Self-managed teams in the context of community care for the elderly : effect on working conditions and quality of care.

Last registered on December 10, 2021


Trial Information

General Information

Self-managed teams in the context of community care for the elderly : effect on working conditions and quality of care.
Initial registration date
December 08, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
December 10, 2021, 2:40 PM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.


There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Institut des politiques publiques

Other Primary Investigator(s)

PI Affiliation
Institut des Politiques Publiques (IPP)
PI Affiliation
Institut des Politiques Publiques (IPP), DREES
PI Affiliation
PI Affiliation
PI Affiliation
Institut des Politiques Publiques

Additional Trial Information

On going
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
The Buurzorg healthcare organisation, created by Jos de Blok in 2006 with a dozen professionals, currently includes 10.000 employees. This success encouraged the adoption of this organisational model by other organizations, in France and around the world. By giving employees more autonomy to organize their schedule and the care plan of the patient, this model of self-managed teams is supposed to empower employees. The employees would be more motivated and satisfied, and the quality of their work would be improved for the beneficiaries.

In France, the long-term care sector is growing and suffering: the working conditions are difficult while wages and career prospects are low. The implementation of the "autonomous teams" model, inspired by Buurtzorg's principles, seems a promising avenue to limit job dissatisfaction and turnover, but the management of schedules and care plans implies developing new skills, and a high level of investment at work. A careful impact evaluation would ensure that this organisational model indeed improves workers satisfaction and the care received by beneficiaries.

This project aims at evaluating the effects of the new working organisation called autonomous teams on the employees (absenteeism, turnover, psychosocial risks, job satisfaction) and on the beneficiaries (hospitalizations, deaths, medication, nursing home entry). It aims at informing public policies from the public health and the health at work standpoints.

To ensure that long-term care organisations, teams and employees that undergo the change in organisations are comparable to those that do not, we implement a randomized controlled trial. This means randomly choosing, among the teams of employees that are willing to adopt the new model, the ones that will actually change their work organisation. The causal impact of the program will be measured by comparing the treated and control employees and care beneficiaries.
External Link(s)

Registration Citation

Breda, Thomas et al. 2021. "Self-managed teams in the context of community care for the elderly : effect on working conditions and quality of care.." AEA RCT Registry. December 10.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


Home care services choose to participate in the experiment, then they choose teams that could switch to the self-managed team model. Among those teams, we randomly pick the teams which will implement the new model. The teams selected to participate to the experiment that do not switch to the new model make up the control group and keep their standard working organisation.

All home care services under investigation take care of dependent elderly persons : those who are followed by treated teams will benefit from their new working organisation. The dependent elderly persons who are cared for by the control group teams will benefit from the standard working organisation.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
1. Employees' job quality
2. Beneficiaries' health and satisfication with the service provided
Primary Outcomes (explanation)
1. We use both objective and subjective measures of job quality related to three domains: job satisfaction (subjective measure), working conditions and work-life balance (objective measure, self-assessed), absenteeism and health. Outcomes are measured based on baseline (before randomization) and endline (18 months after randomization) surveys or administrative data.
1a. Job satisfaction: self-assessed on a likert scale. .
1b. Working conditions: measured through questions related to autonomy at work, work atmosphere, responsibilities, employees' work schedules, commuting time and work-life balance. We have included in our survey instrument questions allowing us to apply the Karasek-Siegrist approach and we will evaluate the treatment effect on the indices and measures of job strain following this approach.
1c. Employees' absenteeism : based on baseline (before randomization) and endline (18 months after randomization) surveys, we will measure how many days the employee was off because of sick leaves or because of work or communting accidents. Using administrative data (if we obtain access to the relevant data), we will also measure how it is likely they have quited their job at different points in time after the beginning of the experiment.

2. Beneficiaries health and satisfaction with care services among disabled people:
2a: Beneficiaries health. We will collect survey data among beneficiaries before randomization and 18 months after. It will allow us to measure self-declared health. Information about death (if possible) and nursing home entry will be collected from the formal care service. Alternatively, we will rely on "exits", i.e. beneficiaries likelihood not to be clients of the home care provider anymore. If possible, administrative data on emergency hospital entry and medication consumption will also be collected. Lastly, relying on administrative data (if possible), information regarding the use of beneficiaries' care plan will allow us measuring to what extent individuals fully use their care plan or not.

2b. Satisfaction with care services among disabled people: we will collect survey data among disabled persons before randomization and 18 months after. It will allow us measuring to what extent disabled people are satisfied with the care service they benefit from: we will know whether one or several employees take care of them, whether the planned schedule frequently changes, to what extent disabled individuals are happy about that. We will also measure satisfaction regarding specific elements such as the relationship with the employee who takes care of the disabled person.

We plan to investigate effect heterogeneity along several dimensions. For employees, we know the type of structure where they work for and its size; we also have information about their professional experience and the type of contract they have. Heterogeneity analysis will be performed regarding individuals' age and family situation. We will also investigate heterogeneity according to baseline characteristics and outcomes measured at baseline using the baseline survey.

As regards beneficiaries, effect heterogeneity will be measured across individual characteristics (age, gender, disability level) and the characteristics of their service provider. We will also measure heterogeneous effects regarding whether the person benefits from any informal care or not. We will finally investigate heterogeneity according to baseline characteristics and outcomes measured at baseline using the baseline survey.

Secondary Outcomes

Secondary Outcomes (end points)
See above (no distinction made between primary and seconday outcomes)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We implement a random draw among teams that have been designated to participate to the experiment in each of the home care service provider participating to the experiment. The randomization is therefore stratified by home care service provider. Within each of them, denote N the number of teams that is pre-selected to take part in the experiment. We select randomly about N/2 teams. The rounding is adjusted at each draw in order to keep an equal overall number of treated and control teams in the experiment. This adjustment exploits the fact that not all home care service provider enters the experiment at the same time. The total number of participating home care service is not pre-specified and will depend on our ability to recruit new participants while the experiment starts among the first of these service providers.

Treated and control teams keep taking care of the same dependent or disabled persons as before.

When the random draw is realized at the individual level, we similarly pick half of the individuals in each health care provider to be subject to the self-managed model. We then reconstruct new teams among workers assigned to the treatment.
Experimental Design Details
Not available
Randomization Method
Randomization done in office by a computer.
Randomization Unit
The randomization unit is the working team (typically comprising 8 to 12 individuals); it can be the individual if possible for the home care services structure.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
About 38 home care service providers. More if possible. The experiments starts while some home care providers have not been recruited yet. We include systematically in the experiment all providers that we approach and accept to apply our experimental design pre-specified in a written document.
Sample size: planned number of observations
152 teams.
Sample size (or number of clusters) by treatment arms
76 treated teams and 76 teams in the control group.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We are able to define a minimum detectable effect (MDE) for the probabilites of employees 1) to quit their job and 2) to be on sick leave. We account for a power of 80% and a significance level of 0.05. The MDE that we can measure depends on the number of teams included in the experiment. The minimum detectable effect that we can measure on the probability of quitting one's job is a drop of 8pp when accounting for 152 teams in our sample. For the same number of teams, we are able to detect a drop in the number of sick leaves from 5.3 to 4.6. As regards beneficiaries, the MDE with the same number of teams is an increase of 20pp of the probability to fully use their care plan.

Institutional Review Boards (IRBs)

IRB Name
Paris School of Economics
IRB Approval Date
IRB Approval Number