Printed from:
Outpatient Management of Depression
10 - Use of Antidepressants With Other Medications

The preceding chapters have focused on the antidepressant alone; however, the majority of patients on an antidepressant seen in primary-care settings are on other medications as well (Table 10.1). In this chapter, the potential downside of polydrug therapy will be discussed as a cautionary note. In the next chapter, the upside of polydrug therapy will be discussed; specifically, from the perspective of adding other drugs to either treat tolerability problems or augment antidepressant efficacy.

This chapter then will focus on the caution that should be exercised when using polydrug therapy. It will discuss the potential for adverse outcome resulting from an unintended drug-drug interaction.

To understand the potential magnitude of this problem, consider the percentage of patients on an antidepressant in different practice settings who are on more than one other medication and thus at risk for having a drug-drug interaction. In both primary-care and general outpatient psychiatry, two thirds of patients on an antidepressant are on at least one other systemically taken, prescription medication. In fact, one third of all patients on an antidepressant are on three or more other systemically taken, prescription medications (Table 10.1).

TABLE 10.1 — Percentage of Patients on Antidepressants Having the Potential to Experience a Drug-Drug Interaction as a Function of Treatment Setting
Clinical Number of Patients Prescribed only an Antidepressant Prescribed at Least One Other Medication Prescribed Three or More Other Medications
Primary-care setting 2045 28% 72% 34%
Psychiatry clinic 224 29% 71% 30%
VA medical clinics 1076 7% 93% 68%
HIV clinic 66 2% 98% 77%
Abbreviations: VA, Veterans Administration; HIV, human immunodeficiency virus.
Other medications include a systemically taken, prescription drug from any therapeutic class. Does not include over-the-counter medications, topicals, or herbs.
Adapted from: Preskorn SH. J Prac Psych Behav Hlth. 1998;4:37-40.

These percentages are even higher in older and more medically ill populations. That fact has three consequences:

The importance of this issue is further underscored by a literature search that revealed that the annual number of publications on drug-drug interactions increased five-fold from 1970 through 1997. Potential reasons include:

Patients with clinical depression may be particularly at risk for being on polydrug therapy for the following reasons:


TABLE 10.2 — Prevalence of Major Depression in Specific Medically Ill Populations
Medical Illness Prevalence
Terminal solid tumors 25% to 38%
Stroke 27% to 35%
Renal disease 5% to 22%
Chronic pain 35% to > 50%
Epilepsy 20% to 30%
Parkinson’s disease 30% to 50%
Myocardial infarction 20%
Diabetes mellitus 10%
Data from: Evans D. Am Soc Clin Psychopharm Progress Notes. 1995;6:22-25; Robertson MM, Trimble MR. Epilepsia. 1983;24(suppl 2):S109-S116; and Series HG. J Psychosom Res. 1992;36:1-16.

For all of these reasons, the prescriber must consider the potential for a drug-drug interaction when selecting an antidepressant for a depressed patient. There are two ways an antidepressant can interact with a coprescribed drug:

Twenty years ago, clinicians were acutely aware of the risk of drug-drug interactions when prescribing antidepressants because their only options were tricyclic antidepressants (TCAs) and monoamine oxidase inhibitors (MAOIs). As outlined in Chapters 6 through 8, the pharmacology of these older antidepressants is such that they cause a number of different types of potentially serious pharmacodynamic drug-drug interactions when used in combination with other medications. For that reason, many practitioners twenty years ago were reluctant to use these antidepressants in elderly or medically ill patients.

In comparison to TCAs, most newer antidepressants have a much more limited number of pharmacologic actions. As explained in Chapter 6, that fact accounts for their better tolerability, wider therapeutic index, and a reduced risk of interacting pharmacodynamically with other coprescribed drugs. For these reasons, clinicians may have been lulled into a false sense of security. However, a number of the newer antidepressants can cause pharmacokinetic drug-drug interactions by virtue of their inhibition of drug-metabolizing cytochrome P450 (CYP) enzymes (Table 6.10).

Cytochrome P450 enzyme-mediated interactions are probably the most common type of pharmacokinetic drug-drug interaction. They are also the most easily misinterpreted type of drug-drug interaction because they can present in a myriad of ways, including:

These interactions typically cause an increase or a decrease in the concentration-dependent effects of the drug whose metabolism was altered. Hence, the prescriber may attribute the adverse outcome to a problem of "sensitivity" or "resistance" on the part of the patient.168,174

Because our understanding of the mechanism underlying such interactions (ie, the inhibition of CYP enzymes) has increased dramatically over the last decade and because of their clinical importance, the remainder of this chapter will be devoted to some actual case examples that demonstrate the problems CYP enzyme-mediated drug-drug interactions can cause.

Case Studies

This patient presented to his clinician with a recurrent episode of major depression.187 He was otherwise physically healthy based on medical history and physical examination. The patient stated he was in danger of losing his job because of poor performance resulting from his lack of energy and motivation.

His psychiatric history revealed that his depressive illness began in his early twenties. The current episode was his fourth and had begun approximately 3 months before he presented for care. According to the patient's medical history, he had responded to previous treatment with either a tertiary amine tricyclic antidepressant (TATCA) alone or a serotonin selective reuptake inhibitor (SSRI) alone, but the response had been incomplete and had taken considerable time.

Due to concerns about his job, the patient desperately wanted to feel better quickly. In response, his prescriber used an augmentation strategy; the combined use of a TCA and an SSRI (Chapter 11). He chose amitriptyline, 150 mg/day, and fluoxetine, 40 mg/day. The patient responded quickly with his only adverse effect being a mild dry mouth. However, several weeks later, the patient was found dead.

There was no evidence of foul play. An autopsy found no anatomical cause of death. The results of the analyses of postmortem blood samples revealed toxic levels of amitriptyline and its active metabolite, nortriptyline. The coroner signed this case out initially as a suicide.


Suicide is an understandable conclusion in this case for several reasons. The high levels of amitriptyline and its metabolites indicate a drug overdose. This patient had several risk factors for suicide (Table 4.1). He had recurrent clinical depression which leads to death by suicide in 15% of cases and had been under considerable psychosocial stress.

However, there were also several factors against this interpretation including the absence of a suicide note, no missing pills from his amitriptyline prescription bottle, and no pill fragments in his stomach.

The definitive answer about the cause of death came by examining the relative ratio of amitriptyline to nortriptyline in the gastric contents, blood, and tissue samples taken at autopsy. Amitriptyline is converted into nortriptyline by CYP 2D6-mediated oxidative drug metabolism. After taking amitriptyline on a regular basis for a week or more, an equilibrium (steady-state) is reached between the central compartment (the blood) and deep compartments in the tissue. Once steady-state is reached, the ratio of amitriptyline to nortriptyline is the same in the deep tissue compartment as it is in the blood.

Since gastric fluid is produced from the plasma, it reflects the ratio of drug in the plasma unless it is distorted by what is consumed. In an acute overdose, the amitriptyline in his stomach would dissolve into the gastric fluid substantially distorting the amitriptyline to nortriptyline ratio. In fact, patients typically die from an acute cardiac arrest before all of the amitriptyline is absorbed. For these reasons, in an acute amitriptyline overdose we would expect that the highest ratio of amitriptyline to nortriptyline would be in the stomach fluid, next highest in the blood, and lowest in the deep compartments. In this case, the ratios in these three compartments were the same, proving that this patient did not die from an acute overdose.

When these facts were presented to the coroner years later, the death certificate was corrected to show death occurred as a result of a chronic, unintentional overdose of amitriptyline consistent with fluoxetine-induced inhibition of CYP 2D6. In fact, the functional dose of amitriptyline in this case was closer to 900 mg/day than to the prescribed dose of 150 mg/day due to the fluoxetine-induced decrease in amitriptyline clearance.186

A dose of 900 mg/day would be expected to produce toxic levels in almost all patients. The time course from starting the combined therapy to the patient's death is due to the fact that it takes several weeks for fluoxetine to accumulate and sufficiently inhibit the metabolism of amitriptyline. That is the reason why the prescriber must consider the long half-life of fluoxetine when starting or stopping fluoxetine.

This patient was in a nursing home for cognitive impairment and delusions.136 The latter problem had been successfully treated with molindone, an intermediate potency antipsychotic medication, which she tolerated with only a mild, intermittent resting tremor. She then developed a depressive episode and was treated with the SSRI, paroxetine. Being aware of the issue of drug-drug interactions, her prescriber reduced the dose of molindone from 30 mg/day to 20 mg/day. Two weeks later, the patient had a prominent 3-cycle-per-second resting tremor, mask-like facies, cogwheel rigidity, and was unable to walk or care for herself.

The clinician attributed these problems to the worsening of underlying Parkinson's disease. As a result, he reduced the dose of molindone to 10 mg/day and added L-dopa/carbidopa and the anticholinergic drug, benztropine. Over the next several days, the patient developed a superimposed delirium. All of her medications were then discontinued and she gradually returned to her baseline status over the next 2 weeks.


The metabolism of molindone has not been characterized; however, this case suggests that CYP 2D6 plays a major role. If that is correct, then the addition of paroxetine 20 mg/day would be expected to substantially increase her molindone levels. In other words, her clinician would have functionally increased her dose of molindone conceivably well beyond 30 mg/day even though he thought he had reduced the dose to 20 mg/day.

An increased accumulation of molindone would produce a greater degree of dopamine receptor blockade in her brain which in turn would have caused the development of marked parkinsonism. The failure to consider this possibility led to the misdiagnosis and the addition of the antiparkinsonian medications which caused her delirium. While benztropine is another drug whose metabolism has not been characterized, some authors have postulated a CYP 2D6 component on the basis of cases of delirium occurring following the addition of fluoxetine or paroxetine to patients on stable doses of benztropine.11,211

As in the first case, this example illustrates how "not seeing" may simply mean "not recognizing." It also shows how the adverse consequence of a drug-drug interaction can be mistaken for another disease. The result can be both poor patient outcome and increased health-care costs.

This patient had grand mal epilepsy and a triad of behavioral problems (irritability, angry outbursts, and agitation) for which he had been treated for some time with phenytoin, 400 mg/day, and carbamazepine, 600 mg/day.224 On this regimen, he improved in terms of both seizure control and behavior. When he later presented with a major depressive episode, his prescriber added the SSRI, fluoxetine 20 mg/day. The depressive episode resolved. He also experienced a further reduction in the frequency of seizures and behavioral problems.

After 1 year of successful maintenance therapy, the fluoxetine was stopped since he had only one depressive episode. One month after discontinuation of the fluoxetine, his phenytoin level had fallen by 50% and was below the usually effective threshold for seizure control. His carbamazepine levels had also fallen, but only by approximately 15%.


Fortunately, a quality-control program detected the potential problem. The alerted clinician rechecked the patient's phenytoin and carbamazepine levels and increased their dose. If that had not occurred, this patient was at risk for both an increase in his seizure frequency and a worsening of his behavioral problems.

In this case, the problem was a loss of the inhibitory effect of fluoxetine on the clearance of the anticonvulsants. During fluoxetine coadministration, the levels of the anticonvulsants had risen which likely accounted for the better control of his seizure disorder and behavioral problems during that time. However, the prescriber did not understand that he needed to increase the doses of the anticonvulsants when he stopped fluoxetine if he wanted to maintain these higher levels.

This case demonstrates that loss of efficacy can result from a CYP enzyme-mediated drug-drug interaction. It also illustrates that a drug like fluoxetine can have different effects on two CYP enzymes. CYP 2C9/10 mediates the metabolism of phenytoin and is substantially inhibited by fluoxetine 20 mg/day, while CYP 3A mediates the metabolism of carbamazepine and is only mildly inhibited by fluoxetine 20 mg/day (Tables 6.9 and 6.10).

Had this patient experienced a seizure, the clinician might have erroneously concluded that the patient had not been compliant when he found low anticonvulsant plasma levels on repeat therapeutic drug monitoring.

This patient was a narcotic addict and had been successfully treated for several years with methadone.24 He then developed a depressive episode that was in turn successfully treated with the SSRI, fluvoxamine, 200 mg/day. After 3 months, the clinician decided to stop the SSRI, feeling treatment was adequate. Within several days, this patient experienced abdominal cramps, loose stools, sweating, tremulousness, chills, and a runny nose.


The patient was in narcotic withdrawal. The reason is that fluvoxamine produces substantial inhibition of CYP 1A2 which metabolizes methadone (Table 6.9). During the 3 months that this patient was on fluvoxamine, his methadone levels had risen as a result of decreased clearance and he had become dependent on the higher methadone levels. When the fluvoxamine was stopped, his methadone clearance increased and his methadone levels fell precipitating withdrawal symptoms.

Had the clinician not thought of this possibility, he might have thought the patient was drug seeking or that the symptoms were due to an intercurrent medical problem. The latter interpretation could have resulted in unnecessary medical testing while the patient suffered needlessly.


The information reviewed in these case studies should help demystify the issue of drug-drug interactions, particularly those mediated by CYP enzymes. The cases illustrate that CYP enzyme-mediated drug-drug interactions can present in myriad ways from sudden and life-threatening toxicity to more subtle but nevertheless clinically important problems. Such interactions can result in poor patient outcome and increased health-care costs, and frequently have the same consequence as would occur with a comparable change in the dose of the affected (or "victim") drug.

While this book is focused on antidepressant pharmacotherapy, the basic concepts covered in this chapter are relevant to all therapeutic classes of medications. By understanding the mechanisms underlying such interactions, clinicians can minimize the risk of such interactions by proper drug selection. Fortunately, there are a number of newer antidepressant options that generally do not produce clinically meaningful inhibition of drug metabolizing enzymes and hence are not at serious risk for causing such interactions (Table 6.10).