`of Multiple Psychotropic Medications
`
`Michael B. Nichol, Ph.D., Glen L. Stimmel, Pharm.D.,
`and Suk C. Lange, Pharm.D., M.P.H.
`
`Background: Recent studies have ques-
`tioned the appropriateness of some types of
`psychotropic medication prescribing, especially
`by general practitioners. The purposeofthis
`study is to investigate factors that predict
`prescribing of multiple psychotropic medica-
`tions, a class that may represent more
`complicated cases.
`Method:This study analyzed data from the
`1989 National Ambulatory Medical Care
`Survey (NAMCS). Multiple logistic regression
`methods were used to determine variables that
`predicted the provision or ordering of multiple
`psychotropic medications during a single office
`visit.
`
`Results: Patients who visited psychiatrists
`were six times more likely to receive psycho-
`tropics in combination than patients visiting
`general practitioners. Patients diagnosed as
`manic were four times more likely to receive
`multiple psychotropics, and those diagnosed as
`schizophrenic were three times morelikely.
`Patients visiting physicians in the Northeast
`and South were significantly less likely to
`receive psychotropics in combination than
`patients in the Midwest.
`Conclusion: Although general practice
`physicians contribute to the use of multiple
`psychotropic medications, patients visiting
`psychiatric specialists are much morelikely to
`be provided combination therapy.
`(J Clin Psychiatry 1995;56:60-66)
`
`
`
`Received April 6, 1993; accepted May |2, 1993. Fram the School
`of Pharmacy (Drs. Nichol, Stimmel, and Lange), the Andrus School of
`Gerontology (Dr. Nichol), the School of Medicine (Dr. Stimmel!), and
`the University Hospital (Dr. Lange), University of Southern California,
`Los Angeles.
`Portions of these data have been presented at the 120th annual
`meeting of the American Public Health Association, Washington, D.C.,
`November 1992.
`Reprint requests to: Michael B. Nichol, Ph.D., Department of
`Pharmaceutical Economics and Policy, USC School of Pharmacy,
`CHP-/40F, 1540 East Alcazar Street, Los Angeles, CA 90033.
`
`A number of health care researchers have noted
`
`that some psychotropic medication prescribing
`appears to be suboptimal, especially with regard to
`prescribing by primary care physicians.'* The limited
`data presently available indicate that prescribers rarely
`note appropriate psychiatric diagnoses in conjunction
`with psychotropic use* and that medications used
`within particular categories (such as antidepressants)
`may haveside effect profiles that negatively affect pa-
`tient outcome.' Several authors have noted that pri-
`mary care physicians represent a significant portion of
`the mental health network and prescribe the majority
`of psychotropic medications,”* although they may not
`have extensive training in the diagnosis and treatment
`of patients with mental diseases.
`Anarea of prescribing that has not been investi-
`gated is the simultaneous use of multiple psychotropic
`medications. By definition, patients requiring multiple
`psychotropic medications can be considered compli-
`cated, and the potential for inappropriate therapy is
`high, if only because of the increased likelihood of
`adverse effects and drug interactions. Several problem
`areas may develop as a result of multiple concomitant
`psychotropic use: duplicative effects (two or more
`medications from the sameclass), offsetting effects
`(two or more medications that have opposite action),
`and lack of therapeutic indication (two or more medi-
`cations in which there is inappropriate use). Multiple
`psychotropic prescribing with these characteristics has
`the potential to create significant negative patient out-
`comes.
`
`There is increasing interest in the special problems
`of patients with co-occurring disorders.*° The lack of
`attention to multiple psychotropic prescribing is un-
`derstandable since it occurs relatively infrequently and
`few data sources provide sufficient detail regarding
`drug use and diagnosis to determine inappropriate use
`categorically. However, concomitant use of psycho-
`tropics provides a distinct opportunity to identify phy-
`sician and patient characteristics that may account for
`this lack of attention, especially insofar as it might be
`possible to determine whether primary care physicians
`contribute disproportionately to problem prescribing.
`The purpose of this research is threefold: to identify
`
`J Clin Psychiatry 56:2, February 1995
`
`60
`
`1 of 7
`
`Alkermes, Ex. 1071
`
`1 of 7
`
`Alkermes, Ex. 1071
`
`
`
`Nichol et al.
`
`the prevalence of multiple psychotropic prescribing by
`drug category, to classify potential inappropriate or
`questionable prescribing of multiple psychotropics,
`and to determine which factors predict multiple psy-
`chotropic prescribing behavior.
`
`CONCEPTUAL BACKGROUND
`
`One model that has been used extensively to predict
`the use of health services is that proposed by
`Andersen’ and expanded by Andersen and Newman’in
`which health care decision making is hypothesized to
`be affected by four sets of variables: predisposing, en-
`abling, need, and health services characteristics, Pre-
`disposing characteristics include such issues as family
`composition, social structure, and health beliefs. The
`enabling category includes family and community re-
`sources. The need category incorporates concepts of
`il]ness and patient response. This model has been em-
`ployed to explain prescription and over-the-counter
`drug use”""' and a variety of other health services."
`The 1989 National Ambulatory Medical Care Sur-
`vey (NAMCS) provides data on each of these four di-
`mensions. For
`the purposes of this study,
`the
`predisposing factors included patient age, gender, and
`ethnicity. Patient age has been shownto be associated
`with increasing use of medications in a variety of stud-
`ies.'*'* Women have been shownto be morelikely to
`use prescriptions’ and are more likely to be prescribed
`psychotropic medications.'*!’ Several authors'*’? have
`included ethnicity variables as covariates in studies of
`psychotropic medication use, but their results were
`conflicting. Therefore, the impact of ethnicity could
`not be predicted.
`Enabling variables encompassed the specialty of the
`physician and the type of health insurance coverage
`used to pay for the patient visit. Physician specialty
`could be expected to be a good predictor of multiple
`psychotropic prescribing because more complicated
`mental health cases that may require mult?ple medica-
`tions are likely to be supervised by psychiatrists. Evi-
`dence from the 1985 NAMCS established that
`psychiatrists prescribed multiple psychotropics more
`frequently than physicians in other specialties.’ It can
`be expected that health maintenance organization
`(HMO) patients may receive less intensive care than
`those whose care is financed through fee-for-service
`insurance plans,”° although treatment outcome appears
`to be comparable.” Less intensive care may mean that
`patients are less likely to receive multiple prescrip-
`tions.
`Needvariables, for the purposes of this model, in-
`clude whether the patient presented a psychiatric com-
`plaint, diagnostic categories
`identified by the
`physician, and whether the appointment was a repeat
`
`visit for the same problem. Presence of a psychiatric
`complaint was used as an explanatory variable be-
`cause it represents a clear indication that patient per-
`ception corresponds to prescribed medication, even
`though the complaint may also be expressed in terms
`of a physical manifestation of illness not easily con-
`nected to mental diagnoses.” Diagnosis identified
`by the physician could be expected to be a strong pre-
`dictor of drug prescribing, except for the fact that ex-
`isting studies indicate that primary care physicians
`may not recognize psychiatric disorders* or may be re-
`luctant to identify mental health diagnoses.’ The fact
`that the physician had previously treated the patient
`for a similar condition may be correlated with the use
`of multiple psychotropics, since a second visit for the
`same diagnosis could imply that the condition may be
`chronic or complicated.
`Two variables were used to measure health system
`characteristics: region of the country and the provision
`of psychotherapy services. Extensive research has
`shownthat health care service utilization varies dra-
`matically throughout the United States.”* Wells et al.”
`used Health Insurance Experiment data to show that
`residents living in sites in western and eastern states
`were morelikely to visit mental health specialists than
`those in midwestern sites, and midwestern residents
`were more likely to visit a mental health specialist
`than their contemporaries in the South. In the multiva-
`riate analyses that follow, patients residing in
`midwestern states are the comparison group.
`The impact of psychotherapy on multiple psycho-
`tropic prescribing was unclear, a priori. Patients who
`receive psychotherapy could be expected to be more
`likely to suffer from complicated conditions in which
`multiple psychotropic medications may be used. On
`the other hand, these patients may be more involved
`in treatment and less likely to use multiple drugs, de-
`pending on the training of the caregiver and the diag-
`nosis,
`
`METHOD
`
`The NAMCSis conducted periodically by the Na-
`tional Center for Health Statistics (NCHS).*° The 1989
`survey included information regarding 38,384 patient
`visits representing estimates of 692 million visits to
`physicians throughout the United States. Seventy-four
`percent of the eligible physicians who were contacted
`(N = 1421) participated in the survey. Sample counties
`were stratified by size, region, and patient demo-
`graphic characteristics, while the physicians sampled
`within counties were stratified by specialty and prac-
`tice size.
`For the purposesof this study, all visits of patients
`18 years of age and older in which a psychotropic
`
`61
`
`2 of 7
`
`Alkermes, Ex. 1071
`
`J Clin Psychiatry 56:2, February 1995
`
`2 of 7
`
`Alkermes, Ex. 1071
`
`
`
`Multiple Psychotropic Prescribing
`
`
`Table 1. Total Psychotropic Medication Prescribing and Proportion Involving Multiple Psychotropics for Patients Older
`Than 18 Years of Age
`
`% Visits Within
`Category With
`Visits Involving
`Visits With Single
`% All Office
`Total Visits
`Category
`Multiple Prescribing
`Multiple Prescribing
`Drug in Category
`Visits
`Involving Drug
`Antidepressant
`29.1
`3,513,041
`8,568,899
`1.74
`12,081,940
`Stimulant
`47.6
`188,381*
`207,429
`-06
`395,810"
`Antipsychotic
`53.3
`1,356,147
`1,187,937
`37
`2,544,084
`Anorectic
`15.7
`157,738"
`846,021
`14
`1,003,759
`Anticholinergic
`25.2
`563,905
`1,676,834
`32
`2,240,739
`Barbiturate
`2.6
`15,961°
`598,924
`.09
`614,885
`Benzodiazepine
`22.0
`2,409,308
`8,521,322
`1.58
`10,930,630
`Anxiolytic
`13.7
`237,048*
`1,487,087
`25
`1,724,135
`Lithium
`71.4
`957,070
`383,123
`ly
`1,340,193
`Hypnotic
`2,701,615
`39
`1,768,610
`933,005
`34.5
`
`Total
`35,577,790
`5.13
`25,246,186
`10,331,604
`29.0
`
`“Unstable estimate, standard error > 30%.
`
`medication was prescribed were abstracted for analy-
`sis. Medications were categorized according to the
`American Society of Hospital Pharmacists’ drug clas-
`sification scheme.*’ Two medications (both combining
`amitriptyline and perphenazine) were included in this
`study as combination products. ICD-9-CM (Interna-
`tional Classification of Diseases, Ninth Revision,
`Clinica] Modification) diagnoses were compiled into
`the same categories used by Hohmannetal.' Physi-
`cian specialty was aggregated into three classes: gen-
`eral practice (general practice, family practice, and
`internal medicine), psychiatric specialty, and other (all
`other specialties).
`The survey method used by NCHSrequires the ap-
`plication of sampling weights to generate estimates of
`nationwide patient visits. In the descriptive tables that
`follow, weighted estimates of total visits and percent-
`ages are reported. The extrapolation of nationalvisits
`from a limited sample creates potential problems with
`regard to the confidence level of estimates for rare oc-
`currences. Estimates of national visits that have rela-
`tive standard errors exceeding 30% are considered
`unreliable by the NCHS,” andare noted.
`Multiple logistic regression methods were applied
`to the unweighted figures because parameter estimates
`and standard errors are not affected by sampling
`weights.** The categories of predictor variables were
`entered into the model sequentially, with the predis-
`posing variables entered first, followed by enabling,
`need, and health services categories. This “step-up”
`procedure provides an estimate of the relative contri-
`bution of each domain to the probability that the at-
`tending physician ordered or provided multiple
`psychotropics during the patient visit. McFadden’s
`pseudo-R?(similarto a log likelihood ratio) was used
`to compare the degree to which these categories im-
`prove predictive power.** Oddsratios and confidence
`intervals were calculated for significant explanatory
`variables using exponentiation.
`
`RESULTS
`
`Prevalence of Multiple Psychotropic Prescribing
`Based on the NAMCS, the NCHS estimated that a
`total of 692.7 million patient visits occurred during
`1989. As noted in Table 1, approximately 35.6 million
`patient visits involved at least one prescription for a
`psychotropic medication. These data show that certain
`categories of medications are more likely to involve
`multiple psychotropic prescribing. For example,
`lithium, antipsychotics, and stimulants are the most
`likely psychotropics to be used in combination with
`other psychotropics. It is important to note that medi-
`cations in these three categories are infrequently pre-
`scribed, as the total use of all three constituted less
`than 1% ofall office visits during 1989 (0.62%). The
`most frequent psychotropic medication combinations
`include antidepressants: benzodiazepines and antide-
`pressants (24.5% of all combined psychotropics), anti-
`psychotics and antidepressants (17.0%), and lithium
`and antidepressants (8.3%). Several other combina-
`tions are worthy of note, although the relatively small
`numberof visits yields unreliable estimates: hypnotics
`and antidepressants (6.1%), antipsychotics and lithium
`(5.1%), and antipsychotics and anticholinergics
`(5.0%). An estimated 402,000 patient visits during
`1989 involved the provision or ordering of two psy-
`chotropic medications within the sameclass.
`
`Potentially Problematic Prescribing
`Potentially problematic combinations could include
`duplicative therapy, offsetting therapy, or therapy in
`which there are no clinically accepted indications. Us-
`ing this definition, nearly 2.8 million (27%) of the
`10.3 million visits in which multiple psychotropic
`medications were prescribed occurred in combinations
`that could be considered therapeutically questionable.
`Among the many combinations of psychotropic
`drugs represented in this sample, three appear with
`
`J Clin Psychiatry 56:2, February 1995
`
`62
`
`3 of 7
`
`Alkermes, Ex. 1071
`
`3 of 7
`
`Alkermes, Ex. 1071
`
`
`
`Nichol etal.
`
`Table 2. Logit Estimates of the Use of Multiple Psychotropic Medications*
`
`Estimate
`
`fi Estimate
`
`B Estimate
`
`Estimate
`
`Odds
`Ratio
`
`Confidence
`Interval
`
`Variable
`Predisposing
`Age
`Sex (1 = female)
`Race (] = minority)
`Enabling
`Other physician specialty
`Psychiatric specialty
`HMO paymentsource
`Medicare payment source
`Medicaid payment source
`Need
`
`308=(3.03) 375° (4,32) 1.45 1.02, 2.07
`
`Psychiatric complaint
`241
`(1.57)
`322
`(2.76)
`Neurologic diagnosis
`337
`(.63)
`301
`(.47)
`Stress diagnosis
`1.06, 2.26
`1.54
`471°
`(6.11)
`434°
`(5,00)
`Depression diagnosis
`1.60, 5.62
`3.00
`912°
`(8.44)
`1.10°
`(11.72)
`Schizophrenia diagnosis
`2.20, 7.71
`4.14
`1.14°
`(13.54)
`1.42"
`(19.60)
`Mania diagnosis
`32 367=(1.44)(1.12)
`
`
`Other psychoses diagnosis
`58
`(1.74)
`17
`(2.95)
`Other psychiatric diagnosis
`.222
`(.63)
`32
`(1.29)
`Repeatvisit
`Health services
`Psychotherapy provided
`Northeastern region
`Southern region
`Western region
`Modelx”
`Pseudo-R?
`*Numbers in parentheses beside parameter estimates represent Wald chi-square estimates. The numberof visits in the sample equals 2085.
`"9 <.001, *p<.01, ‘p< .05.
`
`-.014° (16.29)
`-.019
`(.02)
`-.185
`(1.02)
`
`~.0006
`104
`~.25
`
`(01)
`(.54)
`(1.52)
`
`(.01)
`022
`2.06" (167.72)
`~.009
`(.001)
`~.012
` (.003)
`182
`(.62)
`
`—.0001 (.0007)
`.068
`(.22)
`-.322
`(2.23)
`
`147
`1.461"
`056
`~.007
`218
`
`(55)
`(52.36)
`(.04)
` (,001)
`(.87)
`
`-.000
`071
`-.305
`
` (,000)
`(.24)
`(1.74)
`
`(75)
`173
`1.844" (36.37)
`-.026
` (.009)
`043
`(.04)
`14
`(23)
`
`6.32
`
`3.47, 11.51
`
`
`
`
`
`17.56"
`.0099
`
`260.19"
`47
`
`285.36"
`161
`
`(2.83)
`(21.09)
`(7.10)
`(2.48)
`
`-.49
`-.95*
`~.46°
`-.33
`311.85"
`176
`
`39
`63
`
`.26, 58
`45, 89
`
`sufficient frequency to merit concern: antidepressant
`and antipsychotic combinations; multiple concomitant
`central nervous system (CNS) depressants (including
`barbiturates, benzodiazepines, anxiolytics, and hyp-
`notics); and CNS-stimulant/CNS-depressant combina-
`tions. Of the more than 12 million visits involving
`antidepressant drugs, almost ].3 million (10.7%) also
`received an antipsychotic. Approximately 65% ofthis
`combination was prescribed by psychiatrists and virtu-
`ally all of these were repeat visits for previously
`diagnosed problems. Psychotherapy services were pro-
`vided during 64.0% of these visits.
`A total of 639,580 visits involved multiple CNS
`depressants in 1989. Although this constituted less
`than 2% of al] psychotropic visits, these combinations
`are notable. For example, while only about 4.3% of
`patients receiving a benzodiazepine also received a
`hypnotic drug, such a combination is rarely necessary.
`Less than 30% of this use occurred during visits to
`psychiatric specialists.
`There was sufficient frequency of the combination
`of stimulant/anxiolytic, CNS depressant/stimulant, and
`anorectic/benzodiazepine to deserve mention, even
`though the low numberofvisits renders the estimates
`unreliable. It is notable that most of these combina-
`tions were prescribed by nonpsychiatrists, with the ex-
`ception of the CNS depressant/stimulant category.
`
`Fortunately, very few visits included CNS depressant/
`stimulant combinations (less than 94,000, or less than
`0.3%ofall visits involving psychotropics).
`Nearly 54% of the potentially problematic combi-
`nations were prescribed by psychiatric specialties,
`while 32.0% were prescribed by general practice spe-
`cialties. Approximately 58% of the questionable pre-
`scribing by psychiatrists was represented by the
`antidepressant/antipsychotic combination. Primary
`care physicians prescribed 47.0% of the combinations
`involving two or more medications within the same
`class (e.g., two antidepressants),
`
`Factors Predicting Multiple Use
`The multivariate analysis provides estimates of the
`factors that predict multiple psychotropic prescribing.
`The results in Table 2 show the strongest effects in
`the enabling, need, and health services categories.
`Variables within the predisposing category (age, sex,
`and race) did not predict the use of multiple psycho-
`tropic medications. The greatest
`increase in the
`pseudo-R?statistics occurred when the enabling char-
`acteristics were entered into the model.
`Enabling. Psychiatric specialty was the only en-
`abling variable to provide predictive power. As de-
`noted by the oddsratios, this variable was the single
`strongest predictor of multiple psychotropic use. Spe-
`
`J Clin Psychiatry 56:2, February 1995
`
`4 of 7
`
`Alkermes, Ex. 1071
`
`4 of 7
`
`Alkermes, Ex. 1071
`
`
`
`cifically, patients who visit a psychiatrist were greater
`than six times more likely to receive an order for mul-
`tiple psychotropic medications than patients who visit
`a general practice physician. Payment source did not
`prove to be a significant predictor of multiple use.
`Need. Several variables in the need category dem-
`onstrated predictive power, although the category as a
`whole made only a modest contribution to the model's
`overall predictive power(an increase of less than 2%
`in the pseudo-R?statistic). The diagnostic classes were
`the best predictors in this category. Visits in which the
`physician identified a mania diagnosis were four times
`more likely to result in multiple psychotropic orders
`than visits in which a non-psychiatric diagnosis was
`specified. Visits with schizophrenia diagnoses were
`three times more likely than visits with only non-psy-
`chiatric diagnoses to result in the ordering of multiple
`psychotropic medications. Visits including depression
`diagnoses were approximately one and a half times
`more likely to result in such orders.
`Health services. Regional differences provided the
`greatest predictive power in the health services cat-
`egory. As noted in Table 2, visits occurring in the
`northeastern and southern regions of the United States
`were much less likely to receive orders for multiple
`psychotropic medications than patients
`in the
`midwestern part of the country. Interestingly, visits in
`which psychotherapy was provided were not more
`likely to result in the prescribing of multiple psycho-
`tropics,
`
`DISCUSSION
`
`These results confirm previous studies that showed
`that psychiatric specialties are responsible for the ma-
`jority of the multiple psychotropic prescribing.' Thus,
`it appears that primary care physicians and other non-
`psychiatrists either refer patients who present compli-
`cated psychiatric profiles to specialists, may not
`recognize the more complex diagnostic issues, or sim-
`ply do not use multiple psychotropics to treat such pa-
`tients. The small proportion of multiple psychotropic
`medication visits to nonpsychiatrists provides prelimi-
`nary evidence that these physicians may recognize the
`difficulties associated with treating more complicated
`psychiatric cases.
`The relatively low incidence of multiple psycho-
`tropic prescribing within individual categories should
`be cause for some celebration, but there is cause for
`concern, as well. Nearly 40% of all multiple psycho-
`tropic prescribing reported in the NAMCS data oc-
`curred in categories with no readily apparentclinical
`justification. The results addressing the three poten-
`tially problematic categories (antidepressant and antip-
`sychotic; multiple concomitant CNS depressants; and
`
`Multiple Psychotropic Prescribing
`
`CNS depressant and stimulant combinations) deserve
`particular mention. These combinations are particu-
`larly worrisome becauseof the lack of clinical support
`for their frequent use.
`While there are several indications for the combi-
`nation of an antipsychotic drug with an antidepressant,
`these indications are infrequent. Schizophrenic pa-
`tients, who require chronic antipsychotic drug therapy,
`may at times experience major depressive episodes
`necessitating drug treatment.” Also, patients with a
`major depressive disorder may infrequently experi-
`ence psychotic symptoms.*” Psychotic depression must
`often be treated acutely with both an antidepressant
`drug and an antipsychotic drug, but there is no evi-
`dence to support a role for antipsychotic drug therapy
`in nonpsychotic depression.*! It is impossible in this
`study to determine whether the nearly 11% of patients
`receiving an antidepressant and antipsychotic drug
`meet one of these two conditions. However, the com-
`bination of antidepressant and antipsychotic is often
`used for either anxious nonpsychotic depression or for
`an inadequately diagnosed patient where drug therapy
`is being used as a diagnostic tool. Both situations are
`problematic. Use of an antipsychotic drug for anxiety
`or agitation that accompanies depressionis rarely ap-
`propriate when safer and moreeffective antianxiety
`drugs can be used. Use of an antipsychotic drug,
`merely for its sedative effect and the risk of tardive
`dyskinesia, in a patient without psychotic symptomsis
`of even greater concern than the lack of any docu-
`mented evidence of benefit. Diagnostic uncertainty
`also accounts for some of the use of combination psy-
`chotropic drugs. However, success or failure of com-
`bination therapy does not help clarify the diagnosis as
`would use of a single drugtrial.
`The primary concern regarding the combination of
`CNS depressants is unnecessary drug duplication. At
`best, the use of multiple CNS depressants represents
`unnecessary use of additional drugs. Selection of an
`antianxiety drug should include consideration of de-
`sired onset of effect, elimination half-life, and a dose
`and dosage schedule that matches the symptoms and
`needs of the patient. It is rarely necessary to prescribe
`one antianxiety drug for the daytime and another drug
`for sleep at night. It is also rarely necessary to use
`two benzodiazepines or two anxiolytic drugs together
`except during a transition from oneto another.” For
`some patients, however, multiple CNS depressants
`pose the risk of serious adverse effects such as exces-
`sive sedation, impaired psychomotor performance,
`ataxia, and possible falls.****
`The multivariate results provide strong evidence
`that multiple psychotropic use is related to psychiatric
`specialty. This fact, combined with the descriptive
`data showing that patients in the antidepressant/anti-
`
`J Clin Psychiatry 56:2, February 1995
`
`64
`
`5 of 7
`
`Alkermes, Ex. 1071
`
`5 of 7
`
`Alkermes, Ex. 1071
`
`
`
`Nicholet al.
`
`psychotic combination category generally receive
`psychotherapy and are uniformly repeatpatients, indi-
`cates that psychiatrists may be caring for more com-
`plicated patients who may require unusual therapeutic
`regimens. At the least, the multivariate results illus-
`trate that primary care physicians avoid the more
`complicated cases that require psychotropic combina-
`tions.
`The observation about the complexity of these
`cases is further reinforced by the contribution of need
`variables in predicting the use of multiple psycho-
`tropic medications, particularly with regard to the di-
`agnostic categories. Patients who received a diagnosis
`of either mania or schizophrenia were much more
`likely to receive multiple psychotropics. While bipolar
`patients more routinely receive multiple psycho-
`tropics, in schizophrenia it more likely reflects com-
`plexity. This result is consistent with the complexity
`hypothesis because both mania and schizophrenia are
`more likely to require treatment with multiple psycho-
`tropic medications.
`The regional differences apparent in the multivari-
`ate results deserve special mention. Patients in virtu-
`ally all other regions of the country wereless likely
`than those in the Midwest to be ordered or provided
`psychotropics in combination, although differences
`between the West and Midwest were notstatistically
`significant. These practice variations are not unusual,
`but the consistent direction of the effect indicates that
`practitioners in the Midwest may be less aware of the
`problems associated with prescribing multiple psycho-
`tropic drugs.
`The limitations of the NAMCS data mean that
`these results should be considered exploratory in na-
`ture. Although a large proportion of all psychotropic
`visits involved multiple drugs, these still represent a
`relatively smal! number of office visits across the na-
`tion. Thus, estimates derived from the NAMCS data
`are subject to significant variability, as evidenced by
`the high relative standard error for some categories.
`In addition to the fact that many of these combina-
`tions represent unstable estimates, the NAMCS data
`set suffers some important limitations that must be
`recognized. First, these data were collected at a single
`point in time, which can result in overrepresentation
`of frequent users.** Second, the data set does notin-
`clude aJ! information that would be useful to deter-~
`mine problems of multiple psychotropic prescribing.
`For example, it is impossible to determine whether the
`listed medications were added to an existing regimen
`prescribed by another physician, were substitutes for
`existing medications, or were only used for a brief
`time period. Third, physicians could only report a
`maximum of five medications. This limitation may be
`particularly important when analyzing visits to pri-
`
`mary care physicians, which may generate prescrip-
`tions for numerous conditions. Fourth, physician re-
`juctance to identify a patient with a mental health
`diagnosis may result in underreporting. Fifth, the sur-
`vey was completed by physicians or office staff with
`no verification of responses.
`Research conducted within the last decade supports
`the use of multiple psychotropic medications in some
`cases. However, future research into co-occurring
`mental disorders should pay particular attention to the
`use of multiple psychotropic medications and the
`problemsassociated with such use. Population-based
`research should documentclinical diagnoses, reasons
`for multiple psychotropic use, and the periodicity of
`multiple use (short- or long-term). Particular attention
`should be directed to determining the appropriateness
`of multiple psychotropic prescribing.
`
`Drug names: perphenazine/amitriptyline (Etrafon, Triavil).
`
`REFERENCES
`
`Le
`
`. Hohmann AA, Larson DB, Thompson JW, et al, Psychotropic medi-
`cation prescription in US ambulatory medical care. DICP, Ann
`Pharmacother 1991;25:85-89
`2. Beardsley RS, Gardocki GJ, Larson DB, etal. Prescribing of psycho-
`tropic medications by primary care physicians and psychiatrists. Arch
`Gen Psychiatry 1988:45:1117-1119
`. Eisenberg L. Treating depression and anxiety in primary care: closing
`the gap between knowledge and practice. N Eng] J Med 1992.326
`(16): 1080-1084
`4. Schurman RA, Kramer PD, Mitchell JB. The hidden mental health
`network: treatment of mental illness by nonpsychiatric physicians.
`Arch Gen Psychiatry 1985;42:89-94
`5. Barrett JE, Barrett JA, Oxman TE,et al. The prevalence of psychiat-
`ric disorders in a primary care practice. Arch Gen Psychiatry 1988:
`45:1 100-1 106
`6. Schulberg HC. Methodological problems in conducting research in
`co-occurring mental health disorders. Presented at the 9th annual
`meeting of the Association for Health Services Research; June 9,
`1992; Chicago,Ill
`7. Andersen R. A Behavioral Mode! of Families’ Use of Health Ser-
`vices. Chicago, Ill: University of Chicago; 1968. Center for Health
`Administration, Research Series 25
`8. Andersen R, Newman JF. Societal and individual determinants of medi-
`cal care utilization in the United States. Milbank Q 1973;51:95-124
`9. Bush PJ, Osterweis M. Pathways to medicine use. J Health Soc
`Behav [978;19;179-189
`10. Fleming GV, Giachello AL, Andersen RM, et al. Self care: substitute,
`supplement, or stimulus for formal medical care services’? Med Care
`1984:22:950-966
`it. Nichol MB, McCombs JS, Johnson KA, et al. The effects of consulta-
`lion on over-the-counter medication purchasing decisions. Med Care
`1992;30:989— 1003
`12. Sharpe TR, Smith MC, Barbre AR. Medicine use among the ruralel-
`derly. J Health Soc Behav 1985;26:1 13-127
`13. Stoller EP. Prescribed and over-the-counter medicine use by the am-
`bulatory elderly. Med Care 1988;26:1149-] 157
`14. Kasper JA, Rossiter LF, Wilson R. A Summary of Expenditures and
`Sources of Payment for Personal Health Services From the National
`Medical Care Expenditure Survey: Data Preview 24. Rockville, Md:
`National Center for Health Services Research and Health Care Tech-
`nology Assessment; 1987 (May)
`15. National Center for Health Statistics. Cost and Acquisition of Pre-
`scribed and Non-Prescribed Medicines, United States, June
`1964—June 1965. Publication no. 1000, series 10, no. 33. Washington,
`DC: National Center for Health Statistics; 1966
`
`65
`
`6 of 7
`
`Alkermes, Ex. 1071
`
`J Clin Psychiatry 56:2, February 1995
`
`6 of 7
`
`Alkermes, Ex. 1071
`
`
`
`Multiple Psychotropic Prescribing
`
`16. Hohmann AA. Genderbias in psychotropic drug prescribing in pri-
`mary care. Med Care 1989;27:478-490
`17. Baum C, Kennedy DL, Knapp DE, et al. Prescription drug use in
`1984 and changes over time. Med Care 1988;26:105-114
`18. Reid LD, Christensen DB, Stergachis A. Medical and psychosocial
`factors predictive of psychotropic drug use in elderly patients. Am J
`Public Health 1990;80(11):1349-1353
`19. Lin KM, Poland RE, Lesser IM. Ethnicity and psychopharmacology.
`Cult Med Psychiatry 1986;10(2): 151-165
`20. Norquist GS, Wells KB. How do HMOsreduce outpatient mental
`health care costs? Am J Psychiatry 1991;148:96-101
`21. Wells KB, Manning WG, Valdez RB. The effects of a prepaid group prac-
`tice on mental health outcomes. Health Serv Res 1990;25(4):615-625
`22. Lipowski ZJ. Somatization: the concept andits clinical application.
`Am J Psychiatry 1988;145:1358-1368
`23. Bridges K, Golberg D, EvansB, et al. Determinants of somatization
`in primary care. Psychol Med 1991;21:473-483
`24. Wennberg J, Gittelsohn A. Small area variations in health care deliv-
`ery. Science 1973;182:1102-1108
`25. Wells KB, Manning WG, Duan N,et al. Sociodemographic factors and
`the use of outpatient mental health services. Med Care 1986;24:75-85
`26. National Center for Health Statistics, 1989 NAMCS Micro-Data Tape Docu-
`mentation. Hyattsville, Md: Nationa! Center for Health Statistics; 1991
`27. McEvoy GK, ed. AHFS (American Hospital Formulary Service) Drug
`Information 92, Bethesda, Md: