Evaluating organisational changes using quasi-experimental study designs - evidence from a case study including low back pain patients

In this thesis, we focus on quasi-experimental designs used for evaluation of the impact of organisational change on low back pain (LBP) patients in one region of Denmark. Organisational change is interesting, for several reasons. It affects a highly prevalent and incident patient population, and LBP is one of the most disabling and costly diseases in the world. Although organisational change is frequently implemented, little is known about its effects. The organisational changes were implemented in the Region of Southern Denmark (RSD), which decided to centralise all regional spine departments into a specialist spine hospital. The region further gave general practitioners (GPs) the option to refer LBP patients for a lumbar magnetic resonance imaging (MRI) assessment, without any prior consultation with an office-based spine specialist (rheumatologist or orthopaedic). Chiropractors (CP) were also given the opportunity to refer patients for a lumbar MRI assessment. These organisational changes were followed by a diseases management program (DMP) for LBP patients, developed specifically for the region. This combination was unique in Denmark, as two other regions only implemented the referral option to lumbar MRI; however, easy access to lumbar MRI is controversial in the literature. International clinical guidelines have advocated for limited use of lumbar MRI in the diagnostics of LBP. Lumbar MRI generates high-resolution images of the spine and surrounding tissues. With improving imaging techniques, there is a risk of finding abnormalities, even in individuals without LBP. As a consequence, guidelines only recommend lumbar MRI if doctors suspect that the LBP originates from a severe pathology, such as cancer, fractures, infections, herniations or stenosis. However, these conditions are rare, and most people have LBP without any known pathology, known as unspecific LBP. As we do not know the underlying pathology of unspecific LBP, any abnormalities found on an MRI might trigger a cascade of unnecessary treatment. Scientific literature on the use of lumbar MRI can be divided into two main groups: timing of the lumbar MRI in relation to guidelines and routine use of MRI. Literature on early MRI vs guidelinesrecommended MRI is fairly consistent, showing that early MRI increases chance of lumbar surgery, prescription of pain medications and prolonged periods with disability. Literature on routine use of MRI and its consequences in terms of surgery are somewhat inconsistent. Studies from the US shows that areas with high lumbar MRIs have similarly high surgery rates. However, some studies from outside the US do not find these associations, hence the need for further research. This thesis comprises three studies examining the effects of organisational change, using quasiexperimental studies. First, the use of lumbar MRI is assessed (Study I), then the relationship between use of MRI and lumbar surgery following the organisational changes (Study II), and finally, if the use of lumbar MRI and surgery following the organisational changes affects the sick leave duration following the organisational changes (Study III). All analyses in this thesis are based on an unbalanced panel dataset provided by Statistics Denmark. It comprises all Danish citizens aged 18+ from 2008–2013. We have yearly information on citizens’ socio-demographic characteristics, use of primary care providers, and use of pain medication prescribed by a doctor, and detailed information on secondary care usage (procedure codes and ICD-10 diagnosis). We rely on natural experiment, using organisational changes to allocate GPs to RSD (intervention) or control regions. We use difference in difference (DD) and instrument variable (IV) analyses to assess the effects of the organisational changes from 2010– 2013 (intervention years) and using 2008 and 2009 as pre-intervention years. We find that the lumbar MRI rate significantly increased by 1.8 lumbar MRI per 1,000 population enlisted with a GP, approximately a 10% increase in use of lumbar MRI, compared with control regions. Similarly, we find that the overall use of lumbar surgery significantly increased by 0.24 surgeries per 1,000 population enlisted with GPs in RSD. Examining lumbar stenosis surgery and lumbar herniated disc surgery, we find the opposite effect—an insignificant decrease in lumbar stenosis surgery rates and a significant increase in lumbar herniated disc surgery rates. As lumbar MRI and overall lumbar surgery rates increased following the introduction of the organisational changes, we investigated if the increase in lumbar MRI had a positive effect on the long-term sick-leave length (>3 weeks of consecutive sick leave) for patients with LBP. It is not possible to identify LBP patients using primary care registries. As a consequence, we identify two groups of individuals at risk of long-term sick leave due to LBP: people with less than 11 years of education and those visiting a physiotherapist or CP. Results show significant reductions in sick leave for both groups in RSD, indicating that the increase in lumbar MRIs has a positive effect for those groups in RSD. However, there may be other trends not controlled for in the study, which might confound the results. Overall, the studies show that quasi-experimental studies can be used to assess large organisational changes, especially using the registry data available in most health care settings, where claims-based data are used. For the specific organisational change investigated, we find that the overall increase in use of lumbar MRIs and lumbar surgery seems to reduce long-term sick leave for those at high risk of LBP. There is potential for some confounding of the results, as other interventions or advantages for patient groups, which might affect the results, are not controlled for in the analyses. The thesis advocates that we should continue developing these registries, both at the hospital and primary care levels, as these have the potential to become valuable sources for analyses of the effects of future organisational changes and other health-related matters.