Title: ANDROLOGY LAB CORNER*: One Semen Sample or 2? Insights From a Study of Fertile Men
Abstract: Journal of AndrologyVolume 28, Issue 5 p. 638-643 Free Access ANDROLOGY LAB CORNER*: One Semen Sample or 2? Insights From a Study of Fertile Men Abbie Stokes-Riner, Corresponding Author Abbie Stokes-Riner Department of Biostatistics and Computational Biology, University of Rochester, Rochester, New York Department of Biostatistics and Computational Biology, University of Rochester Medical Center, 601 Elmwood Ave, Box 630, Rochester, NY 14642 (e-mail: [email protected]).Search for more papers by this authorSally W. Thurston, Sally W. Thurston Department of Biostatistics and Computational Biology, University of Rochester, Rochester, New YorkSearch for more papers by this authorCharlene Brazil, Charlene Brazil Center for Health and the Environment, Department of Obstetrics and Gynecology, University of California, Davis, CaliforniaSearch for more papers by this authorDavid Guzick, David Guzick Department of Obstetrics and Gynecology, University of Rochester, Rochester, New YorkSearch for more papers by this authorFan Liu, Fan Liu Department of Obstetrics and Gynecology, University of Rochester, Rochester, New YorkSearch for more papers by this authorJames W. Overstreet, James W. Overstreet Center for Health and the Environment, Department of Obstetrics and Gynecology, University of California, Davis, CaliforniaSearch for more papers by this authorChristina Wang, Christina Wang Department of Medicine, Harbor-UCLA Medical Center and Los Angeles Biomedical Research Institute, Los Angeles, CaliforniaSearch for more papers by this authorAmy Sparks, Amy Sparks Departments of Urology and Obstetrics and Gynecology, University of Iowa, Iowa City, IowaSearch for more papers by this authorJ. Bruce Redmon, J. Bruce Redmon Departments of Medicine and Urologic Surgery, University of Minnesota, Minneapolis, Minnesota.Search for more papers by this authorShanna H. Swan, Shanna H. Swan Department of Obstetrics and Gynecology, University of Rochester, Rochester, New YorkSearch for more papers by this author Abbie Stokes-Riner, Corresponding Author Abbie Stokes-Riner Department of Biostatistics and Computational Biology, University of Rochester, Rochester, New York Department of Biostatistics and Computational Biology, University of Rochester Medical Center, 601 Elmwood Ave, Box 630, Rochester, NY 14642 (e-mail: [email protected]).Search for more papers by this authorSally W. Thurston, Sally W. Thurston Department of Biostatistics and Computational Biology, University of Rochester, Rochester, New YorkSearch for more papers by this authorCharlene Brazil, Charlene Brazil Center for Health and the Environment, Department of Obstetrics and Gynecology, University of California, Davis, CaliforniaSearch for more papers by this authorDavid Guzick, David Guzick Department of Obstetrics and Gynecology, University of Rochester, Rochester, New YorkSearch for more papers by this authorFan Liu, Fan Liu Department of Obstetrics and Gynecology, University of Rochester, Rochester, New YorkSearch for more papers by this authorJames W. Overstreet, James W. Overstreet Center for Health and the Environment, Department of Obstetrics and Gynecology, University of California, Davis, CaliforniaSearch for more papers by this authorChristina Wang, Christina Wang Department of Medicine, Harbor-UCLA Medical Center and Los Angeles Biomedical Research Institute, Los Angeles, CaliforniaSearch for more papers by this authorAmy Sparks, Amy Sparks Departments of Urology and Obstetrics and Gynecology, University of Iowa, Iowa City, IowaSearch for more papers by this authorJ. Bruce Redmon, J. Bruce Redmon Departments of Medicine and Urologic Surgery, University of Minnesota, Minneapolis, Minnesota.Search for more papers by this authorShanna H. Swan, Shanna H. Swan Department of Obstetrics and Gynecology, University of Rochester, Rochester, New YorkSearch for more papers by this author First published: 02 January 2013 https://doi.org/10.2164/jandrol.107.002741Citations: 78 † Andrology Lab Corner: welcomes the submission of unsolicited manuscripts, requested reviews, and articles in a debate format. Manuscripts will be reviewed and edited by the Section Editor. All submissions should be sent to the Journal of Andrology Editorial Office. Letters to the editor in response to articles as well as suggested topics for future issues are encouraged. AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL In clinical settings it is usually recommended that at least 2 semen samples are obtained, because this decreases the error rate in classifying men as fertile, subfertile, or infertile (Berman et al, 1996; World Health Organization, 1999; Sharlip et al, 2002). In fact, some authors have recommended that 3 samples be collected because of the large within-man variability of semen parameters (Keel, 2006). In our prospective study of semen quality in partners of pregnant women (the Study for Future Families, or SFF), we requested 2 semen samples from each participant in order to reduce the variability of our estimates of semen quality. However, population studies of semen quality frequently collect only 1 sample from each participant (Auger and Jouannet, 1997; Jorgensen et al, 2001, 2002). Is this optimal when conducting a population study of semen quality when resources are limited? If the total number of samples to be collected is fixed, it is preferable, from a purely statistical standpoint, to take a single sample from as many men as possible, because a larger sample size will increase the precision of the population estimates. From an epidemiologic point of view, it is necessary to consider whether requesting a second sample will decrease participation or lead to a less representative study population. From a practical standpoint one needs to consider the cost of participant recruitment relative to the cost of sample analysis. In SFF we found that the cost of recruiting a participant far outweighs the cost of analyzing the semen sample(s) he provides, and the added cost for a second sample is relatively minimal. In addition, we found that once a man has agreed to provide 1 semen sample he is likely to agree to provide additional samples (88% did). However, if multiple samples are collected, and particularly if the number of samples per man varies, this raises some statistical issues, which we address here. In this study, we ask 2 questions and use data from SFF to address these. We consider the appropriateness of (1) combining data from men who gave 1 sample and data from men who gave 2 samples in the analysis and (2) combining data from both samples from men who gave 2. For the first question, we ask whether men who give 1 sample and men who give 2 samples differ, on average, with respect to either subject characteristics or semen parameters. For the second, we ask whether semen parameters differed, on average, between the first and second sample among men who gave 2 samples in our study population. Methods Study Population— We examined these 2 questions in SFF, a clinic-based study of partners of pregnant women. Methods have been described elsewhere (Swan et al, 2003). Briefly, women were recruited at prenatal clinics affiliated with university hospitals in Los Angeles, California (Harbor-UCLA Medical Center and Cedars-Sinai Medical Center); Minneapolis, Minnesota (University of Minnesota Health Center); Columbia, Missouri (University Physicians); and Iowa City, Iowa (University of Iowa). Only pregnancies that were conceived without medical intervention were eligible. The male partner was recruited if the woman agreed, and all men were asked to give 2 semen samples. The number of subjects varied by center, as did the proportion of men giving 2 semen samples. For SFF 2 samples were requested but were not required for participation. Center directors had discretion as to the emphasis they placed on obtaining the second sample; in most centers this was strongly emphasized, but in Iowa, which was added later in the study, it was stressed less. Semen Collection and Analysis— Men were requested to observe a 2–5 day abstinence period before providing each semen sample. Prior to each of the 2 visits, which were approximately 3 weeks apart, we mailed instructions regarding specimen collection, including a schedule to assist the subject in timing his last ejaculation prior to the visit. At the time of the visit the importance of accurately reporting the actual abstinence period was stressed, and the men were assured that their sample would not be rejected if they deviated from the recommended abstinence protocol. At the study visit men collected semen samples by masturbation at the clinic, and these were analyzed, on average, within 27 minutes of collection (range, 10–105 minutes). Nine subjects were excluded from our study because the analysis was started less than 10 minutes after the sample was collected. Methods used for semen evaluation have been described previously (Brazil et al, 2004a). Briefly, concentration assessments were evaluated using both hemacytometer and MicroCell counting chambers for the first semen evaluation and only MicroCell chambers for the second semen evaluation. For both methods, the final concentration was the mean of the concentration values from the separate dilutions of 2 drops, when these differed by less than 10%. If the duplicate concentrations varied by more than 10%, a third dilution was prepared and counted and the median of the 3 values used as the estimate of concentration. On average, 200–300 sperm were counted for each dilution. For MicroCell counting, semen was always diluted with equal parts of fixative to immobilize the sperm. Ejaculate volumes were estimated by specimen weight, assuming a semen density of 1.0 g/mL. For this calculation each container was preweighed and the weight (written on the container) was subtracted from the weight of the container plus the sample. In this analysis the percent motile sperm was counted in a MicroCell chamber (Overstreet and Brazil, 1997) and refers to the percentage of sperm with any flagellar movement, whether twitching or progressive. Seminal smears were prepared at the clinical centers and shipped to the Andrology Coordinating Center at the University of California, Davis, for Papanicolaou staining, analysis, and storage. A single technician assessed sperm morphology using the strict morphology method, the method recommended by the World Health Organization (1999; Guzick et al, 2001). A second technician made assessments using the 1987 World Health Organization criteria. For each determination, 200 or 300 sperm were scored. One hundred sperm were scored in each of 2 areas of the slide, and if the differences were acceptable (10% using 1987 World Health Organization criteria and 5 sperm out of 100 using 1999 World Health Organization criteria) the mean was determined. If the difference was not acceptable, an additional 100 sperm were scored from a third area and the median value used. The SFF was approved by human subject committees at all participating institutions, and all subjects signed informed consents. Statistical Analysis— In these analyses we are interested in the 4 semen parameters described above: concentration, motility, morphology, and volume. Because sperm concentration has a skewed (nonnormal) distribution, we transformed concentration using the logarithm (base 10), as has been recommended (Berman et al, 1996). Semen volume, motility, and morphology, which were not markedly skewed, were analyzed without transformation. Following the procedures of previous SFF studies (Swan et al, 2003), all analyses excluded data from men for whom either the first or second abstinence time was missing, unknown, less than 2 hours, or more than 10 days. Men who gave 1 and 2 samples were compared with respect to center as well as the following self-reported subject characteristics: education, race, smoking status, recent fever, history of sexually transmitted disease (STD), age, body mass index (BMI), and recent employment status. To address the first question (ie, whether men who gave 1 and 2 semen samples differed), we first compared all continuous variables—semen parameters, BMI, and age—between the 2 groups using t-tests. Any variable that differed significantly was then examined by center. We also compared the categorical subject characteristics between these 2 groups of men using the χ2 test for independence. We also performed covariate-adjusted analyses to compare these 2 groups. We first used a multivariate logistic regression model (Hosmer and Lemeshow, 2000) to model whether the man gave 1 or 2 samples as a function of the complete set of subject characteristics and center. We then used linear regression to model each semen parameter as a function of the complete set of subject characteristics, center, an indicator variable denoting whether 1 or 2 samples was given, abstinence time, and time to start the analysis. To address the second question (ie, whether semen parameters from the first and second semen samples differed among men who gave 2 samples), we used paired t-tests for univariate comparisons of mean differences of semen parameters between the first and second sample and examined any significant differences by center. In addition, we analyzed the difference in each semen parameter using multiple linear regression models (Weisberg, 2005), adjusting for difference in abstinence time, difference in time to start the analyses, and center. Because the 2 samples were collected only 3 weeks apart (on average), we assumed subject characteristics (eg, age, smoking) had not changed, and these were not included in these models. We included center because any possible changes in how the samples were analyzed in the time between the first and second sample collection would not necessarily be the same across center (eg, the technician may have been different for the first and second samples.) Recent fever, which may have changed between visits, was not included because it was only recorded at study entry. In these multivariate regression models, adjusted differences in semen parameters between the first and second samples would be reflected in the model intercept. Results After exclusions for long (n = 29), short (n = 1), or missing (n = 23) abstinence times, 697 men were available for analysis, of whom 615 provided 2 semen samples, an average of 24 days apart. A total of 177 were from California (96% gave 2 samples), 199 were from Minnesota (93% gave 2 samples), 183 were from Missouri (92% gave 2 samples), and 138 were from Iowa (66% gave 2 samples). Comparisons Between Men who Gave 1 and 2 Samples— The results of the univariate comparisons of subject characteristics for men who gave 1 and 2 samples are summarized in Table 1. The χ2 tests (using a continuity correction) show no appreciable differences except center (P < .001) and a suggestion of a difference in STD history (P = .079). Results of the multivariate logistic regression of the number of samples given are similar to the unadjusted results and are not reported. Centers differed with respect to the proportion of men giving 2 samples primarily because fewer men in Iowa gave 2 samples than men from other centers. Table 1. . Characteristics of study participants Men who Gave 1 Sample (n = 82) Men who Gave 2 Samples (n = 615)* No. % No. % P† Center location California 7 9 170 28 <.0001 Minnesota 13 16 186 30 … Missouri 15 18 168 27 … Iowa 47 57 91 15 … Education College 70 85 499 81 .356 Race Nonwhite 65 21 444 72 .238 Smoking status Nonsmoking 61 74 489 80 .116 ≤10 cigarettes per day 11 13 84 14 … >10 cigarettes per day 10 12 37 6 … Recent fever‡ 2 2 21 3 .888 History of STD 16 20 73 12 .079 Recently employed 72 88 570 93 .150 Mean SD Mean SD Age (y) 31.5 7.0 31.5 5.9 .999 BMI 27.7 4.8 28.6 5.4 .232 *Some of the totals are less than 82 or less than 615 because of missing values †P was determined by t-test for age and BMI and by χ2 test for all other variables ‡Test for recent fever uses the Fisher exact test We also compared the semen parameters between men who gave 1 and men who gave 2 samples. The unadjusted comparisons used paired t-tests assuming common variance (Table 2, columns A and B). There was an indication that strict morphology differed significantly between the 2 groups (P = .050). We examined this difference by center and saw that it was greatest in Minnesota (2.27%). Moreover, this difference in morphology between men giving 1 and 2 samples was not confirmed by a comparison of World Health Organization morphology (1987) measurements (P = .296). Table 2. . Summary of semen parameters Men who Gave 1 Sample (n = 82) Men who Gave 2 Samples (n = 624) A First Sample B First Sample C Second Sample P (t-test) Semen parameters Mean SD Mean SD Mean SD A vs B B vs C* Abstinence time (h) 78.7 31.2 77.6 30.5 71.9 30.6 .764 .464 Concentration† (× 106/mL) 63.4 37.2 64.3 40.0 65.3 43.0 .450 .044 Percent motile 49.0 13.9 50.7 11.4 53.0 10.6 .196 .0005 Morphology (percent normal) WHO 1987 (strict) 9.7 4.2 10.9 5.2 11.1 5.1 .050 .018 WHO 1999 58.5 9.5 57.3 10.4 57.5 9.7 .296 .131 Volume (g) 3.8 1.6 3.9 1.7 3.9 1.6 .426 .389 Time to start (min) 26.4 6.3 27.1 9.6 28.0 9.5 .516 .067 *Uses paired t-test †Tests for concentration were performed on data using the log (base 10) transformation In the adjusted linear regression models for each semen quality outcome, the indicator variable for whether the man gave 1 or 2 samples was not a significant predictor for any outcome (data not shown). As expected, greater abstinence time was associated with a significantly greater log concentration and greater volume (P < .0001 for both) but was not significantly associated with either motility or morphology. A longer time before starting analysis was associated with a lower motility (P = .015). Several self-reported characteristics were significantly associated with semen quality measures; men who had a recent fever tended to have a lower motility (P = .01), smokers had on average a smaller volume (P = .005), and there were some center differences. BMI was not included as a covariate in the linear models because it was missing for many subjects and the t-tests showed no difference in BMI for men who gave 1 or 2 samples. Comparisons of First and Second Semen Samples Among Men who Gave 2 Samples— Semen parameters for the first and second sample, from men who gave both, are summarized in Table 2 (columns B and C). The paired t-test indicated no significant difference between sample 1 and 2 for volume. Mean concentration differed significantly between the samples (P = .044), but the estimated difference was only 1 × 106/mL (64.3 vs 65.3 × 106/mL). Mean morphology differed between samples (P = .018), but again this difference was not confirmed by a test of World Health Organization morphology (P = .131). Mean percent motility differed significantly between the first and second samples (P = .0007), although this represents a difference of only 1% between samples (51% vs 52% motile). We examined the difference in motility for each center individually and found it to be significant only in Iowa (percent motile sample 1 = 49, percent motile sample 2 = 54; P = .0015). Results of modeling the differences between semen parameters on the first and second samples by linear regression are given in Table 3. The difference in abstinence time had a significant effect on the differences in both concentration (P < .0001) and volume (P < .0001), which agrees with previously reported data (Jorgensen et al, 2001; Swan et al, 2003; Carlsen et al, 2005). The relationship between the differences in time to start the analysis and motility (P = .017) is also reported in the literature (Jorgensen et al, 2001; Swan et al, 2003). Consistent with the univariate analysis, the degree to which motility differed between samples varied by center, being most marked in Iowa (P = .0002), where the motility measurements for the second samples were on average larger than those for the first sample. Most importantly, after controlling for abstinence time, time to start the analysis, and center, the intercept was not statistically significant for any of the differences in semen parameter outcomes. This indicates that after adjustment, semen parameters are not significantly different between the first and second samples (P values are given in Table 3). Table 3. . Estimated parameters (and P values) for linear regression models of differences in semen quality* Difference Between First and Second Sample Log Concentration Volume Motility Morphology (Strict) Intercept −0.25 (.203) 0.015 (.867) −1.105 (.129) −0.064 (.765) Difference in abstinence time 0.002 (<.0001) 0.009 (<.0001) −0.013 (.185) −0.001 (.628) Difference in time to start analysis −0.001 (.241) 0.004 (.228) −0.074 (.017) −0.013 (.151) Center location Minnesota −0.008 (.763) 0.043 (.720) 0.110 (.913) −0.294 (.325) Missouri 0.014 (.314) 0.093 (.450) 1.131 (.275) −0.250 (.414) Iowa 0.022 (.505) 0.012 (.936) −4.620 (.0002) −0.412 (.257) *Rows indicate independent variables, and columns indicate model outcomes Discussion In clinical settings, it is standard practice to obtain multiple semen samples in the investigation for an infertile couple, but in population studies of semen quality the desirability of requesting multiple samples is less clear. Obtaining more than 1 semen sample will, on average, reduce measurement error in estimating population semen parameters, but requiring more than 1 sample from study subjects may reduce participation. In SFF about 12% of the 697 men chose to give only 1 of the 2 requested samples. This provided us with the opportunity to address 2 issues that arise when the number of samples per man varies. First, are there systematic ways in which men who give 2 samples differ from men who give only 1? Second, among men who give 2 samples, are there systematic differences in semen quality between these samples? This is important because the choice of the appropriate statistical model for the analysis of these semen samples depends on the answers to these 2 questions. In SFF the percent of men who gave 2 samples varied by center and reflected a between-center difference in the degree to which the need for 2 samples was stressed at the time of recruitment. The greatest difference was between California and Iowa, with 4% and 34% giving only 1 sample, respectively. Aside from center, no subject characteristic differed between men who gave 1 versus 2 samples, although men with a history of STD were somewhat less likely to give 2 samples (P = .08). No other characteristics predicted whether a man would give 1 or 2 samples, either in unadjusted or covariate-adjusted models. Moreover, semen quality did not differ significantly between men who gave 1 or 2 samples, except strict morphology without covariate adjustment. Thus, it is appropriate to combine data from men who gave 1 and 2 samples in an analysis of semen quality. One of the strengths of this study is that these data are drawn from a large population-based study with extensive quality control (Brazil et al, 2004b). A limitation of this study, however, is that all subjects are (presumably) fertile; their partners were pregnant at the time of recruitment. Therefore, the extent to which these results can be generalized to a population of men of unknown fertility is unclear. If this population had included infertile men, the range in semen quality would have been greater, and perhaps a stronger relationship between subject characteristics and semen quality would have been observed, but the impact of this on our 2 study questions is unclear. In principle, men with unknown fertility status might be less likely to return once they are given results of their sperm analysis. But this is unlikely to be relevant here because our study participants were not given their results. Our finding that semen parameters do not vary appreciably between the first and second sample, among men who give 2 samples, is likely to also be seen in a sample from the general population; there is no a priori reason to believe the second sample would systematically differ from the first among infertile men when it did not among fertile men Our results should be reassuring for investigators who are conducting population studies in which men give a variable number of samples. While none of the men in this population gave more than 2 samples, it seems likely that similar resolutions would hold for several samples given in a short time. However, this may change, for example, for sperm donors giving large numbers of samples over many years. The “dividing line” between these situations needs to be determined by studies of such donors. Our findings should also reassure researchers that data from multiple semen samples can be used in a single model as long as the model accounts appropriately for repeated measures and adjusts for all relevant covariates. References Auger J., Jouannet P.. Evidence for regional differences of semen quality among fertile French men. Federation Francaise des Centres d'Etude et de Conservation des Oeufs et du Sperme humains. Hum Reprod. 1997; 12: 740– 745. Berman NG, Wang C., Paulsen CA. Methodological issues in the analysis of human sperm concentration data. J Androl. 1996; 17: 68– 73. Brazil C., Swan SH, Drobnis EZ, Liu F., Wang C., Redmon JB, Overstreet JW, Study for Future Families Research Group. Standardized methods for semen evaluation in a multicenter research study. J Androl. 2004a; 25: 635– 644. Brazil C., Swan SH, Tollner CR, Treece C., Drobnis EZ, Wang C., Redmon JB, Overstreet JW, Study for Future Families Research Group Quality control of laboratory methods for semen evaluation in a multicenter research study. J Androl. 2004b; 25: 645– 656. Carlsen E., Swan SH, Petersen JH, Skakkebaek NE. Longitudinal changes in semen parameters in young Danish men from the Copenhagen area. Hum Reprod. 2005; 20: 942– 949. Guzick DS, Overstreet JW, Factor-Litvak P., Brazil CK, Nakajima ST, Coutifaris C., Carson SA, Cisneros P., Steinkampf MP, Hill JA, Xu D., Vogel DL, National Cooperative Reproductive Medicine Network, Sperm morphology, Motility, and concentration in fertile and infertile men. N Engl J Med. 2001; 345: 1388– 1393. Hosmer DW, Lemeshow S.. Applied Logistic Regression. 2nd ed. Hoboken, NJ: John Wiley & Sons Inc. 2000. Jorgensen N., Andersen AG, Eustache F., Irvine DS, Suominen J., Petersen JH, Andersen AN, Auger J., Cawood EH, Horte A., Jensen TK, Jouannet P., Keiding N., Vierula M., Toppari J., Skakkebaek NE. Regional differences in semen quality in Europe. Hum Reprod. 2001; 16: 1012– 1019. Jorgensen N., Carlsen E., Nermoen I., Punab M., Suominen J., Andersen AG, Andersson AM, Haugen TB, Horte A., Jensen TK, Magnus O., Petersen JH, Vierula M., Toppari J., Skakkebaek NE. East-West gradient in semen quality in the Nordic-Baltic area: a study of men from the general population in Denmark, Norway, Estonia and Finland. Hum Reprod. 2002; 17: 2199– 2208. Keel BA. Within- and between-subject variation in semen parameters in infertile men and normal semen donors. Fertil Steril. 2006; 85: 128– 134. Overstreet JW, Brazil CK. Semen Analysis. In: L. Lipshultz, S. Howard, eds. Infertility in the Male. St Louis, Mo: Mosby; 1997: 487– 490. Sharlip ID, Jarow JP, Belker AM, Lipshultz LI, Sigman M., Thomas AJ, Schlegel PN, Howards SS, Nehra A., Damewood MD, Overstreet JW, Sadovsky R.. Best practice policies for male infertility. Fertil Steril. 2002; 77: 873– 882. Swan SH, Brazil C., Drobnis EZ, Liu F., Kruse RL, Hatch M., Redmon JB, Wang C., Overstreet JW, Study For Future Families Research Group Geographic differences in semen quality of fertile. S. males. Environ Health Perspect. 2003; 111: 414– 420. Weisberg S.. Applied Linear Regression. 3rd ed. Hoboken, NJ: John Wiley & Sons Inc. 2005. Organization World Health. WHO Laboratory Manual for the Examination of Human Semen and Semen-Cervical Mucus Interactions. 2nd ed. Cambridge, United Kingdom: Cambridge University Press, 1987. Organization World Health. WHO Laboratory Manual for the Examination of Human Semen and Semen-Cervical Mucus Interactions. 4th ed. Cambridge, United Kingdom: Cambridge University Press, 1999. Footnotes Supported by Science to Achieve Results (STAR) grant RD832515 from the US Environmental Protection Agency, National Institutes of Health grant M01-RR00400 to the University of Minnesota General Clinical Research Center and M010RR0425 to Harbor-UCLA Medical Center, and grant 2T32 ES007271 from the National Institute of Environmental Health Sciences, National Institutes of Health. Citing Literature Volume28, Issue5September‐October 2007Pages 638-643 ReferencesRelatedInformation