Effectiveness of early detection

The critical question with any intervention for early detection and treatment is whether it does more good than harm as compared with the results of later diagnosis and intervention. The appropriate research tool for answering such a question is the randomized controlled trial. A MEDLINE search of the literature and subsequent secondary searches identified only four trials of sufficient methodologic quality. In each study the basic question was whether the routine testing of patients just before seeing their physician provided any benefit in terms of detection and management of depression beyond that achieved through the usual clinical evaluation. The evaluation of effectiveness incorporated the explicit criteria of Sackett, Haynes and Tugwell.

In a study by Shapiro and associates 1242 patients attending an inner-city primary care teaching facility over 4 months completed the GHQ (General Health Questionnaire) while waiting to see their physician. In addition, each patient underwent a structured interview (the Diagnostic Interview Schedule [DIS]) at home within 2 weeks after visiting their physician. The 488 subjects (39%) with positive GHQ scores were randomly assigned to one of three groups: (a) no feedback to the physician, (b) provision of the GHQ results to the physician immediately after completion of the questionnaire and (c) provision of the DIS findings to the physician immediately after the interview. Of the patients with negative GHQ results 40% were randomly assigned to one of two groups: (a) no feedback and (b) provision of the findings to the physician immediately after the questionnaire. The GHQ and the DIS were repeated after a 6-month follow-up for 6 months.

The provision of GHQ information had no statistically significant effect on the physician’s diagnosis. Among patients over 65 years of age there was a statistically significant increase in the detection rate. However, there was no ultimate effect on patient management, even in the elderly group. Shapiro and associates speculated that the GHQ data may have had a marginal impact for patients with low to medium GHQ scores when the physician was uncertain about the symptoms and complaints.

Hoeper and collaborators used the GHQ (General Health Questionnaire) to screen 1469 patients in a Wisconsin primary care office over a 5-month period. The people were then randomly assigned to either a group in which the physician received the score at the end of the visit or one in which the physician received no information. Physicians completed a standardized record tailored to address mental health issues. Knowledge of a positive GHQ result had no effect on the rate of detecting psychologic distress. The GHQ identified 28% of the patients as having psychologic distress; however, only 16% were identified by the physicians. The impact on management and patient outcome was not assessed. Furthermore, there was no description of patient selection or an attempt to assess the bias of using a special record.

Zung and colleagues asked 1086 patients over a 12-month period to complete the SDS (Zung Self-Assessment Depression Scale) in the waiting room of their family physician. Of the 143 patients (13%) with positive scores 102 were randomly allocated to a group in which the physician was immediately informed of the results and subsequently applied a structured interview during the patient’s visit; the other 41 were allocated to a group in which the physician was not told of the results. At follow-up 4 weeks later the patients were reassessed with the SDS and the charts reviewed for indications that the physician suspected or was treating depression. Among those whose physician was notified of the results 68% had charts indicating physician awareness of depression, as compared with only 15% of those whose physician was not informed. After reassessment at 4 weeks 64% of the identified patients showed clinical improvement, as compared with 18% of those who were in the unidentified (and untreated) group.

Unfortunately, the study by Zung and colleages has several major design flaws and provides inadequate evidence to support its conclusions. Not only was there insufficient validity testing of the SDS (Zung Self-Assessment Depression Scale) but also the study period was extremely brief. There was no indication of patient selection or of the proportion of patients who agreed to participate. Most important, there was a considerable loss to follow-up, so that the results represented only a small portion of the patients entered into the study.

In a rather complex study in Britain a general practitioner had 1093 consecutive patients complete the GHQ in the waiting room before their visit. In addition, the physician briefly assessed their psychologic status during the visit. For half the patients the physician reviewed the GHQ results at the end of the visit and extended the discussion as necessary. The others served as control subjects, the physician receiving no GHQ (General Health Questionnaire) results. After 1 year all of the patients were asked to complete the GHQ again, underwent a brief clinical assessment and were interviewed to assess the duration and severity of any psychiatric symptoms during the preceding year. On the basis of these findings the patients were labelled (retrospectively) as having transient, mild or severe psychologic problems.

During the study period 16% of the patients were treated for new episodes of psychiatric distress without data from the GHQ (General Health Questionnaire). An additional 11% were identified as a result of the GHQ results and further discussion with the physician. Patients with severe problems diagnosed as a result of the GHQ results provided to the physician at the initial visit had lower GHQ scores at 1 year than the control subjects with severe problems.

Although the investigators viewed this approach as being beneficial, there were several methodologic flaws. The control subjects selected for comparison were identified retrospectively, whereas the treatment patients were selected prospectively. The outcome measures relied heavily on patient recall over the preceding 1 year. The methods used to assess the effect of management were relatively indirect.

Although these four trials represent the best evidence currently available, they fail to provide adequate evidence to support the benefit of routine screening for the early detection of depression.

Related posts:

1. Early detection of depression. Part 4 Special problems The early detection of depression is much more complex and challenging than the detection of other conditions seen in the primary care physician’s office. Perhaps the most significant reason for this is how the natural history of depression differs from that...
2. Early detection of depression. Part 2 Are there acceptable routine screening tests? Instruments for routine case-finding in primary care settings must be of acceptable quality, brief and easy to use. Presumably the patients would complete the test while waiting to see their physician. Most self-administered tests have been designed...
3. Early detection of depression. Part 1 Depression is common and can usually be treated effectively. However, in many cases physicians may not recognize depression, especially in its early stages. Because the burden of suffering can be high, including death through suicide, early recognition and treatment are worth while. Such...
4. Childhood Depression It is estimated that 1.9% of primary school children and 4.7% of adolescents are affected by depression; the consequences of depression in this age group can include social dysfunction, academic underachievement and suicidal behavior. Within the last three decades, the recognition of a...
5. Managing Resistant Depression. Part 1 The World Health Organization identified mood disorders as a significant public health problem more than a decade ago. Some evidence suggests that the incidence of mood disorders is increasing. The lifetime prevalence rate for all mood disorders is about 8.3%; for serious depression,...