Correspondence
Insurance and the Risk of Ruptured Appendix
N Engl J Med 1995; 332:395-398February 9, 1995
- Article
To the Editor:
The article by Braveman et al. (Aug. 18 issue)1 does not mention the number of emergency room visits or any admissions in the weeks before the admission for appendicitis. This is important information, because patients with no insurance and those covered by Medicaid have a high number of emergency room visits, as the authors note, and a high number of hospital admissions during which no diagnosis is made. The patients with early signs and symptoms of appendicitis could have had one or more emergency room visits or hospital admissions for undiagnosed appendicitis, only to present later with a frankly perforated appendix. Other abdominal conditions can mimic appendicitis, and emergency ultrasound examination of the appendix has a sensitivity of 80 percent and a specificity of 54 percent for the diagnosis of appendicitis.2
2 ReferencesMumtaz A. Siddiqui, M.D.
Graduate Hospital, Philadelphia, PA 191461
Braveman P ,Schaaf VM ,Egerter S ,Bennett T ,Schecter W . Insurance-related differences in the risk of ruptured appendix. N Engl J Med 1994;331:444-449
Full Text | Web of Science | Medline2
Pignatelli V ,Ruiu U ,Savino A ,Kiferle M ,Orsitto E ,Calderazzi A . The echographic diagnosis of acute appendicitis and its complications. Radiol Med (Torino) 1990;79:215-219
To the Editor:
Braveman et al. reported an increased odds ratio (1.19) for ruptured appendix among black patients with appendicitis as compared with patients in other racial or ethnic groups in noncounty hospitals in California, after adjustments for age, sex, source of payment, and other factors.
The Quality Measurement and Management Program (QMMP) study, which they cite, used data from all cases of appendicitis at 366 U.S. hospitals participating in the study, very few of which were county-owned.1 The ruptured-appendix outcome (either abscess or peritonitis) was adjusted for risk with the use of detailed age tables according to sex. Table 1Table 1
Ratio of Observed to Expected Cases of Ruptured Appendix and Delayed Admissions, According to Race or Ethnic Group. shows the ratio of observed to expected cases of ruptured appendix, according to race or ethnic group, for the entire QMMP study and for the patients at Kaiser Foundation hospitals in the northern California region. The ratios were high among black, Hispanic, and Asian patients in the QMMP study but only among black patients in the Kaiser Foundation hospitals. These high ratios may be explained in part by the greater likelihood of inadequate coverage among the members of minority groups (the ratio for patients with Medicaid was 1.27 in the overall study). However, the difference is puzzling, since all patients at Kaiser Foundation hospitals have almost equally comprehensive benefits.Because we found that delays in surgery after admission were not an important explanation, we sought data on the earliest unscheduled office or emergency room visits at the Kaiser Foundation hospitals that could plausibly be associated with appendicitis. Over 97 percent of the Kaiser admissions (1763 of 1810) were matched to such “first” visits. These admissions were then classified as delayed (first visit before 3 p.m. the day before admission), same day (first visit on the day of admission), or other (first visit after 3 p.m. on the day before admission). Table 1 shows the ratio of observed to expected delayed admission, adjusted for age and sex. There were no significant differences according to race or ethnic group, indicating that the black patients did not have an excess number of delayed admissions after the first visit. Table 1 also shows the ratio of observed to expected cases of ruptured appendix for patients admitted on the same day as the first visit. After adjustment for age and sex, blacks were significantly more likely to have a ruptured appendix than other patients (ratio of observed to expected cases, 1.52).
One possible explanation is that blacks may be more likely to delay seeking care for a given severity of symptoms associated with appendicitis than other patients with the same comprehensive benefits and access to care. Another possible explanation is that abscess or peritonitis develops more rapidly in blacks.
These findings should be considered by investigators concerned about differences in outcomes and in the care received by blacks.
1 ReferencesMark S. Blumberg, M.D.
Peter I. Juhn, M.D., M.P.H.
Kaiser Foundation Health Plan, Oakland, CA 94612To the Editor:
We congratulate Braveman et al. on performing a thorough and comprehensive study. However, we would like to make the following points, which are based on our own study of children presenting with appendicitis at the Children's Hospital of Buffalo, New York.
There is strong evidence that a delay in treatment in the pediatric population results in an increased risk of perforation and higher mortality.1 However, it is not clear whether the delay occurs at the stage when the parent is seeking medical advice, the referring physician is seeking a surgical consultation, or the surgeon is deciding whether to operate.1-3 To resolve this issue we determined the time to presentation and the time to definitive treatment for all cases of confirmed acute appendicitis presenting during the two-year period from January 1992 to December 1993 (Table 1Table 1
Insurance-Related Differences in the Perforation Rate, Time to Treatment, and Length of Hospital Stay among Children with Acute Appendicitis.). The cases were categorized according to insurance coverage.The rate of perforation was higher in the group with Medicaid coverage or no insurance than in the group with coverage from a health maintenance organization (HMO) or a fee-for-service plan (P = 0.085). In addition, patients with Medicaid coverage or no insurance presented significantly later (P<0.008) and stayed in the hospital significantly longer (P<0.05) than those with HMO or fee-for-service coverage. During hospitalization, there were no significant differences in the time taken by the surgeons to decide to operate. All the children with Medicaid or no insurance presented directly to the emergency room, where surgical consultations are requested immediately in cases of abdominal pain. Because of this policy, any delays in obtaining surgical consultations for children in this group could not be attributed to physician-gatekeepers. All the children covered by HMOs had consultations with their physician-gatekeepers initially, usually by phone. Children in the fee-for-service category were usually, but not always, referred to the emergency rooms by their pediatricians.
Our results support the conclusion of Braveman et al. that there was a delay between the onset of symptoms and definitive treatment in the group of children with Medicaid coverage or no insurance. However, the cause of this delay may not have been the “organizational and financial features” of Medicaid, as they suggest. Our findings indicate that there was a delay between the onset of symptoms and the first contact with a health care provider. Perhaps the parents failed to recognize the importance of their children's symptoms or intentionally delayed seeking medical advice (possibly for financial reasons). In contrast, the parents of children enrolled in HMOs and, to a lesser extent, the parents of children covered by fee-for-service plans had ready access by telephone to pediatrician-gatekeepers who could give advice and direct the children to an emergency room if required.
We suggest that in the pediatric population parental awareness and easy access to a physician-gatekeeper are two factors with a strong influence on the rate of perforating appendicitis. Educating parents about the early symptoms of appendicitis and improving access to physician-gatekeepers may reduce the rate of rupture among children with acute appendicitis and concomitantly decrease morbidity and its associated costs.
3 ReferencesStuart J. O'Toole, M.B., B.S.
Hratch L. Karamanoukian, M.D.
Philip L. Glick, M.D.
Children's Hospital of Buffalo, Buffalo, NY 142221
Pledger HG ,Fahy LT ,van Mourik GA ,Bush GH . Deaths in children with a diagnosis of acute appendicitis in England and Wales 1980-4. BMJ 1987;295:1233-1235
CrossRef | Web of Science | Medline2
Brender JD ,Marcuse EK ,Koepsell TD ,Hatch EI . Childhood appendicitis: factors associated with perforation. Pediatrics 1985;76:301-306
Web of Science | Medline3
Linz DN ,Hrabovsky EE ,Franceschi D ,Gauderer MW . Does the current health care environment contribute to increased morbidity and mortality of acute appendicitis in children? J Pediatr Surg 1993;28:321-328
CrossRef | Web of Science | Medline
To the Editor:
Braveman and colleagues have done admirable work in determining the association between insurance coverage and outcome among patients with appendicitis. Readers should be aware, however, of two important caveats in interpreting their results.
The first concerns the interpretation of an odds ratio in the context of a condition with an outcome (ruptured appendix) that is relatively common (occurring in 30 percent of all cases). We are used to thinking of an odds ratio as a reasonable approximation of a relative risk. In his editorial accompanying the article by Braveman et al., Dr. Relman1 makes such an assumption in interpreting two odds ratios (approximately 1.2 and 1.5). He states, “Rupture was 20 percent more frequent among patients covered by private fee-for-service insurance and approximately 50 percent more frequent among those without insurance and those covered by Medicaid.” Unfortunately, this approximation is valid only when the outcome is relatively rare. Table 1 in the article by Braveman et al. provides data for the calculation of both crude odds ratios and relative risks and makes it clear that the multivariate adjustment had a small effect. Therefore, we can be confident that the increase in risk is closer to 10 and 30 percent, respectively (i.e., crude relative risks of approximately 1.1 and 1.3).
The second caveat concerns the interpretation of the outcome. Is a high rupture rate good or bad? The question arises because of the perspective taken in measuring the primary outcome. The study takes a surgeon's perspective and asks the question: What proportion of all cases of acute appendicitis end in rupture? But from an epidemiologic perspective, the relevant measure is not a proportion based on a case series but a population-based rate. This perspective calls for data on the rate of rupture for the total population of interest (e.g., all uninsured patients and all patients with capitated insurance).
The relevance of this issue was recently emphasized by the work of a group of Swedish surgeons.2 They noted that population-based rates of appendiceal rupture are remarkably stable with respect to both time and geography. But the proportion of all cases of appendicitis that result in rupture varies widely and is closely correlated with the threshold at which surgeons decide to operate. A lower rate of exploratory laparotomy results in a lower rate of appendicitis without rupture (because of the elimination of self-limited cases) and a higher proportion of cases resulting in rupture simply because of the decreased denominator. An abridged version of the Swedish time-series data, indicating an overall reduction in the rate of appendectomy, is shown in Table 1Table 1
Adjusted Incidence of Appendicitis and Proportion of Cases Resulting in Rupture from 1970 through 1989..Thus, from the surgeon's perspective, a more conservative practice looks bad (i.e., leads to a higher proportion of cases with rupture), whereas from the epidemiologic perspective, that same practice looks good (i.e., leads to fewer operations, with a stable population-based rupture rate). We agree with the Swedish researchers' conclusion that the proportion of cases resulting in rupture can be misleading as a measure of the quality of management in cases of appendicitis.
2 ReferencesH. Gilbert Welch, M.D., M.P.H.
Department of Veterans Affairs, White River Junction, VT 05009Thomas D. Koepsell, M.D., M.P.H.
University of Washington, Seattle, WA 981051
Relman AS . Medical insurance and health -- what about managed care? N Engl J Med 1994;331:471-472
Full Text | Web of Science | Medline2
Andersson R ,Hugander A ,Thulin A ,Nystrom PO ,Olaison G . Indications for operation in suspected appendicitis and incidence of perforation. BMJ 1994;308:107-110
CrossRef | Web of Science | Medline
To the Editor:
Braveman et al. state that virtually all cases of appendicitis result in hospitalization. This statement is impossible to verify and is not supported by the cited studies. Indeed, one of those studies1 proposes that perforating appendicitis and nonperforating appendicitis are different entities and that selection bias can explain the increase in the perforation rate with the duration of symptoms, because of the spontaneous resolution of nonperforating appendicitis.
We have shown that spontaneously resolving appendicitis is common and that variations in the perforation rate depend primarily on whether or not patients with these milder cases undergo surgery.2 The relative rate of perforation is therefore misleading as a measure of the quality of care for patients with suspected appendicitis.
In a medical care system in which exploratory surgery is performed in patients with dubious symptoms, more of these milder cases will be detected, resulting in a lower perforation rate, than in a system in which patients postpone seeking medical care for mild symptoms or doctors refrain from performing unnecessary exploratory surgery. In the second system the perforation rate will be higher, mainly because the milder cases will resolve spontaneously. This should not be used as a reason to discredit the second system, as long as patients who need surgery are treated in due time.
Delayed treatment of gangrenous appendicitis may, of course, result in an increased rate of perforation. An insurance-related difference in this respect, however, can be identified only by comparing the incidence of perforation in the epidemiologic sense (i.e., the number of perforations in relation to the population in which they have been documented), assuming that the true incidence of appendicitis is similar in the populations being compared.
2 ReferencesRoland Andersson, M.D.
Per Olof Nyström, M.D.
University Hospital, S-581 85 Linköping, Sweden1
Luckmann R . Incidence and case fatality rates for acute appendicitis in California: a population-based study of the effects of age. Am J Epidemiol 1989;129:905-918[Erratum, Am J Epidemiol 1990;131:1102.]
Web of Science | Medline2
Andersson R ,Hugander A ,Thulin A ,Nystrom PO ,Olaison G . Indications for operation in suspected appendicitis and incidence of perforation. BMJ 1994;308:107-110
CrossRef | Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: If, as Dr. Siddiqui suggests, misdiagnosis rates vary according to insurance coverage, this variation reflects serious insurance-related disparities in the quality of care. We agree with Dr. O'Toole and his colleagues that access to a primary care provider is a likely factor in delays in treatment in the Medicaid group; we construe this factor as an organizational or financial feature of Medicaid. We believe that parental awareness or education is unlikely to alter care-seeking practices for a common, painful, life-threatening condition, and our covariables accounted for some variation in education. It would be interesting to know whether the racial differences in perforation that Drs. Blumberg and Juhn observed remained after adjustment for poverty and comorbidity, which we performed in our study.
Drs. Andersson and Nyström do not mention that they found higher perforation rates in their population during the period from 1984 through 1989.1 They and Drs. Welch and Koepsell argue that perforation rates vary according to diagnostic accuracy and the severity of hospitalized cases, which could reflect the management of “spontaneously resolving” appendicitis. Their claim concerning the prevalence of self-limited appendicitis appears to be based solely on the observation of a decreasing incidence of nonperforating appendicitis with a stable incidence of perforating appendicitis in a study in which neither the sources of data for the denominator nor the methods for calculating or standardizing population rates were noted.1 Errors in any of these areas or the exclusion of cases not treated at the single hospital studied could have affected their results.
The incidence of self-limited appendicitis is unknown, because the diagnosis cannot be confirmed. The incidence would have to be very high to explain the 50 percent greater likelihood of rupture we found among uninsured patients and those with Medicaid coverage, as compared with the likelihood of rupture among the patients enrolled in HMOs. It is exceedingly unlikely that the hospitalized patients in the HMO group included many with self-limited appendicitis. Surgeons — especially in HMOs — observe patients before performing surgery for mild, nonworsening symptoms of recent onset. The lower perforation rate among the patients enrolled in HMOs (which have disincentives for unnecessary intervention) than among the patients with fee-for-service coverage is not consistent with the argument presented by Drs. Andersson and Nyström. Although, in the light of this discussion, we agree that Luckmann2 and Addiss et al.3 should not have been cited as claiming that appendicitis results in hospitalization, the reports by Billings4 and Weissman et al.5 support that claim.
The argument regarding the effect of diagnostic accuracy on perforation rates is based on speculation rather than evidence. However, in interpreting perforation rates as a measure of access to and quality of care, it is advisable to control for factors that could influence diagnostic accuracy, as we did in our analyses. Differences in perforation rates among different insurance or socioeconomic groups at the same types of institutions and during the same periods of time strongly indicate unequal access to appropriate care. Drs. Andersson and Nyström should consider whether the higher perforation rates they observed may reflect patient- or treatment-related delays reflecting the increase in copayments and decrease in the availability of beds (Bygren LO: personal communication) that occurred in Sweden during the 1980s.
5 ReferencesPaula Braveman, M.D., M.P.H.
Susan Egerter, Ph.D.
University of California School of Medicine, San Francisco, CA 94143William Schecter, M.D.
San Francisco General Hospital Medical Center, San Francisco, CA 941101
Andersson R ,Hugander A ,Thulin A ,Nystrom PO ,Olaison G . Indications for operation in suspected appendicitis and incidence of perforation. BMJ 1994;308:107-110
CrossRef | Web of Science | Medline2
Luckmann R . Incidence and case fatality rates for acute appendicitis in California: a population-based study of the effects of age. Am J Epidemiol 1989;129:905-918[Erratum, Am J Epidemiol 1990;131:1102.]
Web of Science | Medline3
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
Web of Science | Medline4
Billings J. Consideration of the use of small area analysis as a tool to evaluate barriers to access. New York: United Hospital Fund, July 30, 1990.
5
Weissman JS ,Gatsonis C ,Epstein AM . Rates of avoidable hospitalization by insurance status in Massachusetts and Maryland. JAMA 1992;268:2388-2394
CrossRef | Web of Science | Medline
