Pincus HA, Tanielian TL, Marcus SC, Olfson M, Zarin DA, Thompson J, Zito JM. Prescribing Trends in Psychotropic MedicationsPrimary Care, Psychiatry, and Other Medical Specialties. JAMA. 1998;279(7):526–531. doi:10.1001/jama.279.7.526
From the Office of Research, American Psychiatric Association, Washington, DC (Drs Pincus and Zarin and Ms Tanielian); Western Psychiatric Institute and Clinic, University of Pittsburgh Medical Center, Pittsburgh, Pa (Mr Marcus); New York State Psychiatric Institute and the Department of Psychiatry, Columbia University, New York (Dr Olfson); Department of Psychiatry, University of Maryland at Baltimore (Dr Thompson); and the Department of Pharmacy, Practice and Science, University of Maryland at Baltimore (Dr Magno Zito). Dr Thompson is now with the Office of Education, American Psychiatric Association.
Context.— Psychotropic medications are widely prescribed, but how new classes
of psychotropic medications have affected prescribing patterns has not been
Objective.— To examine changes between 1985 and 1994 (data from 1993 and 1994 were
combined) in the prescribing patterns of psychotropic medications by office-based
primary care physicians, psychiatrists, and other medical specialists.
Design.— National estimates for the number of visits during which a physician
prescribed a psychotropic medication based on the National Ambulatory Medical
Care Surveys conducted in 1985, 1993, and 1994.
Setting.— Office-based physician practices in the United States.
Participants.— A systematically sampled group of office-based physicians.
Main Outcome Measures.— National estimates of visits that included a psychotropic medication.
Results.— The number of visits during which a psychotropic medication was prescribed
increased from 32.73 million to 45.64 million; the proportion of such visits,
as a proportion of all visits, increased from 5.1% to 6.5% (P≤.01). Antianxiety or hypnotic drug visits, previously the largest
category, decreased as a proportion of psychotropic drug visits (P≤.01) and are now surpassed by antidepressant visits. Visits for
depression increased from 10.99 million in 1988 to 20.43 million in 1993 and
1994 (P≤.01). Stimulant drug visits increased
from 0.57 million to 2.86 million (P≤.01). Although
visits for depression doubled for both primary care physicians and psychiatrists,
the proportion of visits for depression during which an antidepressant was
prescribed increased for psychiatrists but not for primary care physicians.
Conclusions.— The patterns of psychotropic medication use in outpatient medical practice
changed dramatically during the study period, especially in psychiatric practice.
PSYCHOTROPIC MEDICATIONS are among the most widely prescribed medications
in the United States. As a class, they represented 8.8% of the prescription
drug market in 1994,1 and their use has been
increasing in recent years. A number of studies have documented differences
over time and among physician groups in the use of psychotropic medications.2,3 In 1993, Olfson and Klerman4
reported a number of changes in the prescription patterns of psychotropic
medications among office-based physicians. Their analyses revealed that between
1980 and 1989 the total number of office visits that included the prescription
of a psychotropic drug remained relatively stable. They noted a decrease in
the proportion of these visits to primary care physicians and an increase
in the proportion of these visits to psychiatrists. However, recent years
have seen enormous changes in the health care system and in the availability
and applications of new and older psychotropic drugs.
Since the 1988 introduction of fluoxetine hydrochloride, the first of
a new class of antidepressant drugs termed selective serotonin
reuptake inhibitors (SSRIs), 4 more antidepressants have been brought
to the market. Moreover, there have been additional indications approved for
other marketed medications (eg, in 1990, alprazolam was approved for the treatment
of panic disorder). At the same time, the treatment research literature on
mental disorders has expanded considerably, providing evidence for the application
of medications in ways that are not formally approved for labeling by the
US Food and Drug Administration (FDA). Much of this information has been systematically
integrated into practice guidelines,5- 7
which also affect the selection of treatments. Furthermore, some states have
implemented regulatory approaches in an attempt to alter prescribing patterns;
one specific example would be the requirements for triplicate prescriptions
for benzodiazepines in New York State.8,9
Finally, the advent of managed care in medicine has affected not only prescribing
practices, through the use of protocols, formularies, etc, but also the relationships
among physician specialties and changes in patient access to specialists.10
In this article, we examine changes between 1985 and 1993 and 1994 in
the prescribing patterns of psychotropic medications by office-based physicians,
specifically psychiatrists, primary care physicians, and other medical specialists.
This article further highlights the changes in antidepressant prescription
patterns among these medical specialists, given the advent of the SSRIs during
the 10-year period examined. Our objective was to assess the differential
trends across physician specialities in the prescription of psychotropic medications
and determine what factors may have led to any observed differences.
The source of data for this report is the National Ambulatory Medical
Care Survey (NAMCS). The NAMCS, which is conducted annually by the National
Center for Health Statistics (NCHS), Hyattsville, Md, samples a nationally
representative group of visits to physicians in office-based practices. The
current report is based on results from the 1985 and the 1993 and 1994 NAMCS
reports. At the time of these analyses, the 1993 and 1994 NAMCS data were
the most recent NAMCS data available for public use. We used the 1985 data
as our baseline, since it was included in the earlier Olfson and Klerman4 study, and we were interested in measuring a decade
The NAMCS reports were conducted via a 3-stage sampling design. A probability
sample was drawn of practicing physicians within primary sampling units, and
a systematic random sample was then drawn of the patient visits to these physicians.
A 1-week period was sampled. Physicians expecting more than 10 visits per
day recorded visits based on a predetermined sampling interval.
Approximately 75000 visits were sampled in the 1985 survey, 36000 in
the 1993 survey, and 34000 in the 1994 survey. Between these implementation
periods, NCHS decreased its physician sample size.11- 13
Therefore, following NCHS recommendations, data from the 1993 and 1994 surveys
were combined to establish a larger base from which to derive annual estimates.
Attending physicians or their office staff completed a 1-page data form
for each patient visit. The form contained items such as the patient's age,
sex, diagnoses, medications, and reasons for visiting the physician. The listed
medications included new prescriptions; ordered, supplied, or administered
medications; and continuing medications with or without new orders. Only minor
modifications were made to the survey form between 1985 and 1993 and 1994.
In the present study, physicians were divided into the following 3 mutually
exclusive groups: psychiatrists (child psychiatry, psychiatry, and psychoanalysis),
primary care physicians (general practice, family practice, adolescent medicine,
pediatrics, geriatric medicine, and internal medicine), and other physicians.
The 1985 NAMCS surveyed 178 psychiatrists, 1053 primary care physicians, and
2873 other physicians. In the 1993 and 1994 NAMCS reports combined, these
numbers were 257, 847, and 3786, respectively. The response rates in 1985
were 74% for psychiatrists, 69% for primary care physicians, and 71% for all
others. In 1993 and 1994, the response rates were similar: 70%, 71%, and 72%,
Psychotropic drugs were classified according to the 1994 Physicians' Desk Reference14 and the Drug Evaluations Annual 1994.15
Antidepressants, antianxiety drugs and hypnotics, antipsychotics, psychostimulants,
and lithium are considered in Table 1.
A psychotropic drug visit is defined as a visit in which at least 1 psychotropic
drug was prescribed, ordered, supplied, administered, or continued; an antidepressant
visit is defined as a visit in which at least 1 antidepressant was prescribed
and so forth. If a particular visit included more than 1 type of psychotropic
medication, for example, an antidepressant and a stimulant, then it was classified
as both an antidepressant and a stimulant drug visit.
Some of the analyses involved aggregating patient visits into broad
categories by first, second, or third diagnosis. Antianxiety drugs and hypnotic
visits were classified as including a mental disorder diagnosis if any of
the listed International Classification of Diseases, Ninth
Revision, Clinical Modification16 diagnosis
codes fell within the range of 290 to 319, inclusive.
To examine the changes in prescribing practices for depression, some
of the analyses focused specifically on patient visits where an ICD-9 diagnosis code indicated 1 of the following disorders in the
first, second, or third diagnosis field: major depressive disorder; dysthymic
disorder; depressive disorders, not otherwise specified; bipolar II disorder;
or bipolar disorder, not otherwise specified (296.58, 296.82, 296.89, 300.40,
or 311). Visits for depression were further grouped by medical specialty and
psychotropic drug class.
One of the primary aims of the NAMCS is to provide national estimates
of the volume and content of office-based care. Because of the complex patient
visit sampling design, the NCHS weights each visit to mirror the US population.
The US Bureau of the Census population estimate for July 1 of the survey year
is used to compute the annual visit prevalence. The percentages reported in
the current report are based on the weighted estimates. Estimates for the
1993 and 1994 period represent the annualized mean of the 2 survey years.
The NCHS provides formulae for the 1985 and 1993 and 1994 NAMCS reports
to calculate SEs of the survey estimates in this complex sample.11- 13
These formulae were used to compute 99% confidence intervals around the survey
estimate and to accommodate for multiple comparisons.
As depicted in Table 2, between
the 2 study periods, overall visits to office-based physicians (including
visits by children) increased slightly. All psychotropic medication visits,
however, increased from 32.73 million to 45.64 million visits; as a proportion
of all visits, they increased from 5.1% to 6.5% (P≤.01).
Visits by females accounted for 67.1% of the psychotropic medication visits
in 1985 and 64.1% of these visits in 1993 and 1994. No significant differences
over time were associated with patient sex. The psychotropic medication visits
of children and adolescents (younger than 18 years) increased significantly
from 1.10 million in 1985 to 3.73 million visits in 1993 and 1994; as a proportion
of all psychotropic medication visits, they increased from 3.4% to 8.2%, respectively
(P≤.01). Psychotropic medication visits to psychiatrists
almost doubled, rising from 7.77 million to 15.09 million (P≤.01), while psychotropic drug visits to primary care and other
specialists had a relatively smaller increase. If SSRIs are excluded from
the total in 1993 and 1994, psychotropic medications remain virtually stable
as a proportion of visits across the 2 time periods (34.19 million visits
in 1993 and 1994 [4.9% of office visits]). The increase in psychotropic medication
visits by children and adolescents can be accounted for by the increase in
these visits to primary care physicians. As a proportion of visits to primary
care physicians, visits by children and adolescents rose from 3.3% in 1985
to 9.6% in 1993 and 1994 (from 0.59 million visits to 2.10 million visits).
Table 3 examines the distribution
of psychotropic drug visits by drug class and physician specialty. Antianxiety
and hypnotic drug visits, which had previously accounted for the most psychotropic
drug visits, decreased as a proportion of psychotropic drug visits (P≤.01) as well as in absolute numbers and were surpassed
by antidepressant drug visits. Primary care visits with antianxiety and hypnotic
drug prescriptions decreased from 12.09 million to 10.21 million visits. While
these changes demonstrate a declining trend in the number of antianxiety and
hypnotic drug visits, the changes were not statistically significant at the P ≤.01 level. Psychiatric visits with an antianxiety
and hypnotic drug prescription represented almost a quarter of all visits
to physicians for this drug class.
The large growth in antidepressant visits can be entirely accounted
for by the use of SSRIs. Antidepressant drug visits increased across all physician
classes. The distribution of those visits, however, changed in important ways.
In 1985, primary care physicians provided 47.5% of all antidepressant drug
visits, the most of all physician specialties. In 1993 and 1994, psychiatrists
provided almost 44% of all antidepressant drug visits, with primary care providing
41%. In addition, a larger proportion of antidepressant drug visits to psychiatrists
involved the use of SSRIs (50.4% [5.56/11.04]) than either primary care (40.5
%) or other physicians (32.9%).
The largest proportional increases occurred among stimulant drug visits,
with a 5-fold increase over this time period, from 1.5% to 5.1% of all psychotropic
drug visits (P≤.01). Overall visits that included
a stimulant medication increased from 0.57 million to 2.86 million during
the 10-year period (P≤.01). This increase is the
result of the significant rise in the number and proportion of stimulant visits
by children and adolescents (from 0.31 million to 2.41 million visits) (P≤.01). Stimulant visits to psychiatrists did not increase
significantly; however, stimulant visits to primary care physicians demonstrated
a 7-fold increase (P≤.01) and similarly a 10-fold
increase to other physicians (P≤.01). Again, these
significant increases can be accounted for by a 10-fold increase in the number
of psychotropic drug visits by children to primary care physicians (from 0.14
million to 1.50 million, P ≤.01) and a 15-fold
increase in the number of these visits to other physicians (from 0.04 million
to 0.61 million) (P≤.01).
Antipsychotic visits did not rise significantly; however, an increasing
proportion of these visits were provided by psychiatrists. Drug visits for
lithium did not rise significantly.
As indicated in Table 4,
the total number of visits for depression increased from 10.99 million in
1985 to 20.43 million in 1993 and 1994 (P≤.01),
with virtually no shift in the proportion of visits across the 3 categories
However, within each physician group there are important changes in
the pattern of psychopharmacological use. While there was approximately a
doubling of depression visits to primary care physicians, there was no significant
change in the proportion of visits in which a psychopharmacological agent
was used. The proportion of those depression visits associated with antidepressant
prescription was also rather stable at approximately 60%. Primary care physicians
reported very limited use of lithium or other mood stabilizers. Benzodiazepine
and antipsychotic use for depression visits showed no significant change as
a proportion of visits regardless of physician specialty.
Overall visits for depression to psychiatrists also doubled (P≤.01). As a proportion of all visits to psychiatrists, depression
visits increased from 35.8% to 52.6% (P≤.01).
The proportion of psychiatric visits for depression that included a prescription
of a psychopharmacological agent increased from 53.5% to 70.9% (P≤.01). In addition to the advent of SSRI use, there was also an
increase in the use of older antidepressants. The use of lithium and mood
stabilizing agents increased, although these agents represented a small proportion
of depression visits.
The "other physicians" group had no significant increase in depression
visits. Less than half of those visits included a psychopharmacological agent
and did not demonstrate significant changes in the distribution across medication
The availability of new SSRIs, beginning with fluoxetine in 1988, sertraline
in 1991, and paroxetine in 1992, has had an enormous impact on the prescription
of psychopharmacological agents. Virtually all the substantial increase in
psychotropic drug prescriptions can be accounted for by the use of these medications.
Interestingly, the impact of these medications differed across the 3 specialty
groups, with the SSRIs having the greatest impact on psychiatry. While data
demonstrate a substantial increase in primary care physicians' use of antidepressants
(the majority of which were SSRIs), the proportion of their depression visits
involving an antidepressant drug did not change.
The greatly expanded use of SSRIs by psychiatrists, as compared with
primary care and other physicians, however, is surprising; the lower adverse
effect profile and the simpler dosing patterns would seem to be attractive
to primary care physicians, increasing their confidence to prescribe these
medications. This might be explained by the fact that psychiatrists are closer
to the initial research literature, which tends to be published in specialty
journals and conducted in specialty populations. Psychiatrists, therefore,
may be more on the leading edge of new technological developments in depression
treatment. Thus, despite the fact that these data were collected 5 to 6 years
after the initial marketing of the SSRIs, it may be that a longer time frame
is necessary for a broader diffusion of new technologies to primary care.
There also may exist other barriers to the appropriate recognition and management
of depression in primary care, such as primary care physicians' interest and
training, societal stigma, and health plan reimbursement.17
Finally, it is possible that the prescribing practices of primary care physicians
appropriately reflect the fact that many "depressed" patients in primary care
settings may not have a depressive condition that warrants the use of a medication.
The largest proportional increase in the use of psychotropic medications
occurred within the stimulant drug class. This increase can be largely accounted
for by increases in the absolute number and proportion of stimulant visits
by children and adolescents. These visits are associated with the diagnosis
of attention deficit hyperactivity disorder18,19
and may provide further evidence of the dramatic increase in the recognition
and prevalence of this condition.20- 23
Significant increases occurred in the absolute number of stimulant drug visits
to primary care physicians and to other physicians (largely neurologists)
but not to psychiatrists. These data do not, however, shed light on whether
the expanded use of these medications is appropriate or on the reasons why
the expansion in use has been primarily in nonpsychiatric specialties.
Looking across the psychotropic drug categories, it is apparent that
there has been a dramatic change in the nature of the practice of psychiatry.
For example, in primary care and other physician specialties, there has been
a continuing decrease in benzodiazepine use,2- 4
with no substantial changes in the pattern of indications. In psychiatry,
however, there is a rising trend in the proportion of antianxiety drug visits.
The expansion of psychopharmacological research in psychiatry across the range
of mental disorders has provided data supporting the expanded indications
for benzodiazepines.24- 28
While alprazolam was approved by the FDA for an additional indication for
panic disorder in 1990, other indications have not been formally approved
for labeling by the FDA; they have nonetheless been incorporated in practice
guidelines,29,30 which have systematically
gathered and assessed relevant data and integrated them into treatment recommendations.
Other factors that may be affecting the pattern of psychotropic drug
use are changes in the structure of health care delivery and health care reimbursement.31 In many managed care organizations, the incentives
of a capitated system would encourage primary care physicians to maintain
responsibility for their patients with mental disorders, treat them efficiently
with medications, and not refer them to specialty care. At the same time,
managed care patients who are referred for specialty care may be more likely
to receive medication treatment by a psychiatrist as the sole form of treatment
or may be receiving psychotherapy from other providers.32
Unfortunately, the data do not permit us to examine the trends in psychotherapy
utilization, since the instrument for characterizing nonpharmacological treatments
changed between the 2 time periods. However, the duration of psychiatrist
visits for depression showed only a slight drop, from an average of 42.5 minutes
in 1985 to 39 minutes in 1993 and 1994.
Another important finding is the substantial growth in the number of
depression visits across all the physician groups. It is unclear, however,
whether this represents expanded case finding and improved treatment, as promoted
by depression awareness programs33 and as requested
by a recent consensus conference.34 While the
number of depression visits did increase significantly for primary care, they
represented only 2% of primary care visits in 1993 and 1994, far below the
prevalence of depression.35 Moreover, there
was no increase in the proportion of primary care depression visits that involved
an antidepressant prescription. For psychiatry, the large increase in depression
visits also reflects a substantial shift in the percentage of psychiatric
visits for depression, 35.8% to 52.6% (P≤.01),
and the percentage of depression visits associated with an antidepressant
medication. However, there was no substantial change in the overall number
of psychiatrist visits. Thus, it is unclear whether there is a real change
in psychiatrists' case load (ie, individuals who have read Listening to Prozac36 or related publicity
are seeking psychiatric care) or whether diagnostic patterns have changed
(ie, individuals who previously might have been given a nondepression diagnosis
are being given a depression diagnosis). While there has been no significant
change in the standard nomenclature (ie, the Diagnostic
and Statistical Manual of Mental Disorders, Fourth Edition37) for depression during this time, the availability
of effective, well-tolerated medications might encourage psychiatrists to
evaluate patients more carefully for depressive symptoms and recharacterize
their condition. However, without additional information (eg, whether the
physicians' diagnostic assessments were accurate), it is premature to draw
any conclusions about appropriateness or quality of care.
In addition, it is important to recognize a number of limitations in
the data from the NAMCS reports. First, they are limited to office-based settings:
they do not include visits in hospital settings, federal facilities, emergency
departments, or outpatient hospital-based clinics. Diagnoses are based on
physician report, without an independent objective assessment of their reliability
or validity. Second, since the office visit rather than the individual patient
is the unit of data collection, it is possible that more frequent users of
care are disproportionately sampled. Since a patient who visits a physician
10 times per year is 10 times more likely to be sampled, it is possible that
the NAMCS sample is weighted in the direction of persons with more chronic
illnesses. Therefore, any inferences regarding the number of patients associated
with these visits must be made cautiously. Despite these shortcomings, the
NAMCS provides the most extensive provider survey of office-based medical
For the past 10 years there has been a dramatic change in the patterns
of use of psychotropic medications in outpatient medical practice. Most significant
have been the changes in psychiatric practice, especially the greatly expanded
use of antidepressant and other medications. These changes reflect the availability
of new agents, the expanded, formally approved uses of marketed medications,
and the growth and application of the research literature in psychiatry supporting
unlabeled uses of medications. It is also likely that changes in reimbursement
have affected the patterns of prescribing, not only in psychiatry but also
in relationship to other specialty groups. However, much more information
is needed to tease apart the impact of these broader health policy shifts.
In addition, more information is needed to understand how these patterns relate
to other changes in the use of nonpharmacological interventions by primary
care physicians, psychiatrists, and other health and mental health providers.
It would be especially useful to have information on the clinical reasoning
behind the selection of these medications. While population-based studies
of physician practice patterns are a first step, new research strategies for
some of these issues will soon be available. For example, the American Psychiatric
Association has developed a national Practice Research Network that is similar
to the Ambulatory Sentinel Practice Network in family practice and the Pediatric
Research in Office Settings network. These in vivo research approaches will
be able to assess the impact of financial arrangements, dissemination of clinical
practice guidelines, and patient and provider characteristics on psychiatrists'
clinical decision making, treatment selection, and, ultimately, outcomes.