A morbidity score for congenital heart surgery based on observed complications

Abstract

OBJECTIVES

The aim of this study was to develop a morbidity score based on observed postoperative complications after congenital heart surgery.

METHODS

Conditions or diseases that impair patients after congenital heart surgery or increase costs of hospital stay were called complications and attributed to scores ranging from 1 (mild) to 4 (severe) points, according to estimated severity or costliness. ‘No complication’ was assigned 0.5 points. From January to March 2011, scores for each observed ‘complication’ for every main (primary) surgical procedure were recorded and a morbidity score was calculated. In conformity with the Aristotle score methodology, if the sum of observed complication scores amounted to more than 5 points, a morbidity score of only 5 points was attributed. The estimated morbidity score was compared with the morbidity score attributed by the Aristotle basic complexity (ABC) score.

RESULTS

One hundred and thirty-nine primary procedures were carried out. The mean ABC and Aristotle comprehensive complexity scores reached 8.31 ± 2.52 and 9.62 ± 3.47 points, respectively. Two patients died. No complication was detected after 46 procedures. Overall, there were 185 listed ‘morbidity’ conditions in connection with the other 93 surgical procedures, rendering a total score of 385 points. The most frequent event was ‘mechanical ventilation 25–95 h’: n = 39. The mean morbidity score was 2.14 ± 1.63. The morbidity scores ranged from 0.5 points (n = 46) to 5 points (n = 23) with a median of 2.0 points. The scores for 11 different procedures that were performed at least five times positively correlated with the corresponding Aristotle morbidity scores: Pearson's coefficient r = 0.75. But the morbidity score for bidirectional cavopulmonary anastomosis (3.14 ± 1.77) was higher than the corresponding Aristotle morbidity score (2.0). It was lower for ‘conduit placement, right ventricle to pulmonary artery’: 1.08 ± 0.97, versus 2.0, and for arterial switch operation: 2.08 ± 1.11, versus 3.0.

CONCLUSIONS

The reported morbidity scores need to be tested on larger series and in different institutions. The introduced morbidity score has the potential to quantify postoperative complications accurately. Its estimation over time can facilitate the assessment of quality of congenital heart surgery. It will allow comparison of morbidity outcomes across institutions with different case-mixes.

INTRODUCTION

Secondary diseases or conditions can affect the physical or psychological well-being of patients who undergo repair or palliation of congenital heart disease, either during surgery or in the early postoperative period. These complications, mild or severe, are regularly reported and their incidence varies to a great deal, particularly in relation to procedural complexity. Morbidity constitutes one of the three parameters evaluated by the Aristotle complexity score. But its estimation until today is indirect. It is based on the duration of stay at the intensive care unit (ICU). An index of morbidity combining prolonged hospital stay and ICU stay was proposed [1–3].

Morbidity should be evaluated according to the observed postoperative complications or unfavourable events. These parameters should be weighed according to their potential to affect patient's quality of life and consume resources [4]. The aim of this work was to develop a score of patient's morbidity which is in conformity with complications that are likely to occur in the course of the surgical management of congenital heart disease.

PATIENTS AND METHODS

Complication list and patients

A team of paediatric anaesthetist, cardiac intensivist and cardiac surgeon met and drew up a list of conditions or diseases that can affect patients in the course of congenital heart surgery or increase costs of hospital stay. The complication list included 47 items divided into 5 categories: cardiac, respiratory, thoracic, neurological, wound complication and infection. Table 1 classifies these morbidity factors according to increasing scores, in alphabetical order, and then provides definitions for some of these complications. At the end of Table 1 is a list of complications observed during the study period and other additional adverse conditions that were not present in the original list of complications prepared prior to the study period. If no unfavourable condition was observed during all the postoperative period, a score of 0.5 points was assigned. Each ‘complication’ item was attributed a score ranging from 1 (mild or low) to 4 (severe or high) points, according to its estimated severity or costliness. In particular, the maximal figure of 4 points was attributed to ‘need of mechanical ventilation for more than 7 days’.

This complications list was tested on all patients who underwent surgery from 1 January to 31 March 2011. Two premature newborns, whose patent ductus arteriosus were closed, were excluded. Every postoperative complication was carefully recorded until discharge from hospital. Our unit carries out almost all kinds of congenital heart surgical procedures, except for cardiac transplantation. The International Congenital Heart Surgery Nomenclature [5] was used. When a patient underwent more than one procedure in the same hospital stay, he was assigned to the procedure with the highest Aristotle basic complexity (ABC) score. This procedure was considered as the main (primary) procedure and the other concomitant procedures as secondary procedures [1]. The following parameters were also collected: age at the time of surgery, ABC and Aristotle comprehensive complexity (ACC) scores, operation times, duration of postoperative mechanical ventilation and length of stay in the ICU.

Data analysis

All complications were recorded, including those associated with secondary procedures, during the whole period of hospital stay. The scores for each observed ‘complication’ were summed up in order to estimate a complication score after each main procedure. The complication scores were thereafter grouped and compared according to age at the time of surgery, ABC and ACC scores, operation data, and duration of mechanical ventilation and ICU stay, using the Mann–Whitney U-test. Correlation and regression between these variables and morbidity scores as well as the corresponding area under the receiver operating characteristic (ROC) curve (the C-index) were estimated. For each parameter, Pearman's correlation coefficient ‘r’ was calculated; the regression function with the highest goodness of fit ‘r²’ was retained.

According to the observed complications, a morbidity score was thereafter attributed to each main procedure. In conformity with the Aristotle scoring [1], a maximum of 5 morbidity points was assigned for observed complication scores, i.e. if the sum of observed complication scores amounted to more than 5 points, a morbidity score of only 5 points was attributed. This results into different figures of complication score and morbidity score when the sum of observed complication scores amounted to more than 5 points. The morbidity score for each procedure was thereafter compared with the morbidity score assigned by the ABC score. The observed morbidity scores that most correlated with the related Aristotle morbidity scores were thereafter used to compute a linear regression between the observed morbidity scores and ABC scores.

The software Graphpad Prism (San Diego, CA, USA) was used to perform the whole statistical analysis. Means were given with standard deviation and percentages with 95% confidence interval (95% CI) when indicated. The significance level was set at a P-value of <0.05.

Written consent was obtained from all patients or parents to use anonymous data for research, teaching and external quality control.

RESULTS

Procedures

In the 3 months observation term, 139 primary (main) procedures, among a total number of 184 operations, were carried out. Twenty-five of these 139 procedures were performed in newborns. Eight (5.8%) were performed without cardiopulmonary bypass (CPB). The most frequent operations included total cavopulmonary connection with extracardiac fenestrated conduit in nine patients (6.5%: 9/139), Norwood procedure and repair of complete atrio-ventricular septal defect, each on eight occasions (5.8%: 8/139) and bidirectional cavopulmonary anastomosis seven times (5.0%: 7/139). Table 2 lists all performed main procedures. Mean ABC and ACC scores reached 8.31 ± 2.52 and 9.62 ± 3.47 points, respectively (Table 2).

Outcome

Two patients died after operation, one after Norwood procedure and the other after repair of common arterial trunk: an early mortality of 1.4% (2/139) (95% CI: 0.2–4.9%). Ten were extubated immediately after surgery, and 57 within 24 h. The mean postoperative ventilation time for survivors was 53 ± 83 h, median: 25.50 h, range 0–604 h. Thirty-three patients stayed 1 day at ICU. ICU stay of more than 1 week was needed after 13 procedures. The mean ICU duration for survivors was 5.7 ± 7.9 days, median 3 days, range 1–49 days. No postoperative complication was detected after 46 (33%) procedures. Overall, there were 185 listed ‘morbidity’ conditions in connection with the other 93 surgical procedures, rendering a total score of 385 points. The following unlisted complications were observed: upper respiratory tract infection (n = 5), norovirus gastro-intestinal infection (n = 3), urinary tract infection (n = 2), presence of methicillin-resistant Staphylococcus aureus (n = 1) and massive ascitis (n = 1).

Complication scores

Table 3 shows the complications and unfavourable events that occurred during the postoperative period. The three most frequent conditions were ‘mechanical ventilation 25–95 h’ (n = 39, score = 2 points), ‘mechanical ventilation 4–7 days’ (n = 17, score = 3 points) and ‘low cardiac output syndrome’ (n = 15, score = 2 points). Table 2 gives for each procedure the related complication score. No complication was detected after closure of atrial septal defect (complication score: 0.5 points).

The complication scores ranged, for the 11 different procedures that took place at least five times, from 1.08 ± 0.97 points (conduit placement, right ventricle to pulmonary artery) to 5.75 ± 3.58 points (Norwood procedure). The mean complication score for the whole cohort reached 2.71 ± 3.26 points.

Complication scores correlation

The observed complication scores were higher after procedures requiring CPB (P = 0.046). Maximal scores were noted in the neonatal period (Fig. 1A). Scores increased with CPB duration (Fig. 1B) and with time spent on mechanical ventilation (Fig. 1C).

As testified by the ROC (Fig. 2A and B), the observed complication scores highly correlated with the Aristotle basic and comprehensive scores, with C-indices of 0.92 and 0.94, respectively. There was a satisfactory correlation with the length of ICU stay, having a C-index value of 0.70 (Fig. 2C).

Morbidity scores

The mean morbidity score was 2.14 ± 1.63. The morbidity scores ranged from 0.5 points (n = 46) to 5 points (n = 23), median: 2.0 points. They positively correlated with the corresponding Aristotle morbidity scores, but Pearson's correlation coefficient was not high: r = 0.46.

The morbidity scores for the 11 different procedures that were performed at least five times are displayed in Table 4. There was a high correlation (Pearson's r = 0.75) between these scores and related Aristotle morbidity scores. The morbidity score was lower than the observed complication score after bidirectional cavopulmonary anastomosis, aortic valvuloplasty, repair of complete atrio-ventricular septal defect, Fontan with external fenestrated conduit and the Norwood procedure. It is to be noted that the morbidity score for bidirectional cavopulmonary anastomosis (3.14 ± 1.77) was higher than the corresponding Aristotle morbidity score (2.0). On the contrary, the score was lower for ‘conduit placement, right ventricle to pulmonary artery’: 1.08 ± 0.97, versus 2.0, and for the arterial switch operation: 2.08 ± 1.11, versus 3.0. By removing these three ‘outliners’, one obtains a very high correlation between the observed morbidity scores and the morbidity scores assigned by ABC model, with a Pearson coefficient r equal to 0.93 (P = 0.0009), for the remaining eight procedures. The score numbers for these eight procedures were used to depict the linear regression (Fig. 3) between the Aristotle basic scores (x) and observed morbidity scores (y) with the following derived equation: y = (−0.23 ± 0.60) + (0.29 ± 0.06)x.

DISCUSSION

With improving outcome after congenital heart surgery, attention will progressively focus on quality of life for early survivors who constitute, nowadays, the vast majority of patients. They represented 97.5 (1783/1828) (95 CI: 96.7–98.2%) of the cohort operated upon in our unit in years (admission-based survival) [6]. To further improve treatment quality, the multidisciplinary team will need, sooner or later, an objective evaluation of postoperative morbidity. This evaluation was already requested in an editorial by Lacour-Gayet et al. [7]. Clarke et al. [3] proposed in 2008 that morbidity score be constituted of the following four components: postoperative hospital length of stay, postoperative time on ventilator, postoperative ECMO and/or ventricular assist device time, and major complications (re-operation, permanent pacemaker for atrio-ventricular block, nerve palsy, neurological disorders and dialysis). But the algorithm is still under development and this kind of morbidity score has not yet been published.

We considered that any surgical intervention is associated with some sort of uncomfortableness and assigned therefore 0.5 score when no unfavourable postoperative event was detected. We chose to not include hospital death in complication list, even if postoperative mortality due to mismanagement constitutes the most devastating event. We are of opinion that concepts of mortality and morbidity should not be mixed. Complications leading to death should be separately evaluated.

This study demonstrates that the chosen complication scores highly correlate with Aristotle basic and comprehensive scores. But the C-index for ICU length of stay, the current morbidity surrogate for the Aristotle model was only satisfactory: 0.70. It was already stressed by Heinrichs et al. [4] that morbidity index based on ICU length of stay is distorted in those centres, like our own, where intermediate care units function. This study adds evidence to known experience that postoperative morbidity is higher after neonatal congenital surgery and after procedures requiring longer CPB times.

The morbidity scores reported here should be compared with those from other units in order to establish a ‘common’ morbidity score for each recorded procedure in congenital heart surgery. Meanwhile, the formula provided in Fig. 3 could be used to extrapolate morbidity scores from ABC scores and allow comparison. O'Brien et al. [2] acknowledged that mortality after the arterial switch operation was lower than labelled by the Aristotle score. We found that the observed morbidity score was also lower. If these findings were confirmed, the basic Aristotle score for this procedure should be diminished from 10 to, for example, 9 points. On the contrary, upgrading the bidirectional cavopulmonary anastomosis procedure might be considered.

Availability of an accepted morbidity score based on observed postoperative complications will pave the way towards accurate estimation of performance in congenital heart surgery, not only based on early survival (‘surgical performance’ [1, 6]) but also on observed ‘non-appearance of morbidity’. The equation should also involve the complexity of performed procedures. It could be formulated as following: Performance = Aristotle complexity score × (1 – morbidity score/5), as the maximal mean morbidity score to be observed is 5 points. In this series, the result of performance based on non-appearance of morbidity would be 4.74 ± 0.28. It would have reached 7.48 ± 2.27 if no complication had been detected.

LIMITATIONS OF THIS STUDY

The list of complications and unfavourable conditions appears to be incomplete. The unlisted events reported in this study shall be added to the next ‘complication catalogue’. It is clear that scores assigned to each complication item may be challenged. But the maximum score of 4 points attributed to cardiopulmonary resuscitation and the use of ECMO and/or other assist devices is unlikely to be contested. A consensus will have to be reached for evaluation of severity of observed negative outcomes after congenital heart surgery, having regard to whether these unfavourable events are associated or not with temporary or permanent sequelae, how they are tolerated by the patient, and how much they increase costs of hospital stay. This study may be considered as the first step towards evaluation of this objective morbidity and should trigger discussion.

This series is limited, with only 11 procedures having been performed at least five times. Results have to be confirmed with larger number of different procedures over a longer period of observation at multiple institutions. Some other complications might have occurred after discharge from hospital and could not be recorded.

CONCLUSION

This study constitutes the first reported attempt to quantify observed morbidity after congenital heart surgery. The completed list of complication scores (footnote, Table 1) will be further applied for a longer period. Resulting morbidity scores bear potential to accurately estimate morbidity in connection with congenital heart surgery. The morbidity scores based on observed complications will contribute to revision of the Aristotle complexity score, in the same way as the observed mortality risks [8]. Estimation over time of such morbidity scores can facilitate the assessment of quality of paediatric cardiac surgical programmes with different case-mixes, provided that it will be accepted by the paediatric cardiovascular surgery community.

Conflict of interest: none declared.

References