Travel distance to the general practitioner. Do patients move closer to the services when starting to use them?

ObjectivesThe main objectives were 1) to calculate and describe travel time by car from the home of Norwegian residents to the office of their named general practitioner (GP); 2) to estimate changes in travel time for residents who started to visit the GP and, if so, to 3) to explore if the residents changed GP or moved to reduce their travel time. MethodsWe used nation-wide individual-level annual registry data 2009-2017 on the exact location of the home of every resident and the GP-office to calculate travel time in minutes by car from home to their assigned GP. First, using data for 2017 only, we calculated travel time at the median and 90th percentile, and by sex, age, immigrant background, county of residence and use of GP in 2017. Second, with annual data 2009-2017, and restricting the sample to residents who had not used their GP over the last two years (t-2 and t-1), we used a difference-indifferences model to estimate changes in travel time in the next two years (t+1 and t+1) for patients with a visit in year t compared with those with no visit in t. Separate models were run for those who changed GP and those who moved from t-2 to t., and for the 20% who lived farthest away in t-1. ResultsWe could calculate the travel time for 3,976,910 residents in 2017, with a median travel time from home to the GP of 4.9 minutes and a travel time at the 90th percentile of 18.3 minutes. In the most sparsely populated northern county of Norway, travel time was about 5 minutes at the median and below 45 minutes at the 90th percentile. Elderly residents and residents who used their GP in 2017 had a somewhat shorter travel time than other groups of the population. Using annual data for 2009-2017 in the difference-in-differences analysis (16,388,151 resident-year observations), travel time dropped by 2.5 minutes (95% confidence interval 2.4 to 2.6) in t+1 and t+2 for patients with a visit in t compared with similar patients with no visit in t. The drop was similar for patients who did and did not change GP, but larger for patients who moved (10.0 minutes; 95%CI 9.7 to 10.4) compared with those who did not move (0.6 minutes; 95%CI 0.5 to 0.7), and particularly large for the 20% living farthest away in t-1 (24.2 minutes; 95%CI 23.3 to 25.2). ConclusionsTravel time from home to ones GP is short for the vast majority of the population in the sparsely populated country of Norway. However, residents move closer to the GP when they start using the services, especially patients who used to live far away. This relocation may reflect strong preferences for proximity to the services, and we conclude that more knowledge is needed to enable transparent balancing of costs and benefits of centralizing GP-services, at least in rural areas. JEL classificationI10, E32, J6