NICE NG222, depression in adults: guideline appraisal and AGREE II scores

NG222 rebuilds the management of depression in adults around one structural simplification. The four traditional severity categories give way to two, less severe and more severe, with a numerical marker offered as an example: a score of 16 on the PHQ-9. Shared decision-making and the patient’s preference sit at the centre of the choice, and each level comes with a table of options that can all be used first line. Two messages stand out. In less severe depression, an antidepressant stops being a routine first reflex. And whenever an antidepressant is stopped, the taper should be slow, supervised and paced by the patient.

Version and corpus, worth knowing before use. The text read here is not the June 2022 version. It carries six amendments, two of which touch passages reported on this page. The October 2025 amendment clarified recommendation 1.5.3 on first-line treatment of less severe depression; the December 2025 amendment strengthened, in both tables, the requirement for regular reviews and the timing of the first review. Only the guideline text itself could be obtained. The evidence reviews, the methods document, the scope and the committee membership are not in the corpus, which leaves three of the six AGREE II domains unscored. The detail is given further down.

• Two severity levels replace the four traditional categories. The first brings together subthreshold and mild depression; the second explicitly covers moderate and severe depression. Thresholds on validated scales serve as indicators of severity, and NICE gives one as an example: a PHQ-9 score below 16 for the first level, 16 or above for the second. The terms section at the end of the recommendations, however, says more severe depression “was defined as” a score of 16 or more.
• Shared decision-making, the patient’s preference and their right to decline treatment are placed at the centre of treatment choice.

• The options in table 1 can all be used as first-line treatments. The table follows the committee’s reading of their clinical effectiveness, their cost effectiveness and implementation considerations, with the option judged most effective and most cost effective at the top, with one exception: guided self-help is placed first for pragmatic reasons, not because of its rank on effectiveness. NICE adds that some people will prefer an option lower down, and that such a preference is legitimate.
• Order of table 1, eleven options: Guided self-help, Group cognitive behavioural therapy (CBT), Group behavioural activation (BA), Individual CBT, Individual BA, Group exercise, Group mindfulness and meditation, Interpersonal psychotherapy (IPT), Selective serotonin reuptake inhibitors (SSRIs), Counselling, Short-term psychodynamic psychotherapy (STPP).
• Recommendation 1.5.3 sets out not to offer antidepressant medication routinely as first-line treatment, and to offer it only when it matches the person’s informed preference. The statement concerns antidepressants as a class, not SSRIs alone. The committee also judged that, if an antidepressant is used at this level of severity, SSRIs would be the preferred choice because of their safety and tolerability.

• Same principle: the ten options in table 2 are all first line, and the order rests on the same composite criteria. Order of table 2: Combination of individual cognitive behavioural therapy (CBT) and an antidepressant, Individual CBT, Individual behavioural activation (BA), Antidepressant medication, Individual problem-solving, Counselling, Short-term psychodynamic psychotherapy (STPP), Interpersonal psychotherapy (IPT), Guided self-help, Group exercise.
• In its rationale, the committee assigns good evidence of effectiveness jointly to three options: CBT combined with an antidepressant, individual CBT, and individual behavioural therapies, all three also appearing cost effective on average. Rank in the table is therefore not, on its own, a hierarchy of evidence.

• For more severe depression, the committee places SSRIs and SNRIs first line on the grounds of their tolerability. Two caveats come with this choice and tend to get lost in summaries: a tricyclic may be preferred for someone who responded well to one in the past and has no contraindication, and mirtazapine is kept out of first line, reserved for further-line treatment.
• On duration, the information to give the patient is that treatment may need to continue for at least six months after remission, with regular review. That second point is exactly what the December 2025 amendment strengthened in both tables. NICE sets no maximum length of treatment.
• On stopping, the taper proceeds in proportional steps, with the example of halving the dose and then finer steps of around a quarter, liquid preparations once doses become very small, and a pace led and agreed by the patient, over weeks or even months. Fluoxetine, with its long half-life, can sometimes be stopped by a period of alternate-day dosing at 20 mg. Paroxetine and venlafaxine carry a higher risk of withdrawal symptoms and call for particular care. Withdrawal symptoms should not be mistaken for relapse.

• When there has been no response after four weeks of an antidepressant at a recognised therapeutic dose, or after four to six weeks of psychological therapy or combined treatment, NICE asks first for a look at personal and contextual factors and at adherence, then, once those are dealt with, for a review of the diagnosis and of possible comorbid conditions.
• Electroconvulsive therapy is to be considered in severe depression under any of three conditions: the person’s choice, based on their past experience of ECT and on what has worked for them before; the need for a rapid response (life-threatening depression being the example given); or the failure of other treatments.
• Depression with psychotic symptoms has a section of its own in the guideline, with referral to specialist mental health services, risk assessment, a programme of coordinated multidisciplinary care and access to psychological treatments once acute symptoms have improved. Bipolar disorder is referred to a separate guideline.
• Esketamine nasal spray is not recommended, but that decision belongs to a technology appraisal distinct from NG222 and concerns treatment-resistant depression specifically. Repetitive transcranial magnetic stimulation is referred to the dedicated interventional procedures guidance.
• PEB has drawn a decision tree in two pathways from this guideline, from a new episode to stopping the antidepressant, with what the guideline does not say at each step.

Scoring method: each of the twenty-three AGREE II items was either scored on the 1 to 7 scale and tied to a passage read in the corpus declared above, or declared not assessable when the document that should have supported it is not in the corpus. The percentages shown are domain scores calculated with the instrument’s formula, (obtained score minus minimum possible score) divided by (maximum possible score minus minimum possible score), multiplied by one hundred. Declared deviations: a single appraiser, where the instrument calls for two to four, and three domains not scored, the required documents not being in the corpus. PEB colour bands applied to domain scores: excellent from 85%, very good from 70 to 84%, good from 55 to 69%, fair from 40 to 54%, weak below 40%. English version checked for conformity with the French page and the source on 15 September 2026.

Agreement on the most important point. NICE and the Haute Autorité de Santé (HAS), the French national health authority, both set aside the antidepressant as a first reflex in the least severe forms. The 2017 HAS guideline advises against prescribing an antidepressant in a mild major depressive episode or for subsyndromal symptoms, and proposes supportive psychotherapy with reassessment at four to eight weeks. NICE reaches the same position by another route, by placing non-pharmacological interventions at the top of a table of options. Two independent bodies, two health systems, five years apart, one conclusion.

Where the two texts part ways. Three differences, checked by automated search of the text of the French recommendations. The first concerns the architecture of non-pharmacological options: NICE provides an ordered table of eleven interventions, whereas the HAS text sticks to supportive psychotherapy and four structured families without ranking them. Guided self-help and behavioural activation do not appear there, and physical activity features only as lifestyle advice. The second concerns stopping antidepressants: NICE quantifies proportional steps, while the HAS keeps to a gradual reduction with no quantified step. The third concerns nomenclature: three severity levels in France, two at NICE, NICE explicitly placing moderate depression in the more severe category, which makes a direct mapping between the two decision trees impossible.

What this gap does not mean. It does not reflect a gap in the French evidence review. The 2017 HAS scientific argument devotes a whole chapter to structured physical activity programmes and concludes that they are recommended as an add-on treatment for a major depressive episode, and as monotherapy in mild to moderate forms. Behavioural activation and self-help are also discussed at length. These interventions were therefore read and assessed by the French working group, then not carried over into the text of the recommendations. The gap between the two guidelines is an editorial choice by the HAS, whose reasons are not set out.

French source for this comparison, added to the corpus: Haute Autorité de Santé. Épisode dépressif caractérisé de l’adulte : prise en charge en soins de premier recours. Text of the recommendations and scientific argument. October 2017. has-sante.fr (in French)

National Institute for Health and Care Excellence. Depression in adults: treatment and management. NICE guideline NG222. London: NICE; published 29 June 2022, amended six times up to December 2025.

Official link: https://www.nice.org.uk/guidance/ng222 · Version appraised: the text as it stands after December 2025 · Accessed on 19 August 2026.

What could not be verified. The evidence reviews and the methods document, not obtained, which leaves the search strategy, selection criteria, method for formulating recommendations and external review unverifiable. Committee membership and the collection of patient views. Declarations of interests and the funding statement. The guideline scope with its exclusion criteria. The visual summaries and implementation tools announced by the document. The technology appraisal on esketamine and the interventional procedures guidance on transcranial magnetic stimulation, cited by NG222 and not read: this appraisal attributes nothing to them beyond what NG222 says about them.