Sandro Sperandei
Sandro Sperandei is a Brazilian public-health researcher who studies physical activity adherence in unsupervised settings, which is to say in ordinary commercial gyms rather than in supervised trials. He is the first author of Sperandei, Vieira and Reis (2016), the retrospective cohort of 5,240 members of a single fitness centre in central Rio de Janeiro that reported 63 per cent stopping before their third month and fewer than 4 per cent lasting a continuous year. It is the closest thing the literature has to a real answer about how quickly people stop going to the gym, and it matters to this publication out of proportion to its size.
Who they are
At the time of the 2016 attrition study he was affiliated with the Institute of Scientific and Technological Communication and Information in Health at the Oswaldo Cruz Foundation, FIOCRUZ, in Brazil. The paper is now also listed among the publications of Western Sydney University.
His subject is the setting almost nobody studies. The adherence literature is mostly built from supervised programmes, clinical cohorts and workplace schemes, where somebody is running the intervention and recording who turns up. A commercial gym is the opposite: people join, pay, attend or do not, and leave without telling anyone. The 2016 paper is what it looks like when somebody goes to the records of one such gym and counts.
His co-authors on it are Marcelo C. Vieira, of the State Institute of Cardiology Aloysio de Castro, and Arianne C. Reis, then at Southern Cross University.
Key papers
Sperandei, S., Vieira, M. C., & Reis, A. C. (2016). "Adherence to physical activity in an unsupervised setting: explanatory variables for high attrition rates among fitness centre members", Journal of Science and Medicine in Sport, 19(11), 916 to 920. PubMed 26874647. First author. A retrospective cohort of 5,240 members of a single fitness centre in central Rio de Janeiro, roughly 550 to 700 active members at any one time, mean age 31, 58.8 per cent women, 74.7 per cent inactive before joining. 63 per cent stopped before their third month; fewer than 4 per cent lasted a continuous year.
Three limits belong with those numbers wherever they are used, and this site prints all three.
Dropout was measured as a cancelled payment rather than as attendance. The study counted cancellations.
The authors did not record attendance frequency. So the curve describes membership survival, not gym-going. Most people stop turning up some time before they stop paying, which means the behavioural cliff is earlier and steeper than 63 per cent by three months, and nothing in the dataset says by how much.
It is a single medium-sized gym in one city. A much better number than the folklore it replaces, and still one site.
A figure of roughly 47 per cent gone within two months circulates alongside the two above. This publication does not print it, because it was not confirmed against the full text.
Key ideas
Adherence in an unsupervised commercial setting is a different measurement problem from adherence in a supervised programme, and the two literatures should not be quoted as though they answered the same question.
What a dropout figure means depends entirely on what was counted. A cancelled direct debit, an attendance threshold and a self-report give different answers from the same population.
The honest number is worse than the folklore it replaces. The commonly repeated six-week quit point has no traceable source, and the real cliff is nearer two months and steeper.
Corroboration can agree on direction and disagree wildly on size. Gjestvang and colleagues (2019), in BMJ Open, found roughly 37 per cent of new members across 25 Oslo clubs still exercising regularly at twelve months, against fewer than 4 per cent still paying at a continuous year in Rio. Both say most people stop. The gap between them is the lesson.
Concepts on this site
Sources
- Sperandei, S., Vieira, M. C., & Reis, A. C. (2016). Journal of Science and Medicine in Sport, 19(11), 916 to 920. PubMed 26874647.
- Gjestvang, C., Stensrud, T., Hansen, B. H., Kolle, E., & Haakstad, L. A. H. (2019). BMJ Open, 9(4), e027987. Corroboration in the same direction at a very different number.
- Vault verification record, 2026-09-11. Citation confirmed; the percentages printed here are the two that were confirmed, and the unconfirmed two-month figure is excluded.