Article type
Year
Abstract
Background: The Patient Health Questionnaire-9 (PHQ-9) is the most commonly used depression screening tool in primary care. Most primary studies and meta-analyses of PHQ-9 accuracy focus on sensitivity and specificity, which estimate the probability of a screening result given true diagnostic status. However, due to the complex nature of conditional probabilities, it can be challenging for clinicians to translate these estimates into more meaningful numbers that reflect the likelihood that a patient has depression given her or his screening score. When high sensitivity and specificity are reported, some clinicians believe it means a positive test is “virtually diagnostic,” even if positive predictive value is low. One way to improve understanding of screening tool accuracy estimates is to present information in a format that is more readily understood, such as natural frequencies.
Objectives: To create a user-friendly knowledge translation tool based on sensitivity and specificity estimates from a large individual participant data meta-analysis of PHQ-9 accuracy. The tool allows clinicians to estimate, for a given PHQ-9 screening cutoff and depression prevalence, how many patients would screen positive versus negative, and how many in each group would be correctly versus incorrectly identified.
Methods: We developed a web-tool with a 100-person diagram that self-populates based on user-entered values of major depression prevalence and PHQ-9 cutoff threshold. The tool provides instructions for use, including advice for estimating underlying prevalence and selecting a cutoff, text to explain the numerical results shown in the diagram, and a FAQ section with basic information about depression screening. Family physicians were consulted to improve the format and presentation and to ensure that the content is clear and addresses needs of clinicians considering using the tool.
Results: The web-tool can be found at depressionscreening100.com/phq.
Illustrated example: As shown in Figure 1, by entering an underlying major depression prevalence of 10% and selecting the standard PHQ-9 cutoff score of 10, 22 of 100 patients would be expected to screen positive on the PHQ-9, 9 (39%) of whom would meet diagnostic criteria for major depression (true positives) and 13 (61%) of whom would not (false positives). Of the 78 patients expected to screen negative, 77 (99%) would be correctly ruled out (true negatives), while 1 (1%) would be a missed major depression case (false negative). Numbers in the diagram automatically update for different combinations of prevalence and cutoff.
Conclusions: The present web-tool improves clinician understanding of complex diagnostic accuracy estimates from meta-analyses by translating results into natural frequencies that are more readily understood and providing guidance on their meaning and use.
Patient or healthcare consumer involvement: We consulted with several family physicians during development.
Objectives: To create a user-friendly knowledge translation tool based on sensitivity and specificity estimates from a large individual participant data meta-analysis of PHQ-9 accuracy. The tool allows clinicians to estimate, for a given PHQ-9 screening cutoff and depression prevalence, how many patients would screen positive versus negative, and how many in each group would be correctly versus incorrectly identified.
Methods: We developed a web-tool with a 100-person diagram that self-populates based on user-entered values of major depression prevalence and PHQ-9 cutoff threshold. The tool provides instructions for use, including advice for estimating underlying prevalence and selecting a cutoff, text to explain the numerical results shown in the diagram, and a FAQ section with basic information about depression screening. Family physicians were consulted to improve the format and presentation and to ensure that the content is clear and addresses needs of clinicians considering using the tool.
Results: The web-tool can be found at depressionscreening100.com/phq.
Illustrated example: As shown in Figure 1, by entering an underlying major depression prevalence of 10% and selecting the standard PHQ-9 cutoff score of 10, 22 of 100 patients would be expected to screen positive on the PHQ-9, 9 (39%) of whom would meet diagnostic criteria for major depression (true positives) and 13 (61%) of whom would not (false positives). Of the 78 patients expected to screen negative, 77 (99%) would be correctly ruled out (true negatives), while 1 (1%) would be a missed major depression case (false negative). Numbers in the diagram automatically update for different combinations of prevalence and cutoff.
Conclusions: The present web-tool improves clinician understanding of complex diagnostic accuracy estimates from meta-analyses by translating results into natural frequencies that are more readily understood and providing guidance on their meaning and use.
Patient or healthcare consumer involvement: We consulted with several family physicians during development.