Fetal exposure to prescription drugs and adult sexual orientation 

LEE ELLIS & JILL HELLBERG / Personality and Individual Differences v.38, i.1, 1jan2005

Division of Social Science, Minot State University, 500 University Avenue, Minot, ND 58707, USA

Received 7 July 2003; Revised 1 March 2004; accepted 10 April 2004. Available online 9 June 2004.

Abstract

This study was undertaken to determine if prenatal exposure to therapeutic drugs contributes to variations in sexual orientation. Especially suspect were drugs that could affect the delicate balance of sex hormone levels that appear to guide the sexual differentiation of the fetal brain. The recollections of 5102 mothers concerning their use of therapeutic drugs during pregnancy were linked to reports of the sexual orientation of their offspring (as provided by either the offspring themselves or by their mothers). About 14% of the mothers recalled having taken at least one of 19 prescription drugs (or classes of drugs) during their pregnancy. Regarding male offspring, little evidence was found that prenatal exposure to any of these medications was associated with variations in sexual orientation. However, even after controlling for age, education, and self-rated recall ability of the mothers, exposure to two types of drugs was significantly related to sexual orientation among female offspring. One type consisted of amphetamine-based diet pills and the other was comprised of synthetic thyroid medications. A month-by-month analysis revealed that during the first trimester consumption of all prescription drugs was unusually high for mothers of female homosexual offspring. Prescription medications that affect the mother’s and/or the female fetuses’ developing immune system may alter the feminization/demasculinization of the brain in ways that cause variations in the offspring’s adult sexual orientation.

Author Keywords: Sexual orientation; Prenatal influences; Prescription drugs; Thyroid medications; Diet medications; DES; Progestins

1. Introduction

According to the neuroandrogenic theory of sexual orientation, the gender toward whom one is sexually attracted is largely determined by perinatal factors rather than being a preference that individuals learn in a sociocultural context (Ellis & Ames, 1987). More specifically, the neuroandrogenic theory asserts that the level of sex hormones to which fetuses are exposed helps to sexually differentiate the brain as well as the genitals. The brain in turn is assumed to play a central role in masculinizing/feminizing behavior, inclining behavior associated with whether one prefers same or opposite sex members as sex partners.

Sexing of the genitals has been shown to occur primarily between the first and the fourth months of gestation, while sexing of the brain seems to takes place from the third through the seventh months (Ellis, 1996). Due to this timing difference, sex hormones (especially testosterone) may sometimes masculinize/defeminize the genitals without having the same sexing effects on the brain. According to the neuroandrogenic theory, males whose brains failed to be fully masculinized/defeminized will exhibit varying degrees of feminine preferences and behavioral characteristics throughout life. These feminine/demasculine preferences may sometimes include males preferring same sex partners upon sexual maturation (instead of preferring opposite sex partners). Theoretically, the reverse set of circumstances could incline females to be attracted to members of their own gender following the onset of puberty.

Ellis and Ames (1987) proposed that several factors contribute to varying degrees of genital-brain sexing inconsistencies. Among the hypothesized factors were maternal consumption of drugs that influence sex hormone levels during critical periods of fetal development. If this proposal is correct, it should be possible to find differences in consumption of drugs by mothers of homosexuals relative to mothers of heterosexuals.

The present study was undertaken to identify prescription drugs consumed during pregnancy by greater proportions of mothers of homosexuals (and bisexuals) than by mothers of heterosexuals. Drugs that would be especially suspected as having the capability of altering offspring sexual orientation would be those that can affect sex hormone levels.

2. Methods

Five thousand one hundred and two (5102) women provided data regarding their use of various prescription drugs during pregnancy. Most (4839) of these women were recruited by first securing a questionnaire from their offspring, who were college students attending one of 22 universities (20 US and 2 Canadian) between 1988 and 1998 (see Ellis & Cole-Harding, 2001). Among the numerous items of information provided by the offspring were those pertaining to their sexual orientation.

In addition to the 4839 mothers recruited through the college student sample, 264 women were obtained through Parents and Friends of Lesbians and Gays (P-FLAG), an international support group for parents who have homosexual and bisexual offspring. Through special permission of the board of directors for P-FLAG, copies of a version of our mother’s questionnaire were sent to 250 United States and Canadian chapters along with requests that the questionnaires be made available to any women willing to participate in our study. No matching questionnaires were sought from the offspring of the P-FLAG mothers. Instead, the sexual orientation of these offspring was determined by asking the mothers themselves.

Regarding the mothers of the offspring derived from our college student sample, the average age of the mothers was 47.51 (SD=7.81). For the P-FLAG mothers, it was 59.29 (SD=7.43). Despite these average age differences, the two groups of mothers were combined in order to make the sample of mothers with homosexual and bisexual offspring as large as possible. This was done because only a few dozen offspring of each sex in the college sample declared themselves to be homosexual or bisexual.

The average years of education for the mothers who were obtained through the college students was 13.62 (SD=2.45) while that for the P-FLAG mothers was 15.52 (SD=2.62). Thirty-one percent of mothers of the college students were divorced compared to 36% of the P-FLAG mothers.

The questionnaires used for the two groups of mothers were identical except for the fact that the P-FLAG mothers’ questionnaire asked the mothers themselves to categorize their offsprings’ sexual orientation. This question was asked on the offsprings’ questionnaire in the case of the college students. On both the P-FLAG mothers’ and the college offsprings’ questionnaires, four categories were provided for identifying the offspring’s sexual orientation: homosexual, bisexual, heterosexual, or uncertain. To verify the reported sexual orientation, we confirmed that all offspring of the P-FLAG mothers were classified as either homosexual or bisexual. Regarding the data provided on the offsprings’ questionnaires, some apparent discrepancies were found between the responses on the sexual orientation question and other questions closely related to sexual orientation, such as whether members of the same or opposite sex are most often featured in ones sexual fantasies. As explained in a forthcoming article, these discrepant responses were excluded from our analysis (Ellis, Robb, & Burke, in press).

To measure maternal consumption of prescription drugs (both oral and injected) during pregnancy, the mothers’ questionnaire allowed women to report having taken DES, progestins, and up to two additional prescription drugs during each month of pregnancy. If mothers reported having taken any prescription drugs other than DES or progestins, they were asked to provide the name of the drug, or to carefully describe it and the reason it was taken. Throughout this paper, prescription drug refers to any drug or other substance that was administered or prescribed by a physician.

Mothers reported the number of dosages of the medication they consumed or were administered during each month of pregnancy. From this information, we determined that 697 women (13.7%) took at least one prescription drug sometime during their pregnancy. This relatively small percentage allowed us to perform only two types of analyses: One involved determining whether or not any specific drugs (or types of drugs) were consumed. The other analysis had to do with the month that one or more drugs were taken during each month of pregnancy.

To estimate the statistical significance of our initial group comparisons, we used the 2 exact test (SPSS-11.5). In order to control for the effects of extraneous factors, logistic regression was employed.

3. Results

The results are presented in three parts. The first part has to do with identifying which of 19 drugs (or drug categories) were found to differ between the mothers of both male and female offspring regarding three categories of offsprings’ sexual orientation (heterosexual, homosexual, and homosexual/bisexual combined). In the second part of the results, logistic regression is used to compare the drugs taken by the mothers of homosexual and heterosexual offspring of both genders after controlling for mother’s age, years of education, and self-rated ability to recall events associated with her pregnancy. Third, we considered the timing of prescription drug consumption for each month of pregnancy.

3.1. Maternal prescription drug consumption and offspring sexual orientation

Table 1 summarizes the results from comparing 19 prescription drugs consumed by mothers relative to the sexual orientation of their male and female offspring. The categories used in the analysis were heterosexual, homosexual/bisexual combined, and homosexual only. Our coverage of the details in Table 1 will focus on comparing mothers of homosexuals and mothers of heterosexuals, since in most cases where significant differences were found the greatest differences were between these two groups. Table 1 shows that when bisexuals were combined with homosexuals, differences between their mothers’ use of prescription drugs and use by mothers of heterosexuals were usually diminished at least slightly in statistical significance.

Table 1. Reported prenatal therapeutic drug consumption by mothers of heterosexuals, homosexuals and bisexuals combined, and homosexuals alone

Drugs consumed		 Mothers of males––%(N)                     	Mothers of females––%(N)                    .
during one or		 Heterosexuals,	 Homo/	    	Homosexuals	Heterosexuals,	 Homo/		Homosexuals
more months		 N = 1406	 bisexuals,  	only, N = 215	N = 3241	 bisexuals,	only, N = 114
of pregnancy                             N = 249                                          N = 196                     .
Antibiotics 		 1.7%(24) 	 1.2%(3)     	1.4%(3) 	2.1%(67) 	1.5%(3) 	1.7%(2)
Anti-convulsants 	 0.1%(1) 	 0%(0) 		0.0%(0) 	0.2%(6) 	0%(0) 		0.0%(0)
Anti-diarrheal		 0.0%(1) 	 0%(0) 		0%(0) 		0.1%(3) 	0.5%(1) 	0.9%(1)
   medication
Anti-nausea and		 7.0%(98) 	 4.4%(11)*	4.2%(9)*	5.9%(192) 	4.5%(9) 	4.3%(5)
   vomiting medication 			 p = 0:037 	p = 0:038
Anti-toxemia medication  0.1%(1) 	 0%(0) 		0%(0) 		0.2%(5) 	0%(0) 		0%(0)
Allergy/cold medication  0.9%(12) 	 1.6%(4) 	1.9%(4) 	0.7%(22) 	1.5%(3) 	0.9%(1)
Birth control pills	 0.1%(2) 	 0%(0) 		0.0%(0) 	0.0%(1) 	0.5%(1) 	0.7%(1)
Diethylstilbestrol	 0.6%(8) 	 0.9%(3) 	1.5%(3) 	0.6%(19) 	2.0%(4)*	3.5%(4)**
   (DES)										p = 0:030	p = 0:006
Diet pills 		0.6%(8) 	 1.6%(4)  	1.9%(4)		0.2%(7) 	1.5%(3)*	1.7%(2)*
					 		p = 0:049			p = 0:015	p = 0:033
Diuretics 		0.5%(7)		 0.8%(2) 	0.9%(2) 	1.4%(45) 	1.0%(2) 	1.7%(2)
Gamma globulin		0.0%(0)		 0.8%(2)*	0.9%(2)*	0.0%(1) 	0.0%(0) 	0.0%(0)
					 p = 0:023	p = 0:018
Heart medication 	0.1%(1) 	 0%(0) 		0%(0) 		0.1%(3) 	0%(0) 		0%(0)
Insulin medication 	0%(0) 	 	 0%(0) 		0%(0) 		0.1%(2) 	0.9%(1) 	0.7%(1)
Migraine medication 	0.1%(2) 	 0%(0) 		0%(0) 		0.1%(2) 	0%(0) 		0%(0)
Narcotics 		0.3%(4) 	 0.0%(0) 	0.0%(0) 	0.5%(15) 	0.5%(1) 	0.0%(0)
Prednisone 		0.1 (1) 	 0.5 (1) 	0.2 (1) 	0.0%(0) 	0%(0) 		0.0%(0)
Progesterone 		0.7%(10) 	 0.4%(1) 	0.5%(1) 	0.8%(25) 	1.5%(3) 	2.8%(3)
   (progestin)
Sedatives 		1.1%(16) 	 2.4%(6) 	2.3%(5) 	1.0%(32) 	0.5%(1) 	0.9%(1)
Synthetic thyroid	0.7%(10) 	 0.9%(2) 	0.9%(2) 	1.0%(31) 	3.1%(6)*	5.2%(6)***
   medication										p = 0:0137	p = 0:001
Overall use of 		13.6%(190) 	 14.1%(35) 	14.9%(32) 	13.6%(442) 	15.39%(30) 	19.1%(22)
   medications

* Significant difference relative to heterosexuals of the same sex (p < 0:05).
** Significant difference relative to heterosexuals of the same sex (p < 0:01).
*** Significant difference relative to heterosexuals of the same sex (p < 0:005).

Only five of the 19 drugs (or drug categories) prescribed to mothers during pregnancy were significantly related to their offsprings’ sexual orientation, three of which were for male offspring and three for female offspring (with one drug related to both). It should be emphasized that because the sample sizes for most categories of drugs were often exceedingly small (especially in the case of mothers whose offspring were homosexual or bisexual), caution must be exercised in offering interpretations. Because our tests of statistical significance were based on 2, caution is especially warranted for cells containing less than five subjects.

Regarding male offspring, the consumption of three substances were significantly related to sexual orientation. One class of such drugs was anti-nausea and vomiting medications, substances reportedly consumed by 7.0% of the mothers of male heterosexuals, but only 4.2% of the mothers of homosexuals (p=0.038). This very surprising finding suggests that such medications could have a "protective effect" with respect to male offsprings’ typical sexual orientation.

Another substance was gamma globulin, a naturally produced substances extracted from the blood of other people which contain antibodies to various viruses. Prior to the eradication of measles and related infectious diseases, gamma globulin was fairly often injected into mothers who were exposed to measles during their pregnancy so as to prevent viral-induced birth defects. Our analysis revealed that none of the 1406 mothers of heterosexual males were given gamma globulin during pregnancy, while two of the 215 mothers of homosexual males (0.8%) recalled having been administered this substance. Despite a fairly high level of statistical significance (p=0.018), the small number of mothers in our entire sample who recalled having been administered gamma globulin during pregnancy cautions against making causal inferences at this point.

The third substance that was statistically related to male sexual orientation consisted of amphetamine-based diet pills (primarily Dexedrine and Tenuate Dospan). Whereas 0.6% of mothers of male heterosexuals consumed diet pills during pregnancy, nearly two percent (1.9%) of the mothers of male homosexuals did so (p=0.049).

Turning to the female offspring, three drugs were administered significantly more often to mothers of homosexuals than to mothers of heterosexuals. One––amphetamine-based diet pills––was just discussed with reference to male offspring. Whereas seven out of 3241 (0.2%) of the mothers of female heterosexuals recalled having taken diet medication during pregnancy, two out of 144 (1.7%) of the mothers of homosexuals did so (p=0.033).

DES was another drug taken more often by mothers of female homosexuals than by mothers of female heterosexuals. This drug was prescribed to millions of pregnant women between the 1940s through the early 1970s, especially for those with histories of miscarriages to help them maintain their pregnancies (Noller & Fish, 1974; Peress, Tsai, Mathur, & Williamson, 1982). However, research in the 1970s and 1980s began to link consumption of DES during pregnancy with uterine cancer in daughters ( Miller, Degenhardt, & Sassoon, 1998), testicular cancer in sons ( Depue, Pike, & Henderson, 1983) and with breast cancer in mothers ( Colton et al., 1993). By the end of the 1970s, prescribing DES to pregnant women had been halted.

In our sample, 0.7% of the mothers of both sexes recalled having been prescribed DES. While no significant links to male sexual orientation were found, more than three times as many mothers of homosexual females took DES during pregnancy as did mothers of heterosexual females (2.8% versus 0.6%). The probability that this difference occurred by chance was quite small (p=0.006). Nevertheless, only four of the 114 mothers of female homosexuals actually took DES.

The third type of prescription medication found statistically associated with female homosexuality consisted of synthetic thyroid medications, primarily Synthroid and Thyroxine. Whereas 31 of the 3241 (1.0%) mothers of heterosexuals took these medications during pregnancy, six of the 144 (5.2%) mothers of homosexuals did so. The probability of this difference having occurred by chance was exceedingly small (p=0.001).

3.2. Controlling for maternal age, education, and recall

Because the present study was non-experimental, one can question whether mothers of heterosexuals differed from mothers of homosexuals and bisexuals in ways that might account for the differences in prescription drug consumption revealed in Table 1. To address this question, we used logistic regression to control for three maternal variables: maternal age, maternal education, and self-rated maternal recall. These three variables were chosen in part because, as noted in the methods section, mothers of the homosexual and bisexual offspring were about eight years older than the mothers of the heterosexual offspring and were more educated by about two years. It is reasonable to believe that the different decades in which most of the mothers of homosexuals would have been pregnant relative to that of the mothers of heterosexuals as well as their somewhat greater years of education could have accounted for some of the differences documented in Table 1.

We were somewhat surprised to find no significant differences in the average amount of self-rated recall of pregnancy by these groups of mothers (t=1.011, p=0.310). Nevertheless, we still considered it prudent to include this as a potentially important control variable.

The results of the logistic regression are shown in Table 2. Because the results shown in Table 1 revealed that in all cases where statistical significance was found, mothers of bisexuals appeared to be roughly intermediate to mothers of homosexuals and heterosexuals, we confined our analysis in Table 2 to a comparison of mothers of homosexuals and mothers of heterosexuals.

Table 2. Logistic regression analyses of prenatal therapeutic drug consumption by mothers of heterosexuals and homosexuals after controlling for maternal age, maternal education and maternal recall (how well her memory served her in answering the retrospective questionnaire)

Drugs consumed		
during one or		
more months of		Mothers of males                                Mothers of females                   .
pregnancy               B       SE      Wald    P       Exp(B)          B       SE      Wald    P       Exp(B)
--------------------------------------------------------------------------------------------------------------
Antibiotics 		0.0266 	0.6703 	0.0016 	0.9683 	1.0270 		0.3607 	0.8653 	0.1738 	0.6768 	1.4343
Anti-convulsants 	2.4444 	22.2404 0.0121 	0.9125 	11.5241 	3.7433 	14.2041 0.0695 	0.7921 	42.2365
Anti-diarrheal		3.5285 	22.2404 0.0252 	0.8739 	34.0737 	-1.8471 1.1854 	2.4279 	0.1192 	0.1577
medication
Anti-nausea/		0.0304 	0.3997 	0.0058 	0.9394 	1.0308 		-0.1542	0.4749 	0.1054 	0.7455 	0.8571
   vomiting medication
Anti-toxemia 		4.3472 	22.2398 0.0382 	0.8450 	77.2633 	4.5959 	15.1016 0.0926 	7609 	99.0780
   medication
Allergy/cold 		0.2154 	0.6990 	0.0950 	0.7580 	1.2404 		-0.1422 1.0601 	0.0180 	0.8933 	0.8674
   medication
Birth control pills	2.2072 	15.6431 0.0199 	0.8878 	9.0898 		-2.1673 1.4260 	2.3098 	0.1286 	0.1145
Diethylstilbestrol	0.1736 	0.8142 	0.0454 	0.8312 	1.1895 		-1.1271 0.7075 	2.5377 	0.1112 	0.3240
   (DES)
Diet pills 		-1.0157 0.7029 	2.0880 	0.1485 	0.3622 		-2.5299 0.8715 	8.4270 	0.0037*** 0.0797
Diuretics 		-1.0464 0.8490 	1.5189 	0.2178 	0.3512 		-0.4701 0.7470 	0.3960 	0.5291 	73.6268
Gamma globulin 		-7.6606 22.2400 0.1186 	0.7305 	0.0005 		4.2990 	36.6575 0.0138 	0.9066 	73.6268
Heart medication 	3.7474 	22.2399 0.0284 	0.8662 	42.4105 	3.2833 	20.6390 0.0253 	0.8736 	26.6628
Insulin medication 	- 	– 	– 	– 	– 		-2.6420 1.5310 	2.9780 	0.0844 	0.0712
Migraine medication	4.9215 	13.3338 0.1362 	0.7121 	137.212 	2.7528 	25.8820 0.0113 	0.9153 	15.6865
Narcotics 		5.4596 	9.5761 	0.3250 	0.5686 	234.997 	3.1823 	9.1674 	0.1205 	0.7285 	24.1012
Prednisone 		-3.0123 1.4231 	4.4807 	0.0343 	0.0492		– 	– 	– 	– 	–
Progesterone 		0.8105 	1.1061 	0.5370 	0.4637 	2.2491 		-1.0881 0.6386 	2.9033 	0.0884 	0.3369
   (progestin)
Sedatives 		-0.8393 0.5852 	2.0568 	0.1515 	0.4320 		1.2442 	1.2946 	0.9237 	0.3365 	3.4702
Synthetic thyroid	0.0071 	0.8453 	0.0001 	0.9933 	1.0071 		-1.3699 0.4841 	8.0084 	0.0047** 0.2541
   medication
Overall use of		-0.2533 0.2409 	1.1053 	0.2931 	0.7763 		-0.5654 0.2584 	4.7865 	0.0287** 0.5681
   medications

Note that the negative Betas indicate that non-use of the drug was more prevalent among mothers of homosexuals.
  * Significant difference relative to heterosexuals of the same sex (p<0:05).
 ** Significant difference relative to heterosexuals of the same sex (p<0:01).
*** Significant difference relative to heterosexuals of the same sex (p<0:005).

Table 2 shows that one drug––Prednisone––was significantly associated with male sexual orientation after imposing the three statistical controls. Prednisone is a adrenocortical steroid that is taken primarily to manage disorders of the immune system. As can be seen by referring to Table 1, this drug was only taken by the mothers of two males in our entire sample: one male was heterosexual and the other was homosexual. While the finding based on logistic regression can be considered interesting inasmuch as Prednisone is similar to the naturally produced stress hormone, cortisol, and prenatal stress has been implicated in the etiology of male homosexuality ( Ellis & Cole-Harding, 2001), the sample size is insufficient for the results to be taken seriously at the present time.

Regarding females, the consumption of two classes of drugs were statistically significant after controlling for maternal age, maternal education, and maternal recall. One class consisted of prescription diet pills (p=0.0037) and the other class was synthetic thyroid medications (p=0.0047). Recall that maternal consumption of these two drugs was also very significantly related to the sexual orientation of female offspring before imposing any statistical controls (Table 1). Nevertheless, logistic regression caused maternal consumption of DES to drop from being statistically significant.

3.3. The timing of prescription drug use

Attention is now directed toward the month of pregnancy when prescription drugs might have their greatest impact on offspring sexual orientation. Unfortunately, the number of mothers who took any one of the 19 medications during a specific month of pregnancy were too few for meaningful analysis. Therefore, we compared the proportion of mothers who consumed any of the 19 types of prescription drugs in terms of whether their offspring were male or female and homosexual or heterosexual. For clarity, drug consumption for offspring of bisexuals were not included in this analysis, but in no case were their results statistically different from the results for mothers of the other groups.

The results are shown in Table 3, with the means presented graphically in Fig. 1. Both the table and graph reveal that maternal consumption of prescription drugs was virtually indistinguishable for mothers of male homosexuals, male heterosexuals, and for mothers of female heterosexuals. Regarding mothers of female homosexuals, however, unusually high rates of prescription drug consumption during the second month of pregnancy (p=0.045) were evident, and there was a trend in the same direction for drug consumption during the third month (p=0.075).

******************

Table 3. Means and standard errors of the mean for average drug dosages consumed during pregnancy by the four groups of mothers

Months of	Mothers of male		Mothers of female	Mothers of male		Mothers of female
pregnancy	heterosexuals		heterosexuals		homosexuals		homosexuals
		Mean 	SEM 		Mean 	SEM 		Mean 	SEM 		Mean 	SEM
-----------------------------------------------------------------------------------------------------
First 		0.055 	0.006 		0.071 	0.004 		0.056 	0.018 		0.087 	0.024
Second 		0.062 	0.007 		0.070 	0.004 		0.074 	0.018 		0.113 	0.025
Third 		0.058 	0.007 		0.074 	0.004 		0.070 	0.018 		0.096 	0.025
Fourth 		0.056 	0.006 		0.064 	0.004 		0.061 	0.017 		0.078 	0.023
Fifth 		0.045 	0.006 		0.055 	0.004 		0.061 	0.016 		0.700 	0.021
Sixth 		0.042 	0.005 		0.050 	0.004 		0.051 	0.015 		0.043 	0.021
Seventh 	0.040 	0.005 		0.049 	0.003 		0.056 	0.015 		0.052 	0.020
Eighth 		0.042 	0.005 		0.051 	0.004 		0.046 	0.015 		0.052 	0.020
Ninth 		0.039 	0.005 		0.047 	0.003 		0.050 	0.014 		0.035 	0.020

Fig. 1. The average dosage of prescription drugs consumed during each month of pregnancy by mothers of four groups of offspring.

These analyses support the conclusion that female offspring are more vulnerable to alterations in sexual orientation via exposure to a variety of prescription drugs, and suggest that this vulnerability is greatest during the first trimester.

In an earlier analysis of other aspects of the present data set, evidence was found to suggest that exposure to a non-prescription drug––nicotine (via maternal smoking)––was also associated with an elevated probability of homosexual preferences in female offspring but not in male offspring (Ellis & Cole-Harding, 2001). As with the present findings, this link was statistically significant only during the first trimester of pregnancy.

4. Discussion

Despite its exploratory nature, this study’s findings could be important. The most noteworthy findings were that if prenatal therapeutic substances affect offspring sexual orientation, the effects are mainly confined to female offspring and are limited to the first trimester of pregnancy. In particular, mothers of the female homosexuals in our sample were significantly more likely than mothers of female heterosexuals to have taken diet pills and thyroid medications during pregnancy. To our knowledge, this is the first study to have implicated either of these substances as affecting any behavioral aspects of sexual development. Experimental studies of laboratory animals could shed further light on such a possibility.

Earlier studies of female lesbians indicate that their mothers were more likely than mothers of females in general to have taken DES during pregnancy (Ehrhardt et al., 1985; Meyer-Bahlburg & Ehrhardt, 1986; for a failure to replicate see Lish et al., 1991). Our findings are mixed on this point. While we did find a significant difference consistent with a DES effect, these differences dropped well below statistical significance once controls were introduced for the mother’s age, years of education, and self-perceived ability to recall her pregnancy.

For both practical and ethical reasons, the design of this study was non-experimental and retrospective. While a prospective design is feasible and could provide great insight into the effects of prenatal factors on offspring sexual behavior, such a design would require monitoring the use of drugs by thousands of mothers throughout pregnancy and then locating their children in young adulthood to determine their sexual orientation.

Due to the retrospective nature of this study, one can conjecture that many of the mothers of homosexuals may have consciously or unconsciously biased their responses concerning drug use to match some preconceptions they may have had about prenatal drugs causing homosexuality in their children. However, we seriously doubt that this occurred for four reasons.

First, prior to our study, there was little objective reason to suspect that prenatal drug exposure would affect sexual orientation. The main exception was in the case of DES, which, as just noted, our logistic regression did not statistically associate with offspring sexual orientation.

Second, the questions that were asked about prescription drug consumption comprised only a small portion (about one-half of a page) of a 10-page questionnaire. Thus, there was little in our questionnaire that would have lead mothers to believe that prescription drug exposure was anything more than an incidental focus of our investigation (which was in fact the case).

Third, the vast majority of drugs exhibited no evidence of having been consumed or injected at greater or lesser amounts among mothers of homosexuals compared to mothers of heterosexuals. Also, one type of drug––medications taken to prevent nausea and vomiting during pregnancy––was reported as having been used by a greater proportion of mothers of male heterosexuals than by mothers of male homosexuals. These are not the sort of patterns one would expect to see if mothers of homosexuals were intentionally inflating or otherwise misreporting their consumption of prescription drugs.

Fourth, more of the evidence for prenatal prescription drug effects was found for female offspring than for male offspring. It would be difficult to believe that the mothers with sons who were homosexual would not have been at least as motivated to recall having used prescription drugs during pregnancy as were the mothers of homosexual daughters.

Determining if prenatal exposure to certain drugs affects the sexual orientation of humans, especially in the case of females, is a worthwhile goal. This is partly because nearly all of the scientific evidence regarding biological contributions to sexual orientation thus far has been limited to males (Bailey & Pillard, 1991; Bailey et al., 1999; Ellis, 1996; LeVay & Hamer, 1994).

From a theoretical standpoint, our findings provide qualified support for the neurohormonal theory of sexual orientation (Ellis & Ames, 1987). On the positive side, the theory asserts that immunological factors can alter sexual orientation. In this regard, both amphetamine-based drugs and thyroid medications are known to affect the mother’s immune system during pregnancy ( Glinoer, 1998; Kubera et al., 2002). Also, the placenta is very immunologically active, thereby preventing harmful immune reactions by the mother toward the fetus ( Mellor and Munn, 2000 and Mellor and Munn, 2001), and its functioning may be altered by either of these medications. Another possibility is that certain immune conditions that diet pills and thyroid medications are intended to treat contribute to variations in sexual orientation. If so, the drugs per se are not directly responsible for lesbianism among female offspring.

On the negative side, the neurohormonal theory asserts that sex hormones, especially testosterone, are central to sexually differentiating the brain. Thus, drugs that are known to alter sex hormone levels––most notably progestin and DES––have been especially suspect as affecting offspring sexual orientation (Reinisch & Karow, 1977), but little of our evidence supported this line of reasoning. Overall, it may be hypothesized that drugs affecting the immune system that are being consumed by the mother during pregnancy could alter brain development of the fetus in ways that affect preferences for sex partners later in life. Nevertheless, because there are neither animal experiments nor prior human studies to suggest that either amphetamines or thyroid drugs are capable of making such alterations, more evidence is needed before considering this a well-supported hypothesis.

Acknowledgements

We thank Myrna Nelson and Craig Kinsley for making helpful comments and the Eugene Garfield Foundation and the Minot State University Small Grants Committee for funding major portions of this research project.

References

Bailey, J.M. and Pillard, R.C., 1991. A genetic study of male sexual orientation. Archives of General Psychiatry 48, pp. 1089–1096.

Bailey, J.M., Pillard, R.C., Dawood, K., Miller, M.B., Farrer, L.A., Trivedi, S. and Murphy, R.L., 1999. A family history study of male sexual orientation using three independent samples. Behavior Genetics 29, pp. 79–86.

Colton, T., Greenberg, E.R., Noller, K., Resseguie, L., Van Bennekom, C., Heeren, T. and Zhang, Y., 1993. Breast cancer in mothers prescribed diethylstilbestrol in pregnancy: Further follow-up. Journal of the American Medical Association 269, pp. 2096–2100.

Depue, R.H., Pike, M.C. and Henderson, B.E., 1983. Estrogen exposure during gestation and risk of testicular cancer. Journal of the National Cancer Institute 71, p. 1151.

Ehrhardt, A.A., Meyer-Bahlburg, H.F.L., Rosen, L.R., Feldmann, J.F., Veridiano, N.P., Zimmerman, I. and McEwen, B.S., 1985. Sexual orientation after prenatal exposure to exogenous estrogen. Archives of Sexual Behavior 14, pp. 57–75.

Ellis, L., 1996. The role of perinatal factors in determining sexual orientation. In: Savin-Williams, R.C. and Cohen, K.M., Editors, 1996. The lives of lesbians, gays, and bisexuals, Harcourt Brace College, Fort Worth, TX, pp. 35–70.

Ellis, L. and Ames, M.A., 1987. Neurohormonal functioning and sexual orientation: A theory of homosexuality–heterosexuality. Psychological Bulletin 101, pp. 233–258.

Ellis, L. and Cole-Harding, S., 2001. The effects of prenatal stress, and of prenatal alcohol and nicotine exposure, on human sexual orientation. Physiology and Behavior 74, pp. 213–226. 

Ellis, L., Robb, B., & Burke, D. (in press). Sexual orientation in a large sample of United States and Canadian college students: Toward refined measurement. Archives of Sexual Behavior

Glinoer, D., 1998. Thyroid hyperfunction during pregnancy. Thyroid 8, pp. 859–864.

Kubera, M., Filip, M., Basta-Kaim, A., Nowak, E., Budziszewska, B., Tetich, M., Holan, V., Korzeniak, B. and Przegalinski, E., 2002. The effect of amphetamine sensitization on mouse immunoreactivity. Journal of Physiology and Pharmacology 53, pp. 233–242.

LeVay, S. and Hamer, D.H., 1994. Evidence for a biological influence in male homosexuality. Scientific American 270, pp. 44–49.

Lish, J.D., Ehrhardt, A.A., Meyer-Bahlburg, H.F.L., Rosen, L.R., Gruen, R.S. and Veridiano, N.P., 1991. Gender-related behavior development in females exposed to diethylstilbestrol (DES) in utero: An attempted replication. Journal of American Academy of Child and Adolescence Psychiatry 30, pp. 29–37.

Mellor, A.L. and Munn, D.H., 2000. Immunology at the maternal–fetal interface: Lessons for T cell tolerance and suppression. Annual Review of Immunology 18, pp. 367–391.

Mellor, A.L. and Munn, D.H., 2001. Extinguishing maternal immune responses during pregnancy: Implications for immunosuppression. Seminar in Immunology 13, pp. 213–218. 

Meyer-Bahlburg, H. and Ehrhardt, A.A., 1986. Prenatal diethlstilbestrol exposure: Behavioral consequences in humans. In: Dorner, G., McCann, S. and Martini, L., Editors, 1986. Systematic hormones, neurotransmitters and brain development, Basel, Karger, pp. 90–95.

Miller, C., Degenhardt, K. and Sassoon, D.A., 1998. Fetal exposure to DES results in de-regulation of Wnt7a during uterine morphogenesis. Nature Genetics 20, pp. 228–230.

Noller, K.L. and Fish, C.R., 1974. Diethylstilbestrol usage: Its interesting past, important present, and questionable future. Medical Clinics of North America 58, pp. 793–810.

Peress, M.R., Tsai, C.C., Mathur, R.S. and Williamson, H.O., 1982. Hirsutism and menstrual patterns in women exposed to diethylstilbestrol in utero. American Journal of Obstetrics and Gynecology 144, pp. 135–140.

Reinisch, J. and Karow, W., 1977. Prenatal exposure to synthetic progestins and estrogens: Effect on human development. Archives of Sexual Behavior 6, pp. 257–288.

Corresponding author. Tel.: +1-701-858-3241; fax: +1-701-839-6933
Pages 225-236
doi:10.1016/j.paid.2004.04.004