Polypharmacy: When is it rational?


Journal of Practical Psychiatry and Behavioral Health, July 1995, 92-98

In this article, the authors discuss when it makes sense to consider using more than one medication to treat a single condition. They give a brief history of the use of polypharmacy in psychiatry and discuss how new discoveries in psychotropic drug development are making polypharmacy an increasingly important topic today. The authors then present a list of 10 criteria to guide the rational use of psychotropic polypharmacy and explain each in detail with examples drawn from clinical practice.
(J Pract Psychiatry Behav Health 1995;1:92-98)

KEY WORDS: polypharmacy, drug interactions, pharmacokinetics, pharmacodynamics

The early years of modern psychopharmacology (i.e., 1950-1970s) were marked by the chance discovery of effective medications. During this phase, the first successful drugs in every major class were discovered: mood stabilizers (e.g., lithium), antidepressants (e.g., amitriptyline and iproniazid), antipsychotics (e.g., chlorpromazine), and anxiolytics (e.g., diazepam). All these drugs had multiple mechanisms of action (e.g., amitriptyline and chlorpromazine) and/or fundamental effects with far-reaching consequences (e.g., lithium, iproniazid, and diazepam) and hence a narrow concentration range. Because of these factors, these early drugs had multiple pharmacodynamic interactions when used in combination with other psychiatric or nonpsychiatric medications. When these drugs were developed, the clinical practice of psychopharmacology was in its infancy and practitioners had a limited understanding of pharmacodynamic and pharmacokinetic principles. Drugs were therefore used together without regard to such principles. For example, patients might be treated with more than one benzodiazepine, each of which had the same mechanism of action. Multiple pharmacodynamic interactions might occur when, for example, a tertiary amine tricyclic antidepressant (TCA) (e.g., amitriptyline), a low-potency antipsychotic (e.g., thioridazine), and an anticholinergic (e.g., benztropine) were used together. Copharmacy at this time was not just irrational but also naive.

Our understanding of the clinically relevant pharmacodynamics and pharmacokinetics of psychotropic medications has developed substantially over the last 40 years. Nonetheless, the rational use of copharmacy is still hampered by a lack of systematic research. Most of the information available about a medication comes from the drug development program that leads to its registration and marketing. These programs almost invariably study the new drug in isolation versus either a placebo and/or a comparator agent. For obvious business reasons, the combined use of two drugs is virtually never studied. Experience with any new psychiatric medication in combination with another medication is limited to a few short-term drug-drug interaction studies that are usually conducted in carefully screened, healthy young volunteers. Federal research to study copharmacy has been lacking and has generally been done in a piecemeal fashion. Systematic data on the safe and efficacious use of copharmacy are therefore quite limited. Given this history, the knee-jerk reaction that copharmacy is bad is not surprising.

In this article, we propose some principles to guide the rational use of polypharmacy. By polypharmacy, we mean the intentional, concomitant use of more than two medications to treat either a patient with more than one pathophysiologically distinct illness or a patient with a single disorder. Copharmacy is a subset of polypharmacy that refers to the intentional, concomitant use of two medications. Clinicians use polypharmacy for several reasons:

  1. To treat two pathophysiologically distinct but comorbid illnesses in the same patient in contradistinction to treating the same condition or two 'comorbid" syndromes (e.g., major depression plus panic disorder) in the same patient
  2. To treat an adverse effect produced by the primary drug (e.g., adding an anticholinergic when a patient develops dystonia on a neuroleptic)
  3. To provide acute amelioration while awaiting the delayed effect of another medication (e.g., using lorazepam in acute mania while waiting for the anti-manic effects of lithium to exert themselves)
  4. To treat intervening phases of an illness (e.g., adding an antidepressant to a mood stabilizer when a bipolar patient develops a depressive episode)
  5. To boost or augment the efficacy of the primary treatment (e.g., combining a selective serotonin reuptake inhibitor (SSRI) and desipramine to treat a patient with major depression)

In this article, we focus on using polypharmacy to increase the efficacy of the primary treatment. Although this is a common reason for using more than one medication, it is the most difficult type of polypharmacy to defend with compelling empirical data.

The intent of copharmacy is to produce a drug-drug interaction that will have beneficial consequences for the patient. Generally, the goal is to produce a pharmacodynamic interaction in which the effect of one drug accentuates or diminishes the effect of another. Alternatively, the goal could also be to produce a pharmacokinetic interaction in which one drug alters the absorption, distribution, metabolism, or elimination of another.

Polypharmacy occasions greater concern than copharmacy because each drug that is added to the patient's regimen increases the likelihood of an adverse outcome and the expense of the treatment.1 For the sake of simplicity, we will focus on copharmacy in this article, but it is important to remember that the cautions we discuss here apply even more to polypharmacy. The reader should also be aware of our biases:

  • Mono-drug therapy: the ideal
  • Copharmacy: commonly needed
  • Triple pharmacy: may be necessary
  • Quadruple pharmacy: first consider that three drugs are not working

One of the most frequent questions we encounter when presenting the pharmacology of a new drug to a clinical audience is "Can you use it in combination with drugs X, Y, and Z?" underscoring the prominent role of polypharmacy in the clinical practice of psychiatry. However, polypharmacy is a strategy that physicians approach with ambivalence, having been taught to abhor it and yet feeling compelled to use it.

Ironically, the development of rationally designed psychopharmaceuticals may make the use of polypharmacy even more necessary than it has been in the past. One goal of rational drug development is to produce new drugs with limited numbers of mechanisms of action that will have a wider therapeutic index and be better tolerated (i.e., both fewer overall numbers and fewer types of adverse effects) while either maintaining or improving efficacy.2 However, because of their reduced range of central nervous system effects, such drugs may have more limited clinical applications as single agents. This fact, coupled with the reduced risk of pharmacodynamic interactions when combining drugs with fewer mechanisms of action, sets the stage for an increased use of combination treatment strategies.

In this article, we propose a way to evaluate whether the combined use of specific drugs can be well defended or is at least reasonable. We begin by discussing Parkinson's disease as a model for the use of rational copharmacy and then outline criteria that can serve as a basis for rational copharmacy with psychotropic medications.


It is rare to use a single drug to treat Parkinson's disease (Table 1). The cornerstone of treatment is a copharmacy product, L-dopa and carbidopa (Sinemet).3 The goal of treatment is to increase central dopamine activity. At least early in the course of the disease, this can be accomplished by supplying the substrate, L-dopa, which is then decarboxylated to dopamine. However, this reaction can occur in the periphery as well as centrally and, because dopamine cannot cross the blood-brain barrier, conversion in the periphery decreases the effective dose of L-dopa.4 Although increasing the dose of L-dopa can overcome this problem, it may also result in an increased incidence of peripheral adverse effects caused by excessive peripheral dopamine agonism. For this reason, carbidopa was added to L-dopa to inhibit dopa decarboxylase activity in the periphery and thus increase the bioavailability of the administered L-dopa to the brain.

TABLE 1. Parkinson's disease as a model of rational copharmacy
Treatment Effect
L-Dopa Increase synthesis of central dopamine (type: pk)
L-Dopa plus carbidopa (Sinemet) Inhibit peripheral decarboxylase to reduce the dose of L-dopa needed to increase synthesis of central dopamine (type: pk)
L-Dopa/carbidopa plus dopamine reuptake inhibitor (e.g., bupropion, amantadine) Potentiate the effect of released central dopamine (type: pk)
L-Dopa/carbidopa plus L-deprenyl Increase synthesis of central dopamine and block its degradation (type: pk)
L-Dopa/carbidopa plus bromocriptine Potentiate central dopamine agonism by addition of direct dopamine agonist (type: pd)
Type refers to type of interaction: pk = pharmacokinetic; pd = pharmacodynamic

Several other ways to rationally augment the central action of L-dopa are shown in Table 1.

It was possible to develop such rational copharmacy and even polypharmacy for Parkinson's disease because the pathophysiology of this condition is relatively simple and well understood. The dysfunction in Parkinson's disease involves a single neurotransmitter. The neuroanatomy and neurophysiology are well understood and can be readily studied and pharmacologically manipulated.5 Although such information is ideally needed to support rational copharmacy, we unfortunately do not know the pathophysiology underlying psychiatric conditions to anywhere near the extent that we understand the pathophysiology of Parkinson's disease and are therefore unable to support the use of copharmacy in psychiatric conditions so elegantly. Increasing our knowledge of the etiology and pathophysiology of psychiatric disorders to this level is a goal toward which we should strive.

The treatment of Parkinson's disease may also provide a model for understanding a frequently troubling and perplexing phenomenon: many clinicians report that antidepressants, particularly fluoxetine, seem to lose their effectiveness over time in a substantial number of patients. Although L-dopa can be a miracle drug early in the treatment of Parkinson's disease, it predictably loses its effectiveness during long-term treatment. The reason is based on the pharmacology of the drug and the nature of the illness: L-dopa temporarily ameliorates the pathophysiology of the condition but does not correct the pathoetiology that results in the loss of central dopamine neurons. As these neurons die, L-dopa can no longer be converted to dopamine and thus its loses its efficacy. It is reasonable to believe that antidepressants may work on the level of pathophysiology and not at the level of pathoetiology. Hence, at least in some patients, antidepressants may simply correct the pathophysiology of a condition that is pathoetiologically progressive. If so, such drugs will predictably lose their efficacy over time.


Criteria for rational copharmacy with psychotropic drugs are outlined in Table 2. We discuss each consideration in detail below.

1. Knowledge that the combination has a positive effect on the pathophysiology or pathoetiology of the disorder. Ultimately, this would be the most reliable and elegant type of information to support the use of copharmacy. It is the type of information that is already available for Parkinson's disease. Unfortunately, we do not yet have sufficient knowledge of the pathophysiology or pathoetiology of psychiatric disorders to guide copharmacy. However, items 2-12 in Table 2 describe other valid principles that can help clinicians decide when to consider using copharmacy.

TABLE 2. Criteria for rational copharmacy in psychiatry
  1. Knowledge that the combination has a positive effect on the pathophysiology or pathoetiology of the disorder.
  2. Convincing evidence that the combination is more effective, including more cost-effective, than monodrug therapy
  3. The combination should not pose significantly greater safety or tolerability risks than monotherapy.
    - Drugs should not have narrow therapeutic indices.
    - Drugs should not have poor tolerability profiles.
  4. Drugs should not interact both pharmacokinetically and pharmacodynamically.
  5. Drugs should have mechanisms of action that are likely to interact in a way that augments response.
  6. Drugs should have only one mechanism of action.
  7. Drugs should not have a broad-acting mechanism of action.
  8. Drugs should not have the same mechanism of action.
  9. Drugs should not have opposing mechanisms of action.
  10. Each drug should have simple metabolism.
  11. Each drug should have an intermediate half-life.
  12. Each drug should have linear pharmacokinetics.

2. Convincing evidence that the combination is more effective, including more cost-effective, than monodrug therapy. The next best basis for deciding to use copharmacy is data from formal studies comparing the efficacy and safety of different combination strategies in adequately powered and properly controlled studies. Unfortunately, the number of such studies is quite limited, and they have tended to be done piecemeal rather than being part of a systematic effort to establish an empirical basis for rational copharmacy when monodrug therapy clearly does not work. Generally, a single study or series of case reports is done by one research group at one site based on the group's interest in a particular combination strategy. Regardless of the outcome, follow-up studies are rarely done, and when such follow-up studies are done, they are either done by the same group, leaving the issue of generalizability in question, or done by another group in a different way with different results. For this reason, we have recommended that a network of clinical psychopharmacology sites patterned after the oncology multisite groups (e.g., the Southwest Oncology Group) be established to conduct such studies.6 Until data from this type of effort are available, clinicians trying to evaluate a particular form of copharmacy will frequently have to rely on small-scale, often open-label studies or on the opinions of "experts."

Since these first two criteria are rarely met, we most often rely on criteria based on pharmacological considerations (Table 2). Such considerations are analogous to those based on a greater knowledge of the pathophysiology of a disorder (see item 1 above). Nonetheless, an important unknown is the width of the leap from the mechanism of action of the drug to its impact on the pathophysiology of the illness. The fact that that leap is so narrow with Parkinson's disease is what makes copharmacy in that condition and the empirical studies that support it so compelling. However, we are at the stage of prephysiology with most psychiatric disorders and must deal with the resulting limitations.

3. The combination therapy should not pose significantly greater safety or tolerability risks than monotherapy. A clinician would obviously have to have a compelling reason to use two or more drugs in combination when each drug has significant safety and/or tolerability problems. This would be particularly true when each of the drugs is associated with the same types of problems (e.g., risk of agranulocytosis, seizures), because the risk in the combination may well have additive and possibly even potentiated results.1 This discussion should not be construed to mean that such a combination is absolutely contraindicated but that the evidence for its usefulness must be based on careful studies in groups of patients or carefully documented in the treatment of individuals in clinical practice (see Table 3 for guidelines for n = 1 studies).

4. The drugs should not interact both pharmacokinetically and pharmacodynamically. The logic behind copharmacy is to achieve a greater overall response by increasing efficacy and/or safety and tolerability through either a pharmacodynamic or a pharmacokinetic interaction. The treatment of Parkinson's disease provides examples of both kinds of drug interactions used intentionally and rationally. One of the four strategies listed in Table 1 is based on a planned pharmacodynamic interaction (i.e., L-dopa/carbidopa plus bromocriptine), and three are based on a pharmacokinetic interaction (i.e., the L-dopa and carbidopa combination itself, L-dopa/carbidopa plus dopamine reuptake inhibitor, and L-dopa/carbidopa plus L-deprenyl). However, a combination that involves both a pharmacodynamic and a pharmacokinetic interaction will be inherently more variable across patients and therefore less predictable. In this section, we first discuss the mechanisms involved in pharmacokinetic interactions, then those involved in pharmacodynamic interactions, and then consider some of the problems involved in using strategies that involve both types of mechanisms.

TABLE 3. Guidelines for using a combination based on the n = 1 trial in clinical practice
  1. Each drug has individually been given an adequate trial and has been found to be inadequately effective.
  2. The combination meets most of the criteria outlined in Table 2 and ideally has supporting data from the literature as to its efficacy, safety, and tolerability.
  3. The combination is found to be superior overall to either agent alone in terms of efficacy, safety, and tolerability.
  4. After a period of stabilization, a trial is made to taper one of the agents to test by dechallenge the continued need for combination therapy.

Pharmacokinetic Interactions

Pharmacokinetic drug interactions are those in which one drug potentiates or diminishes the action of the other drug by affecting its absorption from the site of administration, its disposition within the body, or its metabolism or excretion. Examples of such interactions include those that occur between some SSRIs (e.g., fluoxetine, paroxetine) and TCAs or between thiazide diuretics and lithium.

Pharmacokinetic interactions are based on the fact that one of the drugs has a pharmacodynamic effect on the pharmacokinetics of the other (i.e., the target) drug.2,7 The effect of this strategy is most often to alter the functional activity (either induction or inhibition) of the enzyme that mediates the biotransformation of the target drug as a necessary step in its elimination. Such was the case in the two strategies for polypharmacy in Parkinson's disease mentioned above (i.e., the L-dopa and carbidopa combination itself and L-dopa/carbidopa plus L-deprenyl). The goal may be either to block the formation of a metabolite, as happens when carbidopa is added to L-dopa, or to block the degradation of the desired substance to prolong its biological activity, as happens when L-deprenyl is added to L-dopa/carbidopa.

Although pharmacokinetic strategies are rarely used in psychiatry, an example would be using fluvoxamine to inhibit the enzyme P450 1A2, which mediates the conversion of chlorimipramine to desmethylchlorimipramine, with the rationale being that the demethylated metabolite is a much more potent inhibitor of the norepinephrine uptake pump than the serotonin uptake pump, whereas the converse is true for the parent drug.8 If the beneficial effects of chlorimipramine in obsessive-compulsive disorder are due to its ability to inhibit the serotonin uptake pump, then treatment with that drug might fail in a patient who extensively and rapidly converts it to the demethylated metabolite. We cite this interaction merely by way of example and not as a recommendation, since the same pharmacological goal should be achieved by simply using an SSRI that does not lose its selectivity by biotransformation to such a metabolite.

The problem with pharmacokinetic interactions is that the outcome is dependent on both the concentration of the inhibitor and the activity of that enzyme in the specific patient. There can be substantial variation between patients in such activity because of genetic or environmental influences such as exposure to inducers of the enzyme (e.g., smoking and P450 1A2).

Pharmacodynamic Interactions

In pharmacodynamic drug interactions, the effect of one drug potentiates or diminishes the effect of another drug while not affecting its metabolism or disposition (i.e., pharmacokinetics). For example, a sympathomimetic drug and an anticholinergic drug may additively cause dry mouth, or two sedating drugs (a benzodiazepine and trazodone) can produce additive sedation without affecting each other's pharmacokinetics. An antagonistic pharmacodynamic interaction might be seen with drugs that produce sedation and stimulation, as would occur when a sedating antidepressant is coadministered with a psychostimulant.

Pharmacodynamic interactions are based on the fact that one of the drugs alters the effect of another by affecting the same or a different mechanism of action. The combined use of an SSRI and pindolol to increase antidepressant efficacy is based on a planned pharmacodynamic interaction. The SSRI increases serotonin availability at various serotonin receptors including the presynaptic 5-HT1a (5-HT = serotonin) receptor sites. Increasing serotonin at this receptor will slow the firing rate of these neurons, initially decreasing the effectiveness of the SSRI. Pindolol, although primarily a beta-adrenergic blocker, can also block the 5-HT1a receptor. This action should augment the effects of the SSRI by initially blocking this feedback autoreceptor.9

Interactions Involving Both Pharmacokinetic and Pharmacodynamic Mechanisms

Examples of drugs that can produce both pharmacokinetic and pharmacodynamic interactions include some SSRIs (e.g., fluoxetine), which inhibit one or more P450 enzymes in addition to their intended effect on the serotonin uptake pump, and several anticonvulsants (e.g., carbamazepine), which induce one or more P450 enzymes in addition to their desired anticonvulsant action.10 In general, the use of such drugs as part of a copharmacy strategy should be avoided, because the outcome could be due to either a pharmacodynamic or pharmacokinetic interaction. In fact, those two interactions may have opposing effects on the benefit/risk ratio. For this reason, valproate would generally be preferred to carbamazepine as the first choice for a combination strategy for bipolar disorder unless there are compelling data to support the superiority of the alternative. The rationale is that valproate could add mood stabilization properties while being less likely to alter the pharmacokinetics of the other drug. In a similar way, sertraline would be preferred to fluoxetine in combination therapy, because its usually effective therapeutic dose would provide the serotonin uptake pump inhibition without causing a clinically significant effect on P450 enzymes such as 2D6.8,11

Fluoxetine is the most problematic of all the SSRIs to use in copharmacy for two reasons. First, it inhibits more than one P450 enzyme in addition to its desired effect of inhibition of the serotonin uptake pump.2 Second, the potential adverse consequences of this nonselectivity of action are further aggravated by the extended half-lives of both fluoxetine and norfluoxetine.8 Both of these molecules are active with regard to both the serotonin uptake pump and more than one P450 enzyme. The magnitude and the duration of their effects on these various mechanisms of action are dependent on the concentration and half-life, respectively, of each that is achieved on the dose being taken. Thus, the magnitude of these effects can increase for many weeks after the drug has been started or the dose increased and can similarly persist for many weeks after it has been discontinued or the dose reduced.8 Because fluoxetine follows nonlinear pharmacokinetics, the magnitude and the duration of these effects are increased in a nonlinear fashion with dose increases. Taken together, these factors make copharmacy with this drug particularly complicated.

5. Drugs have mechanisms of action that are likely to interact in a way that augments response. This consideration is relevant for pharmacodynamically based copharmacy strategies. Although having such information about the mechanisms of action of the drugs involved is not as ideal as knowing the impact of the combination on the pathophysiology of the illness, it is nonetheless substantially more rational than simply using trial and error in combination strategies.

One strategy that has been used in developing this type of copharmacy is based on considerations related to the basic neuropharmacology of the drugs involved. This approach involves extrapolating from basic science research on the mechanisms of action of the combination on a specific neurotransmitter system. This appealing approach has had promising results so far, leading to the development of new drugs with combined mechanisms of action as well as to new copharmacy strategies.2 It may be helpful to illustrate this type of strategy with some examples. One strategy for antidepressant treatment was to achieve more rapid production of beta-adrenergic receptor subsensitivity, because such desensitization has been temporally linked with the onset of antidepressant activity. This observation led researchers to try the approach of combining medications to achieve both norepinephrine and serotonin uptake inhibition, such as in combination treatment with desipramine and an SSRI.2 There is now a single agent, venlafaxine, that, at doses of 200 mg/day or more, results in clinically significant inhibition of both norepinephrine and serotonin uptake pumps. Another strategy is to try to enhance central serotonin agonism by combining drugs with complimentary actions on the release of serotonin. This approach led to the development of the following combination approaches: 1) serotonin uptake inhibition plus 5-HT1a receptor blockade (e.g., SSRI plus pindolol); 2) serotonin uptake inhibition plus 5-HT1a receptor agonism (e.g., SSRI plus buspirone); and 3) serotonin uptake inhibition plus serotonin releasing agent (e.g., SSRI plus lithium). The original combination used to test this last proposal was a TCA plus lithium, because it was studied before SSRIs were available. A strategy for treating psychotic disorders, such as schizophrenia, was to try to enhance central dopamine antagonism by an approach that combined dopamine 2 receptor blockade with dopamine depletion (i.e., a combination treatment using a neuroleptic plus reserpine).12

Another strategy that has been used in developing this type of copharmacy is based on extrapolating from the clinical pharmacology of each individual agent. Examples of this type of copharmacy strategy include: 1) combining two drugs with mood stabilization properties in a bipolar patient who has not had sufficient mood stabilization on either drug alone13,14; 2) combining an antipsychotic and an antidepressant in a patient with psychotic depression15,16; and 3) combining two antidepressants with different apparent mechanisms of action in a patient with a major depressive episode that is refractory to monodrug therapy (e.g., a combination of an SSRI plus desipramine [the combination of an SSRI plus desipramine has already been mentioned and is based on basic science considerations as well as this type of extrapolation from clinical psychopharmacology]; a combination of an SSRI plus bupropion [care should be taken when combining bupropion with fluoxetine due to case report data that fluoxetine can increase plasma levels of some of the metabolites of bupropion and may thus increase the risk of seizures]; a combination of a TCA with a monoamine oxidase inhibitor).2

The next four criteria in Table 2 (numbers 6-9) are based on pharmacodynamic principles.

6. Each drug should have only one mechanism of action. The more mechanisms of action that each drug in the combination has, the more likely that there will be an increase in either safety or tolerability problems and the more ways the drugs can interact pharmacodynamically.

7. Drugs should not have a broad-acting mechanism of action. A drug may have only one mechanism of action, but that action may have wide ranging effects on brain function due to its fundamental nature. An example would be monoamine oxidase inhibitors, which profoundly affect four different central neurotransmitters systems (i.e., dopamine, epinephrine, norepinephrine, and serotonin), and SSRIs, which affect all presynaptic serotonin terminals.

8. Drugs should not have the same mechanism of action. It would generally be more reasonable to simply increase the dose of one drug rather than to use two drugs with the same single mechanism of action. The main exception to this principle is when the goal is to take advantage of a difference in the pharmacokinetics of the two drugs to achieve a difference in the magnitude of the effect over a dosing interval. For example, in alcohol detoxification a patient could initially be treated with lorazepam, which has rapid absorption and is not dependent on hepatic bio- transformation for its elimination and rapid absorption, and could then be switched to clonazepam because of its long half-life, which can facilitate gradual subsequent discontinuation to avoid rebound symptoms.17

9. Drugs should not have opposing mechanisms of action. The rationale for avoiding drugs with opposing mechanisms should be obvious. There might be specific circumstances in which such combinations might be found to have benefits; however, the data supporting the efficacy of such combinations would have to be substantial to offset this general principle. For example, the first drug may have a broad effect on a neurotransmitter system (e.g., SSRIs), and the goal of adding the second drug with specific antagonistic properties is to make the combined effects more narrow. Although SSRls are generally thought of as being selective, their basic mechanism of action is blockade of the serotonin pump at all serotonin terminals. As a result, they promote serotonin actions at a wide number of postsynaptic receptors. Some of these actions produce desired effects, whereas others have undesired effects. For example, stimulation of 5-HT2A receptors may interfere with sleep architecture,18 a problem that can be addressed by adding trazodone (which has 5-HT2A blocking action) to a wide number of SSRIs. Stimulation of the 5-HT3 receptor in the brain and/or the gastrointestinal tract seems to be responsible for the gastrointestinal distress that can be produced by SSRIs. This can be addressed by adding cisapride early in treatment and then discontinuing it once receptor tolerance has developed.19

Criteria 10 through 12 are based on pharmacokinetic principles. Each principle is based on making the outcome more predictable within a patient and across patients.

10. Each drug should have simple metabolism. The rationale behind criterion 10 is that many drugs are transformed into metabolites that are biologically active, with activity that can vary substantially from the parent drug. Desmethylchlorimipramine is one example we have already discussed. Another is methylchlorpiperazine, a metabolite of trazodone that is a 5-HT2C agonist and has anxiogenic properties. The presence of these metabolites makes the outcome more variable because the effect of the drug is a function of the relative concentration of the metabolite to the parent drug. This ratio is dependent on the rate of biotransformation, which can vary substantially across individuals.

11. Each drug should have an intermediate half-life. The rationale for generally preferring drugs with an intermediate half-life is that the concentration of each drug will be reasonably stable over a dosing interval, but at the same time the relative concentration of each drug can still be adjusted within a reasonable time frame to achieve the desired magnitude of the combined effect. An intermediate half-life also means that washout can be accomplished within a reasonable time after drug discontinuation if safety or tolerability problems develop in a specific patient.

12. Each drug should have linear pharmacokinetics. If the drug has linear pharmacokinetics, then the magnitude of the effect produced by dose adjustment will also be more predictable. An exception to this guideline is when compliance is an issue. For example, in treating a patient with schizoaffective disorder with both an antipsychotic and an antidepressant, it would be ideal to first be able to test the efficacy, safety, and tolerability of the combination therapy with intermediate-lived formulations of the drugs and then switch to depot formulations. Unfortunately, both types of formulation are available for only a few drugs. If the patient had first been treated with oral haloperidol and sertraline, one possibility would be to switch to depot haloperidol and fluoxetine, which is essentially a depot drug, realizing that fluoxetine will also produce inhibition of more than one P450 enzyme, which will have consequences such as elevating the plasma levels of haloperidol. As mentioned previously, one or more of these combinations might conceivably be found in empirical studies to have benefits that outweigh the concerns encompassed by these principles.


Although there is an extensive literature on copharmacy, the quality of the reports varies widely. Many are single or multiple case reports without any controls. Some are controlled but may not be double-blind. The few that are properly controlled are typically underpowered and not of sufficient duration to convincingly establish the efficacy, safety, and tolerability of the copharmacy over monodrug therapy. In this article, we offer some principles to assist clinicians in evaluating such reports and translating them into clinical practice. Clearly, there is a compelling need for an ongoing means of testing whether specific copharmacy strategies can be supported by empirical data. This need will become even more pressing with the increase in rational drug development in psychiatry. The drugs that are being produced lend them- selves to more effective, safe, and better-tolerated forms of copharmacy in which the clinician will increasingly be able to tailor the treatment to fit the needs of a given patient.

Editor's Note: In this issue, Dr. Preskorn begins a column in which he will discuss new directions and technologies in psychiatry today.


  1. McInnes GT, Brodie MJ, Drug interactions that matter: a critical reappraisal. Drugs 1988;36:83-110
  2. Preskorn SH, Antidepressant drug selection: criteria and options. J Clin Psychiatry. 1994;55(suppl A):6-22, 23-24, 98-100
  3. Physicians’ Desk Reference, 49th Edition. Philadelphia: Medical Economics; 1995:959-63
  4. Jozefowicz RF, Berg M, Dimitsopulos T, et al, Current therapy. In: Joynt RJ. Clinical neurology. Vol 1. Philadelphia: JB Lippincott; 1993
  5. McDowell F, Cedarbaum JM, The extrapyramidal system and disorder of movement. In: Joynt RJ. Clinical neurology. Vol 3. Philadelphia: JB Lippincott; 1993
  6. Preskorn SH, The future and psychopharmacology: potentials and needs. Psychiatr Ann. 1990;20(suppl 11):625-633
  7. Callahan AM, Fava M, Rosenbaum JF, Drug interactions in psychopharmacology. Psychiatr Clin North Am 1993;16:647-71
  8. Hyttel J, Comparative pharmacology of selective serotonin reuptake inhibitors (SSRIs). Nord J Psychiatry. 1993;47 (suppl 30):5-12
  9. Artigas F, Perez V, Alvarez E, Pindolol induces a rapid improvement of depressed patients treated with serotonin reuptake inhibitors. Arch Gen Psychiatry 1994;51:248-51
  10. Gerner RH, Treatment of acute mania. Psychiatr Clin North Am 1993;16:443-60
  11. Preskorn SH, Magnus R, The clinical relevant pharmacology of serotonin selective reuptake inhibitors. Clin Pharmacokinet. In press
  12. Wolkowitz OW, Rational polypharmacy in schizophrenia. Ann Clin Psychiatry 1993;6:79-90
  13. Horst V;D, Preskorn SH, The role of benzodiazepines in the treatment of psychotic disorders. Psychiatric Ann 1993;23:317-24
  14. Reus VI, Rational polypharmacy in the treatment of mood disorders. Ann Clin Psychiatry 1993;5:91-100
  15. Spiker DG, Hanin I, Perel JM, Cofsky JE, Rossi AJ, Sorisio D, Pharmacological treatment of delusional depressives. Psychopharmacol Bull 1982;18:184-6
  16. Spiker DG, Perel JM, Hanin I, et al, The pharmacological treatment of delusional depression. Part II. J Clin Psychopharmacol 1986;6:339-42
  17. Janicak PG, Davis JM, Preskorn SH, Syd FJ Jr, Principles and Practices of Psychopharmacotherapy. Baltimore, Md: Williams and Wilkins; 1993
  18. Hendrickse W, Roffwarg H, Grannemann B, et al, The effects of fluoxetine on the polysomnogram of depressed outpatients: a pilot study. Neuropsychopharmacology 1994;10:85-91
  19. Bergeron R, Blier P, Cisapride for the treatment of nausea produced by selective serotonin reuptake inhibitors. Am J Psychiatry. 1994;151:1084-1086

Copyright and Disclaimer

©2010, Sheldon H. Preskorn, M.D.
site design by CyberKansas Technologies
Questions or Comments about the site?