Discussion

These analyses demonstrate that use of alpha-blockade (either phenoxybenzamine or phentolamine) during paediatric cardiac surgery admissions is independently associated with decreased ICU length of stay, total length of stay, and decreased mortality after adjusting for covariates. This is without a concomitant significant increase in billed charges. Previous reports of the use of alpha-blockade in paediatric cardiac surgery admissions have shown favourable effects. ^{Reference Hoffman, Tweddell and Ghanayem5–Reference De Oliveira, Ashburn and Khalid10}

The univariate analyses demonstrate a pattern consistent with alpha-blockade being more frequently used in cardiac lesions palliated or repaired in the neonatal period such as double outlet right ventricle repair (Risk-adjustment Congenital Heart Surgery-1 category 3), hypoplastic left heart syndrome (Norwood procedure, Risk-adjustment Congenital Heart Surgery-1 category 6), transposition (arterial switch, Risk-adjustment Congenital Heart Surgery-1 category 4), common arterial trunk (truncus arteriosus repair, Risk-adjustment Congenital Heart Surgery-1 category 4), and total anomalous pulmonary venous connection repair (Risk-adjustment Congenital Heart Surgery-1 category 4). These findings are not particularly surprising as these neonatal palliations and repairs continue to represent the congenital heart surgery interventions with greatest morbidity and mortality according to the Risk-adjustment Congenital Heart Surgery-1 classification. ^{Reference Ithuralde and Neirotti11} A study by Ithurrlhade et al. Risk-adjustment Congenital Heart Surgery-1 in the neonatal population showed no statistical significance of mortality in between different paediatric age groups (neonates, infants, and children over 1 year). Furthermore, they also proved that Risk-adjustment Congenital Heart Surgery-1 in the neonatal population was a significantly more powerful mortality predictor than in any other paediatric age group. ^{Reference Ithuralde, Ballestrini and Ithuralde12}

A report of alpha-blockade after cardiopulmonary bypass can be found quite early in the experience with cardiopulmonary bypass. Sakauchi and colleagues hypothesised that the effect of cardiopulmonary bypass may be similar to haemorrhagic shock and thus felt that phenoxybenzamine may be beneficial. They demonstrated increased blood flow to the heart and brain with decreased blood flow to the kidneys and also found that beta-glucuronidase activity increased after cardiopulmonary bypass. Administration of phenoxybenzamine reversed these findings and was found to be helpful in promoting systemic blood follow and oxygen delivery. ^{Reference Sakauchi, Anzai, Oki, Iino and Matsumoto13}

Since this early report, much has been learned about the utility of peri-operative and post-operative alpha-blockade. Peri-operative use of alpha-blockade allows for higher cardiopulmonary bypass pump flows which has been associated with decreased peri-operative metabolic acidosis. ^{Reference Mossad, Motta, Sehmbey and Toscana8,Reference Bisoi, Sharma and Chauhan9,Reference Motta, Mossad, Toscana, Zestos and Mee14} As previously shown, alpha-blockade may have a pulmonary vasodilating effect that can be of benefit in certain patient populations ^{Reference Kiran, Zuber and Kakani6,Reference Kiran, Makhija, Das, Bhan and Airan7,Reference Ogawa, Yamamoto and Asada15–Reference Ruskin and Hutter18} . Alpha-blockade has also been demonstrated to improve coronary flow. ^{Reference Jones, Liang and Maulsby19}

There are some possible side effects associated with the use of alpha-blockade, the most common being hypotension. Phenoxybenzamine-induced hypotension can be ameliorated by fluid administration and vasopressin. ^{Reference O'Blenes, Roy, Konstantinov, Bohn and Van Arsdell20} Phentolamine-induced hypotension can be ameliorated by altering the dose of or discontinuing phentolamine. Fluid administration can also help ameliorate phentolamine-induced hypotension.

Use of and availability of phenoxybenzamine has decreased significantly over the years as the price of the agent has become prohibitive at many centres. The current analyses cover an era in which the use of phenoxybenzamine was nearly absent. The early study period did find some recorded use of phenoxybenzamine although even this was minimal. Phentolamine represents the major agent used for alpha-blockade in the current era.

These analyses represent, to our knowledge, the largest series of congenital heart surgery patients for whom data regarding alpha-blockade have been reported. The data identify patterns in the use of alpha-blockade and also demonstrate independent associations with admission characteristics.

These analyses, however, are not without their limitations. Firstly, the data come from a retrospective administrative database and so incorrect coding, variable coding practices, and non-coding may impact the frequencies and associations (although to a lesser extent) in these analyses. Furthermore, the ability to categorise diagnoses and/or surgeries with this data source is limited. It must also be kept in mind that the patient populations identified in the univariate analyses do not represent patient populations in whom alpha-blockade is more beneficial but simply describes the bias of providers as to what patient populations receive alpha-blockade. Subset analyses could not be reliably conducted to identify specific patient populations that may benefit more than others from alpha-blockade. Additionally, the timing of alpha-blockade cannot be discerned. And thus, whether timing of alpha-blockade impacts the effect cannot be commented upon.

An additional limitation is also the general shortcomings of regression analyses. There is no one ideal approach to regression analyses. We chose to use a stepwise approach since our variables of interest (which included cardiac diagnoses, cardiac surgeries, and comorbidities) were already selected to be variables of particular interest. The a priori selection of variables in the setting of stepwise regressions does seem to help improve model stability and performance. We elected to use a stepwise regression method versus a penalised regression method as stepwise regressions are not always outperformed by penalised regression methods as commonly believed. A study by Morozova colleagues, in fact, demonstrated greater model stability with stepwise regression utilising backward elimination with likelihood ratio testing than when compared to using penalised regression models ^{Reference Morozova, Levina, Uusküla and Heimer21} .

It should be noted that ICU length of stay, total length of stay, inpatient mortality, and billed charges were all significantly higher in the alpha-blockade admissions by univariate analyses. Regression analyses demonstrated that other factors were actually behind these increases and that alpha-blockade was actually independently associated with decreased ICU stay, total length of stay, and mortality. There was no independent association with any significant change in billed charges.