Exploration du lien entre la confiance diagnostique et l’accord interjuges dans l’évaluation de la sévérité du TAG

Abstract

Lorsqu’un clinicien a une forte confiance en son impression diagnostique, cette dernière est-elle plus fidèle ? Cette étude a pour but d’explorer s’il existe un lien entre la confiance diagnostique du clinicien et la fidélité de son impression diagnostique, tel qu’évaluée par accord interjuges. Nos hypothèses sont les suivantes : (1) la confiance diagnostique à une première entrevue diagnostique structurée sera corrélée avec la confiance diagnostique à une deuxième entrevue diagnostique structurée ; et (2) le degré de confiance diagnostique à chacune des deux entrevues prédira l’accord interjuges. La confiance diagnostique fut évaluée par des cliniciens ayant administré indépendamment deux entrevues diagnostiques structurées à 100 patients évalués pour participer à une étude de traitement pour le trouble d’anxiété généralisée. Les résultats révèlent une relation statistiquement significative entre la confiance diagnostique aux deux entrevues indépendantes. Par contre, la confiance diagnostique à chacune des deux entrevues ne corrèle pas avec l’accord interjuges. Lorsque le nombre de troubles comorbides est contrôlé statistiquement, la relation entre la confiance aux deux entrevues n’est plus significative. Il semble donc y avoir une entente sur les tableaux cliniques qui sont plus faciles ou difficiles à diagnostiquer selon le nombre de troubles comorbides.

Background

When providing a diagnostic impression, clinicians are guided by their sense of certainty or confidence. It would be expected, therefore, that when a clinician is confident in his/her diagnostic impression, the latter should stand a greater chance of being reliable. Surprisingly, some medical studies suggest that there is only a moderate relationship between a clinician's level of confidence in his/her diagnostic impression and the reliability/validity of the diagnosis. Generalized anxiety disorder (GAD) is one of the most difficult psychological disorders to assess and diagnose. Thus, GAD offers a unique opportunity to examine the relationship between clinicians’ diagnostic confidence and the reliability of their diagnostic impression. Our hypotheses were: (1) that diagnostic confidence on an initial diagnostic interview (Anxiety Disorders Interview Schedule for DSM-IV; ADIS-IV) would be correlated with diagnostic confidence on a second independent diagnostic interview (Mini International Neuropsychiatric Interview; MINI); and (2) that higher diagnostic confidence at each interview would be associated with greater inter-rater agreement.

Method

As part of a clinical trial for generalized anxiety disorder, we examined whether diagnostic confidence was associated with inter-rater agreement. Ratings of diagnostic confidence (0 to 100%) were provided by a team psychologist using the ADIS-IV and a team psychiatrist using the MINI following independent administrations of the interviews to 100 patients with GAD.

Results

Correlational analyses revealed a significant association between diagnostic confidence on the ADIS-IV and diagnostic confidence on the MINI (r = 0.23, p < 0.05). However, neither diagnostic confidence on the ADIS-IV nor diagnostic confidence on the MINI was associated with inter-rater agreement between the interviews. Conversely, an exploratory analysis showed that the number of comorbid disorders was related to diagnostic confidence on each interview (ADIS-IV: r = −41, p < .001; MINI: r = −36, p < 0.001), with higher comorbidity being associated with lower confidence. Finally, comorbidity accounted for the relationship between confidence levels on each of the interviews.

Conclusion

The results of this study suggest that clinicians tend to agree on which cases are easier or harder to diagnose based on the number of comorbid disorders present. However, confidence in one's diagnosis appears to be unrelated to diagnostic reliability. These findings have several implications. Given that generalized anxiety disorder is difficult to assess and has a very high comorbidity rate, clinicians may want to systematically assess their level of confidence in their diagnostic impression for patients with GAD. Moreover, if future studies replicate the finding that diagnostic confidence is unrelated to inter-rater agreement for the diagnosis of GAD, we suggest that clinicians should be careful not to conclude that their diagnostic impression is accurate based solely on their level of confidence. Ultimately, continued assessment during therapy remains the best strategy to ensure that the treatment offered is tailored to the patient's condition.