THE WATER LEAK TEST, the primary test used to assure quality of condoms, is limited in its ability to detect small holes.1 Holes as small as 20 microns in diameter may be reliably detected.1,2 Although condoms have a history of successful use as contraceptives, viruses are considerably smaller than sperm, which are themselves smaller than 20 microns. Therefore, virus penetration tests were developed to determine the possible existence of previously-undetected small holes that would render condoms ineffective as virus barriers.
The first studies of condom barrier effectiveness challenged condoms with selected human viruses-human immunodeficiency virus (HIV), hepatitis B virus (HBV), and herpes simplex virus (HSV)-that cause sexually transmitted diseases (STDs).3–11 These tests also used protocols designed to simulate various conditions present during sexual intercourse, including movement. In general, the results indicated that latex condoms were effective barriers and that natural membrane condoms might not be.7,8,11 However, there was little, if any, control or knowledge of (1) the transcondom pressure that might cause the virus to pass through a hole, (2) the extent to which the condom surface is challenged, or (3) the sensitivity of the test method. Further, use of pathogenic viruses limited the number of condom samples that could be evaluated.
Scientists at the Center for Devices and Radiological Health at the Food and Drug Administration developed test protocols that were designed to simulate some conditions of actual condom use. Motion was not directly included, but the tests were designed to detect holes under pressures determined to represent the maximum generated by motion during actual use. A number of papers have been published concerning these studies.12–18 The purpose of this review is to summarize the approaches taken, the data obtained, and, based upon the combined results, the risk of exposure to (possibly disease-carrying) semen associated with condom use.
For the following discussion, a distinction will be made between condoms that have no detectable holes by the standard water leak quality assurance test,19 herein referred to as “intact” condoms, and condoms with at least one hole detectable by the water leak test but passed on to consumers because of the finite acceptable quality level (AQL) of the test, herein referred to as “water leakers.”
Certain parameters determine whether a given hole will pass fluid during a test. These parameters include transcondom pressure, fluid viscosity and surface tension, condom shape, and the extent of condom surface tested.12 Through determining appropriate values for these parameters and controlling the values during testing, we standardized the test method, providing a more sensitive test and allowing an estimate of the effective hole size when virus or microsphere passage occurred.12
The method consisted of filling a condom with buffered saline challenge solution that contained virus-sized probes (virus or microspheres) and submerging the condom in saline solution to collect any probe that passed through the condom.12–15 Two types of probes have been used: polystyrene microspheres (beads) with a fluorescent marker12,13 and “live” viruses,11,14–17 either herpes simplex virus, a human pathogenic virus, or one or more surrogate viruses. The standard test hydrostatically pressurized the challenge solution to 60 mmHg.12 The latex condoms and the polyurethane condoms required a restrainer to prevent the condoms from expanding beyond appropriate geometric proportions when pressurized. The natural membrane condom did not require a restrainer.13,14 The resultant test method has been characterized according to limits of sensitivity12,15,16 and has been used to evaluate different types of condoms.13,14,17 However, we note that the sensitivity of the test has been determined with artificially-produced holes15 that may be different from the size and shape of defects caused by manufacture or use.
These tests12,14,15 did not fully reproduce the conditions during actual condom usage, but many of the conditions expected were simulated and others were exaggerated. Those conditions that were exaggerated were such that reasonable extrapolations to conditions of normal use could be made. For example, although the tests were done statically, the pressure used was the maximum pressure achievable in a test simulating intercourse.12 Knowledge of the mean pressure in those tests and knowledge of how fluid passage is dependent upon pressure allows extrapolation of test results to those results expected during use conditions. In this analysis, we express the amount of probe passage in terms of volume of challenge suspension, as a means to normalize for different levels of challenge titer and to simplify risk assessment when the titer of the pathogen of interest is known.14 In addition, the amount of probe passage in 30 minutes was used to calculate the effective hole size so that it could be estimated whether any particular condom would have likely failed the water leak test with a single 20 micron or larger hole.1
The use of fluorescent beads has been described by Retta et al.12 and by Carey et al.13 The beads contained the fluorescent dye embedded in the microspheres. The process was controlled well enough to produce beads of quite uniform size, 110 ± 2 nm for the batch used in these tests (Polysciences, Inc., Warrington, PA). Passage rates were determined from analytical fitting routines of the observed spectrum emanating from the saline bath around the condom and comparison of that spectrum to the known spectrum of the embedded dye. Test sensitivity was such that between 10−5 and 10−6 of the challenge titer could be detected in the surrounding bath. Test sensitivity then was such that passage of roughly 1 μL of challenge fluid could be detected (the equivalent of flow through a single 3.5-micron hole for 30 minutes).13 Some dye could have dissociated from the beads and penetrated the condom through holes too small to pass a virus, thereby distorting the risk analysis by indicating a greater passage rate than appropriate. Carey et al.13 estimated the possible size of this effect. The data presented here represents the worst case finding-that all observed passage was caused by transmission of the whole beads.
Of the different challenge probes used to date,3–6,11,14 the bacteriophage (bacterial virus) φX174, a small (27 nm) surrogate virus, has been most useful. Surrogate viruses have the advantage over human pathogens of allowing the tests to be conducted more safely, more quickly, less expensively, and without complications associated with virus binding to condom surfaces.14,16,20 Surrogate virus tests are also more intuitive and subject to a more straightforward analysis than tests using fluorescent beads. The φX174 bacteriophage is special because of its size (as small as the smallest STD pathogen), stability, ease and cost of assay, and low level of adsorptivity.20,21
The φX174 method can detect passage of as little as 2 nL (2×10−6 mL) of challenge suspension.15 Studies with laser-drilled (nearly cylindrical) holes indicated that the amount of virus passage is consistent with flow of the challenge solution carrying the virus, as expected, and that the method can detect holes as small as 1 to 2 microns in diameter.15 Therefore, the detection limit of this extremely sensitive test is still much larger than the diameters of the viruses of interest (φX174, 0.027 microns; HBV, 0.040; HIV, 0.10; herpes simplex virus, 0.14).22 However, maximum expected flow through a 1-micron hole during intercourse would be about 0.05 nL. If such small holes existed on a massive scale (at least 106 per condom), they could pose a significant public health risk. However, these tests would detect them (in aggregate). It was a search for such holes that motivated the Food and Drug Administration's development of these tests.
Two important limitations in these tests were discovered during the development of the φX174 test: (1) Passage of the challenge virus might be blocked, apparently by dusting powder, reducing virus passage after about 2 minutes of flow.15,16,18 It is not known how common blocked passage is because it was only noted using φX174 as the challenge probe on one brand of latex condom with artificially-induced holes. In addition, it was not observed with several brands of condoms when fluorescent beads were used as the challenge probe.13 (2) Virus or bead adsorption to the condom surface may reduce passage through a hole.16,20,21 Although the beads used in these experiment have recently been shown to adsorb to condom latex (Kaplan and Lytle, unpublished data, November 1998), the challenge virus, φX174, does not.23 The effect of both (1) and (2) on any risk assessment would be dependent upon whether these effects would also be present during actual use.
Fluorescent Bead Penetration Through Condoms in the Laboratory
Testing was performed on 89 nonlubricated latex condoms13 and on 20 nonlubricated natural membrane condoms (previously unpublished results) (see Table 1). Of all the latex condoms tested, none would have failed the water leak test. Twenty-nine latex condoms allowed passage of detectable amounts of challenge suspension, with the median passage equal to 2.2 μL (range, 0.9-24.1). All 20 natural membrane condoms passed detectable amounts, with the median passage equal to 34.0 μL (range, 3.5-68.2).
Virus Penetration Through Condoms in the Laboratory
The results14,17 for passage of φX174 through condoms made of latex, polyurethane, and natural membranes are shown in Table 1. Of the 530 latex condoms tested in two studies, 15 allowed virus passage, 3 of which were “water leakers.” Of the 76 polyurethane condoms tested, 4 allowed virus passage, 1 of which was a “water leaker.” Of the 19 natural membrane condoms tested, 13 passed virus. A far higher percentage of natural membrane condoms allowed virus penetration. In addition, more than 50% of the natural membrane condoms allowed at least 1 μL passage (data not shown), compared to about 4% for polyurethane and 1% to 2% for latex.17 The 4 condoms that allowed the highest levels of passage accounted for more than 99% of the total virus passage among all condoms.17
Extrapolation to Expected Exposure to Semen in Real Life
A summary of the results extrapolated to actual use conditions is also presented in Table 1. Because the laboratory tests used exaggerated values for some test parameters, we computed expected actual leakage rates by applying corrections to the laboratory results. Corrections were based on expected maximum average pressure (12 mmHg) versus the test condition (60 mmHg) and semen viscosity (14 cP versus 1 cP) for all data.12 In addition, for the data with fluorescent beads, correction was made for time (10 minutes of expected use versus 30 minutes of test)13; because most virus passage through artificial holes usually occurred in less than 10 minutes,15,16 no similar time correction was made for virus passage. For “intact” latex condoms which passed virus or virus-sized beads, the extrapolated median passage amounts were approximately 0.01 μL.
Comparison of Exposure to Semen from Holes and Breaks in Latex Condoms
Although this analysis suggests that the presence of a few holes in condoms would result in exposure to relatively small volumes of semen, a condom that breaks during use would be expected to result in much greater exposure to semen. Condoms that break during use might be expected to allow exposure to most of the ejaculate, which averages 3.3 mL.24 Pooled data from five studies25–27 indicated a breakage rate for latex condoms of 1.5% (ranging from 0.9% to 1.9%). One study with Tactylon synthetic rubber condoms indicated a 1.2% breakage rate.25
The comparison of the consequences of breaks to holes can be structured along two approaches. One approach is to compare by exposure to semen on a per condom basis. This approach has value for risk assessment at the public health level, but only applies to infectious agents of low concentration and low infectivity (such as HIV). Table 2 compares expected exposures to semen for latex condoms on a per condom basis. The expected semen exposure per latex condom from breakage during intercourse might be around 1 mL per broken condom or 15 μL per condom used (1.5% × 1.0 mL). The values for expected exposure through holes was calculated using the values in Table 1: the median volume times the fraction of condoms that allowed virus passage. For example, in the larger φX174 study,17 the expected semen exposure from “water leaker” condoms would be 12 μL × 2/470 = 0.051 μL; for “intact” condoms, the expected exposure would be 0.007 μL × 10/468 = 1.5 × 10−4 μL. The estimated semen exposure from the different studies are presented in Table 2. In general, exposure to semen from breakage appears to be greater than from holes, even for holes in natural membrane condoms.
The other approach is to look at the percentage of condoms that allow an “infectious dose” of semen to pass. Results for this approach are presented in Table 3, using values from the larger φX174 study.17 This approach demonstrated that where relatively large semen passage is necessary for disease transmission (0.1 mL or 1.0 mL), only breakage of condoms is important, as was previously determined. However, as smaller volumes are considered, holes contribute more. In fact, holes became as important as breaks for passage of semen amounts less than 0.00001 mL.
Extrapolation to actual use conditions from the passage of viruses and beads through condoms (under exaggerated laboratory conditions) indicates that, except for semen passage less than 0.0001 mL, holes in “intact” condoms are not as great a concern for exposure to semen as exposure through holes in water leakers. The number of the latter condoms allowed to pass the AQL has recently been lowered from <0.4% to ≤0.25%. It seems reasonable to assume that lowering the AQL will lower the number of holes overall, in both water leakers and the intact condoms.
In addition, the exposure amount from any given hole was much smaller than the exposure amount caused by breakage of condoms during actual use. Therefore, the primary finding of the original studies13,14,17 and of this review is that, for infectious agents with low titer and low infectivity (such as HIV), leakage through pores too small to be detected by the water leak test is not the primary public health risk of condom use. Although the 0.25% AQL of the water leak test does create some risk of exposure to HIV, condom breakage appears to be the critical parameter for exposure during intercourse. Partly as a result of this conclusion, the air burst test28 has been implemented recently in the United States as an additional important quality assurance test for condoms because it is believed to relate to strength (and therefore resistance to breakage).
It should be noted that risk is not necessarily proportional to exposure to a volume of semen. Estimation of risk requires further extrapolation because it depends on additional variables, especially the infectious agent of interest. The concentration of virus particles and the probability of an individual virus particle causing an infection clearly need to be considered. For example, exposure to HIV (hypothetical maximum concentration of free virus in semen, 102 HIV/mL29) through a hole might occur in one latex condom of 470 (example from the larger φX174 study17), whereas exposure to HBV (maximum concentration in semen, 106 HBV/mL30) through holes might occur in 8 condoms of 470. However, the difference in actual risk of infection probably would be greater than 8-fold because the infectivity of HBV is greater than that of HIV.31 The situation for natural membrane condoms indicates an even greater difference for HIV and HBV in semen; a small percentage of those condoms might allow exposure to HIV, but most would allow HBV exposure.
The protection afforded by condoms is imperfect. However, condom use provides a significant public health measure to limit the spread of AIDS and other STDs.32 The Food and Drug Administration has expended considerable effort to establish critical test methods for qualifying new condom products. Laboratory use of these tests have confirmed the advantage of condom use. Different methods have corroborated this conclusion.
1. Herman BA, Carey RF, Rinaldi JE. Sensitivity of water leak tests for latex condoms. J Test Eval 1993; 21:124–128.
2. Schmukler R, Casamento J, Baier RE, Beard RB. Testing of the barrier function of condoms: an overview. Annual Meeting of the American Society of Mechanical Engineers, San Francisco, CA, December 10–15, 1989. Paper 89-WA/NDE-1.
3. Conant MA, Spicer DW, Smith CD. Herpes simplex virus transmission: condom studies. Sex Transm Dis 1984; 11:94–95.
4. Katznelson S, Drew WL, Mintz L. Efficacy of the condom as a barrier to the transmission of cytomegalovirus. J Infect Dis 1984; 150:155–157.
5. Minuk GY, Bohme CE, Bowen TJ. Condoms and hepatitis B virus infection. Ann Intern Med 1986; 104:84.
6. Conant M, Hardy D, Sernatinger J, Spicer D, Levy JA. Condoms prevent transmission of AIDS-associated retrovirus. JAMA 1986; 255:1706.
7. Minuk GY, Bohme CE, Bowen TJ, Hoar DI, Cassol S. Condoms and the prevention of AIDS. JAMA 1986; 256:482.
8. Minuk GY, Bohma CE, Bowen TJ, et al. Efficacy of commercial condoms in the prevention of hepatitis B virus infection. Gastroenterology 1987; 93:710–714.
9. Van de Perre P, Jacobs D, Sprecher-Goldberger S. The latex condom, an efficient barrier against sexual transmission of AIDS-related viruses. AIDS 1987; 1:49–52.
10. Rietmeijer CAM, Krebs JW, Feorino PM, Judson FN. Condoms as physical and chemical barriers against human immunodeficiency virus. JAMA 1988; 259:1851–1853.
11. Lytle CD, Carney PG, Vohra S, Cyr WH, Bockstahler LE. Virus leakage through natural membrane condoms. Sex Transm Dis 1990; 17:58–62.
12. Retta SM, Herman WA, Rinaldi JE, Carey RF, Herman BA, Athey TW. Test method for evaluating the permeability of intact prophylactics to viral-size microspheres under simulated physiologic conditions. Sex Transm Dis 1991; 18:111–118.
13. Carey RF, Herman WA, Retta RS, Rinaldi JE, Herman BA, Athey TW. Effectiveness of latex condoms as a barrier to human immunodeficiency virus-sized particles under conditions of simulated use. Sex Transm Dis 1992; 19:230–234.
14. Lytle CD, Routson LB, Cyr WH. A simple method to test condoms for penetration by viruses. Appl Environ Microbiol 1992; 58:3180–3182.
15. Lytle CD, Routson LB, Duff JE, Fleharty B, Cyr WH. A sensitive method for evaluating condoms as virus barriers. J AOAC Int 1997; 80:319–324.
16. Lytle CD, Routson LB, Thomas DP, Regnault WF, Cyr WH. Two parameters limiting the sensitivity of laboratory tests of condoms as viral barriers. J Test Eval 1996; 24:279–286.
17. Lytle CD, Routson LB, Seaborn GB, Dixon LG, Cyr WH. An in vitro evaluation of condoms as barriers to a small virus. Sex Transm Dis 1997; 24:161–164.
18. Mehta RI, Lytle CD, Thomas DP, Myers MR. The cause of cessation of viral passage through holes in latex condoms. J Rubber Res 1998; 1:1–13.
19. Pierdominici VJB Jr, Doucette JM, Boivin WS, Neunaber LM. Water leak testing of male latex condoms. Laboratory Information Bulletin, No. 3970, July 1995.
20. Lytle CD, Routson LB. Minimized virus binding for tests of barrier materials. Appl Environ Microbiol 1995; 61: 643–649.
21. Lytle CD, Truscott W, Budacz AP, Venegas L, Routson LB, Cyr WH. Important factors for testing barrier materials with surrogate viruses. Appl Environ Microbiol 1991; 7:2549–2554.
22. Lytle CD, Cyr WH, Carey RF, et al. Standard quality control testing and virus penetration. In: Protective Gloves for Occupational Use. Mellstrom GA, Wahlberg JE, Maibach HI, eds. Boca Raton: CRC Press, 1994;109–127.
23. Myers MR, Lytle CD, Routson LB. A mathematical model for simulating virus transport through synthetic barriers. Bull Math Bio 1999;61:111–138.
24. Ciba-Geigy. Units of measurement, blood fluids, composition of the body, nutrition. In: Lentner C, ed. Geigy Scientific Tables. Vol. 1. Basel: Ciba-Geigy Limited, 1981:185.
25. Trussell J, Warner DL, Hatcher R. Condom performance during vaginal intercourse: comparison of Trojan-Enz and Tactylon condoms. Contraception 1992; 45:11–15.
26. Steiner M, Piedrahita C, Joanis C, Glover L, Spruyt A. Condom breakage and slippage rates among study participants in eight countries. Intern Family Plan Perspec 1994; 20:55–58.
27. Gerofi J. Condom usage study-abbreviated initial report. Reported to International Standards Organization TC157 Committee on Mechanical Contraceptives, Feb. 29, 1996. Pattaya, Thailand.
28. International Standard ISO 4074 Rubber Condoms-Part 6; Determination of bursting volume and pressure. International Standards Organization, 1984.
29. Levy JA. HIV and the Pathogenesis of AIDS. Washington, DC: ASM Press, 1994:12–16.
30. Jenison SA, Lemon SM, Baker LN, Newbold JE. Quantitative analysis of hepatitis B virus DNA in saliva and semen of chronically infected homosexual men. J Infect Dis 1987; 156:299–307.
31. Holmberg SD, Horsburgh CR Jr, Ward JW, Jaffe HW. Biologic factors in the sexual transmission of human immunodeficiency virus. J Infect Dis 1989; 160:116–125.
32. Gordon R. A critical review of the physics and statistics of condoms and their role in individual versus societal survival of the AIDS epidemic. J Sex Marital Ther 1989; 15:5–30.