The purpose of this study was to use a meta-analytical technique to examine the efficacy of surgical repair of pectus excavatum on pulmonary function. Studies were retrieved via computerized literature searches, cross-referencing from original and review articles. Inclusion criteria were as follows: (1) reporting quantitative measures of preoperative and postoperative pulmonary function; (2) published in the English language; (3) indexed between January 1960 and September 2005; (4) reporting the duration between which preoperative and postoperative assessments were conducted; and (5) describing the pulmonary assessment procedures. The titles and abstracts of potentially relevant articles were reviewed to determine whether they met the criteria for inclusion. Twelve studies representing 313 pectus excavatum patients met the inclusion criteria and were used for the meta-analysis. Random-effects modeling yielded a mean weighted effect size (ES) for pulmonary function which was statistically nonsignificant (ES = 0.08, 95% CI = −0.20 to 0.35; P = 0.58). The findings of the present study indicated that surgical repair of pectus excavatum does not significantly improve pulmonary function. These findings, however, may be a result of testing pulmonary function under conditions in which pectus excavatum does not manifest itself.
Pectus excavatum is a congenital deformity of the chest wall with an incidence of approximately one in every 300–400 Caucasian male births . This condition is more common than Down syndrome which occurs one in every 600–1000 births . Although the pathogenesis of pectus excavatum remains unclear, investigators have hypothesized that the deformity results from unbalanced overgrowth in the costochondral regions. As a result, the chest appears concave and a displaced heart is often palpable on the left mid-axillary line slightly below the axilla. Pectus excavatum occurs more often in males than females (6:1) and accounts for 90% of congenital chest wall deformities [3,4]. Approximately 40% of pectus excavatum patients are aware of one or more members of their family who have pectus deformities; however, a genetic link has not been established . Despite numerous published reports, there is no consensus in the literature as to whether surgical repair improves pulmonary function [5–12]. Thus, it has been suggested by some researchers that correction of pectus excavatum should be considered a cosmetic procedure [13–15].
Although there have been a number of review articles discussing the effects of surgical repair of pectus excavatum on pulmonary function [16–19], they provide only narrative summaries which rely on statistical significance to differentiate between studies. This approach is potentially misleading because statistical significance in any individual study is influenced by multiple factors including sample size and variance [20–23]. Consequently, narrative summaries do not make optimal use of all available quantitative information. Meta-analysis is a statistical technique for literature synthesis in which quantifiable results from individual studies addressing a common problem are statistically analyzed to arrive at conclusions about a body of research . In the present study, meta-analysis was used to (1) aggregate and compare findings on the effectiveness of surgical repair of pectus excavatum on pulmonary function; (2) summarize and draw reliable conclusions on a body of literature where there is a lack of consensus; and (3) improve estimates of treatment effectiveness [24–27]. Meta-analysis represents each study’s findings in terms of an effect size (ES). Briefly, the ES statistic is a statistically standardized measure of the study findings such that the resulting numerical values are interpretable in a consistent fashion across all variables and measures . The ES provides information related to the magnitude and direction of an intervention rather than merely its statistical significance .
Recently, Malek et al.  conducted a meta-analysis to examine the effects of surgical repair of pectus excavatum on cardiovascular function. The results of this study  indicated that, on average, cardiovascular function improved by greater than one-half standard deviation following surgery. These findings supported the hypothesis that relief of cardiac compression brought about by the depressed sternum improves the hemodynamic response of the pectus excavatum patient [9,10,28]. No previous studies, however, have used the meta-analytical technique to examine the efficacy of surgical repair of pectus excavatum on pulmonary function. Therefore, the purpose of this study was to use a meta-analytical technique to examine the efficacy of surgical repair of pectus excavatum on pulmonary function.
Computerized literature searches were performed using Current Contents, EMBASE, HEALTH PERIODICALS DATABASE®, MEDLINE®, Nursing and Allied Health, and SPORTDiscus. In addition, references from retrieved review articles and original investigations were examined. The year 1960 was chosen as the starting date because few, if any, relevant studies which assessed preoperative and postoperative cardiovascular function in pectus excavatum patients were published prior to this time. The following keywords were used either alone or in various combinations for computer searches: pectus excavatum, chest wall deformity, funnel chest, pulmonary function, respiratory, ventilation, lung function, and pectus severity index. The titles and abstracts of studies that were identified in the computerized searches were examined to exclude any articles that were clearly irrelevant. The full texts of the remaining articles were retrieved, and each paper was read to determine whether it contained information on the topic of interest. Because computer searches have been shown to yield less than two-thirds of relevant articles in some research areas , reference lists from original and review articles were reviewed to identify any studies that had not been previously identified and appeared to contain information on the topic of interest. Hand searches of selected journals related to general medicine and surgery were also performed.
Inclusion criteria for this meta-analysis included the following: (1) reporting quantitative measures of preoperative and postoperative pulmonary function; (2) published in the English language; (3) indexed between January 1960 and September 2005; (4) reporting the duration between which preoperative and postoperative assessments were conducted; and (5) describing the pulmonary assessment procedures. Studies published in foreign language journals were not included because of the potential error in the translation and interpretation of findings. Abstracts from conference proceedings, doctoral dissertations, and Master’s theses were also not included because those sources are unlikely to report substantive research findings that are not published elsewhere. Studies meeting the inclusion criteria were examined to ensure that the same subjects were not included in more than one study .
A coding sheet was developed to record information from each article. The time to code each study ranged from approximately 1–3 h. The major categories of variables that were coded include (1) study characteristics (author(s), year, and number of subjects); (2) physical characteristics of subjects (gender and age); (3) type of surgical repair performed; and (4) primary outcomes (pulmonary indices) (Table 1 ).
Because not all studies reported the same index for pulmonary function, we elected to place the indices reported in each study into a global category: pulmonary function. However, all indices used in the current study represent standard pulmonary function characteristics used routinely in clinical settings. This was done to (1) maximize the number of studies that could be used in the meta-analysis; (2) have the best representation of the current findings in this body of literature; and (3) better answer the question, ‘Does surgical repair of pectus excavatum improve pulmonary function?’ The indices included in the present meta-analysis (Table 2 ) represent components that individually or collectively determine pulmonary function (e.g., FEV1, FVC, MVV, FEV1/FVC, TLC, etc.).
Standardized mean change
The primary outcome of interest was possible changes in pulmonary function following surgical repair. The standardized mean change  was calculated using the following formula: ESsg = [(meanpostoperative − meanpreoperative)/spreoperative], where (meanpostoperative − meanpreoperative) equals the amount of change in the mean on a variable of interest for a sample measured preoperatively and then, later, postoperatively, whereas the spreoperative equals the standard deviation for the preoperative time point. The spreoperative was used instead of the pooled standard deviation, because theoretically, spreoperative is unaffected by the treatment (i.e., surgical repair). Because each study in the present meta-analysis produced multiple ESs, we followed the recommendation of Lipsey and Wilson  that an average ES for each study should be calculated to maintain statistical independence.
The standard error (SEsg) and inverse variance ( ) for the average ES of each study was calculated using the following equations : and , respectively. No studies reported the correlation (r) between the preoperative and postoperative scores. Lipsey and Wilson  noted that the computed weighted ES is not very sensitive to the r-value and suggested that an r-value of 0.80 can be substituted in the above formulas when the correlation between the preoperative and postoperative scores are not reported . Based on this recommendation, SEsg and were calculated using a conservative r-value of 0.80.
To determine whether each effect size in a set of effect sizes can be viewed as a measure of a common population effect size (i.e., there is consistency across the studies), a homogeneity statistic Q was calculated. The Q statistic has an approximate chi-square distribution with k − 1 degrees of freedom, where k is the number of effect sizes . If Q exceeds the critical value for chi-square with k − 1 degrees of freedom (P ≪ 0.05), then each observed effect size is assumed to differ from the population mean (i.e., a random-effects model). However, if Q does not exceed the critical value, then the observed effect size estimates the corresponding population effect (i.e., a fixed-effects model) . It has been suggested by Higgins and co-workers [31,32] that, in meta-analyses with a small number of studies, the Q statistic may have low statistical power. Therefore, these investigators developed the I2 statistic which ‘… describes the percentage of total variation across studies due to heterogeneity rather than chance.’  (p. 558). The I2, in the present study, was calculated by using the following equation : I2 = 100% × (Q − df)/Q. The value for I2 ranges from 0 to 100% with a larger number indicating greater heterogeneity .
Publication bias analysis
Publication bias is the tendency for journals to publish only studies that yield statistically significant results . Although publication bias was not anticipated to be a substantive factor in the pectus excavatum literature, this analysis was performed. The approach used to determine publication bias was to examine a funnel plot by plotting the sample size on the vertical axis and the ES measure on the horizontal axis for all studies [24,26,33]. Ordinarily smaller studies show a larger range of ESs than larger studies. Thus, the plot generally resembles an inverted funnel, wide on the bottom and narrow at the top. A gap at the bottom of the funnel on the left side in the range of nonsignificant test results would indicate that small studies with null results may be missing from the published literature .
A categorical model was used to determine the relationship between the surgical repair technique used in each study and the magnitude of the ESs, by using the procedures outlined by Lipsey and Wilson . This model provides a between-group effect analogous to a main effect in an ANOVA design, and a test of homogeneity of ESs within each group. The between-group effects, which can be used to assess differences between study features, are estimated by QB which has an approximate chi-square distribution with k − 1 degrees of freedom, where k is the number of groups defined by study features of interest .
To examine the influence of specific variables (i.e., age and duration between preoperative and postoperative assessment) on changes in pulmonary function, least squares regression models were calculated with each ES weighted by the reciprocal of its variance .
Statistical analysis was performed by using the Statistical Package for the Social Sciences software (version 13.0; SPSS Inc., Chicago, IL, USA). Results were considered statistically significant if P ≪ 0.05. Also, confidence intervals were reported in all cases at the 95% level.
The initial computerized searches identified 171 potentially relevant articles using the search terms ‘pectus excavatum’ and ‘pulmonary.’ Furthermore, reference lists from original and review articles were reviewed to identify any studies that had not been previously identified from the computerized searches. As a result, 19 studies met all of the inclusion criteria [6–8,11,13,34–46]. The studies by Gyllensward et al. [44,45] and Borowitz et al.  did not provide sufficient statistical information to estimate ESs for pulmonary function indices for patients who had surgical repair. Three studies [36,42,43] appeared to have overlapping samples. Therefore, the study by Hu et al.  was used because it provided the necessary statistical information for a valid estimation of ES. The studies by Quigley et al. , and Haller and Loughlin  appeared to use the same sample of pectus excavatum patients. Thus, the information from these two studies were combined and analyzed as one study. A similar approach was used for the studies by Morshuis et al. [11,12]. Also, Kowalewski et al. [7,38] published two studies in 1998 and 1999 using samples that overlapped. Therefore, only the initial study published in 1998 by Kowalewski et al. , which included both moderate and severe pectus excavatum groups, was included in the meta-analysis. Similarly, Sigalet et al.  and Bawazir et al. , and Derveaux et al. [13,47] used overlapping patients; only the studies by Sigalet et al.  and Derveaux et al.  were used in the meta-analysis. Therefore, a total of 12 studies were analyzed for the meta-analysis (Table 1). Six of the studies were conducted in the United States [6,8,15,34,39,40], while the remaining studies were conducted in Belgium , Canada , Japan , Poland , Netherlands , and China . A total of 313 pectus excavatum patients were assessed in the meta-analysis. Four studies did not report a pectus severity index [13,34,36,39], whereas the remaining studies used various techniques for estimating a pectus severity index [6–8,12,37,40,41]. Therefore, the relationship between pectus severity and the average ES for each study could not be examined in the present study. The number of patients assessed preoperatively and postoperatively in each study ranged between 5 and 152. The time between the preoperative and postoperative cardiovascular assessments ranged from 3 to 146.6 months. Four of the 12 studies used the Ravitch repair surgical procedure, two studies used the Nuss procedure, two studies used other methods, one study used the Daniel procedure, and three studies did not report the type of procedure (Table 1). All 12 of the studies reported that subjects were healthy and/or physically active; however, this information was not quantified in terms of duration, frequency, or mode of exercise performed.
Standardized mean gain (pulmonary function)
Measures of the ES were computed for each dependent measure in each study. A total of 52 ESs for pulmonary outcomes were calculated from the 12 studies. Of these, 87% were reported as statistically nonsignificant, whereas 13% were reported as statistically significant. Of the 52 ESs, 42% (n = 22) were in a negative direction, 2% (n = 1) equaled zero, and 58% (n = 30) were in a positive direction (Table 1). The Q statistic was significant (Q = 78.22, P ≪ 0.001) indicating heterogeneity. This was further supported by the I2 statistic (83.3%). Thus, a random-effects model was used to estimate pooled effects and error for the 12 studies. The average ES for each study and the mean weighted effects size for all studies combined are shown in Fig. 1 , along with 95% confidence intervals. The overall mean weighted ES for pulmonary function was not statistically significant (ES = 0.08, 95% CI = −0.20 to 0.35; P = 0.58).
There was no quantitative evidence supporting publication bias as indicated by the near zero correlation between effect size and sample size (r = −0.19, P = 0.52). A funnel plot analysis showed no evidence of missing studies with small or negative effects.
No statistically significant differences or relationships were found when changes in pulmonary function were partitioned or regressed according to (1) surgical procedure used to repair the pectus excavatum; (2) age; and (3) duration between preoperative and postoperative assessment.
The principal finding of the current meta-analysis is that the average effect for surgical repair of pectus excavatum on pulmonary function was very small (0.08) and not statistically significant based on available measures. The nonsignificant finding may be explained, in part, by the conditions during which ventilation was measured. In pectus excavatum patients, the dynamics of ventilation greatly differ when measured at rest, rather than during exercise [9,10]. Better measures of the effects of the chest wall deformity on pulmonary function would be obtained through an incremental exercise test. It is well established in the clinical exercise physiology literature that examination of pulmonary function during incremental cardiopulmonary exercise testing is as important in patient assessment as assessing cardiovascular function [48–50]. Typically, in healthy, active individuals, exercise tolerance is limited by the cardiovascular response with a substantial pulmonary reserve remaining [48,50]. Due to the configuration of the chest wall in pectus excavatum patients, it has been suggested that the pulmonary response is similar to individuals with restrictive lung disease [37,40]. Measures of pulmonary function taken at rest, which are typical for the studies in the current meta-analysis, are not sensitive to pulmonary deficiencies that may appear during exercise.
When assessing pulmonary function in pectus excavatum patients it is advantageous to examine indices such as minute ventilation ( ), tidal volume, respiratory frequency, and pulmonary reserve during an incremental exercise test. is a product of tidal volume and respiratory frequency and during an incremental exercise test may increase by 30-fold as the individual reaches max . This increase is due, in part, to an increase in both tidal volume and respiratory frequency. In pectus excavatum patients, however, the ability to increase tidal volume is limited, because of the chest wall configuration. As a result, the continued increase in over the course of the incremental test is predominantly due to an increase in the respiratory frequency. Malek and Fonkalsrud  reported a decrease in respiratory frequency with a concomitant increase in tidal volume following surgical repair in a moderately severe pectus excavatum patient using the Highly Modified Ravitch Repair (HMRR). Borowitz et al.  reported similar findings for 10 pectus excavatum patients who had undergone surgical repair using the Nuss procedure. It is also beneficial to examine the maximal value (i.e., max) attained at the end of an incremental test to exhaustion which can be compared to the MVV value attained during the pulmonary function test. To evaluate whether the patient exhibited pulmonary limitation to exercise, the investigator can compare the max value attained at max as a percentage of the patient’s MVV value (i.e., %MVV = [ max/MVV] × 100). In healthy individuals, the %MVV at max is approximately 50–70% [48,50,51], whereas in patients with respiratory disease (e.g., COPD) this value may be greater than 80% [48,52,53]. Malek and Fonkalsrud  reported a clinically significant change in %MVV following surgical repair (109% vs 76%), which also resulted in a substantial pulmonary reserve. Following surgical repair via the Nuss procedure, Sigalet et al.  reported a significant decrease in max (71.3 l min−1 vs 60.7 l min−1) in 11 pectus excavatum patients. Thus, to better understand the effects of pectus excavatum on pulmonary function, future studies should examine indices such as tidal volume, respiratory frequency, and max during incremental exercise testing conditions in addition to using routine pulmonary function test.
An interesting finding in the present study was that no statistically significant differences in pulmonary function were observed between surgical techniques. It should be noted, however, that the sample size was small, which limits statistical sensitivity. Nevertheless, from a clinical perspective, significant differences between surgical techniques do exist that warrant brief discussion. Surgical repairs of pectus excavatum have been performed using extensive modifications of the original procedure described by Brown and Cook  and modified by Ravitch  and Welch , with minimal costal cartilage resection, or without costal cartilage resection as proposed by Nuss and co-workers [19,57]. Currently, two procedures are predominantly used to correct pectus excavatum, the Highly Modified Ravitch Repair [1,4] and the Nuss-MIRPE [19,57]. In the past decade, various studies have suggested that the Nuss-MIRPE procedure for repair of pectus excavatum may be a less invasive alternative to the HMRR [19,57]. Reports [58–60] regarding the HMRR and the Nuss-MIRPE procedure, however, indicate that although the operating time is shorter for the Nuss-MIRPE procedure, the length of hospitalization, severity of postoperative pain, complication rate (i.e., pneumothorax and/or bar displacement), need for reoperations, and limitation in activity may be more severe with the Nuss-MIRPE procedure than with the HMRR. Additionally, the time to bar removal, on average, is 24 months for the Nuss-MIRPE procedure [58–60], whereas patients who have the HMRR have the bar removed in 5–6 months following the surgery [1,4,61]. More recently, studies on the Nuss-MIRPE procedure have reported complications such as hemorrhaging and the inability to perform cardiopulmonary resuscitation because of the placement of the sternal bar [62–64]. These incidences have not been reported with the HMRR [1,4,58,65]. Furthermore, studies have reported a small, but clinically significant amount of recurrent pectus excavatum associated with the Nuss-MIRPE technique . Perhaps a salient feature of the HMRR is that less than 3% of patients operated on have had recurrence severe enough to cause any symptoms . Most of these recurrent cases, however, were from early years when the operation was performed on young children and without a support bar (personal communication Dr Eric W. Fonkalsrud, MD, FACS). It is important to note that the HMRR is based on modifying the structural framework of the chest wall , whereas the Nuss-MIRPE procedure avoids the anterior chest incision, cartilage resection, and sternal osteotomy . Therefore, as recommended by Fonkalsrud , studies with longer follow-up periods are needed to determine the advantages and disadvantages of these two surgical techniques in relations to cardiovascular and pulmonary function.
The application of quantitative meta-analysis in the present study offers advantages compared to previous comprehensive reviews of pectus excavatum because to date such reviews of literature have resulted in subjective qualitative summaries based on narrative descriptions of chronologically arranged studies [16–19]. Individual studies reporting pulmonary function in pectus excavatum patients may have a small sample size with large variances. As a result, there is very low statistical power (high Type II error rates) and the failure to attain statistical significance is an inadequate basis for making decisions about the overall efficacy of surgical repair of pectus excavatum. In the present investigation, meta-analysis provides a quantitative synthesis of the entire body of literature on pectus excavatum resulting in a larger effective sample size. Thus, the ES estimates and the statistical conclusions resulting from a meta-analysis are more robust and accurate than those provided from single studies or narrative reviews [24,26].
The present meta-analysis along with the findings of Malek et al.  are the first investigations to provide comprehensive analytical data regarding the physiological effectiveness of surgical repair for pectus excavatum. The results of the current investigation, in conjunction with those of Malek et al. , indicated that surgical repair of pectus excavatum significantly improves cardiovascular function, but not pulmonary function as indicated by the indices reported (Table 2). Future studies on the efficacy of surgical repair of pectus excavatum should consider examining pulmonary function in response to an incremental exercise test rather than using only measures taken at rest.
This research was supported by an ACSM Foundation Research Grant (FRG) from the American College of Sports Medicine Foundation (M.H. Malek).