Risk and expense of unnecessary adjunctive medication (i.e., many patients may not require it).

  If adverse side effects are encountered (e.g., an allergic rash), it is unclear whether these effects were caused by the antidepressant or the adjunctive medication. Thus, both have to be discontinued, and clinicians are naturally reluctant to restart the antidepressant alone because it could in fact have caused the adverse event.

  If the adjunctive medication covers some of the criterion depressive symptoms by which the objectives of acute-phase treatment are gauged, practitioners will be less able to judge with certainty whether a full antidepressant response was obtained until the adjunctive medication is discontinued. This procedure may unnecessarily increase the number of visits or may delay appropriate revision in the treatment plan.

  Adjunctive medications may mask side effects that, if observed, would lead to either dose adjustments or switching treatments. In short, adjunctive medications may obscure information needed to make strategic decisions. For example, the automatic use of sedative-hypnotics in conjunction with fluoxetine to deal with ongoing insomnia that occurs in some patients may inappropriately delay a strategic decision to either decrease the dose or switch to an alternative agent. It is more logical to wait to determine whether the adjunctive medication is needed, rather than automatically prescribe it from the outset.

In addition to side effects, factors affecting medication choice include previous treatment responses, cross-sectional symptom features, patient preference, dosing convenience (which affects adherence), drug interactions (if patients are taking or will be taking other medications), current general medical conditions that may make one side-effect profile preferred over another, and family history of response. A patient’s treatment history is important because prior response typically predicts current response. In addition, a documented failure on a properly conducted trial of a particular antidepressant class (e.g., TCAs or MAOIs) points in the direction, given available data (see Thase and Rush 1995), of choosing an agent from an alternative class. Switching classes, for patients who do not respond to one class, appears to be associated with roughly a 50% response rate with the second drug class. For example, those that do not respond to a TCA have a 50/50 chance of responding to an MAOI. Similar numbers appear to apply to switching from a TCA to an SSRI, or vice versa. However, the likelihood that a patient will respond may depend as well on the symptom features that are present (see below). The types of cross-sectional symptom features often point to a preference of one drug class over another. When atypical symptom features are present, the TCAs appear to be less effective than the MAOIs. The SSRIs may also be more effective than the TCAs in such depressions, although findings from a formal randomized clinical trial to test this notion prospectively are not yet available. Melancholic features in more severely ill patients are associated with a positive response to ECT or TCAs (Rush and Weissenburger 1994). Patients with depression associated with many anxiety symptoms may do less well on TCAs than those without such features (Kupfer and Spiker 1981; Nelson et al. 1994; Roose et al. 1986). Some data (Danish University Antidepressant Group 1986, 1990), but not all, suggest that the TCAs are more effective than the SSRIs in severely depressed inpatients. The history of a first-degree relative’s having responded to a TCA or an MAOI is associated with a better response to the same class of agents in the patient (Stern et al. 1980). Whether family history of response acts as a predictor of response among the newer antidepressants has not been evaluated. Most antidepressants can be administered once a day except for bupropion sustained release and venlafaxine extended release, which are used twice a day. More frequent dosing is associated with poorer adherence. With regard to drug interactions, fluoxetine, paroxetine, and bupropion block the cytochrome P450 (CYP) 2D6 isoenzyme system and are therefore associated with elevations in blood levels of medications (e.g., TCAs, antipsychotics) metabolized solely by 2D6. Other agents (e.g., nefazodone) block other P450 systems (e.g., 3A/4 in this case). Information regarding drug interactions is now found in the Physicians’ Desk Reference (Medical Economics Company 1999). Tactical issues surrounding medication use include dosing steps, drug metabolism and pharmacokinetics, drug interactions, and side effects. The TCAs are typically started at lower doses and increased until the maximally tolerated dose is reached or, in the case of nortriptyline, until a therapeutic level is obtained. Gradual dose escalations are important to ensure adherence by reducing unacceptable initial side effects. Thus, TCAs are associated with a roughly once-a-week visit frequency for outpatients as doses are adjusted. When TCA blood-level monitoring is used, however, the total dose adjustment time can be reduced to 1 week (Nelson et al. 1987; Perry et al. 1984; Warner et al. 1993), with adjustments being made every 3-4 days. For the SSRIs, dosing is less complicated. Dose increments are fewer and the proper dose is more easily attained earlier due to better side-effect profiles (Mulrow et al. 1999). For some antidepressants, the starting dose is the therapeutic dose. Safety in overdose is an issue, especially early in treatment. Thus, agents with better safety in overdose are preferred. Evaluating outcome  Acute-phase treatment (medication, psychotherapy, the combination, or ECT) aims at symptom remission. Response without remission is associated with more functional impairment (Miller et al. 1998) and a worse prognosis. Thus, careful interviewing for all depressive symptoms at key visits is essential. Self-reports or clinician ratings can usefully augment this effort. A recent review of clinically useful scales is now available (American Psychiatric Association 2000). Declaring an unsatisfactory response to acute-phase medication  Failure with a medication in acute-phase treatment may occur because 1) even with a good clinical response, the side effects are not acceptable or an idiosyncratic adverse event may occur; or 2) the medication is tolerated but is ineffective. Side effects, in general, are often encountered within the first weeks of treatment, especially with dosage escalation or as medication levels rise to a steady-state level. Some side effects (e.g., sedation) are dose dependent and can usually be reduced by decreasing the dosage or slowing the rate of dosage escalation. Moderate side effects, when encountered, argue for holding the dosage and allowing time for physiological adaptation, which often results in fewer and less severe side effects. Some side effects (e.g., orthostatic hypotension) are less dose dependent, and tolerance to them does not develop. In these cases, gradual dose escalation is less useful, and a change in treatment is often indicated. Idiosyncratic or serious adverse side effects (e.g., seizures, allergic reactions), although rare, are most likely encountered within the first several weeks of treatment. Lack of efficacy is not uncommon with any medication, but it cannot be fully judged until patients have had several weeks of treatment at adequate dose levels (4-8 weeks). Thus, careful evaluation of symptoms at critical visits (e.g., weeks 4 and 8) during acute treatment, with or without a symptom rating scale, is a useful gauge of the adequacy of response. Psychotherapy It is useful to distinguish between a formal psychotherapy aimed at specific objectives and the general clinical management that is part of any medication treatment. General clinical management includes explanation of the diagnosis and treatment plan, the treatment objectives, and the anticipated treatment period; general counseling and management of both adherence and side effects; and a regular assessment of whether or not the treatment objectives are being met. General clinical management may involve consulting both the patient and significant others. When formal psychotherapy is used alone in the treatment of mood disorders, its objectives are identical to those for medication: symptom remission, psychosocial restoration, and the prevention of relapse or recurrence in the continuation and maintenance phases. When used in combination with medication, psychotherapies can also target other objectives: adherence to medication regimens, improvement of secondary psychosocial sequelae of the disorder (e.g., marital discord, occupational difficulties), or resolution of residual depressive symptoms (Rush 1986). Formal psychotherapy can also be used, over and above clinical management, to increase adherence. Short, multiple-visit psychotherapy packages increase adherence in mood disorder patients (see Cochran 1984). Persons for whom more formal adherence counseling may be needed include those with significant prior or current adherence difficulties and those with relatively fixed negative attitudes toward a treatment that is clearly indicated. Formal psychotherapies that ameliorate the psychosocial consequences of the disorder may include individual, family, couples, or occupational therapies. Some evidence suggests that such treatments, when used in combination with medication to control symptoms, result in improvement of the targeted area (e.g., marital counseling improves marriages) (for a review, see Paykel 1995). Frank and co-workers (1990) (in adults) and Reynolds and colleagues (1999) (in the elderly) found that maintenance-phase interpersonal psychotherapy (IPT) (Klerman et al. 1984), even in patients with highly recurrent forms of major depressive disorder, can delay recurrences, such that psychotherapy alone may usefully provide a medication-free interlude in a medication maintenance regimen for selected patients (e.g., those planning to become pregnant or those needing surgery). Choosing among the psychotherapies  Substantial evidence indicates that depression-targeted psychotherapies produce better symptom response than is observed in waiting-list control subjects (Conte et al. 1986; Depression Guideline Panel 1993; Dobson 1989; D. G. Robinson and Spiker 1985; L. A. Robinson et al. 1990; Rush and Thase 1999; Wexler and Cicchetti 1992; Wexler and Nelson 1993). Psychotherapy trials for depression typically admit somewhat less severely ill patients than do medication trials. Psychotherapy alone as a maintenance treatment has efficacy (Frank et al. 1990), but medications appear to be most effective at prolonging the well interval. When the need for maintenance treatment is anticipated, medication treatment (alone or in combination with psychotherapy) should be considered an essential part of acute-phase treatment. To date, the best-studied psychotherapies are the time-limited, depression-targeted reeducative approaches such as IPT, cognitive therapy, and behavior therapy. A meta-analysis of randomized controlled trials of short-term psychodynamic therapy suggests a lower response rate for this modality than for the more reeducative therapies, but firm conclusions from these studies cannot be drawn because of their inherent methodological difficulties. There are no established clinically useful predictors by which to choose from among the available psychotherapies. However, time-limited therapies are preferred over time-unlimited therapies because efficacy has been established for the former but not for the latter, and because medication is an effective alternative if time-limited psychotherapy fails. Some clinicians believe that exploratory psychotherapies may have a more prominent role in the treatment of Axis II disorders, whereas reeducative therapies may be more useful in the treatment of Axis I conditions. However, there is no firm evidence that psychotherapy alone is preferred over medication when there is a concurrent Axis II disorder. Conversely, psychotherapeutic tactics may be called for with patients who have personality disorders and who are managed with medication, to ensure adherence and to gauge the effects of the medication. Declaring an unsatisfactory response to acute-phase psychotherapy  Psychotherapy, if used alone, should logically be tried for a finite period of time and outcomes assessed, just as with medication. For mood disorders, one of the more serious difficulties may be that a response without full remission is accepted by both patient and provider—a risk similar to that encountered with medication. Secondly, what began as a time-limited approach may actually become a time-unlimited therapy if it is only partially effective when, logically, medications are available. Thus, when psychotherapy alone is used as the treatment for a mood disorder, it is essential both to limit the trial period and to measure the outcome to make a timely revision in the treatment plan if the objectives are not being met. Determining when to declare psychotherapy a failure is a complex problem that is not fully answered by available data. Some patients respond early, whereas others appear to take 8-10 weeks to respond. Declaration of psychotherapy nonresponse is largely based on lack of efficacy, although a few patients will discontinue treatment unilaterally. The preemptive discontinuation rate may be higher in actual practice than in efficacy trials. As with medication, if a patient inappropriately discontinues treatment while symptomatic, it is advisable to try to reengage him or her, because the depression has not remitted and, consequently, the prognosis is poor. What treatment should follow if psychotherapy alone is ineffective? Medication, given its established efficacy, is the next logical step. The psychotherapy may be continued or discontinued when medication is begun. Whether a different kind of psychotherapy would be effective if the initial form of psychotherapy was not effective has not been tested.