Screening for depression in general practice and related medical settings

In Australia, more than 800 000 adults2 and 95 000 children and adolescents3 experience depressive disorders each year, and depression is a leading cause of both ongoing disability and premature mortality.4 Furthermore, depression is seen to make substantial contributions to major medical and behavioural disorders, such as heart disease and alcohol or other substance misuse. Much effort is currently directed towards increasing community awareness of depression5 and improving the quality of medical interventions provided.6 Importantly, in the absence of broad population and enhanced health service approaches, the contribution of depression to overall disability and premature mortality is projected to increase and be second only to cardiovascular disease by 2020.7

Data from Australian general practice8,9 show that most patients with depression or anxiety presenting to medical care do not receive a psychological diagnosis or appropriate interventions. Consequently, some mental health advocates promote the use of simple case-detection systems on both clinical and epidemiological grounds.10-12 The more conservative position has been to reject such calls, arguing conceptual, practical and ethical limitations.1

As the number of relevant studies has increased, it is timely to review this evidence. We sought to answer the following questions:

• Can screening tools efficiently and accurately identify patients with depression?

• Do clinicians use the information obtained from screening tools to assist with diagnosis and treatment planning?

• Is the outcome of patients with depression improved within practice settings that use screening tools (alone or in combination with other techniques)? and

• Is there evidence that the benefit is limited to specific subpopulations of patients?

Methods have been developed to conduct systematic evaluations of reviews.13,14 These methods are efficient and reliable, and avoid the need to consult the original articles (unless considerable disagreement exists between the conclusions of well-conducted reviews). We used these methods, focusing our attention on systematic reviews of "case-finding" and "screening/intervention" procedures for common forms of depression encountered in primary care, general practice or other related general medical settings.

Articles were identified by searching the Medline (1966–2002), EMBASE (1980–2002) and PsycINFO (1966–2002) databases using the keywords depression or depressive disorders, plus screening or mass screening or psychiatric status rating scales or case finding, plus primary health care or family practice or clinical practice or general practice or general medical setting(s). We supplemented these sources by searching the Cochrane databases (to 2002); performing additional specific Medline, EMBASE and PsycINFO searches; scrutinising reference lists of selected articles; and querying experts. We did not seek articles concerning depression screening in association with specific medical conditions such as cancer, stroke or heart disease or their related service systems. In total, 1386 articles were identified.

Eligibility criteria for inclusion of reviews were:

• review of prospective studies with a primary focus of depression screening in general medical settings;

• review of studies of healthy populations or those with known depression;

• publication in a peer-reviewed journal; and

• written in English.

Two authors (T D, C R) independently reviewed the titles and abstracts of the 1386 articles. Those for which both reviewers agreed that eligibility criteria had not been met were excluded. When the reviewers disagreed, the full articles were reviewed by all authors and a decision on inclusion was made by consensus. Eleven reviews satisfied the inclusion criteria, incorporating studies published between 1976 and 2001.

The quality of each review was assessed independently by two authors (T D, C R) using the Oxman and Guyatt Index13 and scoring system as modified by Peach (see Box 1).14 Four reviews15-18 met the criteria of a score of five or more by Peach's system.14 One author (C R) abstracted and tabulated relevant data from the included reviews (Box 2). A second author (T D) checked the accuracy of the tables against the original reviews. All authors then examined the evidence presented in each review to respond to our four questions (Box 3).

Of the four key reviews identified, two15,18 dealt largely with the practical and predictive capacities of screening tools. Both reviews concluded that there were many instruments that were acceptable from both a psychometric and practical implementation perspective. The other two reviews16,17 focused on the effects of screening on outcomes, with split recommendations. The larger and more recent review17 argued for the value of screening procedures in adults when used in conjunction with other practitioner feedback and relevant service enhancement techniques. The other review argued against the value of screening,16 and its conclusions have been the subject of considerable debate.19 It included fewer studies and had a more restrictive view of the questions being addressed.

It is important to note that the United States task force that now recommends screening did so only for adults,20 arguing that there was insufficient evidence of benefit from current treatments to recommend screening of children or adolescents.

Our interpretation of the value of screening rests largely on the more recent and more inclusive review of Pignone et al.17 As argued by these authors, their review used a larger and more appropriate dataset in its meta-analyses (derived from 14 studies). The review by Gilbody et al16 pooled data from only four studies, and commented on negative findings from two other studies. The clear danger was that too few studies were included to reveal a positive effect. It seems clear to us that sufficient data are now available to be more confident about the value of screening.

Critics of screening for depression highlight technical, practical and ethical issues. The most important technical issue has been the limited positive predictive value (35%–50%) of the instruments when used in primary care environments. Although some authors have argued that this is insufficient to recommend routine use,21 others have concluded that the performance of these instruments is adequate, provided that screening is followed by a formal diagnostic interview.19 A wide variety of self-report screening instruments have been developed to increase detection, follow the course of disorders, and more effectively engage patients in their own treatment.12 Such instruments are not "diagnostic" screening tests, but are more akin to other forms of medical examination or pathology testing (eg, recording body temperature, blood pressure or haemoglobin levels), where abnormal results indicate the need for further enquiry (rather than making a specific diagnosis). It is common to combine the use of screening instruments with more formal education in relevant diagnostic systems, improved communication skills and recognition of other patient and doctor factors that promote treatment adherence.22 The introduction of screening tools without attention to such issues would be unlikely to result in the improved outcomes described in the reviews. In the Australian setting, we are now seeing systematic attempts to use screening and outcome tools within this appropriate context.23

A common difficulty for screening systems in mental health is deciding which "disorder" or "diagnosis" is being predicted. The narrow view is that only specific diagnoses (eg, major depression) that are associated with clear treatment guidelines should be identified.6 The broader view is that screening tools should be used to identify a wider range of typically comorbid disorders (eg, depression, anxiety, somatoform disorders, and alcohol or other substance misuse).11 This avoids the need to use multiple screening tools for closely related disorders. Therefore, if the real "diagnosis" of interest is actually a series of common mental disorders, then the prevalence in general practice is high (up to 60% for some forms of depression, anxiety, somatoform disorders, and alcohol or other substance misuse),9 and the positive predictive value of related instruments increases proportionally. The data provided within the reviews presented here deal only with the specific topic of depression. A new wave of such studies in primary care is likely to begin to address this broader issue.9

An important ethical consideration is whether screening tools should be introduced within a health system that lacks commitment to enhanced service provision. Until very recently, the Australian healthcare system has relied almost exclusively on general practitioners to provide care for people with common mental disorders.24 However, GPs have not been supported by adequate professional training, remuneration, logistic, professional and specialist support, or access to specialist treatment systems.25 The weight of evidence is that screening instruments are only likely to be of benefit where they are tied clearly to practice procedures that use the information to promote evidence-based treatments.17 Therefore, screening for depression (or comorbid disorders) can only be recommended in those healthcare systems that also incorporate new practice and practitioner-training systems designed to maximise evidence-based treatments.23

An important difference between screening for depression and other forms of medical screening should also be noted. Mental health screening is not "screening" in the typical sense. Self-report instruments require the person to disclose the symptoms of mental disorder voluntarily. If a person does not wish to reveal that he or she has psychological or related somatic symptoms, he or she can simply not complete the questionnaire or can return negative responses. Importantly, such instruments are predictive of a person's recognition of need for appropriate psychological assessment.10 In essence, such methods are less about screening and more about efficient methods for prompting voluntary disclosure of key psychological symptoms. Other doctor and patient factors clearly influence disclosure of psychological symptoms, so acceptance of screening tools will be affected by those same factors (eg, doctor's communication skills, patient's understanding of the nature of their symptoms, time that the doctor is likely to spend with the patient, doctor's confidence in providing psychological treatments, doctor's training in mental health).9,10,22,26 Medical practitioners are often reported by patients to hold surprisingly negative views about people with mental health problems,27 and such views will also affect disclosure.

Although the case for effective pharmacological and non-pharmacological treatments of common forms of depression, anxiety, and alcohol or other substance misuse in the general practice setting is well established,6 the notion of effective treatments is not well known to people in the community.5 Additionally, many people often perceive medical treatments for depression to be harmful (eg, causing addiction, personality change or severe medical side effects) or leading to other adverse outcomes (eg, through increased stigma or resultant discrimination in employment or access to insurance). This raises another important ethical concern. If providing treatment results in direct medical benefits but also contributes to other social losses, then the practitioner has an additional obligation to ensure that the patient (and in some instances their family) has been provided with adequate information with regard to treatment alternatives (including the consequences of not being treated at all). To date, we have little evidence that systematic information about the risks and benefits of treatments is provided to patients or their families by either primary care or specialist services.28 Ethically, screening in medical settings is not analogous to screening in non-medical environments (eg, telephone screening) or within other health systems (eg, antenatal or postnatal clinics), where the information from the screening tool is not linked directly to the provision of relevant information or treatments.

Given the potential impact of screening on existing workloads, it is likely that GPs will dismiss screening systems as "over-sensitive".29 Although the instruments are not over-sensitive relative to existing standards (such as degree of resultant disability, clinician judgement, patient's perception of need for treatment, or formal psychiatric [DSM-IV30] diagnoses), a case can be made for not treating everyone identified by the systems. Treatment decisions, as in all other clinical situations, require formal diagnoses, clinical judgements, comparison with existing guidelines,6 and negotiation with the person affected. The decision by a clinician not to screen or treat, however, may be skewed by other considerations, such as their own lack of confidence11 or skills in delivering psychological treatments, their inability to access appropriate professional and specialist support services, their misperception of the seriousness of the condition, and a tendency (shared with the rest of the community5) not to see common mental disorders as major health problems.

In our view, the most pertinent ethical and professional dilemma surrounding the introduction of screening in general practice is whether appropriate services will be provided to those who are eventually diagnosed as having depression (or other related disorders). First, the number of people identified by any of the self-report and related diagnostic systems currently available (7%–15%) would be far in excess of the number of patients currently receiving any formal intervention (about 4%).31 Clearly, it would be unethical to introduce screening if practitioners ignored the results, or if health systems did not change rapidly to accommodate the increased need for relevant treatments. However, it may become unethical in the future to ignore high prevalence mental disorders that have serious health consequences. Currently, practitioners (or health systems) that ignore people with hypertension, diabetes or asthma, or ignore child immunisation or cervical cancer screening, or discourage detection of harmful patterns of alcohol or tobacco use, could be described as acting in an unethical manner. Current evidence suggests that it is time for professional leaders in general practice in Australia to reconsider their stance on screening for common mental disorders such as depression.