Screening for depression in patients with coronary heart disease or diabetes does significantly increase the number of patients with recognised depression, but the absolute number is so small that it calls into question the value of screening,a UK retrospective cohort study has found.¹

Data from a large representative sample of Scottish general practices for the 12-month period starting 1 April 2007 was obtained from the PPCIUR primary care database. Patients screened for depression using the two validated questions (see below) were identified by the relevant Read code.

• During the last month, have you often been bothered by feeling down, depressed or hopeless?

• During the last month, have you often been bothered by having little interest or pleasure in doing things?

Patients with a new diagnosis of depression were identified either by a depression diagnosis code or initiation of antidepressant treatment (excluding those commonly prescribed for pain).

The number of patients newly diagnosed or treated for depression during the 28 days following screening was compared with the number diagnosed or treated at other times. During the four-week post-screening period the relative incidence (RI) of a new depression diagnosis was 3.03 (95% CI:2.44-3.78) and the RI of new antidepressant prescribing was 1.78 (95% CI:1.54-2.05).

The total number of patients screened was 67,358. The number needed to screen (NNS) to obtain one new coded diagnosis of depression was 976 (95% CI: 886-1,104). The NNS to start one new patient on antidepressant treatment was 687 (95% CI: 586-853); this includes both coded and uncoded patients.

These figures are an order of magnitude greater than the NNS of 46 reported by a Dutch primary care study of screening for depression in high-risk patients.²

This highlights the dangers of extrapolating results of research studies to real world settings. In the Dutch study, screening was undertaken by research workers rather than practice nurses, and the screening tools were more sophisticated than the two question screen.

There are a number of possible explanations for the remarkably high NNS. Patients may be discouraged from answering positively to the screening questions if they are asked in a perfunctory manner in the context of a complex chronic disease consultation.

For those who do answer in the affirmative, the positive predictive value is low: unless practice nurses have the necessary training and skills to filter out the large number of false positives, both they and their GP colleagues may rapidly become disillusioned by the screening process and subsequently discount the small minority of true positives. Even if clinicians do make a diagnosis of depression, patients identified by screening are often reluctant to accept treatment.²

The authors conclude that the very small yield of new diagnoses does not justify brief screening for depression in patients with chronic disease. I agree. Screening for depression has been ‘an unhelpful diversion’³ and NICE is right to recommend that the DEP1 indicator (see box 1, below left) should be retired from the 2013/14 Quality and Outcomes Framework.⁴