Figure. Patients with uncomplicated diaphragmatic hernia admitted to US hospitals from 1999 to 2008 by procedures per 100 000 procedures.
Paul S, Nasar A, Port JL, et al. Comparative analysis of diaphragmatic hernia repair outcomes using the Nationwide Inpatient Sample database. Arch Surg. Published online March 19, 2012. doi:10.1001/archsurg.2012.127.
eTable. ICD-9-CM Codes Used for Procedures Type, Demographics, and In-hospital Complications of Patient with UDH Undergoing Repair from 1999 to 2008
eFigure 1. Proportion of Thoracic DH Repair Cases as Percentage of Total DH Repair Cases (Thoracic, Laparoscopic and Open DH Repair) by Hospital Truncated at Hospital Number 700
eFigure 2. Proportion of Laparoscopic DH Repair Cases as Percentage of Total Abdominal DH Repair Cases (Laparoscopic and Open DH Repair) by Hospital Truncated at Hospital Number 400
This supplementary material has been provided by the authors to give readers additional information about their work.
Paul S, Nasar A, Port JL, Lee PC, Stiles BC, Nguyen AB, Altorki NK, Sedrakyan A. Comparative Analysis of Diaphragmatic Hernia Repair Outcomes Using the Nationwide Inpatient Sample Database. Arch Surg. 2012;147(7):607-612. doi:10.1001/archsurg.2012.127
Author Affiliations: Division of Thoracic Surgery, Department of Cardiothoracic Surgery (Drs Paul, Port, Lee, Stiles, Nguyen, and Altorki and Mr Nasar) and Department of Public Health (Dr Sedrakyan), New York Presbyterian Hospital–Weill Cornell Medical College, New York.
Objective To determine the comparative effectiveness of various approaches to diaphragmatic hernia (DH) repair, including open abdominal, laparoscopic abdominal, and thoracotomy.
Design, Setting, and Patients Using the Nationwide Inpatient Sample from 1999 to 2008, a comprehensive cohort of 38 764 patients (mean [SD] age, 60.8 [19.5] years) hospitalized with a primary diagnosis of DH who underwent repair was identified.
Main Outcomes Measures Morbidity and mortality of patients who underwent DH repair.
Results Open approaches were the most common, performed in 91% of patients (open abdominal, n = 28 824 [74.4%]; thoracotomy, n = 6573 [17.0%]). Hospital mortality was 1.1% or less for each of the approaches. However, patients who underwent a laparoscopic DH repair had a shorter length of stay (mean [SD], 4.5 [0.10] days) and fewer discharges to skilled nursing facilities than those who underwent open abdominal or thoracotomy repair approaches. Patients who underwent a DH repair through a thoracotomy approach had the longest length of stay (mean [SD], 7.8 [0.11] days) and a higher need for postoperative mechanical ventilation than those undergoing open or laparoscopic abdominal approaches (5.6% vs 3.2% vs 2.3%, respectively; P < .001). In addition, the thoracotomy approach was found to be an independent predictor for the development of a pulmonary embolism.
Conclusions This large national study demonstrates that most DH repairs are performed through open abdominal and thoracic approaches. Laparoscopic approaches are associated with decreased length of stay and more routine discharges than open abdominal and thoracotomy approaches.
Diaphragmatic hernias (DHs) are responsible for 1 in every 2000 inpatient hospital admissions annually. As the population ages, this rate is likely to increase.1- 3 Surgery is the only treatment for DH and can be performed using an abdominal approach (open laparotomy or laparoscopic techniques) or through the left chest with a thoracotomy. While the location of the DH might be related to the choice of the technique, surgical training and convictions are often main factors for deciding on the technique as DHs can be accessed from either chamber of the trunk (abdomen or thoracic cavity). The comparative effectiveness evidence is very limited; in some small institutional case series, laparoscopic repair has been suggested to be superior to other approaches in terms of improved postoperative outcomes.4- 6 We sought to determine the comparative effectiveness of the various approaches to DH repair using one of the largest national databases, the Nationwide Inpatient Sample (NIS) database, representing 20% of all inpatient admissions in the United States. An additional goal of our study was to examine the variation in the choice of approach to DH repair.
The primary source of the cohort for this study was the 1999-2008 NIS, developed by the Agency for Healthcare Research and Quality.7 The NIS is the largest all-payer inpatient care registry in the United States. It contains data on more than 8 million hospital stays from approximately 1000 hospitals and represents 20% of all hospital discharges from nongovernment institutions. Data contained within the NIS include patient and hospital demographic characteristics, admission and treating diagnoses, inpatient procedures, in-hospital mortality, length of hospital stay, hospital charges, and discharge status. The NIS data set has numerous internal quality measures and is validated by comparing other similar efforts such as the National Hospital Discharge Survey and the Medicare Provider Analysis and Review. This study was approved by the institutional review board of Weill Cornell Medical College and conforms to the data use agreement.
The study population consists of patients whose principal admitting diagnosis was uncomplicated DH (types I, II, and III paraesophageal hernia; International Classification of Diseases, Ninth Revision [ICD-9 ] code 553.3) from 1999 to 2008. Patients with DH obstruction (ICD-9 code 552.3) or with gangrene (ICD-9 code 552.3) or congenital hernias (ICD-9 codes 750.6 and 756.6) were excluded. Specific DH repair procedures were identified using the following ICD-9 Clinical Modification (ICD-9-CM) procedure codes: 53.7, 53.72, and 53.75 (open abdominal approach), 53.71 (laparoscopic abdominal approach), and 53.80, 53.83, and 53.84 (thoracotomy approach). As no specific codes existed for laparoscopic DH repair prior to 2008, ICD-9-CM procedure codes 54.21 and 54.51 in conjunction with 53.7 if listed together on the same procedure day were used together to identify laparoscopic procedures as described previously.8 The ICD-9-CM codes used for patient identification are summarized in the eTable). Hospital volume of DH repair cases was organized into quartiles based on the total number of DH repair cases performed during the study period regardless of approach.
Data collected for each patient admission included age at the time of admission, sex, patient comorbidities, in-hospital complications, in-hospital mortality, length of stay, discharge status, and overall hospitalization cost. Patient comorbidities and in-hospital complications were identified from the patient's ICD-9-CM diagnosis codes. Most patient comorbidities were recorded using the NIS comorbidity software version 3.6 format as summarized in the eTable.9
The primary end point of this study was comparison of the in-hospital mortality and morbidity of patients with uncomplicated DH who underwent repair by open abdominal, laparoscopic abdominal, or thoracotomy approaches. Secondary end points included trends in DH repair.
We had a multistep approach to our analyses. In a first stage, a descriptive analysis of baseline characteristics and outcomes was performed. In a second stage, the patients who underwent operations using different approaches were compared with regard to baseline characteristics, comorbid conditions, and surgical procedure–related variables using t tests and χ2 tests (bivariate analysis). In a third stage, a multivariable model was constructed. The variable selection for the multivariable analyses was based on peer-reviewed literature and clinical judgment. Multivariable logistic regression analysis was used to determine independent risk factors for mortality or composite mortality-morbidity end points. The variables considered in the model included the patient age, coronary artery disease, congestive heart failure, diabetes, chronic pulmonary disease, peripheral vascular disease, hospital setting, DH repair approach, and hospital DH repair volume. Colinearity between independent variables was assessed using κ > 0.4. No variables tested were found to be colinear. Methods that adjust for intracluster correlations (multicenter data coming from different hospitals were considered as clustered data) were used at the final stage of analysis (multilevel model). SUDAAN version 9.0.1 software (Research Triangle Institute) was used to convert raw counts generated from the NIS database into weighted counts that were used to generate national estimates. Weighted data were used for all statistical analysis. To test for statistical significance, χ2 analysis was used for categorical data and analysis of variance was used for continuous data. All P values were 2-sided with statistical significance evaluated at the α = .05 level. Bonferroni adjustment was considered when making multiple comparisons.
During the 10-year period of this study, 38 764 patients underwent DH repair as the principal procedure for their admission. Most repairs were through an open abdominal or open thoracic approach in each year of the study period (Figure). Laparoscopic abdominal and open thoracotomy approaches in comparison with open abdominal approaches were most frequently performed in urban teaching hospitals (Table 1 and Table 2).
The baseline characteristics of the 38 764 patients who underwent DH repair are shown in Table 1 and Table 2. We compared open abdominal approaches with either laparoscopic abdominal approaches or open thoracotomy approaches in 2 separate tables. Patients were predominantly white and female with a mean (SD) age of 60.8 (19.5) years. Compared with patients who had an open thoracic repair, those who underwent an abdominal approach, either open or laparoscopic, were slightly older (mean [SD] age, 57.4 [0.23] years for open thoracic repair vs 61.5 [0.12] years for open abdominal approach [P < .001] and 62.0 [0.28] years for laparoscopic abdominal approach [P < .001]). Patients who underwent open abdominal repair were more likely to have associated coronary artery disease than those undergoing laparoscopic abdominal or open thoracic repair (Table 1 and Table 2). Similarly, patients undergoing an open thoracic approach were more likely to have a chronic pulmonary disease than patients undergoing an abdominal approach (Table 1 and Table 2).
Operative mortality was low for all 3 repair approaches and not significantly different between the approaches (open abdominal, 1.1%; laparoscopic abdominal, 0.6%; open thoracic, 1.1%). Compared with patients undergoing open thoracic repair, those who underwent DH repair by an abdominal approach, whether open or laparoscopic, were less likely to need postoperative mechanical ventilation (5.6% for open thoracic repair vs 3.2% for open abdominal repair [P < .001] and 2.3% for laparoscopic abdominal repair [P < .001]) (Table 3 and Table 4). Patients undergoing a thoracic DH repair also had a higher occurrence of pulmonary embolism compared with those undergoing an open abdominal repair (1.0% vs 0.4%, respectively [P = .03]) (Table 4). Finally, patients undergoing laparoscopic abdominal repair had lower rates of postoperative urinary tract infections compared with those undergoing open abdominal repair (1.7% vs 2.9%, respectively [P = .03]). No differences were noted among the 3 DH repair approaches in rates of postoperative pneumonia, deep venous thromboembolism, myocardial infarction, or sepsis (Table 3 and Table 4).
Patient who underwent a laparoscopic abdominal repair had the shortest length of stay compared with patients who underwent either open abdominal or open thoracic repair (4.5 days for laparoscopic abdominal repair vs 5.9 days for open abdominal repair [P < .001] and 7.8 days for open thoracic repair [P < .001]) (Table 3 and Table 4). Laparoscopic abdominal repair was also associated with fewer discharges to skilled nursing facilities than open abdominal or thoracotomy repair approaches (Table 3 and Table 4).
To evaluate the extent of confounding by indication, we performed an analysis of the variation in the choice of abdominal approaches (both open and laparoscopic) or thoracic approach at each center. As shown in eFigure 1, most centers performed abdominal approaches or thoracic approaches almost exclusively. Similarly, an analysis comparing the percentages of the open abdominal approach vs both the open and laparoscopic abdominal approaches at each center was performed. Most centers performed open abdominal or open thoracic approaches almost exclusively (eFigure 2).
To account for differences in the baseline characteristics of the patients undergoing DH repair with respect to outcomes, a multivariable analysis accounting for patient-specific factors as well as hospital characteristics was performed (Table 5 and Table 6). The thoracotomy approach was found to be an independent predictor for the development of a pulmonary embolism and the composite end point (mortality, myocardial infarction, and pulmonary embolism) in the multivariable models (Table 5 and Table 6). No other differences between the repair approaches were found for mortality and serious morbidity.
Age and presence of congestive heart failure were predictors of mortality, myocardial infarction, and the composite end point in all comparisons (Table 5 and Table 6). Coronary disease was a predictor of myocardial infarction. Finally, age was found to be a predictor of pulmonary embolism. No other factors were related to any end points.
In this largest comparative study to date, we found that DHs continue to be repaired mostly through open techniques despite the introduction of laparoscopic techniques more than 2 decades ago. We found that fewer complications and substantially shorter lengths of stay were associated with laparoscopic abdominal DH repair.
Thoracotomy approaches for DH repair are very frequent and associated with higher chances of pulmonary embolism, need for mechanical ventilation, and longer length of stay compared with open or laparoscopic abdominal repair approaches. The increased period of immobilization from requiring mechanical ventilation could potentially predispose the patient to deep venous thrombosis and pulmonary embolism and could at least partially explain this finding. While we did not find substantial differences in mortality, the advantages of laparoscopic and then open abdominal approaches are difficult to ignore.
As the NIS database does not describe the location and extent of the DH, a valid concern is that our results are confounded by these 2 factors. However, the differences in outcome do not appear to be a result of confounding by indication. The conventional wisdom that larger, more difficult DHs should be approached by thoracotomy or open abdominal approaches was not supported when we used a simple but innovative analytical approach to evaluate confounding by indication. We evaluated variation in the choice of the technique and have shown that in the vast majority of centers only 1 approach is predominantly performed. This supports our hypothesis that there is very limited if any confounding by indication related to DH location as all DHs are approached in 1 uniform fashion within each center. Only a fraction of the centers examined used all 3 approaches. This also illustrates that the choice of open thoracic, open abdominal, or laparoscopic abdominal DH repair is more likely to be related to technical familiarity than to DH and patient attributes. We advocate this method for comparative effectiveness evaluations as a valid and unique way to assess confounding by indication.10,11
While to our knowledge this is the first comparative effectiveness study, it is supported by several prior reports in the literature comparing open abdominal DH repair with laparoscopic abdominal or open thoracotomy approaches.1,3- 6,12- 16 Most of these studies have shown that DH surgery is safe and that laparoscopic DH repair has been associated with shorter length of stay and earlier recovery.3,4,6 Our study also extends prior findings related to reduced morbidity and expedited functional recovery associated with the laparoscopic approach.3,4,6
We recognize that there are several limitations to this analysis. Our study is limited to only short-term inpatient mortality and outcomes. In the NIS cohort, we were not able to study long-term outcomes or the functional status after surgery. Hence, quality-of-life measures describing postoperative recovery such as the nature and duration of pain and return to daily activities are not known for each approach. Laparoscopic abdominal DH repair, although not vastly superior in terms of postoperative morbidity or mortality, may be better than open abdominal and thoracic DH repair approaches in this regard. Patients with laparoscopic DH repair were more likely to be discharged home with less need for support in skilled nursing facilities. Moreover, laparoscopic DH repair may be associated with less morbidity in special populations such as elderly patients or those with severe comorbidities. Alternatively, policies for early home discharge may be characteristic of hospitals performing laparoscopic DH repair. As only a handful of hospitals perform all 3 DH repair approaches, no meaningful comparison can be made to analyze this. The ICD-9-CM procedure code for laparoscopic DH repair became available in 2008. For the data prior to 2008, we coded open abdominal DH repair and laparoscopy as a laparoscopic procedure if they were both available. This assumption may not be valid in all cases. However, there is a small increase in the number of cases in 2008 associated with the presence of the new laparoscopic code, and the full benefits of laparoscopic DH repair would be underestimated. Long-term DH recurrence rates are not known for each approach. Long-term studies are needed to address this limitation. Finally, the specialty of the operating surgeon is not known within our cohort, which may be an important determinant of quality of care in addition to patient and hospital factors.
In conclusion, our analysis of the NIS database demonstrates that most DH repairs in the United States continue to be performed using an open abdominal or thoracic approach rather than a laparoscopic abdominal approach. While mortality and morbidity for all 3 DH repair approaches are low, laparoscopic approaches are associated with better short-term postoperative outcomes.
Correspondence: Subroto Paul, MD, Division of Thoracic Surgery, Department of Cardiothoracic Surgery, New York Presbyterian Hospital–Weill Cornell Medical College, 525 E 68th St, M404, New York, NY 10065 (firstname.lastname@example.org).
Accepted for Publication: December 20, 2011.
Published Online: March 19, 2012. doi:10.1001/archsurg.2012.127
Author Contributions: Drs Paul and Sedrakyan and Mr Nasar had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Paul, Port, Stiles, Nguyen, Altorki, and Sedrakyan. Acquisition of data: Paul, Nasar, Nguyen, and Sedrakyan. Analysis and interpretation of data: Paul, Lee, and Sedrakyan. Drafting of the manuscript: Paul, Nguyen, Altorki, and Sedrakyan. Critical revision of the manuscript for important intellectual content: Paul, Nasar, Port, Lee, Stiles, Altorki, and Sedrakyan. Statistical analysis: Paul, Nasar, Nguyen, and Sedrakyan. Obtained funding: Altorki. Administrative, technical, and material support: Paul, Lee, Stiles, Nguyen, Altorki, and Sedrakyan. Study supervision: Paul, Port, and Sedrakyan.
Financial Disclosure: None reported.