Behavioral clinicians around the country are starting to use assessment and outcomes measures. The most commonly used instrument is the PHQ9 which is a self-report measure that rates severity of depression. It is being used for eminently practical reasons: it is quick to take, has sound psychometric properties, and it is a quality measure for ONC Meaningful Use (and NCQA has proposed it to be a HEDIS measure for 2016). Finally, it is clinically useful. The sensitivity/specificity for PHQ9 scores over 10 is 88% which means it is extremely accurate at diagnosing depression. Also, studies show response to treatment in the first month, as reflected in declining PHQ9 scores, predict the likelihood of remission.

We looked at some PHQ9 data from patients seen by CarePaths providers and compared it with PHQ9 data from a primary care setting.(1)


As can be seen, behavioral clinicians see a much more disturbed cohort of depressed patients than do primary care physicians. Consider that behavioral clinicians see:

–16.5 times more severely depressed patients than PCPs

–8.25 times more moderately severely depressed patients than PCPs

–4.17 times more moderately depressed patients than PCPs

–1.46 times more mildly depressed patients than PCPs

What this data suggests is that more depressed patients are being referred to behavioral health clinicians.

The next table shows the outcomes of treatment as measured by PHQ9 change scores. An outcome is defined as follow up PHQ9 two months or longer from the index PHQ9 assessment. Patients are considered improved if their PHQ9 score decreases by 5 or more points, deteriorated if the scores increase by 5 or more points and not changed if the scores change by less than 5 points.


The results show improvement for all levels of severity of disturbance, with clinically significant improvement shown by 51.6% of all clinically depressed patients seen. More disturbed patients (the moderately severe and severe patients) show the greatest gains with clinically significant improvement in about two thirds of patients seen.

These data are rough. It is to be expected that more severe patients will show the greatest gains. Also, we did not calculate other statistics of interest, e.g. percent of patients recovered or in remission as opposed to improved, effect sizes, etc. Nor did we control for the duration of treatment. Follow up PHQ9 administrations varied between 2 months and over 3 years. PHQ9 recovery curves show that treatment gains are often significant for at least five months, so shorter followups underestimate change.

We also did not control for type of treatment: psychotherapy only, medication only, or combined treatment. Practice guidelines are associated with levels of severity on the PHQ9.(2) For example, severe depression should be treated with pharmacotherapy and/or combined treatment, whereas mild and moderate depression allow for monotherapy, either psychotherapy alone or medication alone. Combining this information with outcomes data might show a differential benefit of type of treatment.

We are implementing a national Practice Research Network initiative. Our hope is to identify practices that lead to optimal outcomes in actual practice settings. Participation will entail nothing other than using the PHQ9 in your practice according to a standard protocol, e.g. monthly PHQ9 assessments for 6 months. Those of you who participate will get benchmark data on your practice (how you compare with national norms).

We will be posting more about this project in the next month or two. If you are interested helping us shape this project, please contact us at

Thanks to Dave Johnson, MSW, for PHQ-9 research and key insights into depression treatment in care settings .

  1. NCHS Data Brief, No. 172. December 2014.

  2. Kroenke K, Spitzer RL, Psychiatric Annals 2002;32:509-521.