Depression in primary care: Tools for screening, diagnosis, and measuring response to treatment

Introduction

How many times have physicians wished they could measure a serum sadness level, and show the report to their skeptical patients, thus convincing them that they have a medically treatable disorder? After all, this is the model we use for diabetes, thyroid disease, elevated cholesterol, and many other chronic illnesses. Even when we reach that moment where there is widespread public acceptance that clinical depression, despite its multifactorial causes, is a treatable illness, we will need to use diagnostic tools to confirm our clinical suspicions, inform our patients, and monitor their treatment outcomes.

Symptom-based psychiatric rating scales were developed more than 40 years ago to assign numerical values to a complex range of patient behaviors, affects, and feelings. They have since proliferated into a bewildering array of tools designed for a variety of purposes, some very general in their scope, and some quite narrowly focused. This article is not intended to be a comprehensive review; rather it is an attempt to identify and describe a few such tools that can be easily incorporated into a busy clinician’s repertoire to improve diagnosis and management of depression.

 Why use symptom-based rating scales?

Although we routinely rely on clinical data, most explanations of how to interpret diagnostic data are confined to laboratory and X-ray reports. Yet symptoms and signs usually produce far more powerful support of diagnostic hypotheses than we can ever derive from the laboratory. Rating scales are not intended to provide a substitute for good clinical judgment. Once we have a clinical suspicion that depression plays a role in a particular patient’s problems, we want reliable, accurate ways of confirming the diagnosis and monitoring the patient’s progress over time. We need psychometrically sound, user-friendly tools that give us clinically useful information, and that are reliable, valid, and consistent for a variety of patients and settings when administered by different clinicians.

To choose the correct tools we need to determine the goals of our assessment. Screening tools, which give us a quick indication of whether further assessment is warranted, need to have high sensitivity (few false-negatives). Diagnostic tools need good content validity (do they measure what we think they do?), test-retest reliability (reliability over time), good inter-rater reliability (agreement between clinicians), and high specificity (fewer false-positives). When we use scales to evaluate treatment outcomes we need good test-retest reliability and the scale needs to be sensitive enough to detect clinically significant changes in a variety of domains. The gold standard for comparison of all these tools is always a focused, in-depth interview by an experienced mental health clinician. For busy physicians, an ideal test would be short, straightforward, and reliable for screening, diagnosis, and outcome assessment. 

 Which tools to use?

In most clinical settings we are likely to have multiple goals of assessment and a minimum of time, so careful choice of tools is crucial. Any assessment must be individualized to acknowledge language/cultural differences, intellectual or cognitive impairments, age-specific issues (children, teens, the elderly), co-morbid psychiatric or other illness (anxiety, bipolar disorder), or concurrent substance abuse (e.g., the CAGE questionnaire is useful). Risk of suicide must always be evaluated.

Both self-report inventories and clinician-rated scales are available.[1] Some are in the public domain, while others are protected by copyright and require payment of a fee for their use. Ultimately, a clinically useful diagnostic test must do three things: provide an accurate diagnosis, support application of an efficacious therapy, and ideally, lead to a better outcome for the patient.

The complete evaluation process is impractical for an individual physician to apply to each test he or she uses, so this article will summarize some of the most useful validated tools in depression. As a busy doctor you can choose and become familiar with these few, thereby improving the quality of your patient assessments. Most of the frequently used instruments show robust correlations among themselves, although the self-rating scales show better correlation among themselves than with the clinician-rated scales. Tempting though it may be to use a cut-off score on a self-report inventory as a single means of deriving a diagnosis, it is inadequate and unreliable and should be avoided.

Screening

Two quick questions from Primary Care Evaluation of Mental Disorders (PRIME-MD)[2] can provide us with a highly sensitive (94%) but not very specific (35%) screening test for depression:[3]

1. Have you been bothered by little interest or pleasure in doing things?

2. Have you been feeling down, depressed, or hopeless in the last month?

If a patient responds positively to these two questions, only four follow-up questions—on sleep disturbance, appetite change, low self-esteem, and anhedonia—are needed to confirm a diagnosis of depression. If a patient has a positive response on at least two of these four questions (Table 1), the specificity of a positive test increases to 94%.[3]

Self-rated screening tools are also available.[1] The Hospital Anxiety and Depression scale is the most widely investigated and validated scale for screening; however, it is too long and difficult to score, making it less useful in clinical practice. The 20-item Zung Depression Self-Rating Scale is less commonly used but it is in the public domain. It does not have adequate sensitivity to detect change over time, so it is not considered useful for following response to treatment.

The Geriatric Depression Scale (GDS) is a self-report measure designed to minimize the impact of somatic symptoms associated with aging and illness.[4] It has a yes/no format, and the 15-item version, using a cutoff of five, has good sensitivity and positive predictive values for diagnosis of major depression (Table 2). If a clinician is concerned about cognitive impairment, the Mini Mental State Exam (MMSE), which takes 5 minutes to administer and score, is a useful addition.[5]

 Diagnosis

The full version of the PRIME-MD clinician-rated scale, available in the public domain, contains 26 yes/no questions concerning symptoms experienced in the past month, and incorporates observed and reported behavior.[2] On average, it takes 8.4 minutes to perform. If we have a clinical suspicion that a patient is depressed, the pre-test probability is close to 50%, and the post-test probability after a positive test (using a cut-off value of 5/9) becomes 94%. This is better than most of the routine laboratory tests we use daily in practice. If the score is 4/9 or less, then we need to consider other depressive disorders such as dysthymia, complicated bereavement, adjustment disorder, mixed anxiety/depression, minor depressive disorder, or premenstrual dysphoric disorder. PRIME-MD has been validated in adults and adolescents over age 13; its applicability to seniors and children is limited.

More recently, a streamlined patient self-report version of the PRIME-MD, called the Patient Health Questionnaire (PHQ) has been made available.[6] The PHQ is 3 pages long and covers the five most common psychiatric issues in primary care (depression, anxiety, alcohol, somatoform, and eating disorders). An abbreviated PHQ (the PHQ-9) for depression has been developed that reduces physician time to less than 3 minutes (Table 3).[7] The PHQ-9 also offers a severity score for each symptom, and hence can also be used to follow outcome. The PHQ-9 and other depression tools can be downloaded from the MacArthur Foundation web site (www.depression-primarycare.org). 

 Measuring outcomes

The Hamilton Depression Rating Scale (HAM-D), the oldest, most widely used and validated instrument, has numerous versions, both clinician-rated and self-reported, as well as a computer-administered version.[8] Some versions are currently available in the public domain while others are still copyright protected. The clinician-administered versions are widely used in clinical trials for evaluating response to treatment but they require training to use, take 20 to 30 minutes to administer, and so are less useful for busy family physicians.

Many clinicians prefer to use a patient self-rated scale such as the Beck Depression Inventory (BDI, protected by copyright and requiring permission and payment of a fee to reproduce). The BDI-II is a 21-item self-report measure of the severity of depressive symptoms. It has high sensitivity and specificity and is valid and reliable in assessing the severity of depressive symptoms.[9] Among its shortcomings are its high item difficulty (requires the patient to be able to read and understand the questions) and poor discriminant validity against anxiety.

Currently, the HAM-D and the BDI are probably the best-validated scales to quantitatively assess response to treatment. Response has been defined as a 50% reduction in baseline score on the HAM-D or BDI in most clinical trials—in practice we ideally want to see our patients in remission (e.g., scores within the normal range) rather than just a 50% symptom reduction. There are shorter six- to seven-item versions of the HAM-D and the BDI, but they are not yet widely validated or used in clinical practice, and do not include some important clinical items such as sleep disturbance.

 So what should I use in my office?

On a day-to-day basis, use of the two PRIME-MD screening questions followed by either the rest of the clinician-administered PRIME-MD or the self-report PHQ-9, with evaluation of both alcohol/drug consumption and anxiety by screening questions, remains the briefest, simplest, most accurate way to diagnose major depression in an adult population. Using the self-report BDI or the PHQ-9 to follow scores at baseline and designated follow-up intervals is an accurate and reliable strategy that allows us to identify those individuals who are unresponsive to treatment and/or who require further intervention or consultation. Patients can complete and score the questionnaires themselves in the waiting room prior to seeing their doctor.

Consistent use of this systematic approach to depression management can improve our diagnostic accuracy, save time, help us choose appropriate treatment interventions, and effectively monitor outcomes. This approach should also allow us to further reduce the significant burden associated with depression in primary care.

 Competing interests

Dr Michalak is supported by a Canadian Institutes of Health Research/Wyeth Ayerst Canada Postdoctoral Research Fellowship.

 Table 1. PRIME-MD screening questions to detect depressive symptoms (www.depression-primarycare.org/ap1.html) [9]

 Table 2. Geriatric Depression Scale—GDS, Short Form (www.stanford.edu/~yesavage/GDS.html) [4]

 Table 3. Patient Health Questionnaire–PHQ-9 www.depression-primarycare.org/ap1.html.).[7]

This article has been peer reviewed.

Dr Anderson is a family physician in Sooke, British Columbia. Dr Michalak is a postdoctoral research fellow in the Division of Clinical Neuroscience, Department of Psychiatry, UBC. Dr Lam is professor and head of the Division of Clinical Neuroscience, Department of Psychiatry at UBC and director of the Mood Disorders Centre at UBC Hospital.