When to Switch or Stay: NMDA Therapy Discontinuation Rates

The Clinical Problem: Waiting Too Long to Act

A new analysis published in Psychiatric Times (April 2026) examines the side effect profiles and discontinuation rates of NMDA receptor antagonists — primarily IV ketamine and intranasal esketamine (Spravato) — in patients with treatment-resistant depression (TRD). The core message from clinicians: the field has been too slow to optimize doses and switch agents when initial protocols aren't working, and better tracking tools like the PHQ-9 need to become standard practice, not an afterthought.

For readers researching low-dose ketamine, this analysis matters for a specific reason. Discontinuation decisions — when a patient stops treatment — are often driven by side effects that are manageable if caught early, or by lack of response that might be corrected with dose adjustment. Understanding what the real-world data shows about why patients leave treatment can help you have more informed conversations with your prescriber and set more realistic expectations before you begin.

What the Side Effect Data Actually Shows

Across clinical trials and observational studies of NMDA antagonists, the most commonly reported side effects fall into a predictable cluster: dissociation, dizziness, nausea, and transient elevations in blood pressure. These effects are dose-dependent and, critically, are almost always time-limited — typically peaking around 40 minutes after IV administration or intranasal dosing and resolving within one to two hours. This temporal pattern is one reason low-dose protocols attract interest: attenuating the dissociative peak while preserving antidepressant signal is an active area of clinical refinement.

Discontinuation rates due to adverse events in esketamine's pivotal trials hovered in the 10–15% range, with dissociation and dizziness cited most frequently. IV ketamine data from real-world infusion clinics shows similar patterns, though the rates vary considerably depending on dose (typically 0.5 mg/kg over 40 minutes in standard protocols) and patient population. What the Psychiatric Times analysis underscores is that many of these discontinuations may not have been inevitable — structured side effect monitoring and proactive dose adjustments could have retained a meaningful portion of those patients in effective treatment.

There is also a subtler category worth noting: patients who don't officially discontinue but quietly disengage — missing sessions, extending intervals beyond therapeutic benefit, or never completing an adequate induction series. This informal dropout pattern is harder to track but clinically significant, particularly in ketamine where the evidence base for antidepressant maintenance depends on sustained treatment engagement over time.

The Case for Faster, Data-Driven Decision-Making

The Psychiatric Times piece positions clinicians as needing to move more decisively: optimize doses when response is partial, switch agents sooner when response is absent, and use validated outcome measures like the PHQ-9 to make those calls with data rather than intuition. This framework has direct implications for how ketamine treatment is structured in practice.

In the current landscape, there is meaningful variation in how infusion clinics approach non-responders. Some extend the induction series from the standard six infusions to eight or ten. Others adjust dose upward within safety parameters. Others add adjunctive medications — most commonly oral ketamine, memantine, or traditional antidepressants — to support or extend response. What is less common, but what the data increasingly supports, is a formal decision point built into the protocol: if PHQ-9 scores haven't moved by a specified threshold after a defined number of sessions, a structured conversation about dose optimization or treatment switching should happen automatically rather than waiting for the patient to raise it.

This matters most for patients who are paying out of pocket for IV ketamine infusions. Without a monitoring framework that triggers timely adjustments, it is easy to continue a protocol that isn't working simply because stopping feels like admitting defeat. The evidence base suggests that early, data-informed pivots — up, down, or laterally to a different agent — produce better long-term remission outcomes than passive continuation.

What This Means for Low-Dose Protocols Specifically

Low-dose ketamine — whether administered via infusion at sub-standard doses, oral troches, sublingual formulations, or intramuscular injection — occupies a space where the side effect calculus shifts. The dissociative and hemodynamic effects that drive discontinuation at 0.5 mg/kg IV are generally attenuated at lower doses. This is part of the clinical rationale for low-dose approaches: reduce the tolerability barrier while still engaging the NMDA antagonism and downstream neuroplasticity mechanisms believed to underlie antidepressant effect.

However, lower dose does not mean no monitoring is needed. The Psychiatric Times analysis reinforces that systematic outcome tracking — not just symptom check-ins, but structured use of instruments like the PHQ-9 or MADRS — is what separates clinics operating on evidence from those operating on anecdote. For low-dose protocols in particular, where the therapeutic window may be narrower and the dose-response relationship less well characterized than for standard IV infusion, this kind of structured measurement is arguably more important, not less.

The practical implication: if you are working with a provider on a low-dose maintenance protocol, the same decision-making framework applies. How is response being measured? At what point would your provider adjust your dose, interval, or route of administration? How are side effects — even mild ones — being documented over time? These are not bureaucratic questions; they are the clinical levers that determine whether a treatment course reaches remission or drifts into an ambiguous middle ground that serves no one well.

The full analysis is available at Psychiatric Times. The evidence base for NMDA antagonists in TRD continues to mature; what's changing now is less the pharmacology and more the clinical infrastructure around how to use these tools rigorously.