Sexuality Among Middle-Aged and Older Adults With Diagnosed and Undiagnosed Diabetes: A national, population-based study

NSHAP involved a nationally representative probability sample of community-dwelling adults aged 57–84 years (at the time of screening), generated from U.S. households screened in 2004, described in detail elsewhere (7). Of 4,017 eligible subjects in the sample, 3,005 (1,455 men and 1,550 women) were interviewed at home between July 2005 and March 2006, yielding a weighted response rate of 75.5% (unweighted 74.8%). The protocol was approved by the University of Chicago and NORC institutional review boards; all respondents gave written informed consent.

Details of demographic and sexuality measures have been reported previously (3). Sex was defined as “any mutually voluntary activity with another person that involves sexual contact, whether or not intercourse or orgasm occurs.” Sexually active respondents were asked about the presence of sexual problems selected on the basis of diagnostic and clinical criteria for sexual dysfunction (3). All respondents who had not had sex in the previous 3 months were asked to indicate why from a list of reasons (3). A self-administered questionnaire completed during the in-home interview asked about the frequency of masturbation, defined as “stimulating your genitals (sex organs) for sexual pleasure, not with a sex partner,” and ascertained whether orgasm occurred with masturbation. Questions about sexual activity and problems were refused by 2–7% of respondents; 12–13% declined to answer the questions regarding masturbation.

Individuals were classified as having “diagnosed diabetes,” regardless of their A1C value, if they responded that they had been told by a physician that they had diabetes or if they were using one or more diabetes medications. To identify individuals with “undiagnosed diabetes,” we used an A1C cut point of 6.0% based on the correlation of A1C with the traditional fasting glucose criterion in older individuals. This A1C cut point for identifying undiagnosed diabetes was selected based on a sensitivity/specificity analysis of data from the 1999–2004 U.S. National Health and Nutrition Examination Survey for individuals aged 57–85 (8). By comparing different A1C cut points with diagnosis of diabetes based on fasting glucose levels, an A1C cut point of 6.0% maximized specificity of the assay for detecting undiagnosed diabetes without compromising sensitivity for all cut points examined between 5.0 and 7.0% (specificity 0.91 for men and 0.91 for women; sensitivity 0.68 for men and 0.69 for women) (supplementary Fig. 1, available in an online appendix at http://care.diabetesjournals.org/cgi/content/full/dc10-0524/DC1). Individuals who did not have diagnosed diabetes were classified as having “no diabetes” if their A1C value was <6.0% and were classified as having “undiagnosed diabetes” if their A1C value was ≥6.0% (supplementary Fig. 1A). In light of recent changes in international diabetes care guidelines, we also performed sensitivity analyses using the A1C 6.5% cut point criterion and summarize these in the discussion (9,10).

Details of medication data collection by direct observation and medication classification and coding have been described previously (11). Sixteen percent of all individuals and 26% of those in the analytic sample (17 and 24% weighted, respectively) were taking at least one medication classified as an antidiabetic agent “on a regular schedule, like every day or every week.” These agents (and the weighted proportion of individuals in the analytic sample taking them) included biguanides (13.9%), sulfonylureas (12.5%), thiazolidinediones (7.2%), insulin (5%), antidiabetes combinations (2.6%), meglitinides (0.7%), α-glucosidase inhibitors (0.3%), and miscellaneous antidiabetes agents (0.1%). Of all the individuals using one or more diabetes medications, 95.1% also reported a diabetes diagnosis. Of individuals classified as having diagnosed diabetes, 3.9% were classified on the basis of medication data alone.

Fingerstick dried blood specimens were sought from a random two-thirds of study respondents (n = 2,494), with a cooperation rate of 84.4% (n = 2,105) (supplementary Fig. 2B, available in an online appendix). A1C was obtained using well-validated dried blood spot methods described previously (12,13). Adequate specimens were obtained for analysis from 1,746 respondents.

Physical health was self-rated using the 5-point “excellent,” “very good,” “good,” “fair,” or “poor” scale. Comorbidities were assessed using the Katz modification of the Charlson index (14) (diabetes excluded), and activities of daily living were assessed using the Katz Activities of Daily Living Scale (15). Respondents were asked whether a medical doctor had ever told them they had any of several common diabetes-related complications (Table 1). Communication with a physician about sexual matters since age 50 was assessed as described previously (3). Depressive symptoms were assessed using the 11-item short form of the Center for Epidemiological Studies-Depression (CES-D) index (16), with response options 0–3; a score ≥9 was considered indicative of a clinically significant level of depressive symptoms, consistent with a threshold of 16 on the 20-item scale (17).

The analytic sample consisted of the 1,993 participants for whom diabetes status could be determined based on A1C, medication data, and/or self-report (supplementary Fig. 2A).

Demographic and clinical characteristics were estimated separately within each of the three diabetes status groups by gender. Bivariate associations with diabetes status were tested using the Rao and Scott (18) correction to the χ² test to account for the survey design. Logistic regression was used to model associations between diabetes conditions and sexual activity, behavior, and problems separately by gender.

All models were adjusted for age-group (57–64, 65–74, and 75–85 years), depressive symptoms (CES-D scores <9 vs. ≥9), and the modified Charlson comorbidity index (0, 1–2, and ≥3) except for outcomes with too few individuals in either outcome category to support a fully adjusted model (19). Education and race were also evaluated as potential confounders of the effect of diabetes status. For outcomes with a small number of individuals in either outcome category, confounding effects of each covariate listed above were evaluated separately; unadjusted models are presented for these outcomes unless confounding, defined as a change in the odds ratio of ≥10%, was identified.

All analyses accounted for the survey sampling design through incorporation of sampling strata and clusters, as well as weights that adjusted for a differential probability of selection and differential nonresponse. All reported estimates are weighted. All analyses were performed with STATA statistical software (version 10).

Table 1 summarizes the demographic and health characteristics of the analytic sample, stratified by diabetes status. Self-rated health and capacity for activities of daily living were consistently lower, and the prevalence of several diabetes complications and comorbidities were consistently higher for individuals with diagnosed diabetes compared with those with no diabetes, with intermediate results for those with undiagnosed diabetes.

Men, regardless of age or diabetes status, were more likely than women to be married or living with a partner (Table 1) and were significantly more likely than women to be currently sexually active (Table 2). Sixty-one percent of men (69% of partnered men) and 33% of women (62% of partnered women) with diagnosed diabetes were currently sexually active. Women with diagnosed diabetes were less likely than men with diagnosed diabetes (adjusted odds ratio [AOR] 0.28 [95% CI 0.16–0.48]) and other women (0.63 [0.45–0.87]) to be sexually active. Among sexually active individuals, the majority engaged in sexual activity at least two to three times per month and neither the frequency of sexual activity nor specific partnered sexual behaviors differed by diabetes status or gender.

Among sexually active individuals, partnered sexual behaviors did not differ by gender or diabetes status (Table 2). However, adults with diagnosed diabetes were less likely than others to masturbate (AOR 0.50 [95% CI 0.32–0.78 for women] 0.66 [0.44–0.97 for men]) and to experience orgasm with masturbation (Table 3).

Sexual problems were ascertained only for individuals who were sexually active in the prior 12 months (Table 3). Men with diagnosed diabetes were more likely than other men to report lack of interest in sex (AOR 1.72 [95% CI 1.12–2.63]); among women, interest in sex did not differ by diabetes status. The prevalence of orgasm problems (inability to climax or climaxing too quickly) was similarly elevated among men with diagnosed and undiagnosed diabetes compared with men without diabetes, but erectile difficulties were elevated only among men with a diabetes diagnosis (2.52 [1.53–4.14]).

Among all individuals who had not been sexually active for ≥3, men with diagnosed diabetes were more likely than all other groups (men and women) to report that they had not had sex because of their own physical health problems (60.9 vs. 34.5% for men with undiagnosed diabetes and 39.4% for men with no diabetes, P < 0.001; 16.2% of women with diabetes vs. 8.5% with undiagnosed diabetes and 9.2% with no diabetes). Among women, the common reasons for sexual inactivity were similar between those with diagnosed and no diabetes, but women with undiagnosed diabetes were more likely to report lack of interest as a reason for sexual inactivity (54.5% for undiagnosed diabetes vs. 44.9% for diagnosed diabetes and 38.0% for no diabetes, P < 0.05).

Men with diagnosed diabetes were more than twice as likely (46.8%) as women with diagnosed diabetes (18.8%) to discuss sex with a physician compared with 28.0% of men and only 11.3% of women with undiagnosed diabetes (supplementary Fig. 3A and B, available in an online appendix). Among those who had discussed sexual matters with a physician, 16.7% of men overall (10.0% of men with diagnosed diabetes) compared with 30.5% of women overall (28.4% of women with diagnosed diabetes) reported that the physician initiated the conversation. Approximately one-third of sexually active men and women with sexual problems reported avoiding sex because of problems (Table 3); this number did not vary by diabetes status.

Our findings, based on nationally representative U.S. data, indicate that two-thirds of men and approximately one-third of women aged 57–85 years with diabetes were sexually active. Although diabetes was associated with a higher rate of sexual inactivity, those who were active participated in partnered sexual behaviors and activity at a rate similar to that of those without diabetes. As a group, the majority of individuals with diabetes were married or living with a partner, although women with diabetes were more likely to be alone. Sexually active adults with diabetes had a similar prevalence of sexual problems, and women were more likely than men to avoid sex because of these problems. However, fewer than one in five women with diagnosed diabetes compared with nearly half of men had discussed sex with a physician. Individuals with undiagnosed diabetes, particularly women, were even less likely than others to have discussed sex with a physician.

In this study, we combined self-report measures, medication, and biological measures from a population-based probability sample to stratify individuals as having diagnosed, undiagnosed, or no diabetes. There is not yet full agreement about using A1C to diagnose diabetes in older adults (9,10,20), but our strategy generated estimated prevalences of diagnosed and undiagnosed diabetes comparable to 2005–2006 U.S. population estimates using fasting plasma glucose and/or oral glucose tolerance testing for community-residing individuals aged ≥60 years (among those with A1C results, 20.5% [95% CI 17.5–24%] of 901 women and 25% [21–29%] of 843 men had diagnosed diabetes, whereas 19% [16–22%] of women and 22% [19–25%] of men had undiagnosed diabetes) (21). Repetition of the analyses shown here using a 6.5% glycosylated hemoglobin threshold for diabetes classification in this population yielded few qualitative differences in the outcomes of interest but did result in a far smaller undiagnosed diabetic group (4.7% in women and 5.6% in men) than found by classification based on traditional diagnostic criteria. Using either threshold, as a group, individuals with undiagnosed diabetes are different from those with diagnosed diabetes in two important ways. First, those with undiagnosed diabetes seemed to be earlier in the course of the disease. Second, undiagnosed diabetes is a pathophysiological state that lacks the psychological burden and/or social stigma associated with having diagnosed diabetes (5).

The etiology of sexual problems associated with undiagnosed diabetes (controlling for other physical and psychological factors known to be associated with sexual problems) might reflect a predominant physiological mechanism whereas the etiology of sexual problems associated with diagnosed diabetes might be more likely to have an additional, diabetes-specific psychosocial component. Although cross-sectional data cannot determine the causal direction of such relationships, understanding the sexuality of individuals with undiagnosed compared with that of those with diagnosed and no diabetes can shed light on the pathological mechanisms and the natural history of both diabetes and sexual dysfunction in later life.

In this study, aside from the expected higher prevalence of erectile dysfunction in men with diagnosed diabetes (55%) (1), the prevalence of many sexual problems did not differ significantly according to diabetes status. Dropping out from sexual activity may partly explain the lack of a diabetes association with sexual problems, especially in women. This finding is suggested by the significantly higher prevalence of sexual inactivity among women with diagnosed diabetes compared with that for men and a greater lack of interest in sex among sexually inactive women with diabetes compared with those without. Furthermore, women with diabetes (diagnosed and undiagnosed) were nearly half as likely as other women to report masturbating, suggesting a reduction in sexual drive that was independent of partner status and of knowledge of the disease. The prevalence of masturbation was also lower in men with diagnosed or undiagnosed diabetes compared with that in men without but was three times higher than in women with diabetes (45%).

Interestingly, the rate of erectile dysfunction was not markedly elevated in men with undiagnosed diabetes (36 vs. 32% in men without diabetes), but the inability to experience orgasm was high and comparable to that of men with diagnosed diabetes (29 vs. 16% in men without diabetes). This finding suggests that loss of orgasmic function may not only occur as a consequence of erectile dysfunction as described by others (22) but also may actually precede erectile dysfunction, at least as perceived by some men with diabetes. In women, inability to experience orgasm with masturbation was also significantly higher among those with diagnosed diabetes. Physicians who do ask about sexual function tend to engage patients with partners (23) and focus on male erectile issues for which treatment is readily available. Asking about orgasm function in relation to partnered sex and masturbation and expanding these discussions to include women may assist in prevention of downstream sexual problems and personal and interpersonal distress and in earlier diagnosis of diabetes in some individuals. Although no pharmacological treatment is approved as a remedy for anorgasmia, interventions such as education to inform the patient that anorgasmia is known to occur for a substantial proportion of sexually active individuals with diabetes, directed masturbation, use of a clitoral pump in women, and discussion of ways to enhance sexual arousal and intimacy can be therapeutic.

Medications are another important iatrogenic etiology of later-life sexual problems (24). Glucose-lowering medications are largely thought to improve sexual function by mitigating glycemic-related microvascular damage, as seen in clinical studies of erectile function in men with diabetes (4). The stratification strategy used in our study classifies all individuals taking glucose-lowering medications as having diagnosed diabetes. This strategy would tend to underestimate the association between diagnosed diabetes and sexual function, particularly for the subgroup with uncontrolled diabetes. However, other medications used to treat diabetes, including antihypertensive and cholesterol-lowering drugs, may have deleterious effects on sexual function (24). Because of sample size, this study is limited by an inability to account for the effects of other medications in estimating the association between diabetes status and aspects of sexuality. Prospective clinical trials are needed to fully elucidate the effects and interactions of medications on sexual activity and function among middle-age and older adults with diabetes; virtually nothing is documented about the effects of diabetes medications on sexuality in older women.

The prevalence of specific sexual problems was only assessed for those who were sexually active in the prior 12 months, therefore underestimating the prevalence of sexual problems in this population. Next, in addition to the expected differences in population prevalence estimates for undiagnosed diabetes, reanalysis using a glycosylated hemoglobin cut point criterion of 6.5% results in differences between the groups for some outcomes, in part because of loss of precision in estimates. For example, the rate of erectile dysfunction was still higher in men with undiagnosed diabetes (40.5%) compared with that of men without diabetes (32.1%), but the AOR comparing these two groups was no longer significant (0.63 [95% CI 0.25–1.58]). No substantive differences were found in diabetes status comparisons among sociodemographic, health, or communication variables. As with virtually all clinical and population-based research on human sexuality, these data were self-reported, although the interview methods are widely accepted as being valid (25). Use of a population-based probability sample adds to prior knowledge (1,2) about later-life diabetes and sexuality by disaggregating individuals with undiagnosed or preclinical diabetes from those with no diabetes. This study builds on prior work by filling a void of information about older women's sexuality and gender differences in sexuality among middle-age and older adults with diabetes. Further research should be powered to also look at age-group comparisons.

In conclusion, many middle-aged and older adults with diabetes are sexually active. Sexual problems are common but are infrequently discussed with physicians, especially among women. Physician knowledge about sexuality in relation to diabetes should improve patient education and counseling, as well as the identification of symptoms that could signal undiagnosed disease or a high risk for disease. Attention to potentially treatable sexual problems in middle-aged and older adults with diabetes should improve quality of life and enhance overall diabetes management.

NSHAP is supported by the National Institutes of Health, including the National Institute on Aging, the Office of Research on Women's Health, the Office of AIDS Research, and the Office of Behavioral and Social Sciences Research (grant 5R01-AG-021487). NSHAP is also supported by NORC, whose staff was responsible for the data collection. Donors of supplies include Orasure, Sunbeam Corporation, A&D/LifeSource, Wilmer Eye Institute at the Johns Hopkins Bloomberg School of Public Health, Schleicher & Schuell Bioscience, Biomerieux, Roche Diagnostics, Digene Corporation, and Richard Williams. S.T.L., A.G., and D.M.Q. were supported by the National Institutes of Health, National Institute on Aging University of Chicago–NORC Center on Demography and Economics of Aging Core on Biomarkers in Population-Based Aging Research (5P30-AG-012857). S.T.L. is also supported by the National Institute on Aging (grant 1K23-AG-032870-01A1). A.V., E.S.H., M.L.D. D.M.Q., and M.H.C. are supported by the National Institute of Diabetes and Digestive and Kidney Diseases Diabetes Research and Training Center (grant P60-DK-20595) E.S.H. is also supported by a the National Institute of Diabetes and Digestive and Kidney Diseases (project grant R01-DK-081796-01A1). M.H.C. is also supported by a Midcareer Investigator Award in Patient-Oriented Research from the National Institute of Diabetes and Digestive and Kidney Diseases (K24-DK-071933).

The investigators maintained complete independence from the funding organizations in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, and approval of the manuscript.

No other potential conflicts of interest relevant to this article were reported.

S.T.L. developed the study concept and design, acquired data, analyzed and interpreted data, drafted the manuscript, obtained funding, provided administrative, technical, or material support, and provided overall study supervision. H.T. analyzed and interpreted data, provided statistical analysis, wrote the manuscript, and reviewed/edited the manuscript. A.G. analyzed and interpreted data and wrote the manuscript. A.V. analyzed and interpreted data, provided statistical analysis, and reviewed/edited the manuscript. E.S.H. analyzed and interpreted data, reviewed/edited the manuscript, and lent administrative, technical, or material support. M.L.D. provided statistical analysis and reviewed/edited the manuscript. D.M.Q. analyzed and interpreted data and reviewed/edited the manuscript. M.H.C. obtained funding, analyzed and interpreted data, reviewed/edited the manuscript, and provided administrative, technical, or material support.

We acknowledge research assistance provided by Jessica Schwartz, Katherine Githens, Andreea Mihai, and Emily Abramsohn, MPH, employed by Dr. Lindau at the University of Chicago. They contributed technical assistance in preparing the manuscript for publication and editing the bibliographic material.