Integration of the Intrinsic Surgical Risk With Patient Comorbidities and Severity of Congenital Cardiac Disease Does Not Improve Risk Stratification in Children Undergoing Noncardiac Surgery

To the Editor We read with interest the article by Faraoni et al.1 In this retrospective analysis, the authors reported that among children with congenital heart disease (CHD) undergoing noncardiac surgery, accounting for the intrinsic surgical risk did not improve the predictive performance of a risk scoring system for 30-day mortality. This topic is both interesting and important, but it is our opinion that further clarification is required to explain these results. The author’s conclusion is predicated on the intrinsic surgical risk being coded as a binary variable (low and high). However, dichotomization of a continuous variable can result in substantial loss of information and statistical power.2 This may explain why the predictive performance of the model did not improve when accounting for intrinsic surgical risk. Had the authors chosen a different functional form for the intrinsic surgical risk, the predictive performance of the model may have been higher, and the conclusion different. We recommend that the author’s research question be explored in a way that avoids dichotomization of intrinsic surgical risk, perhaps consistent with the original study by Nasr et al3 where the intrinsic surgical risk was classified into least 3 categories. An additional (albeit counterintuitive) conclusion of the article was that children with CHD undergoing “high-risk” noncardiac surgeries were actually at a lower risk of 30-day mortality. In their analysis, the authors used relative value units (RVUs) associated with current procedural terminology (CPT) codes to produce the intrinsic surgical risk. Using RVUs as a surrogate for surgical complexity may be problematic given that RVUs are part of the architecture of the surgery billing infrastructure. Perhaps if the authors had used the same methodology in a prior publication from their group,3 the results of the present analysis might be different. In this earlier analysis, RVUs were not integrated into the risk calculation. It should also be mentioned that the authors’ definitions of severity of CHD are based on the American College of Surgery National Surgical Quality Improvement Program Pediatric (ACS NSQIP-P) definitions of severity. The severity of cardiac lesions is classified largely by anatomic and radiologic findings while some of the metrics commonly used in clinical practice are omitted. This definition will not account for clinically relevant factors such as weight gain, physical activity tolerance, and ventricular function. It also does not account for physical examination findings such as resting heart rate, respiratory rate, work of breathing, etc. Our concern is that patients with physiologically minor CHD may fall into the major or severe category and vice versa. Specifically, we are considering patients with disease trajectories that are worsening or improving around the time of noncardiac surgery. As with any hierarchical classification system, there is some gray area in the space between 2 levels. For example, a patient with any degree of pulmonary hypertension will be classified as severe CHD. This will include patients with mild or improving pulmonary hypertension that does not require medication, supplemental oxygen, and is not a barrier to growth and development. Additionally, patients with coarctation of the aorta commonly have a residual pressure gradient in the immediate aftermath of surgical correction. This could be problematic for the severity categorization because the patient has a corrected defect but has a residual hemodynamic abnormality that will improve with time. Finally, the severe category includes patients with “ventricular dysfunction requiring medications.” This is sufficiently ambiguous to allow for a wide variety of patients to fall into the severe category. Ionotropes, beta-blockers, diuretics, angiotensin converting enzyme (ACE) inhibitors, and digoxin are just a few of the drugs that may be used in patients with ventricular dysfunction, but a patient requiring an epinephrine infusion is different than one on maintenance digoxin. Similar arguments can be made for all of the disease states listed in the minor, major, and severe classification system. In conclusion, we applaud the authors’ endeavor to contribute to the literature on postoperative mortality in children with CHDs. However, we thought it important to emphasize some limitations of the definitions used and functional form of intrinsic surgical risk, which must be considered when interpreting the findings of their study. Jesse W. Hawkins, MDPeter D. Winch, MD, MBADepartment of Anesthesiology and Pain MedicineNationwide Children’s HospitalColumbus, Ohio[email protected]