“The PHQ-9 Came Down” Is Not a Response

A falling score is a signal, not a response or a remission.

An outpatient psychiatrist opens a follow-up note. Last visit the PHQ-9 was 18. Today it is 12. The assessment line reads “responding to treatment.” The team huddle uses the same word. No paper is named. No interval is written. The difficulty item is blank. Item 9 is scored and then left there. The number moved. The word was added.

This fictional composite is not a patient case. It is a score change treated as a verdict.

A score can fall. The chart still has to say what that fall was called, and by whom.

Sources reviewed August 19, 2026, with the EMA guideline’s Rev. 3 status rechecked August 22 and 23, 2026: the 2001 PHQ-9 validation paper, the 2002 Kroenke and Spitzer review that states the authors’ rules of thumb, the developers’ instruction manual, the APA 2010 major-depressive-disorder guideline on measurement, the 2022 VA/DoD guideline’s PHQ definitions, and the EMA 2025 depression-trial guideline on “response” on a usual rating scale. This is not a scoring manual, not a treatment algorithm, and not a measurement-based-care manifesto.

This article is not about passive suicidal ideation or outpatient disposition. That is Passive Suicidal Ideation Is Not a Disposition. Item 9 is a scored item. A scored item is not a disposition. The question here is narrower: once a PHQ-9 total is lower than last time, what has actually been shown?

Kroenke, Spitzer, and Williams, in the Journal of General Internal Medicine in 2001, describe the Patient Health Questionnaire as a self-administered version of the PRIME-MD diagnostic instrument for common mental disorders. The PHQ-9 is the depression module. Each of the nine DSM-IV criteria is scored from 0 (not at all) to 3 (nearly every day). The total ranges from 0 to 27. The same nine items, they write, can establish depressive-disorder diagnoses and grade depressive symptom severity. The paper analyzed 6,000 patients in primary care and obstetrics-gynecology clinics, with criterion validity against 580 mental-health-professional interviews. It concludes that the PHQ-9 is also a reliable and valid measure of depression severity, and that those characteristics plus its brevity make it a useful clinical and research tool.

That is what the opened paper says the instrument is. It does not say that a falling total is response.

The 2001 paper is cross-sectional. The authors state that longitudinal studies are needed to establish sensitivity to change. They offer a preliminary approach: a PHQ-9 score less than 10 and a 50 percent decline from the pretreatment score as clinically significant improvement. Any proposed threshold, they write, requires prospective verification. They note that this approach would be consistent with that established for the Hamilton Rating Scale for Depression. That sentence sits in a limitations paragraph. It is not a finding of the study, and it is not a definition of remission.

For most analyses they divided scores into 0–4, 5–9, 10–14, 15–19, and 20 or greater. Table 2 labels those intervals minimal, mild, moderate, moderately severe, and severe. The Discussion says scores of 5, 10, 15, and 20 represent valid and easy-to-remember thresholds demarcating the lower limits of those four severity levels. Those are severity bands printed in that paper. They are not response criteria.

The developers’ instruction manual scores the PHQ-9 the same way and repeats the 5, 10, 15, and 20 cutpoints. It says the PHQ-9 is used as a severity measure and as an outcome tool to determine treatment response. It does not, on the pages opened, give a numeric threshold for response or remission.

The American Psychiatric Association’s 2010 major-depressive-disorder guideline lists the PHQ-9 among scales that can help determine trajectory and effects of treatment. Self-rated scales, it says, require review, interpretation, and discussion with the patient. Integrating such measurements is recommended with moderate clinical confidence. The guideline does not define PHQ-9 response or remission. DSM is not used here as a source for those words.

Eighteen to twelve is a six-point change. In the 2001 Table 2 bands, 18 sits in moderately severe (15–19) and 12 sits in moderate (10–14). The total moved one severity band. The note said “responding.”

A point change is a difference between two numbers. It is not, by itself, proof that treatment worked. It is not recovery. It is not a HAM-D result. It is not a protocol endpoint unless a protocol that was opened says so.

The European Medicines Agency’s January 2025 guideline on clinical investigation of medicinal products in depression states that usually an improvement of 50 percent or more on a usual depression rating scale is applied to define individual treatment response. Acceptable scales named there are the Hamilton Rating Scale of Depression, preferably the 17-item, and the Montgomery–Åsberg Depression Rating Scale. Criteria for response and remission must be pre-specified and justified in the study protocol.

That 50 percent rule is EMA trial language on a usual rating scale, with HAM-D and MADRS named. It is not a PHQ-9 rule. Do not silently apply it to a PHQ-9 total.

The words have definitions in named papers. They are not interchangeable, and they are not a universal clinical law.

Kroenke and Spitzer, in Psychiatric Annals in 2002, write that they currently consider a decline of at least 5 points necessary to qualify as a clinically significant response, that a score less than 10 qualifies as a partial response, and that a score less than 5 qualifies as remission. The next sentence: these numbers are obviously simple rules of thumb that require clinical evaluation of the individual patient.

That is the 2002 paper’s rule of thumb. It is not the 2001 preliminary approach, which required both a score less than 10 and a 50 percent decline. The two opened Kroenke papers do not use the same sentence.

The 2022 VA/DoD clinical practice guideline, Appendix I, page 141, states its own list. Response: PHQ score improvement of ≥50 percent from baseline. Remission: PHQ score of ≤4, maintained for at least one month. Recovery: PHQ score of ≤4, maintained for at least six months. Partial response: less than 50 percent improvement. Page 127 of the same appendix says the PHQ-9 is an indicator of severity or response to treatment and should not be used in isolation to make a diagnosis. The guideline’s qualifying statements say it is not intended to define a standard of care.

Eighteen to twelve clears the 5-point decline the 2002 paper calls necessary to qualify as a clinically significant response. Read strictly, clearing a necessary condition is not meeting a definition. Read the way the rule of thumb is commonly applied, the six-point decline would be called a response under the 2002 paper — and 12 is not the score below 10 that same paper calls a partial response. Under the 2022 VA/DoD definitions, the same decline is 33 percent, below the 50 percent line, and falls under partial response, not response. It meets neither requirement of the 2001 preliminary approach. It is not a score less than 5, or ≤4, and nothing in the composite was maintained for a month or six months. The same score pair takes different labels depending on which opened document supplies the word. The huddle named none of them.

This is evidence, not a ranking. If a chart uses “response” as a defined outcome, it has to make the operative definition identifiable. If none is identifiable, “responding” is an interpretation sitting on a subtraction.

Operational questions, not treatment advice.

The instrument asks how often the person has been bothered over the last two weeks. Two totals without an interval are two numbers, not a change over a stated window.

The total is the sum of nine items. A number built from eight items, or filled in later by someone else, is not the 0-to-27 score those papers describe.

The 2001 paper includes a tenth question: how difficult the problems have made it to do work, take care of things at home, or get along with other people. The instruction manual states that this difficulty item is not used in calculating any PHQ score. A falling total with a blank difficulty item has not recorded function.

Item 9 is one of the nine scored symptoms. The 2001 diagnostic algorithm counts it if present at all. The manual says a final decision about actual risk of self-harm requires a clinical interview. This article does not decide disposition. A total can fall while item 9 is still present, still unscored, or still uninterpreted.

Who administered each score is not in a naked total. Kroenke 2001 used a patient-completed form in clinic. A portal form, a nurse read-aloud, and a psychiatrist filling boxes after the visit are not the same act.

What treatment was in place at each score is not in the subtraction. This article does not say to start, stop, or switch a drug. Writing “responding” does not record what was being done.

Operational failures, not universal legal conclusions.

A subtraction is not a defined outcome unless the note makes the operative definition identifiable.

Table 2 classifies severity. The Discussion’s 5, 10, 15, and 20 thresholds demarcate severity. They do not define response.

Kroenke and Spitzer called those numbers simple rules of thumb that require clinical evaluation of the individual patient.

EMA’s 50 percent language is on a usual depression rating scale, with HDRS and MADRS named. It is not, in that guideline, a PHQ-9 rule.

That list is the 2022 VA/DoD appendix. It is not a default for every outpatient note.

The manual keeps the difficulty item out of the score. Item 9 is a symptom item. Neither is the huddle word “responding.”

Companion worksheet: present / absent for the named definition and its paper, the interval, nine completed items, the difficulty item, item 9 interpreted, who administered each score, and what treatment was in place. The worksheet is not a scoring manual and does not authorize a treatment change.

If any answer is unknown, do not treat the falling total as a response.

Educational Disclaimer: The Psychiatric Record provides general educational information for psychiatric and mental-health professionals. Content does not constitute medical, legal, regulatory, compliance, billing, or other professional advice; does not establish a standard of care; and is not a substitute for independent professional judgment. Requirements and appropriate practices may vary by jurisdiction and circumstance. Verify current authoritative sources.

This article is not a scoring manual and not individualized treatment advice. It does not recommend a drug, a switch, or a measurement program.