Apparent truth about congenital diaphragmatic hernia: a population-based database is needed to establish benchmarking for clinical outcomes for CDH

Apparent Truth About Congenital Diaphragmatic Hernia: A Population-Based Database Is Needed to Establish Benchmarking for Clinical Outcomes For CDH By The Ontario Congenital Anomalies Study Group* Toronto, Canada

Background: The authors hypothesize that recent single or multiinstitution-based reports of improved survival of congenital diaphragmatic hernia (CDH) patients are biased by patient selection, practice, and referral patterns. Here the authors report a population-based analysis of the clinical outcomes of CDH in the province of Ontario for 1996.

interval, 0.5, 3.7; P ⬎ .01). Neonatal death (⬍28 postnatal days) accounted for the majority, 13 of 14 (92.8%) of deaths. Six of 14 (42.9%) CDH-associated deaths, however, were not accounted by the institutional-based reporting. In addition, institutional-based survival rates for CDH varied from 62.5% to 100%.

Methods: A retrospective analysis of cross-sectional data from the Bureau of Vital Statistics of Ontario and all 5 pediatric surgical institutions in Ontario for 1996 was performed.

Conclusions: Our results indicate the existing bias associated with institution-based reporting and database of CDH. The “hidden mortality” associated with CDH is still present. A population-based database is needed to establish the benchmarking for management of CDH. J Pediatr Surg 39:661-665. © 2004 Elsevier Inc. All rights reserved.

Results: Twenty-four CDH-associated deaths were registered in Canada in 1996. Fourteen of 24 occurred in Ontario (58.3%). Of 30 institutionally identified CDH in Ontario, 8 patients died (26.7%). CDH-associated infant mortality rate was 6.6 of 100,000 live births in Canada compared with 10 of 100,000 live births for Ontario (Relative risk, 1.4; confidence

C

ONGENITAL DIAPHRAGMATIC hernia (CDH) occurs in 1 per 2,000 to 4,000 live births and is associated with significant morbidity and mortality.1-5 Approximately 40% to 50% of CDH patients have associated anomalies, most commonly involving the cardiac system, followed by the genitourinary, gastrointestinal, and central nervous systems.6 Although chromosomal anomalies frequently occur, the defects involving the cardiac system have been shown previously to be more important in terms of prognosis.6,7 What was once believed to be a simple mechanical defect with the loss of domain for organs in one body compartment is now evolved into a much more complex clinical challenge, representing interactions of biological and physiological factors leading to pulmonary hypoplasia, pulmonary vascular abnormality, and a high, yet varied, mortality rate.1,4,5 Outcome and management strategies continue to vary widely between institutions, as there is no common protocol in managing these complex patients, including the use of such interventions as prenatal tracheal occlusion, inhaled nitric oxide, extracorporeal membrane oxygenation, and high-frequency oscillation.3,8,9 Although there is a general trend toward a delayed repair of CDH after a period of stabilization, the use of delayed versus early surgical correction remains to be clearly defined.5,10 Despite the lack of a standardized approach to CDH, the overall survival rate has increased from approxiJournal of Pediatric Surgery, Vol 39, No 5 (May), 2004: pp 661-665

INDEX WORDS: Congenital diaphragmatic hernia, mortality rate, population-based data.

mately 50% to 60% to 75% to 90% in the last 5 to 7 years.11-13 However, this apparent improvement is not consistently observed across all institutions.1,4,14 Thus, whether this represents a “true” improvement or whether it is a result of changing patient populations, variability in hospital-based practice patterns, selection bias, or other factors, remains to be verified. In this study, we sought to determine if clinical outcomes based on institution-based reporting accurately reflect a true improvement, and what factors have led to such variability between institutions. In addition, institutional differences in clinical outcome were also examined, as were varia-

*Leslie Scott, London Health Sciences Center, London Ontario; Douglas Mah, Brian Cameron, McMaster University, Hamilton, Ontario; Noelle Grace, North York General Hospital, Toronto, Ontario; Juan Bass, Children’s Hospital of Eastern Ontario, Ottawa, Ontario; Dan Panaeru, Queen’s University, Kingston, Ontario, and Peter Masiokos, Desmond Bohn, Paul Wales, and Peter C. W. Kim, Hospital for Sick Children, Toronto, Ontario, Canada. Presented at the 35th Annual Meeting of the Canadian Association of Paediatric Surgeons, Niagara-on-the-Lake, Ontario, Canada, September 18-21, 2003. Address reprint requests to Peter C.W. Kim, Hospital for Sick Children, 555 University Ave, Room 1526, Toronto, M5G 1X8, Ontario, Canada. © 2004 Elsevier Inc. All rights reserved. 0022-3468/04/3905-0002$30.00/0 doi:10.1016/j.jpedsurg.2004.01.032 661

662

SCOTT ET AL

Table 1. Comparison of Patient Characteristics

No. of cases M:F Age of birth (mean) Hernia (right sided) Hernia (left sided) 5-min calculated apgar (mean) Birth weight (mean) Associated anomalies

Current Study

Wilson et al (1997)4

Azarow et al (1997)1

Boloker et al (2002)2

30 1.7:1 38.5 20.0% 80.0% 6.7 3097.6 33.0%

196 — — 13.8% 86.2% 6.1 — 39.0%

223 — 38.7 15.2% 84.8% 6.1 3048 16.0%

120 — 37.9 17.5% 80.8% 7.1 3002 19.2%

tions in clinical management, specifically the use of permissive hypercapnea. MATERIALS AND METHODS The Bureau of Vital Statistics has compiled a database of postnatal deaths beyond 20 weeks’ gestation, causes of death, and live births in Canada. Using these statistics, the number of CDH-associated deaths in Canada and Ontario was recorded for the year 1996. Approximately 40% of the national population resides in this geographically defined area with a pediatric population of 2.6 million (based on 1996 Census data). Deaths related to CDH per 100,000 live births were also noted. The Ontario Congenital Anomalies Study Group, which consists of pediatric surgeons and intensivists representing all 5 pediatric tertiary care centers in the province, was formed to develop and to promote multiinstitution-based and population-based studies of various congenital anomalies. CDH cases in all pediatric institutions in Ontario were then reviewed for the year 1996. Hospital records were used to determine patient demographics, Apgar scores, the use of prenatal diagnosis, location of the diaphragmatic hernia, age at surgery, associated anomalies, age of death and autopsy reports, and the use of permissive hypercapnea with or without paralysis.

Statistical Analysis Statistical analysis of data was performed with SPSS for Windows (V.10) statistical programs. Continuous variables were compared using Student’s t test for independent variables. Categorical variables were analyzed using Fisher’s Exact test or ␹2 analysis with Yates correction. A P value of .05 was considered significant.

RESULTS

Patient Characteristics In 1996, a total of 30 cases of CDH were identified in all participating pediatric institutions in Ontario. These cases represent a complete and inclusive number for all CDH in the geographically defined area, the Province of Ontario. To determine whether these patients from the multiinstitutions were comparable with previously reported patients in the literature, we compared our patient characteristics with those in 3 recent large CDH case series (Table 1).1,2,4 Patient age and weight at birth, sidedness, calculated Apgar scores at 5 minutes, and associated anomalies at birth were compared (Table 1). The patient characteristics from our patient population from all pediatric institutions from the geographically defined area were comparable and similar to those in

previously reported series, except for the percentage of patients presenting with associated anomalies.1,2,4 Differences Between Case Series and PopulationBased Reporting Twenty-four CDH-associated deaths were registered in Canada for the year 1996 according to the Vital Statistics database (Table 2). Fourteen of the 24 occurred in Ontario (58.3%), including 9 boys and 5 girls. Of the total of 30 CDH cases reported by the Ontario institutions, there were 8 recorded deaths (26.7%) in the institution-based reporting. Therefore, 6 of the 14 deaths (42.9%) in Ontario were unaccounted for, clearly showing a major shortcoming in institution-based reporting. Given the fact that population-based data from the department of Vital Statistics represents an underestimation of true incidence of the disease, the “hidden incidences and mortality” associated with CDH does not appear to have changed significantly despite recent reports of improved survival rates.15 Interestingly and surprisingly, CDH-associated infant mortality rate was noted to be higher in the province of Ontario compared with the national rate (10 v 6.6 per 100,000 live births; RR ⫽ 1.4 [95% CI ⫽ 0.5 to 3.7]; (P ⬎ .01). Although this relative risk or odds ratio of approximately 30% to 40% higher risk of CDH-associated death in Ontario compared with national average was a consistent finding over a much longer period, only a general trend, not a statistical significance, was noted. The majority of the CDH deaths occurred in the neonatal period (⬍28 postnatal days), accounting for 13 of the 14 (92.9%) Ontario CDH deaths. This was true also for the whole country with 23 of 24 (95.8%) deaths occurring in the neonatal period. This appears to suggest Table 2. Institution Versus Population-Based Reporting of CDH Cases and Deaths

Total No. of CDH cases reported No. of CDH-associated deaths CDH deaths (M:F) No. of neonatal CDH deaths CDH deaths per 100,000 live births

Institution

Ontario

Canada

30 8 1.7:1 8 —

— 14 1.8:1 13 10

— 24 1.2:1 23 6.6

CDH OUTCOMES

663

Table 3. Survival Rate of CDH Patients Treated in Ontario Pediatric Referral Centers Institutions

Survival Rate (%)

A B C D Total

62.5 71.4 100 78.6 78.1

that 2 potential major factors in the attrition rate for CDH may be perioperative mortality and preterm termination. The survival rates between Ontario institutions were highly variable, and the overall survival rate based on institution-based reporting was approximately 78% (Table 3). The CDH survival rates between Ontario institutions were highly variable, ranging from 62.5% to 100%. The obvious explanation for this variability relates to differences in practice pattern, referral pattern, and, more importantly, patient volume, where numbers of CDH patients ranged from 1 to 14 in 1996. This observation further represents weakness in any institution-based reporting. Potential Factors Responsible for Improved Survival Rate Although a number of new technical advances have been introduced and advocated as responsible for improved survival rate in institution-based reporting, any comparative analysis of these variables has not been performed previously. To determine whether improved survival rates for CDH and the variability in relative survival rates among institutions were attributable to specific clinical practice pattern, risk factors and differences in management between institutions were analyzed. We examined the use of hypercapnea with or without paralysis, because this ventilatory modality has been suggested to be an important variable in recent improvements in CDH survival.2 No centers used permissive hypercapnea without paralysis. The use of hypercapnea with paralysis ranged from 37.5% to 100% in all participating institutions (Table 4). Both hypercapnea and paralysis were analyzed as independent variables in assessing the effect on survival. Relative risk analyses suggest that hypercapnea and paralysis may be protective Table 4. Effects of Hypercapnea With or Without Paralysis on Outcome of CDH Patients Hypercapnea

Survived

Died

Yes No Paralysis Yes No

10* 12

5 3

5† 10

3 5

*P ⫽ .682, RR ⫽ 0.83 (95% CI ⫽ 0.54 to 1.3). †P ⫽ .274, RR ⫽ 0.72 (95% CI ⫽ 0.54 to 0.96).

Table 5. Effect of Early Versus Late Surgery, Sidedness, and Associated Anomalies on Outcome of CDH Patients

Timing of surgery Late surgery Early surgery No surgery Sidedness R-sided L-sided Anomalies Associated anomalies No associated anomalies

Survived

Died

19* 3 0

3 1 4

3† 19

3 5

5‡

4

17

4

*P ⫽ .006, ␹ ⫽ 10.2. †P ⫽ .3, RR ⫽ 0.63 (95% CI ⫽ 0.79 to 7.3). ‡P ⫽ .195, RR ⫽ 2.3 (95% CI ⫽ 0.74 to 7.3). 2

by reducing the risk of mortality by approximately 17% (95 CI ⫽ 0.54 to 1.3) and 28% (95% CI ⫽ 0.54 to 0.96). The Fischer’s Exact test clearly indicates that both hypercapnea and paralysis did not significantly influence the survival rate (P ⫽ .682 and P ⫽ .274, respectively). Previous studies investigating the benefits of early versus late surgery (defined as surgery before or after 24 hours of life) have yielded conflicting results. Although there is an increasing preference to stabilizing the patient before operating, it remains controversial whether delaying surgical intervention decreases mortality. In this study, 22 patients underwent late surgery (73.3%), 4 had early surgery (13.3%), and 4 died before any surgical correction (Table 5). One would intuitively assume that the early group would be biased toward a better survival rate because patients who are clinically stable would be corrected earlier. Alternatively, the delayed group may represent a self-selected group with better survival rates. Interestingly, the delayed group had significantly better survival rates compared with early surgery or no surgery, favoring the latter hypothesis (P ⫽ .006). Conflicting studies also have been reported on whether the site of the diaphragmatic defect is a significant factor in survival rate.1,4,15,16 Although right-sided defects accounted for 20% of the diaphragmatic hernias in institution-based reporting in Ontario in 1996 and right side has a higher CDH-associated mortality rate, there was no statistical difference in mortality rate when compared with left-sided defects (P ⫽ .3). The high prevalence of associated anomalies with CDH has been documented in the past, with heart defects being most prevalent and having the worst prognosis. Previous trials have found a decreased survival rate in CDH patients with associated anomalies and it is a common confounding factor when comparing outcomes between institutions. Nine patients were found to have associated anomalies in this study, with 6 having congenital heart defects. The prevalence of associated anom-

664

SCOTT ET AL

alies varied between institutions, ranging from 0% to 38% of patients. Patients with other anomalies were found to have a lower survival rate (55%) in relation to patients with isolated CDH (81%). However, although there was 2-fold increased risk of death with the presence of associated anomaly (RR ⫽ 2.33 [95% CI ⫽ 0.743 to 7.332], the presence of associated anomalies in this study did not reach statistical significance (P ⫽ .195). DISCUSSION

Over the last decade, various single institutional– based case series have reported a significant improvement in the survival rate of patients with CDH compared with previous periods.1,2,4 This improvement has been attributed to various factors, including the use of inhaled nitric oxide, high-frequency oscillatory ventilation, extracorporeal membrane oxygenation, permissive hypercapnea, exogenous surfactant therapy, and corticosteroid therapy.3,8,9 Despite the plethora of major pediatric centers rushing to advertise their improved survival rates in both peer-reviewed and non–peer-reviewed media, many if not most studies lack critical comparisons of clinically relevant variables accountable for the improved outcome. These studies often fail to investigate CDH in the context of geographically defined populations over a set period to determine the true incidence, natural history, and true impact of treatment protocols on clinical outcomes. In this study, we attempted to look at these issues by comparing and analyzing all institution-based or population-based data from a geographically defined population base over a set period. We report here that there is persistent bias associated with institutional-based reporting and databases, and that there continues to be a significant “hidden mortality” associated with CDH. Institution-based reports, in general, report higher survival rates than those that are population based. Harrison et al17 first introduced the concept of “hidden mortality” that accounts for the group of patients who are not reported on by tertiary centers. Previous research has found that the “hidden mortality” of CDH can be determined, though not accurately measured, by populationbased reporting.15,16,18 In our study, the number of registered CDH-associated deaths from the Vital Statistic Bureau, which compiles population-based data, was compared with those reported by the Ontario pediatric centers for 1996. Of the 24 deaths in Canada, 14 occurred in Ontario, and only 57.1% of these deaths were accounted for by institutional-based reporting. The fact that the death of 43% of these potentially correctable anomalies occurred even before reaching these pediatric institutions raises a more difficult challenge of determining where and why these CDH-associated mortalities occur. A logical deduction or explanation, given the centralized health care system in Ontario, is preterm counseling and

termination. This is consistent with a prospective study by Harrison et al,15 which reported that 34% of CDH infants died in utero or soon after birth. The challenge for tertiary centers, therefore, is to educate and provide appropriate support for primary and secondary care providers if the shared goal is to truly improve CDHassociated outcome. Of interest, population-based reporting found that the province of Ontario had a higher number of CDHassociated deaths compared with the national average in 1996 (10 v 6.6 per 100,000 live births). This was not an isolated finding because a longer term analysis over 8 years showed a trend toward higher provincial CDHassociated mortality in the geographically defined area compared with the national averages. The exact reason for this finding is not clear at the moment. However, it is hypothesized that the increased mortality rate in Ontario can be attributed, at least in part, to the large variation in geographic and institutional referral practices. Past studies have warned of misinterpreting variations in outcome owing to the presence of case selection artifacts19,20; hence, the differences in survival rate may simply be a result of higher-risk patients within certain centers because of the availability of various treatment modalities, such as extracorporeal membrane oxygenation (ECMO). In terms of “visible” mortality for CDH, there continues to be a wide range of survival rates between centers. Geographic differences may again play a role, as some hospitals may treat all patients within their surrounding area or may select for a certain category of patients. The survival rates in this study ranged from 62.5% to 100%. It is recognized that this wide range can in part be attributed to the varying number of patients seen per center, although other variables can play important roles in affecting survival rates. The lack of consensus as to how CDH should be managed is a vital factor. Although a comparison of all treatment modalities currently available for CDH is beyond the scope of this study, the use of permissive hypercapnea was examined between institutions. On review of its use in Ontario, it was found that no center used the recommended permissive hypercapnea without paralysis,2 whereas the use of hypercapnea with paralysis ranged between 37.5% and 100%. However, this “gentle” ventilation strategy with permissive hypercapnea, which avoids overdistension of the lung and iatrogenic barotrauma, was not a significant variable in terms of improved survival rates. In addition, the avoidance of paralysis was not a significant variable in determining survival of CDH patients. The benefit of delayed surgery in the CDH population also continues to be of some controversy. Whereas emergency surgery was the standard of treatment in the early 1980s, as our understanding of CDH grew, so did the value of preoperative stabilization.10,14,21,22 Nonetheless,

CDH OUTCOMES

665

a Cochrane review5 noted that there was no clear evidence that favors early (before 24 hours of life) versus late surgery, although both the quantity and the quality of studies investigating this strategy of treatment is lacking. In contrast, our study points to a significant survival advantage toward delayed correction beyond 24 hours. Given the current trend toward surgery after a period of initial stabilization, it is not clear whether this finding represented a selection bias or the patients in the early surgery group were not truly stable. Our finding is consistent with recent recommendations supporting delayed repair after preoperative stabilization. However, the optimal timing of delayed repair, with and without ECMO, requires further analysis because studies have reported delays from 24 to more than 96 hours,23 with the CDH study group noting a mean age at surgery of 73 hours for patients not treated with ECMO.7,14

The current bias in institution-based reporting has led to misleading increases in survival rate that does not include the changing, yet still prevalent, “hidden mortality” of CDH. A population-based database that will record both the patients within the pediatric centers, as well as those that do not present to these institutions (stillborns, in utero death, or termination), should be created to provide an understanding of the true mortality in CDH and provide a benchmark to which future treatments will be compared. This study provides only a snapshot of the changing trends and outcomes of CDH over the past decade. A more extensive analysis of CDH data over the last 10 years should be performed to confirm if “hidden mortality” has changed over the years and what factors are most influential in determining the outcome of patients with congenital diaphragmatic hernias.

REFERENCES 1. Azarow K, Messineo A, Pearl R, et al: Congenital diaphragmatic hernia—A tale of two cities: The Toronto experience. J Pediatr Surg 32:395-400, 1997 2. Boloker J, Bateman DA, Wung JT, et al: Congenital diaphragmatic hernia in 120 infants treated consecutively with permissive hypercapnea/spontaneous respiration/elective repair. J Pediatr Surg 37:357-366, 2002 3. Bohn D: Congenital diaphragmatic hernia. Am J Respir Crit Care Med 166:911-915, 2002 4. Wilson JM, Lund DP, Lillehei CW, et al: Congenital diaphragmatic hernia—A tale of two cities: The Boston experience. J Pediatr Surg 32:401-405, 1997 5. Moyer V, Moya F, Tibboel R, et al: Late versus early surgical correction for congenital diaphragmatic hernia in newborn infants (Review). The Cochrane Database of Systematic Reviews 1, 2003 6. Fauza DO, Wilson JM: Congenital diaphragmatic hernia and associated anomalies: Their incidence, identification, and impact on prognosis. J Pediatr Surg 29:1113-1117, 1994 7. Congenital Diaphragmatic Hernia Study Group: Estimating disease severity of congenital diaphragmatic hernia in the first 5 minutes of life. J Pediatr Surg 36:141-145, 2001 8. Lally KP: Congenital diaphragmatic hernia. Curr Opin Pediatr 14:486-490, 2002 9. Smith NP, Jesudason EC, Losty PD: Congenital diaphragmatic hernia. Pediatr Respir Rev 3:339-348, 2002 10. Langer JC, Filler RM, Bohn DJ, et al: Timing of surgery for congenital diaphragmatic hernia: Is emergency operation necessary? J Pediatr Surg 23:731-734, 1988 11. Downward CD, Jaksic T, Garza JJ, et al: Analysis of an improved survival rate for congenital diaphragmatic hernia. J Pediatr Surg 38:729-732, 2003 12. Frenckner B, Ehren H, Granholm T, et al: Improved results in patients who have congenital diaphragmatic hernia using preoperative stabilization, extracorporeal membrane oxygenation, and delayed surgery. J Pediatr Surg 32:1185-1189, 1997 13. Reyes C, Chang LK, Waffarn F, et al: Delayed repair of

congenital diaphragmatic hernia with early high-frequency oscillatory ventilation during preoperative stabilization. J Pediatr Surg 33:10101016, 1998 14. Clark RH, Hardin WD, Hirshl RB, et al: Current surgical management of congenital diaphragmatic hernia: A report from the Congenital Diaphragmatic Hernia Study Group. J Pediatr Surg 33: 1004-1009, 1998 15. Harrison MR, Adzick S, Estes JM, et al: A prospective study of the outcome for fetuses with diaphragmatic hernia. JAMA 271:382384, 1994 16. Jaffray B, Mackinlay GA: Real and apparent mortality from congenital diaphragmatic hernia. Br J Surg 83:79-82, 1996 17. Harrison MR, Bjordal RI, Langmark F, et al: Congenital diaphragmatic hernia: The hidden mortality. J Pediatr Surg 13:227-230, 1978 18. Steinhorn RH, Kriesmer PJ, Green TP, et al: Congenital diaphragmatic hernia in Minnesota. Impact of antenatal diagnosis on survival. Arch Pediatr Adolesc Med 148:626-631, 1994 19. Evans SM, Jaffray B: Sample selection and the natural history of congenital diaphragmatic hernia: Evidence of systematic selection bias in reported survival. The 50th Annual International Congress of the BAPS, Oral presentation, Estoril, Portugal, July 15-18, 2003 20. Jaffray B, Stege G, Fenton A: Nihlism in the 1990s: The true mortality of congenital diaphragmatic hernia. The 50th Annual International Congress of the BAPS, Oral presentation, Estoril, Portugal, July 15-18, 2003 21. Sakai H, Tamura M, Hosokawa Y, et al: Effect of surgical repair on respiratory mechanics in congenital diaphragmatic hernia. J Pediatr Surg 111:432-438, 1987 22. Tracy TF Jr, Bailey PV, Sadiq F, et al: Predictive capabilities of preoperative and postoperative pulmonary function tests in delayed repair of congenital diaphragmatic hernia. J Pediatr Surg 29:265-269, 1994 23. Nio M, Haase G, Kennaugh J, et al: A prospective randomized trial of delayed versus immediate repair of congenital diaphragmatic hernia. J Pediatr Surg 29:618-21, 1994