Skip to content
Back to the blog Blog

Why a questionnaire isn't enough to detect mental health problems in primary care

More than half of potential mental disorders go unnoticed in primary care visits. We review what screening questionnaires offer, where they fall short, and what a conversation adds when conducted with clinical rigor.

For most people, the first place they turn when something isn't right emotionally isn't a psychologist's or psychiatrist's office. It's their family doctor's office. Sometimes they come in talking about anxiety or sadness. More often, they talk about sleeping poorly, always feeling tired, or pain that won't go away.

There, during a visit with little time and many concerns to address at once, something important is decided: whether that distress is detected early or goes unnamed for months. Screening questionnaires exist to help detect it. They're useful, but they're not enough. In this article, we explain why and what a well-conducted conversation can offer.

Mental health care starts in primary care

According to data from Spain's National Health System, mental health problems affect 20% of the population age 15 and older and are the third most frequently recorded chronic health problem in primary care, behind only high cholesterol and hypertension.

Primary care is the natural entry point. It's the most accessible level of care, the one that knows each person's history and can support them over time. It's also the one under the most pressure: short visits, packed schedules, and patients coming in with several concerns in a single visit.

More than half of cases go unnoticed

A study published in the journal Atención Primaria measured how well family doctors identify mental disorders in their patients. It analyzed 628 people ages 18 to 65 seen at health centers in Madrid. First, they completed screening questionnaires, and then potential cases were confirmed through a structured diagnostic interview.

The results were clear:

  • Nearly one in three people (31.7%) had a potential psychiatric disorder.
  • The family doctor had identified 44.2% of those cases.
  • The remaining 55.8% had not been detected during the visit.
  • Agreement between the doctor's assessment and the diagnostic interview was weak.

Not all problems were missed at the same rate. Some depressive conditions, such as dysthymia or melancholic depression, were identified more successfully. Problematic alcohol use and bulimia were among the least detected: precisely the conditions people often find hardest to discuss.

The data were collected in 2010, but the message remains relevant because the reasons emotional distress goes unnoticed are still present:

  • Lack of time. Exploring someone's emotional state requires open-ended questions and time to listen to the answers.
  • Symptoms in disguise. Insomnia, fatigue, pain, or digestive problems can be how an emotional problem manifests.
  • Stigma. Many people don't mention what they're going through unless asked directly, and sometimes not even then.
  • Several concerns in one visit. When lab results need reviewing, a prescription needs renewing, and knee pain needs addressing, emotional concerns get left for another day.

What a questionnaire does well

Screening questionnaires were developed to help address this problem. Tools such as the PHQ-9 for depression or the GAD-7 for anxiety have real advantages:

  • They're quick. They take just a few minutes to complete.
  • They're validated. How they work and what each score means have been studied.
  • They're comparable. They allow a person's progress to be tracked using the same measure.
  • They structure the assessment. They help ensure important questions aren't overlooked.

That's why they're included in many clinical guidelines, and it makes sense for them to remain there. The question is whether they're enough on their own.

Where a questionnaire falls short

Questionnaires have known limitations that are worth keeping in mind:

  • They only ask what they ask. A scale doesn't capture context: a recent bereavement, a separation, loneliness, a problem at work, or the burden of caring for a family member.
  • They depend on how they're understood. Phrases such as “little interest or pleasure in doing things” don't mean the same thing to everyone. Reading comprehension, age, and language influence the response.
  • They're a snapshot. They reflect how the person is doing on the day they fill them out, not how things have changed since the last visit.
  • They encourage rushed responses. When faced with a form, many people select the middle option or the one they think is expected of them, especially on topics associated with stigma.
  • They don't confirm anything on their own. A cutoff score separates what's likely from what's unlikely, but the result always requires a professional's assessment.

The Madrid study itself illustrates this well: the questionnaires helped identify potential cases, but a clinical interview was needed to confirm them. The questionnaire opens the door. Understanding what's happening requires a conversation.

What a conversation adds

When people can describe how they're feeling in their own words, information emerges that a form doesn't capture. A conversation makes it possible to:

  • Ask follow-up questions based on the answer. If someone says they're sleeping poorly, you can explore how long it's been happening, what's keeping them awake, and how it affects their daily life.
  • Capture the context. What's changed in their life, who they can rely on, and what's worrying them.
  • Adapt to each person. By voice or in writing, at their own pace and in their own language—something especially important for older adults or people less comfortable with digital tools.
  • Reduce pressure. Sharing something difficult is easier without the time pressure of an office visit.
Questionnaire Conversation
Questions Fixed, the same for everyone Adapt to what the person shares
Context Not captured Part of the response
Language The questionnaire's The person's
Accessibility Requires reading and writing Voice or text, in multiple languages
Result A score A score with context

The challenge: conversation without losing rigor

An open-ended conversation has a drawback: it's difficult to compare and measure. Two interviews with the same person can take different directions, and what gets captured depends heavily on who asks the questions and how. In mental health, where an assessment guides clinical decisions, that variability is unacceptable.

The key is to combine the best of both worlds: a natural conversation for the person and, underneath it, consistent clinical criteria that turn what they share into reliable information for the professional.

How we do it at Nexi Health

At Nexi Health, we're developing Bruna, a conversational AI for mental health, with one clear principle: the person speaks naturally, but the clinical process is never improvised.

  • The conversation is open-ended; the assessment isn't. Bruna asks questions tactfully and adapts to what each person shares, but the assessment follows validated rules and instruments. Given the same information, the result is always the same.
  • Everything is traceable. The professional can review what was asked, what was answered, and the basis for each conclusion.
  • The professional decides. Bruna provides a suspected diagnosis, the symptoms recorded, and follow-up recommendations. The doctor assesses and decides.
  • Before and between visits. The person can talk to Bruna before the visit, so the professional comes in with more context, and also between visits to track their progress.
  • Accessible. By voice or text, from a mobile phone or browser, and in more than ten languages.
  • Integrated into the health center. It works as a web application or connects to the electronic health record through HL7/FHIR, with data hosted and processed in the European Union.

If you'd like to know what requirements a tool like this must meet before being used with patients, we explain them in how an artificial intelligence tool is clinically validated.

Listen earlier to detect better

Questionnaires will remain necessary: they provide structure, measure symptoms, and allow comparisons. But in primary care, where timely detection of distress is at stake, more than a score is needed. Listening is essential—and it needs to happen before the person arrives for the visit.

The challenges raised at the SEMG National Congress are still present: greater demand, less time, and professionals who need to come to each visit with more information. Technology can help address them if it's designed to support professionals, not replace them.

References

  1. Garrido-Elustondo S, Reneses B, Navalón A, Martín O, Ramos I, Fuentes M. Family physicians' ability to detect psychiatric disorders. Atención Primaria. 2016;48(7):449-457.
  2. Ministry of Health. Key data from the National Health System. 2025.
  3. Kroenke K, Spitzer RL, Williams JBW. The PHQ-9: validity of a brief depression severity measure. Journal of General Internal Medicine. 2001;16(9):606-613.

See what we are building

Explore how Bruna, Nexi Health's conversational AI, supports people's mental health.

Latest articles

Don't miss a thing.

News on AI in mental health, product updates and clinical findings, straight to your inbox.

By subscribing you accept the privacy policy.