The worldwide prevalence and epidemiology of erectile

International Journal of Impotence Research (2000) 12, Suppl 4, S6±S11
ß 2000 Macmillan Publishers Ltd All rights reserved 0955-9930/00 $15.00
www.nature.com/ijir
The worldwide prevalence and epidemiology of erectile dysfunction
JB McKinlay1*
1New
England Research Institutes Inc., Watertown, Massachusetts, USA
This paper: (i) describes the worldwide prevalence of erectile dysfunction (ED); (ii) presents agespeci®c incidence rates for ED in the US; (iii) summarizes some key epidemiologic correlates of ED
in the general population; and (iv) considers the possibility that ED may be a biobehavioral marker
(sentinel event) of subsequent cardiovascular disease in men. Clinical, anthropometric, life style
and hormonal data are presented from the milestone Massachusetts Male Aging Study (MMAS), a
large (over 1000) prospective cohort of randomly sampled community-dwelling, normally aging
men. Newly updated population prevalence and (more importantly) age-speci®c incidence rates
are reported. We also estimate the likely magnitude of ED that will accompany the worldwide
globalization of aging. Key correlates (predictors) of incident ED, especially vasculogenic
in¯uences, are identi®ed and discussed. In conclusion, ED is a common biobehavioral
phenomenon and there are strong physiological and epidemiological reasons for considering it
a major marker (or predictor) of subsequent cardiovascular disease in men. International Journal
of Impotence Research (2000) 12, Suppl 4, S6±S11.
Keywords: incidence; prevalence; questionnaire; depression; cardiovascular risk; impotence
Introduction
There was very little reliable information on erectile
dysfunction (ED) available until the very end of the
twentieth century.1 Pioneering contributions during
the 1950s were not extended into later decades and
half-century-old ®ndings had little relevance to very
different subsequent cohorts. That situation has
changed dramatically Ð more has been learned
about ED (a medical condition formerly termed
`impotence') during the past decade than was
known during the rest of the twentieth century
combined. Prior to about 1985, most information on
ED was based on clinical impressions and studies of
convenience samples of atypical patients.
The Massachusetts Male Aging Study (MMAS)
was the ®rst major epidemiologic investigation of
ED2,3 and it is the largest population-based study to
date with information on the prevalence, incidence
and etiology of ED. The overall research design of
the MMAS is summarized in Figure 1. The MMAS is
a random-sample prospective study of health and
aging in middle-aged men, addressing speci®c
conditions of importance to older men (prostate
cancer (PCa), hormone status, and erectile function)
in the context of comprehensive information on
physiology, behavior, illness, medication, anthropometrics, sexual activity, psychology and lifestyle.4 It
*Correspondence: JB McKinlay, New England Research
Institutes Inc., 9 Galen Street, Watertown, Mass, USA.
E-mail: [email protected]
is the only large random sample study to provide
longitudinal data regarding the epidemiology of ED
and other aspects of sexual function in men over the
age of 60. The MMAS baseline study (1987 ± 1989)
established a cohort of 1709 men aged 40 ± 70, of
whom 1156 were re-interviewed in 1995 ± 1997. The
study has provided important new cross-sectional
and longitudinal information regarding changes in
hormone status and sexual function with respect
to aging, the relation of these changes to ED and
prostate cancer, and preliminary evidence for a
relation between ED and cardiovascular disease.
Prevalence and worldwide burden
The prevalence of minimal, moderate and complete
ED from the baseline (T1) MMAS (1987 ± 1989) is
summarized in Figure 2. It should be noted that
these estimates improve upon and re®ne our earlier
published estimates,2 mainly as a result of a better
method (a new discriminant function) for estimating
ED in a population-based epidemiological study.5
Our new approach is less susceptible to selection
bias, results in less missing data and has greater
construct validity than our earlier widely accepted
method.2 The prevalence and degree of ED (minimal, moderate and complete) clearly increases with
age. While 8% of men in their forties report either
moderate or complete ED, this approaches 40% in
men aged 60 ± 69 y. In short, men between the ages
Epidemiology of erectile dysfunction
JB McKinlay
S7
Figure 1 Description of MMAS cohorts: MMAS-1 (1987 ± 1989); MMMAS-2 (1995 ± 1997), and a proposed MMAS-3 (2001 ± 2003).
Figure 2 Prevalence of erectile dysfunction: Massachusetts Male Aging Study, 1987 ± 1989 (n ˆ 1626); source Kleinman et al. J Clin
Epidemiol 2000.5
of 60 and 69 y experience more than four times the
level of moderate or complete ED compared to their
40 ± 49-y-old counterparts.
Using recent MMAS estimates, Figure 3 shows
the likely worldwide (age-adjusted) prevalence of
ED between 1995 and 2025 (employing the most
conservative, ie lowest, United Nations population
projections).6 The worldwide prevalence of ED will
probably increase from 152 million men in 1995 to
322 million men with ED in 2025, representing an
increase of 170 million.7 Much of this increase will
occur in the developing world and is associated
with the aging of the world's population. Other
related factors (secular increases in weight, dietary
changes, smoking behavior and an emerging
diabetes pandemic) are also contributing to the
increase.
The MMAS has provided the ®rst incidence rates
for ED (Table 1). Incidence rates, which comprise
new cases of a condition, or a transition from a
Table 1 Estimated number of new cases of ED annually in the
US; MMAS rates and 1990 US Census data (men, aged 40 ± 69)
Age
Number of
men at risk
Age-speci®c
incidence rate
for ED=1000
Expected number
of new ED cases
40 ± 49
50 ± 59
60 ± 69
11 928 664
7 540 415
5 282 236
12.4
29.8
46.4
147 915
224 704
245 096
Total
24 751 351
617 715
nondiseased to a diseased state, obviously require
prospective data for their calculation. Prior to the
MMAS, the only published estimate of the incidence of ED was from a convenience sample of 3250
men aged 26 ± 83 y visiting a preventive medicine
clinic in Texas between 1987 and 1991.8 Johannes et
al 9 has recently reported age-category speci®c
incidence rates for ED and shows how they differ
International Journal of Impotence Research
Epidemiology of erectile dysfunction
JB McKinlay
S8
Figure 3 Prevalence of erectile dysfunction: between 1995 and 2025, the world total increase ˆ 169,932,786. Source: World Population
Prospects: the 1996 Revision. New York: United Nations, 1997. The most conservative (lowest) projection was employed. Prevalence
estimates of ED were employed for data from MMAS 1987 ± 1989 for ages 40 ± 69 and MMAS 1995 ± 1997 for ages 70 ± 79 y.
Figure 4 From observation to experimentation: history of the development of scienti®c knowledge concerning erectile dysfunction (ED).
Moving left to right, this diagram represents the chronologic improvement in the knowledge base about ED, from uncontrolled
observations to experimental evidence from ®eld trials.
by sociodemographic status (education, income)
and also health status.
Improving the epidemiological knowledge of
ED: from observation to experimentation
Figure 4 summarizes the recent development of
scienti®c knowledge concerning ED. Moving from
the left toward the right-hand side of Figure 4
(indicated with bold arrow) represents improvement
in the knowledge base about ED. We have thus
moved from uncontrolled observations to crosssectional surveys and prospective studies towards
International Journal of Impotence Research
experimental evidence from ®eld trials. Prior to
about 1980, and with the notable exceptions of
Kinsey in 194810 and Masters and Johnson in
1966,11 little was known about ED and clinical
opinion prevailed. Real progress was made after the
1992 NIH Consensus Conference on ED,12 during
which a working de®nition was agreed upon. With
this de®nition, `the persistent inability to attain and
maintain a penile erection adequate for sexual
performance' investigators could begin to measure
ED. That landmark conference also called for `major
research efforts to improve diagnostic classi®cation
for epidemiologic purposes'. It was a milestone
event because it underscored key features of ED: (a)
that it is a subjectively identi®ed state for each man
Epidemiology of erectile dysfunction
JB McKinlay
(the means for its `objective' measurement with
technology still require improvement and validation); (b) that it focuses on behavior (penile
erections) and not underlying aspects of libido
(sexual drive or interest); and (c) it recognized that
ED, like most medical conditions, is not an either=or
condition. Degrees of ED are recognized in the
MMAS, where important distinctions are made
between Complete, Moderate, Minimal and No ED.
The MMAS often combines `Complete ED' and
`Moderate ED' into a single category. `Minimal ED'
is often considered a transitional state (in¯uenced
perhaps by mood or episodic depression) and
usefully grouped together with `No ED'.
Once reliable incidence rates were established, it
was possible to relate the presence of various
chronic disease conditions (eg, heart disease, diabetes, hypertension and depression) to the development of new cases of ED. Various risk factors and
lifestyles have been linked to the development of
chronic diseases. For example, elevated cholesterol,
smoking, lack of physical activity and hypertension
are thought to be precursors of chronic heart disease
(CHD)13 and this same report also suggested a strong
relationship between ED and cardiovascular disease. It was possible with the MMAS to determine
whether these risk behaviors (which are related to
major diseases) are also linked to the development
of ED, independent of the chronic diseases for
which they are precursors. This linkage marked a
distinct step to the right in Figure 4.
Elsewhere, we looked at the relationship of both
active and passive smoking to the development of
ED. The adjusted rate of incident ED was signi®cantly higher in cigarette smokers (24%) than in
nonsmokers (14%), with an adjusted odds ratio
(OR) ˆ 1.97 and 95% con®dence interval (CI) of
1.07 ± 3.63. Passive exposure to cigarette smoke (if
present both at home and at work) doubled the odds
of incident ED (adjusted OR ˆ 2.07, 95% CI 1.04 ±
4.13). Cigar smoking was also strongly predictive of
incident ED, with adjusted OR ˆ 2.45, 95% CI 1.9 ±
5.50. Former smokers were no more likely than nonsmokers to incur ED. These results are therefore
strongly suggestive that passive smoking has the
same effect on ED as active smoking. Among
smokers, each increase in passive exposure increased the likelihood of incident ED. Cigar smokers
had an equivalent increase in incident of ED
(OR ˆ 2.45). It is noteworthy that all three forms of
smoking (cigarettes, cigars and passive exposure)
had equivalent independent effects on ED.
Weight is a strong independent behavioral predictor of ED. The adjusted rate of incident ED in
men who were overweight (BMI > 30.0 kg=m2) at
baseline was 22%, signi®cantly higher than those
who were not overweight (12%), with adjusted
OR ˆ 1.96,95% CI 1.17 ± 3.28.
While these new results represent a desirable
move from cross-sectional to prospective analyses of
cohort data, they still remain observational. The
next step, representing a type of `quasi-experiment',
was to see if changes in the risk factors that are
known to be associated with ED also result in
changes in incident ED. Using MMAS data, Derby
and her colleagues14 recently examined the extent to
which changes in smoking, heaving drinking,
sedentary lifestyle and obesity are associated with
incident ED. Only physical activity (an increase
from the ®rst time point (T1) to the second time
point (T2)) seemed to be bene®cial, ie, those
becoming more active over time seemed to have
less ED, compared with those who did not change
their activity levels.
S9
Measuring ED in epidemiologic ®eld studies
Several well-known questionnaires are available for
the measurement of ED.15,16 These instruments
were developed in clinical settings, mainly for
self-selected atypical populations seeking treatment
for ED: they have limited use in ®eld studies with
population-based samples (not patients). Major
limitations of these questionnaires are: (a) they are
sometimes lengthy and increase respondent burden
(with multiple detailed queries on sensitive topics);
(b) if items are omitted or skipped (common in selfadministered questionnaires) it is often dif®cult to
calculate an overall score (resulting in missing
data); (c) wordy questionnaires present dif®culties
for poorly educated respondents and increase
translational misunderstandings when used in
cross-cultural studies. There is a need for a simple
appropriate method for validly measuring degrees
ED in population-based epidemiologic studies.
Asking one single question has been shown to
provide useful information in many other areas of
health care. We developed such a question for
population studies of ED and have carefully
assessed its metric properties against the two
established indices previously developed and validated in clinical settings.17 It should be noted that
the MMAS single question approach may be
inappropriate for clinical studies, which require
instruments that are sensitive enough to detect
changes with medical and surgical treatments, or
which must be speci®c enough to optimize case
®ndings where screening is followed by clinical
evaluations. A single, direct question approach
provides a practical means of assessing ED in
population-based studies. With further development it should be possible to include a single
question concerning ED in large national health
studies (eg, NHANES) to provide estimates of ED
for selected population groups and geographic
regions.
International Journal of Impotence Research
Epidemiology of erectile dysfunction
JB McKinlay
S10
Acknowledgements
Figure 5 Only a tiny proportion of US males seek treatment for
ED: results from a national survey; n ˆ 1384 (men ˆ 639). Source:
American Association of Retired Persons (AARP).18
Health services research and ED
Very little health services research has focused
speci®cally on ED. The willingness of older men
to: (a) actually seek medical care; and (b) receive
appropriate effective treatment once the health
system has been accessed, are now important public
health issues, especially with the availability of
apparently effective, well-tolerated user-friendly
oral preparations. We have recently considered
some of the health services research and policy
implications for many countries of the availability of
new treatments for ED.6
While the MMAS could not include questions
concerning health care utilization for ED, some new
data are available for the United States from a
national survey sample recently conducted for the
American Association of Retired Persons (AARP,
1999).18 The results are summarized in Figure 5 and
show that only about 10% of men report seeking or
receiving treatment for `sexual problems', and only
half of these (5% of all US males > 45 y of age) have
ever tried or are using sildena®l (ViagraTM).
Summary
Some results from the largest population-based
longitudinal study of male health and aging (the
ongoing MMAS) are presented. Thus far, the
information gathered provides the following
important information about ED in our society:
(i) ED is a prevalent medical condition;
(ii) it has behavioral precursors that are amenable
to change;
(iii) it has serious implications for quality of life;
(iv) despite the availability of new pharmacologic
interventions, the vast majority of aging men
(even those reporting complete or moderate ED)
do not seek medical care for the condition.
International Journal of Impotence Research
The Massachusetts Male Aging Study (MMAS) is
supported by grants from the National Institutes of
Health, NIDDK (DK 44995) and NIA (AG 04673). All
members of the MMAS Team made important
contributions to this paper: Andre Araujo, Isik
Aytack, Carol Derby, Henry Feldman Irwin Goldstein, Katherine Johannes, Kenneth Kleinman, Robert Krane, Christopher Longcope, and Elizabeth
Mohr. Lisa Marceau and Agnes Migliaccio assisted
with the development of the ®nal manuscript. For
information on the MMAS contact Dr John B
McKinlay ([email protected]).
References
1 Krane RJ, Goldstein I, de Tejada IS. Impotence. New Engl J
Med 1989; 321: 1648 ± 1659.
2 Feldman HA et al. Impotence and its medical and psychosocial correlates: results from the Massachusetts Male Aging
Study. J Urol 1994; 151: 54 ± 61.
3 McKinlay JB. Medical and psychological in¯uences on erectile
dysfunction Ð results from a community study. The 9th
Scienti®c Congress of the Kuwait Medical Association. Sociol
Epidemiol Biostat 1994; 353 ± 358.
4 McKinlay JB, Longcope C, Gray A. The questionable physiologic and epidemiologic basis for a male climacteric syndrome:
preliminary results from the Massachusetts Male Aging Study.
Maturitas 1989; 11: 103 ± 105.
5 Kleinman KP, Feldman HA, Johannes CB, McKinlay JB. A new
surrogate variable for erectile dysfunction status in the Massachusetts Male Aging Study. J Clin Epidemiol 2000; 53: 78 ± 78.
6 McKinlay JB et al. International differences in the epidemiology of male erectile dysfunction (MED). (Abstract.) Int J Impot
Res 1998; 10(Suppl 1): S42.
7 Aytac IA, McKinlay JB, Krane RJ. The likely worldwide increase
in erectile dysfunction between 1995 and 2025 and some
possible policy consequences. Br J Urol Int 1999; 84: 450 ± 456.
8 Wei M et al. Total cholesterol and high density lipoprotein
cholesterol as important predictors of erectile dysfunction. Am
J Epidemiol 1994; 140: 930 ± 937.
9 Johannes CB et al. Incidence of erectile dysfunction in men
ages 40 ± 69: longitudinal results from the Massachusetts Male
Aging Study. J Urol 2000; 163: 460 ± 463.
10 Kinsey AC, Pomeroy WB, Martin CE. Sexual Behavior in the
Human Male. WB Saunders: Philadelphia, 1948.
11 Masters WH, Johnson VE. Human Sexual Response. Little
Brown: Boston, MA, 1966.
12 NIH Consensus Development Panel on Impotence. JAMA
1993; 270: 83 ± 90.
13 Feldman HA et al. Erectile dysfunction and coronary risk:
prospective results from the MMAS. Prev Med 2000; 30: 328 ±
338.
14 Derby CA et al. Modi®able risk factors and incidence of erectile
dysfunction: implications for prevention. Urology (in press).
15 Rosen RC et al. The International Index of Erectile Function
(IIEF): a multidimensional scale for assessment of erectile
dysfunction. Urology 1997; 49: 822 ± 830.
16 O'Leary MP et al. A brief male sexual function inventory for
urology. Urology 1995; 46: 694 ± 706.
17 Derby CA et al. Measurement of erectile dysfunction in
population-based studies: the use of a single question selfassessment in the Massachusetts Males Aging Study. Int J
Impot Res 2000 (in press).
Epidemiology of erectile dysfunction
JB McKinlay
18 American Association of Retired Persons, Modern Maturity
Magazine, Washington DC, 1999.
Appendix
Open discussion following Dr McKinlay's
presentation
Dr Wilson: We convert your single question into,
`how many times out of 10 are you successful at
intercourse?'. If they answer `2 out of 10' or `8 out of
10' successes, those results can be correlated with
penile Doppler ®ndings and we can also predict
their responsiveness to Viagra. If they answer 4 or 5
out of 10, Viagra is questionable, and we have to use
injections; if they are never successful, we use an
implant because nothing else works.
Dr Sharlip: When you present this material to
epidemiological groups, do they have any criticisms
or point out any limitations of your methodology?
Dr McKinlay: Because this is a population study
and not a convenient sample of patients, one
problem is the response rate. Another problem is
the diurnal variation of the hormone levels.
Dr Sharlip: How reliable is the application of your
data to populations in other parts of the world, such
as Asia?
Dr McKinlay: The Massachusetts study is a very
small proportion of the general population and even
a random sample in Massachusetts under-represents
Hispanic and African American men. ED may be
higher in those populations because there are racial
and ethnic differences in diabetes, smoking, weight
and hypertension. But when we extrapolate to
worldwide estimates, we take the most conservative
projections and I am con®dent that our projections
of 300 million cases of ED by 2025 is a considerable
underestimate.
Dr Sharlip: There may be 300-plus million men in
the world who have erectile dysfunction, but how
many of them care?
Dr McKinlay: A lot of men think it's an aging
condition. Clearly, frequency of intercourse and
erections goes down with age but so do their
expectations. But, we believe that ED, independent
of every other risk factor, predicts subsequent heart
disease. So they shouldn't dismiss it, because it may
be the single most important biobehavioral marker
of the largest killer of aging men.
S11
Dr Wagner: What about depression, does that also
correlate with treated and untreated?
Dr McKinlay: Probably Ð yes, but to do a good
analysis on depression drugs was dif®cult because
there were so many different drugs and the numbers
were small for any single drug.
Dr Montague: Does physical activity reverse or
slow down the progression from a moderate=severe
to a minimal or moderate situation?
Dr McKinlay:
ED.
It reduces the probability of getting
Dr Melman: How did you de®ne physical activity?
It seemed from your data that men who exercised
and then stopped were worse off than men who
never exercised at all.
Dr Broderick: I also raise a technical point about
how you described to your interviewees what
200 kcal per day meant.
Dr McKinlay: The instrument used was a validated
one, the Pathenbagger Scale, which asks how much
they walked and how often they sweated and all of
these things. When you crunch the numbers you can
actually calculate kilocalories of energy expended.
Dr Broderick: It brings up a paradox though,
because that's certainly asking more than one
question. So the attempt to reduce the questions to
enhance epidemiologic data is wonderful, but even
within your own study, to ascertain the most
protective thing takes a trained interviewer, or many
questions.
International Journal of Impotence Research