Biomechanical Modeling to Inform Pulmonary Valve Replacement in Tetralogy of Fallot Patients After Complete Repair

Abstract

Background:

A biomechanical model of the heart can be used to incorporate multiple data sources (electrocardiography, imaging, invasive hemodynamics). The purpose of this study was to use this approach in a cohort of patients with tetralogy of Fallot after complete repair (rTOF) to assess comparative influences of residual right ventricular outflow tract obstruction (RVOTO) and pulmonary regurgitation on ventricular health.

Methods:

Twenty patients with rTOF who underwent percutaneous pulmonary valve replacement (PVR) and cardiovascular magnetic resonance imaging were included in this retrospective study.

Results:

RV contractility before PVR (mean 66 ± kPa, mean ± standard deviation) was increased in patients with rTOF compared with normal RV (38–48 kPa) (P < 0.05). The contractility decreased significantly in all patients after PVR (P < 0.05). Patients with predominantly RVOTO demonstrated greater reduction in contractility (median decrease 35%) after PVR than those with predominant pulmonary regurgitation (median decrease 11%). The model simulated post-PVR decreased EDV for the majority and suggested an increase of Q_eff—both in line with published data.

Conclusions:

This study used a biomechanical model to synthesize multiple clinical inputs and give an insight into RV health. Individualized modeling allows us to predict the RV response to PVR. Initial data suggest that residual RVOTO imposes greater ventricular work than isolated pulmonary regurgitation. Biomechanical models specific to individual patient and physiology (before and after PVR) were created and used to estimate the RV myocardial contractility. The ability of models to capture post-PVR changes of right ventricular (RV) end-diastolic volume (EDV) and effective flow in the pulmonary artery (Qeff) was also compared with expected values.

Stenosis of the pulmonary valve is a relatively common congenital condition isolated, or as a part of the constellation of tetralogy of Fallot (TOF). The initial pressure overload is corrected surgically often using a transannular patch, causing varying degrees of pulmonary regurgitation (PR). Over time, valve-sparing techniques have been preferred, but at the cost of potentially higher rates of residual right ventricular outflow tract obstruction (RVOTO). Chronic RV volume overload leads to RV dilation and chronic RVOTO can cause RV hypertrophy. To avoid RV failure and even to allow reverse remodeling, patients undergo pulmonary valve replacement (PVR) in adolescence or early adult life. There is a robust debate regarding the optimal timing of PVR.^1–7 Among the main clinical indicators considered in decision making are the degree of RV dilation, RV ejection fraction (EF), and pulmonary valve regurgitation fraction (RF).⁸ However, the ability to predict post-PVR reverse remodeling remains elusive and is witnessed in only 60% of patients regardless of the threshold for intervention.² Upstream, the debate as to the optimal initial repair remains whether transannular patch with mild residual obstruction is preferable to annulus-sparing approaches in borderline valves. Current practice is to favor valve preservation, but recent long-term outcome data suggest that RV hypertrophy may be a predictor of adverse events.⁹

Although the current guidelines focus on the direct measures taken from imaging data, the underlying physiology is not the primary driver of interventions. Incorporating physical and physiologic assumptions of cardiovascular function in the framework of biomechanical modeling^10,11 has the potential to augment the interpretation of clinical data, such as by estimating some clinically relevant functional quantities, eg, myocardial contractility.^12–14 The contractility in the present work represents the active stress generated by the myocardial sarcomere unit during contraction. The contractility in a given patient can be increased chronically (eg, in ventricular overloading such as due to a valvular defect) and adapts to the actual physiologic state by inotropic stimulation.¹⁴ We remark that the level of myocardial contractility—as used in this work—was demonstrated to correlate with the maximum time derivative of ventricular pressure, max (dp/dt),¹⁴ which is a widely accepted surrogate measure of global ventricular contractility. Assessing mechanical properties of the hearts of patients with tetralogy of Fallot after complete repair (rTOF) provides additional metrics which may provide new insight into the initial procedure choice in cases of borderline pulmonary valve annulus.

Objectives

The goal of this retrospective observational work was to quantify the level of RV myocardial contractility before PVR and immediately after PVR by coupling the available clinical data with a biomechanical model. We hypothesize that the RV contractility is chronically increased in rTOF patients with PR or residual RVOTO and that it will decrease immediately after PVR. This reduction toward the normal values will also translate into the reduction of stroke work that the heart needs to exert.

Methods

Data

A group of 20 patients with rTOF who underwent percutaneous PVR were included in this retrospective study. The data collections were performed under the ethical approvals of Institutional Review Boards of the UT Southwestern Medical Center, Dallas (STU-2020-0023), and the University of Texas, Austin (IRB 2020-06-0128). The IRBs waived the need for a consent to use the anonymized retrospective data.

All patients underwent cardiovascular magnetic resonance (CMR) examination 4–6 months before PVR. CMR was processed semi-automatically by using the CVI42 software (Circle Cardiovascular Imaging, Calgary, Alberta, Canada) and a finite element method for image registration.^15,16 The results of CMR analyses are RV time-vs-volume plots and time-vs-flow through the pulmonary artery (PA) together with the flow integrated throughout the cycle containing the forward and regurgitant components (Q_for and Q_back, respectively). The effective flow through the pulmonary circulation (ie, Q_for – Q_back) will be denoted as Q_eff.

During the percutaneous PVR, the right-heart pressures (containing right atrial, RV, and PA pressures) were taken before and after deploying the valve. Figure 1 displays an example of processed clinical data of a selected patient.

Figure 1.

Example of processed clinical data (A) before and (B) after pulmonary valve replacement.

Supplemental Tables S1 and S2 provide information about the basic demographics, baseline anatomy, possible palliation, type and age of TOF repair, reintervention (if any), and the type of prosthesis in each patient.

Data of healthy subjects were obtained from a population study¹⁷ that contained the values (weighted means) of RV and PA end-systolic pressures (ESPs), RV end-diastolic volume (EDV) and end-systolic volume (ESV), and RV myocardial mass.

Biomechanical model

The biomechanical heart model of reduced order^18,19 was used. While the geometry and kinematics of RV are reduced, the constitutive mechanical laws are preserved as in the full 3D model²⁰ (Fig. 2). Specifically, myocardial tissue was modeled by a viscoelastic material with active contractile component representing the actin-myosin interaction (consistent with the sliding filament theory by A.F. Huxley^21,22). The myocardial internal stresses—passive (given by the tissue stiffness of the viscoelastic material) and active (generating the active shortening of the myocardial fibers leading to heart contraction)—are in equilibrium with the external loading (the pressure exerted on the endocardial surface) and inertia forces. The model of the RV was connected to a Windkessel model of the circulatory system,²³ which consisted of proximal resistance and capacitance R_prox, C_prox (representing the main and branch pulmonary arteries) with pressure P_ar, distal resistance and capacitance R_dist, C_dist (representing the remaining pulmonary circulation) with pressure P_dist, and terminal venous pressure P_vs, as depicted in Figure 2. The detailed description of the model is provided in Supplemental Appendix S1, and rigorous formulations can be found in our previous work.^18,20,24

Figure 2.

Model of right ventricle coupled with atrioventricular and arterial outflow valves and with circulation system represented by a Windkessel model. P_V, P_at, P_ar, P_dist, and P_vs stand for pressures in ventricle, atrium, large arteries, distal circulation, and venous system, respectively, $R_{for}^{TV}$, $R_{back}^{TV}$, $R_{for}^{RVOT}$, and $R_{back}^{RVOT}$ are forward and backward resistances of the tricuspid valve and forward and backward resistances of the ventricular outflow tract, respectively, and R_prox, R_dist, C_prox, and C_dist are proximal and distal resistances and capacitances of the circulation.

Model calibration to data of individual patients

The biomechanical model of RV and pulmonary circulation was adjusted manually to the data of each individual patient by a sequential calibration.¹⁹ The pulmonary Windkessel model was adjusted while imposing the measured PA flow. Using cine magnetic resonance imaging, we prescribed the wall thickness and the ventricular volume in reference configuration (when zero intraventricular pressure is assumed).²⁵ The RV preload was prescribed according to the measured RV pressure in diastole. Passive myocardial properties (ie, myocardial stiffness) were calibrated so that the simulated EDV matched the measurement. RVOT was modeled as an outlet valve allowing the forward and regurgitant flow²⁴ (with resistances $R_{for}^{RVOT}$ and $R_{back}^{RVOT}$, respectively). The backward resistance $R_{back}^{RVOT}$ was adjusted to match the backward flow waveform. $R_{back}^{RVOT}$ was set to a high value if no PR was present (effectively zero backward conductance). Similarly, a negligible $R_{for}^{RVOT}$ was used if no pathologic RVOTO was present. The tricuspid valve (TV) was modeled as an inlet valve with resistances $R_{for}^{TV}$ and $R_{back}^{TV}$; respectively.²⁶ The myocardial contractility was adjusted according to the measured RV stroke volume (SV) and RV ESP. Physiologic assumptions of mechanochemical coupling of the actin-myosin complex in the sarcomere are translated into the mechanical system generating force—the active contractility—which combines with the passive viscoelastic properties of the tissue.

The post-PVR model was obtained by recalibrating the pre-PVR model with the aim to match the post-PVR pressure measurements. We preserved the passive myocardial properties, prescribed the preload as in the measurement, $R_{back}^{RVOT}$ was set to its maximum (to eliminate PR) and $R_{for}^{RVOT}$ adjusted to match the pressure difference between RV and PA. Finally, the contractility was adjusted so that the simulated RV ESP matched the data.

The mechanical parameters of these patient-specific models give an insight into the cardiovascular physiology of each patient. The simulated RV EDV and pulmonary flow after PVR were used to validate the model prediction against published data by Lurz et al.²⁷ Further details of the model calibration, including quantitative values of parameters for each patient (Supplemental Table S3), are presented in the Supplemental Appendix S1.

Statistical analysis

Wilcoxon signed-rank tests were conducted for the changes in model-derived contractility and PA ESP at P < 0.05. Bland-Altman plots were constructed to evaluate the difference between simulated and measured functional indicators.

Results

Direct analysis of clinical data

The results of direct analysis of clinical data are summarized in Supplemental Table S1. Pre-PVR CMR revealed that 45% of patients had mild PR, with RF of 10%–30%, 15% of patients had moderate PR (RF 30%–40%), and 40% had severe PR (RF > 40%). Considering the RV size, 25% of patients had moderate RV dilation (EDV indexed to the body surface area, RV EDVi, 120–140 ml/m² ) and 20% had moderate to severe RV dilation (RV EDVi 140–150 ml/m² ), while the remaining patients had normal or mildly dilated RV. Finally, 30% of patients had moderate pulmonary stenosis, with the RV to PA pressure gradient of ≥ 25 mm Hg. The degree of RVOTO was evaluated based on the ratio of RV to LV ESP, where ratios ≥ 50% and < 50% were considered to represent high and low, respectively, degrees of RVOTO. The comparison of RV SV and outflow volume Q_for revealed mild TR in 4 patients.

The patients were divided into 3 groups according to the level of RVOTO and PR: low degree of RVOTO and at least moderate PR (group A); high degree of RVOTO and mild PR (group B); and high degree of RVOTO and at least moderate PR (group C). Groups A-C contained 9, 8, and 3 patients, respectively.

Model-derived ventricular contractility

We successfully calibrated the model for all patients before and after PVR. Figure 3 and Supplemental Figures S1–S3 show the simulated cardiac cycles compared with data in selected patients. Bland-Altman plots in Figure 4 and quantitative summary in Table 1 show the mean bias between the simulations and measurements before and after PVR.

Figure 3.

Measured data (dashed lines) and simulation (solid lines) for patient 16. PA, pulmonary artery; PVR, pulmonary valve repair; RV, right ventricle.

Figure 4.

Bland-Altman plots for RV EDVi, ESVi, PA Q_for, Q_back, RV ESP, PA ESP (all pre-PVR), and for RV ESP and PA ESP (post-PVR). Solid horizontal lines represent the mean of the difference between data and simulation, top and bottom dashed horizontal lines show the limits of agreement at 95% prediction interval (± 1.96 times SD). EDVi, end-diastolic volume indexed to body surface area; ESP, end-systolic pressure; ESVi, end-systolic volume indexed to body surface area; PA, pulmonary artery; PVR, pulmonary valve replacement; Q_back, backward flow; Q_for, forward flow; RV, right ventricular.

Table 1.

Bland-Altman quantitative statistics summarizing the mean bias ± standard deviation and limits of agreement (95% confidence intervals) between simulations and measurements

Pre-PVR
RV EDVi, ml/m2	0.01 ± 0.04	−0.08 to 0.01
RV ESVi, ml/m2	−0.77 ± 1.41	−3.53 to 2.00
RV ESP, mm Hg	0.06 ± 1.10	−2.11 to 2.23
PA ESP, mm Hg	−0.08 ± 0.95	−1.93 to 1.78
PA Qfor, ml/cardiac cycle	3.02 ± 5.40	−7.60 to 13.70
PA Qback, ml/cardiac cycle	0.01 ± 5.00	10.20 to 10.20
Post-PVR
RV ESP, mm Hg	−0.05 ± 0.60	1.23 to 1.13
PA ESP	0.41 ± 0.67	−0.90 to 1.72

EDVi, end-diastolic volume indexed to body surface area; ESP, end-systolic pressure; ESVi, end-systolic volume indexed to body surface area; PA, pulmonary artery; PVR, pulmonary valve replacement; Q_back, backward flow; Q_for, forward flow; RV, right ventricle.

Calibrating the model by using the data from a healthy population revealed the contractility of healthy RV to be in the range of 38–48 kPa. The values of RV contractility in all patients (assessed by models calibrated to pre- and post-PVR data) are plotted against RV ESP in Figure 5A. Pre- and post-PVR median contractilities were 66 and 51 kPa, respectively. Figure 5B shows a strong positive correlation (R² = 0.95; P < 0.001) between the RV ESP and contractility rescaled by the ratio of myocardial wall thickness over ventricular radius.

Figure 5.

Model-derived right ventricular (RV) (A) contractility and (C) stroke work for each patient. Filled and empty circles correspond to pre- and post-PVR values, respectively. The number labels of the circles correspond to individual patients. The values for healthy RVs are represented by the area encompassed within the dashed lines. (B) RV contractility rescaled by the ratio of myocardial wall thickness over ventricular chamber radius against RV end-systolic pressure, where gray line is Pearson correlation with R² = 0.95 and P < 0.05. PR, pulmonary regurgitation; PVR, pulmonary valve replacement; RVOTO, right ventricular outflow tract obstruction.

Stroke work was calculated as the area encompassed within the simulated ventricular pressure-volume (P-V) loops and is plotted against RV ESP in Figure 5C. Pre- and post-PVR median stroke works were 327 and 233 mJ, respectively. The stroke works obtained from the healthy subjects were in the range of 103–150 mJ.

The values of mean contractility and stroke work for each group, relative to the maximum values of the normal population, are presented in Table 2.

Table 2.

Median right ventricular (RV) contractility and relative contractility compared with normal RV contractility (48 kPa), and median RV stroke work and relative stroke work compared with normal RV stroke work (150 mJ)

	Contractility, kPa		Contractility vs normal		Stroke work, mJ		Stroke work vs normal
	Pre-PVR	Post-PVR	Pre-PVR	Post-PVR	Pre-PVR	Post-PVR	Pre-PVR	Post-PVR
Group A	50	46 (P = 0.008)	+4%	0%	200	166 (P = 0.027)	+34%	+ 10%
Group B	71	52 (P = 0.016)	+48%	+8%	444	266 (P = 0.016)	+ 196%	+77%
Group C	75	53 (P = 0.250)	+56%	+10%	706	448 (P = 0.250)	+370%	+199%
All patients	66	51 (P = 0.000)	+38%	+5%	327	233 (P = 0.000)	+118%	+55%

Group A: no right ventricular outflow tract obstruction (RVOTO) and high pulmonary regurgitation (PR), group B: high RVOTO and no PR, group C: high RVOTO and high PR.

Significance is assumed at P < 0.05.

Post-PVR changes in RV EDV and effective PA flow

The patient-specific post-PVR models were used to assess the changes of RV volumes and PA flow after PVR based on captured post-PVR invasive pressures. Figure 6 shows the model-derived changes of RV EDVi and PA Q_eff. The average EDP decrease in our cohort was lower than that in the study by Lurz et al.²⁷: 14.2% vs 26.5%. Consequently, the average EDV decrease was also lower: 3% decrease suggested by the model vs 12% reported by Lurz et al. The model suggested an increase of PA Q_eff after PVR, which fell between the pre-PVR values of Q_for and Q_eff—in line with Lurz et al.

Figure 6.

(A) Model-predicted postepulmonary valve replacement (PVR) changes in right ventricular end-diastolic volumes indexed to body surface area (EDVi). Central line inside each box indicates the median, and the bottom and top edges of the boxes show 25th and 75th percentiles, respectively. Black circles connected by black lines show the EDVi change from before to after PVR for individual patients. (B) Post-PVR model-predicted change in effective flow (Qeff) in each patient group. Orange bars: Qeff_pre-PVR/Qfor_pre-PVR corresponding to the complement of regurgitation fraction (1 – Qback/Qfor), where Qfor and Qback are forward and backward flows, respectively. Blue bars: Qeff_post-PVR (model) is scaled by Qfor_pre-PVR (data) (consistently with the orange bars) to demonstrate model-predicted significant increase of Qeff_post-PVR. Note that model-derived Qeff/Qfor = 1 for post-PVR (owing to the assumption of zero pulmonary regurgitation post-PVR). Asterisks indicate significant difference at P < 0.05.

Discussion

This study applied patient-specific biomechanical modeling on a group of patients with rTOF indicated for PVR. We aimed to assess whether the proposed data-model coupling framework provides any additional clinical indicators of the effects of PVR.

CMR- and pressure-derived clinical indicators suggested 3 groups of patients based on the grade of PR and severity of RVOTO. The patient-specific models created separately for pre- and post-PVR physiology revealed that PVR triggered an immediate adaptation of RV contractility for a majority of patients, with a decrease of the median contractility by 23%. Table 2 reveals that the patients with a high degree of RVOTO showed a larger contractility decrease (groups B and C) and the contractility of all patients decreased close to the normal values (Table 2 and Figure 5).

The stroke work represents the mechanical energy generated by the ventricle during a heartbeat. The analysis of P-V loops suggested that even though the patients with a high degree of RVOTO (groups B and C) experienced a significant decrease of the median stroke work (Table 2), the actual stroke work after PVR remained elevated owing to a limited decrease of RV pressure after PVR. The patients with predominantly PR and a low degree of RVOTO (group A) showed a decrease of the median stroke work by 17%. The resulting stroke work was only ~ 10% above normal values (Table 2), thanks to the normal RV pressure after PVR.

The strong correlation between the ventricular pressure and the rescaled contractility is in line with Laplace’s law of myocardial wall stress being directly proportional to the level of pressure developed in the chamber for a given geometry during ventricular systole. Thus, before PVR we observed higher levels of contractility in patients with increased RV systolic pressure and in those with dilated RVs (patients 12 and 14 from group B). The RV contractility decreased after PVR primarily owing to the release of RVOTO. The sole effect of regurgitation on the system was visible in the patients from group A. The median contractility and stroke work in that group appeared to be only up to 4% and 34% higher, respectively, than a range of reference healthy RV contractility and stroke work (Table 2). This suggests that, in infants with TOF with borderline pulmonary valve annulus, inserting a transannular patch might be preferential, because the RV is likely to well tolerate the created PR. However, 2 patients with the most dilated RVs (RV EDVi > 140 mL/m²) appeared to be the outliers, with pre-PVR contractilities of 77 and 73 kPa (patients 12 and 14, respectively). It is likely that before PVR, the majority of patients in group A had been preserving their cardiac output at moderately elevated myocardial stresses. Those with dilated RVs had been progressively becoming less efficient and had elevated metabolic demands. Therefore, our model revealed that the myocardial contractility was substantially increased in the patients with dilated RVs before PVR. However, the critical range of RV dilation is unclear from this study and is the subject of our ongoing work. Furthermore, we showed that the greatest reduction of contractility occurred because of removal of RVOTO, while removing the regurgitation itself did not lead to a significant immediate decrease of contractility in the majority of patients.

Not having post-PVR CMR prevented us from performing a detailed validation of our post-PVR models. However, a partial validation was possible thanks to the study by Lurz et al.,²⁷ in which CMR was performed immediately after the intervention on the pulmonary valve. Our models showed a decrease of RV EDV for patients experiencing a decrease in RV EDP and an increase of PA Qeff after PVR, both of which qualitatively match with the study of Lurz et al. The lower decrease of EDP (as measured in our patients compared with those of Lurz et al.) explains the lower EDV decrease in our group of patients.

Limitations

A number of limitations should be addressed. First, an assumption of spherical RV might have overestimated the contractility because the spherical shape is mechanically more efficient than a crescent shape. Figure 7 demonstrates the effect of interobserver variability in estimating RV myocardial mass: The difference between pre- vs post-PVR contractility were preserved in a relative sense, and this would be expected also if an accurate RV geometry was used. It is expected that the active stress developed in various parts of the RV varies (eg, inflow part may be very contractile, while the outflow will be hypo- or dyskinetic due to the passive transannular patch). Such a heterogeneity cannot be captured by our simplified model. However, our pilot study demonstrates that using such a modeling approach is directly feasible in the clinical setup. The level of complexity allows medical doctors to set up the patient-specific models. This may facilitate launching a large-scale and multisite clinical study counting hundreds of cases, which would be out of reach for a number of complex models.

Figure 7.

Sensitivity of the model-derived contractility to the variation of right ventricular (RV) mass. Pree— and post—pulmonary valve replacement (PVR) contractilities (filled and empty points, respectively) were calibrated for the following type of mass input for each patient: RV mass represented by RV free wall (RVFW; black points), RVFW mass increased by the mass of half septum mass (blue points), and RVFW with full septum (red points). Solid lines between the points show the decrease of the model-derived contractility after PVR.

Percutaneous PVR is mostly not used in patients with very dilated RVs and RVOTs, who rather undergo surgical PVR. In this pilot study we took advantage of the accessibility of measured pressures after PVR. However, we acknowledge that the population in this study was biased and did not allow for general conclusions regarding relative impact of PR and pulmonary stenosis over the whole range of this patient population. In a future study, we aim to use an estimation of RVOT pressure gradient based on the measured flow profile through the RVOT²⁸ and the estimated RV end-diastolic pressure according to the right atrial volume and flow through the tricuspid valve.²⁹

Another important concern is the time period of 4–6 months between the acquisition of CMR and catheterization data in this proof-of-concept work. The progressive RV remodeling might have caused an underestimation of the current RV volumes, which may in turn affect the predicted model-derived properties. Finally, the validation of our post-PVR model was limited by the absence of post-PVR CMR. In the future we aim to perform an additional CMR within 48 hours after PVR, which would be plausible for most patients. RV electromechanical dyssynchrony is another major pathophysiologic factor that can lead to further pathologic RV remodeling. Cardiac models with a detailed electrical component provide perspective in optimizing RV cardiac resynchronization therapy,³⁰ and modeling could even contribute to the decision making of a possible combination of PVR and RV cardiac resynchronization therapy.

Future perspectives

In addition to RV contractility at rest, a reduced exercise capacity may play a significant role in the response to PVR and will be studied.¹⁴ The present work demonstrates an application of models capturing an immediate state of the cardiovascular system to inform about the current physiologic state without explicitly considering the previous progress of ventricular remodeling. Models of long-term evolution^31,32 have the potential to include information about the initial state of pathology, type of repair, and the evolution throughout the life of patient (Supplemental Table S2). They could also be useful in better understanding the reverse remodeling after PVR—a crucial step in predicting the long-term effect of PVR and subject of our ongoing research.^33,34 Advanced imaging techniques such as the assessment of myocardial fiber directions by magnetic resonance diffusion tensor imaging³⁵ could be considered in the future.

Even though the present study demonstrated a substantially increased work that the RV must exert in patients with residual RVOTO and therefore favors transannular patch in the case of borderline annulus size, we are aware that our proof-of-concept modeling study cannot have immediate implications on patient care. Furthermore, it is known that a mild-to-moderate stenosis in the early postoperative period may spontaneously regress and the strategy of preservation of annular function may be compatible with long-term relief of RVOTO.

Conclusion

Combination of computational models and clinical data is feasible in a cohort of patients with rTOF. The present study quantifies the level of RV overload with chronic valvular disease and how the level of overloading decreases after intervention on the valve. Though not directly predicting who should undergo PVR and who should wait, we think our physiologic finding is of clinical interest as a step toward optimal clinical evaluation and management of patients with valvular heart disease. Furthermore, the need to aim for long-term relief of RVOTO is highlighted, perhaps prejudicing the initial surgical approach. Predictions based on coupling clinical data and biomechanical models have the potential to become part of clinical assessment to contribute to optimizing and personalizing the clinical management of every patient.