Mental Health Assessment Instruments

Mental health instruments do three different jobs and are often used as though they did one. Screening tools such as the PHQ-2, the Edinburgh Postnatal Depression Scale and the SCOFF Questionnaire are built to be short and to over-identify rather than miss, so a positive result is a reason to assess further, not a diagnosis. Severity ratings such as the Hamilton scales, the Montgomery-Asberg Depression Rating Scale and the Brief Psychiatric Rating Scale are built to track change in someone whose diagnosis is already established. Wellbeing measures such as the WHO-5 Well-Being Index and the Warwick-Edinburgh Mental Well-Being Scale measure something else entirely, and a high score on one is not the absence of illness on another.

The library also holds instruments that sit at the boundary, such as the Perceived Stress Scale and the Copenhagen Burnout Inventory. These measure states that are real and measurable but are not clinical diagnoses, and treating a cut-off on them as a case definition is one of the more common errors in this area.

Clinician-rated and self-report instruments are not interchangeable even when they cover the same construct. Clinician-rated scales such as HAM-D and MADRS require trained raters and inter-rater reliability checks, and in multi-site studies rater drift is a recognised source of noise that has to be managed actively rather than assumed away. Self-report instruments avoid that problem and introduce another, because they are sensitive to how the question is asked, the mode it is asked in, and who else is in the room.

In research these instruments are usually collected on a fixed schedule, which raises a practical issue that matters more here than in most areas: repeated administration of a symptom questionnaire can itself affect the participant. Suicide risk items in particular need a defined escalation pathway written into the protocol before the first questionnaire is sent, not improvised when a score comes back.

Check the reference population first. Many of these instruments were developed and normed in one country, one language and one age band, and the cut-offs travel less well than the item wording does. A score that identifies probable depression in one population may perform quite differently in another, and using an imported cut-off without local validation is a measurement decision, not a neutral default.

Check the recall period second. Instruments that ask about the last two weeks, the last month and the last seven days cannot be compared with each other, and swapping one for another mid-study breaks the time series.

Check licensing and scoring rights third. Several instruments in this group are commercially licensed, and some require a qualified user level for purchase or scoring. Finally, be careful with abbreviated forms. A short version is a separate instrument with its own psychometrics, and its cut-offs are not the long version's cut-offs divided down.

What is the difference between a screening tool and a severity scale?

A screening tool is designed to find people who might have a condition, so it accepts false positives to avoid false negatives. A severity scale is designed to quantify how unwell someone already known to have the condition is, and to detect change over time. Using a screener to track treatment response, or a severity scale to screen a general population, gives poor results in both directions.

Can I use a self-report scale instead of a clinician-rated one?

Only if the study was designed for it. Self-report and clinician-rated instruments measuring the same construct correlate imperfectly and respond to treatment at different rates, so they cannot be substituted for one another partway through, and results from one should not be pooled with results from the other.

Do I need permission to use these instruments?

Often yes. Several widely used mental health instruments are commercially licensed, some restrict who may administer or score them, and translations are frequently licensed separately. Confirm the terms before data collection begins.

How should suicide risk items be handled in a digital questionnaire?

Decide in advance what happens when an item is endorsed, who is notified, how quickly, and what the participant sees on screen. This needs to be written into the protocol and the participant information, and the system has to be able to act on it in real time rather than at the next data export.