Selection biases in semen study?

References 1. Check JH, Davies E, Adelson H. A randomized prospective study comparing pregnancy rates following clomiphene citrate and human menopausal gonadotrophin therapy. Hum Reprod 1992;7:801–5. 2. Check JH, Nowroozi K, Choe J, Lurie D, Dietterich C. The effect of endometrial thickness and echo pattern on in vitro fertilization outcome in donor oocyte-embryo transfer cycle. Fertil Steril 1993;59:72–5. 3. Nakamura Y, Ono M, Yoshida Y, Sugino N, Ueda K, Kato H. Effects of clomiphene citrate on the endometrial thickness and echogenic pattern of the endometrium. Fertil Steril 1997;67:256 – 60. 4. Benda JA. Clomiphene’s effect on endometrium in infertility. Int J Gynecol Pathol 1992;11:273– 83. 5. Bonhoff AJ, Naether OGJ, Johannisson E. Effects of clomiphene citrate stimulation on endometrial structure in infertile women. Hum Reprod 1996;11:844 –9. PII S0015-0282(98)00111-3 Selection Biases in Semen Study? published between 1981 and 1990 that were summarized by Carlsen et al. (2). The claim that the men studied by Rasmussen et al. were ‘‘a random sample that is representative of the Danish male population,’’ is unsubstantiated and seems unjustified in light of the above. Although it is possible that there has been no decline in the sperm concentration of Danish males born between 1950 and 1970, this has yet to be established using a representative population and standard methods. Shanna H. Swan, Ph.D. Reproductive Epidemiology Section California Department of Health Services Berkeley, California January 27, 1998 To the Editor (Letter 1): In the December 1997 issue of Fertility and Sterility, Rasmussen et al. (1) present data intended to support their conclusion that, ‘‘sperm quality has not changed in the Danish male population during the last 20 –30 years.’’ The high estimate of sperm concentration reported by these authors would appear to support this conclusion. However, it is questionable whether these results can be generalized to the entire population of Danish males. References First, the mean sperm concentration reported by these authors (183.73106/mL) is inconsistent with the published literature. Among the 61 studies summarized by Carlsen et al. (2), none reported a sperm concentration higher than 1343106/mL. More recently, in a report of sperm densities from 10 cities around the world, the highest sperm concentration was 107.33106/mL (3). The reason for the exceptionally high estimate reported by Rasmussen et al. is apparent from the authors’ discussion of their methods. They state that sperm counts were obtained by utilizing a Horwell counter, which is not commonly used for this purpose. ‘‘To some extent, the high sperm concentration found in our study may be attributed to the use of the Horwell counting chamber. In an unpublished study, . . . the Horwell chamber overestimated the concentrations by .100%’’ (p. 1062). If the Rasmussen et al. estimate is reduced by 50% to adjust for this bias, the resulting estimate of 91.03106/mL is more consistent with published estimates. Rasmussen et al. (1) reported no change in semen quality among male partners of women who received IVF treatment at Odense University Hospital. They concluded that no change had occurred in semen quality of Danish men. Second, in addition to using an unusual chamber for counting sperm, the population utilized by these authors is highly selected. Subjects were partners of women receiving IVF during the 1990 –1996 time period; ‘‘The study comprises the data on the semen samples delivered and used in connection with the first IVF treatment’’ (p. 1060). It appears that males with low counts may be underrepresented in this population, reflecting referral bias. The distribution includes only 5.6% of men with concentrations #203106/mL. This is less than one-third the estimate for the proportion of samples below this concentration (18%) based on 33 studies 1158 Letters to the Editor 1. Rasmussen PE, Erb K, Westergaard LG, Laursen SR. No evidence for decreasing semen quality in four birth cohorts of 1,055 Danish men born between 1950 and 1970. Fertil Steril 1997;68:1059 – 64. 2. Carlsen E, Giwercman A, Keiding N, Skakkebaek N. Evidence for decreasing quality of semen during past 50 years. Br Med J 1992;305: 609 –13. 3. World Health Organization Task Force on Methods for the Regulation of Male Fertility. Contraceptive efficacy of testosterone-induced azoospermia and oligozoospermia in normal men. Fertil Steril 1996;65:821–9. To the Editor (Letter 2): Perhaps their conclusion should be taken with a grain of salt. The authors claim that their population (partners of women referred for IVF treatment) was a ‘‘random sample that is representative of the Danish male population.’’ However, their subjects do not in any way represent men from the general population. According to general Danish referral guidelines adopted by the university clinics in Århus and Copenhagen, one of the criteria for admission to IVF treatment was that the male partners should have good semen quality. Although the term good semen quality was interpreted with some flexibility, couples were generally not accepted for IVF treatment if the male partner had severe oligozoospermia (,5 million per mL) (J. Starup, A.N. Anderen. Personal communication). These criteria were used until 1994. If a change in semen quality had indeed occurred in the general population during the period in question and resulted in more men with low semen quality, such men would obviously be underrepresented in the subjects studied by Rasmussen et al. This fact probably helps explain the high sperm counts (mean, 183.7 million per mL) reported by the authors, which were strikingly higher than those of previous studies on healthy Danish men where Vol. 69, No. 6, June 1998 the mean sperm counts were between approximately 60 and 80 million per mL (3– 4). Niels E. Skakkebæk, M.D. Aleksander Giwercman, M.D. Tina Kold Jensen, M.D. Elisabeth Carlsen, M.D. Niels Jørgensen, M.D. Department of Growth and Reproduction Rigshospitalet University of Copenhagen Niels Keiding E-mail: [email protected] Department of Biostatistics University of Copenhagen February 2, 1998 References 1. Rasmussen PE, Erb K, Westergaard LG, Laursen SB. No evidence for decreasing semen quality in four birth cohorts of 1,055 Danish men born between 1950 and 1970. Fertil Steril 1997;68:1059 – 64. 2. Bonde JP. Semen quality and sex hormones among mild steel and stainless steel welders: a cross sectional study. Br J Ind Med 1990;47: 508 –14. 3. Bonde JP. Semen quality in welders before and after three weeks of non exposure. Br J Ind Med 1990;47:515– 8. 4. Jensen TK, Giwercman AJ, Carlsen E, Scheike T, Skakkebæk NE. Semen quality among members of organic food associations in Zealand, Denmark. Lancet 1996;347:1844. Reply of the Authors to Letters 1 and 2: The authors of the two letters above claim that our statement, ‘‘Sperm quality has not changed in the Danish male population during the last 20 –30 years’’ (1), is unsubstantiated and unjustified (or at least should be accompanied with a grain of salt) because of the use of the Horwell counting chamber and because of a highly selected study population. As a hemocytometer, the Horwell chamber possesses the same sources of error as most other types of counting chambers used for counting spermazoa. The admittedly high mean sperm concentration found in our study and the knowledge of possible discrepances in sperm counts using different counting chambers (2, 3) made us compare the Horwell chamber with three other types of counting chambers. Using latex beads with a known concentration of 3.53106/mL, we found that all four types of counting chambers overestimated the concentration to a variable extent. The CellVision chamber was most accurate as mean concentration was found to be 1.02 times greater than expected. The mean concentration was 1.26 times greater using Makler, 1.42 times greater using Improved Neubauer, and 2.37 times greater than expected using Horwell. This implies that the mean concentration of our study (183.73106/mL) determined with the Horwell chamber corresponds to a mean concentration of 79.13106/mL using CellVision, 97.73106/mL using Makler, and 110.13106/mL using the Improved Neubauer. Interestingly, despite the type of chamber used, all values are well above the mean concentration of 663106/mL reFERTILITY & STERILITYt ported by Carlsen et al. (4). However, these discrepancies do not affect the important message of our work—that no change over time in the sperm concentration was found. We allege that our study population is one of the most representative study populations for the average man among the many studies of semen quality performed during the last decades. The study population consists of 1,055 consecutive men who share two features: [1] they are married to a woman referred for IVF treatment due to tubal infertility and [2] they are required to deliver a semen sample. All couples were offered IVF notwithstanding the semen quality. We cannot exclude some selection bias (who can?) as men with known poor semen quality might have chosen not to seek IVF treatment due to a belief of a low success rate or lack of desire to use donor semen or donor backup semen. However, the distribution of semen qualities in our material, including 1.8% of semen from men with azoospermia, does not seem to show any substantial difference to other studies. If men with oligozoospermia were underrepresented in the first part of the study period due to restricted criteria for treatment (which could have been the truth in Copenhagen and Århus but not in Odense), a significant fall in sperm concentration over time should have been the result. This would have been erroneously interpreted as more evidence for a global fall in semen quality. This was not the case. We conclude that neither the use of the Horwell chamber nor our selection of study population can in any way affect our finding of an unchanged sperm concentration in four birth cohorts of Danish men. However, the questions raised once again focus on the many confounding errors affecting measurement of semen quality. When sperm concentration, even in a large study like ours, can vary as much as 50% between different half-year periods in the same group of men, combined with all the possible known errors of laboratory handling, it is more than questionable what information is gained by epidemiologic studies of sperm concentration and it certainly raises questions about the value of meta-analyses in this field. Per Emil Rasmussen, M.D. Karin Erb, M.Sc. Lars G. Westergaard, M.D., Ph.D. Sten B. Laursen, M.Sc. Fertility Clinic Department of Obstetrics and Gynecology Odense University Hospital Denmark February 25, 1998 References 1. Rasmussen PE, Erb K, Westergaard LG, Laursen SR. No evidence for decreasing semen quality in four birth cohorts of 1,055 Danish men born between 1950 and 1970. Fertil Steril 1997;68:1059 – 64. 2. Imade GE, Towabola OA, Sagay AS, Otubu JAM. Discrepancies in sperm count using improved Neubauer, Makler and Horwells counting chambers. Arch Androl 1993;31:17–22. 3. Seaman EK, Goluboff E, BarChama N, Fisch H. Accuracy of semen counting chambers as determined by the use of latex beads. Fertil Steril 1996;66:662–5. 1159 4. Carlsen E, Giwercman A, Keiding N, Skakkebæk N. Evidence for decreasing quality of semen during past 50 years. Br Med J 1992;305: 609 –13. To the Editor (Letter 3): We would like to correct certain statements that were made about our publications in the interesting article by Rasmussen et al. (1). On page 1062 the investigators state that we reported sperm concentrations in our article (2) between 0 and 1803106/mL in a normal subject over the course of a year. These data were extracted from a graph published by Dr. Paulsen in the World Health Organization (WHO) manual (3). The subject never demonstrated azoospermia; the minimum value was 13106/mL. The data on this subject was not used in the specific studies included in our Fertility and Sterility article. After the article in the WHO manual was written, we rigorously assessed the methodology of analyzing sperm concentration in a 1996 subsequent study (4). We demonstrated that the within-subject variability of sperm concentration can be reduced by using log sperm concentration in analysis, as well as using the geometric mean of several samples. The authors’ citation of our paper on page 1063 (2) was also incorrectly referenced. They stated correctly that we saw a slight, statistically significant rise in sperm concentration from 1972 to 1993; however, they did not include our statement that this increase was not considered to be biologically important. Nancy G. Berman, Ph.D. Department of Pediatrics Harbor–UCLA Medical Center Torrance, California Christina Wang, M.D. Department of Medicine Harbor–UCLA Medical Center Torrance, California C. Alvin Paulsen, M.D. Department of Medicine University of Washington Seattle, Washington February 11, 1998 Reply of the Authors to Letter 3: Berman, Wang, and Paulsen indicate correctly that the reported sperm concentration in a normal subject over the course of 2 years is extracted from a graph published in the World Health Organization manual (1). Regrettably, this does not appear in our article (2). Due to a typographic error, references 15 and 16 were misplaced. The correct minimum concentration was as mentioned, 13106/mL. In the article by Paulsen, Berman, and Wang (3), a slight, statistically significant rise in sperm concentration from 1977 to 1993 was reported in men from the greater Seattle area. We fully agree that this increase is not to be considered biologically important. We question, for the same reason, the biological and clinical relevance of the observed changes of sperm concentrations over time shown in other studies. However, we do not question the need for more reliable and reproducible parameters for the continuing measurement of possible environmental influences on male gonadal function. Per Emil Rasmussen, M.D. Karin Erb, M.Sc. Lars G. Westergaard, M.D., Ph.D. Sten B. Laursen, M.Sc. Fertility Clinic Department of Obstetrics and Gynecology Odense University Hospital Denmark February 25, 1998 References 1. World Health Organization. Laboratory manual for the examination of human semen and semen– cervical mucus interaction. 3rd ed. New York: Cambridge University Press, 1992. 2. Rasmussen PE, Erb K, Westergaard LG, Laursen SB. No evidence for decreasing sperm quality in four birth cohorts of 1,055 Danish men born between 1950 and 1970. Fertil Steril 1997;68:1059 – 64. 3. Paulsen CA, Berman NG, Wang C. Data from men in greater Seattle area reveals no downward trend in semen quality: further evidence that deterioration of semen quality is not geographically uniform. Fertil Steril 1996;65:1015–20. PII S0015-0282(98)00112-5 Is IVF/ICSI Always the First Step to Treat Couples With Infertility Due to Retrograde Ejaculation? To the Editor: References 1. Rasmussen PE, Erb K, Westergaard LG, Laursen SB. No evidence for decreasing sperm quality in four birth cohorts of 1,055 Danish men born between 1950 and 1970. Fertil Steril 1997;68:1059 – 64. 2. Paulsen CA, Berman NG, Wang C. Data from men in greater Seattle area reveals no downward trend in semen quality: further evidence that deterioration of semen quality is not geographically uniform. Fertil Steril 1996;65:1015–20. 3. World Health Organization. Laboratory manual for the examination of human semen and semen– cervical mucus interaction. 3rd ed. New York: Cambridge University Press, 1992. 4. Berman NG, Wang C, Paulsen CA. Methodological issues in the analysis of sperm concentration data. J Androl 1996;17:68 –73. 1160 Letters to the Editor We were interested to read the recent article of Jimenez et al. (1) concerning the usefulness of freezing sperm recovered from postejaculatory urine of a patient with retrograde ejaculation. The count of the sperm recovered was good and the percentage of live spermatozoa was high. Motility was very poor, however. The investigators confirmed that contact with the urine may have had a detrimental effect on the viability and motility of the sperm. Nevertheless, they decided not to assess the sperm characteristics before exposure to the urine Vol. 69, No. 6, June 1998