Complication rates rise with age and Haller index in minimally invasive correction of pectus excavatum: A high-volume, single-center retrospective cohort study

,

Results: A total of 2013 patients were corrected by minimally invasive repair of pectus excavatum with a median age (interquartile range) for correction of 16.6 (5) years.Overall compounded complication rate occurred at a frequency of 16.4%, of which 9.3% required invasive reinterventions (Clavien-Dindo classification !IIIa).The complication rate related to bar insertion was 2.6-fold higher compared with bar removal (11.8% vs 4.5%, respectively).Multivariable analysis revealed age (adjusted odds ratio, 1.05; P < .001),precorrection Haller Index (adjusted odds ratio, 1.10; P < .033),and early-phase institutional experience (adjusted odds ratio, 1.59; P < .002)as independent predisposing risk factors.The optimal age of correction was 12 years, and the compounded complication rate correlated exponentially with age with a doubling time of 7.2 years.Complications increased 2.2-fold when the Haller index increased to 5 or more units.
Conclusions: Minimally invasive repair of pectus excavatum is associated with a high compounded complication rate that increases exponentially with age and high Haller Index.Consequently, we recommend repair during late childhood and early adolescence, and emphasize the importance of informing patients and relatives about the significant risks of adult correction as well as the need of 2 consecutive procedures taking the complication profile into account before planning surgery.(J Thorac Cardiovasc Surg 2024;-:1-13) Age at Correction 24  28   The compounded complication rate for patients aged 12 to 25 years (age interval containing 90% of population).Red line marking best fit of exponential equation.

CENTRAL MESSAGE
Higher age and Haller index, and lower institutional experience are associated with risk of complications in relation to correction by MIRPE.

PERSPECTIVE
In MIRPE, the lowest incidence of complications is observed in late childhood and young teenagers, and the predictive complication factors of age and severity should be considered in preoperative assessment and preparation.
See Commentary on page XXX.
Body ideals and self-perception are becoming increasingly important, and cosmetic procedures have increased extensively over recent years. 1 Surgical correction of the most common congenital chest wall deformity pectus excavatum (PE) is no exception and has increased immensely over the last decades. 2From a cosmetic and psychologic point of view, correction of PE has a significant impact on patient self-perception and self-esteem. 3,4In addition to the psychological effects, the literature frequently reports of improvements in various cardiopulmonary parameters after surgical correction of PE. 5,6 However, despite numerous descriptions of patients reporting subjective improvement of exercise tolerance after PE correction, 7 convincing and consistent objective data to corroborate this perception physiologically have been elusive. 6,8inimally invasive repair of pectus excavatum (MIRPE) is the most frequently used surgical technique for correction 9 and has gained widespread popularity and acceptance. 10,11MIRPE is characterized by small skin incisions, decreased blood loss, and shorter operative time; does not require cartilage resection; and is associated with low postoperative morbidity and length of stay. 9,12owever, the full MIRPE treatment regimen consists of 2 consecutive surgical procedures, that is, pectus bar insertion for correction followed by pectus bar removal after 2 to 3 years. 13Although MIRPE was originally introduced for correction of PE in prepubertal and early pubertal patients, the technique was subsequently applied in adults with minor modifications and is now regarded as the standard treatment regardless of age. 10,11,14,15everal small studies have reported increased complications in relation to correction of PE in older age. 11,15Other studies report correction at an older age to be safe and feasible. 14In this context, there has been extensive debate in the medical literature regarding the optimal age of correction to minimize the risk of complications. 2,12,16Up until now, the literature has addressed the complication rate in relation to age by a dichotomous approach between young and old patients and depending on the cutoff age with an odds ratio from 1.09 to 4.30. 2 Likewise, a dichotomous approach typically is used to describe complications in relation to PE correction and bar removal with complication rates ranging from 2.4 to 27% and 0 to 59%, respectively. 2,17To date, a compounded complication rate (CCR) systematically describing the summarized complications of the 2 consecutive surgical procedures needed for the full MIRPE regimen has not been put forward.Therefore, the aim of this single-center retrospective cohort study was to examine the CCR and to assess the risk factors associated with MIRPE.

Design and Oversight
This retrospective, single-center cohort study was conducted at the Department of Cardiothoracic and Vascular Surgery, Aarhus University Hospital, Denmark.All patients undergoing MIRPE between 2001 and 2022 were identified and included using national procedure codes (correction and bar extraction, KGAF10 and KGAF03, respectively).Patients were excluded due to misclassification or missing medical charts (study oversight and data flow, Figure 1).To provide a comprehensive analysis that encompasses the entirety of our center's experience, we chose not to exclude patients providing data to previously published articles from our center.The study was conducted in accordance with the Declaration of Helsinki, approved by the Institutional Board of Patient Safety, Aarhus University Hospital, and reported in conformity with Strengthening the Reporting of Observational studies in Epidemiology guidelines for observational studies.
Demographic, baseline, and perioperative variables in relation to both procedures were recorded (Table 1).By assessing chest x-rays preoperatively and 6 weeks postoperatively, the pre-and postcorrection Haller Indexes were calculated on the basis of patient chest widths and minimal sternovertebral depth (mSVD).Complications in relation to bar insertion and removal were graded according to the Clavien-Dindo classification and categorized as shown in Figure 2, C (bleeding, infection, displacement, overcorrection, pain, and other).The Clavien-Dindo classification ranges from minor complications (grade I, any deviation from the normal course) through major complications (grade III-IV) to death (grade V).An overview of the Clavien-Dindo classification is shown in Figure 1.Complications occurring before or causing premature bar removal were attributed to the bar insertion procedure.Because both procedures are required to complete the MIRPE regimen, we defined the CCR as a composite of complications in relation to bar insertions (PE correction) and bar removal procedures for each patient, as well as complications occurring in the intermediate period and until at least 30 days after bar removal.When biprocedural complications occurred, the most severe complication was graded according to Clavien-Dindo classification when reporting CCR.

Patient Management and Surgical Technique
Patients were referred by general practitioners, pediatricians, pediatric orthopedic surgeons, cardiologists, or thoracic surgeons to our congenital chest wall outpatient clinic (Figure 1).If the PE condition was suspected to be a part of syndromic disease or other specific conditions, further workup was conducted.All patients were routinely examined with chest radiography in 2 dimensions before and 6 weeks after a full MIRPE regimen.
The MIRPE with modifications used at our department has been described in detail. 18,19Briefly, the bar insertion procedure was performed using general anesthesia and epidural analgesia.A thoracoscope was introduced into the right pleural cavity, and through bilateral incisions, an introducer was visually guided through the pleural cavity from the right side and past the anterior mediastinal soft tissue in close contact with the sternum to the left side.After preliminary correction of the deformity using the introducer, the definitive asymmetrical prebent correctional metal bar (Zimmer Biomet) was placed retrosternal.The standard procedure was to place a Routinely, the bars were removed under general anesthesia after 3 years and the procedure was typically performed by residents.The scar from the primary operation on the side with the stabilizer(s) was reopened/excised and stabilizer(s), steel wire(s), and bar(s) were dissected and extracted. 18atients were routinely discharged on the day of bar removal, and chest x-ray and other examinations were performed on clinical indication only.At discharge, patients received paracetamol and nonsteroid anti-inflammatory drugs with gradual reduction and full discontinuation after approximately 1 week.Allowed therapeutic regimens are: drugs as antiemetics, antipyretics, analgetics, diuretics and electrolytes and physiotherapy.This grade also includes wound infections opened at the bedside.

Statistics and Data Analysis
Requiring pharmacological treatment with drugs other than such allowed for grade I complications.Blood transfusionsand total parenteral nutritionare also included.Numbers in the diagram refer to CCR (%).CNS, Central nervous system; IC, intensive care; ICU, intensive care unit.

Grade IV
and chi-square test were performed to assess risk factors associated with complications, as appropriate.Univariable and multivariable logistic regression were performed to identify potential risk factors for complications (Clavien-Dindo classification !1).A separate multivariable analysis was performed for complication of Clavien-Dindo class III or greater to assess risk factors for more severe complications grades.The design of the multivariable analysis model was based on clinical relevance of variables for analysis.We chose to include variables related to (1) patient characteristics (ie, age and gender), ( 2) the degree of PE severity (precorrection Haller Index), and (3) intraoperative variables (number of bars inserted and degree of change in mSVD).Missing data were assumed to be missing at random and listwise deletion was applied in the multivariable analysis.

Patient Characteristics
A total of 2013 patients corrected for PE between 2001 and March 2022 were included (22 patients were excluded of 2035 primary identified patients) and with a male:female ratio of 5.7 (85% male and 15% female, Figure 1 flow chart).Median follow-up time was 4.6 years (range, 3 months to 17 years) from the last performed procedure to final follow-up in 2022.Age range was 7 to 58 years with a median (interquartile range) age for correction of PE of 16 years (5).The baseline characteristics of all patients (n ¼ 2013) and patients with data available for CCR calculation (n ¼ 1804) are presented in Table 1, and variables in relation to bar insertion, bar removal, and procedural risk factors are grouped accordingly.Median precorrection Haller Index was 3.3, and after correction the Haller Index was reduced to a median of 2.2 (Figure 3).In univariable analysis, all variables in Table 1 show significant differences between patients without and with complications (P .005)except for sex and postcorrection mSVD (P ¼ .36 and .43,respectively).

Complication Rates
Of all patients (n ¼ 1804) completing MIRPE, 295 (16.4%) experienced at least 1 complication related to bar insertion or removal, and 12 patients (0.7%) experienced biprocedural complications.Complications requiring reinterventions (graded as Clavien-Dindo classification !III) accounted for 70% of all complications registered (Figure 2, A).The majority of complications were seen in relation to bar insertion compared with removal (11.8% vs 4.5%, Figure 2, B and D).Moreover, 33 patients (1.8%) required premature (<2 years) removal of bars as a consequence of complications.During the early inclusion period, as outlined in Figure 4, A, a small but significant year-by-year reduction of CCR was seen during the first 7 years (early phase), which corresponded to 471 patients (26%) corrected with MIRPE.This

Risk Factor Analysis
In multivariable analysis (Table 3, Section 1) including 1770 patients with Clavien-Dindo classification I to IV (missing data n ¼ 34, 1.8%), age (adjusted odds ratio [OR adj ] ¼ 1.05 per year), precorrection Haller Index (OR adj ¼ 1.10 per unit Haller Index), and early-phase correction (OR adj ¼ 1.59, early vs late-phase experiences) were independent risk factors for complications when adjusting for sex, age, D-correction mSVD, and number of bars inserted.A similar pattern was found when multivariable analysis with the same variables was repeated on complications of Clavien-Dindo class III or more (Table 3, Section 2).
To further describe the nature of the individual risk factors, subanalyses were performed for age, precorrection Haller Index and institutional experience (learning curve).First, when CCR (%) was plotted as a function of age, an exponential tendency was uncovered with a complications doubling time of 7.2 years for patients aged 12 to 25 years (90% of the population) (Figure 5, A-C).
Second, when a similar analysis was performed for precorrection Haller Index, a severity relationship was uncovered with stable CCR of 14.4% until Haller Index greater than 4.5, after which a 2.2-fold increase in CCR was observed (Figure 5, D-F).To convert the data into a general clinically appreciable presurgical assessment tool, the CCR of age and precorrection Haller Index were transferred into a 2-variable table for overall Clavien-Dindo classification and Clavien-Dindo classification III or greater (Table 4).
Third, regarding the learning curve, it was found that for the correctional procedure, age less than 20 years and Haller Index 4.5 or less subgroups followed the general learning curve (Figure 4, B-D).Conversely, the year-by-year complications rates for bar removal, age 20 years or more, and Haller Index greater than 4.5 subgroups diverged from the general learning curve (Figure 4, B-D).

DISCUSSION
This retrospective, high-volume, single-center cohort study examined the CCR and risk factors associated with 2 consecutive surgical procedures of the full MIRPE regimen.The overall CCR occurred at a frequency of 16.4% of which 9.3% required invasive reinterventions (Clavien-Dindo classification !IIIa), that is invasive surgical, radiological, or endoscopic interventions.However, the CCR decreased over time as institutional experience increased reaching a stable level after 7 years of 15.2%.The learning curve was seen only for bar insertion, ages less than 20 years, and Haller Index 4.5 or less.Multivariable analysis revealed age (OR adj ¼ 1.046) and precorrection Haller Index (OR adj ¼ 1.12), and early-phase correction (OR adj ¼ 1.59) as independent predisposing risk factors of overall compounded complications when adjusting for variables.Similar predisposing risk factors were present for more severe complications graded as Clavien-Dindo classification III or greater.Furthermore, in a

Minimally Invasive Pectus Excavatum Repair
Predisposing complication risk factors for minimally invasive pectus excavatum repair are increasing age, severity of pectus excavatum and institutional experience

Compounded Complication Rate
The overall CCR of 16.4% in this study is below the weighted insertion complication rate of 19.7% in a metaanalysis of 15 studies. 2 However, in 9 of the 15 studies, pneumothorax was reported as a complication with an average complication rate of 7.9% and significantly higher in adults compared with adolescents. 2Several studies reported postoperative pneumothorax frequencies of 40 to 59% with the majority resolved within 24 hours after surgery even if they were of considerable size. 20,21We regarded most pneumothoraces as a natural consequence of interpleural space manipulation and only accounted those requiring drainage as complications.Nonetheless, accounts of pneumothorax with severe clinical implications have been reported in cases with parenchymal lung injury occurring concurrently with the unification of the pleurae during MIRPE.
When separating CCR into insertion and removal, complication rates were 11.8 and 4.5%, respectively.In regard to the former, infection was the single most frequent complication in this study with a rate of 3.3%.The infectious complications ranged from superficial skin infections treated with oral antibiotics to deep chest wall or pleural infections that require vacuum treatment, removal of pectus bar(s), or other surgical interventions (Table 3).There are  multiple reports of infections in relation to PE correction with variations from no infections to a rate of 22%. 22his study provides a systematic account of infectious complications associated with the 2-stage procedure of MIRPE and the infectious-related CCR of 4.8% is, as mentioned, mainly conveyed by the primary procedure.This relatively high rate could be linked to the introduction of a foreign object in proximity to the skin.However, our data do not appear to deviate from the infectious complication rate commonly reported in the existing MIRPE literature.Surprisingly, recent reviews and meta-analyses fail to detect any disparity in infectious complication rates between MIRPE and the 1-stage Ravitch procedures.Thus, it is worth noting that these investigations do not comprehensively consider the unique aspects of the 2-stage MIRPE procedure.However, the high variations of reported infection indicate a lack of attention and definition of infectious complications from surgeons and out-ofhospital registration.In all, this supports the notion of general underestimation of complication rate.
In regard to bar removal complication rates, our results are in accordance with other studies reporting complication rates of 3.5 to 4.1%. 23,24However, one retrospective study reported a removal complication rate as high as 17.5%. 17In agreement with the mentioned studies, the removal complication types of this study mainly consisted of bleeding and wound infection in 1.4 and 1.5%, respectively (Table 2).rates, severe and potentially life-threatening complications do occur occasionally.Specifically, in our cohort, 6 incidences of Clavien-Dindo classification IV complication due to major bleedings occurred.Thus, it is crucial for the operative team to be ready and properly equipped to handle these potentially life-threatening situations.Because MIRPE requires 2 consecutive surgical procedures, it is important to apply the CCR when comparing it with the alternative open Ravitch approach.The open Ravitch procedure is performed with different modifications as a 1-or 2-stage procedure, and its overall complication rate is similar to MIRPE in several reviews and meta-analyses.Notably, studies have not considered the 2-stage procedure of MIRPE when making comparisons with Ravitch, which further underlines the importance of applying the CCR for these interventions to facilitate comparison.Nevertheless, it appears that hospital stays and blood transfusion requirements are greater in relation to Ravitch when compared with MIRPE.Unfortunately, there is currently a lack of randomized studies that directly compare MIRPE with Ravitch.

Age and Optimal Timing of Correction
Optimal timing for correction has been an issue of immense debate in the literature. 2,12,16Several small retrospective studies and 1 meta-analysis reported increased complication rates in older patients. 2Previous studies describing complications in relation to age compared adults with children or adolescents, thereby stratifying age as a dichotomous relation. 2,12,16This approach might simplify the true nature of the relationship between age and complirate with the potential of missing crucial information of relevance for presurgical assessment and recommendation.However, this study is the first of its kind systematically fusing insertion and removal complications into a continuous age and CCR relationship.Arguments in favor of using MIRPE in prepubertal and early pubertal patients include greater chest wall compliance, residual growth, and remodeling potential at a younger age. 256][27] Regarding the latter, Papic and colleagues 28 have demonstrated that older patients with MIRPE necessitate a greater amount of pain medication compared with their younger counterparts.This age-related impact does not seem to manifest in the context of Ravitch repair.Furthermore, adult chest deformities may be more complex indicated by greater rates of anterior chest wall asymmetry. 29All together, these factors may explain the increased rates of complications in older patients. 21][32] This notion is supported by a recent meta-analysis showing that adults were more likely to be corrected with 2 or 3 bars, which, however, was associated with an increased complications rate. 2 Our study verifies a significant association between numbers of bars and CCR in the univariable analysis; however, despite observing an increased CCR with advancing age and a concomitant increase in the numbers of pectus barre with age, the multivariable analysis indicates that these 2 factors are, nevertheless, independent.

THOR
Therefore, age stands as a predisposing factor for complications, whereas the numbers of bars represent a dependent complication risk variable in our cohort (Table 3).

Implication of Haller Index
Several studies have compared precorrection Haller Index or mSVD in adolescents and adults and found no significant differences in age. 22,33To our knowledge, no previous study has linked precorrection Haller Index as an indicator of increased complications in relation to MIRPE.However, multivariable analysis and a detailed subanalysis show a clear and highly significant relationship between CCR and precorrection Haller Index.The relationship revealed an abrupt 2.2-fold increase in CCR at precorrection Haller Index greater than 4.5, making precorrection Haller Index assessment essential.It should be noted that according to the literature, an increase in the Haller index as individuals mature and a small but significant decrease of Haller Index during adulthood have been described 34 (þ PLoS One); however, we observed no alteration in the Haller index with advancing age in our specific group.The results of this study undoubtedly underline the considerable risk of complications related to full MIRPE.Furthermore, age, precorrection Haller Index, and low institutional experience are the main complication determinants, and appropriate counseling in patients with a precorrection Haller index of more than 4.5 and an age of more than 20 years should be performed.Awareness of a 2-stage procedure and risk of complications should be stressed when informing patients before MIRPE.With respect to the high risk of complications and the reduced CCR within the first 7 years of practice, this study points to highvolume centers as preferable for correction of PE and emphasizing the importance of centralizing the treatment.Interconnecting CCR of age and precorrection Haller Index reveal a detailed risk factor profile and provide a clinically applicable presurgical Age-Haller diagram.

Study Limitations
First, the study is unique because it systematically describes the CCR in relation to the full 2-stage MIRPE regimen.Moreover, this study clearly defines complications in accordance with the Clavien-Dindo classification and has set an uniform standard for complications, which enables an analysis of data with greater accuracy and strength.Furthermore, data are collected from a highvolume center with a long tradition of MIRPE.This is emphasized by the long period of inclusion and the number of patients included and only surpassed by 1 multiinstitutional database study and a single meta-analysis study both without access to patient charts. 2,12Despite the considerable number of patients included in the study, 90% of the patients were aged 12 to 25 years, which is why rigorous extrapolation of CCR to ages less than 12 years and more than 25 years should be made with caution.The majority of patients in our cohort undergoing MIRPE had no comorbidities or previous medical conditions and seems comparable to the background population. 35Additionally, it must be expected that some complications of Clavien-Dindo classification less than III are treated in the primary health care system or at other hospitals without reporting to our department, consequently underestimating less severe complications and possibly overemphasizing complications rates of Clavien-Dindo classification III or greater.Consequently, it is highly plausible that this present study generally underrates the true CCR of MIRPE.Although radiography has demonstrated validity and high interpersonal precision in determining Haller index, the gold standard remains computed tomography scanning.In our study, the Haller index is determined via conventional radiography, which may introduce a minor uncertainty and limitation in determination of a precise Haller index.However, we do not find it likely that this uncertainty influences the identification of preoperative Haller index as a predisposing risk factor.Because of the retrospective study design, it was not possible to precisely evaluate the continuous evolution of the procedure during the 20þ-year study period, thus limiting our assessment of the observed learning curve.
The application of the CCR and Clavien-Dindo classification should be the new standard applied in future studies when describing and assessing surgical complications in this field.Our results regarding the CCR, risk factors, and proposed precorrection Age-Haller diagram are to be

CONCLUSIONS
This study represents the first to systematically compile the complications associated with the 2-stage MIRPE procedure into a CCR.Furthermore, this study comprehensively highlights the significant risk of complications associated with MIRPE in patients with increasing age, Haller Index 5 or greater, and low institutional experience.
the normal postoperative course without the need for pharmacological treatment or surgical, endoscopic and radiological interventions.

FIGURE 1 .
FIGURE 1. Data flow chart.Enrollment of patients receiving full treatment regimen with insertion and correction of PE followed by pectus bar removal after 3 years.The flow of patients from primary contact to referring physician followed by treatment flow, extraction of data, and analysis.Complication related to pectus correction and bar(s) removal are graded according to Clavien-Dindo classification and complication grades are defined in the table.Numbers in the diagram refer to CCR (%).CNS, Central nervous system; IC, intensive care; ICU, intensive care unit.

FIGURE 2 .
FIGURE 2. Compounded complication rates categorized according to Clavien-Dindo classification and specific complication types.A, CCR (%) as a function of different Clavien-Dindo classification stages (I-IV) and biprocedural complications (patients exposed to complications at correction procedure and bar removal procedure).B, Complication rates in relation to PE correction as a function of Clavien-Dindo classification.C, Complication rates in relation to PE correction as a function of different complication types.Each complication type is divided into Clavien-Dindo classification (I-IV).D, Complication rates in relation to bar removal as a function of Clavien-Dindo classification.E, Complication rates in relation to bar removal as a function of different complication types.Each complication type is divided into Clavien-Dindo classification (I-IV).CDC, Clavien Dindo classification.

FIGURE 3 .
FIGURE 3. Definition and analysis of risk factors for CCR related to MIRPE.

TABLE 4 .
Age-Haller complication diagram for compounded complication rateSection 1prospective studies.Furthermore, new modified techniques to decrease complications should be the subject of future randomized studies.

TABLE 1 .
Baseline characteristics describing variables of demography, bar insertion, bar removal, and procedural risk factors in relation to the total cohort ± complications Univariable analysis was performed on all variables.Statistical methods and significance level are noted.IQR, Interquartile range; mSVD, minimal sternovertebral distance.4TheJournal of Thoracic and Cardiovascular Surgery c -2024Thoracic Media et al THORFigure 2, B and C, and Table 2 give a detailed description of the specific complications in relation to the 2-stage procedure including complication types and Clavien-Dindo classification grades.

TABLE 2 .
Detailed qualitative description of complications related to pectus correction and removal of bar(s)

TABLE 3 .
Multivariable analysis was performed on selected variables of clinical relevance for compounded complication rate units for age, precorrection Haller Index, and early-phase institutional correction are increase year À1 , increase unit À1 , and early-phase institutional correction versus latephase institutional correction (Figure4), respectively.Section 1 shows OR for all complications graded with Clavien-Dindo classification I to IV. Section 2 shows OR for complications graded with Clavien-Dindo classification III or greater.OR, Odds ratio; SE, standard error; mSVD, minimal sterno vertebral distance.The Journal of Thoracic and Cardiovascular Surgery c -2024 OR

5 .
Subanalyses of the independent precorrection risk factors of age and Haller Index.A, Age distribution of the included cohort (n ¼ 1804).B, Overall CCR determined at age of correction.Red rectangles in A and B indicate the 90% age distribution of the cohort.C, Zooming in on the 90% age distribution of A and red line marking best fit of an exponential equation.D, Haller Index distribution before correction with MIRPE.E, Haller Index distribution after pectus correction with MIRPE.F, Overall CCR as function of precorrection Haller Index.Data fitted to a Sigmoidal function that defines lower (14.35%;range, 12.14-16.38)and upper (33.30%; range, 30.46-36.28)average complication rate as a function of precorrection Haller Index with HI 50 of 4.9 and r 2 ¼ 0.94.