Stepped Care for Anxiety: Does Starting With Less Intensive Treatment Work?

Treatment for anxiety exists on a spectrum of intensity: from a self-help book at one end to specialist inpatient care at the other. Stepped care models formalize this spectrum into a pathway, starting everyone at the least intensive level and moving up only as needed. The promise is efficiency — most people should recover without expensive one-to-one therapy. The question is whether the efficiency comes at a cost.

How the model works

The stepped care model typically has three to five steps:

Step 1: Recognition and active monitoring. Identification of the problem, psychoeducation, and watchful waiting to see if symptoms resolve without treatment.

Step 2: Low-intensity interventions. Guided self-help (workbooks or online programs with brief therapist support), computerized CBT, psychoeducational groups, or structured exercise programs.

Step 3: High-intensity interventions. Individual CBT (typically 12–16 sessions), other evidence-based psychotherapy, or medication.

Step 4: Specialist treatment. Complex interventions for people who have not responded, including combined treatments, specialist services, or day programs.

Patients are assigned to the lowest appropriate step based on initial assessment. Those who do not improve after an adequate trial at one step are “stepped up” to the next. NICE clinical guidelines recommend this model for generalized anxiety disorder, panic disorder, and social anxiety disorder [1].

The evidence base

The evidence for stepped care specifically for anxiety disorders is thinner than you might expect given how widely the model is implemented.

Systematic evidence. A systematic review and meta-analysis of stepped care for depression found that stepped care was comparable to other delivery models in outcomes, with some evidence of cost savings [2]. Evidence specific to anxiety disorders is sparser. A Dutch trial randomized primary care patients with anxiety and depression to stepped care versus care as usual, finding modest advantages for stepped care on some outcomes but not all [4].

The theoretical case. The strongest argument for stepped care is logical rather than empirical: not everyone with anxiety needs 16 sessions of individual therapy. If 50% recover with guided self-help, offering everyone individual therapy wastes half the resources. Bower and Gilbody laid out this efficiency argument clearly, noting that the model’s value depends on the proportion of patients who respond at each step [5].

IAPT as a natural experiment. The Improving Access to Psychological Therapies program in England is the largest implementation of stepped care for anxiety and depression in the world. It has treated over one million people per year, with publicly reported outcome data [6]. Recovery rates hover around 50%, and reliable improvement rates are higher. These are real-world figures, not trial results, and they represent a massive scaling of evidence-based treatment.

However, IAPT is not a randomized trial of stepped care. There is no comparison arm where the same population receives a different delivery model. The outcomes tell us that stepped care can be implemented at scale with reasonable results, but not whether a different model would do better or worse.

Where the model is uncertain

Step-up failure. A significant number of patients who do not respond at Step 2 do not step up to Step 3. They drop out of the system entirely. In IAPT data, a substantial proportion of people referred to low-intensity treatment do not complete it, and not all of those who don’t respond are offered higher-intensity care [3]. The stepped model assumes that stepping up happens, and when it doesn’t, people who need more treatment don’t get it.

Delay. For patients who ultimately need high-intensity therapy, stepped care means they spend weeks or months in a lower-intensity treatment before reaching what they actually need. If the initial assessment could accurately predict who needs to start at a higher step, those patients could be matched directly. But prediction is poor (see predictors of therapy response).

Demoralization. Failing at a lower step may itself be harmful. A patient who tries guided self-help and doesn’t improve may conclude that they are too difficult to treat, that therapy doesn’t work for them, or that their problem is more severe than it is. This demoralization effect is theoretically plausible but has not been systematically studied.

One size versus matching. Stepped care is a one-size-fits-all pathway applied to everyone. An alternative model — stratified care — would use initial assessment to match patients directly to the right intensity. Stratified care requires accurate prediction of treatment response, which we do not yet have, so stepped care wins by default.

What the data from scale shows

The IAPT program provides the best available data on stepped care in practice. Key findings:

• Recovery rates (crossing from clinical to non-clinical range) are about 50%
• Access has expanded dramatically: over 1.2 million referrals in 2022–23
• Wait times vary by region, with significant inequities
• Low-intensity treatments (Step 2) account for the majority of completed treatment courses
• Patients with more complex presentations or comorbidities tend to have lower recovery rates

These figures are reasonable but not exceptional. The 50% recovery rate has been roughly stable since the program’s early years, despite ongoing quality improvement efforts. Whether a non-stepped model could achieve higher rates with the same resources is unknown.

How to think about this

Stepped care for anxiety is a pragmatic solution to a resource problem. It is not clearly superior to other models in clinical outcomes, but it is more efficient, allowing more people to access treatment within a constrained budget. The evidence base for the model itself is modest — it rests more on logical necessity and large-scale implementation data than on randomized comparisons of delivery models.

For more on the treatments delivered at each step, see CBT versus medication trials and online versus in-person therapy.

When to look for more support. If you have been offered a low-intensity treatment and it is not helping, ask about stepping up to individual therapy. Not responding to one level does not mean you will not respond to the next. Talk to your GP or the service that referred you. If you are in crisis, call or text 988 (US) or Samaritans on 116 123 (UK and Ireland).