[Sensitivity and specificity of the worry and anxiety questionnaire (WAQ) in a sample of health-care users in the province of Quebec].

INTRODUCTION

Generalized anxiety disorder (GAD) is a prevalent anxiety disorder characterized by persistent, excessive worrying. Even if GAD's ill consequences on health and quality of life are well documented, this disorder is still difficult to identify in primary care. The worry and anxiety questionnaire (WAQ) is a questionnaire assessing specific GAD symptoms, as defined by the Diagnostic and Statistical Manual of Mental Disorders-4th Edition (DSM-IV).

OBJECTIVES

This study aimed at assessing the capacity of the French version of the WAQ's to identify individuals with GAD in a sample of individuals reporting a certain level of anxiety. A second objective was to identify which of its items better distinguish individuals with GAD from those without. According to these results, different scoring algorithms have been developed and their effect on the WAQ's sensitivity and specificity indicators has been explored.

DESIGN OF THE STUDY

The sample was drawn from a mother study in which 1110 health-care users completed questionnaires while waiting for a medical consultation with a family physician. Of those, a subsample of 219 individuals reported anxiety symptoms typical of GAD, as assessed by the WAQ. Among those who agreed to participate in the study's second phase (n=176), 100 were randomly selected and invited within one to three months to a clinical interview assessing their anxiety symptoms more thoroughly. Thirty-three individuals accepted and thus formed the present sample. The clinical interview was the anxiety disorders interview schedule (ADIS). The ADIS is a semi-structured diagnostic interview following the DSM-IV criteria. It assesses all anxiety disorders and includes screening questions on mood, substance use and psychotic disorders. Participants also completed the WAQ for a second time at the time of the interview.

RESULTS

Nineteen individuals received a diagnosis of GAD after completing the ADIS while 13 did not. Sixteen of the 19 individuals with GAD were correctly identified with the WAQ, compared to eight out of 13 for individuals without GAD. Sensitivity of the WAQ's actual scoring algorithm is thus of 84.2% and its specificity of 61.5%. The number of false negatives produced by the WAQ in this sample (3/19, 15.8%) was lower than the number of false positives (5/13, 38.5%). Positive and negative predictive power is thus of 76.2% and 72.7%, respectively. Receiver-operating characteristic (ROC) curves analyses indicated that the most useful items to identify individuals with GAD were those assessing the presence of excessive worrying, the number of days disturbed by worries, the degree with which worries interfere with daily functioning and the degree of control over worries. Knowing an individual's outcome on the WAQ increases the probability of correctly identifying an individual with or without GAD by 8.5 times compared to mere chance. A new scoring algorithm, where the cut-off score on the excessive worrying item was increased by one unit, considerably improves the WAQ's specificity (84.6%), without altering its sensitivity by much (78.9%). This new scoring algorithm thus increases the probability of correctly identifying individuals with and without GAD to 20.6 times (again compared to chance).

CONCLUSION

The WAQ is thus a useful instrument in screening GAD, even in a sample of anxious individuals. Its original scoring algorithm shows excellent sensitivity, a valued quality in an instrument used for screening. On the other hand, it is possible to increase the specificity of the WAQ by raising the cut-off point on the excessive worrying item, making the instrument useful as a diagnostic aid or as a screening questionnaire for GAD, in particular among a sample displaying anxiety. This questionnaire is thus an easy-to-complete and adaptable instrument that can be used by family physicians to help them identifying individuals with GAD.