Dear Ashley and fellow health economists,
Doing analysis at a cluster level -- in this case health facilities -- is
fine. If the data are provided by others (e.g. the facility) as aggregates
for each facility, and no person-level information is ever accessed by the
study investigators, this type of study would not be considered human
studies research in the US.
For the analysis, each facility is a unit of observation. Often a concern
in such studies is statistical power, the number of facilities or facility
years must be sufficient to meet the study's needs.
Attached is a recent paper of my colleagues and me using facility level data
as an example. Further studies are developing the cost-effectiveness of
provider incentives from these data. I am sure other readers could supply
hundreds of examples if needed.
Best wishes,
Prof. Donald S. Shepard
Schneider Institutes for Health Policy, Heller School, MS 035, Rm 275
Brandeis University, Waltham, MA 02454-9110 USA
Web: www.brandeis.edu/~shepard; email: [log in to unmask]; tel:
+1-781-736-3975; fax: +1-888-429-2672
-----Original Message-----
From: The international health economics discussion list.
[mailto:[log in to unmask]] On Behalf Of Ashley Agus
Sent: Wednesday, January 08, 2014 7:46 AM
To: [log in to unmask]
Subject: cost-effectiveness analysis at the cluster-level?
Dear all,
I am involved in a funding application for a cluster RCT. The principal
investigator wants to get consent at the cluster level (nursing homes) and
not at an individual level, thus patient-level data on costs and outcomes
will not be available. Does anyone have experience or advice and what
meaningful analysis of costs and outcomes can be done at the cluster level?
Perhaps I am overthinking this!
Many thanks, Ashley
|