Changes in depression questionnaire scores often disagree markedly with patients’ own ratings of changes in their mood, a prospective cohort study has found.

Patients were recruited from GP surgeries in three English cities (Bristol, Liverpool and York). Inclusion criteria were: age 18-70; recorded depression symptoms or depression diagnosis within the past 12 months; no history of bipolar disorder, psychosis, eating disorder or substance abuse.

Of 7,721 potential participants who were contacted by letter, a total of 554 (68% women), mean age 48 years, were eventually enrolled in the study. At baseline, 377 (69%) were taking antidepressants.

Participants were assessed at baseline and at 2, 4 and 6 weeks’ follow-up. At each assessment they were asked to complete the patient health questionnaire (PHQ-9) and the Beck depression inventory (BDI-II).

They were also asked a patient global impression question, namely: ‘compared to when we last saw you 2 weeks ago how have your moods and feelings changed?’ The Likert-type response options were: feeling a lot better; feeling slightly better; feeling about the same; feeling slightly worse; feeling a lot worse.

The minimal clinically important difference (MCID) in rating scores was defined as 20%. At each follow-up, rating scores were categorised as: clinically important decrease (≥ 20% decline over the past 2 weeks); no clinically important change (decline or increase of < 20%); clinically important increase (≥ 20% increase).

 At the 2-week follow-up there was disagreement between the questionnaire score changes and  patients’ self-rated mood changes in more than half of cases:

• 51% (95% CI: 46-55%) of the PHQ-9 score change categories

• 55% (95% CI: 51-60%) of the BDI-II categories

For the PHQ-9 a different conclusion from the rating scale was found in:

• 53% of patients reporting that they felt a lot better

• 65% of those feeling slightly better

• 30% of those reporting no change

• 75% of those feeling slightly worse

• 82% of those feeling a lot worse

Similar results were obtained at the 4-week and 6-week follow-up assessments.

An increase in anxiety symptoms measured with the 7-item generalised anxiety disorder assessment (GAD-7) was associated with decreased odds of reporting feeling better.

An improvement in either the physical or mental health-related quality of life scores (SF-12) was associated with increased odds of reporting feeling better.

Both NICE (‘questionnaires may be useful to monitor the person’s progress’)¹ and the British Association for Psychopharmacology (‘case management should include standardised assessment of symptoms’)² recommend the use of depression rating scales when assessing response to treatment. However, GPs consider their clinical judgement to be more important than symptom scores³ and a systematic review found conflicting evidence on whether follow-up severity scores influenced management.⁴ The present study casts further doubt on the utility of questionnaire monitoring.

In my opinion, the main argument in favour of questionnaires is that they are an important component of case management and collaborative care which have been consistently shown to improve both treatment adherence and outcomes.²

Questionnaires can also help to structure an assessment according to the DSM criteria.

However, they are no substitute for clinical judgement and the study authors recommend that clinicians should be cautious in interpreting changes in questionnaire scores without further clinical assessment.

This should include an assessment of comorbid anxiety symptoms and should give due weight to the patient’s own assessment of their progress.

Dr Phillip Bland