Minimum and optimal numbers of psychiatric beds: expert consensus using a Delphi process

Adrian P. Mundt, Enzo Rozas Serri, Matías Irarrázaval, Richard O’Reilly, Stephen Allison, Tarun Bastiampillai, Seggane Musisi, Ashraf Kagee, Andrei Golenkov, Joseph El-Khoury, Seon Cheol Park, Lydia Chwastiak, Stefan Priebe

Research output: Contribution to journalArticlepeer-review

13 Citations (Scopus)


The required minimum number of psychiatric inpatient beds is highly debated and has substantial resource implications. The present study used the Delphi method to try to reach a global consensus on the minimum and optimal psychiatric bed numbers. An international board of scientific advisors nominated the Delphi panel members. In the first round, the expert panel provided responses exploring estimate ranges for a minimum to optimal numbers of psychiatric beds and three levels of shortage. In a second round, the panel reconsidered their responses using the input from the total group to achieve consensus. The Delphi panel comprised 65 experts (42% women, 54% based in low- and middle-income countries) from 40 countries in the six regions of the World Health Organization. Sixty psychiatric beds per 100 000 population were considered optimal and 30 the minimum, whilst 25–30 was regarded as mild, 15–25 as moderate, and less than 15 as severe shortage. This is the first expert consensus on minimum and optimal bed numbers involving experts from HICs and LMICs. Many high-income countries have psychiatric bed numbers that fall within the recommended range. In contrast, the number of beds in many LMIC is below the minimum recommended rate.

Original languageEnglish
Pages (from-to)1873-1879
Number of pages7
Issue number4
Early online date21 Jan 2022
Publication statusPublished - Apr 2022


  • psychiatric
  • inpatient
  • beds
  • mental health
  • outreach
  • treatment
  • hospital


Dive into the research topics of 'Minimum and optimal numbers of psychiatric beds: expert consensus using a Delphi process'. Together they form a unique fingerprint.

Cite this