1 INSERM (National Institute For Health and Medical Research) U569 IFR69, 94276 Le Kremlin-Bicêtre, 2 Service de Biologie de la Reproduction-CECOS, Hôpital Cochin, EA 1752 Université Paris 5, 75014 Paris, 3 CECOS Midi-Pyrénées, Research Group Human Fertility (E1 3694), 4 Centre de Stérilité Masculine, Service d'Urologie Andrologie, Hôpital Paule de Viguier, 31059 Toulouse, 5 Service de Biologie de la Reproduction-CECOS, Hôpital de l'Hôtel-Dieu, 35000 Rennes and 6 Département de Médecine et Biologie de la Reproduction, Faculté de Médecine-CECOS, 69373 Lyon, France
7 To whom correspondence should be addressed at: INSERM U569, 82 rue du Général Leclerc, 94276 Le Kremlin-Bicêtre cedex, France. Email: muller{at}vjf.inserm.fr
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key words: bias (epidemiology)/fecundity/participation/selection bias/semen studies
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
To date, few studies have dealt with selection bias in semen studies. Five studies (Handelsman et al., 1985; Handelsman, 1997
; Larsen et al., 1998
; Cohn et al., 2002
; Lalos et al., 2003
) have shown that men who volunteer for semen studies tend to be younger, to have experienced a long period of infecundity (measured as a time to pregnancy, TTP, of >6 months; Larsen et al., 1998
) and to have family members who have had problems conceiving a child.
The work presented here is based on data from a French cross-sectional multicentre study (Reproduction de l'Homme, REPRHOM), the principal aim of which was to compare the fecundity (semen quality and TTP) of the partners of pregnant women in four French towns. The selection phenomena associated with voluntary participation were analysed at three levels: (i) refusal to participate in the study but completion of a short refusal questionnaire; (ii) completion of study questionnaires only; and (iii) completion of study questionnaires and collection of semen sample.
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Design
We used the same standardized recruitment protocol in each centre to minimize between-centre differences. There were no incentives at all to participate. An interviewer described the study to the pregnant women, checked that the couple was eligible and then asked them to participate (Figure 1). If the couple agreed to participate, three questionnaires (for the man, the woman and the mother of the man) were given to the woman. About 1 week later, the interviewer contacted the couple to ask the man to attend an appointment at the local CECOS for the collection of a semen sample. If the man refused, the interviewer tried to persuade the couple to complete and to return the questionnaires. If the questionnaires were still not returned after several telephone calls, the interviewer attempted to get the man to complete a refusal questionnaire by telephone. This questionnaire asked about the man's age, level of education and smoking habits. In all cases in which the man agreed to participate, he provided written consent to the medical team. The study protocol was approved by the French regulatory and consultative ethics bodies.
|
Semen collection. Each male volunteer who agreed to give a semen sample came to the local CECOS during his partner's pregnancy or, at the latest, 1 month after delivery. The semen sample was obtained by masturbation after a recommended period of 210 days of sexual abstinence.
Definition of studied variables
Information was obtained for three groups of men: (i) men who refused to participate but who completed the refusal questionnaire; (ii) those who completed the study questionnaires only; and (iii) those who gave a semen sample and completed the study questionnaires.
The following information was recorded in all questionnaires (refusal and complete questionnaires): age of the man, his educational level, his smoking status and the study centre. Of men who refused to participate, 44% agreed to complete the refusal questionnaire. More detailed information was available on the completed questionnaires. Women were the first to be approached by the interviewers. As the study did not directly concern them, the vast majority of them (96%) agreed to ask their partner to participate. Therefore, we considered in this work that the choice to participate was the man's decision.
Number of previous pregnancies included all pregnancies for which the man was the father (other than that for which the man had been included in the REPRHOM study) with either their current partner or a previous one. Unfavourable pregnancy outcomes included all spontaneous miscarriages, ectopic pregnancies, abortions for medical reasons and stillbirths resulting from every pregnancy for which the man was responsible. Men who had had cryptorchidism, inguinal hernia, varicocele, testicular torsion, testicular cancer, hypospadias, gonococcal infection, epididymitis or orchitis were considered to have a history of andrological disease. Consultation for infecundity included declarations made by the men concerning their entire lifetime. TTP was estimated for the current pregnancy on the basis of declarations made by the pregnant women. It was not possible to estimate TTP for 40 couples, and 51 couples were using contraception at the time of conception. The statistical analyses of fecundity thus included 894 couples for whom TTP was known. Smoking habits were recorded for the 3 month period preceding conception. Exposure to tobacco in utero was recorded on the questionnaire filled in by the volunteers mothers and concerned the period of intra-uterine development of the man.
Statistical analysis
We first compared the men who completed the refusal questionnaire with the group of participants (questionnaires only and questionnaires + semen sample) on the basis of all the information available in the refusal questionnaire. The adjusted odds ratios (ORs) for participation were calculated by logistic regression. Variables included in the model were man's age, man's education level, man's smoking status and study centre.
We then compared the detailed information from the complete questionnaires for the two groups of participants (semen collection + questionnaires versus questionnaires only). The adjusted ORs associated with the characteristics of the men agreeing to provide a semen sample with respect to those completing questionnaires only were estimated by logistic regression. The variables taken into account in the logistic regression analysis were the man's age, man's education level, study centre, history of andrological disease, unfavourable pregnancy outcomes, consultation for infecundity, number of previous pregnancies, smoking status of the man, exposure to tobacco in utero and conception problems in the man's family.
To complete this analysis, the TTP of the current pregnancy was studied using the discrete Cox model with a logistic link (Scheike and Jensen, 1997) on the total number of months of attempts, according to outcome of the attempt (failure/success). The fecundability ratio (FR) obtained is the ratio between the monthly probability of conceiving for a couple who participate in semen collection compared with that of a couple who only complete questionnaires. TTP was censored after 13 months, as couples not managing to conceive may change their behaviour or lifestyle in a way that might affect fertility after this time. The variables taken into account in the discrete Cox model were the same as in the previous logistic regression model plus frequency of sexual intercourse and history of gynaecological disease, variables known to be confounding factors. All ORs and FRs were estimated with 95% confidence intervals (CIs) using STATA SE 8.2® software (Stata Corporation, College Station, TX).
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
Comparison of participants who gave semen sample with participants who only completed questionnaires
The men who provided semen had a higher educational level (test for trend: P<0.001) than those who only completed questionnaires (Table IIIA). They were also more likely to have been exposed to tobacco in utero, even with a large confidence interval (OR 1.64, 95% CI 0.962.78). Their current and previous partners were more likely to have experienced unfavourable pregnancy outcomes (OR 1.68, 95% CI 1.142.49) than those of the men completing questionnaires only. The couples who agreed to provide a semen sample had a similar fecundability to those who only completed questionnaires (Table IIIB).
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Most papers dealing with the bias associated with voluntary participation in semen studies were based on highly diverse populations: men from cohorts created for other studies (Cohn et al., 2002), sperm donors not proven to be fecund (Handelsman et al., 1985
; Lalos et al., 2003
), workers exposed to pollutants (Larsen et al., 1998
) and volunteers for trials of contraception methods for men (Handelsman, 1997
). It is therefore difficult to compare our results with those of other studies.
Were there differences between participants and those who refused to participate?
It is noteworthy that information was available for only 44% of the men who refused to participate. There may be differences between the men for whom no information was available and the other men who refused, but this cannot be checked.
Men with a low educational level were more likely to refuse to participate than highly educated men (Table II: OR secondary school/further education 0.52, 95% CI 0.360.74). To our knowledge, no other study on human sperm has considered this issue. Our questionnaires were quite long, taking 45 min to complete. It is therefore possible that the complexity of the questionnaires selected the most educated subjects. However, a similar difference in educational level was also observed between subjects who agreed to give a semen sample and those who only completed the questionnaires (Table IIIA: OR secondary school/further education 0.41, 95% CI 0.240.69). In addition, more educated individuals may be more aware of the arguments in favour of active participation in research (Jouannet et al., 2001
). Moreover, there was no incentive to participate, but a recent study by Eustache et al. has found that monetary compensation did not influence participation rate to semen study (Eustache et al., 2004
).
Male age is known to affect fecundity (De La Rochebrochard and Thonneau, 2003; Eskenazi et al., 2003
; Hassan and Killick, 2003
), but the proportion of men for which such an effect was possible (men over the age of 40) was very low in our study (5%). In contrast to Larsen et al. (1998)
who found that the participation rate decreased with increasing age, participation in REPRHOM did not differ in terms of male age. Smoking has a negative effect on sperm quality (Kunzle et al., 2003
). In our study, those refusing to participate were more likely to be smokers than were participants. It is therefore possible that the sperm of those refusing to participate is of lower quality than that of those who agreed to participate. Educational level and men's smoking status should be taken into account as potential confounders in future analyses.
Were the participants who only completed questionnaires as fecund as those who provided a semen sample?
Men who donated sperm were no more likely to declare a family history of conception problems than were those who completed questionnaires only. Our initial hypothesis was that men who agreed to provide a semen sample were more likely to have a personal or a familial history of fecundity problems (Jouannet et al., 2001). This was shown in two studies of sperm donors in infertility clinics who donate out of a spirit of altruism (Handelsman et al., 1985
), often because they have been sensitized to the issue as a result of sterility among their relatives (Lalos et al., 2003
). However, the samples studied were very different: partners of pregnant women recruited for a fertility study on the one hand, and semen donors at infertility clinics on the other hand. The reasons for participating in a scientific research programme are undoubtedly different from the reasons for donating semen. As for those who refused to participate and participants, men with a low educational level were more likely to participate in questionnaires only than highly educated men who agreed more often to give a semen sample (Table IIIA: OR secondary school/further education 0.41, 95% CI 0.250.68).
The FR for the current pregnancy did not differ according to the level of participation (Table IIIB). In an occupational Danish study (Larsen et al., 1998), men who had experienced a long period of infecundity were more likely than other men to agree to semen collection. There may be two reasons for these different results. First, participation tends to be higher in occupational studies because individuals are aware that they have impaired health, and may wish to explain their state of health in terms of professional exposure (Bonde et al., 1996
). Secondly, the REPRHOM sample consisted of fecund couples (pregnant women and their partners), and the known effects of various factors on TTP have been shown to disappear or to be reversed if infecund couples are excluded. This is the case, for example, for maternal age (Juul et al., 2000
).
The participants who provided a semen sample and those who completed the questionnaires only did not differ in terms of andrological history or male smoking status (Curtis et al., 1997; Kunzle et al., 2003
). Female partners of men who donated semen were more likely to have experienced unfavourable pregnancy outcomes than were the partners of those who completed questionnaires only (OR 1.68, 95% CI 1.142.49). This suggests that these men were more likely to have experienced infecundity in the past than the others. Furthermore, exposure to tobacco in utero was slightly (but not significantly when adjusted) more frequent among men agreeing to provide semen than among the men who only completed the questionnaires (adjusted OR 1.64, 95% CI 0.962.78). Maternal smoking probably has a deleterious effect on sperm quality in the son (Storgaard et al., 2003
; Jensen et al., 2004
). It is therefore possible that sperm quality was lower in those who agreed to give a semen sample than in those who only completed the questionnaires: this cannot be tested as no semen sample was available for the latter.
In view of differences of unfavourable pregnancy outcomes and maternal smoking, sperm quality may be lower for participants providing a semen specimen than for men completing questionnaires only. Low semen quality is known to affect TTP (Bonde et al., 1998; Slama et al., 2002
), but our analysis of TTP for this sample did not confirm the hypothesis that fecundity is different between participants who completed questionnaires only and men who provided a semen sample. It is possible that the semen quality variation studied was not sufficient to have a detectable effect on TTP. It may not be possible to extrapolate the distribution of sperm characteristics for the participants providing semen samples to the entire population of participants or to the whole of the population from which this sample originates. This element should be studied and considered in subsequent analyses of sperm quality. Given the similarity of fecundity in the two groups of participants, the analysis of TTP for all study participants should not be affected by selection bias in the REPRHOM sample.
In conclusion, our results show that men who volunteer to participate, even in part, in semen studies differ from those who refuse to participate (from which they originated), particularly in terms of educational level and smoking status. Moreover, men who agree to complete questionnaires only and men who agree to semen collection differ in terms of certain socio-demographic characteristics (educational level) and variables associated with fecundity (unfavourable pregnancy outcomes, exposure to tobacco in utero), despite having a similar FR (similar TTP). The results of semen analysis for this population sample cannot be extrapolated to the whole population from which the volunteers originate. As seen in our study, there may be certain differences in fertility between the refusal group and participants (in the REPRHOM sample, the refusal group could be less fecund than participants, but individuals who provided a semen sample could be less fecund than those who completed questionnaires only). Therefore, as selection bias could involve bilateral deviation in fertility characteristics, we recommend that the effect of this selection bias be systematically analysed for each sample in every new study. Above all, more information is required on participating men and men who refused to participate, and participation rates should be reported in semen studies to make it possible to interpret the results correctly.
![]() |
Acknowledgements |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
Notes |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Andersen AG, Jensen TK, Carlsen E, Jorgensen N, Andersson AM, Krarup T, Keiding N and Skakkebaek NE (2000) High frequency of sub-optimal semen quality in an unselected population of young men. Hum Reprod 15, 366372.
Andolz P, Bielsa MA and Vila J (1999) Evolution of semen quality in North-eastern Spain: a study in 22,759 infertile men over a 36 year period. Hum Reprod 14, 731735.
Auger J, Kunstmann JM, Czyglik F and Jouannet P (1995) Decline in semen quality among fertile men in Paris during the past 20 years. N Engl J Med 332, 281285.
Bonde JP, Giwercman A and Ernst E (1996) Identifying environmental risk to male reproductive function by occupational sperm studies: logistics and design options. Occup Environ Med 53, 511519.[Abstract]
Bonde JP, Ernst E, Jensen TK, Hjollund NH, Kolstad H, Henriksen TB, Scheike T, Giwercman A, Olsen J and Skakkebaek NE (1998) Relation between semen quality and fertility: a population-based study of 430 first-pregnancy planners. Lancet 352, 11721177.[CrossRef][ISI][Medline]
Bujan L, Mansat A, Pontonnier F and Mieusset R (1996) Time series analysis of sperm concentration in fertile men in Toulouse, France between 1977 and 1992. Br Med J 312, 471472.
Carlsen E, Giwercman A, Keiding N and Skakkebaek NE (1992) Evidence for decreasing quality of semen during the past 50 years. Br Med J 305, 609613.[ISI][Medline]
Cohn BA, Overstreet JW, Fogel RJ, Brazil CK, Baird DD and Cirillo PM (2002) Epidemiologic studies of human semen quality: considerations for study design. Am J Epidemiol 155, 664671.
Curtis KM, Savitz DA and Arbuckle TE (1997) Effects of cigarette smoking, caffeine consumption, and alcohol intake on fecundability. Am J Epidemiol 146, 3241.[Abstract]
De La Rochebrochard E and Thonneau P (2003) Paternal age 40 years: an important risk factor for infertility. Am J Obstet Gynecol 189, 901905.[CrossRef][ISI][Medline]
Eskenazi B, Wyrobek AJ, Sloter E, Kidd SA, Moore L, Young S and Moore D (2003) The association of age and semen quality in healthy men. Hum Reprod 18, 447454.
Eustache F, Auger J, Cabrol D and Jouannet P (2004) Are volunteers delivering semen samples in fertility studies a biased population? Hum Rep, in press.
Fisch H, Goluboff ET, Olson JH, Feldshuh J, Broder SJ and Barad DH (1996) Semen analyses in 1,283 men from the United States over a 25-year period: no decline in quality. Fertil Steril 65, 10091014.[ISI][Medline]
Handelsman DJ (1997) Sperm output of healthy men in Australia: magnitude of bias due to self-selected volunteers. Hum Reprod 12, 27012705.[Abstract]
Handelsman DJ, Dunn SM, Conway AJ, Boylan LM and Jansen RP (1985) Psychological and attitudinal profiles in donors for artificial insemination. Fertil Steril 43, 95101.[ISI][Medline]
Hassan MA and Killick SR (2003) Effect of male age on fertility: evidence for the decline in male fertility with increasing age. Fertil Steril 79 (Suppl 3), 15201527.[CrossRef][ISI][Medline]
Jensen TK, Jorgensen N, Punab M, Haugen TB, Suominen J, Zilaitiene B, Horte A, Andersen AG, Carlsen E, Magnus O et al. (2004) Association of in utero exposure to maternal smoking with reduced semen quality and testis size in adulthood: a cross-sectional study of 1,770 young men from the general population in five European countries. Am J Epidemiol 159, 4958.
Jorgensen N, Andersen AG, Eustache F, Irvine DS, Suominen J, Petersen JH, Andersen AN, Auger J, Cawood EH, Horte A et al. (2001) Regional differences in semen quality in Europe. Hum Reprod 16, 10121019.
Jorgensen N, Carlsen E, Nermoen I, Punab M, Suominen J, Andersen AG, Andersson AM, Haugen TB, Horte A, Jensen TK et al. (2002) EastWest gradient in semen quality in the NordicBaltic area: a study of men from the general population in Denmark, Norway, Estonia and Finland. Hum Reprod 17, 21992208.
Jouannet P, Wang C, Eustache F, Kold-Jensen T and Auger J (2001) Semen quality and male reproductive health: the controversy about human sperm concentration decline. Apmis 109, 333344.[CrossRef][ISI][Medline]
Juul S, Keiding N and Tvede M (2000) Retrospectively sampled time-to-pregnancy data may make age-decreasing fecundity look increasing. European Infertility and Subfecundity Study Group. Epidemiology 11, 717719.[CrossRef][ISI][Medline]
Kunzle R, Mueller MD, Hanggi W, Birkhauser MH, Drescher H and Bersinger NA (2003) Semen quality of male smokers and nonsmokers in infertile couples. Fertil Steril 79, 287291.[CrossRef][ISI][Medline]
Lalos A, Daniels K, Gottlieb C and Lalos O (2003) Recruitment and motivation of semen providers in Sweden. Hum Reprod 18, 212216.
Larsen SB, Abell A and Bonde JP (1998) Selection bias in occupational sperm studies. Am J Epidemiol 147, 681685.[Abstract]
Richthoff J, Rylander L, Hagmar L, Malm J and Giwercman A (2002) Higher sperm counts in Southern Sweden compared with Denmark. Hum Reprod 17, 24682473.
Sandler DP (2002) On revealing what we'd rather hide: the problem of describing study participation. Epidemiology 13, 117.[CrossRef][ISI][Medline]
Scheike TH and Jensen TK (1997) A discrete survival model with random effects: an application to time to pregnancy. Biometrics 53, 318329.[ISI][Medline]
Slama R, Eustache F, Ducot B, Jensen TK, Jorgensen N, Horte A, Irvine S, Suominen J, Andersen AG, Auger J et al. (2002) Time to pregnancy and semen parameters: a cross-sectional study among fertile couples from four European cities. Hum Reprod 17, 503515.
Storgaard L, Bonde JP, Ernst E, Spano M, Andersen CY, Frydenberg M and Olsen J (2003) Does smoking during pregnancy affect sons' sperm counts? Epidemiology 14, 278286.[CrossRef][ISI][Medline]
Swan SH, Elkin EP and Fenster L (2000) The question of declining sperm density revisited: an analysis of 101 studies published 19341996. Environ Health Perspect 108, 961966.[ISI][Medline]
Swan SH, Brazil C, Drobnis EZ, Liu F, Kruse RL, Hatch M, Redmon JB, Wang C and Overstreet JW (2003) Geographic differences in semen quality of fertile U.S. males. Environ Health Perspect 111, 414420.[ISI][Medline]
Submitted on June 28, 2004; accepted on August 19, 2004.