The effect of hospital volume of pediatric appendectomies on the misdiagnosis of appendicitis in children

The effect of hospital volume of pediatric appendectomies on the misdiagnosis of appendicitis in children

Douglas S. Smink

Appendicitis is common in children, but it is often misdiagnosed. Of the >70 000 pediatric appendectomies performed annually in the United States, (1) as many as 24% are negative appendectomies, revealing a normal appendix. (2-5) Such rates of misdiagnosis are accepted because of the common belief that a reduction in the incidence of misdiagnosis results in an increase in appendiceal perforation and its complications. (6) Furthermore, misdiagnosis is accepted because many clinicians consider a negative appendectomy to have few harmful effects. Some studies now question the inverse relationship between these 2 outcomes, suggesting that misdiagnosis can be improved without affecting the risk of perforation. (7-9) In addition, recent research shows that the misdiagnosis of appendicitis has considerable clinical and economic costs. (10)

Studies have demonstrated an association between higher hospital volume and improved patient outcome for certain surgical procedures. (11-13) Based on this research, the Leapfrog Group, a consortium of public and private purchasers of health insurance, encourages patients to be treated at high-volume centers for certain procedures, (14) Although most volume-outcome research has focused on adult procedures, recent investigations have suggested a similar relationship between volume and outcome in pediatric conditions. (15-18) Nonetheless, the effect of hospital volume on outcome is not known for the majority of pediatric surgical procedures. In this study we test the hypothesis that higher hospital volume of pediatric appendectomies is associated with lower rates of misdiagnosis of appendicitis in children.


We conducted a retrospective cohort study of children undergoing an appendectomy for suspected appendicitis using data from the Healthcare Cost and Utilization Project Kids’ Inpatient Database (KID). (19) The KID is a pediatric discharge database from 2521 community- hospitals in 22 states from January 1 to December 31, 1997; all patients in the database were [less than or equal to] 18 years old. From each hospital in the database, the KID contains a random sample of 80% of pediatric nonbirth discharges. For each discharge, the database includes up to 15 diagnosis and 15 International Classification of Diseases, Ninth Revision (ICD-9) procedure codes, with the first code representing the principal diagnosis or procedure. (19)

The study cohort consisted of children 1 to 18 years old with a principal ICD-9 procedure code for nonincidental appendectomy: 47.0 (appendectomy), 47.01 (laparoscopic appendectomy), or 47.09 (other appendectomy). (20) Patients undergoing an incidental appendectomy (an appendectomy at the time of another procedure) were identified with separate ICD-9 codes and were not included in the analysis. Two patients with no gender reported were excluded from the analysis. For other variables with missing data, we included a separate category indicating missing values. Analysis of race, the most commonly unreported variable, showed that 91% of the 7882 patients with unreported race were treated at hospitals that did not report race for any of their patients. The characteristics of these hospitals were similar to the other hospitals in our sample, so patients with missing race were included. In addition, we assessed the sensitivity of our final results by repeating the analysis with patients with missing race excluded.

A case of appendicitis was defined by a diagnosis code for appendicitis: 540.0 (acute appendicitis with generalized peritonitis), 540.1 (acute appendicitis with peritoneal abscess), 540.9 (acute appendicitis without mention of peritonitis), 541 (appendicitis, unqualified), and 542 (other appendicitis). (20) We assumed that a patient having one of these discharge diagnosis codes was diagnosed correctly with appendicitis at the time of surgery. The remaining appendectomy patients did not have a discharge diagnosis code for appendicitis. These patients were classified as misdiagnosed in accordance with previous studies. (21,22) Misdiagnosis was the primary outcome in our analysis.

For every hospital in the database, we determined the number of nonincidental pediatric appendectomies performed in 1997. Because the KID contains a random 80% sample of all nonbirth discharges, the estimated annual volume of appendectomies was calculated for each hospital by multiplying the KID volume of appendectomies by 1.25. Hospital volume was analyzed as both a continuous and a categorical variable; because both approaches yielded similar results, only the categorical results are presented. Hospitals were categorized into 5 mutually exclusive volume groups based on the estimated annual hospital volume of pediatric appendectomies: lowest (<1 appendectomy per month), low (at least 1 appendectomy per month but <1 per week), medium (at least 1 but <2 appendectomies per week), high (at least 2 but <3 appendectomies per week), and highest ([greater than or equal to] 3 appendectomies per week). We selected these categories to represent the spectrum of hospital frequency of pediatric appendectomies before analysis of hospital characteristics or misdiagnosis rates. Highest-volume hospitals were used as the reference group.

We considered patient age, gender, race, zip code median income, insurance payer, and transfer from another hospital as covariates in our analysis. We categorized these variables as shown in Table 1. The treating hospital’s classification as a teaching hospital was considered in our analysis as well. We report children’s hospital status but did not include this variable in the bivariate or multivariate analyses. Patient characteristics were compared among hospital volume levels by using analysis of variance and Cochrane-Armitage tests of trend.

We developed a logistic regression model at the patient level to determine the predictors of misdiagnosis and used generalized estimating equations to control for the effect of clustering within hospitals. (23,24) Covariates significant at the 0.10 level in bivariate analyses were entered in the multivariate model in a stepwise manner, interaction terms for age and gender as well as race and volume were considered also. Interaction terms allow the association between an independent variable and the outcome to differ at each level of another variable. For instance, we postulated that the effect of gender on misdiagnosis might differ in each age group. Variables and interaction terms significant at the 0.05 level in the multivariate analysis were included in the final model. Although it was not statistically significant in the multivariate analysis, we retained insurance status in the final model because of the broad literature on differences in care by insurance. All statistical analyses were performed using SAS 8.2 (SAS Institute Inc, Cary, NC).

The research protocol was approved as exempt from review by the institutional review boards of Children’s Hospital Boston and Harvard Pilgrim Health Care.


Our sample included 37 109 children 1 to 18 years old who underwent a nonincidental appendectomy in 1997. The characteristics of these patients are presented in Table 1. The mean age was 12.0 years, and 58.5% of the patients were male. Of patients with a recorded race, the majority (51.3%) were white, 5% were African American, and 17.5% were Hispanic. Almost a quarter (22.5%) were insured by Medicaid, and 13 411 (36.1%) were treated at a teaching hospital. Few (1.8%) of the patients were transferred from another hospital before appendectomy.

Of the 37 109 patients, 3103 (8.4%) were misdiagnosed (Table 1). The most common principal discharge diagnoses for patients who were misdiagnosed include other abdominal symptoms or conditions: unexplained abdominal pain (35.1%), mesenteric adenitis (22.8%), lymphoid hyperplasia (10.6%) or other diseases of the appendix (9.9%), gastroenteritis (4.4%), and ovarian cyst (3.3%).

Patient and hospital characteristics varied among volume categories (Table 2). The 24 children’s hospitals ranged in volume from low to highest, with 8 (33%) in the highest-volume category. Teaching hospitals were represented in all hospital volume categories. The mean age of patients decreased with increasing hospital volume levels (P for trend < 0.001), whereas the percentage of Medicaid-insured patients increased (P for trend < 0.001).

Table 3 describes the distribution of hospital volume of pediatric appendectomies. The 37 109 appendectomies were performed at 2217 hospitals, with the estimated annual hospital volume ranging from 1 to 302. Of these hospitals, 1060 (47.8%) were lowest-volume; they performed only 4786 (12.9%) of the appendectomies. Low-volume hospitals constituted 974 (43.9%) of the hospitals and performed 19 869 (53.5%) of the appendectomies. In contrast, the high- and highest-volume hospitals comprised only 21 (0.9%) and 13 (0.6%) hospitals, respectively. Collectively, these hospitals performed >11% of the total appendectomies. Of note, children’s hospitals comprised 8 (62%) of the 13 highest-volume hospitals.

The unadjusted odds ratios (ORs) suggest an association between hospital volume and misdiagnosis. Before controlling for patient or hospital characteristics, patients treated at lowest- (OR: 1.9; 95% confidence interval [CI]: 1.2-3.0; P < .01), low- (OR: 1.9; 95% CI: 1.2-3.0; P < .01), and medium- (OR: 1.7; 95% CI: 1.1-2.7; P = .03) volume hospitals had a significantly increased likelihood of misdiagnosis compared with patients treated at highest-volume hospitals (Table 3). Misdiagnosis at high-volume hospitals (OR: 1.6; 95% CI: 0.9-2.7; P = .09) was also more common than at highest-volume hospitals, but this difference was not statistically significant.

In addition to hospital volume, the final adjusted model included gender, age, gender/age interaction, race, and insurance status (Table 4). The effect of gender on misdiagnosis differed across age groups. Compared with 15- to 18-year-old males, 10- to 14- and 15- to 18-year-old females had significantly increased odds of misdiagnosis (OR: 2.3; 95% CI: 2.0-2.6; P < .01 and OR: 3.3; 95% CI: 2.9-3.8; P < .01, respectively). In the 1- to 4- and 5- to 9-year-old age groups, both males and females had an increased likelihood of misdiagnosis compared with the reference group, 15- to 18-year-old males. Race also had a significant effect, with misdiagnosis less common in Hispanics (OR: 0.57; 95% CI: 0.50-0.65; P < .01) and patients of other minority races (OR: 0.66; 95% CI: 0.54-0.81; P < .01) compared with whites. In this analysis, insurance status did not have a significant effect on misdiagnosis.

After adjusting for patient gender, age, gender/ age interaction, race, and insurance status, lower hospital volume was associated with an increased likelihood of misdiagnosis (Table 4). Patients treated at lowest- (OR: 1.5; 95% CI: 1.0-2.2; P < .05) and low-(OR: 1.6; 95% CI: 1.1-2.3; P = .02) volume hospitals were significantly more likely to be misdiagnosed than those treated at highest-volume centers. The results for medium- (OR: 1.5; 95% CI: 1.0-2.2; P = .06) and high- (OR: 1.4; 95% CI: 0.9-2.2; P = .17) volume hospitals were similar to those of lower-volume institutions but were not statistically different from highest-volume hospitals. Results were not altered substantially when patients with missing race were excluded.


Numerous studies have demonstrated that higher hospital volume of surgical procedures is associated with improved patient outcomes. (11-13,15,16) Most of these studies have focused on treatment of complex diseases rather than diagnosis of a seemingly low-complexity condition such as appendicitis. This analysis of a large, nationally representative pediatric database suggests that the number of pediatric appendectomies performed at a hospital is associated with the rate of misdiagnosis of appendicitis. After controlling for patient characteristics, patients treated for suspected appendicitis at lowest- and low-volume hospitals were approximately 50% more likely to be misdiagnosed than those treated at highest-volume centers. Medium- and high-volume hospitals showed trends toward increased risk of misdiagnosis as well.

This study also provides useful insight into the distribution of pediatric appendectomies among hospitals. Of the 2521 hospitals included in the KID, 2217 performed at least I pediatric appendectomy in 1997. Most (91.7%) of these hospitals were lowest- or low-volume by our classification, performing <1 pediatric appendectomy per week. However, these lowest- and low-volume hospitals performed 66% of all appendectomies. Thus, the majority of pediatric appendectomies occur at hospitals that perform them infrequently. Although patient and hospital characteristics varied among hospital volume levels, our analysis was designed to adjust for these factors.

In addition to showing that higher hospital volume is associated with a lower risk of misdiagnosis, this study demonstrates the importance of patient characteristics as well. Infants and young children, who might not be able to describe their symptoms, have an increased risk of misdiagnosis, as do adolescent females, in whom the diagnosis can be complicated by gynecological and obstetric conditions. We found that Hispanics and patients of other minority races have significantly decreased odds of misdiagnosis. One possible explanation for these results is that cultural beliefs, language barriers, or limited access to medical care cause patients of minority race to seek care later in the course of their disease. At advanced stages, appendicitis is generally easier to diagnose, making misdiagnosis less common in this group. Because the KID includes only 22 states, the racial and ethnic distribution in our sample does not represent the entire US population, which may affect our ability to generalize the race and ethnicity results. Further work in other patient samples is needed to confirm and clarify these results.

Ours is the first investigation to focus primarily on the relationship between hospital volume and misdiagnosis of pediatric appendicitis. Although our conclusions contrast with a previous study, hospital volume was not a focus of that analysis. (4) Here we contribute data from a large, nationally representative sample of pediatric patients and hospitals. We were able to control for many of the patient characteristics that influence misdiagnosis. Our volume categories were selected to represent a natural breakdown of hospital frequency of appendectomies and to be reproducible for use in future analyses. These categories did not result in the separation of children’s or teaching hospitals into any single volume category.

As with any study using administrative data, this analysis has limitations. We did not have final pathology reports to confirm the presence or absence of appendicitis. To be consistent with the existing literature, we used the ICD-9 code definitions of appendicitis and misdiagnosis that have been described previously. (21,22) Patients with a normal appendix could have been miscoded as having appendicitis or vice versa. Moreover, we assumed that all patients with an ICD-9 code for a nonincidental appendectomy were treated for suspected appendicitis. Some appendectomies may have been performed for another purpose or as part of another procedure. In an attempt to minimize the miscoding of procedures, we modified the definition of appendectomy used by Addiss et al (21) and Flum et al (22) to include only patients with a principal procedure of nonincidental appendectomy excluding those for whom appendectomy was listed as a secondary procedure.

We were surprised somewhat by the low misdiagnosis rate in our sample, which may differ from studies that used sources other than claims data. Because of the potential sources of misclassification, the absolute misdiagnosis rate in our analysis may not reflect the misdiagnosis rate in the pediatric population precisely. Nonetheless, the misdiagnosis rate seen here was within the range of published series, (2-5) and the principal diagnoses of misdiagnosed patients were consistent with clinical expectations. Our main objective was to compare relative rates of misdiagnosis across hospitals. There is no reason to expect that the miscoding of diagnoses and procedures is more common at any hospital volume level. In fact, no systematic coding errors were detected in an audit of pediatric coding in New York. (16) As long as systematic coding errors do not exist, comparisons of misdiagnosis across hospitals should be valid.

Other factors such as the volume of appendectomies per surgeon, children’s hospital status, or hospital location or region could have contributed to the diagnostic accuracy seen at highest-volume hospitals. Our goal was to focus on hospital volume, the major component of recent quality initiatives, (14) rather than other characteristics that could be associated with volume. Although children’s hospitals may possess specific qualities that improve diagnostic accuracy, we were unable to control for children’s hospital status in our model due to colinearity with volume. We did not control for volume of procedures per surgeon, because these data were not available in the KID. In addition, our hospital volume calculations did not include adult appendectomies. Some of our low-volume hospitals may have performed large numbers of adult appendectomies. Because of limitations of the database, hospital volume of adult appendectomies was not available. We believe, however, that because of the differences between children and adults, pediatric appendectomy volume is the most relevant volume measure to associate with the misdiagnosis of pediatric appendicitis.

We used 1997 data, the most recent data available at the time of our analysis. Increased utilization of imaging studies and clinical algorithms could have resulted in changing diagnostic strategies in recent years, suggesting the need for similar analyses as more current data become available.

A hypothetical application of the diagnostic accuracy of highest-volume hospitals to patients treated at lowest- and low-volume hospitals yields interesting insights. Of the 24 655 children treated at lowest-and low-volume hospitals, 2187 were misdiagnosed. Our data suggest that if the diagnostic accuracy seen at highest-volume hospitals were available at lowest- and low-volume institutions, only 1453 of those children would be expected to have been misdiagnosed, implying that 734 unnecessary appendectomies could have been avoided. Of note, improved diagnostic accuracy at highest-volume hospitals may be reflected in increased costs. This must be weighed against the clinical and economic costs that have been associated with misdiagnosis. (10)

We are not recommending that all children with appendicitis be treated at highest-volume centers. The highest-volume hospitals do not have the capacity to treat these additional patients, and many patients do not have timely access to a highest-volume hospital. We do not know whether the lower rates of misdiagnosis at highest-volume hospitals are a result of differences in clinical skills, diagnostic testing patterns, or other processes of care. A previous study showed that care in a group of hospitals can be improved by identifying the best clinical practices at member institutions. (25) Performing such an analysis of highest-volume hospitals could enable dissemination of their most successful processes of care, resulting in improved diagnosis of pediatric appendicitis at all hospital-volume levels.


The majority of pediatric appendectomies are performed at lowest and low pediatric volume hospitals. Patients treated at those hospitals have an increased likelihood of misdiagnosis compared with patients treated at highest pediatric volume hospitals. Identifying the successful practices at highest-volume hospitals could improve the diagnosis of appendicitis throughout the pediatric population.

TABLE 1. Characteristics of 37 109 Children Undergoing

a Nonincidental Appendectomy in 1997

All Appendectomies Misdiagnosis No.

No. (% of Total) * (% Misdiagnosed)

Total 37 109 (100.0) 3103 (8.4)


1-4 y 1633 (4.4) 146 (8.9)

5-9 y 8783 (23.7) 644 (7.3)

10-14 y 14 610 (39.4) 1114 (7.6)

15-18 y 12 083 (32.6) 1199 (9.9)


Female 15 408 (41.5) 1875 (12.2)

Male 21 701 (58.5) 1228 (5.7)


African American 1861 (5.0) 155 (8.3)

Hispanic 6491 (17.5) 327 (5.0)

Other 1835 (4.9) 110 (6.0)

White 19 040 (51.3) 1809 (9.5)

Missing 7882 (21.2) 702 (8.9)

1990 Zip Code Median Income ($)

0-25 000 9777 (26.3) 838 (8.6)

25 001-30 000 6943 (18.7) 594 (8.6)

30 001-35 000 5953 (16.0) 513 (8.6)

[greater than or equal to] 12 412 (33.4) 995 (8.0)


Insurance payer

Medicaid 8367 (22.5) 616 (7.4)

Other 4314 (11.6) 321 (7.4)

Private 24 294 (65.5) 2147 (8.8)

Transfer from another hospital

Yes 653 (l.8) 52 (8.0)

No 35 620 (96.0) 2983 (8.4)

Treated at teaching hospital

Yes 13 411 (36.1) 982 (7.3)

No 23 698 (63.9) 2121 (9.0)

* Totals may not add to 100% because missing data.

TABLE 2. Characteristics of Hospitals by Volume Category

Children’s Teaching

Hospital Hospital Mean Age Medicaid

(n = 24) (n = 416) of Patients * Patients *

No. (%) No. (%) (y) (%)

Lowest 0 (0) 81 (19.5) 12.9 19.9

Low 2 (8.3) 233 (56.0) 12.3 20.9

Medium 7 (29.2) 78 (18.8) 11.7 22.7

High 7 (29.2) 15 (3.6) 10.8 26.2

Highest 8 (33.3) 9 (2.2) 10.1 40.6

* P for trend < .05

TABLE 3. Hospital Volume, Misdiagnosis Rates, and Unadjusted

Odds of Misdiagnosis


Hospitals Appendectomies Misdiagnosis Unadjusted

(% of Total) (% of Total) Rate (%) OR (95% CI)

Total 2217 37 109 8.4

Lowest 1060 (47.8) 4786 (12.9) 8.8 1.9 (1.2-3.0)

Low 974 (43.9) 19 869 (53.5) 8.9 1.9 (1.2-3.0)

Medium 149 (6.7) 8246 (22.2) 7.9 1.7 (1.1-2.7)

High 21 (0.9) 2116 (5.7) 7.6 1.6 (0.9-2.7)

Highest 13 (0.6) 2092 (5.6) 4.8 Reference


TABLE 4. Risk Factors Associated With Misdiagnosis of Pediatric


Adjusted OR P Value

(95% CI)


Lowest 1.5 (1.0-2.2) .048

Low 1.6 (1.1-2.3) .02

Medium 1.5 (1.0-2.2) .06

High 1.4 (0.9-2.2) .17

Highest 1.0 Reference group


Female 1-4 y 2.3 (1.7-3.0) <.01

Female 5-9 y 1.8 (1.5-2.1) <.01

Female 10-14 y 2.3 (2.0-2.6) <.01

Female 15-18 y 3.3 (2.9-3.8) <.01

Male 1-4 y 2.0 (1.6-2.6) <.01

Male 5-9 y 1.3 (1.1-1.6) <.01

Male 10-14 y 1.0 (0.9-1.2) .97

Male 15-18 y 1.0 Reference group


African American 0.93 (0.77-1.11) .41

Hispanic 0.57 (0.50-0.65) <.01

Other 0.66 (0.54-0.81) <.01

White 1.0 Reference group


Medicaid 0.98 (0.88-1.08) 0.68

Other 0.89 (0.79-1.01) 0.07

Private 1.0 Reference group

ABBREVIATIONS. KID, Kids’ Inpatient Database; ICD-9, International Classification of Diseases, Ninth Revision; OR, odds ratio; CI, confidence interval.


Dr Smink is supported by Agency for Healthcare Research and Quality grant T32 HS00063.


(1.) HCUPnet, Healthcare Cost and Utilization Project. Rockville, MD: Agency for Healthcare Research and Quality; 2002. Available at: http://

(2.) Putnam TC, Gagliano N, Emmens RW. Appendicitis in children. Surg Gynecol Obstet. 1990;170:527-532

(3.) Neilson IR, Laberge JM, Nguyen LT, et al. Appendicitis in children: current therapeutic recommendations. J Pediatr Surg. 1990;25:1113-1116

(4.) Pearl RH, Hale DA, Molloy M, Schutt DC, Jaques DP. Pediatric appendectomy. J Pediatr Surg. 1995;30:173-178

(5.) Korner H, Sondenaa K, Soreide JA, et al. Incidence of acute nonperforated and perforated appendicitis: age-specific and sex-specific analysis. World J Surg. 1997;21:313-317

(6.) Velanovich V, Satava R. Balancing the normal appendectomy rate with the perforated appendicitis rate: implications for quality assurance. Am Surg. 1992;58:264-269

(7.) Colson M, Skinner KA, Dunnington G. High negative appendectomy rates are no longer acceptable. Am J Surg. 1997;174:723-726

(8.) Hale DA, Jaques DP, Molloy M, Pearl RH, Schutt DC, d’Avis JC. Appendectomy. Improving care through quality improvement. Arch Surg. 1997;132:153-157

(9.) Pittman-Waller VA, Myers JG, Stewart RM, et al. Appendicitis: why so complicated? Analysis of 5755 consecutive appendectomies. Am Surg. 2000;66:548-554

(10.) Flum DR, Koepsell T. The clinical and economic correlates of misdiagnosed appendicitis: nationwide analysis. Arch Surg. 2002;137:799-804

(11.) Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The empirical relation between surgical volume and mortality. N Engl J Med. 1979;301:1364-1369

(12.) Hannan EL, O’Donnell JF, Kilburn H Jr, Bernard HR, Yazici A. Investigation of the relationship between volume and mortality for surgical procedures performed in New York State hospitals. JAMA. 1989;262: 503-510

(13.) Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mortality in the United States. N Engl J Med. 2002;346: 1128-1137

(14.) Birkmeyer JD, Finlayson EV, Birkmeyer CM. Volume standards for high-risk surgical procedures: potential benefits of the Leapfrog initiative. Surgery. 2001;130:415-422

(15.) Jenkins KJ, Newburger JW, Lock JE, Davis RB, Coffman GA, Iezzoni LI. In-hospital mortality for surgical repair of congenital heart defects: preliminary observations of variation by hospital caseload. Pediatrics. 1995;95:323-330

(16.) Hannan EL, Racz M, Kavey RE, Quaegebeur JM, Williams R. Pediatric cardiac surgery: the effect of hospital and surgeon volume on in-hospital mortality. Pediatrics. 1998;101:963-969

(17.) Phibbs CS, Bronstein JM, Buxton E, Phibbs RH. The effects of patient volume and level of care at the hospital of birth on neonatal mortality. JAMA. 1996;276:1054-1059

(18.) Bratton SL, Haberkern CM, Waldhausen JH, Sawin RS, Allison JW. Intussusception: hospital size and risk of surgery. Pediatrics. 2001;107: 299-303

(19.) HCUP Kids’ Inpatient Database. Rockville, MD: Agency for Healthcare Research and Quality; 2001

(20.) ICD-9-CM: International Classification of Diseases, Ninth Revision, Clinical Modification. Salt Lake City, UT: Medicode; 1996

(21.) Addiss DG, Shaffer N, Fowler BS, Tauxe RV. The epidemiology of appendicitis and appendectomy in the United States. Am J Epidemiol. 1990;132:910-925

(22.) Flum DR, Morris A, Koepsell T, Dellinger EP. Has misdiagnosis of appendicitis decreased over time? A population-based analysis. JAMA. 2001;286:1748-1753

(23.) Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13-22

(24.) Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42:121-130

(25.) O’Connor GT, Plume SK, Olmstead EM, et al. A regional intervention to improve the hospital mortality associated with coronary artery bypass graft surgery. The Northern New England Cardiovascular Disease Study Group. JAMA. 1996;275:841-846

Douglas S. Smink, MD, MPH * ([double dagger])([section])([parallel]); Jonathan A. Finkelstein, MD, MPH ([parallel][paragraph]); Ken Kleinman, ScD ([paerallel]); and Steven J. Fishman, MD ([double dagger])

From the * Harvard Pediatric Health Services Research Fellowship, ([double dagger]) Department of Surgery, and ([paragraph]) Division of General Pediatrics, Children’s Hospital Boston, Boston, Massachusetts; ([section]) Department of Surgery, Brigham and Women’s Hospital, Boston, Massachusetts; and ([parallel]) Department of Ambulatory Care and Prevention, Harvard Pilgrim Health Care and Harvard Medical School, Boston, Massachusetts.

Received for publication Feb 4, 2003; accepted May 20, 2003.

Reprint requests to (S.J.F.) Department of Surgery, Children’s Hospital Boston, 300 Longwood Ave, Boston, MA 02115. E-mail: steven.fishman@

COPYRIGHT 2004 American Academy of Pediatrics

COPYRIGHT 2004 Gale Group