Letters

Questionnaires for depression and anxiety

BMJ 2001; 323 doi: http://dx.doi.org/10.1136/bmj.323.7305.167/b (Published 21 July 2001) Cite this as: BMJ 2001;323:167

Systematic review is incomplete

  1. Michael Pignone (pignone{at}med.unc.edu), assistant professor of medicine,
  2. Bradley N Gaynes, assistant professor of psychiatry,
  3. Kathleen N Lohr, professor of health policy and administration,
  4. C Tracy Orleans, senior scientist,
  5. Cynthia Mulrow, professor of medicine
  1. University of North Carolina, Chapel Hill, NC 27599-7110, USA
  2. Robert Wood Jones Foundation, Princeton, NJ 08543, USA
  3. University of Texas Health Science Center-San Antonio, San Antonio, TX 78284, USA
  4. University of Pennsylvania, Comprehensive Cancer Center, 3400 Spruce Street/11 Gates, Philadelphia, PA 19104, USA
  5. University of Auckland, Private Bag 92019, Auckland, New Zealand

    EDITOR—Gilbody et al published a systematic review of the effect in primary care settings of routinely administered questionnaires on the recognition, treatment, and outcome of psychiatric disorders, particularly depression.1 They reviewed randomised trials published throughout 2000 and concluded that the routine administration of such questionnaires is a costly exercise that has not been shown to influence clinicians' behaviour.

    On behalf of the United States Preventive Services Task Force, we performed a broader systematic review of the effectiveness of routine screening for depression. In addition to those reported in Gilbody et al, we identified six randomised trials of screening that examined recognition, treatment, or clinical outcomes.27 Each study used a validated screening instrument and gave feedback to providers of the screening results; some also confirmed results from the screening instrument with a criterion standard or gave systematic support to providers and patients to improve the quality of care after recognition of the diagnosis or condition. Gilbody et al cited the study by Wells et al in their discussion but did not include it in their analysis; other studies were not addressed in the report. Since several of these studies had positive impact on at least one major outcome, we are concerned that not including them may have affected the conclusions of the review. Gilbody et al included one study of depression screening that we did not identify.8 Even with the addition of the extra studies, we agree with Gilbody et al that interpreting the results of these screening trials is difficult because of the heterogeneity of outcome measures and times at which outcomes were assessed. In addition, because the minimal difference in outcome rates that is considered clinically important has not been defined, the question of whether existing trials have enough power to exclude an important effect remains unclear.

    Gilbody et al surmise that because the post-test probability of major depression is only 50% after a positive screen, physicians may tire of sorting out actual cases from those who would not benefit clinically. We argue, by contrast, that a 35-50% probability of major depression is quite high enough to justify the next step—namely, a diagnostic interview lasting 10 minutes. Not acting on the results of a positive screen may be related to lack of familiarity with diagnostic interviewing, lack of skill in treating depression, or failure to appreciate the morbidity associated with depressive disorders. Systematic efforts to provide additional support for treatment and follow up, as employed in the study by Wells et al, may improve the likelihood of improved outcomes.

    References

    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.

    Routine screening entails additional pitfalls

    1. James C Coyne (Jcoyne{at}mail.med.upenn.edu), codirector, behavioural sciences and health services research,
    2. Steven C Palmer, chief postdoctoral fellow,
    3. Richard Thompson, postdoctoral fellow, department of psychiatry
    1. University of North Carolina, Chapel Hill, NC 27599-7110, USA
    2. Robert Wood Jones Foundation, Princeton, NJ 08543, USA
    3. University of Texas Health Science Center-San Antonio, San Antonio, TX 78284, USA
    4. University of Pennsylvania, Comprehensive Cancer Center, 3400 Spruce Street/11 Gates, Philadelphia, PA 19104, USA
    5. University of Auckland, Private Bag 92019, Auckland, New Zealand

      EDITOR—Gilbody et al, in their meta-analysis, challenge the conventional wisdom that routine screening is an efficient means of improving the outcome of depression in general medical settings.1 We agree that the allocation of resources that would be required to make such screening feasible is not warranted. It seems, however, that Gilbody et al could have made an even stronger case for their conclusions.

      Specifically, they may have overestimated the efficiency of screening and underestimated the pitfalls inherent in its implementation. The positive predictive value of an elevated score on a screening instrument is typically less than 35%, rather the 50% they assumed in analysis.2 Furthermore, the considerable professional resources consumed in resolving false positive cases decrease the resources available for efforts to improve the outcome of already detected patients. The PRIME-MD study showed this problem. More than 80% of the 1000 patients screened positive for a mental disorder, and most of these cases proved to be false positives. Clinicians averaged an additional eight minutes in their follow up interviews with these patients, an increase of over 50% in the length of a typical primary care visit. This effort, however, ultimately yielded only 16 new prescriptions for antidepressant medication.3 If implemented without significant (and costly) support, this programme would drastically reduce the time and resources available for follow up with previously identified patients. Another study recently showed how difficulties in obtaining patient acceptance and in integrating follow up interviews into the competing demands of biomedical care can substantially reduce the yield of new patients in need of treatment.4

      Finally, there is the comparatively recent phenomenon, at least in North America, of the prevalence of antidepressant prescription equalling or exceeding the prevalence of depression among general medical patients.5 Some of this undoubtedly reflects increased detection of depressed patients and alternative uses for antidepressant drugs, but much of the increase represents inappropriate and inadequate treatment for depression.

      This is especially important given repeated demonstrations that increased detection does not translate into improved outcomes in non-specialty settings. This situation is likely to be aggravated through the implementation of routine screening. Without efforts aimed at improving the outcome of already detected depression in routine general medical care and at reducing the inappropriate prescription of antidepressants, routine screening remains part of the problem, rather than the solution to untreated or inadequately treated depression outside of specialty mental health settings.

      References

      1. 1.
      2. 2.
      3. 3.
      4. 4.
      5. 5.

      Two screening questions may be helpful

      1. Bruce Arroll (b.Arroll{at}auckland.ac.nz), associate professor of general practice and primary health care
      1. University of North Carolina, Chapel Hill, NC 27599-7110, USA
      2. Robert Wood Jones Foundation, Princeton, NJ 08543, USA
      3. University of Texas Health Science Center-San Antonio, San Antonio, TX 78284, USA
      4. University of Pennsylvania, Comprehensive Cancer Center, 3400 Spruce Street/11 Gates, Philadelphia, PA 19104, USA
      5. University of Auckland, Private Bag 92019, Auckland, New Zealand

        EDITOR—In response to the systematic review by Gilbody et al on routine administered questionnaires for depression and anxiety, screening has not improved outcomes of patients with depression. 1 2 Since that report there has been one study that asked patients a single question: “Have you felt depressed or sad much of the time in the past year?”3 The screened group were more likely to recover (48% v 27%, P<0.05), presumably because those with major depression were more likely to be recognised (45% v 24%) and treated (55% v 28%, P<0.02).3 Compared with the gold standard diagnostic interview schedule the single screening question had sensitivity of 85% of those with depression and a specificity of 66%.

        Another screening tool of two questions has been developed in a written form as opposed to an oral question.4 These two questions are, “During the past month have you often been bothered by feeling down, depressed or hopeless?” and, “During the past month have you often been bothered by little interest or pleasure in doing things?” A “yes” to either of these has a sensitivity and specificity of 96% and 66% for depression in patients in whom substance abuse is excluded. Written questionnaires are too time consuming for primary care physicians. The two screening question (consisting of two questions) is easy to remember and has a positive predictive value of 33% for a prevalence of depression of 15%. Any patient with a positive response to either of these questions can be asked a few further questions from the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, to rule depression in or out. This is a much more practical solution, and we are hoping to test these two questions in a clustered randomised controlled trial in general practice.

        References

        1. 1.
        2. 2.
        3. 3.
        4. 4.