|
 |
|
|||||||
|
|||||||||
© 2003 American Public Health Association
RESEARCH AND PRACTICE |
Lisa M. Lee, J. Stan Lehman, and Patricia L. Fleming are with the Division of HIV/AIDS Prevention—Surveillance and Epidemiology, Centers for Disease Control and Prevention, Atlanta, Ga. Andrew B. Bindman is with the Department of Epidemiology and Biostatistics, University of California, San Francisco.
Correspondence: Reprint requests should be sent to the Office of Communications, National Center for HIV, STD, and TB Prevention, Centers for Disease Control and Prevention, Mail Stop E-07, Atlanta, GA 30333.
 |
INTRODUCTION |
|---|
 TOP INTRODUCTION METHODSRESULTSCONCLUSIONSReferences |
|---|
Since 1981, the national HIV/AIDS reporting system (HARS) has provided data to track the progression of the AIDS epidemic, detect patterns of transmission, assess prevention programs, provide an epidemiological basis for planning, and allocate federal resources.1 The Centers for Disease Control and Prevention actively evaluate the quality of the data to ensure that the objectives of the system are based on accurate and complete information.2–5
The HARS relies on medical record reviews by health care providers or trained health department personnel for completion of case reports. Validity studies of medical record data have shown that accuracy and reliability vary according to the type of information and the diagnoses examined.2,4,6 To assess the accuracy of 2 HARS variables on which data are frequently stratified, we compared HARS data on race/ethnicity and transmission mode to self-reported data collected during a survey of people with AIDS.
|
METHODS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSCONCLUSIONSReferences |
|---|
We used data from the AIDS Patient Survey (APS), a study conducted in Arizona, Colorado, Mississippi, Missouri, New Mexico, North Carolina, Oregon, and Texas.7 The survey was designed to study the association between HIV reporting policies and timeliness of testing and medical care. Because the APS data on race and transmission mode were collected and maintained independently from the HARS, we used these sources to assess the concordance between 2 methods of data collection (self-report in the APS and medical record review in the HARS). Stratifying on transmission mode, we sampled 3321 persons reported with AIDS from May 1995 through December 1996, aged at least 18 years, whose AIDS diagnosis was within 12 months before the report to the health department. We excluded 520 persons due to age, death, or report date errors. Of 2801 eligible persons, 1913 (68.3%) were interviewed; 6 interviews were incomplete, leaving 1907 for analysis.
In the APS, questions about race and ethnicity were asked separately, and reporting multiple races was possible; in the HARS, race/ethnicity comprised mutually exclusive categories of White non-Hispanic, Black non-Hispanic, Hispanic, Asian/Pacific Islander, and American Indian/Alaska Native. We combined categories in the APS, assigning persons self-reporting Hispanic ethnicity into the Hispanic category and all others to their self-reported racial category. Persons reporting multiple races maintained multirace designations. In the APS, the mode of transmission was ascertained from a series of questions about risks for HIV acquisition; reporting of multiple modes was possible. In the HARS, persons reporting multiple modes of transmission were assigned to a hierarchy reflecting the most probable mode. The hierarchy included male-to-male sexual contact (abbreviated here as MSM), injection drug use (IDU), MSM/IDU, heterosexual contact (HC) with a person with known risk for or infected with HIV, and Other. Multiple modes of transmission were incompletely ascertained in the HARS. Therefore, we combined responses to APS questions to reflect the HARS hierarchy.
We calculated the proportion of the sample whose classification in the HARS matched their selfreport in the APS. Because we considered selfreported race/ethnicity in the APS to be the gold standard,8 we calculated the Youden J statistic (J),9 a percent agreement measure that gives equal weight to positive and negative matches, and its 95% confidence interval adjusted for multiple comparisons. Because the gold standard for the mode of transmission is unknown, we assessed agreement beyond chance using Cohen’s 
statistic10 and its 95% confidence interval.
|
RESULTS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSCONCLUSIONSReferences |
|---|
Of 1907 persons interviewed, 83% were male; 54% reported White race, 34% Black, 10% Hispanic, 1.7% American Indian/Alaska Native, 0.3% Asian/Pacific Islander; 49% reported MSM, 13% IDU, 9% MSM/IDU, 23% HC, and 6% Other.
Of 1010 persons self-reporting White race, 999 (98.9%) were classified as White in the HARS (J = 0.92, 95% confidence interval [CI] = 0.90, 0.95, Table 1
). Of 633 persons self-reporting Black race, 623 (98.4%) were classified as Black in the HARS (J = 0.97, 95% CI = 0.95, 0.98). Of 192 persons selfreporting Hispanic ethnicity, 165 (86%) were so classified in the HARS (J = 0.85, 95% CI = 0.78, 0.91). Agreement was lower among Asian/Pacific Islander and American Indian/Alaska Native (2 of 6 [33%], J = 0.33, 95% CI = 0, 0.83, and 11 of 31 [35%], J = 0.35, 95% CI = 0.13, 0.57, respectively). Of 34 persons reporting multiple races, 33 (97.1%) were classified in the HARS into 1 of their selfreported categories.
|
Of 927 men self-reporting MSM, 92% were so classified in the HARS (Table 2
); of 168 men reporting IDU, 76% were so classified in the HARS; of 174 self-reporting MSM/IDU, 55% were so classified in the HARS, and an additional 40% were classified as either MSM or IDU but not both; of 302 men reporting HC, 29% were so classified in the HARS, and most of the remaining were classified as Undetermined. We observed a of 0.57 (95% CI = 0.53, 0.60) on agreement of transmission mode among men.
|
Of 135 women self-reporting HC, 77% were classified as HC in the HARS; of 75 women reporting IDU, 91% were so classified in the HARS. Of 108 women reporting Undetermined/Other risk (including heterosexual sexual contact in the absence of partners’ risks), 29.6% were classified as Undetermined/Other and 63.9% as HC in the HARS (
= 0.44; 95% CI = 0.36, 0.52). |
CONCLUSIONS |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSCONCLUSIONSReferences |
|---|
Self-reported and AIDS surveillance system classification of race/ethnicity and transmission mode agreed well in larger groups. The data from smaller groups and the HC group had poorer agreement. Among men, agreement was moderate for transmission mode, in part because some men self-reported dual modes but were classified in the HARS as one or the other, not both.
Our findings of high concordance of transmission mode among those reporting a mode other than HC are consistent with those of other studies.2,4 HC is difficult to validate, in part because both male-tofemale sexual contact and knowledge of a partner’s transmission mode or sexual contact with a known seropositive partner are required for HC classification. Because of the difficulty of ascertaining partners’ risks, we need a standard definition of high-risk heterosexual sexual acts to use as a proxy for increased probability of heterosexual transmission in the absence of knowledge of partners’ risks. As in an earlier study,2 agreement on race/ethnicity was very good for Blacks and Whites and good for Hispanics. We found the poorest agreement among other racial/ethnic groups, but the numbers were small, yielding unstable estimates.
People who did not complete the in-person interview were more likely to be reported in the HARS as Hispanic or Asian/Pacific Islander race/ethnicity and IDU or Undetermined transmission mode compared with those who completed interviews. This potentially limits generalizability. Additionally, small numbers in certain groups limit inferences.
Accurate demographic data are critical for assessing epidemiological trends and accurately addressing prevention and care needs of communities affected by HIV. To get the best information for less-frequent race/ethnicity and transmission mode groups, we need multiple strategies, including medical record documentation, interviews, self-administered questionnaires, and matching to other data sources that provide services to specified populations. These strategies will become increasingly important as the epidemic continues to shift to racial/ethnic minorities.
|
Acknowledgments |
|---|
This analysis was presented at the 2001 National HIV Prevention Conference, August 12–15, 2001, Atlanta, Ga.
Members of the Multistate Evaluation of Surveillance for HIV Study Group are Andrew B. Bindman, MD, Frederick M. Hecht, MD, Dennis Osmond, PhD, Karen Vranizan, MA, Dennis Keane, MPH (University of California San Francisco); John Ward, MD, MPH, Patricia L. Fleming, PhD (Centers for Disease Control and Prevention, Atlanta, Ga); Denise K. Boyd, MS, MPH, Vjollca Berisha, MD, MPH (Arizona Public Health Department); Kenneth Gershman, MD, MPH, Melanie Mattson (Colorado Public Health Department); John Newman, Craig Thompson (Mississippi Public Health Department); Robert Hamm, MD, MPH, Kristen Wendt, MPH, Linda Bell (Missouri Public Health Department); Michael Samuel, DrPH, Mark Stenger, MS (New Mexico Public Health Department); Delbert E. Williams, PhD, Evelyn Foust, MPH, Judy Owen-O’Dowd (North Carolina Public Health Department); Steven Modesitt, RN, MPH, Roger Wirt, PhD, David Fleming, MD (Oregon Public Health Department); Ann S. Robbins, PhD, Sharon A. King, MA, Douglas Hamaker (Texas Public Health Department).
Human Participation Protection
This study was approved by the University of California institutional review board and state and local review boards as required.
|
Footnotes |
|---|
Contributors
L. M. Lee posed the research question, analyzed the data, and wrote the brief. J. S. Lehman, A. B. Bindman, and P. L. Fleming designed the study, including the data collection instrument, and reviewed and edited the brief.
Accepted for publication November 4, 2002.
|
References |
|---|
|
TOPINTRODUCTIONMETHODSRESULTSCONCLUSIONSReferences |
|---|
1. Centers for Disease Control and Prevention. Guidelines for national human immunodeficiency virus case surveillance, including monitoring for human immunodeficiency virus infection and acquired immunodeficiency syndrome. MMWR Morbid Mortal Wkly Rep. 1999;48(RR13):1–31.
2. Klevens RM, Fleming PL, Li J, et al. The completeness, validity, and timeliness of AIDS surveillance data. Ann Epidemiol. 2001;11:443–449.[Web of Science][Medline]
3. Jara MM, Gallagher KM, Schieman S. Estimation of completeness of AIDS case reporting in Massachusetts. Epidemiology. 2000;11:209–213.[Web of Science][Medline]
4. Klevens RM, Fleming PL, Neal JJ, Li J, the Mode of Transmission Validation Study Group. Is there really a heterosexual AIDS epidemic in the United States? Findings from a multisite validation study, 1992–1995. Am J Epidemiol. 1999;149:75–84.
5. Buehler JW, Berkelman RL, Stehr-Green JK. The completeness of AIDS surveillance. J Acquir Immune Defic Syndr. 1992;5:257–264.
6. Klungel OH, de Boer A, Paes AH, Seidell JC, Bakker A. Cardiovascular diseases and risk factors in a population-based study in The Netherlands: agreement between questionnaire information and medical records. Netherlands J Med. 1999;55:177–183.[Web of Science][Medline]
7. Bindman AB, Osmond D, Hecht FM, et al. Multistate evaluation of anonymous HIV testing and access to medical care. JAMA. 1998;280:1416–1420.
8. Kaufman JS. How inconsistencies in racial classification demystify the race construct in public health statistics [editorial]. Epidemiology. 1999;10:101–103.[Web of Science][Medline]
9. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;32–35.
10. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20:37–46.[Web of Science]
This article has been cited by other articles:
|
|
 |
|
  L. Espinoza, H. I. Hall, F. Hardnett, R. M. Selik, Q. Ling, and L. M. Lee Characteristics of Persons With Heterosexually Acquired HIV Infection, United States 1999-2004 Am J Public Health, January 1, 2007; 97(1): 144 - 149. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |