PATIENT VERSUS CLINICIAN RATED DEPRESSION SCORES: A COMPARISON OF PARTICIPANT SCORES ON THE CARROLL DEPRESSION SCALE AND THE HAMILTON DEPRESSION RATING SCALE

Abstract

Introduction In clinical practice, physicians and other medical providers rely heavily on self-assessment tools to screen and assess severity of psychiatric illness, including depression. Ensuring that these self-rating depression scales correlate to investigator rated depression scales is important for diagnosis, validity, and patient care. Several studies have demonstrated a strong correlation between adult patient rated scales compared to clinician completed depression scales which served to establish the validity of self-reported tests. To the best of our knowledge, no studies have been performed in depressed older adults to compare the correlation of self-report versus clinician completed depression scales. We hypothesized that older adults would rate depression higher on self-reported scales compared to clinician completed depression scales and that variables such as cognition would contribute to these findings. Methods Using data collected from the Neurobiology of Late Life Depression (NBOLD) study, we compared the 61-item self-report Carroll Depression Scale – Revised (CDS-R) scores against the 17-item clinician rated Hamilton Depression Rating Scale (HAM-D) scores in 100 depressed patients that were obtained on the same day as part of the study baseline assessment battery. Based on the crosswalk between CDS-R and HAM-D, the first 52 questions from the CDS-R were compared with the HAM-D. Spearman correlation, weighted kappa coefficient (κ), scatter plot, and linear regression were used to determine the agreement between CDS-R scores and HAM-D scores, and evaluate potential influencing factors for any discrepancy. Both total score and the 17 individual domain-specific subscale scores were examined. Results Consistent with the hypothesis, CDS-R total score (mean 19.6, SD 6.8) were found to be, on average, six points higher than HAM-D total score (mean 13.4, SD 3.9). This six point difference between CDS-R and HAM-D was consistent along the range of depression severity, as demonstrated in figure 1. Comparing domains within each assessment tool, suicidal tendency, weight loss, early insomnia had the highest correlation, while agitation, insight, and general somatic symptoms had lower correlations. The relationship between CDS-R total score and HAM-D total score was not affected by age, gender, Montgomery and Ǻsberg Depression Rating Scale (MADRS), and Cumulative Illness Rating Scale (CIRS) total scores. Cognition, as indicated by CERAD total score, was the only statistically significant covariate for predicting CDS-R total score in a model that included HAM-D score, which supported the hypothesis. Conclusions Self-assessment screening tools, including those for depression, are important tools for researchers and clinicians alike. These screening tools, such as the Carroll Depression Scale, have demonstrated positive correlation compared with clinician completed scales, such as the Hamilton Depression Rating Scale. Similarly, this data analysis demonstrated positive correlation between the CDS-R and HAM-D in older adults. Limitations of this study include small sample size and mild to moderate depression scores in this ambulatory population. Additional studies would be beneficial to compare other depression rating scales, especially in older adults. Other research is needed to investigate variables that may attribute to higher rates of self-reported depression symptoms. This research was funded by Not Applicable