BRIEF REPORT HIV, prevalence
Prevalence of HIV Antibody in a Noninner-City University Hospital Emergency Department Study hypothesis: To determine the prevalence of antibody to human i m m u n o d e f i c i e n c y virus (HIV) in trauma and nontrauma patients not identified as having known HIV infection in a noninner-city university teaching hospital emergency department, and to determine the frequency with which treating emergency physicians are knowledgeable of patients' risk factors for HIV infection. Population: ED patients between 18 and 59 years old with injuries that m e t trauma center triage criteria or with nontrauma-related illness who had blood drawn for physician-requested laboratory tests and for whom an extra aliquot of blood was available for HIV antibody testing. Methods: All serum samples were first tested for antibody to HIV by enzyme-linked immunosorbent assay. If positive, the specimen was retested. All repeatedly reactive specimens were analyzed by Western blot test. The treating physician completed a questionnaire regarding the patient's illness and risk factors for HIV infection after the patient's care was completed. Results: Two of 100 major trauma patients (confidence interval, 0% to 5%) and seven of 100 nontrauma patients (confidence interval, 2% to 12%) had antibody to HIV. The seropositive rate by age and clinical group varied from 0% to 12.5%, with the highest rates in the 30- to 39-year-old group of nontrauma patient s. The difference in proportions of seropositivity between the sexes was not statistically significant. Physicians obtained information regarding homosexual or bisexual behavior, IV drug use, and hemophilia from 52% of the nontrauma patients and only 17% of trauma patients. None of 30 trauma patients for whom data were available and only two of the 100 nontrauma patients gave a history of any high-risk behavior. Conclusion: Although the sampling technique we used has limitations, the prevalence of HIV infection in our noninner-city ED is similar to that recently reported from inner-city EDs. This is in contrast to previous reports of low rates of HIV infection among ED patients in nonurban settings. Physician assessment of risk factors was incomplete in the majority of our patients. Patients rarely acknowledged any high-risk behavior. It is essential that emergency health care workers take m a x i m u m diligence to prevent exposure to blood and other body fluids from all ED patients. [Baraft LJ, Talan DA, Torres M: Prevalence of HIV antibody in a noninner-city u n i v e r s i t y h o s p i t a l e m e r g e n c y d e p a r t m e n t . A n n Emerg M e d July 1991;20: 782- 786.]
Larry J Baraff, MD, FACEP* David A Talan, MD*t Mina Torres* Los Angeles, California From the Departments of Pediatrics* and Medicine,t UCLA School of Medicine, UCLA Medical Center; and Olive ViewUCLA Medical Center, Los Angeles, California. Received for publication July 23, 1990. Revisions received November 13, 1990, and February 5, 1991. Accepted for publication February 24, 1991. Address for reprints: Larry J Baraff, MD, FACER UCLA Emergency Medicine Center, 924 Westwood Boulevard, Suite 300, Los Angeles, California 90024.
INTRODUCTION Exposures to blood in the emergency department are numerous, especially when providing care to critically injured trauma patients. The majority of.ED patients are cared for without knowledge of their complete medical history, including their human immunodeficiency virus (HIV) antibody status. We have previously shown that ED health care workers often provide care without strict adherence to the universal precautions recommended by the Centers for Disease Control3, ~ It is estimated that between 800,000 and 1.2 million US citizens axe infected with HIV.3 The majority of these individuals are asymptomatic and do not know that they are infected. The rate of infection is greatest in
20:7 July 1991
Annals of Emergency Medicine
782/113
HIV A N T I B O D Y Baraff, Talan, & Torres
high-risk groups, which include homosexuals or bisexuals, IV drug users, hemophiliacs, and sexual partners of the aforementioned. 4 Previous articles have reported the seroprevalence of antibody to HIV in selected ED patients not known to be i n f e c t e d to range f r o m 2.1% to 8.9%. 5-9 The rate of seropositivity is greatest in men, adults between 20 and 45 years old, and victims of penetrating trauma, especially in eastcoast inner-city hospitals, where the infection rate has been reported to be as high as 19% in trauma victims between 25 and 35 years old.7, 8 We undertook this study to determine the prevalence of antibody to HIV in t r a u m a patients who m e t trauma center triage criteria and nontrauma ED patients not identified by their treating emergency physician as having k n o w n HIV infection. We hoped that by providing this information to ED health care workers and members of the trauma team at our institution we would improve compliance with universal precautions. We also sought to determine the c o m p l e t e n e s s of r i s k - a s s e s s m e n t practices of emergency physicians to determine if such assessment could be used to reliably detect the majority of HIV-infected ED patients. MATERIALS A N D M E T H O D S The UCLA Emergency Medicine Center is a Level I trauma center and comprehensive ED as defined by California state regulations. It serves a generally affluent suburban area of Los Angeles County. The insurance mix of our patients is as follows: private insurance, 12%; Medicare, 11%; Medicaid, 16%; contracts, 25%; and uninsured, 36%. The insurance mix of trauma patients is: private insurance, 23%; Medicare, 10%; Medicaid, 20%; contracts, 26%; and uninsured, 21%. In 1989, there were 39,221 ED patient visits and 815 trauma patients who met t r a u m a center triage criteria, 20% of whom suffered penetrating trauma. This study was conducted from January 21 through July 11, 1989. Only ED patients between 18 and 59 years old w i t h injuries w h o m e t trauma center triage criteria or had noninjury-related illness were considered eligible for the study. Only patients who had blood drawn for physician-requested laboratory tests and for whom an extra aliquot of 114/783
blood was available for HIV antibody testing were entered into the study; the nursing staff routinely draw extra tubes of blood from all patients who have blood tests ordered. Undergraduate student research assistants were present in the ED during selected four-hour time periods convenient for them, between noon and midnight. After a patient's ED care was completed, the research assistants requested that the treating physician complete a questionnaire regarding the patient's illness and risk factors for HIV infection. Questions included the patient's age; sex; whether the patient had AIDS, AIDSrelated complex, or known HIV infection; and whether the patient was in any one of four high-risk groups (ie, homosexual or bisexual men, IV drug user, hemophiliac, or sexual partner of any of the former three). Risk factors were determined by the treating physician during history taking f r o m the p a t i e n t s and t h e i r friends or family, review of the patient's chart, and occasionally from discussion with the patient's personal physician. The investigators made no attempt to interfere with or influence the manner in which the risk f a c t o r a s s e s s m e n t was performed. Patients were classified as having known HIV infection if they were so identified by the emergency physician after completing their care. All other patients were considered to have unknown HIV antibody status. Because all patients with known HIV infection were excluded from this study, all patients who proved to have HIV antibody were defined as having unrecognized HIV infection. Blood was collected from the nontrauma group only when the research assistants were present in the ED. Therefore, data regarding risk factors were available for all 100 nontrauma patients. Extra blood specimens from trauma patients were saved by the nursing staff when the research assistants were not present in the ED. Therefore, data regarding risk factors were not available for all trauma patients. Nurses were instructed not to save blood from t r a u m a patients with AIDS, A I D S - r e l a t e d c o m p l e x , or known HIV infection. Sera were collected until there were 100 serum specimens from eligible patients for both groups of ED patients? Blood Annals of Emergency Medicine
TABLE 1. Proportion of ED patients with HIV antibody by age, sex, and type Age ( y r )
Nontrauma %
Trauma %
18 - 19
0/11
0
0/11
0
20 - 29
2/44
4.5
0/43
0
30 - 39
3/24
12.5
1/27
3.7
40 - 49
1/21
4,8
1/15
6.7
50 - 59
O/O
0
0/4
0
Male
4/39
10,3
2/85
2,4
Female
3/61
4,9
0/15
0
Total
7/100
7.0
2/100
2.0
Sex
samples were labeled only with the patient's age, expressed as the decade of life (ie, a 35-year-old was labeled 3 0 - 3 9 ) , t h e p a t i e n t ' s sex, a n d whether the patient was a trauma or nontrauma patient. Therefore, all serologic testing was conducted anonymously, and it was impossible to link the patients' antibody status with their identity. The study was approved by the hospital's institutional review board, and informed consent was not obtained from the patients. All serum samples were first tested for a n t i b o d y to HIV by e n z y m e linked immunosorbent assay (ELISA, Abbott Laboratories, Abbott Park, Illinois). If positive, the specimen was retested by ELISA. All repeatedly reactive specimens were analyzed by Western blot test. Specimen with multiple bands of antibody reactive to p24, gp41, gp120, and gpl60 were considered seropositive for HIV antibody. Seroprevalence rates specific to age group, sex, and study subgroup were c a l c u l a t e d . S e r o p r e v a l e n c e rates within these groups were compared by X2 test for proportions. Seroprevalence specific to are was analyzed by forming ten-year age intervals and c a l c u l a t i n g the age-specific seroprevalence rate for each interval. Ninety-five percent confidence intervals were calculated for each seroprevalence rate. RESULTS One hundred trauma and 100 nontrauma ED patients were recruited during the study. The t r a u m a patients ranged in age from 18 to 55 years (mean age, 31.4 years; SD, 9.7 years). There were 85 men and 15 20:7 July 1991
HIV A N T I B O D Y Baraff, Talan, & Torres
TABLE 2. Risk factor assessment of selected ED patients by emergency physicians % Nontrauma (N - 100)
% Trauma (N - 30)
Yes
No
Known
Yes
No
Known
Homosexual or bisexual
1.0
61.0
62.0
0
20.0
20.0
IV drug use
2.0
59.0
61.0
0
30.0
30.0
Hemophilia Partner at risk
0 1.0
63.0 30.0
63.0 31.0
0 0
36.7 10.0
36.7 10.0
Three risk factors
......
52.0
......
16.7
Four risk factors
......
30.0
......
10.0
TABLE 3. Prevalence of antibody to HIV in selected hospital populations not known to be infected Prevalence
Mean Age (yr) (Range)
Reference
VA Hospital, Washington, DC
2.5%
59.5 (21 - 96)
5
ED medical, trauma, and obstetric admissons
Charity Hospital, New Orleans
2.1%
30.9 (19
54)
6
ED patients
Johns Hopkins, Baltimore Johns Hopkins, Baltimore
4.0%
35.8 (15
95)
7
7.8%
*
8 9
Population
Location
Admitted patients
ED patients, critically ill ED patients
Inner city
4.2% to 8.9%
NSt
ED patients
Suburban
0.4% to 6.4%
NS
9
ED patients
UCLA, Los Angeles
4.5%
31.4 (18 - 55)
...
*HIV positivity by age group: 15 45: 13.2%; ~ 45: 0%. ~NS, not specified,
women in this group. The nontrauma patients ranged in age from 18 to 49 years (mean age, 30.2 years; SD, 9.4 years). There were 42 men and 58 women in this group. The insurance mix of the 100 nontrauma patients was private insurance, 23%; Medicare, 7%; Medicaid, 28%; contracts, 15%; and uninsured, 27%. The proportion of contract patients was significantly less and the proportion of Medicaid patients was significantly more than the overall ED patient population (P < .025). The results of the HIV antibody testing are presented (Table 1). Two of the 100 trauma patients had antibody to HIV (confidence interval [CI], 0% to 5%), and seven of the nontrauma subgroup had evidence of infection (CI, 2% to 12%). The seropositive rate by age and clinical group varied from 0% to 12.5%, with the highest rates in the 30 to 39-yearold group of nontrauma patients. Of the seven antibody-positive patients in the nontrauma group, four were men and three were women; all were between 20 and 49 years old. Both of 20:7: July 1991
the antibody-positive individuals in the trauma group were men between 30 and 49 years old. The difference in proportions of seropositivity between the sexes was not statistically significant. The results of the physicians' ass e s s m e n t of risk f a c t o r s in the trauma and nontrauma subgroups is presented (Table 2); only the trauma patients who were enrolled when research assistants were present to collect risk factor data are included. For the nontrauma group, physicians obtained information regarding homosexual or bisexual behavior, IV drug use, and hemophilia from approximately 60% of patients and regarding all three of these risk factors from 52% of patients. Information regarding whether a sexual partner was in one of these three risk groups was obtained from only 31% of patients, and information regarding all four risk factors was obtained from 30% of patients. The physicians' acquisition of risk factor information was less complete for the 30 trauma patients, for whom Annals of Emergency Medicine
these data were available. Information regarding homosexual or bisexual behavior, IV drug use, and hemophilia was obtained from only 20% to 37% of patients, and information regarding all three risk factors was obtained from only 17% of patients. Information regarding whether a sexual partner was in one of these three risk groups was obtained from only 10% of these patients. None of the 30 trauma patients gave a history of any high-risk behavior. Only two of the 100 nontrauma patients gave a history of any high-risk behavior. DISCUSSION At the beginning of the 1990s, the AIDS epidemic continues to spread t h r o u g h o u t the world and in the United States. The proportion of the US population that is infected continues to increase and is currently estimated to be 0.4% .3 The majority of infected individuals are asymptomatic because the mean time to onset of s y m p t o m a t i c illness is approximately 11 years, t° The prevalence of antibody to HIV in high-risk groups in the United States is homosexuals, 20% to 50%; IV drug users, 5% to 60%; and h e m o p h i l i a c s , 35% to 70% .4 Several previous studies have dealt with the prevalence of antibody to HIV among ED patients; these are summarized (Table 3). In general, the seroprevalence among ED patients ranges from 0.4% in suburban hospitals to 8.9% in inner-city hospitals. 5-9 It is highest among patients with penetrating trauma, in male patients, and in individuals between 20 and 45 years old.ha The lower incidence in trauma patients in our population probably reflects the nature of trauma seen at our medical center. The majority of our patients have blunt trauma sustained as a result of vehicular injuries. Furthermore, because the rate of s e r o p o s i t i v i t y among IV drug users is lower in California than on the east coast, we would expect a lower proportion of victims of penetrating trauma to be seropositive. This seroprevalence is higher than that reported from other selected populations in Los Angeles: 0.02% of selected hospital admissions (excluding patients with AIDS, AIDS-related complex, or diagnoses likely to be associated with HIV infection) at three sentinel hospitals; 11 3.8% of men 784/115
HIV ANTIBODY Baraff, Talan, & Torres
and 1.0% of women attending sexually t r a n s m i t t e d disease clinics; 12 0.1% of adolescents incarcerated in juvenile hall; 13 0.04% of parturients in county hospitals in Los Angeles; 14 and 0.13% of women in a selective voluntary HIV testing program at a public prenatal and family planning clinic, is The subgroup of n o n t r a u m a patients studied here was not a random sample and was not representative of the total population of n o n t r a u m a patients in our ED. By design, it was limited to the 18- to 59-year-old age group (and in fact included only patients between 18 and 49 years old) and only i n c l u d e d p a t i e n t s from whom blood was obtained for laboratory tests. HIV seroprevalence is known to be a function of age group, with the highest rates in those 25 to 54 years old.7,n, 16 We excluded patients older than this from our study for this reason. The other studies referenced (Table 3) did not do this and thus are not completely comparable. The fact that the HIV seroprevalence in our noninner-city ED is similar to that reported from inner-city EDs may be explained in part by the insurance status of ED patients, which may be viewed as a proxy for socioeconomic group: Regardless of geographic location, EDs are used as the means of primary health care by the uninsured and "underinsured." This is more likely to be true at teaching hospitals. Sixty-five percent of the nontrauma patients we sampled fell into these two payor categories (selfpay and Medicaid). In this respect, a large proportion of our patient population is similar to that seen in innercity EDs. We do not know the payor status of the HIV-positive patients; however, Kelen et al found that HIV infection was more likely in uninsured patients (49.1% vs 36.0%).17 The overall seropositivity rate for men was 4.8% (six of 124) and for women was 3.9% (three of 76). The male-to-female ratio of unrecognized HIV infection in this study was 1.2:1, which contrasts with national data for hospitalized patients of 7:1 and for AIDS patients of 8:1.11,16 Few of our patients are IV drug users, and all known IV drug users were excluded from the sample populations. This suggests that HIV infection is being transmitted in the heterosexual population in Southern California. These findings emphasize the need 116/785
to be diligent in the use of universal precautions in the emergency care setting. We have previously shown that gloves are used by ED personnel during IV access and p h l e b o t o m y only 50% of the time in the care of nontrauma patients and 75% of the time during the care of trauma patientsJ The use of the other universal precautions (ie, gown, mask, and protective eyewear) during the resuscitation of trauma patients is less prevalent. We hope the knowledge of the frequency of HIV infection in ED patients, including in noninner-city hospitals, will improve the use of universal precautions. An important finding in this study is that the assessment of risk factors was not complete in the majority of ED patients and was rarely done in v i c t i m s of t r a u m a t i c injuries. We evaluated risk factor assessment to determine how often this was done by emergency physicians. Of the 30 trauma patients, only 10% had complete data regarding risk factors; of these, none gave any history of being in a high-risk group. Complete risk factor assessment was only accomplished for 30 of the 100 nontrauma patients. Only two patients were identified as having risk factors. Because of the efforts taken to protect p a t i e n t i n d i v i d u a l i t y , we do n o t know whether either was seroposirive. However, at least five of the seven seropositive n o n t r a u m a patients were not k n o w n to be in a high-risk group. This is similar to the findings of Kelen et al (ie, risk factor assessment and patient presentation cannot reliably identify even the majority of HIV-positive patients). 7 In a follow-up study that included rigorous risk assessment, risk assessment predicted 76% of those infected. 16 Thus, even rigorous risk assessment will fail to identify a large minority with HIV infection. The fact that physicians did not do complete risk factor assessment in 90% of t r a u m a victims, a k n o w n high-risk group, and 70% of nontrauma patients demonstrates that emergency physician history often does not include HIV infection risk assessment for either major trauma or medical patients, probably because physicians do not consider HIV infection to be related to the patient's medical condition. We do not think it appropriate to perform HIV risk assessment on every ED patient. InAnnals of Emergency Medicine
stead, we r e c o m m e n d that emergency physicians consider all patients at risk for HIV infection and encourage strict adherence to universal precautions. These data may help in promoting compliance with universal precautions at noninner-city facilities that may have been lulled into complacency by previous reports showing very rare occurrence of HIV among ED patients in such facilities. CONCLUSION We have demonstrated that HIV seroprevalence in a Southern California noninner-city Level I trauma center ED is 2% in critical trauma patients and 7% in a selected population of nontrauma patients. Although the sampling technique we used has limitations, the prevalence of HIV infection in our noninner-city ED is similar to that recently reported from inner-city EDs. This is in contrast to previous reports of low rates of HIV infection among ED patients in a nonurban setting. Assessment of risk factors was incomplete in a majority of our patients. When asked, patients rarely acknowledged any high-risk behavior. It is essential that emergency health care personnel take maximum diligence to prevent exposure to blood and other body fluids from all ED patients. The authors acknowledge the assistance of Robert Chang in collection of specimens and questionnaires, Alice Garakian of Dr Yvonne Bryson's laboratory for assistance with the ELISAassays, and Larry Magpantay of Dr Steven Miles's laboratory for his performance of the Western blot assays.
REFERENCES 1. Baraff LJ, Talan DA: Compliance with universal precautions in a university hospital emergency department. Ann Emerg Med 1989~18:654-657. 2. Centers for Disease Control: Guidelines for prevention of transmission of hurnan immunodeficiency virus and hepatitis B virus to health care and public safety workers. M M W R 1989;38:1-37. 3. Centers for Disease Control: HIV prevalence, projected AIDS case estimates: Workshop, October 31-November 1, 1989. M M W R 1990;39:1110-1119. 4. Quinn TC: The epidemiology of the human immunodeficiency virus. Ann Emerg Med 1990~19:225-232. 5. Gordin FM, Gilbert C, Hawley HP, et ah Prevalence of human immnnodeficiency virus and hepatitis B virus in unselected hospital admissions: Implications for mandatory testing and universal precautions. J Infect Dis 1990;161:14-17. 6. , Risi GF, Gaumer RH, Weeks S, et al: Human immunodeficiency virus: Risk of exposure among health care workers at a southern urban hospital. South Med J 1989;82:1079-1082. 7. Kelen GD, Fritz P, Quqish B, et al: Unrecognized human immunodeficiency virus infection in emergency
20:7 July 1991
HIV ANTIBODY Baraff, Talan, & Torres
department patients. N Engl 7 Med 1988~318:1645-1650.
cisco. [AMA 1990;236:1497-1501.
(abstract). I m Conf AIDS 1989;5:79.
8. Kelen GD, Fritz F, Quqish B, et ah Substantial increase in h u m a n i m m u n o d e f i c i e n e y virus infection (HIV-I} infection in critically ill emergency patients: 1986 and 1987 compared. A n n Emerg Med 1989,18: 378-382.
11. St Louis ME, Olivo N, Critchley S, et ah Methods of surveillance for HIV infection at US sentinel hospitals. Public Health Rep 1990;105:140-146.
15. Fehrs L, Hill D, Kerndt P, et ah HIV screening prograin at a Los Angeles prenatal/family planning center (abstract). Int Conf AIDS 1989;5:65.
12. Ford WL, Rose T, Kerndt P, et ah HIV seroprevalence in sexually transmitted disease ciinics in Los Angeles County (abstract). Int Conf AIDS 1989;5:90.
16. Centers for Disease Control: Pilot study of a household survey to determine HIV seroprevalence.MMWR 1991~40:1-5.
9. Marcus R, Bell DM, Culver DH: Frequency of emergency care providers' contact with blood of patients in n letted with human immunodeficiency virus (abstract). Ann Emerg Med 1990;19:454. 10. Letup G1a, Payne SF, Rutherford GW, et ah Projections of AIDS morbidity and mortality in San Fran-
20:7 July 1991
13. Baker CJ, Re O, Morris R, et al: HIV seroprevalence and demographic survey of incarcerated adolescents (abstract). Int Conf AIDS 1989;5:775. 14. Hill D, Kerndt P, Frenkel LM, et ah HIV seroprevalence among parturients in Los Angeles County, 1988
Annals of Emergency Medicine
17. Kelen GD, DiGiovanna T, Bisson L, et ah Human immunodeficiency virus infection in emergency department patients: Epidemiology, clinical presentations, and risk to health care workers: The Johns Hopkins experience. JAMA 1989;262:516-522.
786/117
Prevalence of HIV Antibody in a Noninner-City University Hospital Emergency Department Study hypothesis: To determine the prevalence of antibody to human i m m u n o d e f i c i e n c y virus (HIV) in trauma and nontrauma patients not identified as having known HIV infection in a noninner-city university teaching hospital emergency department, and to determine the frequency with which treating emergency physicians are knowledgeable of patients' risk factors for HIV infection. Population: ED patients between 18 and 59 years old with injuries that m e t trauma center triage criteria or with nontrauma-related illness who had blood drawn for physician-requested laboratory tests and for whom an extra aliquot of blood was available for HIV antibody testing. Methods: All serum samples were first tested for antibody to HIV by enzyme-linked immunosorbent assay. If positive, the specimen was retested. All repeatedly reactive specimens were analyzed by Western blot test. The treating physician completed a questionnaire regarding the patient's illness and risk factors for HIV infection after the patient's care was completed. Results: Two of 100 major trauma patients (confidence interval, 0% to 5%) and seven of 100 nontrauma patients (confidence interval, 2% to 12%) had antibody to HIV. The seropositive rate by age and clinical group varied from 0% to 12.5%, with the highest rates in the 30- to 39-year-old group of nontrauma patient s. The difference in proportions of seropositivity between the sexes was not statistically significant. Physicians obtained information regarding homosexual or bisexual behavior, IV drug use, and hemophilia from 52% of the nontrauma patients and only 17% of trauma patients. None of 30 trauma patients for whom data were available and only two of the 100 nontrauma patients gave a history of any high-risk behavior. Conclusion: Although the sampling technique we used has limitations, the prevalence of HIV infection in our noninner-city ED is similar to that recently reported from inner-city EDs. This is in contrast to previous reports of low rates of HIV infection among ED patients in nonurban settings. Physician assessment of risk factors was incomplete in the majority of our patients. Patients rarely acknowledged any high-risk behavior. It is essential that emergency health care workers take m a x i m u m diligence to prevent exposure to blood and other body fluids from all ED patients. [Baraft LJ, Talan DA, Torres M: Prevalence of HIV antibody in a noninner-city u n i v e r s i t y h o s p i t a l e m e r g e n c y d e p a r t m e n t . A n n Emerg M e d July 1991;20: 782- 786.]
Larry J Baraff, MD, FACEP* David A Talan, MD*t Mina Torres* Los Angeles, California From the Departments of Pediatrics* and Medicine,t UCLA School of Medicine, UCLA Medical Center; and Olive ViewUCLA Medical Center, Los Angeles, California. Received for publication July 23, 1990. Revisions received November 13, 1990, and February 5, 1991. Accepted for publication February 24, 1991. Address for reprints: Larry J Baraff, MD, FACER UCLA Emergency Medicine Center, 924 Westwood Boulevard, Suite 300, Los Angeles, California 90024.
INTRODUCTION Exposures to blood in the emergency department are numerous, especially when providing care to critically injured trauma patients. The majority of.ED patients are cared for without knowledge of their complete medical history, including their human immunodeficiency virus (HIV) antibody status. We have previously shown that ED health care workers often provide care without strict adherence to the universal precautions recommended by the Centers for Disease Control3, ~ It is estimated that between 800,000 and 1.2 million US citizens axe infected with HIV.3 The majority of these individuals are asymptomatic and do not know that they are infected. The rate of infection is greatest in
20:7 July 1991
Annals of Emergency Medicine
782/113
HIV A N T I B O D Y Baraff, Talan, & Torres
high-risk groups, which include homosexuals or bisexuals, IV drug users, hemophiliacs, and sexual partners of the aforementioned. 4 Previous articles have reported the seroprevalence of antibody to HIV in selected ED patients not known to be i n f e c t e d to range f r o m 2.1% to 8.9%. 5-9 The rate of seropositivity is greatest in men, adults between 20 and 45 years old, and victims of penetrating trauma, especially in eastcoast inner-city hospitals, where the infection rate has been reported to be as high as 19% in trauma victims between 25 and 35 years old.7, 8 We undertook this study to determine the prevalence of antibody to HIV in t r a u m a patients who m e t trauma center triage criteria and nontrauma ED patients not identified by their treating emergency physician as having k n o w n HIV infection. We hoped that by providing this information to ED health care workers and members of the trauma team at our institution we would improve compliance with universal precautions. We also sought to determine the c o m p l e t e n e s s of r i s k - a s s e s s m e n t practices of emergency physicians to determine if such assessment could be used to reliably detect the majority of HIV-infected ED patients. MATERIALS A N D M E T H O D S The UCLA Emergency Medicine Center is a Level I trauma center and comprehensive ED as defined by California state regulations. It serves a generally affluent suburban area of Los Angeles County. The insurance mix of our patients is as follows: private insurance, 12%; Medicare, 11%; Medicaid, 16%; contracts, 25%; and uninsured, 36%. The insurance mix of trauma patients is: private insurance, 23%; Medicare, 10%; Medicaid, 20%; contracts, 26%; and uninsured, 21%. In 1989, there were 39,221 ED patient visits and 815 trauma patients who met t r a u m a center triage criteria, 20% of whom suffered penetrating trauma. This study was conducted from January 21 through July 11, 1989. Only ED patients between 18 and 59 years old w i t h injuries w h o m e t trauma center triage criteria or had noninjury-related illness were considered eligible for the study. Only patients who had blood drawn for physician-requested laboratory tests and for whom an extra aliquot of 114/783
blood was available for HIV antibody testing were entered into the study; the nursing staff routinely draw extra tubes of blood from all patients who have blood tests ordered. Undergraduate student research assistants were present in the ED during selected four-hour time periods convenient for them, between noon and midnight. After a patient's ED care was completed, the research assistants requested that the treating physician complete a questionnaire regarding the patient's illness and risk factors for HIV infection. Questions included the patient's age; sex; whether the patient had AIDS, AIDSrelated complex, or known HIV infection; and whether the patient was in any one of four high-risk groups (ie, homosexual or bisexual men, IV drug user, hemophiliac, or sexual partner of any of the former three). Risk factors were determined by the treating physician during history taking f r o m the p a t i e n t s and t h e i r friends or family, review of the patient's chart, and occasionally from discussion with the patient's personal physician. The investigators made no attempt to interfere with or influence the manner in which the risk f a c t o r a s s e s s m e n t was performed. Patients were classified as having known HIV infection if they were so identified by the emergency physician after completing their care. All other patients were considered to have unknown HIV antibody status. Because all patients with known HIV infection were excluded from this study, all patients who proved to have HIV antibody were defined as having unrecognized HIV infection. Blood was collected from the nontrauma group only when the research assistants were present in the ED. Therefore, data regarding risk factors were available for all 100 nontrauma patients. Extra blood specimens from trauma patients were saved by the nursing staff when the research assistants were not present in the ED. Therefore, data regarding risk factors were not available for all trauma patients. Nurses were instructed not to save blood from t r a u m a patients with AIDS, A I D S - r e l a t e d c o m p l e x , or known HIV infection. Sera were collected until there were 100 serum specimens from eligible patients for both groups of ED patients? Blood Annals of Emergency Medicine
TABLE 1. Proportion of ED patients with HIV antibody by age, sex, and type Age ( y r )
Nontrauma %
Trauma %
18 - 19
0/11
0
0/11
0
20 - 29
2/44
4.5
0/43
0
30 - 39
3/24
12.5
1/27
3.7
40 - 49
1/21
4,8
1/15
6.7
50 - 59
O/O
0
0/4
0
Male
4/39
10,3
2/85
2,4
Female
3/61
4,9
0/15
0
Total
7/100
7.0
2/100
2.0
Sex
samples were labeled only with the patient's age, expressed as the decade of life (ie, a 35-year-old was labeled 3 0 - 3 9 ) , t h e p a t i e n t ' s sex, a n d whether the patient was a trauma or nontrauma patient. Therefore, all serologic testing was conducted anonymously, and it was impossible to link the patients' antibody status with their identity. The study was approved by the hospital's institutional review board, and informed consent was not obtained from the patients. All serum samples were first tested for a n t i b o d y to HIV by e n z y m e linked immunosorbent assay (ELISA, Abbott Laboratories, Abbott Park, Illinois). If positive, the specimen was retested by ELISA. All repeatedly reactive specimens were analyzed by Western blot test. Specimen with multiple bands of antibody reactive to p24, gp41, gp120, and gpl60 were considered seropositive for HIV antibody. Seroprevalence rates specific to age group, sex, and study subgroup were c a l c u l a t e d . S e r o p r e v a l e n c e rates within these groups were compared by X2 test for proportions. Seroprevalence specific to are was analyzed by forming ten-year age intervals and c a l c u l a t i n g the age-specific seroprevalence rate for each interval. Ninety-five percent confidence intervals were calculated for each seroprevalence rate. RESULTS One hundred trauma and 100 nontrauma ED patients were recruited during the study. The t r a u m a patients ranged in age from 18 to 55 years (mean age, 31.4 years; SD, 9.7 years). There were 85 men and 15 20:7 July 1991
HIV A N T I B O D Y Baraff, Talan, & Torres
TABLE 2. Risk factor assessment of selected ED patients by emergency physicians % Nontrauma (N - 100)
% Trauma (N - 30)
Yes
No
Known
Yes
No
Known
Homosexual or bisexual
1.0
61.0
62.0
0
20.0
20.0
IV drug use
2.0
59.0
61.0
0
30.0
30.0
Hemophilia Partner at risk
0 1.0
63.0 30.0
63.0 31.0
0 0
36.7 10.0
36.7 10.0
Three risk factors
......
52.0
......
16.7
Four risk factors
......
30.0
......
10.0
TABLE 3. Prevalence of antibody to HIV in selected hospital populations not known to be infected Prevalence
Mean Age (yr) (Range)
Reference
VA Hospital, Washington, DC
2.5%
59.5 (21 - 96)
5
ED medical, trauma, and obstetric admissons
Charity Hospital, New Orleans
2.1%
30.9 (19
54)
6
ED patients
Johns Hopkins, Baltimore Johns Hopkins, Baltimore
4.0%
35.8 (15
95)
7
7.8%
*
8 9
Population
Location
Admitted patients
ED patients, critically ill ED patients
Inner city
4.2% to 8.9%
NSt
ED patients
Suburban
0.4% to 6.4%
NS
9
ED patients
UCLA, Los Angeles
4.5%
31.4 (18 - 55)
...
*HIV positivity by age group: 15 45: 13.2%; ~ 45: 0%. ~NS, not specified,
women in this group. The nontrauma patients ranged in age from 18 to 49 years (mean age, 30.2 years; SD, 9.4 years). There were 42 men and 58 women in this group. The insurance mix of the 100 nontrauma patients was private insurance, 23%; Medicare, 7%; Medicaid, 28%; contracts, 15%; and uninsured, 27%. The proportion of contract patients was significantly less and the proportion of Medicaid patients was significantly more than the overall ED patient population (P < .025). The results of the HIV antibody testing are presented (Table 1). Two of the 100 trauma patients had antibody to HIV (confidence interval [CI], 0% to 5%), and seven of the nontrauma subgroup had evidence of infection (CI, 2% to 12%). The seropositive rate by age and clinical group varied from 0% to 12.5%, with the highest rates in the 30 to 39-yearold group of nontrauma patients. Of the seven antibody-positive patients in the nontrauma group, four were men and three were women; all were between 20 and 49 years old. Both of 20:7: July 1991
the antibody-positive individuals in the trauma group were men between 30 and 49 years old. The difference in proportions of seropositivity between the sexes was not statistically significant. The results of the physicians' ass e s s m e n t of risk f a c t o r s in the trauma and nontrauma subgroups is presented (Table 2); only the trauma patients who were enrolled when research assistants were present to collect risk factor data are included. For the nontrauma group, physicians obtained information regarding homosexual or bisexual behavior, IV drug use, and hemophilia from approximately 60% of patients and regarding all three of these risk factors from 52% of patients. Information regarding whether a sexual partner was in one of these three risk groups was obtained from only 31% of patients, and information regarding all four risk factors was obtained from 30% of patients. The physicians' acquisition of risk factor information was less complete for the 30 trauma patients, for whom Annals of Emergency Medicine
these data were available. Information regarding homosexual or bisexual behavior, IV drug use, and hemophilia was obtained from only 20% to 37% of patients, and information regarding all three risk factors was obtained from only 17% of patients. Information regarding whether a sexual partner was in one of these three risk groups was obtained from only 10% of these patients. None of the 30 trauma patients gave a history of any high-risk behavior. Only two of the 100 nontrauma patients gave a history of any high-risk behavior. DISCUSSION At the beginning of the 1990s, the AIDS epidemic continues to spread t h r o u g h o u t the world and in the United States. The proportion of the US population that is infected continues to increase and is currently estimated to be 0.4% .3 The majority of infected individuals are asymptomatic because the mean time to onset of s y m p t o m a t i c illness is approximately 11 years, t° The prevalence of antibody to HIV in high-risk groups in the United States is homosexuals, 20% to 50%; IV drug users, 5% to 60%; and h e m o p h i l i a c s , 35% to 70% .4 Several previous studies have dealt with the prevalence of antibody to HIV among ED patients; these are summarized (Table 3). In general, the seroprevalence among ED patients ranges from 0.4% in suburban hospitals to 8.9% in inner-city hospitals. 5-9 It is highest among patients with penetrating trauma, in male patients, and in individuals between 20 and 45 years old.ha The lower incidence in trauma patients in our population probably reflects the nature of trauma seen at our medical center. The majority of our patients have blunt trauma sustained as a result of vehicular injuries. Furthermore, because the rate of s e r o p o s i t i v i t y among IV drug users is lower in California than on the east coast, we would expect a lower proportion of victims of penetrating trauma to be seropositive. This seroprevalence is higher than that reported from other selected populations in Los Angeles: 0.02% of selected hospital admissions (excluding patients with AIDS, AIDS-related complex, or diagnoses likely to be associated with HIV infection) at three sentinel hospitals; 11 3.8% of men 784/115
HIV ANTIBODY Baraff, Talan, & Torres
and 1.0% of women attending sexually t r a n s m i t t e d disease clinics; 12 0.1% of adolescents incarcerated in juvenile hall; 13 0.04% of parturients in county hospitals in Los Angeles; 14 and 0.13% of women in a selective voluntary HIV testing program at a public prenatal and family planning clinic, is The subgroup of n o n t r a u m a patients studied here was not a random sample and was not representative of the total population of n o n t r a u m a patients in our ED. By design, it was limited to the 18- to 59-year-old age group (and in fact included only patients between 18 and 49 years old) and only i n c l u d e d p a t i e n t s from whom blood was obtained for laboratory tests. HIV seroprevalence is known to be a function of age group, with the highest rates in those 25 to 54 years old.7,n, 16 We excluded patients older than this from our study for this reason. The other studies referenced (Table 3) did not do this and thus are not completely comparable. The fact that the HIV seroprevalence in our noninner-city ED is similar to that reported from inner-city EDs may be explained in part by the insurance status of ED patients, which may be viewed as a proxy for socioeconomic group: Regardless of geographic location, EDs are used as the means of primary health care by the uninsured and "underinsured." This is more likely to be true at teaching hospitals. Sixty-five percent of the nontrauma patients we sampled fell into these two payor categories (selfpay and Medicaid). In this respect, a large proportion of our patient population is similar to that seen in innercity EDs. We do not know the payor status of the HIV-positive patients; however, Kelen et al found that HIV infection was more likely in uninsured patients (49.1% vs 36.0%).17 The overall seropositivity rate for men was 4.8% (six of 124) and for women was 3.9% (three of 76). The male-to-female ratio of unrecognized HIV infection in this study was 1.2:1, which contrasts with national data for hospitalized patients of 7:1 and for AIDS patients of 8:1.11,16 Few of our patients are IV drug users, and all known IV drug users were excluded from the sample populations. This suggests that HIV infection is being transmitted in the heterosexual population in Southern California. These findings emphasize the need 116/785
to be diligent in the use of universal precautions in the emergency care setting. We have previously shown that gloves are used by ED personnel during IV access and p h l e b o t o m y only 50% of the time in the care of nontrauma patients and 75% of the time during the care of trauma patientsJ The use of the other universal precautions (ie, gown, mask, and protective eyewear) during the resuscitation of trauma patients is less prevalent. We hope the knowledge of the frequency of HIV infection in ED patients, including in noninner-city hospitals, will improve the use of universal precautions. An important finding in this study is that the assessment of risk factors was not complete in the majority of ED patients and was rarely done in v i c t i m s of t r a u m a t i c injuries. We evaluated risk factor assessment to determine how often this was done by emergency physicians. Of the 30 trauma patients, only 10% had complete data regarding risk factors; of these, none gave any history of being in a high-risk group. Complete risk factor assessment was only accomplished for 30 of the 100 nontrauma patients. Only two patients were identified as having risk factors. Because of the efforts taken to protect p a t i e n t i n d i v i d u a l i t y , we do n o t know whether either was seroposirive. However, at least five of the seven seropositive n o n t r a u m a patients were not k n o w n to be in a high-risk group. This is similar to the findings of Kelen et al (ie, risk factor assessment and patient presentation cannot reliably identify even the majority of HIV-positive patients). 7 In a follow-up study that included rigorous risk assessment, risk assessment predicted 76% of those infected. 16 Thus, even rigorous risk assessment will fail to identify a large minority with HIV infection. The fact that physicians did not do complete risk factor assessment in 90% of t r a u m a victims, a k n o w n high-risk group, and 70% of nontrauma patients demonstrates that emergency physician history often does not include HIV infection risk assessment for either major trauma or medical patients, probably because physicians do not consider HIV infection to be related to the patient's medical condition. We do not think it appropriate to perform HIV risk assessment on every ED patient. InAnnals of Emergency Medicine
stead, we r e c o m m e n d that emergency physicians consider all patients at risk for HIV infection and encourage strict adherence to universal precautions. These data may help in promoting compliance with universal precautions at noninner-city facilities that may have been lulled into complacency by previous reports showing very rare occurrence of HIV among ED patients in such facilities. CONCLUSION We have demonstrated that HIV seroprevalence in a Southern California noninner-city Level I trauma center ED is 2% in critical trauma patients and 7% in a selected population of nontrauma patients. Although the sampling technique we used has limitations, the prevalence of HIV infection in our noninner-city ED is similar to that recently reported from inner-city EDs. This is in contrast to previous reports of low rates of HIV infection among ED patients in a nonurban setting. Assessment of risk factors was incomplete in a majority of our patients. When asked, patients rarely acknowledged any high-risk behavior. It is essential that emergency health care personnel take maximum diligence to prevent exposure to blood and other body fluids from all ED patients. The authors acknowledge the assistance of Robert Chang in collection of specimens and questionnaires, Alice Garakian of Dr Yvonne Bryson's laboratory for assistance with the ELISAassays, and Larry Magpantay of Dr Steven Miles's laboratory for his performance of the Western blot assays.
REFERENCES 1. Baraff LJ, Talan DA: Compliance with universal precautions in a university hospital emergency department. Ann Emerg Med 1989~18:654-657. 2. Centers for Disease Control: Guidelines for prevention of transmission of hurnan immunodeficiency virus and hepatitis B virus to health care and public safety workers. M M W R 1989;38:1-37. 3. Centers for Disease Control: HIV prevalence, projected AIDS case estimates: Workshop, October 31-November 1, 1989. M M W R 1990;39:1110-1119. 4. Quinn TC: The epidemiology of the human immunodeficiency virus. Ann Emerg Med 1990~19:225-232. 5. Gordin FM, Gilbert C, Hawley HP, et ah Prevalence of human immnnodeficiency virus and hepatitis B virus in unselected hospital admissions: Implications for mandatory testing and universal precautions. J Infect Dis 1990;161:14-17. 6. , Risi GF, Gaumer RH, Weeks S, et al: Human immunodeficiency virus: Risk of exposure among health care workers at a southern urban hospital. South Med J 1989;82:1079-1082. 7. Kelen GD, Fritz P, Quqish B, et al: Unrecognized human immunodeficiency virus infection in emergency
20:7 July 1991
HIV ANTIBODY Baraff, Talan, & Torres
department patients. N Engl 7 Med 1988~318:1645-1650.
cisco. [AMA 1990;236:1497-1501.
(abstract). I m Conf AIDS 1989;5:79.
8. Kelen GD, Fritz F, Quqish B, et ah Substantial increase in h u m a n i m m u n o d e f i c i e n e y virus infection (HIV-I} infection in critically ill emergency patients: 1986 and 1987 compared. A n n Emerg Med 1989,18: 378-382.
11. St Louis ME, Olivo N, Critchley S, et ah Methods of surveillance for HIV infection at US sentinel hospitals. Public Health Rep 1990;105:140-146.
15. Fehrs L, Hill D, Kerndt P, et ah HIV screening prograin at a Los Angeles prenatal/family planning center (abstract). Int Conf AIDS 1989;5:65.
12. Ford WL, Rose T, Kerndt P, et ah HIV seroprevalence in sexually transmitted disease ciinics in Los Angeles County (abstract). Int Conf AIDS 1989;5:90.
16. Centers for Disease Control: Pilot study of a household survey to determine HIV seroprevalence.MMWR 1991~40:1-5.
9. Marcus R, Bell DM, Culver DH: Frequency of emergency care providers' contact with blood of patients in n letted with human immunodeficiency virus (abstract). Ann Emerg Med 1990;19:454. 10. Letup G1a, Payne SF, Rutherford GW, et ah Projections of AIDS morbidity and mortality in San Fran-
20:7 July 1991
13. Baker CJ, Re O, Morris R, et al: HIV seroprevalence and demographic survey of incarcerated adolescents (abstract). Int Conf AIDS 1989;5:775. 14. Hill D, Kerndt P, Frenkel LM, et ah HIV seroprevalence among parturients in Los Angeles County, 1988
Annals of Emergency Medicine
17. Kelen GD, DiGiovanna T, Bisson L, et ah Human immunodeficiency virus infection in emergency department patients: Epidemiology, clinical presentations, and risk to health care workers: The Johns Hopkins experience. JAMA 1989;262:516-522.
786/117
