Skip to content
PaperFren

Research method

Patient Health Questionnaire (PHQ-9)

6 studies

Ask PaperFrenWatch this

The PHQ-9 is a nine-item self-report instrument for depressive-symptom severity, not a treatment. Papers in this set use it as an eligibility threshold (often PHQ-9 ≥10), as a primary outcome (mean change, response, remission, reliable improvement), and as a reference scale against which a new mood measure is correlated. GAD-7 often travels with it; a PHQ-9 benefit without a GAD-7 benefit — or the reverse — is a result, not a rounding error.

Anxiety- and depression-treatment studies need a shared severity language. PHQ-9 answers 'how do these symptoms score on a standard questionnaire at this week?' Its main limitation is that a falling score is not a mechanism, a single-arm drop to PHQ-9 <5 is not a randomised effect, and a 2.5-point mean difference can coexist with a non-significant remission contrast.

Studies

6

Findings

4

6 supporting · 0 challenging · 0 qualifying citations

Open tensions

2

Currently

What we know

  1. As a reference instrument, PHQ-9 correlated with a brief mobile Immediate Mood Scaler at r=.59 (GAD-7 r=.57) among 110 clinical and community adults. That cross-sectional correlation supports complementary in-the-moment monitoring; it does not show that IMS improves treatment outcomes, and the mixed epilepsy/community sample is not routine outpatient anxiety care.
  2. Randomised PHQ-9 contrasts in this set are modest-to-moderate and endpoint-specific. Kokoro-app adjunctive CBT in 164 participants with antidepressant-refractory major depression scored about 2.5 PHQ-9 points lower at week 9 (SMD 0.40) with more response but not significantly more remission. Internet CBT for depression in diabetes (91 with baseline data after screening 106) produced Hedges g=0.78 on PHQ-9 and g=0.72 on GAD-7; 66% completed all six lessons, and 51% of iCBT completers with data showed reliable PHQ-9 improvement versus 18% of treatment-as-usual.
  3. IntelliCare used PHQ-9 both to enter and to judge outcome. A single-arm field trial of 105 adults (PHQ-9 ≥10 and/or GAD-7 ≥8) found significant paired-score improvement, with 37% reaching PHQ-9 <5 and 42% GAD-7 <5 by end of treatment. A later 2×2 factorial RCT found PHQ-9 and GAD-7 fell across arms; coaching produced larger GAD-7 reductions than self-guided care, while weekly recommendations strengthened PHQ-9 gains and raised median sessions (median 216; last use day 56). Coaching did not clearly beat self-guidance on depression.
  4. PHQ-9 can stay flat. New dementia carers on Talking Point showed no PHQ-9 or GAD-7 change at 12 weeks (61 provided outcomes); 17 of 58 never visited the forum. Among users, more time correlated with greater PHQ reduction, but most individuals showed no reliable symptom change. No control group and low engagement limit any claim that forums treat carer depression.

Largest unresolved question

Mean PHQ-9 change, response, remission and 'score <5' are different claims. Kokoro-app's 2.5-point ITT advantage came without a significant remission difference. IntelliCare's uncontrolled 37% below 5 is a threshold count, not a randomised remission rate. iCBT's 51% versus 18% is reliable improvement among completers with data, not the same as g=0.78 on the full ITT sample. Treating those percentages as one 'PHQ-9 worked' statistic collapses the disagreements the papers actually report.

Common misconceptions

  • PHQ-9 is a treatment for depression or anxiety.

    It is a questionnaire. Kokoro-app, iCBT, IntelliCare and Talking Point are interventions; PHQ-9 is how several of them scored symptoms. IMS even uses PHQ-9 as a reference, not as care.

  • If 37% of users reach PHQ-9 <5, the app remitted depression in 37% of patients.

    That figure is from a single-arm IntelliCare field trial with coaching and no control. Expectancy and concurrent care remain alternative explanations. Kokoro-app's randomised 2.5-point gap did not even carry a significant remission contrast.

  • A significant mean PHQ-9 difference is the same as clinical recovery, and PHQ-9 always tracks anxiety.

    Kokoro-app separated response from remission. iCBT's reliable-improvement gap (51% vs 18%) is a completer statistic alongside g=0.78. Coaching in the factorial trial moved GAD-7 more clearly than PHQ-9, so depression-scale success is not automatic anxiety-scale success.

Related