Advertisement

Underuse and overuse of diagnostic testing for coronary artery disease in patients presenting with new-onset chest pain∗

      Abstract

      PURPOSE: To determine the extent of overuse and underuse of diagnostic testing for coronary artery disease and whether the socioeconomic status, health insurance, gender, and race/ethnicity of a patient influences the use of diagnostic tests.
      SUBJECTS AND METHODS: We identified patients who presented with new-onset chest pain not due to myocardial infarction at one of five Los Angeles–area hospital emergency departments between October 1994 and April 1996. Explicit criteria for diagnostic testing were developed using the RAND/University of California, Los Angeles, expert panel method. They were applied to data collected by medical record review and patient questionnaire.
      RESULTS: Of the 356 patients, 181 met necessity criteria for diagnostic cardiac testing. Of these, 40 (22%) failed to receive necessary tests. Only 7 (3%) of the 215 patients who received some form of cardiac testing had tests that were judged to be inappropriate. Underuse was significantly more common in patients with only a high school education (30% vs 15% for those with some college, P = 0.02) and those without health insurance (34% vs 15% of insured patients, P = 0.01). In a multivariate logistic regression model, only the lack of a post–high school education was a significant predictor of underuse (odds ratio 2.2, 95% confidence interval 1.0 to 4.4).
      CONCLUSION: Among patients with new-onset chest pain, underuse of diagnostic testing for coronary artery disease was much more common than overuse. Underuse was primarily associated with lower levels of patient education.
      After controlling for disease burden or health status, the use of cardiovascular procedures and other health-care resources varies with several patient characteristics. These include geographic region (
      • Carlisle D.M
      • Valdez R.B
      • Shapiro M.F
      • Brook R.H
      Geographic variation in rates of selected surgical procedures within Los Angeles County.
      ,
      • Leape L.L
      • Park R.E
      • Solomon D.H
      • Chassin M.R
      • Kosecoff J
      • Brook R.H
      Does inappropriate use explain small area variations in the use of health care services?.
      ), ethnicity (
      • Wenneker M.R
      • Epstein A.M
      Racial inequalities in the use of procedures for patients with ischemic heart disease in Massachusetts.
      ,
      • Ayanian J.Z
      • Epstein A.M
      Differences in the use of procedures between women and men hospitalized for coronary heart disease.
      ,
      • Ayanian J.Z
      • Udvarhelyi S
      • Gatsonis C.A
      • Pashos C.L
      • Epstein A.M
      Racial differences in the use of revascularization procedures after coronary angiography.
      ,
      • Carlisle D.M
      • Leake B.D
      • Shapiro M.F
      Racial and ethnic differences in the use of invasive cardiac procedures among cardiac patients in Los Angeles county, 1986–1988.
      ), gender (
      • Shaw L.J
      • Miller D.D
      • Romeis J.C
      • Kargl D
      • Younis L.T
      • Chaitman B.R
      Gender differences in the non-invasive evaluation and management of patients with suspected coronary artery disease.
      ,
      • Ayanian J.Z
      • Epstein A.M
      Differences in the use of procedures between women and men hospitalized for coronary heart disease.
      ), health-insurance status (including type of health insurance) (
      • Langa K.M
      • Sussman E.J
      The effect of cost-containment policies on rates of coronary revascularization in California.
      ), and socioeconomic status (
      • Bombardier C
      • Fuchs V.R
      • Lillard L.A
      • Warner K.E
      Socioeconomic factors affecting the utilization of surgical operations.
      ,
      • Gittelsohn A.M
      • Halpern J
      • Sanchez R.L
      Income, race, and surgery in Maryland.
      ). Physician practice style has been cited as the major underlying cause of geographic variation in utilization patterns (
      • Wennberg J
      • Gittelsohn A
      Variations in medical care among small areas.
      ).
      Whereas rates of use of many services are lower among minorities and the poor, less is known about the extent to which variation is related to clinical need. More specifically, it is not known whether these or other patient characteristics are related to failure to receive clinically necessary procedures, which we call underuse, or to receipt of unnecessary procedures (overuse), or both.
      The possibility that there is substantial underuse is suggested by several studies that evaluated the use of invasive cardiac procedures in New York State in 1990. Whereas there was little overuse of these procedures, lower rates of use among women, African-Americans, Latinos, and poorer patients suggest that rate disparities may have been due to underuse (
      • Leape L.L
      • Hilborne L.H
      • Park R.E
      • et al.
      The appropriateness of use of coronary artery bypass graft surgery in New York State.
      ,
      • Hilborne L.H
      • Leape L.L
      • Bernstein S.J
      • et al.
      The appropriateness of use of percutaneous transluminal coronary angioplasty in New York State.
      ,
      • Bernstein S.J
      • Hilborne L.H
      • Leape L.L
      • et al.
      The appropriateness of use of coronary angiography in New York state.
      ). More recent studies in Los Angeles County have shown that underuse of angiography is related to not having a cardiologist as a regular source of care and that underuse of revascularization is related to African-American ethnicity and being a patient at a public hospital (
      • Borowsky S.J
      • Kravitz R.L
      • Laouri M
      • et al.
      Effect of physician specialty on use of necessary coronary angiography.
      ,
      • Laouri M
      • Kravitz R.L
      • French W.J
      • et al.
      Underuse of coronary revascularization procedures; application of a clinical method.
      ).
      Patients with cardiac symptoms may fail to get needed care at any time from their initial presentation to the receipt of revascularization procedures. Likewise, at each step of this process, patients may receive tests or procedures that are not indicated.
      We examined the underuse and overuse of invasive and noninvasive diagnostic tests among patients who presented to an emergency department in Los Angeles County with new-onset chest pain. We sought to determine the extent of underuse and overuse of diagnostic testing for coronary artery disease in the evaluation of chest pain, and whether such patient characteristics as gender, race or ethnicity, socioeconomic status, and health-insurance status were associated with the underuse or overuse of such diagnostic testing.

      Methods

      Criteria development

      We used the previously described RAND/University of California, Los Angeles (UCLA), method to develop highly specific and clinically relevant criteria for the determination of appropriateness and necessity (
      • Park R.E
      • Fink A
      • Brook R.H
      • et al.
      Physician ratings of appropriate indications for six medical and surgical procedures.
      ,

      Chassin MR, Kosecoff J, Park RE, et al. Indications for selected medical and surgical procedures: a literature review and ratings of appropriateness: coronary artery bypass surgery. Publication R-3204/2-CWF-HF-HCFA-PMT-RWJ. Santa Monica: RAND Corporation, 1986.

      ,
      • Shekelle P.G
      • Schriger D.L
      Evaluating the use of the appropriateness method in the Agency for Health Care Policy and Research Clinical Practice Guideline Development Process.
      ,
      • Coulter I
      • Adams A
      • Shekelle P
      Impact of varying panel membership on ratings of appropriateness in consensus panels a comparison of a multi- and single disciplinary panel.
      ). A comprehensive review was performed of English-language peer-reviewed published papers related to the use of diagnostic testing for coronary artery disease. Using information from this review and consultations with cardiologists, we developed a list of 723 indications for the use of diagnostic testing for coronary artery disease in patients presenting with acute chest pain. These explicit indications encompassed eight aspects of the clinical presentation: the age and gender of the patient, the character and duration of the pain, the time interval between pain onset and emergency department presentation, the number of risk factors for coronary artery disease, electrocardiographic abnormalities, and whether the patient had an occupation where public safety would likely be jeopardized by an alteration in consciousness. These indications were grouped into four subjective categories according to acuity of symptoms and likelihood of coronary artery disease: unstable angina, or either at high, moderate, or low probability of having coronary artery disease.
      An expert panel of Los Angeles County physicians consisting of 4 cardiologists, 3 general internists, and 2 family practitioners was convened to rate these indications. Indications were rated using a two-stage modified Delphi RAND/UCLA rating on a 1 (highly inappropriate) to 9 (highly appropriate) scale (

      Brook RH. The RAND/UCLA appropriateness method. In: McCormick KA, Moore SR, Siegel RA, eds. Methodology Perspectives. Rockville, MD: Agency for Health Care Policy and Research, Public Health Service, U.S. Department of Health and Human Services, 1995:59–70; AHCPR Publication 95-0009.

      ). Panelists rated an indication appropriate (a score of 7, 8, or 9) if its benefits outweighed its risks and there were no preferable alternatives. Panelists rated indications inappropriate (a score of 1, 2, or 3, meaning that diagnostic testing for coronary artery disease should not be performed) if the benefits did not outweigh the inherent risks, or uncertain (a score of 4, 5, or 6). Indications found to be appropriate were rated by the expert panel at a second meeting for necessity again on a 1 (definitely not needed) to 9 (absolutely needed) scale. Necessity was defined as the obligation of a physician to recommend diagnostic testing for coronary artery disease because it has a high probability of significant benefit, that is, it is clearly the best option for the patient. An indication was classified as necessary if the median panel rating of necessity was 7, 8, or 9. At the completion of the expert panel process, each of the 723 indications was classified as necessary, appropriate but not necessary, uncertain, or inappropriate. Underuse was defined as the failure to provide diagnostic testing for coronary artery disease for a patient with a necessary indication. Overuse was defined as the provision of diagnostic testing for coronary artery disease when the indication was inappropriate. Diagnostic testing was considered less than appropriate when the indication was either inappropriate or uncertain.
      An example of an inappropriate indication for the use of diagnostic testing would be a 50-year-old woman with sharp, right-sided pleuritic chest pain lasting 2 to 3 seconds, whose pain began >48 hours before emergency room evaluation, who had a normal electrocardiogram, no cardiac risk factors, and a low-risk occupation (defined as one in which public safety would not be jeopardized if the patient experienced a sudden cardiac event, such as syncope). If this subject received diagnostic cardiac testing, it would be an example of overuse. Uncertain indications included a 60-year-old man with atypical chest pain whose electrocardiogram had a single premature atrial complex but no changes consistent with acute or previous ischemia, who had no cardiac risk factors, and a low-risk occupation. An appropriate but not necessary indication for diagnostic testing for coronary artery disease would be a 65-year-old woman with squeezing retrosternal pain of >1 minute duration that radiated to her left shoulder but with no history of pain with exertion, relief from nitrates, or ischemic changes on her electrocardiogram, and one cardiac risk factor. Necessary indications included a man with >1 minute of exertional left chest pressure that radiated to his left arm that was associated with ischemic electrocardiographic changes. If diagnostic testing for coronary artery disease was not provided to this subject, it would constitute underuse. Additional examples of the indication classification scheme are presented in the Appendix.

      Sample

      Five urban Los Angeles–area hospitals (two public, one private not-for-profit, one university medical center, and one not-for-profit health maintenance organization [HMO] facility), whose emergency departments treated approximately 180,000 adults per year, participated in the study. Each hospital performed both noninvasive and invasive testing for the diagnosis of coronary artery disease on site. The sample consisted of all patients meeting the age criteria (men 40 to 75 years and women 50 to 75 years) who were listed on emergency department logs of these hospitals with a primary complaint of chest pain (including the following synonyms: angina, unstable angina, rule-out myocardial infarction) from October 1994 through March 1996. Patients were excluded if they 1) had a diagnosis of chest pain due to a noncardiac cause at the time of the emergency department evaluation; 2) had a diagnosis of acute myocardial infarction upon discharge; 3) had a history of known cardiac disease or had received a previous diagnostic evaluation for ischemic heart disease; or 4) failed to receive an electrocardiogram as part of their initial evaluation. Emergency department logs and clinical records at each facility were reviewed weekly by specially trained abstractors to identify all patients meeting eligibility criteria.
      Eligible patients were mailed a questionnaire (written in either English or Spanish at an eighth-grade reading level) and accompanying consent form 4 weeks after their emergency department presentation. The questionnaire asked if they had received any diagnostic testing for coronary artery disease (including the type received) and included items that described aspects of the care they received and their socioeconomic and demographic characteristics. Examples of items include the following: “Did you have a treadmill test or stress test at any time after your visit to the emergency room on _____?” (preceded by detailed schematic and verbal descriptions of exercise stress testing); “Did you have an angiogram any time after your visit to the emergency room on _____?” (preceded by detailed schematic and verbal descriptions of an angiogram); “What kind of health insurance do you currently have?” (followed by eight categories); “How much school did you have?” (followed by seven categories from none, eighth grade or less, to a professional degree); “What language do you usually speak?” (followed by English, Spanish, or other).
      Efforts to contact nonrespondents were made by use of reminder letters, postcards, and repeated telephone calls by a bilingual research assistant.
      Receipt of informed consent was required by institutional review boards and participating hospitals before medical record abstraction. Abstracted information was reviewed for accuracy by project physicians (kappa = 0.80). Explicit clinical criteria were then used to assign to each patient one of the 723 indications previously described. Medical records were also reviewed for evidence of any recommendation that the patient undergo further cardiac testing.

      Analytic methods

      Two analytic methods were used to evaluate the association between predictor variables and the necessary and appropriate use of diagnostic cardiac testing. Chi-square and Fisher’s exact tests were used to examine bivariate associations between categorical variables. Multiple logistic regression analysis was then performed to evaluate whether specific variables of interest (education level, insurance status, and race/ethnicity) were independent predictors of underuse or overuse. For these multivariate comparisons, education level was collapsed into college/no college, insurance status into insured/uninsured, and race/ethnicity into nonminority/minority (African-American or Latino). Goodness of fit was assessed with the Hosmer-Lemeshow statistic and model performance with the c statistic and Somers’ D. Sensitivity analysis was performed to assess the resilience of significant bivariate associations.

      Results

      During the 18-month enrollment period, 12,430 patients with a complaint of chest pain were evaluated in the participating emergency departments. Of these, 6,764 patients were not eligible for the study because they were either too old or too young. An additional 4,931 patients were excluded because they were subsequently diagnosed with myocardial infarction, had a previous diagnosis of cardiac disease, or had undergone previous diagnostic testing to evaluate the possibility of coronary artery disease.
      Each of the remaining 735 potential subjects was mailed a packet including the questionnaire and the consent form; 174 (24%) subjects could not be contacted due to inaccurate addresses or telephone numbers (including 6 patients who were determined to have died during the follow-up period). Of the 561 patients who responded, 63% provided consent to participate in the study, resulting in a final study sample of 356 patients.
      Similar numbers of men and women participated; more than half were members of ethnic minority groups (Table 1
      Table 1Demographic and Socioeconomic Characteristics of the 356 Patients
      HMO = health maintenance organization.
      NumberPercent
      Female gender17950
      Race/ethnicity
      White15543
      African-American9627
      Latino6719
      Asian or Pacific-Islander319
      Age (years)
      40–496217
      50–6424569
      65–754914
      High school education or less15544
      Payer
      Medicaid185
      Medicare288
      HMO10529
      Private5014
      None14942
      legend HMO = health maintenance organization.

      Acknowledgements

      We would like to thank Janet Stone, Susan Diaz, and Drs. Edd Post and James T. Niemann for their critical contributions to this project.

      Appendix.

      Examples of indications for cardiac diagnostic testing among patients presenting to an emergency department with a complaint of chest pain

      Necessary
      Complete appropriateness and necessity indications are available from authors upon request.
      Appropriate but Not NecessaryUncertain AppropriatenessInappropriate
      A 70-year-old man with increasingly frequent chest tightness on mild exertion of >10 minutes duration radiating to his left shoulder and T-wave inversion on his electrocardiogram.A 53-year-old woman with no cardiac risk factors with left-sided chest pain of >1 minute duration that radiated to her jaw occurring 3 days ago and nonspecific T wave changes on her electrocardiogram.A 51-year-old woman with 3 seconds of substernal pressure who has diabetes and hypertension and a normal electrocardiogram.A 41-year-old man with 2 hours of left chest pain while lying in bed not relieved by nitroglycerin, a normal electrocardiogram, and no cardiac risk factors.
      A 65-year-old diabetic woman with new-onset sharp mid-chest pain at rest, left ventricular hypertrophy on electrocardiogram, and who smokes cigarettes.A 60-year-old hypertensive woman with substernal pain lasting <1 minute while shopping who has an old Q wave on her electrocardiogram.A 64-year-old male smoker with new burning right chest discomfort during intercourse, and nonspecific ST segment changes.A 50-year-old woman with right-sided pleuritic pain while walking, with a normal electrocardiogram and no cardiac risk factors.
      A 40-year-old man with a history of hypertension and a family history of ischemic heart disease who has pressure-like pain while jogging 3 days previously and a normal electrocardiogram.A 71-year-old woman with mid-chest pain relieved by a friend’s nitroglycerin 1 week previously with a normal electrocardiogram and no cardiac risk factors.A 66-year-old woman with upper back pain while walking, a normal electrocardiogram, and diabetes.A 42-year-old male runner with 30 seconds of sharp left chest pain while competing, a normal electrocardiogram, and no cardiac risk factors.
      A 50-year-old female pilot who smokes and has a family history of heart disease who has 30 seconds of nonradiating retrosternal pain while playing basketball and has Q waves on her electrocardiogram.A 64-year-old male custodian with right-sided pain while mopping floors 1 week previously, high blood pressure, and inverted T waves on his electrocardiogram.A 49-year-old man with right-sided chest pain while watching TV the previous weekend, no cardiac risk factors and nonspecific ST segment or T-wave findings on his electrocardiogram.A 65-year-old woman with high blood pressure who recently had 10 seconds of gas-like left chest pain and a normal electrocardiogram.
      Complete appropriateness and necessity indications are available from authors upon request.
      Their mean (± SD) age was 56 ± 8 years. Slightly more than half reported some college or post–high school education. Three quarters of the patients with health insurance had some form of private insurance.
      Almost 59% of patients were hospitalized immediately after their emergency department evaluation. Slightly more than half (56%) had normal electrocardiograms, 22% had nonspecific electrocardiographic changes (such as left ventricular hypertrophy or nonspecific ST-segment or T-wave changes), and 22% had electrocardiographic changes consistent with active or previous coronary artery disease (including T-wave inversion, poor R-wave progression, and Q waves). Thirteen percent of patients were judged to have unstable angina, 13% were thought to be at high probability of having coronary artery disease, 28% at moderate probability, and 48% at low probability.
      Of the 356 patients, 60% received a test to diagnose or exclude coronary artery disease: 43% received one or more noninvasive cardiac stress tests, 12% received both a noninvasive stress test and angiography, and 5% were evaluated by angiography alone. Sixty-five percent of all tests and 80% of the coronary angiograms were performed during the emergency department visit or the resulting hospitalization.

      Overall prevalence of underuse and overuse of testing for coronary artery disease

      Fifty-one percent of subjects (181 of 356) met the criteria for necessary diagnostic testing for coronary artery disease (Table 2). Of those meeting necessity criteria, 78% received one or more necessary diagnostic tests, and 22% did not (underuse). Of the 215 patients who underwent diagnostic testing, only 7 (3%) received testing for indications judged inappropriate by our expert panel criteria (overuse). Of all subjects who underwent diagnostic testing, indications were less than appropriate (inappropriate or uncertain) in 23%.
      Table 2Use of Diagnostic Testing by Indication Category
      Necessary = an indication for which testing is both appropriate and necessary. Appropriate = an appropriate but not necessary indication for testing. Uncertain = an indication for which testing is neither appropriate nor inappropriate. Inappropriate = an indication for which testing is inappropriate.
      NecessaryAppropriateUncertainInappropriate
      Number of patients in category1814310923
      Number who received testing (%)141 (78%)25 (58%)42 (39%)7 (30%)
      Necessary = an indication for which testing is both appropriate and necessary. Appropriate = an appropriate but not necessary indication for testing. Uncertain = an indication for which testing is neither appropriate nor inappropriate. Inappropriate = an indication for which testing is inappropriate.

      Demographic and socioeconomic factors related to underuse and overuse of testing

      Patients without post–high school education (Table 3) were significantly more likely than their counterparts to fail to receive needed testing (30% vs 15%, P = 0.01). Similarly, patients without health insurance were more than twice as likely as those with health insurance (34% vs 15%, P <.01) not to receive necessary diagnostic testing. Among insured patients, there were significant differences by type of health insurance. HMO enrollees, the privately insured, and Medicare beneficiaries had a lower rate of underuse (a weighted average of 11%) compared with Medicaid recipients (42%, P = 0.02), although the number of Medicaid recipients was quite small (n = 12). Underuse rates were also higher among African-American and Latino patients but lower among Asian and Pacific-Islanders compared with white patients, and those younger than 65 years, but these differences were not statistically significant. All overuse occurred among those with at least some post–high school education (Table 4). No significant differences in rates of underuse or overuse were observed among the five participating hospitals.
      Table 3Underuse of Diagnostic Testing, by Selected Patient Characteristics
      Abbreviation as in Table 1.
      Number of Subjects in Whom a Test Was NecessaryPercent of Subjects with a Necessary Indication Who Were Not Tested
      Overall18122
      Gender
      Female9426
      Male8718
      Race/ethnicity
      White7818
      African-American5529
      Latino3526
      Asian or Pacific-Islander128
      Age, years
      40–492623
      50–6412325
      65–74329
      Education
      No college8630
      Some college9515
      P <0.05 for comparison of no college versus some college, and for comparison of Medicaid vs other insurance.
      Insurance type
      Medicaid1242
      P <0.05 for comparison of no college versus some college, and for comparison of Medicaid vs other insurance.
      Medicare1613
      HMO6611
      Private1513
      Insured
      No6834
      P <0.01 for comparison of insured versus noninsured.
      Yes11115
      P <0.05 for comparison of no college versus some college, and for comparison of Medicaid vs other insurance.
      P <0.01 for comparison of insured versus noninsured.
      legend Abbreviation as in Table 1.
      Table 4Nonappropriate Use of Diagnostic Testing, by Patient Characteristics
      Abbreviation as in Table 1.
      Number of Subjects TestedOveruse (percent of those tested in whom the test was inappropriate)Uncertain Use (percent of those tested in whom the test was uncertain)Less Than Appropriate Use (percent of those tested in whom the test was inappropriate or uncertain)
      Overall21532023
      Gender
      Female10722022
      Male10851823
      Race/ethnicity
      White9822123
      African-American5341923
      Latino3701111
      Asian or Pacific-Islander2343135
      Age, years
      40–492901414
      50–6414932326
      65–743751419
      Education
      No college8601616
      Some college12952227
      Insurance type
      Medicaid1003333
      Medicare21101424
      HMO8252121
      Private2502828
      Insured
      No7212324
      Yes14341822
      legend Abbreviation as in Table 1.

      Multivariate analysis

      In logistic regression models that included education, insurance status, minority status, gender, and age, only level of education was associated with underuse, or inappropriate use, of diagnostic testing. Underuse was more likely to occur among those without college education (odds ratio 2.2, 95% confidence interval 1.0 to 4.4; Table 5). Similarly, those with college education were more likely to have less than appropriate use of diagnostic testing (odds ratio 2.1, 95% confidence interval 1.1 to 4.3).
      Table 5Multivariate Associations between Selected Patient Characteristics and Underuse or Overuse (nonappropriate use)
      Underuse = not receiving a necessary test. Nonappropriate use = receiving a test for an inappropriate or uncertain indication. College = some college or vocational post–high school education. Minority = African-American and Latino patients.
      Odds Ratio (95% confidence interval)
      Adjusted for age and gender.
      Underuse
      No college2.2 (1.0–4.4)
      Uninsured1.4 (0.7–2.9)
      Minority1.5 (0.7–3.2)
      Nonappropriate use
      College2.1 (1.1–4.3)
      Insured0.9 (0.5–1.8)
      Nonminority1.2 (0.6–2.4)
      Underuse = not receiving a necessary test. Nonappropriate use = receiving a test for an inappropriate or uncertain indication. College = some college or vocational post–high school education. Minority = African-American and Latino patients.
      Adjusted for age and gender.

      Potential causes of underuse

      Among the 40 patients who met necessity criteria but were not tested, the medical records of 27 (68%) contained no recommendation or order for additional diagnostic testing. Of the other 13 patients, 9 spoke either Spanish or another non-English language as their primary language. Underuse of diagnostic testing was more common among the 14% of patients meeting necessity criteria who were not admitted to the hospital (60% compared with 16% for those who were hospitalized immediately after their emergency room evaluation, P <0.01). No emergency room or inpatient records documented that patients refused diagnostic testing when it was recommended to them.

      Sensitivity analysis

      To evaluate the importance of potential bias introduced by the response rate, sensitivity analysis was performed to assess the strength of the association between underuse and educational level. We found that differences in the rates of underuse by education level would have to be substantially different among nonrespondents from those among respondents to render our findings nonsignificant. For example, the rate of underuse among non-college-educated nonrespondents would have to have been <13% (compared with the 30% rate among similar respondents) to eliminate the significance of the difference in underuse between the educational-level groups. Alternatively, underuse among college-educated nonrespondents would have to have been >38%, compared with the 15% rate we found among respondents.

      Discussion

      Although disparities in the use of cardiac procedures have been observed for low socioeconomic status, uninsured, ethnic minority, and female patients, most reports have focused on comparative rates of invasive procedure use rather than underuse of needed procedures. We studied overuse and underuse of diagnostic procedures at the time that patients presented to an emergency department with a potentially serious disease. Evaluation at this stage of care is important, because inappropriate use of diagnostic testing can initiate a “therapeutic cascade” that compounds the extent of improper treatment (
      • Verrilli D
      • Welch H.G
      The impact of diagnostic testing on therapeutic interventions.
      ).
      We found a large amount of underuse and comparatively little overuse. The extent of underuse (22% of patients who needed diagnostic tests did not receive them) is particularly alarming. In these patients, the diagnosis of coronary artery disease may lead to treatment that is life saving. Failure to receive needed testing means that patients will not receive a proper diagnosis or will receive it at a later time, perhaps after further progression of disease. They may also fail to receive needed care in a timely fashion. On the other hand, we found a low rate of overuse among those who did receive diagnostic testing, suggesting that few resources were wasted in the diagnostic evaluation of patients in our sample. These results on overuse are similar to those for invasive cardiovascular procedures performed in New York State (
      • Leape L.L
      • Hilborne L.H
      • Park R.E
      • et al.
      The appropriateness of use of coronary artery bypass graft surgery in New York State.
      ,
      • Hilborne L.H
      • Leape L.L
      • Bernstein S.J
      • et al.
      The appropriateness of use of percutaneous transluminal coronary angioplasty in New York State.
      ,
      • Bernstein S.J
      • Hilborne L.H
      • Leape L.L
      • et al.
      The appropriateness of use of coronary angiography in New York state.
      ).
      Among the demographic characteristics that we studied, level of education, but not insurance status, was an independent predictor of underuse. Lower educational level may be a proxy for the difficulty that patients have in paying for health care, including diagnostic testing. It may also be a marker for other factors associated with testing, such as income, understanding of complicated medical conditions, and access to alternative provider opinions, legal interventions, or appeal processes. These findings suggest that universal health insurance may not eliminate disparities in use of health-care services related to socioeconomic status.
      Our results suggest that a lack of recognition of the need for diagnostic testing by clinicians may cause underuse. Although it is possible that physicians make recommendations for further testing without recording them in the medical record, it seems likely that at least some physicians failed to make such recommendations. In addition, communication barriers between the physician and the patient may contribute even when such recommendations do occur. Impaired access to outpatient diagnostic testing after emergency room discharge may also be important. Patient refusal to receive tests, whether due to cultural or other factors, appears to be less important as a cause of underuse.
      African-Americans and Latinos had higher rates of underuse when compared with white patients. However, these trends did not achieve statistical significance even when African-Americans and Latinos were aggregated. Any racial or ethnic disparities in test utilization appear to be overshadowed by the greater contributions of educational status in this analysis.
      Our findings should be interpreted with caution for several reasons. First, although the participating hospitals were recruited to reflect the diversity of delivery systems in the Los Angeles area and to include a large number of HMO patients, it is possible that different results would have been observed if patients were recruited from other Los Angeles hospitals or from a more diverse geographic area. A second limitation is that our study was confined to patients who presented to hospital emergency departments. Their care may differ from that of patients presenting for chest pain evaluation to physician offices or outpatient clinics. Third, many potential subjects did not respond to our invitation to participate in this study. This may be attributable to the additional respondent burden imposed by requiring eligible patients to provide consent for review of their medical records in addition to completing a mailed questionnaire. At the same time, our sensitivity analysis demonstrated that the results would have had to have been substantially different among nonrespondents to change the overall findings. Finally, our multivariate analysis of the association between socioeconomic factors and underuse may have been limited by colinearity among education, health-insurance status, and patient race or ethnicity.
      Nonetheless, our findings suggest that there may be a substantial number of patients with undiagnosed and untreated coronary artery disease even after they have sought emergency department and hospital-based medical care for chest pain. This problem may be especially important among patients with lower educational levels. These findings underscore the failure of the US health-care system to ensure the provision of medical care for all patients who may need it.

      References

        • Carlisle D.M
        • Valdez R.B
        • Shapiro M.F
        • Brook R.H
        Geographic variation in rates of selected surgical procedures within Los Angeles County.
        Health Serv Res. 1995; 30: 27-42
        • Leape L.L
        • Park R.E
        • Solomon D.H
        • Chassin M.R
        • Kosecoff J
        • Brook R.H
        Does inappropriate use explain small area variations in the use of health care services?.
        JAMA. 1990; 263: 669-672
        • Wenneker M.R
        • Epstein A.M
        Racial inequalities in the use of procedures for patients with ischemic heart disease in Massachusetts.
        JAMA. 1989; 261: 253-257
        • Ayanian J.Z
        • Epstein A.M
        Differences in the use of procedures between women and men hospitalized for coronary heart disease.
        NEJM. 1991; 325: 221-225
        • Ayanian J.Z
        • Udvarhelyi S
        • Gatsonis C.A
        • Pashos C.L
        • Epstein A.M
        Racial differences in the use of revascularization procedures after coronary angiography.
        JAMA. 1993; 269: 2642-2646
        • Carlisle D.M
        • Leake B.D
        • Shapiro M.F
        Racial and ethnic differences in the use of invasive cardiac procedures among cardiac patients in Los Angeles county, 1986–1988.
        Am J Public Health. 1995; 85: 352-356
        • Shaw L.J
        • Miller D.D
        • Romeis J.C
        • Kargl D
        • Younis L.T
        • Chaitman B.R
        Gender differences in the non-invasive evaluation and management of patients with suspected coronary artery disease.
        Ann Intern Med. 1994; 120: 559-566
        • Ayanian J.Z
        • Epstein A.M
        Differences in the use of procedures between women and men hospitalized for coronary heart disease.
        NEJM. 1991; 325: 221-225
        • Langa K.M
        • Sussman E.J
        The effect of cost-containment policies on rates of coronary revascularization in California.
        NEJM. 1993; 329: 1784-1789
        • Bombardier C
        • Fuchs V.R
        • Lillard L.A
        • Warner K.E
        Socioeconomic factors affecting the utilization of surgical operations.
        NEJM. 1977; 297: 699-705
        • Gittelsohn A.M
        • Halpern J
        • Sanchez R.L
        Income, race, and surgery in Maryland.
        Am J Public Health. 1991; 81: 1435-1441
        • Wennberg J
        • Gittelsohn A
        Variations in medical care among small areas.
        Sci Am. 1982; 246: 120-134
        • Leape L.L
        • Hilborne L.H
        • Park R.E
        • et al.
        The appropriateness of use of coronary artery bypass graft surgery in New York State.
        JAMA. 1993; 269: 753-760
        • Hilborne L.H
        • Leape L.L
        • Bernstein S.J
        • et al.
        The appropriateness of use of percutaneous transluminal coronary angioplasty in New York State.
        JAMA. 1993; 269: 761-765
        • Bernstein S.J
        • Hilborne L.H
        • Leape L.L
        • et al.
        The appropriateness of use of coronary angiography in New York state.
        JAMA. 1993; 269: 766-769
        • Borowsky S.J
        • Kravitz R.L
        • Laouri M
        • et al.
        Effect of physician specialty on use of necessary coronary angiography.
        J Am Coll Cardiol. 1995; 26: 1484-1491
        • Laouri M
        • Kravitz R.L
        • French W.J
        • et al.
        Underuse of coronary revascularization procedures; application of a clinical method.
        J Am Coll Cardiol. 1997; 29: 891-897
        • Park R.E
        • Fink A
        • Brook R.H
        • et al.
        Physician ratings of appropriate indications for six medical and surgical procedures.
        Am J Public Health. 1986; 76: 766-772
      1. Chassin MR, Kosecoff J, Park RE, et al. Indications for selected medical and surgical procedures: a literature review and ratings of appropriateness: coronary artery bypass surgery. Publication R-3204/2-CWF-HF-HCFA-PMT-RWJ. Santa Monica: RAND Corporation, 1986.

        • Shekelle P.G
        • Schriger D.L
        Evaluating the use of the appropriateness method in the Agency for Health Care Policy and Research Clinical Practice Guideline Development Process.
        Health Serv Res. 1996; 31: 453-468
        • Coulter I
        • Adams A
        • Shekelle P
        Impact of varying panel membership on ratings of appropriateness in consensus panels.
        Health Serv Res. 1995; 30: 577-591
      2. Brook RH. The RAND/UCLA appropriateness method. In: McCormick KA, Moore SR, Siegel RA, eds. Methodology Perspectives. Rockville, MD: Agency for Health Care Policy and Research, Public Health Service, U.S. Department of Health and Human Services, 1995:59–70; AHCPR Publication 95-0009.

        • Verrilli D
        • Welch H.G
        The impact of diagnostic testing on therapeutic interventions.
        JAMA. 1996; 275: 1189-1191