passes classes of persons usually consid- ered to be .... technical assistance in reproductive health, ... Schein, 330 Harper P1, Highland Park, NJ. 08904.
etters to the Editor
Letters to the Editor wiU be reviewed and are published as space permits. By submitting a Letter to the Editor, the author gives permission for its publication in the Journal. Letters should not duplicate material being published or submitted elsewhere. Those referimng to a recent Journal article should be received within 3 months of the article's appearance. The Editors reserve the right to edit and abridge and to publish responses. Submit three copies. Both text and references must be typed double-spaced. Text is limited to 400 words andfewer than 10 references. Reprints can be ordered through the author whose address is listed at the end of the letter.
Clarifying the Legal Definition of "Undocumented Aliens" and Regarding Siddharthan Alalasundaram's article, "Undocumented Aliens and Uncompensated Care: Whose Responsibility?"' the authors use premises and definitions relating to "undocumented aliens" that differ from those used in a legal context. Reliance on a definition of "undocumented" that encompasses essentially all aliens other than those with visitor visas and green cards results in an inflated estimate of the undocumented population. The authors' conclusions are somewhat misleading as a result. Individuals who are legally resident, with or without documentation, are not usually considered to be undocumented for purposes of immigration classification.2 Individuals may be present in the United States lawfully with other types of documentation.3 As an example, individu1032 American Journal of Public Health
als admitted to the United States as refugees are in the United States legally and are documented.4 Publicly funded health insurance programs such as Medicaid, although theoretically able to verify persons' legal status through a verification system of the Immigration and Naturalization Service (INS) known as Systematic Alien Verification for Entitlements,5 are often unable to do so owing to problems at the INS with lost files, the INS's failure to enter updated information into the database, and the lack of data relating to nativeborn citizens. Undocumented aliens who are considered under public benefit law to be "permanently residing under color of law" in the United States, although undocumented, may be eligible for Medicaid benefits in addition to those under the emergency provisions of that statute.6 Their eligibility for publicly funded medical care will not be reflected in their INS record because of definitional differences. The authors' conclusion that "close to 6% ... of all uncompensated care provided in Dade County [in FY 1990] involved those residing in Dade County in violation of the immigration laws of the United States" does not follow logically from the data they have presented. Their definition of "undocumented" encompasses classes of persons usually considered to be documented. It is not a violation of our immigration laws, as this article implies that it is, for an individual to enter into the United States in one status and to decide at a later date to apply for political asylum, absent fraud or material misrepresentation.7 Immigration law is extraordinarily complex.8 It is often difficult to determine an individual's true legal status. Any discussion of undocumented individuals' use of health resources is likely to be politically charged.9 Therefore, it is crucial that the methods used in such
research be as precise and accurate as possible. O Sana Loue, JD, PhD Jonathon Foerstel JD At the time this was written, Sana Loue and Jonathon Foerstel were with the Legal Aid Society of San Diego, Inc, San Diego, Calif. Requests for reprints should be sent to Sana Loue, JD, PhD, Department of Epidemiology and Biostatistic, School of Medicine, Case Western Reserve University, 10900 Euclid Ave, Cleveland, OH 44106-4945.
References 1. Siddharthan K, Alalasundaram S. Undocumented aliens and uncompensated care: whose responsibility? Am J Public Health. 1993;83:410-412. 2. Loue S. Access to health care and the undocumented alien. J Legal Med. 1992;13: 271-332. 3. Immigration and Nationality Act, section 101, 8 USC § 1101. 4. Immigration and Nationality Act, section 207,8 USC § 1157. 5. Immigration Reform and Control Act of 1986, section 121. 6. Omnibus Budget Reconciliation Act of 1986, Pub L No 99-509, section 9406, 100 Stat 1874, 2507, amending 42 USC § 1396 (b). 7. Matter of Salim, 18 I & N Dec 311 (BIA 1982). 8. Steel RD. Steel on Immigration Law 2d. 2nd ed. New York, N.Y.: Clark Boardman Callaghan; 1992. 9. Dalton R. State program turns into "MediWorld." San Diego Union-Tribune. April 12, 1993: Al.
The Protective Effect of
Nonoxynol-9 against HIV Infection Women need products within their control to protect themselves during sexual encounters with men who may be unwilling to wear condoms, which are known to reduce the incidence of human immunoJune 1994, Vol. 84, No.6
Letters to the Editor
TABLE 1-Incidence Rates of HIV Infection per 100 Person-Years of Observation, by Level of Condom and Nonoxynol-9 Spermicide Use Level of Nonoxynol-9 Spermicide Use in Coital Acts Not Protected by Condomsb
Level of Condom Usea 0% to 50% 51% to 75% More than 75%
0% to 50% (n)
51% to 75% (n)
More than 75% (n)
36.7 (15) 17.8 (36) 0.0 (22)
18.9 (26) 11.9 (44) 0.0 (15)
0.0 (23)
infection.2 In a second study of lower-dose (150 mg) nonoxynol-9 suppository use and HIV incidence among female sex workers, there was evidence of a protective effect.3 That analysis, however, did not distinguish between acts in which both condoms and nonoxynol-9 were used and acts in which only one method or neither method was used. We report a reanalysis of the data here. By considering as condom use all coital acts in which condoms were used, whether alone or jointly with nonoxynol-9, we focus on condom-free encounters and compare HIV incidence rates according to the extent of nonoxynol-9 use. The rate of HIV infection decreased as condom use increased (Table 1). Controlling for the consistency of condom use, HIV incidence decreased with more consistent use of nonoxynol-9, with the single exception of the lower right table cell representing women who used condoms for more than 75% of coital acts and used nonoxynol-9 for more than 75% of the remaining coital acts. These results were essentially unchanged when we restricted attention to the 60-day observation period prior to the last follow-up visit. The HIV incidence rate increased sharply as the percentage of unprotected coital acts rose. The HIV rate ratio was 7.0 (95% confidence interval = 2.1, 23.0) for women with 30% or more unprotected acts compared with women reporting less than 10% unprotected acts. June 1994, Vol. 84. No. 6
Adjusted 95% HIV ConfiRate dence Ratio Interval
3.6 (41) 9.4 (48)
Note. The numbers in parentheses refer to the number of women in each cell. aThe categories of condom use are based on the percentage of all coital acts in which condoms were used, either alone or jointly with nonoxynol-9 suppositories. bThe categories of nonoxynol-9 use are based on the percentage of coital acts not protected by condoms in which nonoxynol-9 was used.
deficiency virus (HIV).1 Whether vaginal spermicides containing nonoxynol-9 reduce the risk of acquiring HIV infection is a matter of acute interest. Nonoxynol-9 inactivates HIV at low concentrations in vitro, yet an epidemiological study found that use of a vaginal contraceptive sponge with 1000 mg of nonoxynol-9 did not protect women sex workers against HIV
TABLE 2-HIV Rate Ratios from Proportional Hazards Regression, by Level of Condom and Nonoxynol-9 Spermicide Use
We used proportional hazards regression4 to calculate HIV rate ratios accounting simultaneously for condom use, nonoxynol-9 use, and the number of sexual partners (Table 2). Given the small number of HIV infection events, the confidence intervals for the barrier rate ratios were wide. Consistent nonoxynol-9 use was associated with a marked reduction in the rate of HIV infection, as was consistent condom use, and the nonoxynol-9 rate ratio was comparable to the condom rate ratio. When condom and nonoxynol-9 use were analyzed as continuous variables, there again was a protective effect against HIV with increasing consistency of use of each method (data not shown). Women who frequently use a nonoxynol-9 product may develop genital irritation,5 but there was little irritation reported by these study participants, who had a median of four sexual partners per week. Importantly, there was no evidence of a higher infection rate with more frequent nonoxynol-9 insertion (data not
shown). The current results confirm the earlier analysis3; as is the case for our accompanying analysis of nonoxynol-9 use and endocervical gonorrhea,6 this reanalysis has the added strength of specifying coital acts in which both barrier methods, nonoxynol-9 only, or neither barrier method was used. When condoms are not used, consistent use of a nonoxynol-9 spermicide can prevent HIV infection. Additional studies, preferably randomized studies, are needed to confirm these results. O PaulJ. Feldblun, PhD Sharon S. Weir, A, MPH The authors are with Family Health International, Research Triangle Park Branch, Durham, NC.
Level of nonoxynol-9 usea Morethan75% 50°hto75% Level of condom useb Morethan75%
50%oto75% 15+ partners/
0.38 0.42
0.1,1.3 0.1,1.3
0.27 0.39 4.05
0.1,1.2 Q.1, 1.2 1.4,11.7
monthc aThe categories of nonoxynol-9 use are based on the percentage of acts not protected by condoms in which nonoxynol-9 suppositories were used. The referent group is women who used nonoxynol-9 for less than 50% of coital acts not protected by condoms. bThe categories of condom used are based on the percentage of all coital acts in which condoms were used, either alone or jointly with nonoxyonol-9 suppositories. The referent group is women who used condoms for less than 50% of all coital acts. cA dichotomous variable indicating whether the mean number of partners per month throughout follow-up was 15 or more. The referent group is women with fewer than 15 partners per month.
Requests for reprints should be sent to Paul J. Feldblum, PhD, Family Health International, PO Box 13950, Research Triangle Park Branch, Durham, NC 27709. Note. The views expressed in this letter do not necessarily reflect those of the funding agency. Editor's Note. See related editorial by Potts (p 890) and article by Weir et al. (p 910) in this issue.
Acknowledgments Partial support for this work was provided by Family Health International (FHI) with funds from the US Agency for International Development. FHI is an international nonprofit organization that conducts research and provides technical assistance in reproductive health, family planning, sexually transmitted diseases, and AIDS.
References 1. Centers for Disease Control and Prevention. Update: barrier protection against HIV infection and other sexually transmitted diseases. MMW 1993;42:589-591, 597. 2. Kreiss J, Ngugi E, Holmes K, et al. Efficacy of nonoxynol-9 contraceptive sponge use in
American Journal of Public Health 1033
LAcuen to te Ior preventing heterosexual acquisition of HIV in Nairobi prostitutes. JAM4. 1992;268:477482. 3. Zekeng L, Feldblum PJ, Oliver RM, Kaptue L. Barrier contraceptive use and HIV infection among high-risk women in Cameroon. AIDS. 1993;7:725-731. 4. SAS/STAT Software: The PHREG Procedure, Version 6. Cary, NC: SAS Institute Inc; 1991. SAS Technical Report P-217. 5. Roddy RE, Cordero M, Cordero C, Fortney JA. A dosing study of nonoxynol-9 and genital irritation. Int JSex Transm DisAIDS. 1993;4:165-170. 6. Weir SS, Feldblum PJ, Zekeng L, Roddy RE. The use of nonoxynol-9 for protection against cervical gonorrhea. Am J Public Health 1994;84:910-914.
Smoking and Drug Interactions Virginia Ernster, in her excellent editorial in the September issue, summarized the increased risk of various health endpoints-such as cancer, heart attack, stroke, and emphysema-faced by smokers.1 Another endpoint often neglected is that of altered drug response. Tobacco smoke is a potent inducer of hepatic enzymes.Z3 As a result, the metabolism of several medications is enhanced, which often leads to decreased efficacy. For example, smokers may require from one third to twice the dose of theophylline needed by nonsmokers.35 The response to certain pain medication, such as propoxyphene, is diminished in smokers.6 And the interaction between smoking and oral contraceptives is a complex and deadly one; women more than 35 years old who smoke more than 15 cigarettes a day are clearly at increased risk of myocardial infarction.7 The alteration of drug response in smokers must be added to the overwhelming weight of evidence against smoldng. l
Jeff Schein Mr Schein is a doctoral candidate in Public Health at UMDNJ-Robert Wood Johnson Medical School, Piscataway, NJ. Request for reprints should be sent to Jeff Schein, 330 Harper P1, Highland Park, NJ 08904.
References 1. Ernster VL. Women and smoking. Am J
Public Healh. 1993;83:1202-1204. Editorial.
2. Dawson GW, Vestal RE. Smoking and drug metabolism. Phannacol Ther. 1982;15:207-
221. 3. Miller LG. Recent developments in the study of the effects of cigarette smoking on clinical pharmacokinetics and clinical pharmacodynamics. Clini Phannacokun. 1989;17: 90-108. 4. Jenne J, Nagasawa H, McHugh R, et al. 1034 American Journal of Public Health
Decreased theophylline half-life in cigarette smokers. Life Sci 1975;17:195. 5. Jusko WJ, Schentag JJ, Clark JH, et al. Enhanced biotransformation of theophylline in marihuana and tobacco smokers. Clin Pharmacol Ther. 1978;24:406. 6. Boston Collaborative Drug Surveillance Program. Clinical depression of the central nervous system due to diazepam and chlordiazepoxide in relation to cigarette smoking and age. NEnglJMed. 1973;288:277. 7. Rubenstein E, Federman DD. Cardiovascular medicine. In: Scientific American Medicine. New York, NY: Scientific American Inc; 1990: chap 1, p 14.
Access to Comprehensive Health Services Is Fundamental I am writing to commend the editors and contributors for an exceptional March 1993 issue of the Journal and to comment on the dearth of primary care physicians graduating from medical training in the United States and the problems of the British Health Service. Susserl points out that, according to health outcome measures in Britain, the lower socioeconomic strata have failed to close the health gap between themselves and higher socioeconomic groups despite full access to care. He attributes this failure to a decline in the power and effectiveness of the public health authorities. I surmise that Susser's point is to warn public health practitioners that upcoming US reforms are likely to fail if they do not also strengthen the role and authority of the public health sector (i.e., health promotion, health education, populationwide disease prevention, system evaluation, and health standards enforcement). I do subscribe to this view. However, as Susser's first general requirement of health care as a right is access, health care problems in the United States today are more fundamentally the result of a discriminatory health care system. This, coupled with the failure to strengthen the training in and fair distribution of primary care disciplines, suggest that-in the face of uncontrolled cost-more than simple structural change will be needed. Rosenblatt et al.2 and Geiger3 point out that the National Institutes of Health research funding for medical schools totally overwhelms the small amount of funds available for primary care program development. This resource allocation problem has distorted the policy programs of our schools of public health, as well as of our medical schools, because
policy and program development have been driven more by funding than by public need. Most readers will agree that health outcome measures may not equilibrate without major public health authority and financing. But the problems of the health care system in the United States cannot be fixed by public health advocacy separately since equal access to health services is a more fundamental prerequisite. Providing sufficient primary care practitioners (who ought to be well trained in prevention and public health) to meet this need is required. O Marc Sapir, MD, MPH The author is with the Center for Elders' Independence, Oakland, Calif. Requests for reprints should be sent to Marc Sapir, MD, MPH, 1326 Spruce St, Berkeley, CA 94709. Note. The views expressed here are solely the author's.
References 1. Susser M. Health as a human right: an epidemiologist's perspective on the public
health.AmJPublicHealth 1993;83:418-426. 2. Rosenblatt RA, Whitcomb ME, Cullen TJ, Lishner DM, Hart LG. The effects of federal grants on medical schools' production of primary care physicians. Am JPublic
Health. 1993;83:322-328. 3. Geiger HJ. Why don't medical students
choose primary care? Am J Public Health. 1993;83:315. Editorial.
Vitamin Supplement Use and Mortality 1. Study That Found No Relaionship Is Chalenged In concluding that vitamin supplements have no relationship to mortality, Kim et al.1 overlooked my findings based on the same First National Health and Nu-
trition Examination Survey (NHANES I) Epidemiologic Follow-up Study (NHEFS) database. My analysis shows that there is a weak relationship in the cohort as a whole and a stronger one in a portion of the cohort. To demonstrate this relationship, I present here results based on a follow-up of the NHEFS cohort through 1987,3 the same follow-up period used by Kim et al. Of the 11 348 persons in the NHEFS cohort, 474 persons were lost to follow-up after enrollment and a total of 4333 men and 6541 women were followed. First, note that the corrected Table 1 of Kim et al. (shown here as Table 2 in Kim's response) shows lower age-specific June 1994, Vol. 84, No. 6
