9
3 Data and empirical strategy
3.1
Data
We use data from the health care data bases in the Skåne county council,
which contains detailed individual level information about all health care
visits in the county from 1 January 2008 to 31 August 2013. We have also
added information about drug prescriptions from the National Board of
Health and Welfare from 1 January 2008 to 31 August 2013. From the
Social Insurance Agency, we have further collected information about all
sickness and disability benefit spells from 1 January 2000 to 31 August
2013,
7
along with a large set of individual characteristics such as age,
education, marital status, employment status, earnings and municipality of
residence. In all, we have rich information on matters closely related to the
individuals’ health and labor market position, factors that should be
important analyzing future sickness absence, health care and drug
consumption.
The population of interest is individuals aged between 20 and 64 who had a
registered health care visit with a mental illness diagnosis that could qualify
for CBT or a pain-related diagnosis that could qualify for MDT within the
medical rehabilitation guarantee between 1 January 2010 and 30 June
2011 or between 1 January and 31 December 2012.
8,9,10
In total 21
percent of the individuals in the target population for CBT received CBT
within the medical rehabilitation guarantee at some point during this
period, and 3 percent of the target population for MDT received MDT. Given
that the shares are not higher, this may be due to a diagnosis not detailing
a patient’s suitability to receive treatment, or to demand exceeding the
supply of treatment.
Unfortunately, there is no information about employment status at the time
of the health care visit, unless for unemployed
11
individuals who were
registered at the Public Employment Service. When patients registered for
the medical rehabilitation guarantee, they were asked about their sources
of income. Among those who were not reported as unemployed, 75 percent
7
This only includes sickness spells longer than the employer period, which is two
weeks.
8
Since the supply of CBT expanded gradually, few patients received CBT or MDT in
the fall 2009 and the exclusion of these patients does not affect the results.
Analyzing the impact of CBT, health care visits where the patient has more than 10
previous health care visits with a mental illness diagnosis were eliminated from the
beginning, to increase comparability between the groups already before the
matching.
9
The sample is a combination of the samples used in Hägglund et al (2012) and
Hägglund et al (2014). Therefore, there is a gap in the sampling of individuals
during the second half of 2011.
10
See Appendix A for a list of the diagnoses and ICD codes.
11
Unemployed individuals have the right to be on sick leave in Sweden.