August 16, 2011 – A new analysis, published in the British Medical Journal (BMJ), reports that flawed research studies have exaggerated the degree to which depression screening questionnaires are able to accurately detect people with untreated depression. The number of untreated patients who would actually be detected using these questionnaires may be less than half the number predicted by existing studies. In addition, fewer than one in six patients who screen positive for depression using standard questionnaires likely have the condition.
Led by Dr. Brett Thombs of the Lady Davis Institute for Medical Research of McGill University in Montreal, researchers concluded that almost all existing studies are flawed in a way that exaggerates the accuracy of these questionnaires, and that most patients flagged as possibly depressed do not, in fact, have depression. Indeed, they called into doubt the efficacy of screening which is increasingly relied upon to detect depression. They reached these conclusions upon reviewing the published medical literature on depression screening questionnaires.
“Screening for depression may detect some patients with untreated depression, which is what we want.” said Thombs. “On the other hand, the ability of screening questionnaires to detect untreated patients has been exaggerated and their use could subject many patients without depression to inappropriate labelling and treatment with anti-depressant medications, with their attendant side effects.”
Anti-depressant use has increased dramatically in recent years. Approximately 15% of American adults above the age of 35 are currently taking anti-depression medication, the bulk of which is prescribed by primary care physicians who are encouraged to use screening questionnaires to identify patients with untreated depression.
Questionnaires are useful if they can accurately identify depressed patients who have not already been identified and treated by their doctors, while, at the same time, minimizing the number of people falsely identified as depressed. Thombs and his team reviewed almost 200 studies on the accuracy of depression screening questionnaires that were included in seventeen evidence reviews published from 2005 to 2009. They found that only eight of 197 studies (4%) were conducted exclusively with patients whose depression status was unknown when the questionnaires were administered. More than 95% of studies also included patients already diagnosed with, or treated for, depression, a factor which is known to exaggerate the accuracy of testing.
“Patients who are already treated for depression are the easiest to recognize. Physicians do that without a screening questionnaire,” said Thombs. “Testing depression screening questionnaires on patient samples that include already recognized and treated patients isn’t a fair test of their accuracy. We need to know how screening would perform among patients whose depression is not obvious to doctors, and that is a more difficult task.”
Thombs added, “Few people would be impressed by an accurate forecast of yesterday’s weather. Yet, this is the standard that has been applied to evaluations of depression screening questionnaires. We need to do a much better job. Depression is too serious a condition for us not to know how well our screening tests actually work.”
The Canadian Institutes of Health Research and the Fonds de la Recherche en Santé Québec provided funding that supported work on this study. In addition to Thombs, other researchers who contributed to this study were Erin Arthurs, a member of the research staff of the Lady Davis Institute for Medical Research; Ghassan El-Baalbaki, a McGill postdoctoral fellow; Anna Meijer, a graduate student from the University of Groningen, The Netherlands, Roy C. Ziegelstein, M.D., of Johns Hopkins University, and Russell J. Steele, Ph.D. of McGill University.
Research Communications Officer
Lady Davis Institute for Medical Research
Tel.: 514-340-8222 x 8661
Image source: By Hendrike 11:08, 3 May 2006 (UTC) (Own work) [GFDL (www.gnu.org/copyleft/fdl.html) or CC-BY-SA-3.0 (www.creativecommons.org/licenses/by-sa/3.0/)], via Wikimedia Commons