How many general practitioners will read this week's paper by Arroll et al on diagnosing depression (p 1144)1 and sigh, “Another thing we have to do”? On the face of it we have a simple and accurate “test” for depression that is effective. Does this then impose a burden on the backs of general practitioners, another duty that will attract censure if not done properly? We think not. Rather, we ask what can be dropped, and how we can simplify our clinical work?
Making diagnoses has often been taught as a long and complicated process. “First take a thorough history” is the unhelpful advice from textbooks. What does that mean? Ask what is often a muddle of questions, and listen to a lot of information. Then pursue some lines of inquiry and not others. Finally synthesise this to come up with a list of possibilities (the differential diagnoses).
How effective any of this is remains largely unknown. Currently deciding what diagnostic manoeuvre to undertake is largely based on habit and ritual than anything rational. But that is changing, as this paper and others add evidence to improve the diagnostic process.
In diagnostic testing more is not necessarily better. Clinical prediction rules begin with a large battery of items. By statistically separating the wheat from the chaff, this battery can usually be reduced to just a few important items. For example, the Mini-Cog—which uses only a three item recall and a clock face drawing—shows promise as a simple check for dementia.2 The two minutes it takes is more feasible in general practice than performing a full mini-mental state examination. The simplicity has other advantages: one of us recently used it in general practice to detect dementia in a Vietnamese patient, something that would otherwise have been impossible without a professional interpreter and using the mini-mental state exam. We, especially clinicians in primary care, should applaud the development and compilation of further simple instruments.3
Can such simple tools be applied to the notoriously difficult area of depression? Again the answer is yes. Most have similar accuracy, according to a recent review of 11 paper based instruments for detecting depression, although they vary greatly in complexity (1-30 items), and the time needed.4 Among these, the two questions asked by one are particularly simple, appealing, and accurate.5 But mode of questioning is important. Answers people give verbally may differ from those given on paper. So showing that we can effectively ask the two questions verbally is an important step forward.
However, using this information, especially interpreting the answers, is important: the questions are sensitive but not specific. Hence, a negative on both questions makes depression very unlikely, but a positive, even for both questions, means only that we need to explore more fully the possibility of depression, rather than diagnose it on this basis alone.
Such tools are helpful, but much more work is needed in both primary research and systematic reviews. We are at the dawn of a new phase of evidence based practice: the diagnostic age. The Cochrane Collaboration has recently resolved to add diagnosis to intervention as a proper field for systematic reviews.6 As we master the many difficulties attached to doing this, we will start to assemble an easily accessible list of things that are useful and those that are useless. Will we find more that is useless than effective? Probably.
As this information becomes available, and the information is disseminated in a way that is accessible and useful to clinicians, so we will start to behave differently. We predict that this concentrated diagnostic information will have more impact on our day to day business than did the first wave of similar information for effective interventions. For one thing, deciding the diagnosis and prognosis takes up more time than actually initiating treatment. Some of the things we take for granted now will soon become outdated and old fashioned, and new ways of doing them will appear. The paper by Arroll et al is one of the first of these. We predict that similar papers will come in thick and fast.
Primary care p 1144
Competing interests: None declared.
References
- 1.Arroll B, Khin N, Kerse N. Screening for depression in primary care with two verbally asked questions: cross sectional study. BMJ 2003:327: 1144-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Scanlan J, Borson S. The Mini-Cog: receiver operating characteristics with expert and naive raters. Int J Geriatr Psychiatry 2001;16: 216-22. [DOI] [PubMed] [Google Scholar]
- 3.Ebell MH. Evidence-based diagnosis: a handbook of clinical prediction rules. Berlin, New York, Vienna: Springer, 2001.
- 4.Williams JW Jr, Noel PH, Cordes JA, Ramirez G, Pignone M. Is this patient clinically depressed? JAMA 2002;287: 1160-70. [DOI] [PubMed] [Google Scholar]
- 5.Whooley MA, Avins AL, Miranda J, Browner WS. Case-finding instruments for depression. Two questions are as good as many. J Gen Intern Med 1997;12: 439-45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Deeks J, Gatsonis C, Clarke M, Neilson J. Cochrane systematic reviews of diagnostic test accuracy. Cochrane Collaboration Methods Groups Newsletter 2003;7:8-9. www.cochrane.de/newslett/MGNews_2003.pdf (accessed 5 Nov 2003).