Address reprint requests to George R. Seage III, MPH, Boston Depart- ment of ..... PA, Robert McCarthy, RNP, J. Davis Allen, MD, Helen Fitch, RN, and Joan.
Increased Suppressor T Cells in Probable Transmitters of Human Immunodeficiency Virus Infection GEORGE R. SEAGE III, MPH, C. ROBERT HORSBURGH, JR, MD, ANN M. HARDY, DRPH, KENNETH H. MAYER, MD, M. ANITA BARRY, MD, MPH, JEROME E. GROOPMAN, MD, HAROLD W. JAFFE, MD, AND GEORGE A. LAMB, MD Abstract: To evaluate behavioral and immunologic factors related to transmission of human immunodeficiency virus (HIV) by homosexual intercourse, we studied a population of 329 homosexual/ bisexual men (155 partner-pairs) seen in a community health center and medical outpatient clinic. Logistic regression analysis showed that behavioral risk factors for infection in the 130 HIV-infected men included: receptive anal intercourse (OR 4.6, 95% CI-1.8, 12.1); receptive fisting (OR 2.5, CI-1.l, 7.0); nitrite use (OR 2.3, CI-1.2, 4.6); history of gonorrhea or. syphilis (OR 2.3, CI-1.4, 3.9); and history of sexual contact with men from areas with many AIDS cases
(OR 1.9, CI-1.0, 3.5). Comparing seven men who were probable transmitters of HIV and 11 men who had not transmitted HIV to their uninfected partners despite unprotected insertive anal intercourse, we found no differences in HIV isolation from peripheral blood mononuclear cells, circulating HIV antigen detection, or presence of neutralizing antibody to HIV. Helper T-cell numbers were not significantly different between the two groups, but transmitters had more suppressor T-cells than did nontransmitters. (Am J Public Health 1989; 79:1638-1642.)
Introduction The major behavioral risk factors in the transmission of human immunodeficiency virus (HIV) in homosexual/ bisexual men include receptive anogenital sexual contact and a large number of sex partners. 1-3 However, some men with many receptive anal sexual contacts remain uninfected while others with infrequent contact become infected.45 Thus, while particular sexual behaviors may increase the likelihood of HIV transmission, other factors presumably contribute to the efficiency of sexual transmission of the virus. Transmission of HIV requires that the infected partner transmit the virus and that the exposed partner be susceptible to the infection. Studies of susceptibility to HIV infection of exposed persons have failed to reveal either immunologic or genetic susceptibility factors.6'7 However, several studies have shown a correlation between worsening clinical status or declining numbers of helper T cells in the HIV-infected partner and increased transmission of HIV.Y2 In this study, we evaluated the role of specific sex practices in HIV transmission among homosexual men. We also examined the relation of immunologic laboratory parameters to transmission by comparing seropositive men who appeared to have infected their partners with seropositive men who did not infect their partners despite unprotected insertive anal intercourse.
Community Health Center (FCHC) and/or the New England Deaconess Hospital. AIDS was diagnosed according to the Centers for Disease Control (CDC) classification system (Group IV subgroups C-1 and D-2); AIDS-related complex (ARC) was defined as CDC classification group III (persistent generalized lymphadenopathy), IV-A (fever, weight loss or diarrhea), or C-2.'3 Asymptomatic men were those without AIDS or ARC. Sexual contact was defined as orogenital, anogenital, or oroanal contact involving an exchange of body fluids and/or penetration. We recruited asymptomatic participants from: men who had at least one visit to the FCHC from May 1984 through March 1987; men referred by friends or physicians outside of the FCHC; advertisements soliciting participation in the study placed in two newspapers; and men referred by persons already enrolled in the study. The protocol for this study was approved by the Human Studies Committees of all participating institutions. Data Collection At enrollment, all participants were interviewed to obtain data on demographic characteristics, occupational and travel histories, sexual behaviors, and drug and alcohol use. A complete medical history was obtained for the five years before study enrollment, and a standard clinical examination was performed. All participants were counseled regarding high- and low-risk sex practices and ways to reduce their risk for HIV infection. Laboratory Studies All participants were evaluated with a complete blood count (CBC) and lymphocyte subset determination, performed using specific monoclonal antibodies (Ortho Diagnostics, Raritan, NJ). Antibodies to HIV were determined by an enzyme immunoassay (EIA) (Abbott Laboratories, Chicago, IL). All serum samples initially positive by EIA and selected other serum samples (as specified later) were examined by Western blot using previously described methods.'4 Culture of HIV was performed in H9 cells, and reverse transcriptase activity was measured by standard techniques.15 Antigen-capture assay for HIV employed a sandwich EIA using polystyrene beads coated with human antiHIV (Abbott Laboratories, Chicago, IL). 16 Neutralizing antibody assays were performed in H9 cells with the HTLVIIIB isolate of HIV using a modification of the method of
Methods Selection of Study Participants
Men were eligible for enrollment in our study if they were homosexual or bisexual, were 18 years of age or older, and resided within the Boston, Massachusetts Standard Metropolitan Statistical Area. Men were seen at the Fenway From the Community Infectious Disease Epidemiology Program, Division of Community Medicine, Boston Department of Health and Hospitals; AIDS Program, Center for Infectious Diseases, CDC, Atlanta; Epidemiology and Statistics Section, Boston University School of Public Health; Fenway Community Health Center, Boston; and Division of Hematology and Oncology, New England Deaconness Hospital, Boston. Address reprint requests to George R. Seage III, MPH, Boston Department of Health and Hospitals, 818 Harrison Avenue, Boston, MA 02118. This paper, submitted to the Journal November 28, 1988, was revised and accepted for publication May 19, 1989. X 1989 American Joumal of Public Health 0090-0036/89$1.50
AJPH December 1989, Vol. 79, No. 12
SUPPRESSOR CELLS IN HIV TRANSMITTERS
Robert-Guroff.17 The assay was performed by mixing 100 50 percent tissue culture infectious doses of HIV with 200 ml of heat-inactivated human serum at various dilutions, and incubating at 4°C for 1.5 hours. Virus was then absorbed onto 1 x 106 H9 cells for one hour and incubated at 37°C. Neutralization titer was defined as positive when the assay blocked 90 percent or more of the control virus infection at a 1:20 dilution.
had ARC, and 275 were asymptomatic. Six participants were linked to three partners, 12 to two partners, 113 to one partner, and 43 to no partner, for a total of 155 partner-pairs. The mean age of the study participants was 32.8 years; mean level of education 16 years; median income $10,000$19,999; and mean length of time residing in metropolitan Boston 144 months. Ten percent of the men had been married; 316 (96%) were White, 9 (3%) were Black, and four
(1%) were Hispanic. One hundred thirty (40%) men were HIV-seropositive, while 199 (60%) were seronegative. Of the 155 partner-pairs, 62 (40%) were concordant seronegatives, 35 (23%) were concordant seropositives, and 58 (37%) were discordant (one seropositive and one seronegative partner). Thirteen of 24 partners (54%) of men with AIDS were seropositive (95% CI 34, 74); 12 of 31 partners (39%) of men with ARC were seropositive (CI 22, 57); 10 of 24 partners (42%) of asymptomatic seropositive men were seropositive (CI 23, 61).
Confidence intervals for infection rates were calculated based on the chi-square function method of Miettinen. " Exposure variables were categorized depending on the distribution of the variable without regard to serologic status. Unadjusted prevalence odds (exposure odds) ratios were estimated and exact 95% confidence intervals (CI) calculated by the method of Rothman.'9 Adjusted estimates were maximum likelihood estimates from logistic regression and 95% CI were calculated using the standard errors of the beta estimates.2O Continuous laboratory variables were compared using the Wilcoxon 2-sample test. Dichotomous laboratory variables were compared by Fisher's exact test.20
To determine which potential risk factors were associated with seropositivity, we analyzed all 329 men in the cohort. In bivariate analysis, six variables were associated with antibody to HIV: number of sex partners in the previous five years; receptive anal sexual activity; receptive oral sexual activity; receptive fisting; sexual contact with persons from areas of high HIV prevalence; history of gonorrhea or syphilis; use of nitrites; and use of other illicit drugs (Table 1).
Results All interview information and laboratory samples were obtained at the initial enrollment visit. Three hundred twentynine men were enrolled in the study: 23 men had AIDS, 31
TABLE 1-Distribution (%), Crude and Adjusted Relative Odds for HIV Infection by Behavioral Risk Factors
HIV-Antibody Status History over Past Five Years
No. of male partners -30 >30 No. of men with whom the subject had a specific sex practice Receptive penile oral 1-10 >10
Percent Seropositive (N = 130)
Percent Seronegative (N = 199)
6.1 34.1 59.9
18.8 57.4 23.8
1.00 1.84 7.82
Receptive penile anal 0 1-10 >10
Receptive fisting never ever
Sex with partners from areas of higher HIV prevalence never ever
History of gonorrhea or syphilis never ever Nitrite use never ever
Illicit drug usec never ever
2.30 2.31 1.37
(.56, 1.23) -
(.62, 3.67) (1.78, 12.06) (1.08, 7.03)
(1.,01, 3.50) (1.36, 3.88) (1.17, 4.60) (.59, 3.22)
a) Adjusted using a logistic regression model that included age and the variables listed in this table. b) New York City, San Francisco, Los Angeles, and/or Miami. c) Marijuana, cocaine, PCP, LSD, and/or quaaludes.
AJPH December 1989, Vol. 79, No. 12
SEAGE, ET AL.
A logistic regression model was developed to assess potential confounding by any of the variables significantly associated with HIV antibody in the bivariate analysis. Factors that correlated with seropositivity in the multiple regression model were: receptive anal sexual activity; receptive fisting; nitrite use; history of gonorrhea or syphilis; and sexual contact with persons from areas of high HIV prevalence (Table 1). Characteristics of Transmitters
We examined partner-pairs in which both partners were seropositive to identify possible transmitters. Ten partnerpairs were identified in which one partner had had over 80 percent of his receptive anal contacts within the relationship (in no case did both partners have over 80 percent of their receptive anal contacts with each other). The partner whose contacts were in large part within the relationship was therefore presumed to have been infected by the other (more sexually active) partner, who was designated the likely transmitter. Peripheral blood cells were available from seven of the transmitters; of these, four had a partner who had no receptive anal intercourse with any other partner, while the other three had partners with 4, 13, and 14 episodes outside the relationship. Partner-pairs in which one partner was seropositive and one was seronegative were examined to identify men who had exposed their partner to HIV without transmitting infection. All seronegative men were confirmed to be uninfected by negative Western blot and HIV antigen assay. Since we could not be certain that the seropositive partner was positive more than a short time before entry into the study, we excluded partner-pairs who had not had at least six episodes in which the seronegative partner had receptive anal intercourse with an insertive seropositive partner during the six months before entry into the study. The HIV seropositive partners in these partner-pairs were designated the likely nontransmitters. Sixteen nontransmitters were identified. Peripheral blood cells were available from 11 of these men. One ofthe transmitters had AIDS and three had ARC; no nontransmitters had AIDS but three had ARC. Transmitters had longer relationships with their partner (mean + SD: 58 ± 24 months) than did nontransmitters (36 ± 36 months), but transmitters had fewer total episodes of insertive anal intercourse with their partner (151 ± 185 episodes) than nontransmitters (192 ± 192 episodes). Neither difference was statistically significant. No differences were noted between transmitters and nontransmitters in history of syphilis, gonorrhea, hepatitis B, herpes simplex, herpes zoster, or cytomegalovirus infection, or infectious mononucleosis. Furthermore, transmitters and nontransmitters had similar histories of fever, skin rash, and enlarged lymph nodes. Transmitters did not differ from nontransmitters with respect to circulating HIV antigen, presence of neutralizing antibody to HIV, or ability to recover virus from peripheral blood mononuclear cells (Table 2). Phenotyping of peripheral
blood mononuclear cells of transmitters revealed a median of 1150 total T cells, 430 helper T cells, 960 suppressor T cells, and a median helper/suppressor ratio of 0.40; nontransmitters had a median of 1170 total T cells, 390 helper T cells, 650 suppressor T cells, and a median helper/suppressor ratio of 0.80 (Figure 1). While total and helper T-cell numbers were similar in both groups, transmitters had higher suppressor T-cell numbers than did nontransmitters (Wilcoxon, p = < .02). Helper/suppressor ratios were lower in transmitters than in nontransmitters (Wilcoxon, p = .12). Five of the six transmitters with elevated suppressor cells were re-evaluated after 12 months; only one had persistently elevated suppressor cells.
Discussion Several studies have shown that the major risk factors for acquisition of infection among homosexual men are exposure to a large number of sex partners and exposure to infected persons by receptive anal intercourse.1-3 Demographic and behavioral analysis of our cohort affirms that it is similar to previously reported cohorts. We assessed risk by estimating exposure odds ratios for HIV infection. The interpretation of such ratios can be difficult when both the temporal sequence and the duration of exposure are unknown. In some cases, the relevant period of risk may have been overestimated, resulting in exposure odds ratios that underestimate the true values. The risk of acquiring HIV infection was positively correlated with increased number of sex partners with whom the person was the receptive anal sex partner. In our analysis, both fisting and use of nitrites were also associated with seropositivity after controlling for confounding. Fisting might facilitate HIV transmission by causing rectal trauma, thereby allowing virus to bypass mucosal defenses. The biologic mechanism for the increased risk in nitrite users is unclear; the role of nitrites as potential effect modifiers needs to be assessed further. While these practices are associated with an increased risk of becoming infected with HIV, they do not assure it. Many persons appear to be sexually exposed to HIV on multiple occasions, yet do not become infected.4'5 Possible explanations for this phenomenon are that some persons are more efficient transmitters than others; that some persons, when exposed, are more susceptible than others; or both. 2400
Parameter HIV culture
Circulating HIV antigen
Neutralizing antibody to HIV
1/5 (20) 1/7 (20) 3/3 (100)
3/9 (33) 4/13 (31) 3/7 (43)
T NT HELPER T CELLS
T - TRANSMITTER NT NON TRANSMITTER
SUPPRESSOR T CELLS -
NT T TOTAL T CELLS
TABLE 2-Distribution (%) of Laboratory Parameters in Transmitters and
SAIDS HIV SEROPOSITIVE, NOT AIDS
FIGURE 1-T-cell Phenotypic Markers in HIV Transmitters and Nontransmitters; Transmitters had Higher Numbers of Suppressor T-cells
AJPH December 1989, Vol. 79, No. 12
SUPPRESSOR CELLS IN HIV TRANSMITTERS
Since previous studies have failed to identify genetic or immunologic susceptibility factors for HIV infection,6'7 we focused our investigation on factors relating to transmission. Partner-pairs in which both partners were seropositive were analyzed to select a group of men who were most likely to have transmitted infection to their partners. Such a determination would ideally require that a seropositive man have only been exposed to a single seropositive partner during the time the virus could be assumed to have been circulating in the community. However, monogamous relationships were rare in our cohort. We therefore computed a scale of relative fidelity for each partner. Seven partner-pairs included a man who had had more than 80 percent of his receptive anal contacts since the end of 1981 with one seropositive person (four had 100 percent of their contacts with the presumed transmitter). Therefore, although the date of HIV transmission could not be definitively established and contact with other seropositive partners could not always be completely excluded, we feel the anal insertive partners in these relationships were likely HIV transmitters. Because of the retrospective nature of the study, the dates of infection of the possible transmitters and nontransmitters were unknown and the exact duration of exposure could not be established. However, the duration of relationships was not significantly different between transmitters and nontransmitters. Moreover, similar numbers of potentially infective episodes were reported for both transmitters and nontransmitters. We found increased suppressor cells in HIV transmitters. This suggests a role for host immunity in modulating infectiousness by controlling the amount of virus present in host secretions and thereby regulating the dose of virus to which a sex partner is exposed. We were not able to measure the amount of virus in the semen of our transmitters and nontransmitters. However, no relationship was seen between the ability to culture virus from the blood, the presence of neutralizing antibody, or the presence of p24 antigen and the ability to transmit HIV. This may have been due to the small number of samples available for study; alternatively, the presence of detectable HIV in blood may not be related to the presence of infectious virus in semen. Increased suppressor T-cell numbers have been reported in primary HIV infection.2' The transmitters in our study therefore may have recently experienced acute HIV infection. Alternatively, since many infectious processes (notably herpes virus infections) can temporarily increase suppressor cells,22'23 increased suppressor cells in our transmitters may be a marker of an intercurrent infectious process other than HIV infection. Such a process might increase HIV replication by activating HIV amplification genes24'25 or by interfering with the as yet unknown mechanisms that permit the host to at least temporarily control HIV replication. Our results suggest that some men with higher numbers of suppressor cells may be more infectious to their sex partners than other infected men. Francis, et al, have also reported increased suppressor cells in mothers who appeared to have transmitted HIV to their neonates.26 On the other hand, several studies have found that HIV-infected men with low helper T-cell numbers or advanced HIV infection are more likely to transmit HIV to their female partners.='2 Our HIV-infected subjects were largely asymptomatic with relatively normal numbers of helper T cells; thus we would not have expected to detect transmission by many men with AIDS or low numbers of T-helper cells. Our observations thus support the hypothesis that HIV infectivity is not limited AJPH December 1989, Vol. 79, No. 12
to the period of immunodeficiency seen with advanced HIV infection.27 ACKNOWLEDGMENTS The authors thank Scott Holmberg, MD, for assistance with this study and review of the manuscript. We thank Martha Moon, RNP, P. Clay Stephens, PA, Robert McCarthy, RNP, J. Davis Allen, MD, Helen Fitch, RN, and Joan Pellitier, RN for their clinical help. Patrick Riggs, Daniel Reed, Janet Bath, and Julie Marston interviewed the study participants, while Lindsey Schneider, Paige M. Bracci, and Tim Bush performed data entry, management, and programming and assisted with the data analysis. W. Mead Morgan consulted with us regarding data analysis. We especially want to thank the staffs of the Fenway Community Health Center and the New England Deaconess Hospital for their invaluable assistance in the performance of this study. Most of all, we are indebted to the participants in the study for their time, effort, and trust in us to perform this difficult and demanding research.
I. Goedert JJ, Samgadharan MG, Biggar RJ, et al: Determinants of retrovirus (HTLV-III) antibody and immunodeficiency conditions in homosexual men. Lancet 1984; 2:711-716. 2. Kingsley LA, Detels R, Kaslow R, et al: Risk factors for seroconversion to human immunodeficiency virus among male homosexuals. Lancet 1987; 1:345-349. 3. Winkelstein W, Lyman DM, Padian N, et al: Sexual practices and risk of infection by human immunodeficiency virus: the San Francisco men's health study. JAMA 1987; 257:321-325. 4. Peterman TA, Stoneburner RL, Allen JR, et al: Risk of human immunodeficiency virus transmission from heterosexual adults with transfusionassociated infections. JAMA 1988;259:55-58. 5. Friedland GH, Klein RS: Transmission of the human immunodeficiency virus. N Engl J Med 1987;317:1125-1135. 6. Eales L-J, Nye KE, Pinching AJ: Group-specific component and AIDS: erroneous data. Lancet 1988; 1:936. 7. Fischl MA, Dickinson GM, Scott GB, Rlimas N, Fletcher MA, Parks W: Evaluation of heterosexual partners, children, and household contacts of adults with AIDS. JAMA 1987; 257:640-644. 8. Goedert JJ, Eyster ME, Biggar RJ, Blattner WA: Heterosexual transmission of human immunodeficiency virus: association with severe depletion of T-helper lymphocytes in men with hemophilia. AIDS Res Human Retrovir 1987; 3:355-361. 9. Laga M, Taelman H, Bonneux L, et al: Risk factors for HIV infection in heterosexual partners of HIV infected africans and europeans [Abstract]. In: IV International Conference on Acquired Immunodeficiency Syndrome. Abstracts Volume 1. Stockholm: Swedish Ministry of Health and Social Affairs, 1988; 260. 10. Osmond D, Bacchetti P, Chaisson RE, et al: Time of exposure and risk of HIV infection in homosexual partners of men with AIDS. Am J Public Health 1988; 78:944-948. 11. Fischl M, Fayne T, Flanagan S, et al: Seroprevalence and risks of HIV infections in spouses of persons infected with HIV [Abstract]. In: IV International Conference on Acquired Immunodeficiency Syndrome. Abstracts Volume 1. Stockholm: Swedish Ministry of Health and Social Affairs, 1988; 274. 12. Hira S, Wadhawan D, Nkowane B, Kamanga J, Perine P: Heterosexual transmission of HIV in Zambia [Abstract]. In: IV International Conference on Acquired Immunodeficiency Syndrome. Abstracts Volume 1. Stockholm: Swedish Ministry of Health and Social Affairs, 1988; 261. 13. Centers for Disease Control: Classification system for human T-lymphotrophic virus type III/lymphadenopathy-associated virus infections. MMWR 1986; 35:334-339. 14. Tsang VCW, Peralta JM, Semons AR: The enzyme-linked immunoelectrotransfer blot techniques (EITB) for studying the specificities of antigens and antibodies separated by gel electrophoresis. Methods Enzymol 1983; 92:377-391. 15. Lasky LA, Groopman JE, Fennie CW, et al: Neutralization of the AIDS retrovirus by antibodies to a recombinant envelope glycoprotein. Science 1986; 233:209-212. 16. Paul DA, Falk LA, Kessler HA, et al: Correlation of serum HIV antigen and antibody with clinical status in HIV infected patients. J Med Virol
17. Groopman JE, Benz PM, Ferriani R, Mayer K, Allan JD, Weymouth LA: Characterization of serum neutralization response to the human immunodeficiency virus (HIV). AIDS Res 1987; 3:71-85. 18. Miettinen 0: Theoretical Epidemiology: Principles of Occurrence Research in Medicine. New York: John Wiley, 1985. 19. Rothman KJ, Boice JD: Epidemiologic Analysis with a Programmable Calculator. NIH Pub. no. 79-1644. Washington, DC: US Department of Health, Education, and Welfare, Public Health Service, 1979.
SEAGE, ET AL. 20. SAS Institute Inc.: SURGI Supplemental Library User's Guide, version 5. Cary, NC: SAS Institute Inc, 1986. 21. Cooper DA, Gold J, MacLean P, et al: Acute AIDS retrovirus infection. Definition of a clinical illness associated with seroconversion. Lancet 1985; 1:537-540. 22. Reinherz EL, O'Brien C, Rosenthal P, Schlossman SC: The cellular basis for viral-induced immunodeficiency: analysis by monoclonal antibodies. J Immunol 1981;125:1269-1274. 23. Camey WP, Rubin RH, Hoffman RA, Hanson WP, Healey K, Hirsch MS: Analysis of T lymphocyte subsets in cytomegalovirus mononucleosis. J Immunol 1981; 126:2114-2116. 24. Ostrove JM, Leonard J, Weck KE, Rabson AB, Gendelman HE: Acti-
vation of the human immunodeficiency virus by herpes simplex virus type 1. J Virol 1987; 61:3726-3732. 25. Davis MG, Kenney SC, Kamine J, Pagano JS, Huang E-S: Immediateearly gene region of human cytomegalovirus trans-activates the promoter of human immunodeficiency virus. Proc Natl Acad Sci USA 1987; 84:8642-8646. 26. Francis H, Lubaki N, Duma MP, et al: Immunologic profiles of mothers in perinatal transmission of HIV infection [Abstract]. In: III International Conference on Acquired Immunodeficiency Syndrome. Abstracts volume. Washington, DC: US Department of Health and Human Services, 1987; 214. 27. Burke DS, Redfield RR: Transmission of human immunodeficiency virus [Letter]. N Engl J Med 1988; 318:1203-1204.
1987 National Medical Expenditure Survey Data Available on Magnetic Tape The 1987 National Medical Expenditure Survey (NMES), a nationwide survey sponsored by the National Center for Health Services Research, is available from the National Technical Information Service (NTIS) on magnetic tape. The Institutional Population Component was designed to yield national estimates of the use of, and expenditures for, care during 1987 by persons who reside in nursing homes or facilities for the mentally retarded. The files contain data from two Phase I instruments: the Facility Questionnaire and the Baseline Questionnaire. * The Facility Questionnaire was collected from approximately 1400 participating facilities, and the data records contain bed size, type of ownership, facility certification, services routinely provided, average cost, and other information. * The Baseline Questionnaire file contains data collected from approximately 6,900 persons who were residing in the sampled facilities, and include sociodemographic information such as age, race, sex, information on residences prior to admission, measures of functional limitations and chronic condition, and the work history of persons in facilities for the mentally retarded. Documentation is provided with the tape and includes a description of NMES, questionnaires, and technical information on sample design, weights, and variance estimation. The NMES is available in EBCDIC or SAS format, and is contained on 9 track, half-inch tape only. Specify density requirement (1600 or 6250 bpi) when ordering. PB89-178461KNC, National Medical Expenditure Survey: Institutional Population Component, Phase I (EBCDIC Files) 1987, $325 plus $3 handling fee. PB89-178487KNC, National Medical Expenditure Survey: Institutional Population Component, Phase I (SAS Files) 1987, $325 plus $3 handling fee. Order from: NTIS, Springfield, VA 22161, (703) 487-4650. For a free Catalog of all Data Files produced by the National Center for Health Services Research, please request PR-791KNC.
AJPH December 1989, Vol. 79, No. 12