Regional Variation in Utilisation of Healthcare Services - A study on hospital care in Norway

Abstract

<p><i>Background:</i> In 1938, J. Alison Glover published a study on incidence of tonsillecto-mies in school children, demonstrating variation in treatment rates across school districts. Consequently, it rates raised questions as to what determines when to treat a patient. After Wennberg and Gittelsohns’ paper ‘Small Area Variations in Health Care Delivery’ was pub-lished in 1973, it became clear that regional variation was not restricted to school districts nor surgical treatments. While the phenomenon of regional variation is well documented, there are still knowledge gaps on the cause and effect of this variation. Thesis fills in some of these omissions by ex-ploring the implications that regional variation has for patients, asking whether there are pat-terns in utilisation rates, and examining how treatment characteristics affect these patterns. Lastly, it considers how variation can be reduced. The original research provides new knowledge on the phenomenon, which hopefully can help policy makers construct a fair and efficient healthcare system. <p><i>Materials and methods:</i> The materials used in this thesis was gathered by the Norwegian Patient Register (NPR), the Norwegian Quality Register for Spine Surgery (NORSpine), and Statistics Norway. Data from NPR contains hospital administrative data on all treatments fitting the inclusion criteria for lumbar disc herniation-, spinal stenosis-, meniscus-, shoulder-, ear drain-, heavy eyelid- and cataracts -surgery, as well as tonsillectomies. For spine surgeries, NORSpine have developed a suitable inclusion criteria. NORSpine is a clinical quality register, containing clinical, demo-graphic and socioeconomic patient level data, as well as multiple patient reported outcome measures. For the six latter treatments, selection criteria was copied from the Norwegian Health Atlas. Data from NPR, NORSpine and Statistics Norway was used in Paper 1 and Paper 3. Data from NPR and Statistics Norway was used in Paper 2. The methods used to explore the research questions were, generalized estimation equations (Paper 1), principal component analysis (Paper 2), logistic regression and stochastic gradient boosting (Paper 3). <p><i>Results:</i> Relatively high regional utilisation rates were associated with reduced patient need. It was found that patients in high-rate regions were found to have better health at baseline and achieved lower health gain after treatment. For primarily elective treatments, high treatment rates are likely supply-driven, and there is a substitution effect across medical specialties. A suggested remedy to reduce this variation has been decision aid-tools. In the case of lumbar spine surgery, survey data alone is not enough to make accurate predictions on whom to treat. <p><i>Conclusion:</i> As a result of regional variation in treatment rates, region of residence affects patients’ likeli-hood of receiving treatment. The current method of financing hospital care is not designed to reduce the inequalities that arise from such variation. The most suitable approach to reduce variation is to increase focus on shared decision making between patient and physician, and facilitate experience sharing among physicians across hospital regions.

<p>Regional variasjon i bruk av helsetjenester eksisterer for nesten alle behandlinger i alle land. Mens noe av variasjonen er et resultat av mer sykdom i noen regioner, er det antatt at store deler av den totale variasjonen skyldes årsaker ikke relatert til pasienters helse. Selv om fenomenet er kjent, er det er fortsatt mye ukjent om hvorfor de oppstår, hva konsekvensen er for pasienter og hvordan man skal fjerne den uønskede variasjonen. <p>I denne avhandlingen undersøkes derfor hvordan pasienter påvirkes av regional variasjon, hvorvidt vi kan lære noe ved å se på behandlingsrater for flere behandlinger samtidig, og det drøftes hvordan variasjonen kan reduseres. Ved å bruke unik data fra norske helse-, norske kliniske kvalitet- og administrative registre presenteres original forskning som kan brukes til å skape et bedre helsesystem.