Major depressive disorder is common in primary care outpatients (4 to 8 percent), and nonmajor forms are even more common (6 to 14 percent). Both recognition and diagnosis of depression rest on an awareness of risk factors for depression, as well as elicitation of the key signs, symptoms, and history of illness. The primary risk factors for depression are:
A variety of clinical clues may further alert the practitioner to the likelihood of a mood disorder.
Guideline: A history of major depressive episodes increases risk for subsequent major depressive episodes. (Strength of Evidence = A.)
One major depressive episode is associated with a 50 percent chance of a subsequent episode; two episodes, with a 70 percent chance of a subsequent episode; and three or more episodes, with a 90 percent chance of recurrent depression over a lifetime (NIMH Consensus Development Conference, 1985).
Guideline: A history of mood disorders in first-degree relatives increases the probability of a patient’s developing a mood disorder. (Strength of Evidence = A.)
First-degree relatives of bipolar disorder patients are at substantially higher risk for developing either a recurrent major depressive disorder (roughly 12 percent) or bipolar disorder (roughly 12 percent). Strong scientific evidence points to a genetic vulnerability to bipolar disorder. For those with more recurrent forms of major depressive disorder, genetic factors also appear to play a significant role. For those with less recurrent forms of major depressive disorder, the role of genetic factors is unclear. It is known, however, that patients who develop major depressive disorder before age 20 have a greater familial morbidity for depression (Goodwin and Jamison, 1990).
Guideline: A history of suicide attempts should trigger the practitioner to inquire specifically about depressive symptoms. (Strength of Evidence = A.)
Suicide attempts are frequently associated with mood disorders. In addition, a history of suicidal ideation and/or attempts increases the patient’s risk for subsequent suicidal ideation/attempts. Additional risk factors for suicide include:
The signs and symptoms of depression can be sought by direct interview, which may be supplemented with self-report ratings and/or by a history obtained from the patient’s spouse or a friend. Depressive disorder is diagnosed on the basis of positive evidence—not by exclusion. For example, patients with or without other general medical conditions either do or do not have a depressive syndrome. Depression should not be explained away or discounted as a commonplace “reaction” to the concurrent general medical condition. Similarly, life events, including losses of important people or key roles, usually precede the onset of major depressive episodes. Thus, the presence of selected untoward life events should not be used to explain away the major mood episode.
Guideline: The clinical interview is the most effective method for detecting depression. The interview elicits the nine criterion symptoms of major depressive disorder and the longitudinal course of illness. Similarly, interviewing for symptoms and course of illness is essential to identifying bipolar, dysthymic, and other mood disorders. Specific questions regarding the criterion symptoms are asked. Since either a depressed, blue, or sad mood or a loss of interest or pleasure is required, these symptoms are elicited first. The clinician who suspects or diagnoses a depressive disorder is advised to perform and record the results of a mental status examination, which includes whether the patient has suicidal ideation/intention; is oriented, alert, cooperative, and communicative; exhibits a normal level of motor activity; and is psychotic. (Strength of Evidence = A.)
Some patients may initially den, the depressed mood, but may identify the somatic symptoms (sleep, appetite, and weight changes). Upon further discussion, the interviewer should return to the issues of mood and interest. Patients may initially complain about sleep, appetite, energy, concentration, and sex drive changes or about chronic or intermittent pain or anxiety. The clinician should be alert to considering the diagnosis of depressive illness in these patients.
If symptoms of depression are present, it is important to determine their time course. How long has the patient been in an episode of major depressive disorder? Were there prior episodes? What degree of interepisode recovery has occurred? How severe are the current symptoms? The more severe forms of depression, characterized by marked suicidal thinking, multiple neurovegetative symptoms, and markedly impaired functioning, argue strongly for the use of medication. A chronic or multiepisode history argues for more prolonged treatment. Symptom severity should be gauged by either clinical interview or rating scales, since severity plays a role in treatment planning (Depression Guideline Panel, forthcoming).
When there is historical, symptomatic, or physiologic evidence to suggest an underlying general medical disorder as the cause of the depression, physical examination and laboratory tests to detect the specific disorder(s) should be used as appropriate in the differential diagnosis. The presence of another medical condition does not exclude the depressive syndrome. In fact, it increases the probability that a depressive syndrome may be present. The depression is either present or absent, based on the patient’s signs/symptoms. If the patient has a depression, the first steps in its treatment differ according to whether another general medical condition is present and thought to be causal, whether alcohol or substance abuse is present, and whether another nonmood psychiatric disorder is present and causal.
Easily administered self-report questionnaires can be used as a lowcost, but valuable, case-finding tool to help clinicians better detect currently depressed patients. Numerous scales are available for assessment of depression (Beckham and Leber, 1985; Marsella, Hirschfeld, and Katz, 1987). It is not uncommon for primary care practitioners to have patients complete self-report depression rating scales in the waiting room to elicit these symptoms. These self-reports are not diagnostic, but they are very sensitive to depressive symptoms.
These self-report scales are also useful in detecting milder mood conditions. The four used most commonly in ambulatory medical care settings are the General Health Questionnaire (GHQ), which has one subscale for depression; the Center for Epidemiological StudiesDepression Scale (CES-D); the Beck Depression Inventory (BDI), and the Zung Self-Rating Depression Scale (ZSRDS) (Coulehan, Schulberg, and Block, 1989). Patient responses to these instruments cannot be used directly to formulate a diagnosis of depression; however, a high score on one of these instruments warrants a clinical interview based on DSM-III-R criteria. Some of these limitations have been addressed by newly developed measures—the Inventory to Diagnose Depression (Zimmerman, Coryell, Stangl, et al., 1987) or the Inventory for Depressive SymptomsSelf-Report (Rush, Giles, Schlesser, et al., 1986). Although initial results are promising, these instruments have not yet been applied extensively in primary care settings.
All four self-report measures commonly used in primary care settings (GHQ, CES-D, BDI, and ZSRDS) help to identify potentially depressed patients with varying degrees of power exceeding chance. All have positive predictive values between two and four times the 6 to 8 percent base rate of major depression. That is, the false-positive rate for detecting major depression with self-reports may be as high as two- to threefold. They detect nearly every patient with major depressive disorder, but they are not very specific. They identify many patients who ultimately prove not to have a mood disorder. In essence, there are few false-negative results, but many (25 to 40 percent) false-positive results.
Fifteen studies have rigorously evaluated the predictive value of selfreport questionnaires for depression in primary care populations by using the diagnosis from a structured psychiatric assessment as the “gold standard” against which the self-report formulation is compared (Coulehan, Schulberg, and Block, 1989). This body of research leads to the following recommendations for case finding with self-report measures:
Several available rating scales are to be completed by the clinician: the Hamilton Rating Scale for Depression (HRS-D) (Hamilton, 1968), the Montgomery-Asberg Depression Rating Scale (MADRS) (Montgomery and Asberg, 1979), the Inventory for Depressive Symptomatology-Clinician Rated (IDS-C) (Rush, Giles, Schlesser, et al., 1986), the Bech-Rafaelsen Depression Scale (BRDS) (Bech, Kastrup, and Rafaelsen, 1986), and depression items of selected structured interviews such as the Present State Examination (PSE) (Wing, Birley, Cooper, et al., 1967) and the Schedule for Affective Disorders and Schizophrenia (SADS) (Endicott and Spitzer, 1978). These may be more sensitive to improvement in the course of treatment and may have slightly greater specificity than do self-reports in detecting depression.
A systematic approach to the identification of depressive disorders can be derived from the information obtained on these questionnaires and scales:
Differential diagnosis of depressive disorders rests largely on clinical phenomenological grounds. The practitioner is advised to proceed according to the following steps:
This process is summarized in Table 8.
Guideline: The descriptive diagnosis of depression is based entirely on the patient’s signs, symptoms, and personal history. Therefore, the practitioner should spend substantial time carefully interviewing the patient and, where appropriate, other informants (i.e., close relatives or friends) before embarking on extensive biologic, neuropsychological, or psychological testing. The main principle is to conduct only a limited number of basic laboratory tests to detect potential general medical causes for the depression unless specific risk factors, specific positive symptoms on the medical review of systems, unusual symptom profiles, or an atypical course of illness is present, in which case selected additional tests are called for to answer specific diagnostic questions. (Strength of Evidence = B.)
There are two kinds of laboratory tests: (1) those that screen for underlying medical causes for the depression and (2) those that identify biologic abnormalities characteristic of depression.
At the point at which a differential diagnosis is framed, with specific alternative conditions being relatively likely based on history, physical examination, or other risk factors, the primary care provider may use selected tests to determine whether these associated conditions are indeedpresent. In addition, various laboratory tests, such as the thyrotropinreleasing hormone stimulation test, the dexamethasone suppression test, and the sleep electroencephalogram, identify biologic abnormalities characteristic of the depressed state (see Rush, Cain, Raese, et al., 1991; Goodwin and Jamison, 1990 for reviews). These tests are not recommended for routine use in primary care outpatients. They lack sufficient specificity and sensitivity to be useful as screening tools.
Table 8. Steps in detecting and treating depressive conditions
The general principle of laboratory testing is that the likelihood of tt disorder for which the test is being conducted should be relatively high the patient being tested because virtually all tests have certain falsepositive as well as false-negative rates. If the false-positive rate of the te is 5 percent and the incidence of the condition in the population is 1 percent, 5 of 100 patients will test positive when only 1 in 100 really ha the condition. The clinician should have a rational basis, independent of the laboratory test per se, for selecting and conducting the test. For example, if a depressed patient shows significant symptoms suggesting a cementing process or a clinical history suggestive of higher cortical impairment, further neuropsychological or necrologic laboratory tests, sun as the electroencephalogram or magnetic resonance imaging, may be use: either to establish or exclude the diagnosis, or to identify the causes of t! dementia. If, on the other hand, the patient simply complains of concentration problems that are not out of the ordinary for a person with typical major depressive disorder, the use of such tests is both inappropriate and likely in at least some patients to lead to false-positive findings, thereby requiring additional testing, unnecessary specialized consultation, delay in appropriate treatment, and unnecessary worry to patients and family members.
Large epidemiologic studies in the general population suggest that previously unrecognized laboratory abnormalities occur in between 0.8 al 4.0 percent of the population (Korvin, Pearce, and Stanley, 1975). A significant portion of these abnormalities, however, either are predictable based on clinical assessment (Dolan and Mushlin, 1985) or represent false-positive test findings for the reasons cited previously. Those populations at highest clinical risk for meaningful abnormalities on these screening laboratory examinations include those who cannot care for themselves entirely, those who currently abuse alcohol or other substances, and those who merit a more thorough evaluation because of age or clinical circumstances (Ferguson and Dudleston, 1986; Koran, Sox, Marton, et al., 1989; Roca, Breakly, and Fischer, 1987; Sox, Koran, Sox, et al., 1989).
These general population findings are supported by three studies in which a psychiatric population was prospectively investigated for identification of previously unrecognized illness (Ferguson and Dudleston, 1986; Kolman, 1985; Willett and King, 1977). The principal findings in these studies, which involved more than 1,300 adult psychiatric inpatients, suggest an abnormality rate of 6 to 17 percent in this psychiatric population. Most of these abnormal findings, however, were predictable based on history and physical examination or were false-positives. A side finding was that many test results were never reviewed by the ordering physician.
Laboratory tests should be tailored to the population served and should be specifically based on patient findings. Reliance on laboratory tests in general evaluation of the condition of depressed patients should be greater if:
The issue of thyroid disease and depression is a case in point. Studies using thyroid function tests to screen nearly 15,000 general medical patients for either hyper- or hypothyroidism indicate that such a procedure is not cost-effective in identifying patients with the disorder (0.3 percent detection of hyperthroidism [women/men = 10/1]; 0.5 percent detection of hypothyroidism [women/men = 2/1].
At least two studies indicate thath using fatigue and depression as prescreening symptoms does little to improve the likelihood of identifying hyper- or hypothyroidism with thyroid function tests as a general screen. The first study included 250 patients and found no increase in the identification rate of patients with hyper- or hyprthyroidism when mental disorders or tiredness were present compared with the general population (Eden, Sundbeck, Lindstedt, et al., 1988). The second study suggested that, while the likelihood of identifying true cases of disease on laboratory testing improved when the symptoms of hyper- or hypothyroidism were elicited in advance, depression and fatigue were not among the most highly discriminant symptoms (Drake, Miller, and Evans, 1982).
These studies suggest thath the optimal means to screen for thyroid disorders are through history and physical examination. The one exception is in the identification of hypothyroidism in women over the age of 50, in whom the identification rate is 10.3 percent in previously unidentified cases (Eden, Sundbeck, Lindstedt, et al., 1988; Hodkinson and Denham, 1977; Taylor, Tomson, and Caird, 1974). Thus, for women over 50 with depressive symptoms, thyroid function tests have a key role in detecting thyroid disease. The prevalence of this illness in women over age 50 is the same in women with major depressive disorder. The same cannot be said of older men, who are far less likely to suffer thyroid disease as a group than are older women.
When symptoms of thyroid disease are present, case finding with thyroid function tests increases to 12 percent for hyperthyroidism (Nuutila, Irjala, Viikari, et al., 1988) and to 3 to 4 percent for hypothyroidism (Goldstein and Mushlin, 1987; Nuutila, Irjala, Viikari, et al., 1988). As the symptoms of thyroid disease increase in number, case finding with such laboratory testing increases (more than five symptoms, 50 percent identification rate) (Drake, Miller, and Evans, 1982). Although fatigue is seen as a symptom of thyroid disease, in isolation it does not increase case finding with laboratory tests over that obtained in the general population (Sugarman and Berg, 1984).
When a treatment trial is conducted, the patient's symptomatic response should be monitored. If the patient does not respond fully, then the diagnostic workup (history, medical review of systems, and physical examination) should be repeated. At that juncture, a differential diagnostic picture may appear that was not obvious at initial evaluation. Appropriate laboratory testing for the differential diagnostic questions raised at the reassessment should then be undertaken.
As noted earlier, some laboratory tests that may be relatively specific at depressive conditions are now under clinical investigation. These include selected sleep electroencephalogram features and failure to suppress cortisol following a dexamethasone challenge, among others (Rush, Cain, Raese, et al., 1991). While many severely depressed patients, especially those with psychotic or melancholic symptoms, display abnormalities on these tests, the tests are not indicated as routine screens for depression because they lack specificity and have lower sensitivity in the less severely ill. However, they can play a role in selected differential diagnostic situations (for example, using the dexamethasone suppression test to differentiate psychotic depression from schizophrenia or using the sleep electroencephalogram to differentiate sleep/wake disorders such as sleep apnea from depression), which are typically managed by specialists rather than primary care providers.
Most psychological tests are standardized in the general population (Dobson, 1985). In general, recommendations for the use of psychological and neuropsychological tests in screening for depressive disorders are similar to those for laboratory tests. They are not recommended for routine use in screening for depression. However, in selected cases, psychological and neuropsychological tests may be very useful in differential diagnosis of depression. Neuropsychological tests, such as the modified Halstead-Reitan battery, should be considered when it is necessary to distinguish dementia from depression. Depressed patients have been found to show significantly different test profiles on the Minnesota Multiphasic Personality Inventory (MMPI), the Symptom Checklist 90, and the Millon Clinical Multiaxial Inventory than do nondepressed patients with psychiatric diagnosis (Weitzler, Strauman, and Dubro, 1989). While these psychological tests predicted depression at a rate surpassing chance, such instruments were not substitutes for a thorough clinical evaluation. In other words, these tests may confirm, but they do not independently render, the diagnosis.
Perhaps the most commonly faced differential diagnosis in primary care practice is that between depressive and anxiety disorders. Dobson (1985) determined that the self-report symptom ratings for these two disorders have poor divergent (discriminative) validity, which is not surprising, since many depressed patients also have anxiety symptoms or suffer from concurrent, co-morbid anxiety disorders.
With regard to distinguishing subtypes of depression, the MMPI and Rorschach are of limited value in differentiating major depressive from bipolar disorder and melancholic from nonmelancholic major depressive disorder. In one study, patients with major depressive disorder showed elevated scores on most MMPI scales, compared to those with bipolar disorder, depressed phase (Donnelly, Murphy, and Goodwin, 1976), although subsequent studies failed to replicate this finding (Lumry, 1978; Silver, Isaacs, and Mansky, 1981). Another study found Rorschach responses to be of no value in distinguishing melancholic from nonmelancholic depressions (Carney, Roth, and Garside, 1965).
It is not appropriate to use psychological tests, including the MMPI, as the sole means to determine whether a patient with a concurrent symptomatic nonpsychiatric medical illness "really" has a depression, because symptoms from the medical illness itself affect test results. To illustrate, medical patients typically have a depressive/hypochondriacal pattern on the MMPI, because the MMPI somatic symptoms that emanate from the nonpsychiatric medical illness contribute to the depressive/hypochondriacal symptom scales.
In summary, the available empirical data do not support the routine use of self-report or projective instruments to differentiate depression from anxiety disorders. These tests can distinguish depressive from other nonanxious psychiatric conditions, and they can enhance a careful clinical formulation rendered by a clinician well trained in the diagnosis of depressive disorders. Where the differential diagnosis is in doubt, psychological tests may help to tip the balance in favor of one or the other condition. Like laboratory tests, however, psychological tests should not be used for routine screening purposes or administered for all differential diagnostic situations.
Guideline: A critical element in the differential diagnosis of depressive disorders is ongoing clinical reassessment. (Strength of Evidence = B.)
Most patients with major depressive disorder respond partially to medication within 2 to 3 weeks, and full symptom remission is typically seen in 6 to 8 weeks. Most patients receiving time-limited psychotherapy respond partially by 5 to 6 weeks and fully by 10 to 12 weeks. Patients who fail to show this pattern can be detected through careful interviewing or by clinical or self-report rating scales. In these patients, a reevaluation is indicated. The clinical reevaluations may include repeating a thorough general medical and psychiatric history, physical examination, and a more thorough medical laboratory evaluation. For patients on selected medications, blood level measurements may help gauge whether the serum level of antidepressant is in the therapeutic range.
A significant subset of patients with major depression also exhibits maladaptive personality traits or disorders. When the underlying mood disorder is successfully treated, the expression of these maladaptive traits may partially or completely resolve (Joffe and Regan, 1988; Thompson, Gallagher, and Czirr, 1988). However, studies suggest that patients with such preexisting personality disorders are less likely to exhibit a full therapeutic response in affective symptoms to either medication or time-limited psychotherapy, or they may take longer to respond fully. In those who respond only partially to treatment and in whom personality disorders are suspected, psychological testing may be useful to determine the presence of personality disorders. If present, the case for combined treatment with medication and psychotherapy may be stronger.
These are general principles whose application to individuals requires judgment, logic, and flexibility. First, it is necessary to define a response, which is usually thought to be at least a 50 percent improvement in baseline (pretreatment) symptom severity or a global judgment that the patient is at least much improved. A remission is defined as either the presence of few or no symptoms or a global patient report of marked improvement. In some cases, other ongoing problems, such as chronic severe life stresses, may slow the response or impair the likelihood of attaining a fully asymptomatic state in 10 to 12 weeks. In such patients, longer observation periods may be needed while treatment continues.
Finally, improvement in some symptoms (e.g., insomnia) may not be the best way to judge overall treatment effectiveness. For example, the side effects of medication (e.g., sedation) or psychotherapy (e.g., improved sense of hope or optimism) may result in abatement of selected symptoms while failing to remove all symptoms of the depressive disorder.
The timing and type of reassessment, and the interpretation of the results of this reassessment are important. Based on panel consensus and an awareness that most psychopharmacology studies have used assessments conducted every 1 to 2 weeks to evaluate response in efficacy trials, it is recommended that medication treatment visits or telephone contacts initially be weekly to ensure adherence, adjust dosage, and detect and manage side effects. After 3 to 4 weeks, visits may be less frequent for most patients. The degree of response/remission can be assessed at each visit, as well as any evidence of side effects. Whether the clinical interview is supplemented with a symptom-rating scale or not, the practitioner should inventory all of the criterion symptoms of depression included in the patient's initial complaint. For those with personality disorders or severe ongoing life stresses, as long as the patient is showing some significant improvement over baseline, the treatment need not be changed for 8 to 10 weeks. For others, an earlier judgment can be made to change the treatment if no meaningful response is found (or to change earlier, if adverse side effects are encountered).