Application of Hill's Criteria
Because direct experimental evidence is unavailable, applying Hill's criteria to the literature reviewed provides a framework for assessing whether the relationship is a causation or an association. The nine criteria were applied to the evidence in the literature we reviewed, and the results are summarized below.
Strength of Association. The strength of the association is the first criterion used by Hill in determining whether an association is causal.5 Most of the literature indicates that both STD and GUDs are associated with an increased risk for the transmission of HIV, even when adjusted for confounders such as sexual behavior, although the effects are less pronounced in male‐to‐female transmission, possibly because of diagnostic difficulties and asymptomatic disease among women. Compared with men exposed heterosexually, women may have a greater predisposition for heterosexually acquired HIV in the absence of a GUD. This may reduce the differential magnitude of observed risk imparted by a GUD in women. Female‐to‐female transmission appears to be unreported in the literature.
It is difficult to arrive at a level of effect because of the differences in study methods, populations, and statistical techniques used for analysis, but estimates of the effect from the more stringently controlled studies range .31 from approximately 2.0 to 3.5, with some groups at higher risk (homosexual men) in the 8.0 to 13.0 range.31 These reasonably large odds ratios, especially in the groups at higher risk, support a possible causal relationship.
Torian et al16 showed that the magnitude of HIV sero‐prevalence has increased over time among the subset of patients with GUDs, even though the overall HIV sero‐prevalence and overall GUD incidence have declined during the same period. In addition, they showed that the association between nonulcerative STD and HIV has decreased over time.
Consistency. An observed association is consistent if it has “been repeatedly observed by different persons, in different places, circumstances and times.”5 The consistency of the association between GUD and HIV infection has been established by a positive result in different populations and in different locations. A statistically significant association between GUD, HSV, or syphilis, is reported in multiple studies. Statistical significance was reported in 11 of the 14 articles that analyzed GUD and HIV,6,8,13–16,18,21,25,29,30 7 of the 12 articles that analyzed HSV and HIV,2,3,9,11,19,23,27 and 11 of the 18 articles that analyzed syphilis and HIV.2,7,11,19,20,23–28 The magnitude of association is similar across studies. No study has reported a statistically significant negative relationship between HIV and GUD, HSV, or syphilis.
Specificity. Specificity addresses the association of a particular factor to a particular disease. For this review, do specifically ulcerative STD increase transmission of HIV, or do other factors, such as nonulcerative STD or behavior, also increase HIV transmission? Numerous other factors have been proposed as associated with an increased risk for HIV acquisition. Some of these factors, such as a large number of lifetime sex partners, also would be expected to be associated with the incidence of GUD. To account for the confounding effect of these factors and to help estimate the specific role of HSV, syphilis, or unspecified GUD in HIV acquisition, some investigators performed multivariate analyses to adjust simultaneously for potential confounders.
Eighteen articles2,3,6,8,11,13–16,20,21,23–25,27–30 reported the use of logistic regression analysis to adjust for multiple confounders in 23 separate analyses. Of these, all 10 logistic analyses of any GUD were significant,6,8,13–16,21,25,29,30 as were all 5 analyses of HSV2,3,11,23 and 5 of 8 analyses of syphilis.2,20,23–25,27,28 The results of 20 analyses indicated a significant association between GUD and HIV infection.2,3,6,8,11,13–16,20,21,23–25,27–30 The three analyses that yielded nonsignificant results from a logistic regression analysis all examined syphilis as the dependent variable.2,23,25 In two of these analyses,23,25 adjusting for confounders eliminated a significant association observed in bivariate analyses of the association of syphilis and HIV seropositivity. These analyses suggest that GUD is a specific risk for HIV infection after adjustment for other confounding factors.
Nonulcerative STD and HIV infection were not specifically addressed in the studies included in our review unless these diseases were analyzed in addition to the GUDs that formed the basis for this review. Wasserheit31 comprehensively reviewed studies of nonulcerative STD. She included in her review only studies that controlled for potential effects of sexual behavior, suggesting that other mechanisms may account for the association between all STD, ulcerative and nonulcerative, and HIV transmission. In her review, she concluded that the body of evidence supports increased risk for HIV transmission associated with both nonulcerative and ulcerative STD. She summarizes that STD may be associated with HIV transmission by recruitment of HIV‐susceptible or HIV‐infected cells or by the disruption of epithelial barriers in the genital tract. She concluded that the “efficiency with which GUD increases HIV transmission, however, may be due in part to the simultaneous operation of both of these mechanisms.”
Temporality. The temporal relationship of GUD and HIV infection is important because the GUD must proceed or coincide with HIV infection if the GUD is to enhance infectivity. Many case‐control and cross‐sectional studies examined and found a relationship between history of GUD with current HIV seropositivity; however, this does not show that GUD preceded HIV. To examine this adequately, a study must be conducted to look for the presence of a GUD before conversion. Of the 14 studies we reviewed that examined the relationship of GUD to HIV, five looked specifically at the presence of genital ulcers before HIV seroconversion.8,9,12,13,15 Plummer et al13 and Telzak et al15 found a statistically significant positive relationship. Kassler et al8 reported a nonsignificant association in their matched analysis (OR = 2.7; 95% CI, 0.6 to 11.9) but a strong association in a multiple logistic analysis (OR = 11.3; 95% CI, 1.6 to 80.2). However, the wide confidence interval indicates that the study population was too small or that the incidence of GUD was too low to detect such an association with adequate power. Keet et al9 found no association between preexisting GUD and subsequent risk for HIV seroconversion. The authors, note, however, that, unlike some other cohorts of gay men, this group had a high prevalence of HSV‐2 infection before HIV seroconversion, which may have created a “saturation effect.” Laga et al12 found an association that was statistically nonsignificant, possibly because of lack of statistical power.
The presence of the HSV before HIV seroconversion was examined in 6 of 12 articles reviewed that looked specifically at HSV‐2.3,9,10,11,15,17 Holmberg et al,3 Kuiken et al,11 and Keet et al9 reported a significant association between HSV‐2 before HIV seroconversion. Keet et al9 also looked at HSV‐2 seroconversion before HIV seroconversion but found no association. Kingsley et al,10 Telzak et al,15 and VanRaden et al17 did not find an association with HIV, but, as stated previously, two of these studies were conducted in homosexual men, and VanRaden et al17 could been subject to misclassification by relying on personal history of HSV‐2 without clinical evidence.
Of the 18 studies reviewed that looked specifically at the association of syphilis and HIV, six examined the temporal relationship of a syphilis diagnosis before HIV seroconversion.7,8,11,12,15,17 Darrow et al7 and Kuiken et al11 showed a significant association. Four studies did not show an association.8,12,15,17
None of the study findings that have been reviewed report whether the outcome preceded the exposure—that is, whether an increased HIV seroconversion rate occurs in those in whom GUDs subsequently develop.
Biologic Gradient. Hill's fifth criterion states that “if the association is one which can reveal a biological gradient, or dose‐response curve, then we should look most carefully for such evidence.”5 Plummer et al13 support an association between GUD and the risk for HIV seroconversion by showing a dose‐response relationship between the annual frequency of GUD and HIV seroconversion. Their results showed that the mean number of annual ulcer episodes was 1.32 ± 0.55 in women who seroconvert and 0.48 ± 0.21 in women who are seronegative (P < 0.02). The seroconversion rates were approximately 60%, 80%, or 100%, when the annual ulcer rate was <1, 1 to 3, or >3, respectively. Boulos et al19 also support this idea by showing higher odds ratios observed for the association of HSV‐2 with HIV (OR = 7.5; 95% CI, 3.4 to 18.5) compared with the association of syphilis with HIV (OR = 3.4; 95% CI, 1.2 to 9.3). This is “consistent with the premise that repeated disruptions of genital mucous membranes in women increases the efficiency of sexual transmission of HIV‐1” because persons with HSV‐2 have more outbreaks of ulcerative lesions than do those with syphilis.19
Biologic Plausibility. The literature support the hypothesis that STD, and particularly GUDs, facilitate HIV transmission by compromising the physical or immunologic integrity of the genital mucosa. Biologic mechanisms, both for STD and GUDs, have been suggested in an attempt to understand how the transmission of HIV is increased. Several authors speculated that the STD‐HIV interaction might be explained by the fact that STD induce macrophage and T‐lymphocyte activation. Both cell types have been shown in vitro to be more susceptible to HIV infection than unstimulated cells.1,2 Sexually transmitted diseases also may increase susceptibility to HIV by causing immunosuppression.4 It has been suggested, as well, that the transmission could be caused by the interaction of HIV and STD pathogens.31 Holmes postulates that increased transmission with nonulcerative STD results from the same general mechanisms as the classic GUDs.32 At his presentation at the First National Conference on Human Retroviruses (Washington DC, 1993), he speculated that STD such as gonorrhea, chlamydia, or trichomoniasis, typically considered nonulcerative diseases, enhance the risk for HIV transmission through inflammation of epithelium of the urethra, cervix, or penis. Levine et al33 also studied non‐GUD STD enhancement of HIV transmission and theorized that increased susceptibility may be caused by the significant increase in endocervical CD4 lymphocytes.
Biologic mechanisms by which GUDs could facilitate HIV transmission are accepted widely. It has been shown that the integrity of the genital mucosa protects against infections34 and that ulceration breaks the integrity of this mucosa, thus allowing easier access of HIV into the bloodstream.3
Analogy. The association between GUD and increased risk for HIV infection also meets the criterion of analogy. Findings of other studies indicate that STD are associated with an increased risk for infection with human T‐lymphotrophic virus type‐I (HTLV‐I), another retrovirus.35,36 Moreover, nonulcerative STD have been associated with an increased risk for infection with other STD.37
Coherence. For Hill's criterion of coherence, he stressed that “the cause‐and‐effect interpretation of our data should not seriously conflict with the generally known facts of the natural history and biology of the disease.”5 An association of GUD and HIV infection does not conflict with generally accepted biologic facts of the natural history of HIV infections. A primary route of heterosexual virus transmission is through the membranes of the genital tract.
Experiment. Although the focus of this review comprises observational studies (because it is generally more accepted that experimentation would be unethical in humans), Grosskurth et al38 reported a clinical trial evaluating the impact of treatment of STD, including GUDs, on HIV transmission. This trial showed that improved STD treatment reduced the HIV incidence by approximately 40% (P = 0.007). This study does not appear to be confounded by sexual behavior because no change in reported sexual behavior was observed in either the treated or the untreated groups.
Discussion and Conclusions
Most of the literature we reviewed provides evidence of a causal relation between HIV infection and GUD, syphilis, HSV, or all three and shows a statistically significant interaction even when adjusted for confounders such as sexual behavior. Applying Hill's criteria to the literature provided additional evidence that genital ulcers are associated with an increased risk for the development of HIV infection.
The Evidence Needed: Future Studies
Future studies of the association between GUD and HIV infection should be conducted carefully to detect a clear temporal association between previous GUD and subsequent HIV status. Several studies have assessed HIV seroprevalence by looking for coexisting GUD and HIV infections. The resultant measures of association are unlikely to produce accurate estimates because the time to seroconversion for HIV typically exceeds the incubation period for GUDs by several weeks.26 Other potential confounders or effect modifiers, e.g., the number of lifetime sex partners and substance abuse, also should be assessed. Martin et al39 suggest that among a population with a high prevalence of cocaine abuse, GUD “simply could be a marker for additional high‐risk sexual behavior.”
Because of the nature of studying STD, few of the studies examining the association between GUD and HIV infection had the option of studying patients across time or obtaining extensive histories and information on other cofactors from patients. In this setting, cohort studies of patients with STD are expensive and impractical. Moreover, few studies are truly population based, and the cross‐section of the population represented at STD clinics, for example, is unlikely to reflect the risk profile that exists in the community at large. Such a truly representative, population‐based study may not even be possible for most STD because many STD go undiagnosed or unreported by private physicians.40 Genital ulcer disease, particularly primary and latent syphilis and HSV, often go unnoticed by patients. In some instances, HSV or latent syphilis can persist for years before diagnosis. Sexually transmitted diseases also can be self‐treated using antibiotics purchased on the street or through other means.41 Studies of STD clinics are particularly problematic in assessing risk in the population because all persons who are infected with HIV—but in whom STD do not develop or who do not seek treatment for STD—serve to increase the measured association between HIV and GUDs among patients at STD clinics.6
Despite these problems in measuring the true risk for HIV infection associated with preexisting GUD, the evidence provided by the published literature strongly supports increased risk. In addition to a literature review, we support the hypothesis that genital ulcers are associated with an increased risk for HIV infection because of our examination of the literature using the criteria for causal inference proposed by Hill.5 These criteria consist of strength of association, consistency, specificity, temporality, biologic gradient, plausibility, analogy, coherence, and experimentation and are considered to provide a widely accepted framework for assessing a causal association when experimental evidence is unavailable.42 All nine of the criteria for causal inference were met, providing additional evidence that genital ulcers are associated with an increased risk for the development of HIV infection.
1. Lukehart SA, Baker-Zander SA, Cheri Lloyd RM, Sell S. Characteristics of lymphocyte responsiveness in early experimental syphilis. J Immunol 1980; 124:461–467.
2. Stamm WE, Handsfield HH, Rompalo AM, Ashley RL, Roberts PL, Corey L. The association between genital ulcer disease and acquisition of HIV infection in homosexual men. JAMA 1988; 260:1429–1433.
3. Holmberg SD, Stewart JA, Gerber AR, et al. Prior herpes simplex virus type 2 infection as a risk factor for HIV infection. JAMA 1988; 259:1048–1050.
4. Hirsch MS, Schooley RT, Ho DD, Kaplan JC. Possible viral interaction in the acquired immunodeficiency syndrome (AIDS). Rev Infect Dis 1984; 6:726–731.
5. Hill AB. Statistical evidence and inference. In: Priciples of Medical Statistics. 9th ed. New York: Oxford University Press, 1971; 309–323.
6. Cameron DW, Simonsen JN, D'Costa LJ, et al. Female to male transmission of human immunodeficiency virus type 1: Risk factors for seroconversion in men. Lancet 1989; 2:403–407.
7. Darrow WW, Echenberg DF, Jaffe HW, et al. Risk factors for human immunodeficiency virus (HIV) infections in homosexual men. Am J Public Health 1987; 77:479–483.
8. Kassler WJ, Zenilman JM, Erickson B, Fox R, Peterman TA, Hook EW. Seroconversion in patients attending sexually transmitted disease clinics. AIDS 1994; 8:351–355.
9. Keet IPM, Lee FK, van Griensven GJP, Lange JMA, Nahmias A, Coutinho RA. Herpes simplex virus type 2 and other genital ulcerative infections as a risk factor for HIV-1 acquisition. Genitourin Med 1990; 66:330–333.
10. Kingsley LA, Armstrong J, Rahman A, et al. No association between herpes simplex virus type 2 seropositive or anogenital lesions and HIV seroconversion among homosexual men. J Acquir Immune Defic Syndr 1990; 3:773–779.
11. Kuiken CL, van Griensven GJP, de Vroome EMM, Coutinho RA. Risk factors and changes in sexual behavior in male homosexuals who seroconverted for human immunodeficiency virus antibodies. Am J Epidemiol 1990; 132:523–530.
12. Laga M, Monoka A, Kivuvu M, et al. Non-ulcerative sexually transmitted diseases as risk factors for HIV-1 transmission in women: Results from a cohort study. AIDS 1993; 7:95–102.
13. Plummer FA, Simonsen JN, Cameron DW, et al. Cofactors in male-female sexual transmission of human immunodeficiency virus type 1. J Infect Dis 1991; 163:233–239.
14. Simonsen JN, Plummer FA, Ngugi EN, et al. Human immunodeficiency virus infection among lower socioeconomic strata prostitutes in Nairobi. AIDS 1990; 4:139–144.
15. Telzak EE, Chiasson MA, Bevier PJ, Stoneburner RL, Castro KG, Jaffe HW. HIV-1 seroconversion in patients with and without genital ulcer disease. Ann Intern Med 1993; 119:1181–1186.
16. Torian LV, Weisfuse IB, Makki HA, Benson DA, DiCamillo LM, Toribio FE. Increasing HIV-1 seroprevalence associated with genital ulcer disease, New York City, 1990–1992. AIDS 1995; 9:177–181.
17. VanRaden M, Kaslow R, Kingley L, et al. The role of ulcerative genital diseases in promoting acquisition of HIV-1 by homosexual men. (Abstract Th.A.O. 17). In: Program and Abstracts of the Fifth International Conference of AIDS. Montreal, Canada: June 4–9, 1989.
18. Bassett MT, Latif AS, Katzenstein DA, Emmanuel JC. Sexual behavior and risk factors for HIV infection in a group of male factory workers who donated blood in Harare, Zimbabwe. J Acquir Immune Defic Syndr 1992; 5:556–559.
19. Boulos R, Ruff AJ, Nahmias A, et al. Herpes simplex virus type 2 infection, syphilis, and hepatitis B virus infection in Haitian women with human immunodeficiency virus type 1 and human T lymphotropic virus type 1 infections. J Infect Dis 1992; 166:418–420.
20. Elifson KW, Boles J, Sweat M. Risk factors associated with HIV infection among male prostitutes. Am J Public Health 1993; 83:79–83.
21. Greenblatt RM, Lukehart SA, Plummer FA, et al. Genital ulceration as a risk factor for human immunodeficiency virus infection. AIDS 1988; 2:47–50.
22. Hayes CG, Manaloto CR, Basaca-Sevilla V, et al. Epidemiology of HIV infection among prostitutes in the Philippines. J Acquir Immune Defic Syndr 1990; 3:913–920.
23. Hook EW, Cannon RO, Nahmias AJ, et al. Herpes simplex virus infection as a risk factor for human immunodeficiency virus infection in heterosexuals. J Infect Dis 1992; 165:251–255.
24. Hunter DJ, Maggwa BN, Mati KG, Tukei PM, Mbugua S. Sexual behavior, sexually transmitted diseases, male circumcision and risk for HIV infection among women in Nairobi, Kenya. AIDS 1994; 8:93–99.
25. Kreiss JK, Koech D, Plummer FA, et al. AIDS virus infection in Nairobi prostitutes: Spread of the epidemic to East Africa. N Engl J Med 1986; 314:414–418.
26. Miotti P, Dallabetta G, Ndovi E, Liomba G, Saah AJ, Chiphangwi J. HIV-1 and pregnant women: Associated factors, prevalence, estimate of incidence and role in fetal wastage in central Africa. AIDS 1990; 4:733–736.
27. Quinn TC, Glasser D, Cannon RO, et al. Human immunodeficiency virus infection among patients attending clinics for sexually transmitted diseases. N Engl J Med 1988; 318:197–203.
28. Quinn TC, Cannon RO, Glasser D, et al. The association of syphilis with risk for human immunodeficiency virus infection in patients attending sexually transmitted disease clinics. Arch Intern Med 1990; 150:1297–1302.
29. Ryder RW, Ndilu M, Hassig S, et al. Heterosexual transmission of HIV-1 among employees and their spouses at two large businesses in Zaire. AIDS 1990; 4:725–732.
30. Simonsen JN, Cameron DW, Gakinya MN, et al. Human immunodeficiency virus infection among men with sexually transmitted diseases: Experience from a center in Africa. N Engl J Med 1988; 319:274–277.
31. Wasserheit JN. Epidemiological synergy: Interrelationships between human immunodeficiency virus infection and other sexually transmitted diseases. Sex Transm Dis 1992; 19:61–77.
32. Holmes K. Sexually transmitted diseases as a co-factor in HIV infection. Presented at the 1st National Conference on Human Retroviruses, Washington DC, 1993.
33. Levine WC, Pope V, Bhoomkar A, et al. Increase in endocervical CD4 lymphocytes in women with non-ulcerative STD. In: Program and Abstracts of the Tenth International Conference on STD. Yokohama, Japan: August 7–12, 1994. Abstract 457C.
34. Miller CJ, Alexander NJ, Sutjipto S, et al. Genital mucosal transmission of simian immunodeficiency virus: Animal model for heterosexual transmission of human immunodeficiency virus. J Virol 1989; 63:4277–4284.
35. Brodine SK, Oldfield EC, Corwin AL, et al. HTLV-I among U.S. marines stationed in a hyperendemic area: Evidence for female-to-male sexual transmissions. J Acquir Immune Defic Syndr 1992; 5:158–162.
36. Murphy EL, Figueroa JP, Gibbs WN, et al. Sexual transmission of human T-lymphotrophic virus type I (HTLV-I). Ann Intern Med 1989;111:555–560.
37. Hillis SD, Nakashima A, Marchbanks PA, Addiss DG, Davis JP. Risk factors for recurrent Chlamydia trachomatis
infections in women. Am J Obstet Gynecol 1994; 170:801–806.
38. Grosskurth H, Mosha F, Todd J, et al. Impact of improved treatment of sexually transmitted diseases on HIV infection in rural Tanzania: Randomised controlled trial. Lancet 1995; 346:530–536.
39. Martin DH, DiCarlo RP. Recent changes in the epidemiology of genital ulcer disease in the United States: The crack cocaine connection. Sex Transm Dis 1994; 21:S76–S80.
40. Fife D, MacGregor RR, McAnaney J. Limitations of AIDS reporting under favorable conditions. Am J Prev Med 1993; 9:317–320.
41. Zenilman JM, Bonner M, Sharp KL, et al. Penicillinase-producing Neisseria gonorrhoeae
in Dade County, Florida: Evidence of core-group transmitters and the impact of illicit antibiotics. Sex Transm Dis 1988; 15:45–50.
© Copyright 1996 American Sexually Transmitted Diseases Association
42. Steinberg J, Goodwin PJ. Alcohol and breast cancer risk—Putting the current controversy into perspective. Breast Cancer Res Treatment 1991; 19:221–231.