Measurement-based care: bringing PHQ-9 and GAD-7 to the workplace

Ask most organisations how their people are doing and you get a feeling. The last engagement survey looked fine. The town hall had good energy. Nobody has resigned this month. These are real signals, but they are the workplace equivalent of a doctor saying a patient looks well from across the room. Measurement-based care in the workplace asks for something harder and more useful: an actual number, tracked over time, that you can act on.

Clinicians have long known that feelings are a poor instrument. A therapist who relies on memory and impression can miss slow drift in either direction. The fix in clinical practice is to measure the thing you care about with a validated tool, repeat the measurement, and let the trend, not the vibe, drive what you do next. That is the whole idea behind a measurement-based care workplace programme that brings PHQ-9 and GAD-7 to work, and it is more rigorous than much of what People teams currently rely on.

Measurement-based care (MBC) is a clinical method, not a wellbeing slogan. It means using a validated, repeatable instrument to track a patient's symptoms at each contact, comparing the latest score against the last one, and adjusting treatment when the numbers are not moving. If depression scores stay flat after several sessions, that is a signal to change the plan rather than to wait and hope. The data informs the decision.

The two best-known instruments are the PHQ-9 and the GAD-7. The PHQ-9 is a nine-item screen for depression; the GAD-7 is a seven-item screen for anxiety. Both are short, both are free to use, both have been validated across large and diverse populations, and both produce a single score on a known scale that maps to severity bands. A person scoring 18 on the PHQ-9 is in a different place than one scoring 6, and a move from 18 to 9 over a couple of months is a measurable improvement rather than a hopeful impression.

The discipline matters more than any single questionnaire. The point is not that the PHQ-9 is magic. It is that you measure the same thing the same way, repeatedly, and you change course when the trend says you should. Research on MBC in mental health suggests that feeding validated scores back into care improves outcomes, in part because it catches people who are quietly not getting better.

The case for measurement at work is not sentimental, it is economic and it is large. The WHO estimates that depression and anxiety cost the global economy roughly US$1 trillion a year in lost productivity, and that around 12 billion working days are lost annually to these conditions. A WHO-led analysis published in The Lancet Psychiatry estimated that scaling up treatment returns roughly US$4 for every US$1 invested, largely through restored productivity and better health. Work by firms such as Deloitte on the employer cost of poor mental health points in the same direction: presenteeism, absenteeism and attrition carry a real and measurable bill.

In India that bill lands in specific, countable places. Replacing a skilled employee costs a meaningful fraction of annual salary; commonly cited estimates from sources like SHRM and Gallup put replacement costs at roughly one-half to two times salary depending on the role, though figures vary widely by study and seniority. Add the loaded cost of PF and ESIC contributions, the productivity dip while a backfill ramps, and the strain that shift work and long commutes place on people, and the financial logic of catching distress early becomes hard to argue with. The question is not whether wellbeing affects the business. It is whether you can see it before it shows up in the attrition report.

Most organisations cannot, because they measure engagement once a year and read it like a horoscope. Measurement-based care offers a more honest alternative: validated instruments, taken more than once, read as a trend.

Here is the line that has to be drawn clearly, because it is the difference between a useful programme and a creepy one. A therapist sees an individual's PHQ-9 score because they are treating that individual. An employer must never see an individual's score, full stop. The workplace version of measurement-based care only works on anonymised, aggregated data: cohort-level trends, never named individuals.

What an organisation can responsibly learn is directional and population-level. Aggregate anxiety in a particular function has risen quarter on quarter. A site that went through a brutal release cycle shows a cohort dip that has not recovered. Burnout signals, which the WHO describes in ICD-11 as a syndrome with three dimensions (energy depletion or exhaustion, increased mental distance or cynicism about one's job, and reduced professional efficacy), are concentrating somewhere specific. None of that requires knowing who. It only requires enough people in a cohort that no single person can be identified.

This is also where India's Digital Personal Data Protection Act 2023 stops being a compliance footnote and becomes design guidance. Health data is sensitive, consent should be meaningful, and aggregation with a minimum cohort threshold is the kind of practice that helps keep a wellbeing programme on the right side of both the law and basic trust. This is general design thinking rather than legal advice, and specifics should be checked with counsel. Get it wrong and participation collapses, which destroys the data anyway.

The clinical lesson that travels best is this: a score you do not act on is wasted. In therapy, a flat PHQ-9 across several sessions is a prompt to change the treatment. At work, a rising aggregate GAD-7 in one team is a prompt to ask what changed there, whether it was workload, a reorganisation, a manager transition, or a punishing on-call rotation, and to do something specific about it.

This is where psychological safety connects to measurement. The concept is most associated with Amy Edmondson's research, and Google's Project Aristotle reported it as the strongest of the factors behind effective teams. A measurement programme that people trust will be used honestly. A measurement programme people fear will be gamed or ignored, and the data will quietly become fiction. The two reinforce each other: safety makes the measurement real, and acting visibly on the measurement builds the safety.

None of this needs to be elaborate to start. A validated baseline, repeated at a sensible cadence, read at cohort level, with a named owner who acts on a worrying trend, already puts an organisation ahead of one that measures wellbeing by how the last all-hands felt. Platforms like AhaTherapy exist to run that loop properly, but the discipline is the point, not the tool.

Measurement-based care did not become standard in mental health because it was fashionable. It gained ground because tracking validated outcomes and adjusting to them tends to produce better results than relying on impression alone. The workplace inherits that lesson cleanly. You do not get to manage what you only sense. Bring the same instruments, the same repetition and the same willingness to change course to the question of how your people are doing, keep the individual data out of the employer's hands, and wellbeing stops being a feeling you defend in a budget meeting and becomes evidence you can actually act on.