All readings

Learn

What the evidence tells us

What reviews and trials have found about family constellations, and why their limits keep this from being a proven treatment.

The short answer

There are few studies, nearly all small, and the most careful assessments conclude that the evidence is not enough to say family constellations work as a treatment or are safe for mental health problems. That does not show they are useless; it means nothing is established. A favorable result in one study is not the same as demonstrated effectiveness. 1 2

What was studied is a specific format: a group therapy held as a seminar, usually two or three days long, with up to 25 or 30 people who do not know one another. A person presenting a situation chooses others in the group to stand in for members of their family or social system, and a facilitator suggests adjustments to positions and conversations starting from that first arrangement. 2

Bert Hellinger (1925–2019) developed the method. He spent 16 years as a missionary for a Catholic order among the Zulus in South Africa and later trained as a psychoanalyst. His own school says he refined his approach over the years in seminars and practice. 3

What the 2021 review found

The systematic review, published in Family Process, searched several databases for studies through April 2020. Of 4,197 records, 12 studies met the criteria, covering 10 independent samples and 568 participants in all. What they measured varied widely, from self-image to psychological symptoms to perceived quality of family relationships. 1

Nine of the 12 studies showed a statistically significant improvement after the intervention; those that did not were of lower methodological quality. A meta-analysis of five studies on general psychopathology found a moderate effect. 1

Seven studies also looked for possible negative effects, and four reported mild or moderate ones in a small share of participants, between 5 and 8 percent, that might have been linked to the intervention. The authors concluded that the data point toward an effective intervention, but that the quantity and quality of the evidence are low. 1

A health agency’s assessment

Spain’s health technology assessment agency at the Instituto de Salud Carlos III looked for evidence on depression, anxiety, and stress. It found a single systematic review, the 2021 one, and rated its methodological quality critically low on the AMSTAR scale. It found no randomized trials or other reviews assessing these conditions. 2

Of the 12 studies in that review, only two involved people with a history or symptoms of mental illness. On safety, the studies that measured it recorded mild or moderate adverse events at low frequency, but the thin evidence does not allow a conclusion that the intervention is safe in those conditions, or effective for symptoms of depression, anxiety, and stress. 2

The report is part of the Ministry of Health’s plan to protect health against pseudotherapies. It is an assessment of the evidence available for those conditions; it does not address every possible use. 2

A later trial and a preprint

A preregistered randomized trial from 2024 assigned 80 people to a one-day workshop adapted for the pandemic. The group that took part improved on general symptoms (one month later and, less strongly, at six months), but the improvement was small and mostly faded by six months. Comparing change between the two groups showed no significant benefit, though statistical power was low. The authors warn that pandemic-related changes limit how far the result can be generalized. 4

In April 2026 a preprint appeared: a text circulated before other specialists have reviewed it, so its conclusions may still change. Its authors registered their protocol in advance and included five randomized trials, four of which provided outcome data, covering 429 participants in all. They judged every trial at high risk of bias and every comparison very low certainty, and concluded there is no reliable evidence to support its use. On adverse effects, one trial reported worsening in 3 to 4 percent of participants and another reported negative changes in 25 to 36 percent through ad hoc questions; the data could not be pooled. 5

How to read these numbers

Cochrane’s handbook sorts the certainty of evidence into high, moderate, low, and very low. Randomized trials start at high and drop for risk of bias, inconsistent results, indirect evidence, imprecision, or publication bias. “Very low” means the evidence is too uncertain to say much about the effect. 6

Cochrane also warns that “no evidence of an effect” is not the same as “evidence of no effect,” and asks that results be described by the same standard whichever way they point: not calling favorable findings “promising” while calling neutral findings on harms “no effect.” That caution is useful for reading supporters and critics alike. 7

To credit a change to an intervention, it is compared with a control group as similar as possible in every other respect. Without one, improvement can have other causes. Also, measures such as the Brief Symptom Inventory or the OQ-45, which appear in these studies, are questionnaires the person fills in: they reflect how the respondent feels, not something observed from outside. 8 9

What this means for Family

Family offers a reflection space without promising relief, healing, or changes in your relationships. This digital map is not an intervention evaluated in those studies: the research concerns in-person workshops with a facilitator, and it establishes no outcomes for arranging pieces on a screen.

You can use the map to express your own view; the site does not score whether it helped or declare that you have resolved a situation. If someone offers you a service, you can ask about their training, the outcomes they promise, the studies they rely on, how they handle your privacy, and what happens if you want to stop. Asking does not commit you to taking part.

This is general information, not an assessment of your health. If something worries you, talk with a qualified professional; the map does not replace that care.

Common questions

Do family constellations work?

It has not been established. A 2021 review found favorable results in several studies, but of low quality, and a Spanish health agency concluded the evidence cannot show effectiveness for depression, anxiety, or stress.

Are family constellations safe?

The evidence is thin. Studies that measured it recorded mild or moderate negative effects in a minority of participants, and the Spanish agency’s report could not conclude the approach is safe in the conditions it assessed.

What is a preprint, and why does it matter?

It is a paper released before other specialists have reviewed it. It can offer useful data, but its conclusions may still change after review.

Is Family a family constellation?

No. Family is a digital map for reflecting at your own pace, with no facilitator and no interpretation. It is not an intervention evaluated in those studies and promises no results.

What can I ask someone who offers a workshop?

You can ask about their training, the outcomes they promise, the studies they rely on, how they protect your privacy, and what happens if you want to stop.

Sources

  1. Konkolÿ Thege et al. · Family Process (2021)
  2. AETS–ISCIII · Eficacia y seguridad de las constelaciones familiares
  3. Hellinger school — Short biography of Bert Hellinger
  4. Konkolÿ Thege & Szabó · Journal of Psychiatric Research (2024)
  5. Souza et al. · medRxiv (2026, preprint)
  6. Cochrane Handbook — Chapter 14: Completing summary of findings and grading the certainty of the evidence
  7. Cochrane Handbook — Chapter 15: Interpreting results and drawing conclusions
  8. NCCIH — How to make sense of a scientific journal article: design of the study
  9. Crameri et al. — The Brief Symptom Inventory and the Outcome Questionnaire-45 (Psychiatry Journal, 2016)

Sources checked November 18, 2026.

Open the free map