content, pulmonary and systemic blood flow and vascular resistance. Pulmonary and systemic flow and resistance were calculated for each estimated value of oxygen consumption. That is, the matrix was mounted so that the final results corresponded to the 5 estimates of pulmonary and systemic flow and flow ratios and pulmonary and systemic vascular resistance and resistance ratios. Statistics Results of analysis of oxygen consumption in patients with ventricular septal defects are expressed as mean and standard deviation. Betweenage group and withinage group differences were firstly analyzed by adjusting a general linear model to the obtained data. If differences were found to be significant, appropriate multicomparison tests were used to further investigate withingroup (between methods) disparities. A significance level of 0.05 was adopted. Results Representative examples of data input and output are shown in [Figure 1] and [Figure 2]. By entering data from 65 children with ventricular septal defects (demographics and heart rate), oxygen consumption values predicted on the basis of 5 different formulas became immediately available. Mean values of predictions by the different methods (formulas) are shown in [Table 1] for specific age groups (08 years). For each age group and every individual patient, it was possible to obtain several estimates of oxygen consumption, one of them (the lowest one) corresponding to the most pessimistic prediction of pulmonary vascular resistance (highest value) using the Fick principle. Thus, using a preestablished matrix containing the formulas, the betweengroups and betweenmethods variability of predicted oxygen consumption could be investigated for a large number of patients, with the most impressive variability between methods being observed in the first age group [Table 1]. Next, we used the automatically calculated pulmonary and systemic arteriovenous differences in oxygen content, the mean pressures (pulmonary artery, left atrium, aorta and right atrium) and the 5 estimates of oxygen consumption to generate for the individual patient, 5 different predictions for pulmonary and systemic blood flow, pulmonarytosystemic flow ratio, pulmonary and systemic vascular resistance and pulmonarytosystemic resistance ratio. [Table 2] shows 2 of the 5 instantaneously generated estimates of hemodynamic parameters in a representative patient, with data (demographics, heart rate, hemoglobin concentration, results of gas analyses and pressures within the cardiac chambers) collected under 2 conditions: Baseline and during nitric oxide inhalation. Discussion In this report, a computerbased matrix was developed for rapid calculations of hemodynamic parameters in patients with congenital heart disease subjected to cardiac catheterization. The main goal was to quickly obtain multiple estimates of several hemodynamic parameters without errors due to calculations. The majority of formulas included in the matrix corresponded to regression models developed for prediction of oxygen consumption. Predicted values of oxygen uptake have been largely considered as unreliable. [7] Actually, in some instances, these values have been shown to be higher than the measured ones, [8] indicating that there was underestimation of pulmonary vascular resistance using the Fick principle. On the other hand, it is notorious that cyanotic patients have higher levels of oxygen uptake than expected on the basis of predictions. Based on these discrepancies, some specialists would prefer to rely on the pulmonarytosystemic vascular resistance ratio (which does not require oxygen consumption to be calculated) for decision making. Less conservative specialists argue that mild variations in oxygen consumption are unlikely to have a significant impact on final calculations, and the need for an exact value of pulmonary vascular resistance may be overrated if all other clinical and noninvasive data are available. By using a rapidcalculation method, we were able to obtain replicate estimates of the same parameter (e.g., oxygen consumption, pulmonary and systemic flow and vascular resistance) and establish the upper and lower limits of a likely range for a given patient. For example, for the case depicted in [Table 2], the upper and lower estimates of pulmonary vascular resistance during inhalation of nitric oxide were 1.14 and 1.71 Wood units, respectively. That is, if that patient had univentricular physiology and was being considered for completion of cavopulmonary anastomoses (socalled Fontan operation, where precise preoperative measurements are needed), we would be relatively comfortable in terms of defining operability, since for him, the likely range was below 2.0 Wood units, which is considered as adequate for this type of operation. Despite the facilities available while managing the prepared matrix, we remain concerned about the unrestricted use of predicted oxygen consumption. [1],[7],[8] This practice should be avoided in at least 3 situations: 1 cyanotic patients; 2 patients with comorbidities, metabolic disturbances or under the use of any drugs that might potentially affect the heart rate; 3scenarios where very precise hemodynamic estimates are needed and the obtained ranges override acceptable limits. In conclusion, in the vast majority of patients with congenital cardiac shunts associated with pulmonary hypertension, assignment to correction (surgical or catheter interventional) is based on a broad spectrum of information obtained from noninvasive evaluation. In a smaller group of patients, however, a clinical history of pulmonary congestion and failure to thrive is not present, and sometimes mild peripheral oxygen desaturation is associated with bidirectional shunting across the defect. Measurement of pulmonary vascular resistance is mandatory in these instances. If for any reasons pulmonary vascular resistance cannot be calculated on the basis of measured oxygen consumption, it is probably better to have estimates at the upper and lower limits of a likely range for that specific patient. The matrix proposed in this report has the advantage of allowing rapid calculations of multiple hemodynamic parameters using different values of predicted oxygen consumption, provided that the limitations of using predicted consumption are taken into consideration. Further studies are obviously necessary to evaluate the impact on outcomes of having intervals for hemodynamic parameter estimation instead of single values. Availability For academic purposes, the prepared computerbased matrix can be obtained free of charge at [email protected] . References


