A low relapse rate can sound like a strong reason to choose treatment. Without a definition, it may say very little. Who was followed? What counted as relapse? Did observation last a month or a year? A major depression trial illustrates why those details belong beside the percentage, not in an unread footnote.
Key findings
478: primary-care patients randomized in the ANTLER trial of long-term antidepressant maintenance versus discontinuation.
39% versus 56%: participants experiencing relapse during 52 weeks in the maintenance and discontinuation groups, respectively.
17 percentage points: the absolute difference between those published group percentages.
2021: the trial’s reporting year. These are historical results from a selected depression population, not a current addiction relapse rate or an individual treatment prediction. [1]
One trial, two strategies, one year
The comparison concerns two defined medication strategies, not two clinics. Source: ANTLER report. [1]
Participants were well enough to consider stopping long-term antidepressant treatment. The trial assigned 238 to maintenance and 240 to discontinuation, which involved gradual reduction followed by placebo.
The result is informative because the population, strategies, and observation period are specified. It is not a universal relapse rate for depression. It also cannot establish what would happen with every stopping schedule or among patients with a different treatment history.
The same result can produce three different headlines
| Description | Calculation | Meaning |
|---|---|---|
| Absolute difference | 56% minus 39% = 17 percentage points | Difference between the group percentages |
| Relative increase | 17 divided by 39, approximately 44% | Discontinuation relative to maintenance |
| Relative reduction | 17 divided by 56, approximately 30% | Maintenance relative to discontinuation |
The 44% increase and 30% reduction describe the same pair of percentages from different reference points. Neither is a 44-point or 30-point difference.
These simple calculations use rounded published values. They do not replace the trial’s statistical models or uncertainty estimates. Whenever a relative figure is used, the absolute rates should remain visible.
That principle applies to treatment advertising. A claim that a program reduces relapse should name the comparator, the population, and the period. Without them, the headline invites a conclusion the evidence may not support.
Relapse and withdrawal need different questions
ANTLER reported more withdrawal symptoms in the discontinuation group as well as a difference in relapse. NICE recommends considering withdrawal when symptoms emerge after reducing or stopping an antidepressant. Assessment and an individualized, monitored plan matter. [1] [2]
A headline cannot determine which explanation applies to an individual. Automatically labeling every symptom after a treatment change as relapse can distort the clinical picture. Automatically dismissing it as withdrawal would be another mistake.
This article is not a reason to change medication independently. Someone considering a change should discuss the benefits, risks, preferences, and treatment history with the prescriber. The trial does not tell everyone to continue or everyone to stop.
One episode and a sustained return are not the same endpoint
Substance-use studies may measure a first episode of use, renewed heavy use, abstinent days, or a longer-term pattern. Each answers a different question. A report should explain both the event and how it was identified.
A review of continuing care describes varied outcomes, including abstinent days, risky drinking, and re-entry to treatment. That variation is a reason to resist a universal relapse statistic. [3]
An isolated event can still carry serious risk. Its significance depends on the substance and the person’s circumstances. Defining an endpoint accurately is not the same as minimizing it.
Longer follow-up can raise the number without proving worse care
A one-month rate and a one-year rate describe different opportunities to observe recurrence. Comparing them as though they used the same window can reward the service that looked for less time.
The starting point matters too. Observation may begin at admission, discharge, initial improvement, or the end of a maintenance phase. A fair comparison aligns those periods or uses an appropriate analysis of time to an event.
For someone returning to a leadership role or family responsibilities, the relevant question is whether the study covers life after treatment. A discharge result may be useful while leaving that later period unobserved.
Who enters the denominator can change the story
Some rates include everyone admitted. Others include only treatment completers, people who first achieved remission, or respondents to a later survey. Each can be valid when labeled correctly. They cannot silently stand in for one another.
Missing follow-up is unknown. It is neither proof of relapse nor evidence that recovery continued. A useful report gives the number eligible, the number actually assessed, and the approach used for missing outcomes.
Starting severity and previous treatment also matter. A service accepting more complex patients should not be judged only against a raw percentage from a narrowly selected group. The denominator describes clinical context, not just arithmetic.
Returning to care can be the right response
Continuing care can aim to identify renewed need and respond earlier. Under that model, appropriate treatment contacts may increase alongside better overall outcomes. Counting every return as failure could penalize the intended response.
For an internationally mobile patient, the plan should specify a contact, an escalation route, and the receiving clinician where relevant. The label aftercare is not enough to establish those arrangements.
The arrangements still need evaluation. A written plan does not prove it was followed, just as extra contact does not guarantee benefit. Follow-up should examine the person’s course rather than count appointments alone.
What affluent families should request
The ANTLER participants were not recruited as an executive or high-net-worth sample. Its percentages should not be presented as the relapse risk of wealthy professionals.
The client-relevant lesson concerns informed choice. A private program’s setting, discretion, and fee do not explain how its outcome percentage was calculated. Ask for the actual population and follow-up behind any claimed advantage.
| Disclosure | Question answered |
|---|---|
| Event definition | What counted as relapse? |
| Population | Who was eligible and included? |
| Treatment | What did each group receive? |
| Time window | When did observation start and end? |
| Assessment | How were events identified? |
| Missing information | Who was not reached and how was that handled? |
| Uncertainty | How precise is the comparison? |
This is a proposed disclosure framework, not a validated ranking of clinics. It applies to THE BALANCE as well as other providers.
What the numbers cannot tell us
The trial does not compare public and private clinics, measure addiction recovery, or predict an individual outcome. Its arithmetic does not establish the effect of every medication schedule. A difference between published rates from separate providers would need its own defensible comparison.
Nor should an outcome become an identity. Someone experiencing renewed symptoms needs assessment and appropriate support. Accurate reporting can acknowledge a difficult event without labeling the person a failure.
The bottom line
Ask what a relapse percentage contains before asking whether it is low. Population, treatment, definition, and follow-up are part of the finding. The most useful provider is not necessarily the one advertising the smallest number, but the one that explains it and has a clear response when further care is needed.
For journalists
Key comparison: 39% versus 56% relapse during 52 weeks in the 2021 ANTLER trial, an absolute difference of 17 percentage points.
Important caveat: selected primary-care depression patients, not an addiction statistic or HNW sample.
Suggested attribution: THE BALANCE analysis of published ANTLER results and outcome definitions.
Methodology and sources
This narrative analysis uses published trial findings, clinical guidance, and a continuing-care review. Relative changes are calculated from rounded published percentages. No new patient dataset or provider comparison was analyzed.
