|
|
| |||||||||||||||||||||||||||||||||||||||
Background We studied differences in the incidence of appendiceal perforation in patients with acute appendicitis according to their insurance coverage.
Methods In a retrospective analysis of hospital-discharge data, we examined the likelihood of ruptured appendix among adults 18 to 64 years old who were hospitalized for acute appendicitis in California from 1984 to 1989.
Results After controlling for age, sex, psychiatric diagnoses, substance abuse, diabetes, poverty, race or ethnic group, and hospital characteristics, we found that ruptured appendix was more likely among both Medicaid-covered and uninsured patients with appendicitis than among patients with private capitated coverage (odds ratios, 1.49 [95 percent confidence interval, 1.41 to 1.59] and 1.46 [95 percent confidence interval, 1.39 to 1.54], respectively). After adjustment for the above factors, the risk of appendiceal rupture associated with a lack of private insurance was elevated at both county and other hospitals, but admission to a county hospital was an independent risk factor. In all income groups, appendiceal rupture was more likely with fee-for-service than capitated private coverage (overall odds ratio, 1.20 [95 percent confidence interval, 1.15 to 1.25]).
Conclusions Among patients with appendicitis an increased risk of ruptured appendix may be due to insurance-related delays in obtaining medical care. Both organizational and financial features of Medicaid and various types or levels of private third-party coverage may be involved. The significant association between ruptured appendix and insurance coverage after adjustment for socioeconomic differences suggests barriers to receiving medically necessary acute care that should be considered in current deliberations on health policy.
Uninsured people and those covered by Medicaid receive less medical care than people with private insurance coverage1,2,3,4,5,6. Medicaid recipients also receive less preventive care than the privately insured,7,8,9 use emergency departments frequently,10 and have high rates of nondiscretionary hospital admission11,12. Differences in medical care according to type of insurance have also been found among the privately insured13,14,15,16,17. Patients in health maintenance organizations (HMOs) are hospitalized less often18,19 and have fewer cesarean sections than patients in fee-for-service plans,13 but the rate of total hospital services is not necessarily lower in HMOs1,17,20,21. As compared with members of fee-for-service plans, HMO enrollees receive more preventive services17,22.
Few studies have shown a correlation between type of insurance and nonelective medical services1,5. Several retrospective studies have found worse health in the uninsured than in the privately insured, indicating that disparities in care were likely, but these analyses did not distinguish acute from preventive and long-term care3,12,23,24,25,26,27. Difficulties in the receipt of preventive and long-term care may not be relevant to access to acute care for serious conditions. Furthermore, it is difficult to link most adverse health outcomes, which have complex causes, only to a lack of care due to inadequate insurance.
We used statewide California hospital-discharge data to study differences according to type of insurance coverage in the incidence of appendiceal rupture among adults 18 to 64 years of age who were hospitalized for acute appendicitis. Appendiceal perforation can be associated with severe morbidity28,29 and elevated mortality,29,30,31,32 and it can be prevented by timely treatment of symptomatic acute appendicitis. The lifetime risks of acute appendicitis are 7 and 9 percent for U.S. women and men, respectively33. As described in the literature, the natural history of acute appendicitis makes it unlikely that socioeconomic status, previous medical conditions, use of preventive health care, or personal health habits influence the development of this condition28,29,30,31,32,33,34,35,36,37,38. Once symptoms of appendicitis occur, postponed surgery (delayed for more than 12 to 24 hours28,29,31) is the strongest predictor of perforation35,36,37,38. Apart from delays in treatment after the appearance of symptoms, increasing age among adults,28,29,30,32 anatomical factors such as retrocecal appendix, and possibly, male sex28,30,31,32,33 are the only recognized risk factors.
Ruptured appendix has not been adequately studied as a marker of access to medical care. Billings and colleagues rejected appendicitis itself as an "ambulatory-care sensitive condition" because it is not preventable,11,34,39 but Rutstein et al. suggested that deaths during appendectomy could serve as an indicator of the quality of medical care40. The study by Weissman et al. of several conditions, including ruptured appendix, showed higher rates of preventable hospitalization among uninsured and Medicaid-covered persons overall and for several specific conditions12. However, the association between insurance coverage and ruptured appendix, when considered alone (without adjustment for hospital, socioeconomic status, or personal factors other than the patient's age and sex), was not significant.
Methods
Study Data and Sample
We used publicly available computerized data on all discharges from civilian acute care hospitals in California. After the exclusions described below, the sample consisted of all 96,587 hospitalizations of California residents 18 to 64 years old with a principal diagnosis of acute appendicitis (code 540.0, 540.1, or 540.9 of the International Classification of Diseases, 9th Revision, Clinical Modification41) from 1984 through 1989 (Table 1). Patients who were 65 or older were excluded because so few of them lack Medicare coverage; children were excluded because the dynamics of children's and adults' care may differ. Virtually all cases of appendicitis in the United States are thought to result in hospitalization12,29,30,33,34. We excluded cases of appendicitis not specified to be acute and appendectomies without a principal diagnosis of acute appendicitis in order to eliminate incidental appendectomies performed during hospitalization for other reasons and laparotomies (with or without appendectomy) for abdominal pain in the absence of acute appendicitis.
|
The outcome measure was appendiceal rupture (code 540.0, acute appendicitis with peritonitis, reflecting gross perforation; or code 540.1, acute appendicitis with peritoneal abscess, reflecting microperforation of the intestinal wall) in all patients hospitalized for acute appendicitis. Peritonitis and abscess are reliably reported in discharge data30. Supplementary analyses examined the risk of peritonitis alone; abscess without peritonitis was too rare (less than 5 percent of the sample) to examine with adjustment for important covariates.
Independent variables included insurance coverage and other personal characteristics (age, sex, psychiatric diagnoses, substance abuse, poverty level of community of residence, and race or ethnic group). We also included variables representing characteristics of medical care delivery systems or settings that could play a part in the risk of ruptured appendix; these systems variables included hospital characteristics and circumstances of admission.
Insurance
The type of insurance coverage (the expected principal payer on admission) was categorized as none (uninsured, paying out of pocket or medically indigent); Medicaid; or private coverage, which was further categorized as fee-for-service or capitated (through an HMO or other prepaid plan). We excluded 4153 records for which insurance coverage was unknown, was poorly defined, or was in a category with too few numbers for separate analysis.
Age
Age in years was included as a continuous variable, on the basis of a linear relation between increasing age and appendiceal rupture in the study age group30,42.
Secondary Diagnoses
Secondary diagnoses included psychiatric diagnoses or substance abuse (including alcohol abuse) (codes 290 to 319), which could affect the seeking of care and diagnosis; and diabetes mellitus (code 250), which could increase the risk of appendiceal rupture due to autonomic neuropathy or compromised intestinal vasculature.
Community Poverty Level
On the basis of 1990 census data for ZIP Code areas, we defined three groups according to the percentage of adults living in poverty in the community of residence of each case subject: 22 percent or more, 10 to 21.9 percent, and fewer than 10 percent of adults at or under the federal poverty level. Supplementary analyses used three other measures: the percentage of all persons living in poverty, the percentage of households receiving public assistance, and median household income. Information about poverty or income was missing for 5244 records.
Race or Ethnic Group
Race or ethnic group was categorized as black, Asian, white, or Hispanic. We excluded 1428 records of subjects of unknown, Native American, or other race (because the numbers were too small to study separately).
Hospital Characteristics
On the basis of characteristics in the 1989 hospital-discharge data, hospitals were classified according to ownership (county- or city-owned, referred to as "county," or other), teaching status (whether they had accredited residency programs in emergency medicine, family practice, general surgery, or internal medicine),43 total volume of admissions (the number of nonobstetrical admissions of adults 18 to 64 years old, classified as low [<1800], medium [1800 to 7700], or high [>7700] volume, which was shown in preliminary analyses to be highly correlated with the volume of admissions for appendicitis), and percentage of emergency admissions (the number of admissions from the emergency room divided by the total volume of admissions, classified as low [<30 percent], medium [30 to 60 percent], or high [>60 percent]).
Circumstances of Admission
Admission from the emergency department was distinguished from admission through other sources. Weekend admissions (Friday through Sunday) were distinguished from other admissions.
Statistical Analysis
We examined the risk of perforation in acute appendicitis that was associated with the type of insurance, controlling for age, sex, psychiatric diagnoses, substance abuse, diabetes, poverty level of community of residence, race or ethnic group, hospital characteristics, and the circumstances of admission. Multiple logistic regression (SAS Logist software)44 was used to determine adjusted odds ratios and 95 percent confidence intervals.
Limitations of the Data
Hospital-discharge data do not reveal the length of time from the onset of a symptom until a symptomatic person first sought care, from the time care was first sought until the initial evaluation, or from the initial evaluation until definitive diagnosis and treatment. Information was not available on individual incomes. Census data according to ZIP Code are widely used to indicate socioeconomic status; however, there can be socioeconomic heterogeneity within ZIP Codes.
Results
Patients with Ruptured Appendix
Ruptured appendix occurred in 34.3 percent of the uninsured patients, 33.6 percent of the Medicaid patients, 29.3 percent of the patients with private fee-for-service insurance, and 25.8 percent of the patients with private capitated-payment insurance hospitalized for acute appendicitis (Table 1). In-hospital mortality from appendicitis was rare (8 deaths per 10,000 hospitalizations overall), but the rate appeared to be higher with perforation (2 deaths per 1000) than without (2 deaths per 10,000).
Results of Multivariate Analysis
A higher risk of ruptured appendix in patients without capitated insurance coverage was observed in multivariate models that controlled for differences in age, sex, psychiatric diagnoses, substance abuse, diabetes, poverty level of community of residence, race or ethnic group, hospital characteristics (including admission to a county hospital), and circumstances of admission. Both patients with appendicitis who had Medicaid coverage and those who were uninsured were about 1.5 times more likely to have appendiceal rupture than those with capitated private coverage (odds ratios, 1.49 [95 percent confidence interval, 1.41 to 1.59] and 1.46 [95 percent confidence interval, 1.39 to 1.54], respectively) (Table 2). In addition, fee-for-service private insurance, as compared with capitated private insurance, was associated with moderately increased risk (odds ratio, 1.20 [95 percent confidence interval, 1.15 to 1.25]) (Table 2); this relation was observed in all groups classified according to the poverty level of the community of residence (Table 3).
|
|
Overall, blacks with acute appendicitis had a moderately higher risk of appendiceal rupture than whites (odds ratio, 1.14 [95 percent confidence interval, 1.06 to 1.24]), but otherwise race or ethnic group was not significant. Admission to a county hospital for acute appendicitis was associated with a substantially elevated risk of rupture (odds ratio, 1.71 [95 percent confidence interval, 1.60 to 1.83]), whereas admission to a teaching hospital or from an emergency room was associated with a somewhat reduced risk; a high volume of hospital admissions was associated with a somewhat elevated risk of rupture. Other hospital or admission factors were not associated with significant differences in risk.
In analyses stratified according to type of hospital ownership, an elevated risk of ruptured appendix was associated with a lack of private insurance at both county and noncounty hospitals (Table 4). However, differences associated with some other variables were observed in separate analyses of county and noncounty hospitals. For example, Table 4 shows that blacks had a significantly increased risk of perforation only at noncounty hospitals. Admission from the emergency room was protective at both types of hospital.
|
Discussion
Lack of any medical insurance, coverage by Medicaid, and coverage by fee-for-service private insurance were found to be significant risk factors for ruptured appendix in patients hospitalized for acute appendicitis. Multivariate analysis showed that this correlation was independent of multiple patient and hospital characteristics, community poverty level, and other socioeconomic measures.
In contrast to most illnesses, appendicitis and appendiceal rupture are not likely to be related to income-related differences in health or personal habits. It is also unlikely that insurance-related differences in unmeasured individual medical risks can explain our findings. For example, anatomical risks such as retrocecal appendix would not vary according to type of insurance. Our analyses controlled for age and sex, the only other risk factors discussed in the literature, as well as for psychiatric diagnoses, substance abuse, diabetes, and other relevant characteristics of the patients and hospitals we studied. Substance abuse is particularly likely to be underreported among persons with private insurance; thus, any bias in estimating the risk of ruptured appendix in patients without private insurance would be conservative.
Differences in the reporting of complications in hospital records are unlikely to explain these findings, because peritonitis and abscess are reliably reported in discharge data30. Furthermore, we controlled for several characteristics of the hospitals and patients in our study. These controls should have reconciled a considerable amount of the variation among physicians in diagnostic and reporting practices. We studied only admissions for acute appendicitis, rather than appendectomies, to avoid including admissions for reasons other than acute appendicitis. If privately insured patients with abdominal pain were more likely to be hospitalized, and if reimbursement incentives caused some negative or equivocal results of abdominal explorations to be coded as acute appendicitis without complications, then the rate of ruptured appendix among the privately insured patients could have been spuriously deflated by a larger denominator of "appendicitis" admissions. This might explain the elevated risk among uninsured patients as compared with those with fee-for-service private coverage, but not the risk among the uninsured and those covered by Medicaid as compared with those with capitated private coverage; incentives in capitated plans are less likely to reward unnecessary hospitalization or overdiagnosis.
The findings could be explained by insurance-related delays in seeking medical care. Uninsured patients might delay seeking care at private and county hospitals because of concern about their ability to pay. During the 1980s, California county hospitals were pressed to recover costs45 and began actively to bill patients. The findings of this study are consistent with our clinical experience that low-income uninsured people are reluctant to seek care, even for serious emergencies and even at county hospitals, fearing bills they cannot pay. Lack of a primary care physician10 could also be associated with delays in seeking or receiving medical help among patients covered by Medicaid and the uninsured. Private physicians can be reluctant to accept patients with Medicaid coverage because of low rates of reimbursement or administrative burdens46,47. Logistic barriers, such as lack of transportation or an insufficient number of physicians in poor communities,44,45 could contribute to delays in seeking medical care, but they should have been controlled for by our use of the socioeconomic variables. Another potential source of delay for patients without private insurance is longer waiting times to be diagnosed or to be referred by community physicians to surgeons.
Delays in emergency rooms are not likely to account for our results, given the protective effect associated with admission from the emergency room. Pieper et al. implicated preoperative delay in the hospital as a factor in appendiceal rupture32; our supplementary analyses suggested an increased risk of perforation when surgery was performed two or more days after admission, but we have no information on hours of delay on the first day. Furthermore, we have no information on reasons for delays after admission to the hospital, which could have been related to operative risk or atypical presentation. Long delays between admission and surgery are unlikely to explain our findings, because the pattern of risks did not change when the analysis included a variable representing days from admission to surgery.
The increased risk of appendiceal rupture among the uninsured and those covered by Medicaid suggests the possible influence of factors related to low income rather than inadequate insurance itself. However, these increased risks persisted when we adjusted for the poverty and income levels of the community, the patients' race or ethnic group, and admission to a county hospital, all of which are highly correlated with socioeconomic status. Furthermore, the differences in the risk of ruptured appendix according to type of insurance were evident even at county hospitals, where most patients, regardless of insurance coverage, have low incomes. It is thus unlikely that the differences in risk according to type of insurance were confounded by socioeconomic status.
Admission to a county hospital was the only systems variable that was consistently correlated with a high risk of appendiceal rupture. The increased risk at county-owned facilities may reflect an increased prevalence of unmeasured factors related to the seeking or delivery of care in the population served at those institutions48; such factors could include a lower likelihood of having a consistent primary care physician, even among patients with third-party coverage. Linguistic barriers could delay the diagnosis of acute appendicitis, but the elevated risk of appendiceal rupture at county hospitals persisted despite control for race or ethnic group. Delays in diagnosis or treatment after admission to county or other hospitals cannot be ruled out,49 but they appear less likely than delays before arrival at the hospital. At both types of hospital we found a decreased risk associated with admission from the emergency room, and the elevated risk at county hospitals did not change when we included the number of days from admission to surgery in a supplementary model.
Among patients with private insurance, those with capitated coverage had a lower risk of appendiceal perforation than those with fee-for-service insurance. Differences in personal factors (including general health status) are unlikely to explain these differences, given the outcome measure (ruptured appendix). Deductibles and higher copayments in fee-for-service plans may contribute to delays by patients in seeking care14,15,50,51. Furthermore, large staff-model HMOs often provide urgent care facilities that are separate from the hospital emergency room; such facilities may increase the likelihood of an early evaluation for abdominal pain.
Causal inferences cannot be made from our findings. However, the results suggest that there are avoidable delays between the onset of symptoms and definitive treatment for acute appendicitis in patients covered by Medicaid and patients without any medical insurance. To a lesser but still significant extent, patients covered by fee-for-service plans also appear to be at a disadvantage as compared with those covered by capitated private plans. Organizational and financial features of Medicaid and private plans may account for these findings. Although it is impossible to identify the causes of delays, postponement in seeking medical care seems more likely than delays in the hospital. Delays in reaching the hospital once care has been sought cannot be ruled out, however. Even if all of the increased risk of appendiceal rupture was due to delays in seeking care, our results point to insurance-related barriers to the appropriate use of medically necessary acute care. These hindrances should be considered in current deliberations on health policy.
We are indebted to Mark Blumberg, M.D., Director of Special Studies, Kaiser Permanente Research Division, and Jonathan Rodnick, M.D., Professor and Chair, Department of Family and Community Medicine, University of California, San Francisco, for helpful comments and references; to Ellen Shaffer, M.P.H., for critical review of an earlier draft of the manuscript; and to Robert Lynch for assistance in preparing the manuscript.
Source Information
From the Department of Family and Community Medicine, School of Medicine, University of California, San Francisco (P.B., S.E., T.B.); the Departments of Medicine and Surgery, San Francisco General Hospital Medical Center (V.M.S., W.S.); and the Department of Surgery, University of California, San Francisco (W.S.) -- all in San Francisco.
Address reprint requests to Dr. Braveman at the Department of Family and Community Medicine, University of California, San Francisco, Box 0900, San Francisco, CA 94143-0900.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Insurance and the Risk of Ruptured Appendix
Siddiqui M. A., Blumberg M. S., Juhn P. I., O'Toole S. J., Karamanoukian H. L., Glick P. L., Welch H. G., Koepsell T. D., Andersson R., Nyström P. O., Braveman P., Egerter S., Schecter W.
Extract |
Full Text
N Engl J Med 1995;
332:395-398, Feb 9, 1995.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
