Traumatic Brain Injury and Major Depressive Disorder

To the Editor: Dr Bombardier and colleagues¹ reported that 53% of 559 adults hospitalized with complicated mild to severe traumatic brain injury (TBI) met criteria for major depressive disorder (MDD) during the 12 months following hospitalization. Patients were administered the Patient Health Questionnaire (PHQ-9) as many as 9 times during the 12-month follow-up and were classified as MDD cases if they met the study PHQ-9 threshold criterion at least once.

The PHQ-9, however, is not a clinical interview designed to diagnose MDD. Rather, it is a screening questionnaire, which is intended to differentiate persons likely to have MDD from those at lower risk. A study by Fann et al² found that the PHQ-9 was 93% sensitive and 89% specific for MDD in TBI. Based on those sensitivity and specificity rates, in a sample with 10% MDD prevalence, for example, 19% of patients would screen positive on the PHQ-9, almost double the actual MDD prevalence.

The 53% rate reported by Bombardier et al,¹ however, was a cumulative rate. Between 52% and 77% of the 559 patients in the study were screened with the PHQ-9 at each of 9 screenings. Based on an assumed 89% specificity of the PHQ-9 among TBI patients,² approximately 11% of noncases at each assessment would have had false-positive depression screens and incorrect classification as MDD cases. Similarly, among patients who never had MDD during the 12-month study period, but who were screened 3 times, 30% would be expected to have at least 1 false-positive PHQ-9 screen and, thus, incorrect MDD classification. The expected rate of at least 1 false-positive screen would be 44% for patients screened 5 times, 56% for patients screened 7 times, and 65% for patients screened 9 times.

Bombardier et al concluded that the cumulative rate of MDD in their sample was 7.9 times what would be expected in the general population based on the 12-month rate of 6.7% reported in the National Comorbidity Survey Replication (NCS-R).³ However, the latter rate was obtained using a validated structured clinical interview for MDD, rather than a screening tool, and was based on a single 12-month retrospective assessment, rather than cumulative surveillance, both of which would produce substantially lower rates compared with the methods employed by Bombardier et al. Depression following TBI is an important public health problem. Rates of cumulative positive screens, however, are not equivalent to rates of MDD based on validated diagnostic methods.