Download Simon-Wastila, Lina. 2000. "The Use of Abusable Prescription Drugs

JOURNAL OF WOMEN’S HEALTH & GENDER-BASED MEDICINE
Volume 9, Number 3, 2000
Mary Ann Liebert, Inc.
The Use of Abusable Prescription Drugs:
The Role of Gender
LINDA SIMONI-WASTILA, B.S.Pharm., Ph.D.
ABSTRACT
It is well documented that women face greater medical exposure to psychotropic drugs than
do men, but little research examines whether women also have increased use of prescription
drugs with abuse potential. The objectives were to examine gender differences in the use of
abusable prescription drugs and to assess how use varies by gender and if patterns of use
vary across therapeutic drug classes. With data from the 1987 National Medical Expenditures
Survey (NMES), logistic regression analysis is used to model the influence of gender and
other sociodemographic and diagnostic variables on the probability of drug use. Women are
48% more likely than men to use any abusable prescription drug, controlling for demographics, health status, economic status, and diagnosis. Additional analyses reveal that being
female is a statistically significant predictor of anxiolytic and narcotic analgesic use but not
of sedative-hypnotic or stimulant use. Marital status, age, urbanicity, employment status, and
having a regular source of care explain gender differences in the use of abusable prescription
drugs. Both healthcare and substance abuse treatment providers should be cognizant that
women may have greater exposure to these potent prescription medicines.
INTRODUCTION
I
that women face
greater medical exposure to psychotropic
drugs than do men.1–11 Yet little information exists on whether women also have greater use of
prescription drugs with abuse potential. Many
studies of psychotropic drugs fail to differentiate
between abusable (e.g., anxiolytic and sedativehypnotic) and nonabusable (e.g., antidepressant
and antipsychotic) therapeutic classes. Further,
psychotropic studies often do not include drugs,
such as the narcotic analgesics and central nervous system (CNS) stimulants. Thus, although
evidence documents gender differences in psychotropic drug use, little of consequence is availT HA S BEEN W ELL DO CU M EN TED
able on gender differences in the use of abusable
prescription drugs. Such analysis is important in
understanding if and how women and men differ in the use of abusable prescription drugs and
may highlight the need for different approaches
to prescription drug abuse detection, prevention,
and treatment.
Because of the scarcity of research examining
prescription drug abuse, it is not known which
factors influence such use. Although not ideal,
studies of psychotropic drug use, especially abusable psychotropic classes, may shed light on these
gaps. The literature suggests that gender differences in coping with and expressing anxiety and
distress,4,5,12,13 willingness to seek medical care,1
perceptions of illness,1 and physician prescribing
Schneider Institute for Health Policy, Brandeis University, Waltham, Massachusetts.
289
290
SIMONI-WASTILA
bias 3,10 contribute to women’s greater psychotropic use. A recent study affirmed gender
differences in the use of anxiolytics and found
that predisposing factors responsible for use,
such as health insurance status and diagnosis, differ by gender as well.6
It is important to note that gender, an important determinant of psychotropic drug use, may
not be the only significant factor. Other variables
predicting drug use include increasing age, socioeconomic status, psychological and social
well-being, mental and physical health, psychic
distress, and polydrug use.3– 5,8,9,11,14–16 Many of
these variables, however, are inextricably linked
with gender. For example, inappropriate drug
use in elderly women has been well documented.17–19 One study presents evidence suggesting that substance abuse, including abuse of
prescription drugs, in older women is a growing
problem, especially among impoverished and minority women.20 Other studies suggest that female gender and age together increase the chance
of receiving a prescription drug with abuse potential.18,21
The purpose of this study is to examine gender differences in the use of abusable prescription drugs. Specifically, this study elicits factors
associated with the use of any abusable drug, as
well as the use of four therapeutic drug categories
with abuse potential—the narcotic analgesics,
CNS stimulants, anxiolytics, and sedative-hypnotics. This study assesses whether female abusable prescription drug use equals male use, how
correlates of use vary by gender, and whether
these patterns are similar across therapeutic drug
classes. Findings from this study may inform researchers, policymakers, and providers about
gender-sensitive differences in the prevalence,
prevention, education, and treatment of prescription drug abuse.
MATERIALS AND METHODS
Data from this study come from the 1987 National Medical Expenditures Survey (NMES), a
data collection effort sponsored by the Agency for
Health Care Policy and Research (AHCPR). The
NMES, which uses a national stratified multistage area probability design, is a survey of the
healthcare utilization and expenditures of approximately 14,000 households and 38,446 noninstitutionalized civilian individuals.22 The sam-
ple used in this analysis consists of 22,460 adults
. 17 years of age. Of these, 12,392 respondents,
or 55.2%, are women.
The primary dependent variable in this analysis is the probability of whether an individual obtained any abusable prescription drug in 1987.
This dependent variable is composed of four therapeutic categories of interest—narcotic analgesics
(e.g., morphine, acetaminophen with codeine,
propoxyphene), CNS stimulants (e.g., amphetamine, methylphenidate), anxiolytic agents (e.g.,
diazepam, alprazolam, triazolam), and sedativehypnotics (e.g., secobarbital, phenobarbital, chloral hydrate). Each of these therapeutic categories
is analyzed as a separate dependent variable. The
drug categories represent prescription drugs
deemed to have abuse potential by the Drug Enforcement Agency (DEA).23 The system used for
processing prescribed medicine data was originally developed for another federal data collection effort and is described elsewhere.24
Selection of explanatory variables and covariates predicting abusable prescription drug use
was guided by earlier studies1–4,6,11 and from review literature.7,14,20 The independent variable of
interest is gender, with male as the reference category. Other demographic variables include race
(nonwhite is reference), age groups 18–24, 25–34,
and $ 65 (35–64 is reference), marital status (nonmarried is reference), geographic location (rural
is reference), and education (less than high school
education is reference). Socioeconomic covariates
include employment status (unemployed is reference), health insurance status (uninsured is reference), whether the individual has a regular
source of care (no regular source is reference), and
poor/fair health status (good/excellent health
status is reference).
In addition to sociodemographic variables, the
analysis controls for diagnoses, which are self-reported conditions subsequently coded into International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) codes. Up to four
diagnostic codes per medication are collected in
NMES. The diagnoses used in this analysis are selected based on their association with the use of
the four categories of abusable drugs, as documented by literature and medical compendia.
Thus, mental illness includes all self-reports of
depression, neurosis (primarily anxiety), psychalgia, psychoses, and other psychiatric conditions. Physical illness includes those conditions
associated with the use of the types of abusable
291
GENDER AND ABUSABLE PRESCRIPTION DRUGS
prescription drugs, such as cardiovascular disease, arthritis, back pain, cancer, insomnia, fatigue, and obesity. The reference category is all
remaining diagnoses.
Logistic regression analysis is used to estimate
the probability of abusable prescription drug use.
The probability of overall drug use is evaluated,
as are the use of the four classes of abusable prescription drugs. For those analyses in which gender is statistically significant, separate logistic regression equations for men and women are
estimated to elicit the factors that vary by gender.
As described, the explanatory variables incorporated in predicting overall drug use and drug
use within each therapeutic category include gender, age, race, urbanicity, marital status, educational attainment, employment status, availability of health insurance, regularity of healthcare,
health status, and mental and physical health status variables. For overall drug use, diagnoses are
coded into two categorical variables—any mental diagnosis and any physical diagnosis. In predicting the probability of any therapeutic drug
class use, specific diagnoses germane to the drugs
of interest are included. For example, diagnoses
in which narcotic analgesics are therapeutic
agents are incorporated into the model predicting the probability of narcotic analgesic use.
Thus, mental diagnoses of psychalgia and physical diagnoses of arthritis, back pain, and cancer
are reported in estimating the probability of any
narcotic analgesic use. Parallel models are estimated, without the inclusion of gender as an independent variable, in the probability of drug use
by each gender for those drug classes in which
gender is statistically significant.
In these analyses, full-model estimates are used
with all explanatory variables in order to demonstrate statistically significant and insignificant associations with the dependent variables. Odds ratios (OR) and 95% confidence intervals (CI) are
reported for all predictor variables in the model
that have a p value # 0.05. Goodness-of-fit is assessed using the chi2 -distributed log likelihood
test.25 SAS statistical software is used to analyze
the data.26
RESULTS
Bivariate analysis
males differ along a
1). Relative to men,
cantly older than 65
reveals that males and fenumber of variables (Table
women tend to be signifiyears of age (24.5% versus
T A BLE 1. B IV A R IA TE A N A LY SIS O F P ERC E N TA G E O F M A LES
A N D F EM A LES , W ITH D EM O G R A P H IC , S O C IO EC O N O M IC ,
C LIN IC A L , A N D D R U G U SE C H A RA C T ER ISTIC S (n 5 22,460)
Variable
Demographic variables
Age (years)
18–24 a
25–34
35–64 a
65 1 a
Whitea
Urban
Married a
High school graduate a
Socioeconomic status
Employeda
Insured a
Has regular source of care a
In poor/fair health a
Mental diagnoses
Any mental diagnosis a
Psychalgia a
Other psychiatric a
Depression a
Neurosisa
Psychosis
Physical Diagnosis
Any physical diagnosisa
Cardiovascular a
Arthritisa
Back pain a
Gastrointestinal
Cancer a
Drug use
Any prescription druga
Any abusable drug a
Any narcotic a
Any anxiolytica
Any sedative
Any stimulant
% of males % of females
(n 5 10,068) (n 5 12,392)
13.8
22.2
43.6
20.4
71.5
73.6
65.1
35.2
12.1
21.6
41.8
24.5
69.3
73.8
53.6
31.3
68.7
79.3
74.7
22.3
50.6
83.3
83.5
26.6
2.6
0.8
0.8
0.6
0.4
0.1
5.7
2.2
1.3
1.4
1.0
0.1
23.6
17.1
4.4
2.7
2.2
1.1
29.9
21.5
8.6
3.2
2.9
1.5
52.5
12.5
8.8
4.2
0.8
0.2
70.9
20.1
13.4
7.9
1.5
0.5
a Chi-square analysis shows gender difference statistically significant at the p # 0.05 level or better.
20.4%), nonwhite (30.7% versus 28.5%), single
(46.4% versus 34.9%) and less well educated
(68.7% versus 64.8%). In addition, women are significantly less likely than men to be employed
(50.6% versus 68.7%). Despite lower levels of employment, however, women are significantly
more likely than men to have a regular source of
care (83.5% versus 74.7%) and have health insurance (83.3% versus 79.3%). Although women appear to have better medical care access than men,
they report significantly poorer health (26.6% versus 22.3%). Women tend to report significantly
more mental and physical conditions of all types,
with the exception of psychosis and gastroin-
292
SIMONI-WASTILA
testinal conditions. Reflecting these diagnostic
differences, women use significantly more prescription drugs of all types than do men (70.9%
versus 52.5%), as well as those prescription drugs
with abuse potential (20.1% versus 12.5%). When
looking at abusable prescription drugs by class,
however, there are no statistically significant gender differences in the use of stimulants or sedatives.
Multivariate analysis shows that women are
more likely to use abusable prescription drugs
than are males (model log likelihood 5 2453.2, 14
degrees of freedom [df]; p 5 0.0001) (Table 2). The
probability of using any abusable prescription
drug is 48% greater (CI 5 1.37, 1.61) for women
than men, all else being constant. A number of
other variables are statistically significant predictors of abusable prescription drug use. Among demographic factors, those aged 25–34 are 31% more
likely (CI 5 1.18, 1.46) than those aged 35–64 to use
any abusable prescription drug. Those 65 and
older, however, are 81% (CI 5 0.73, 0.90) as likely
T A BLE 2. L O G ISTIC R EG RE SSIO N A N A LY SIS O F F A C TO RS
A SS O C IA TE D W ITH TH E PR O BA BILIT Y O F A N Y A BU SA BLE
PR ES C R IPT IO N D RU G U SE (n 5 22,460) a
Variable
Demographic variables
Female
Age (years) b
18–24
25–34
65 1
White
Urban
Married
High school graduate
Socioeconomic status
Employed
Insured
Has regular source of care
In poor/fair health
Diagnosis
Any mental diagnosis c
Any physical diagnosisd
OR (95% CI)
1.48 (1.37, 1.61)
0.98
1.31
0.81
1.47
0.92
1.03
1.06
(0.84,
(1.18,
(0.73,
(1.34,
(0.85,
(0.95,
(0.98,
1.21)
1.46)
0.90)
1.61)
1.00)
1.11)
1.16)
0.81
1.26
1.43
1.82
(0.74,
(1.12,
(1.27,
(1.66,
0.89)
1.42)
1.60)
1.99)
5.62 (4.88, 6.47)
2.84 (2.60, 3.10)
a
The model log likelihood test for the model is 2453.2
with 14 df (p 5 0.0001). OR and CI in bold are parameter
estimates significant at p # 0.05.
b
Age 35–64 is reference.
c Any mental diagnosis includes one or more of ICD-9
codes for the following: depression, neurosis, psychalgia,
and other psychiatric.
d Any physical diagnosis includes one or more ICD-9
codes for the following: cardiovascular, arthritis, back
pain, gastrointestinal, cancer, insomnia, fatigue, and
obesity.
as their middle-aged peers to use these drugs.
Whites are 47% more likely (CI 5 1.34, 1.61) than
nonwhites to use abusable prescription drugs.
Socioeconomic factors are also important predictors of any abusable prescription drug use. Employed people are 81% as likely (CI 5 0.74, 0.89)
as their unemployed peers to use any abusable
prescription drug. On the other hand, insured individuals (OR 5 1.26; CI 5 1.12, 1.42), those with
a regular source of care (OR 5 1.43; CI 5 1.27,
1.60), and those reporting poor/fair health (OR 5
1.82; CI 5 1.66, 1.99) all are significantly more
likely to report any abusable drug use.
Finally, those reporting any mental medical
condition are considerably more likely than those
without a mental illness diagnosis to report any
abusable prescription drug use (OR 5 5.26; CI 5
4.88, 6.47). Similarly, those reporting any of the
selected physical diagnoses are also more likely
to report any abusable drug use (OR 5 2.84; CI 5
2.60, 3.10). Variables not significant in predicting
any drug use are age 18–24, urbanicity, marital
status, and educational attainment.
Separate logistic regression equations are estimated to determine if gender differences persist
in the use of each therapeutic category. Analyses
reveal that gender is not a statistically significant
predictor of sedative-hypnotic or stimulant use
but does play a role in influencing the use of narcotic analgesics and anxiolytics. The findings for
these latter two therapeutic categories are reported in Table 3.
Logistic regression analysis shows that the
probability of receiving any narcotic analgesic is
41% greater for women than men (CI 5 1.28, 1.54)
(model log likelihood 5 1716.0, 16 df; p 5 0.0001).
In addition, being younger than 35 years of age
(age 18–24: OR 5 1.23; CI 5 1.05, 1.45; age 25–34:
OR 5 1.44; CI 5 1.28, 1.61), white (OR 5 1.39;
CI 5 1.25, 1.54), married (OR 5 1.14; CI 5 1.04,
1.26), having a regular source of care (OR 5 1.29;
CI 5 1.14, 1.47), and insured (OR 5 1.29; CI 5
1.13, 1.47) increase the likelihood of narcotic analgesic use. Conversely, being elderly (OR 5 0.79;
CI 5 0.70, 0.90) and employed (OR 5 0.85; CI 5
0.77, 0.95) decrease the likelihood of narcotic use.
Those in poor/fair health have a 69% greater
probability of using narcotic analgesics than do
their healthier colleagues (CI 5 1.53, 1.88). Mental and physical conditions that positively influence narcotic use include psychalgia (OR 5 4.96;
CI 5 3.96, 6.24), arthritis (OR 5 2.87; CI 5 2.51,
3.29), cancer (OR 5 3.22; CI 5 2.49, 4.17), and
293
GENDER AND ABUSABLE PRESCRIPTION DRUGS
T A BLE 3. L O G ISTIC R EG R ESSIO N A N AL YSIS
P R O BA BILIT Y O F A N Y N A R C O TIC A N A LG E SIC
F A C TO R S A SS O C IA TE D W ITH TH E
A N XIO LY TIC U SE (n 5 22,460) a
Any narcotic
OR (95% CI)
Any anxiolytic
OR (95% CI)
1.41 (1.28, 1.54)
1.51 (1.32, 1.73)
1.23
1.44
0.79
1.39
0.96
1.14
1.08
(1.05,
(1.28,
(0.70,
(1.25,
(0.87,
(1.04,
(0.98,
1.45)
1.61)
0.90)
1.54)
1.06)
1.26)
1.19)
0.20
0.52
0.95
1.62
1.03
1.03
0.91
(0.13,
(0.41,
(0.90,
(1.38,
(0.90,
(0.90,
(0.79,
0.32)
0.65)
1.23)
1.90)
1.18)
1.18)
1.06)
0.85
1.29
1.29
1.69
(0.77,
(1.13,
(1.14,
(1.53,
0.95)
1.47)
1.47)
1.88)
0.68
1.18
1.93
2.48
(0.58,
(0.95,
(1.51,
(2.17,
0.80)
1.48)
2.47)
2.85)
Variable
Demographic variables
Female
Age (years)b
18–24
25–34
65 1
White
Urban
Married
High school graduate
Socioeconomic status
Employed
Insured
Has regular source of care
In poor/fair health
Mental diagnosis
Psychalgia
Depression
Neurosis
Other psychiatric
Physical diagnosis
Cardiovascular
Arthritis
Back pain
Cancer
Insomnia
OF
AND
4.96 (3.96, 6.24)
NIc
NI
NI
3.49 (2.58, 4.71)
6.40 (4.69, 8.73)
37.55 (25.55, 55.18)
11.13 (8.21, 15.07)
NI
2.87 (2.51, 3.29)
4.77 (4.15, 5.65)
3.22 (2.49, 4.17)
NI
1.86 (1.62, 2.13)
NI
NI
2.19 (1.60, 2.99)
22.22 (16.09, 30.68)
a
The model log likelihood test for the narcotic model equals 1716.0 with 16 df
(p 5
0.0001); the anxiolytic model log likelihood test is 2709.5 with 19 df
(p 5 0.0001). OR and 95% CI in bold are parameter estimates significant at
p # 0.05.
b Age 35–64 is reference.
c
NI, variable not included in model.
back pain (OR 5 4.77; CI 5 4.15, 5.65). Nonsignificant factors predicting narcotic analgesic
use include urbanicity and educational attainment.
Controlling for other factors, logistic regression
analysis shows that being female rather than male
increases the likelihood of anxiolytic use by 51%
(CI 5 1.32, 1.73) (model log likelihood 5 2709.5,
19 df; p 5 0.0001). Other significant and positive
predictors of anxiolytic use include being white
(OR 5 1.62; CI 5 1.38, 1.90) and having a regular
source of care (OR 5 1.93; CI 5 1.51, 2.47). Being
in poor/fair health greatly increases the probability of anxiolytic use (OR 5 2.48; CI 5 2.17,
2.85), as do diagnoses of depression (OR 5 6.40,
CI 5 4.69, 8.73), neurosis (OR 5 37.55; CI 5 25.55,
55.18), psychalgia (OR 5 3.49; CI 5 2.58, 4.71),
other psychiatric conditions (OR 5 11.13; CI 5
8.21, 15.07), cardiovascular disease (OR 5 1.86;
CI 5 1.62, 2.13), cancer (OR 5 2.19; CI 5 1.60,
2.99), and insomnia (OR 5 22.22; CI 5 16.09,
30.68). Statistically significant negative influences
include all ages younger than 35 (age 18–24: OR 5
0.20; CI 5 0.13, 0.32; age 25–34: OR 5 0.52; CI 5
0.41, 0.65) and being employed (OR 5 0.68; CI 5
0.58, 0.80). Nonsignificant predictors of anxiolytic
use include age $ 65, urbanicity, marital status,
educational attainment, and insurance status.
To further elucidate how men and women differ in the use of abusable prescription drugs, separate logistic regression equations are estimated
for each therapeutic category in which gender is
a statistically significant explanatory variable. For
ease of interpretation, Table 4 compares the OR
and associated 95% CI only for those explanatory
variables that differ by gender in statistical significance or direction at p # 0.05.
Although many variables similarly predict any
294
T A BL E 4.
SIMONI-WASTILA
G EN D ER D IF FE RE N C ES
P R EDI C TIN G TH E PR O BA BILIT Y O F A N Y A BU SA BLE D R U G , N A R C O TIC ,
(95% CI) F O R S TA TISTIC A LL Y S IG N IF IC AN T V A RIA BLES a
IN
Any abusable drug
Variable
Male
Any narcotic
Female
Age 18–24
Age 651
Married
NS
Male
Female
1.50
(1.15, 1.96)
NS b
NS
Employed
NS
Regular Source of Care
NS
0.85
(0.74, 0.96)
1.38
(1.16, 1.63)
Male
0.72
(0.54, 0.96)
1.40
(1.06, 1.86)
Female
NS
NS
0.88
(0.79, 0.98)
reported only for parameter estimates significant at p #
NS, nonsignificant.
a OR
b
A N X IO LY TIC : OR
Any anxiolytic
1.39
(1.16, 1.66)
Urban
1.19
(1.02, 1.38)
NS
OR
abusable prescription drug use by both genders,
there are some notable differences. For example,
females who live in urban areas are 88% as likely
as their female rural peers to use any abusable
prescription drugs (CI 5 0.79, 0.98). Urbanicity,
however, is not influential in predicting abusable
prescription drug use by men. On the other hand,
married men are 19% more likely to use any abusable prescription drug than are single men (CI 5
1.02, 1.38), whereas marital status plays no
unique role in predicting use by women.
There are also significant gender differences in
narcotic analgesic use. For instance, married men
are 39% more likely (CI 5 1.16, 1.66) to use narcotics than are single men; once again, marital status plays no role in predicting narcotic use by
women. Employed women, on the other hand,
are 85% as likely to use narcotic analgesics as are
unemployed women (CI 5 0.74, 0.96), whereas
employment status has no bearing on male narcotic use. Finally, females with a regular source
of care are 38% more likely than females without regular care to receive any narcotic analgesic (CI 5 1.16, 1.63), although having a regular
source of care is not significant for males.
Married men are 40% more likely to engage in
anxiolytic use than are their single peers (CI 5
1.06, 1.86). In addition, men aged $ 65 years are
72% as likely (CI 5 0.54, 0.96) to use anxiolytics
as younger men. Neither marital status nor older
age influence females’ probability of anxiolytic
use.
0.05.
DISCUSSION
This study demonstrates that women are more
likely than men to use an abusable prescription
drug. Controlling for diagnosis, demographic
variables, health insurance, and health status, being female increases the odds of using any abusable prescription drug by 48% relative to being
male. However, women do not use more of all
types of abusable prescription drugs; rather,
women are more likely than men to use narcotics
and anxiolytics.
Gender is not the only significant variable predicting abusable prescription drug use. Not surprisingly, those who are employed are significantly less likely to use these potent drugs,
perhaps because they are healthier than their unemployed peers. Having financial access in the
form of health insurance and having a regular
source of care also increase the probability of
drug use. Those who reported poor or fair health,
a proxy for severity-of-illness, also are more likely
to use prescription drugs with abuse potential.
Although this analysis demonstrates that being
white (rather than nonwhite) increases the probability of any abusable drug use, as well as the
use of anxiolytics and narcotics, other studies
have found the influence of race to be questionable. Several studies have found that nonwhites
receive significantly fewer psychoactive drugs
than whites,6,8,11,16,27 but at least three studies
have demonstrated that nonwhites receive more
GENDER AND ABUSABLE PRESCRIPTION DRUGS
narcotic analgesics than their white peers.28–30
Further analysis of racial and ethnic differences
is needed to clarify this issue.
Of interest is the finding that age varies in significance and direction by therapeutic category.
Those aged 25–34 are significantly more likely
and those aged $ 65 are less likely to use any
abusable prescription drug relative to their middle-aged (34–65) peers. When looking at each
therapeutic category separately, however, this
pattern generally does not hold. Anxiolytic users
tend to be middle-aged (35–64), and narcotic
users tend to be , 35. The elderly across all drug
categories are significantly less likely to use abusable drugs relative to those aged 35–64. This latter finding is of considerable interest, as a number of researchers have suggested that elderly
people are particularly susceptible to exposure to
psychotropic drugs in general4,17– 20 and anxiolytics in particular.8,11 Other studies, however,
have shown either no such association between
psychotropic drug use and advancing age 6 or that
psychotropic drug use peaks at middle age and
then declines subsequently.31
Diagnoses, which also vary by gender, are
highly predictive of abusable drug use. This
analysis, however, does not necessarily suggest
that drug use parallels disease prevalence. For example, although the lifetime prevalence rate of
anxiety is more than two times higher in women
than men (1.8% versus 0.8%, respectively),32 this
study finds that women are only 51% more likely
than men to use an anxiolytic. Thus, unless physicians are systematically misdiagnosing physical
and mental conditions in men and women, differences in disease prevalence alone do not explain gender differences in the prescribing of
abusable prescription drugs.
The variables predicting abusable prescription
drug use are essentially the same for men and
women. There are some subtle gender differences, however, that require further explanation.
Most notably, marital status appears to make a
difference by gender, with married men more
likely to use any abusable prescription drug, narcotic analgesics and anxiolytics than unmarried
men, and with marital status having no real effect on women’s use of drugs. Whether married
men are under more stress or suffer more illness
than their single peers, or are encouraged by their
spouses to seek treatment for conditions that use
potentially abusable prescribed medicines, are
questions subject to further analysis.
295
It is likely that gender differences in the use of
abusable prescription drugs may lie in intervening paths. For example, women and men may differ in their propensity to identify and report conditions that may require treatment with an
abusable prescription drug. Men and women
may also differ in their propensity to seek care in
the first place. Several studies, however, confirm
that women’s higher psychotropic use rates remain after controlling for office visits.3,33
It is not clear what role the medical use of abusable prescription drugs play in their actual abuse
and dependency. Some researchers suggest that
opiate-naive people receiving narcotics for
chronic medical conditions rarely become addicted to these substances.34,35 There are no data
that prove or disprove that addiction to prescription drugs is primarily precipitated by prescriptions received in a medical context. Only one
study to date has described both medical and
nonmedical use of prescription drugs.36 However, the study did not attempt to examine the relationships between medical and nonmedical use
or the factors explaining such use.
The study has several limitations. For one,
physician specialty is unknown, and several studies have shown that physician specialty influences the amount and type of drugs prescribed.6,28,37,3 8 Further, the study does not
estimate patient access to physician care, which
other research suggests is an important determinant of prescription drug use.3,28 The gender of
the prescribing physician also is unknown; some
studies suggest that the interaction between
physician gender and patient gender may influence the amounts and types of treatments provided by physicians.39,40 In addition, the data are
over a decade old. Unfortunately, the 1987 NMES
is the only database available to examine personlevel prescription drug use,* and this analysis
represents the most current examination of gender differences in drug prescribing. Research using the National Ambulatory Medical Care Survey (NAMCS), a survey of office-based physician
visits conducted annually, has shown that office
visits resulting in any psychotropic prescription
increased from 32.7 million in 1985 to 45.6 million in 1994.41 This same study also showed that
*At the time of this analysis, the necessary data files of
the Medical Expenditures Panel Survey (MEPS) were not
yet publicly available. The MEPS uses more recent data
(1996) and is the latest version of the NMES series.
296
anxiolytic use had declined over time but was
only surpassed by the use of antidepressant medications. Findings from the NAMCS suggest that,
if anything, the use of any abusable prescription
drugs may be an even greater problem now, with
the possible exception of anxiolytic use. No studies, however, have examined the use of narcotic
analgesics over time.
This study is also limited by the use of respondent self-reported data, which may result in
underreporting. In particular, self-report of mental health diagnoses may be suspect because of
the stigma that is often associated with behavioral
conditions. However, underreporting is unlikely
to be a significant problem with the NMES as a
result of the care taken by NMES data collectors
in interviewer training, questionnaire design, adjustments for missing responses, and the use of
respondent diaries to prevent recall errors.42
Specifically, a random sample of diagnoses was
verified using claims and medical records, and
comparability was quite high.42 Prescribed medicine information was obtained directly from the
dispensing package.16
This study demonstrates that, controlling for diagnosis, sociodemographics, and other factors,
women are at increased risk for use of abusable
prescription drugs. Specifically, this study shows
that women may be at increased risk for the use of
two types of abusable prescribed drugs, narcotic
analgesics and anxiolytics. What this study does
not do is look at the actual misuse or abuse of prescription drugs. Instead, it examines the use of prescription drugs that have serious potential for
abuse. Further exploration is necessary; specifically, analysis of the gender role in the context of
prescription drug misuse and abuse should be conducted. Other research should examine the possibility that women with abuse tendencies choose
prescription drugs over the illicit drugs and alcohol favored by men. It is important to determine
whether medical exposures to potentially abusable
prescription drugs is a gateway to their actual
abuse, and if propensity to abuse varies by gender.
If medical exposure is a pathway to problem use
or abuse, detection, prevention, and intervention
approaches can be developed for use by physicians
and other health and medical professionals. Just as
important is research that examines whether gender differences in the use of abusable prescription
drugs are clinically or socially relevant and
whether men and women are receiving sufficient,
safe, and appropriate prescription drug care.
SIMONI-WASTILA
ACKNOWLEDGMENTS
Many thanks to Lois McNally for her careful
attention to detail in editing both text and tables.
I thank the National Institute on Drug Abuse (DA
09886-04) for their financial support of this work.
REFERENCES
1. Cafferata GL, Meyers SM. Pathways to psychotropic
drugs. Understanding the basis of gender differences.
Med Care 1990;28:285 .
2. Cottler LB, Robins LN. The prevalence and characteristics of psychoactive medication use in a general
population study. Psychopharmacol Bull 1983;19:746 .
3. Hohmann AA. Gender bias in psychotropic drug prescribing in primary care. Med Care 1989;27:478 .
4. Mellinger GD, Balter MB, Manheimer DI, et al. Psychic distress, life crisis, and use of psychotherapeutics. Arch Gen Psychiatry 1978;35:1045 .
5. Mellinger GD, Balter MB, Uhlenhuth EH. Prevalence
and correlates of the long-term regular use of anxiolytics. JAMA 1984;251:375 .
6. Simoni-Wastila LJ. Gender and psychotropic drug
use. Med Care 1998;36:88 .
7. Svarstad BL, Cleary PD, Mechanic D, Robers PA. Gender differences in the acquisition of prescribed drugs:
An epidemiological study. Med Care 1987;25:1089 .
8. Swartz M, Landerman R, George LK, et al. Benzodiazepine anti-anxiety agents: Prevalence and correlates
of use in a southern community. Am J Public Health
1991;81:592.
9. Takala J, Ryynanen O-P, Lehtovirta E, et al. The relationship between mental health and drug use. Acta
Psychiatri Scand 1993;88:256 .
10. Van der Waals F, Mohrs J, Foets M. Sex differences
among recipients of benzodiazepines in Dutch general practice. Br Med J 1993;307:363 .
11. Wells KB, Kamberg C, Brook R, et al. Health status,
sociodemographic factors, and the use of prescribed
psychotropic drugs. Med Care 1985;23:1295 .
12. Bush PJ, Osterweis M. Pathways to medicine use. J
Health Soc Behav 1978;19:179 .
13. Verbrugge LM. Gender and health: An update on hypotheses and evidence. J Health Soc Behav 1985;26 :
156.
14. Cooperstock R, Parnell P. Research on psychotropic
drug use: A review of findings and methods. Soc Sci
Med 1982;16:1179 .
15. Hallfors D. Factors Affecting Long-Term Use of Benzodiazepines Among the Elderly. Unpublished dissertation. The Florence Heller Graduate School for
Advanced Studies in Social Welfare, Brandeis University, October, 1992.
16. Olfson M, Pincus HA. Use of benzodiazepines in the
community. Arch Intern Med 1994;254:1235 .
17. Bernstein LR, Folkman S, Lazarus RS. Characterization of the use and misuse of medications by an elderly, ambulatory population. Med Care 1989;27:654 .
GENDER AND ABUSABLE PRESCRIPTION DRUGS
18. Jorgensen TM, Isacson DGL, Thorslund M. Prescription drug use among ambulatory elderly in a swedish
municipality. Ann Pharmacoepidemiol 1993;27:1120 .
19. Willcox SM, Himmelstein D, Woolhandler S. Inappropriate drug prescribing for the communitydwelling elderly. JAMA 1994;272:292 .
20. Szwabo PA. Substance abuse in older women. Clin
Geriatr Med 1993;9:197 .
21. Koch H, Campbell WG. Utilization of psychotropic
drugs in office-based ambulatory care. National Ambulatory Medical Care Survey, 1980 and 1981. NCHS
Advance data. Hyattesville, MD: National Center for
Health Statistics, 1983.
22. United States Department of Health and Human Services, Association for Health Services Research. National Medical Care Expenditures Survey, 1987:
Household Survey, Expenditures, Sources of Payment, and Population Data [Public Use Tape 18]. Volume 1: Questionnaires, Data Collection Methods, and
Other Attachments. Rockville, MD: USDHHS, AHCPR, 1993.
23. Fink JL, Marquardt KW, Simonsmeier LM, eds. Pharmacy Law Digest. St. Louis, MO: Facts and Comparisons Division, JB Lippincott Co, 1987.
24. Koch HK. The collection and processing of drug information: National Ambulatory Medical Care Survey. Vital and Health Statistics, Series 2, number 90,
DHHS Pub. No. (PHS)82-1364, 1980.
25. Ben-Akiva M, Lerman SR. Discrete choice analysis:
Theory and application to travel demand. Cambridge,
MA: MIT Press, 1991.
26. SAS Institute. SAS procedure guide, Version 6. 3rd
ed. Cary, NC: SAS Institute Inc., 1990.
27. Khandker RK, Simoni-Wastila L. Differences in prescription drug utilization and expenditures between
blacks and whites in the Georgia Medicaid population. Inquiry 1998;Spring:78.
28. Wastila LJ, Bishop C. The influence of multiple copy
prescription programs (MCPPs) on analgesic utilization. J Pharmaceut Care Pain Symptom Control 1996;
4:3.
29. Koch H. The management of chronic pain in officebased ambulatory care. National Ambulatory Medical Care Survey, 1981. NCHS Advance data No. 123.
DHHS Pub. No. (PHS) 86-1250. Hyattsville, MD: National Center for Health Statistics, 1986.
30. Koch H, Knapp DA. Utilization of controlled drugs
in office-based ambulatory care. National Ambulatory Medical Care Survey, 1985. NCHS Advance data
No. 177. DHHS Pub. No. (PHS) 89-1250. Hyattsville,
MD: National Center for Health Statistics, 1989.
297
31. Parry HJ, Balter M, Mellinger G, et al. National patterns of psychotherapeutic drug use. Arch Gen Psychiatry 1973;28:769 .
32. Robins LN, Regier DA, eds. Psychiatric disorders in
America. New York: Free Press, 1991.
33. Travis CB. Women and health psychology: Mental
health issues. Hillsdale, NJ: Lawrence Erlbaum Associates, 1988.
34. Porter J, Jick H. Addiction rare in patients treated with
narcotics. N Engl J Med 1980;302:133 .
35. Kanner RM, Foley KM. Patterns of narcotic drug use
in a cancer pain clinic. Ann NY Acad Sci 1981;362:161 .
36. Clayton RR, Voss HL, Robbins C, Skinner WF. Gender differences in drug use: An epidemiological perspective. In: Ray BA, Braude MC, eds. Women and
drugs: A new era for research. National Institute on
Drug Abuse Research Monograph Series No. 65. U.S.
Government Printing Office, 1986.
37. Beardsley RS, Gardocki GJ, Larson DB, et al. Prescribing of psychotropic medication by primary care
physicians and psychiatrists. Arch Gen Psychiatry
1988;45:1117.
38. Hohmann AA, Larson DB, Thompson JW, Beardsley
RS. Psychotropic medication prescription in U.S. am bulatory medical care. Ann Pharmacother 1991;25:85 .
39. Delgado A, Lopez-Fernandez LA, De Dios Luna J. Influence of the doctor’s gender in the satisfaction of the
users. Med Care 1993;31:795 .
40. Franks P, Clancy CM. Physician gender bias in clinical decisionmaking: Screening for cancer in primary
care. Med Care 1993;31:213 .
41. Pincus HA, Tanielian TL, Marcus SC, et al. Prescribing trends in psychotropic medicines: Primary care,
psychiatry, and other medical specialties. JAMA 1998;
279:526.
42. National Medical Expenditure Survey, 1987: Household survey, expenditures, sources of payment, and
population data [Public Use Tape 18]. Rockville,
MD: Agency for Health Care Policy and Research,
1993.
Address reprint requests to:
Linda Simoni-Wastila, B.S.Pharm., Ph.D.
Schneider Institute for Health Policy
MS 035
Brandeis University
415 South Street
Waltham, MA 02454-911 0