Education
Evaluating diagnostic accuracy in appendicitis using administrative data11

https://doi.org/10.1016/j.jss.2004.08.020Get rights and content

Background

Research techniques using administrative data have been used to assess quality of care in the management of appendicitis, but their validity has not been assessed. This study assessed the validity of a research technique using administrative data to determine whether patients undergoing an appendectomy actually had appendicitis or if they had a negative appendectomy (NA).

Materials and methods

A retrospective study of patients at Group Health Cooperative of Puget Sound undergoing appendectomy from 1991–1999 was conducted to compare the accuracy of administrative codes with data abstracted from medical records.

Results

Of 1823 nonincidental appendectomies (mean age 31 ± 18.6 years, 49.6% female), 280 did not have appendicitis by criteria applied to administrative data (15.4%). The accuracy of this method for determining appendicitis was determined by medical record evaluation revealing sensitivity (98.6%), specificity (48.6%), positive predictive value for appendicitis (83.8%), and negative predictive value for NA (70.4%). When the administrative-data technique did not classify a patient as having had appendicitis, chart-abstracted data indicated appendicitis in 6.4%, and an incorrectly labeled incidental appendectomy in 23.2%. The administrative-data technique for detecting NA erroneously included many cases of incidental appendectomy and missed many cases of clinically confirmed NA. More than half of clinical NAs were missed by the administrative technique.

Conclusions

In this setting, the sensitivity of administrative-data techniques for the detection of appendicitis was excellent, but their adequacy for identifying patients undergoing NA was limited. The use of this technique as a quality measure should be reconsidered.

Introduction

The lifetime risk of developing appendicitis has been estimated to be 6–9% [1], and appendectomy remains the most frequently performed emergency abdominal surgical procedure. Because it is such a commonly occurring disease and its diagnosis and treatment involves caregivers from many specialties, complex quality of care issues remain to be addressed. Studying the incidence of uncommon but preventable complications in the management of appendicitis requires a large number of patients, which in turn requires either burdensome chart reviews or more costly, prospective data gathering. The use of administrative data for the study of appendectomy and appendicitis may be appealing because these data provide population level information without many of the cost and logistical difficulties of other study designs. Administrative data in appendicitis have been used to assess the basic epidemiology of appendicitis [1], temporal changes in diagnosis and management [2, 3], variations in resource utilization and patient outcome [4, 5], and disparities in care [6, 7].

In evaluating diagnostic accuracy in appendicitis, techniques based on administrative data typically are used to count the number of patients who had a nonincidental appendectomy (appendectomy not being performed incidental to other surgical procedures) but did not have appendicitis. This form of misdiagnosis manifests as a “negative appendectomy” (NA) and may be an important marker of the quality of care as it entails an operative procedure in the absence of the assumed pathology. The frequency of NA also may be important in tracking the accuracy of different diagnostic regimens for right-sided abdominal pain and of the effectiveness of advanced diagnostic technology such as computed tomography and ultrasound in general practice.

This study was designed to assess the validity of a technique first described by Addiss et al. [1] using administrative data to assess the frequency of appendicitis in patients undergoing appendectomy and rates of NA. This approach essentially relies on a high level of sensitivity and specificity of diagnostic codes for appendicitis. With this technique, a patient is described as being accurately diagnosed if an appendectomy procedure code is associated with an appendicitis diagnostic code. A patient is classified as having had a NA if the patient’s administrative record includes a procedure code for a nonincidental appendectomy without an associated diagnostic code of appendicitis. To date, there has been no validation of this technique. Because the validity of diagnostic coding has come under increased scrutiny, we proposed this study to evaluate the accuracy of this technique.

Section snippets

Study design

This methodologic substudy was a component of a larger project performed at the Group Health Cooperative (GHC) of Puget Sound, using a computerized, population-based database to evaluate the incidence of NA. That study involved chart analysis of a subset of patients who underwent appendectomy in the 1990s, and the methodologic substudy was designed to compare the administrative coding of appendicitis and NA with clinical-pathologic data.

Setting

Group Health Cooperative (GHC) of Puget Sound is a large

Results

Between 1991–1999, 1823 nonincidental appendectomies were performed in the GHC system (mean age 31 ± 18.6 years, 49.6% female). Of these, 280 were classified as NA by the administrative-data method (15.4% of total; 8.3% of men versus 24% of women [P < 0.001]). Medical records of 614 patients were reviewed, including all 280 patients who were thought to have had a NA based on administrative coding and a random sample of 334 patients found to have to have appendicitis by administrative codes (

Discussion

Administrative data contain information usually gathered about patients for the purpose of billing and tracking resource utilization. Although primarily captured for other purposes, the use of administrative data to assess quality of care is attractive because elements of the administrative record can be applied to clinical questions. While the use of administrative data are appealing because of its low cost, ease of access and population-based implications, there may be serious problems with

References (14)

  • D.G. Addiss et al.

    The epidemiology of appendicitis and appendectomy in the United States

    Am. J. Epidemiol. Nov.

    (1990)
  • D.R. Flum et al.

    Has misdiagnosis of appendicitis decreased over time? A population-based analysis

    JAMA

    (2001)
  • S.W. Wen et al.

    Diagnostic accuracy and short-term surgical outcomes in cases of suspected acute appendicitis

    CMAJ

    (1995)
  • D.R. Flum et al.

    The clinical and economic correlates of misdiagnosed appendicitisnationwide analysis

    Arch. Surg.

    (2002)
  • R.G. Buckley et al.

    The risk of appendiceal rupture based on hospital admission source

    Acad. Emerg. Med.

    (1999)
  • U. Guller et al.

    Laparoscopic versus open appendectomyoutcomes comparison based on a large administrative database

    Ann. Surg.

    (2004)
  • A. Gadomski et al.

    Ruptured appendicitis among children as an indicator of access to care

    Health Serv. Res.

    (2001)
There are more references available in the full text version of this article.

Cited by (0)

1

The Robert Wood Johnson Foundation provided funding for this project. The views expressed in this article are those of the authors and not necessarily those of the Robert Wood Johnson Foundation or the University of Washington.

View full text