Araujo, Andre B. BA; Durante, Richard PhD; Feldman, Henry A. PhD; Goldstein, Irwin MD; McKinlay, John B. PhD
CES-D = Center for Epidemiological Studies-Depression scale; MED = male erectile dysfunction; DHEA-S = dehydroepiandrosterone sulfate; HDL = high-density lipoprotein cholesterol; MMAS = Massachusetts Male Aging Study; NHANES II = second National Health and Nutrition Examination Survey.
The role of affective factors in the etiology of MED has been extensively debated over the past few decades. Using data from the MMAS, a large, multidisciplinary study designed to examine a wide range of health outcomes in a community-based sample of normally aging men, we focus on the relationship between depressive symptoms and male erectile dysfunction.
By using MMAS data, we are able to address several of the methodological shortcomings of past studies that have reported an association between depression and erectile dysfunction [1-15]. First, because male erectile dysfunction is now considered a multifactorial condition [16-19], there is a need for multidisciplinary study of that condition. By using the broad spectrum of data in the MMAS, we are able to control for demographic, anthropometric, hormonal, health, and lifestyle factors, all of which could potentially mediate any association between depressive symptoms and erectile dysfunction. Second, with the large randomly selected sample of men who participated in the MMAS, we are able to avoid sampling issues (eg, difficulty in generalizing findings due to convenience sampling; the inability to conduct subgroup analyses because of small sample sizes) that have compromised results from past studies. Third, in contrast to much of the previous research, highly reliable and valid measures of depressive symptoms and erectile dysfunction were used in the MMAS.
With this as background, our primary objectives with this study are: a) to determine whether male erectile dysfunction is associated with depressive symptoms, and b) to determine whether this association is independent of aging and para-aging factors (ie, other factors that are highly correlated with age), such as weight, comorbidity, life events, hormone levels, and prescription medications.
The baseline phase of the MMAS, a random-sample survey of health and aging in men aged 40 to 70 years, was conducted from 1986 to 1989 in 11 cities and towns in the Boston Metropolitan area [20-25]. Communities were selected randomly by use of probabilities proportional to the population within each of six strata defined by community size and median income. Sampling fractions were adjusted to produce a uniform age distribution over the three decades. Introductory letters were sent to 5287 men, followed by a telephone call encouraging participation. No financial incentive was offered. Of 4104 successful telephone contacts with subjects or household members, 1526 subjects and 23 household members declined to cooperate, whereas 756 subjects were deceased, inaccessible, ineligible, or unable to complete the interview. A total of 1709 respondents enrolled in the study and completed the in-home protocol. This corresponds to a response rate of 53%, a rate common for epidemiological field studies involving phlebotomy and a restricted window for interview (morning blood sampling).
The distributions of body-mass index, blood pressure, and serum cholesterol in the MMAS closely matched those of NHANES II. Small discrepancies were within the limits of secular trend in the 10 years separating MMAS from NHANES II. The MMAS participants were typically Caucasian (95%), employed (78%), and married (75%). Nearly half were Catholic (48%). Most had completed high school (71%), and many had earned at least a bachelor's degree (42%). There was low representation of racial minorities (4%). All of the above characteristics are consistent with the composition of the Massachusetts population.
Comparison of the MMAS sample with the Massachusetts census listing, which constituted the sampling frame, indicated that respondents were more likely to be employed than ineligibles, refusals, and no-contacts. To determine whether the sample was biased as to the state of health, an attempt was made to conduct brief interviews with 400 randomly selected nonrespondents in towns where the response rate was under 60%. Of these 400, 206 were successfully interviewed, 119 refused, 57 could not be contacted, and 19 were ineligible or deceased. The nonrespondents were slightly less likely than respondents to rate themselves in good-to-excellent health (52% vs. 56%) but also less likely to report heart disease (10.2% vs. 12.7%) or cancer (1.9% vs. 6.1%). No difference was detected in the self-reported prevalence of diabetes, high blood pressure, arthritis, or restricted activity attributed to health.
A trained field technician/phlebotomist visited each subject in his home within 2 hours of his awakening to control for diurnal variation in hormone levels . Two separate samples, drawn 30 minutes apart to control for episodic secretion, were frozen for storage, and later pooled for analysis. Serum total cholesterol, HDL cholesterol, and various measures of hormone, metabolite, and binding-protein concentration were assayed at the University of Massachusetts Endocrine Laboratory. Height, weight, and waist and hip circumference were measured by standardized methods developed for large-scale field work . Age, education level, and change in marital and employment status in the past 12 months were noted. Additional questions assessed the subject's satisfaction with his life, and access to emotional support. Written informed consent was given by all study participants.
Health status was ascertained through self-report  by asking, "Have you ever been told by a health professional that you have any of the following medical conditions?" Nine conditions were listed, including heart disease, diabetes, and high blood pressure. For each affirmative response, the interviewer determined whether the subject was currently receiving treatment for that condition. Prescription and nonprescription medications were inventoried by the interviewer and later coded according to the American Hospital Formulary Service Pharmacologic-Therapeutic Classification .
The presence of depressive symptoms was indicated by a score of 16 or higher on the CES-D scale . This instrument measures current level of depressive symptomatology in community populations and does not indicate a diagnosis of clinical depression. However, it does discriminate between clinically depressed patients and others and correlates highly with more extensive scales designed to measure depression [31,32]. Furthermore, the criterion-related validity of the CES-D has been shown to be very satisfactory in a sample of older adults  and it has been shown to be a good indicator of depressive symptomatology across a range of demographic categories .
A self-administered questionnaire on sexual activity was given to the subject for completion in private, which was then given to the interviewer. The nine items directly related to potency (see Appendix) were linked to a direct assessment of erectile dysfunction by means of a separate calibration study through quadratic discriminant analysis (see  for a detailed description on how this variable was constructed). Briefly, this study consisted of 398 men presenting at the Boston University Medical Center urology clinic during a 6-month period in 1990 who were asked to complete a self-administered questionnaire consisting of the nine items relating to potency and one additional item, asking them to rate themselves as 1, not impotent; 2, minimally impotent; 3, moderately impotent; or 4, completely impotent. Of these 398 questionnaires, 303 (76%) were returned with complete responses. An algorithm was derived from the results of this study and using this, each MMAS subject was assigned to one of four categories: no erectile dysfunction, minimal erectile dysfunction, moderate erectile dysfunction, or complete erectile dysfunction. Men in the latter two categories were defined as having erectile dysfunction for the present analyses.
Level of physical activity was ascertained by recall of the past 7 days' activities, including frequency and duration. Total energy expenditure for the week was estimated by rating moderate, vigorous, and heavy physical activity at 4, 6, and 10 kcal/kg/hr respectively, similar to the scale used in recent publications .
Current cigarette smoking was noted. The subject's customary alcohol intake was estimated by self-report of beer, wine, and liquor consumption, accounting for frequency, quantity, and binge drinking using the Khavari formula . "Light" drinking was attributed to those consuming, on average, 0.5 to 1.5 oz ethanol (1-3 drinks) per day.
All analyses were conducted on a fixed sample of 1265 men (74% of the 1709 men) who had no missing data for the two primary variables, erectile dysfunction and depressive symptoms. Of the 444 omitted subjects, 419 had missing data on erectile dysfunction, the majority because they had no current sexual partner (N = 291), or did not answer the questions pertaining to partner satisfaction (N = 43). Thirty-nine men had missing data on depressive symptoms, and 14 had missing data on both variables. Having missing data on depressive symptoms was not associated with erectile dysfunction or any of the confounders, suggesting that the loss of those 39 cases did not bias our analysis. Having missing data on erectile dysfunction was significantly associated with the presence of depressive symptoms, physical inactivity, and increasing age, all of which were directly associated with erectile dysfunction in the available data. It follows that erectile dysfunction was associated with depressive symptoms in the missing cases, just as in the available cases. Even if no association existed in the missing cases, the overall relationship would remain statistically significant, which we confirmed by Monte Carlo simulation. Missing indicators for erectile dysfunction, depressive symptoms, or both, were randomly assigned according to the marginal proportions, ie, assuming no association in the missing cases. The crude odds ratio (OR), although reduced from 2.03 to 1.5 on average, remained significant (p < .04) in 24 of 25 Monte Carlo replications. We conclude that it is highly unlikely that our results can be attributed to response bias.
Crude associations were assessed by simple logistic regression as listed in Table 2. Multiple regression analysis was conducted to determine whether the association between erectile dysfunction and depressive symptoms could be attributed to confounding by illness, medication, hormonal, demographic, psychosocial, or lifestyle factors as listed in Table 3.
OR for erectile dysfunction were constructed from parameters of the fitted logistic regression model, with asymptotic 95% confidence intervals (CI). For age, OR is expressed per 5-year interval. For medication and other dichotomous predictors, OR is expressed as odds in favor of erectile dysfunction, given presence versus absence of the trait. For illnesses and other polytomous predictors, OR is expressed with respect to an appropriate reference category. Tests of significance were performed on the logistic model parameters using the asymptotic Wald chi squared statistic. "Significant" and "nonsignificant" as used in this report refer to inferences based on the 95% CI of the Wald test with 5% Type I error rate per predictor variable. Statistical Analysis System (SAS) software was used for all computations .
The prevalence of erectile dysfunction in the MMAS sample in 5-year age intervals nearly doubled, as illustrated in the top graph of Figure 1. From 25% of those aged 40 to 44 years, the prevalence of erectile dysfunction increased to 47% for ages 65 and above (see  for additional statistics on erectile dysfunction). The prevalence of depressive symptoms was constant across the age range averaging approximately 12%. (See Appendix for a brief summary of study statistics on depressive symptoms).
The direct correlation between erectile dysfunction and depressive symptoms is illustrated in the bottom graph of Figure 1. Within each decade of age, there is a positive association between depressive symptoms and erectile dysfunction; ie, those in the upper quintiles of CES-D score had higher rates of erectile dysfunction and this is independent of age. In simple logistic regression, depressive symptoms were strongly predictive of erectile dysfunction with an OR of 2.03 (95% CI 1.39-2.96). The association was consistent across strata of age, education, illness, and physical activity (Table 1).
Descriptive statistics are provided in Table 2 for a comprehensive list of potential confounders including illnesses, hormones, medications, demographics, lifestyle, and psychosocial measures. The potential for confounding is indicated by the strength of the association of each variable with both erectile dysfunction and depressive symptoms.
Erectile dysfunction was directly associated with age, diabetes, heart disease, high blood pressure, and related medications, as well as with several other illnesses. Erectile dysfunction was also significantly associated with low serum DHEA-S. Physically active men and light drinkers were less likely to have erectile difficulty. Among psychosocial predictors, erectile dysfunction was inversely correlated with life satisfaction, optimism, and unconcern with the future ("don't think about it"). Finally, erectile dysfunction was inversely associated with sexual thoughts and sexual desire.
Depressive symptoms showed strong associations with diabetes, heart disease, and high blood pressure. Aside from vasodilators and antidepressants, medications were generally not predictive of depressive symptoms. Hormone levels were not predictive of depressive symptoms. Physically active men and light drinkers were less likely to be experiencing depressive symptoms. Current smoking, low education, and recent changes in marital or employment status were also strongly associated. The depressive symptoms indicator was strongly correlated with both the life satisfaction scale and the scale of future expectations, but not with frequency of sexual thoughts or sexual desire.
(Table 3) contains the results from multiple logistic regression conducted with the entire list of significant predictors from Table 2. Depressive symptoms, a strong predictor of erectile dysfunction in simple regression (Table 3, Model A; OR 2.03, 95% CI 1.39-2.96), remained so when controlled for all the other predictors in the multiple regression (Table 3, Model B). The estimated OR for erectile dysfunction, controlled for the other predictors, was 1.82 in the presence of depressive symptoms as compared with the absence of depressive symptoms (95% CI 1.21-2.73). The other significant predictors identified by multiple regression were age (OR 1.15, 95% CI 1.06-1.24) and other para-aging phenomena, such as illness and physical activity (OR 0.55, 95% CI 0.36-0.85). Other significant predictors include heart disease, medication use, and alcohol intake. Conditional odds ratios for erectile dysfunction in the multiple regression analysis (Table 3, Model B) were similar to the crude odds ratios obtained for each predictor individually (Table 2). Hormone levels (p > .20) and antidepressant use (p > .75) did not predict erectile dysfunction when added to the regression model and did not affect the odds ratio or significance level for depressive symptoms. The subject's self-reported frequency of sexual thoughts (p < .0005) and sexual desire (p < .0001) showed a strong inverse association with erectile dysfunction, but also failed to diminish the independent predictive effect of depressive symptoms (Table 3, Models E and F).
Two additional models were constructed to test the confounding of depressive symptoms with the scales of life satisfaction and future expectation (Table 3). In each case, the odds ratio for depressive symptoms was reduced to a value exceeding unity but not statistically significant (Table 3, Models C and D). This gives partial substantiation for the CES-D as an indicator of depressive symptoms in this sample. The statistical significance of age, illness, and physical activity remained at comparable levels to the previous model (Table 3, Model B). The five-level life satisfaction scale was a statistically significant predictor of erectile dysfunction p = .001 (4 df), the most predictive level being "very pleased" (OR 0.53, 95% CI 0.38-0.75). The six-level scale of future expectations was also significant at p = .001, optimistic feelings being associated with the lower odds of erectile dysfunction. An expression of unconcern with the future ("don't think about it") had the strongest inverse association with erectile dysfunction, with odds ratios of 0.29 relative to those who were "unsure" about the future (95% CI 0.15-0.55).
The results of this study suggest that erectile dysfunction is associated with depressive symptoms. Moderate to complete erectile dysfunction is 1.82 times more likely in those who exhibit depressive symptomatology compared with those who do not. This association is independent of both age and para-aging factors (eg, demographic, anthropometric, and lifestyle factors, health status, medication use, hormones commonly associated with age) that, based on our analyses, had the potential to confound this association.
There are basically two types of explanatory models that can account for this relationship . One model emphasizes the ways in which this association may be explained either cognitively or behaviorally. For example, depressed individuals may be overly critical of themselves or have the tendency to self-focus (spectatoring). These attributes/behaviors, in the context of sexual relations, may lead to performance anxiety, which may inhibit the man's ability to obtain an erection. This relationship may also manifest itself through the effect of depression on libido or sexual desire. (1) Given that sexual desire is an important component of erectile response, a low level of desire may prevent erection. The second explanatory model is more biologically oriented and thus, explanations are stated in terms of the pathophysiological effects of depression. Depression is associated with a variety of neurophysiological disturbances; it can often impair the functioning of the autonomic nervous system . As a result, the parasympathetic nervous system may be unable to facilitate the relaxation of the penile smooth muscle tissue necessary for erection .
Dynamic Model of Male Erectile Dysfunction
We now present a model for understanding the complex and dynamic relation between depressive symptoms and male erectile dysfunction, which is based on a theoretical model of the disablement process outlined by Verbrugge and Jette . Use of this model is based on two assumptions that have been made frequently in the recent literature on erectile dysfunction. First, researchers have argued that between depression and erectile dysfunction there is a two-way, mutually reinforcing relationship [9,16,17], and second, that erectile dysfunction has a multifactorial etiology that includes affective components [16-19]. Thus, we recognize in this model that depressive symptoms may cause or exacerbate erectile dysfunction, and that erectile dysfunction may cause or exacerbate depressive symptoms (see Figure 2). Moreover, the results reported in this study support the multifactorial conceptualization of erectile dysfunction. We contend that the appeal of this model lies in the unique way in which it depicts both the dynamism and the multifactorial nature of the depressive symptoms/erectile dysfunction relationship.
In this model, there are three constructs specified that can hasten or arrest the impact these two conditions have on one another: a) risk factors, b) interventions, and c) exacerbators . Each of these constructs is represented in Figure 2 by a circle with an arrow protruding from it. Based on the findings reported here, examples of risk factors for erectile dysfunction include age, disease status, and level of physical activity (see Table 2). Interventions may include seeking treatment for one's erectile difficulty, which will presumably have an impact on the individual's depressive symptoms. Similarly, the effect that depressive symptoms have on erectile dysfunction should be diminished if an individual takes steps to address those depressive symptoms. Finally, the likelihood of erectile dysfunction causing the emergence of depressive symptoms may be augmented by a recent employment change (exacerbator; see Table 2).
Expanding the Definition of Male Erectile Dysfunction
Considering erectile dysfunction as a dynamic condition with a multifactorial etiology highlights the need to clarify the apparent theoretical and empirical confusion over its definition. It is conventionally defined as the consistent inability to achieve or sustain an erection of sufficient rigidity for sexual intercourse . However, we contend that this definition is missing a critical element-specifically, recognition of the role of the social context in which the condition arises. To illustrate this, we adopt the distinction between disease and illness that was expressed by Schneider and Conrad . According to these investigators, disease is a biomedical entity (although subject to social forces), whereas illness is the subjective experience of disease X. Current definitions of erectile dysfunction assume that it is simply a disease (ie, they deal with the physiological underpinnings of the inability to attain and maintain an erection), whereas we claim that it is both a disease and an illness (ie, the expression of that physical limitation in a social context). According to this expanded definition, whether an individual will consider himself to be suffering from erectile dysfunction is a function of the quality of his erections in relation to the parameters delineated by his social environment. Stated in a slightly different manner, it is a function of the male's perception that his erection is sufficiently rigid to permit satisfactory sexual performance , whether this includes coitus or not. By ignoring the social context in which erectile dysfunction arises, we believe it is possible to misconstrue the true nature of the problem.
Implications and Limitations
Given the strength of the association between depressive symptoms and erectile dysfunction, and the fact that at this point we are unable to determine which comes first, clinicians with patients who present with erectile difficulty should consider screening for depressive symptoms and those with patients who present with depressive symptomatology should consider screening for erectile dysfunction. Furthermore, on the basis of the analysis presented herein, treatment regimens may need to be altered to avoid the possible compounding effects of certain medications.
The findings presented here also have implications for medical education. The patients of soon-to-be physicians would certainly benefit were they to be attended by physicians who realize that male erectile dysfunction: a) is not something that exists "out there" as an objective fact, but is rather a process of subjective evaluation of the individual's erection with reference to his social milieu; b) may lead to declines in mental and physical health; and c) may affect or be the result of a wide range of aging and para-aging phenomena.
One limitation of the present study pertains to our effort to control for those subjects who may have been receiving treatment for their depressive symptoms or erectile dysfunction. Treatments may be of an invasive or noninvasive nature. We were able to avoid the confounding effects of the former by controlling for antidepressant medications (at the time of study, invasive treatments for erectile dysfunction were rarely, if ever, instituted). However, we were unable to provide information regarding whether subjects were engaged in some type of noninvasive treatment (ie, therapy) for either condition. Therefore, this association may not have proved significant if our analyses included this variable.
Another limitation to the present study is that we are unable to deal effectively with the fact that our measure of erectile dysfunction is not designed to account for the possibility that the presence of depressive symptomatology may cause an individual to distort his recall of the negative aspects of his sexual functioning.
Sexual-Activity Questions Related Directly to Potency Used to Construct Measure of MED
1. In an average week, how often do you have sexual intercourse or activity? (Enter number in box.)
2. During an average 24-hour day, how often do you have a full hard erection? (Enter number in box.)
3. During the last 6 months, have you ever had trouble getting an erection before intercourse begins?
c. Have not had sexual intercourse in last 6 months
4. During the last 6 months, have you ever had trouble keeping an erection once intercourse has begun?
c. Have not had sexual intercourse in last 6 months
5. How frequently do you awaken from sleep with a full erection?
b. 2 or 3 times per week
c. Once a week
d. 2 or 3 times a month
e. Once a month
f. Less than once per month
g. Not at all in the last 6 months
6. How satisfied are you with your sex life?
a. Extremely satisfied
b. Somewhat satisfied
c. Neither satisfied nor dissatisfied
d. Somewhat dissatisfied
e. Extremely dissatisfied
7. How satisfied are you with your sexual relationship with present partner or partners?
a. Extremely satisfied
b. Somewhat satisfied
c. Neither satisfied nor dissatisfied
d. Somewhat dissatisfied
e. Extremely dissatisfied
8. How satisfied do you think your partner(s) is (are) with your sexual relationship?
a. Extremely satisfied
b. Somewhat satisfied
c. Neither satisfied nor dissatisfied
d. Somewhat dissatisfied
e. Extremely dissatisfied
9. Has the frequency of your sexual activity with a partner been
a. as much as your desire;
b. less than you desire;
c. more than you desire?
Summary Statistics for the CES-D When Treated as a Continuous Variable
Scores on the CES-D may range from 0 to 60. The range for the 1670 subjects in the MMAS with complete data on the CES-D was 0 to 47. The distribution of CES-D scores was highly skewed, with the usual cut-off value of 16 being close to the 90th percentile. The median CES-D score was 4. Scores of 1 and 10 correspond to the 25th and 75th quartiles, respectively.
(1) As stated in the Results section, sexual desire was associated with erectile dysfunction but not with depressive symptoms. This finding is, without doubt, counterintuitive and certainly requires explanation. As noted by one anonymous reviewer, this raises the question of whether the measures of depression and/or sexual desire are appropriate or, alternatively, that the evidence in the literature of a common occurrence of low sexual desire in depression is misleading or incorrect. Our measure of sexual desire does not seem to be inappropriate. We assessed sexual desire with the conventionally used question: "How frequently do you feel sexual desire? This feeling may include wanting to have sexual experience (masturbation or intercourse), planning to have sex, feeling frustrated due to lack of sex, etc." After a thorough review of the scant scientific literature on the affective components of sexual desire (especially depression), we think this null finding can be explained in one of two ways: a) The CES-D differs from other more clinically oriented measures of depression [for example, a diagnosis of Primary Affective Disorder Depression, use of the SCL-90, in its full or abbreviated form as the Brief Symptom Inventory (BSI)] in that it captures many individuals who exhibit relatively mild depressive symptoms, in addition to the more severe cases. Traditionally, loss of sexual interest has been subsumed under anhedonia and has been associated with the more severe endogenous or melancholic forms of depression (see ). Therefore, using the CES-D may have diluted the association normally found between depression and sexual desire. The CES-D is, however, probably the most validated instrument in all of social science research and is an appropriate measure for assessing depressive symptoms in large epidemiological studies. b) The literature pertaining to the role of depression in sexual desire, as was the case with the literature on male erectile dysfunction and depression, suffers considerable (and quite similar) methodological flaws. Research is often limited to patient samples [3,8,9,38-40] and, in many cases, depression is poorly operationalized, as in the case of mood induction experiments [8,9]. Thus, previous research may have shown a stronger relationship between depression and sexual desire than actually exists in the population. Although our null finding is probably due to one or both of these two issues, we believe that until additional rigorous investigation is conducted that bears directly on this issue, we are not in a position to judge which is the culprit.
1. Heiman JR, Rowland DL: Affective and physiological sexual response patterns: The effects of instructions on sexually functional and dysfunctional men. J Psychosom Res 27:105-116, 1983
2. Munjack DJ, Oziel LJ, Kanno PH, et al: Psychological characteristics of males with secondary erectile failure. Arch Sex Behav 10:123-131, 1981
3. Nofzinger EA, Thase ME, Reynolds CF, et al: Sexual function in depressed men. Arch Gen Psychiatry 50:24-30, 1993
4. Mathew RJ, Weinman ML: Sexual dysfunctions in depression. Arch Sex Behav 11:323-328, 1982
5. Thase ME, Reynolds CF, Jennings JR, et al: Diagnostic performance of nocturnal penile tumescence studies in healthy, dysfunctional (impotent), and depressed men. Psychiatry Res 26:79-87, 1988
6. Thase ME, Reynolds CF, Jennings JR, et al: Nocturnal tumescence is diminished in depressed men. Biol Psychiatry 24:33-46, 1988
7. Thase ME, Reynolds CF, Glanz LM, et al: Nocturnal penile tumescence in depressed men. Am J Psychiatry 144:89-92, 1987
8. Meisler AW, Carey MP: Depressed affect and male sexual arousal. Arch Sex Behav 20:541-554, 1991
9. Meisler AW: The effects of elated, depressed, and neutral affect on sexual arousal in sexually functional men (impotence). Syracuse University (0659) 53:11-B of Dissertation Abstracts Int., 1992, 5985
10. Karacan IF, Scott FB, Hartse K, et al: Prevalence of erectile dysfunction in sleep disorders center. In Guilleminault CC, Lugaresi E (eds), Sleep/Wake Disorders: Natural History, Epidemiology, and Long-term Evolution. New York, Raven Press, 1983, 261-273
11. Wolchik SA, Beggs VE, Wincze JP, et al: The effect of emotional arousal on subsequent sexual arousal in men. J Abnorm Psychol 89:595-598, 1980
12. Buvat J, LeMaire A, Buvat-Herbaut M, et al: Comparative investigations in 26 impotent and 26 non-impotent diabetic patients. J Urol 133:34-38, 1985
13. Tamburello A, Seppecher MF: The effects of depression on sexual behavior: Preliminary results of research. In Geine R, Wheeler CC (eds), Progress in Sexology. New York, Plenum Press, 1977,107-128
14. Roose SP, Glassman AH, Walsh BT, et al: Reversible loss of nocturnal penile tumescence during depression: A preliminary report. Neuropsychobiology 8:284-288, 1982
15. Reynolds CF, Frank E, Thase ME, et al: Assessment of sexual function in depressed, impotent, and healthy men: Factor analysis of a brief sexual function questionnaire for men. Psychiatry Res 24:231-250, 1988
16. Smith, AD: Psychologic factors in the multidisciplinary evaluation and treatment of erectile dysfunction. Urol Clin North Am 15:41-51, 1988
17. Barlow DH: Causes of sexual dysfunction: The role of anxiety and cognitive interference. J Consult Clin Psychol 54:140-148, 1986
18. Hengeveld MW: Erectile dysfunction: A sexological psychiatric review. World J Urol 227:232, 1983
19. Jacobs JA, Fishkin R, Cohen S, et al: A multidisciplinary approach to the evaluation and management of male sexual dysfunction. J Urol 129:35-37, 1983
20. McKinlay JB, Feldman HA: Age-related variation in sexual activity and interest in normal men: Results from the Massachusetts Male Aging Study. In Rossi AS (ed), Sexuality Across the Lifecourse: Proceedings of the MacArthur Foundation Research Network on Successful Mid-Life Development, 1992. New York, University of Chicago Press, 1994, 261-285
21. Feldman HA, Goldstein I, Hatzichristou DG, et al: Impotence and its medical and psychosocial correlates: Results of the Massachusetts Male Aging Study. J Urol 151:54-61, 1994
22. McKinlay JB, Longcope C, Gray A: The questionable physiologic and epidemiologic basis for a male climacteric syndrome: Preliminary analysis from the Massachusetts Male Aging Study. Maturitas 11:103-115, 1989
23. Longcope C, Goldfield SRW, Brambilla DJ, et al: Androgens, estrogens, and sex hormone-binding globulin in middle-aged men. J Clin Endocrinol Metab 71:1442-1446, 1990
24. Ascherio A, Stampher MJ, Colditz GA, et al: Nutrient intakes and blood pressure in normotensive males. Int J Epidemiol 20:886-891, 1991
25. Gray A, Jackson DN, McKinlay JB: The relation between dominance, anger, and hormones in normally aging men: Results from the Massachusetts Male Aging Study. Psychosom Med 53:375-385, 1991
26. Bremner WJ, Vitiello MV, Prinz RN: Loss of circadian rhythm in blood testosterone levels with aging in normal men. J Clin Endocrinol Metab 56:1278-1281, 1983
27. McKinlay JB, Kipp DM, Johnson P, et al: A field approach for obtaining physiological measures in surveys of general populations: Response rates, reliability, and costs. In Proceedings of the Fourth Conference on Health Survey Research Methods. Washington, DC, United States Department of Health and Human Services Publication No. PHS 84:3346, 1984
28. Hunt SM, McKenna SP, McEwen J, et al: A quantitative approach to perceived health status: A validation study. J Epidemiol Community Health 34:281-286, 1980
29. McEvoy GK (ed): American hospital formulary service drug information. Bethesda, MD, American Society of Hospital Pharmacists, 1989
30. Comstock GW, Helsing KJ: Symptoms of depression in two communities. Psychol Med 6:551-564, 1976
31. Weissman MM, Sholomskas D, Pottenger M, et al: Assessing depressive symptoms in five psychiatric populations: A validation study. Am J Epidemiol 106:203-214, 1977
32. Radloff LS: The CES-D scale: A self-report depression scale for research in the general population. Appl Psychol Meas 1:385-401, 1977
33. Beekman AT, Deeg DJ, Van Limbeek J, et al: Criterion validity of the Center for Epidemiologic Studies Depression scale (CES-D): Results from a community-based sample of older subjects in The Netherlands. Psychol Med 27:231-235, 1997
34. Feldman HA, Goldstein I, Hatzichristou DG, et al: Construction of a surrogate variable for impotence in the Massachusetts Male Aging Study. J Clin Epidemiol 47:457-467, 1994
35. Pate RR, Pratt M, Blair SN, et al: Physical activity and public health: A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. J Am Med Assoc 273:402-407, 1995
36. Khavari KA, Farber Pd: A profile instrument for the quantification and assessment of alcohol consumption. J Stud Alcohol 39:1525-1539, 1978
37. SAS Institute, Inc: SAS/STAT User's Guide, Version 6, 4th ed. Vol. 2. Cary, NC, SAS Institute, 1989, 1071-1126
38. Howell JR, Reynolds CF, Thase ME, et al: Assessment of sexual function, interest and activity in depressed men. J Affect Disord 13:61-66, 1987
39. Mathew RJ, Largen J, Claghorn J: Biological symptoms of depression. Psychosom Med 41:439-443, 1979
40. Casper RC, Redmond E, Katz M, et al: Somatic symptoms in primary affective disorder. Arch Gen Psychiatry 42:1098-1104, 1985
41. Thase ME, Howland RH: Biological processes in depression: An updated review and integration. In Edward E, Leber WR (eds), Handbook of Depression, 2nd Edition. New York, Guilford Press, 1995, 213-279
42. Saenz de Tejada I, Goldstein I, Blanco R, et al: Smooth muscle of the corpora cavernosae: Role in penile erection. Urol Clin North Am 15:1-7, 1988
43. Verbrugge LM, Jette AM: The disablement process. Soc Sci Med 38:1-14, 1994
44. Krane RJ, Goldstein I, Saenz de Tejada I: Impotence. N Engl J Med 321:1648-1659, 1989
45. Schneider JW, Conrad P: Medical and sociological typologies: The case of epilepsy. Soc Sci Med 15A:211-219, 1981
46. NIH Consensus Development Panel on Impotence: Impotence. J Am Med Assoc 270:83-90, 1993