The salutary effect of peritoneal dialysis catheters on enhanced recovery among high-risk pediatric patients undergoing the left coronary transfer procedure: a cohort study

Indications for prophylactic PDC insertion in children with ALCAPA were primarily as follows: (1) oliguria or anuria, (2) edema caused by fluid accumulation, (3) electrolyte disturbances (acidosis or hyperkalemia), and (4) high-risk characteristics, including extremely young age (< 6 months), low weight not corresponding to age and severely deteriorating cardiac dysfunction [left ventricular ejection fraction (LVEF) < 20%]. However, whether to prophylactically place a PDC in the operating room was determined by the surgical team, which included cardiac surgeons and anesthesiologists, based primarily on the patient’s risk profile, with the aim of preventing adverse events. After patients were transferred to the ICU, intensivists at the bedside also participated in decision-making regarding PDC placement. The process of catheter placement was carried out by cardiac surgeons. Next, the catheters were either opened to drain or, alternatively, used for dialysis in the ICU. Therapeutic dialysis was initiated for patients who experienced the following events: urine output less than 1 mL/kg/h for 4–8 h, oliguria with poor response to loop diuretics, acidosis or hyperkalemia. Dialysis was performed intermittently at our institute. The indication for PDC removal was confirmed relief of the patient’s critical illness, including urine output normalization, hyperkalemia or acidosis disappearance, and the absence of fluid overload or edema.

The surgical strategy to correct ALCAPA is reconstruction of the double coronary system. The alternative Takeuchi procedure, usually called intrapulmonary repair, has been abandoned at our institution due to poor surgical outcomes.

It should be mentioned that during cardiopulmonary bypass (CPB), perfusionists carried out modified ultrafiltration for all cohorts. ECMO was initiated if patients had difficulty weaning off CPB, low cardiac output syndrome or severe hemodynamic instability.

Diuretics were administered as first-line therapy to treat suspected AKI. Then, the intensivists and surgeons decided whether to initiate dialysis if patients were unresponsive to high-dose diuretics and the subsequent presence of anuria or oliguria and edema still persisted.

Baseline serum creatinine was determined as the latest value before surgery; this value was not missing for any of the patients. The highest creatinine level on postoperative day (POD) 7 of admission to the ICU was obtained to identify AKI. The pediatric-modified risk, injury, failure, loss and end stage (pRIFLE) criteria, which have been demonstrated to be sensitive in pediatric cohorts, were used to diagnose AKI and categorize patients accordingly. Risk was defined as an estimated glomerular filtration rate (eGFR) that had decreased by 25–49% compared with its baseline value, injury as a decrease of 50–74%, and failure as a decrease ≥75% or an eGFR < 35 mL/min/1.73 m² [9]. Any degree of failure or loss was perceived as severe AKI. The Schwartz model formula was applied to determine the eGFR [10]. Fluid balance between the operative period and POD 3 was calculated, and negative fluid balance was set as a value less than 0%. For patients with PDC placement in the ICU, the timing of the placement was also obtained. Vasoactive-inotropic score (VIS) = dopamine (μg/kg/min) + dobutamine (μg/kg/min) + 100 × epinephrine (μg/kg/min) + 10 × milrinone (μg/kg/min) + 10,000 × vasopressin (U/kg/min) + 100 × norepinephrine (μg/kg/min) [11]. A prolonged duration of mechanical ventilation was defined as more than 100 h because this cutoff is the upper 75% limit of the interquartile range, and 25% of the cohort exceeded this value. Cardiac improvement was the change in LVEF during the interval between the preoperative visit (T1) and hospital discharge or the 1-year follow-up time point (T2). Thus, the change in LVEF was calculated by the following equation: 100 %  ∗ (LVEF_(T2) − LVEF_(T1))/LVEF_(T1). The degrees of mitral regurgitation were categorized as none/trivial, mild, moderate and severe. The amelioration of mitral regurgitation at the 1-year follow-up was defined as follows: (a) the classification was degraded for patients diagnosed preoperatively as having mild, moderate or severe regurgitation; and (b) there was no further upgrade or aggressive deterioration in patients with no/trivial regurgitation.

Given that patients with PDC placement recover efficiently in the ICU, they will have a decreased length of stay in the general wards before successful hospital discharge. The ratio of general ward/ICU length of stay was therefore adopted as a primary outcome in the present study.

Variables following a normal distribution are expressed as the mean ± standard deviation. However, a majority of the continuous data had a skewed distribution and were thus presented as the median (interquartile range). Frequencies and percentages were also used, as appropriate, for categorical data. For continuous variables, t-tests or Mann-Whitney U tests were used to assess differences between groups. Comparisons of categorical data were made by the chi-square test in combination with Yate’s continuity correction or Fisher’s exact test.

All statistical analyses were performed using SPSS 24.0 software (IBM, Armonk, NY, USA). Figures were generated with GraphPad Prism 7.0 in conjunction with Microsoft office word 2016.

An LVEF ≤25% enhanced the odds of PDC placement by nearly 10-fold [OR = 9.88, 95% CI 2.13 to 45.76; P = 0.003]. This model had good discrimination (AUC-ROC 0.918) and calibration (Hosmer-Lemeshow statistic P = 0.594). Additionally, concomitant mitral repair was associated with lower odds of PDC usage in the univariable analysis (OR = 0.25, 95% CI 0.07 to 0.85, P = 0.026) (Table 3).

Compared with patients without PDC, those with PDC placement had significantly extended lengths of stays in the ICU (20.0 days (8.5, 62.0), 2.0 days (1.0 to 5.0), P <  0.001) and general ward (10.0 days (7.0, 21.0), 6.0 days (5.0 to 7.0), P < 0.001).

The primary outcome, the ratio of ward/ICU length of stay, was significantly lower in patients who underwent PDC implantation [0.65 (0.25 to 1.0), 2.5 (1.0 to 6.0), P < 0.001] than in those who did not. An LVEF > 50% (B = 1.82, 95% CI 0.52 to 3.12; P = 0.007) and age > 12 months (B = 1.61, 95% CI 0.31 to 2.91; P = 0.016) were both significant predictors of an increased ratio of general ward/ICU length of stay. However, the univariable analysis showed that PDC placement was strongly associated with a decreased ratio of general ward/ICU length of stay (B = − 2.88, 95% CI − 4.46 to − 1.31; P < 0.001) (Table 4). The variance inflation factor of this multivariable linear regression was < 2 for all the included covariables; thus, among these variables, there was no possibility for multicollinearity.

Among the 82 patients surviving after surgery, the proportions of PDC placement in patients ≤12 months and those > 12 months were 32.5% (13/40) and 4.8% (2/42), respectively. Additionally, the incidences of PDC placement in those with LVEF ≤50% and LVEF > 50% were 37.5% (15/40) and 0% (0/42), respectively. The distributions of length of stay in the ICU and general ward categorized by age and LVEF are depicted in Fig. 2. Patients with LVEF > 50% had a significantly higher ratio of ward/ICU length of stay than those with LVEF ≤50% [3.5 (1.42 to 6.0), 0.86 (0.52 to 2.0), P < 0.001]. Additionally, compared with patients > 12 months, those ≤12 months had a lower ratio of ward/ICU length of stay [1.07 (0.58 to 2.0), 4.25 (1.25 to 6.0), P < 0.001].

Subgroup analyses were performed in surviving pediatric patients with a high-risk profile, namely, those with LVEF ≤50% (n = 40) or those younger than 12 months (n = 40) (Supplemental Tables 1 and 2). Remarkably, PDC placement was an independent predictor of a reduced ratio of ward/ICU length of stay in both of these subgroups (B = − 1.62; 95% CI − 2.77 to − 0.46; P = 0.008 for LVEF ≤50%; B = − 1.57; 95% CI − 2.88 to − 0.26; P = 0.02 for age ≤ 12 months).

The inherent retrospective characteristics of this study contributed to the predominant limitation. Another shortcoming was that no specific effect of PDC placement on short- and long-term mortality was estimated, as there were only 2 in-hospital deaths. Moreover, the complex association between AKI and PDC implantation cannot be further solved due to chronological uncertainty. After all, more than 60% of PDCs are placed in pediatric patients lacking acute kidney failure [4]. Finally, we did not acquire accurate serum creatinine levels or urine output counts at the time of PDC implantation. Oliguria was diagnosed based on our intensivists’ crude estimation while referring to the standard rule of urine output < 1 mL/kg/h for 4–8 h.