Community factors associated with long-term physician retention

Introduction
Physician retention is a complicated issue related to many different variables.[1-3] Attracting a physician to a community is the first step in the physician-retention process. For example, growing up in a rural community, and having undergraduate, post-graduate, or locum exposure to rural medical practice are factors that contribute to a physician choosing to practice in a rural community.

Once physicians are recruited to a community, job factors are important determinants as to whether they will stay in the community. Excessive workload, burnout, long work hours, too much call responsibility, and not enough time off for vacations or continuing medical education are commonly cited reasons why physicians leave rural practice.[2,4-8]

Personal background, training, and professional satisfaction are, however, not the only things that determine whether a physician is going to move to and stay in a given community. The desire for greater recreational, cultural, and educational opportunities for family; desire for improved climate and geography; desire for less isolation; and desire to be closer to family and friends also contribute to a physician’s decision to work in a given community.[2,9-11]

We recently showed, using physician directory data, that communities with fewer than 7000 people have the lowest year-to-year physician retention rates and highest year-to-year physician recruitment rates.[12] The objective of this current study was to see whether physician directory data can be used to also show relationships between long-term physician retention and various community factors; namely, population, work situation, quality of school, recreational opportunities, climate, isolation, and cultural milieu.

Methods
The population studied were communities that had fewer than 30,000 people and had either a hospital or diagnostic-and-treatment centre. The communities studied were from British Columbia Health Regions located outside the Lower Mainland and southern Vancouver Island; namely, East Kootenay, West Kootenay-Boundary, North Okanagan, South Okanagan-Similkameen, Thompson, Coast Garibaldi, Central Vancouver Island, Upper Island/Central Coast, Cariboo, North West, Peace Liard, and Northern Interior health regions.[13-15]

Communities lacking a hospital or diagnostic-and-treatment centre were omitted from the study. This was done to avoid including bedroom community data; that is, communities that are so close to a larger community that they could be considered regional extensions of the larger community and so do not warrant having a hospital or treatment centre.[16]

Community population was obtained from the 1996 British Columbia census data.[17] From the 1996 British Columbia census data, we also obtained the percentage of the population who are Aboriginal. School ratings for secondary schools in each community were obtained from Fraser Institute’s web site.[18] The higher the Fraser Institute school rating score, the better academically that school is rated. In the few communities that had two secondary schools, the school with the higher score was used, as we reasoned that physicians wanting to get their children into the better-rated school would find a way to do so.

Latitude of communities studied was obtained from the Gazetteer of Canada, British Columbia edition.[19] Ski-hill communities, i.e., communities located within 1 hour of a ski hill, were identified from the British Columbia Recreational Atlas.[20]

Community isolation was studied by correlating physician retention with Northern and Isolation Allowance assignments. A Northern Isolation Allowance (NIA) community is a community in which doctors are given an extra fee in addition to that provided to physicians working in less “needy” communities. The British Columbia Medical Services Plan (MSP) has developed a rurality index score, which it uses to determine whether a community is northern and/or isolated enough to qualify for the NIA, and just how much that community’s NIA shall be.[21]

Factors considered include number of general practitioners in the community, number of specialists in the community, distance (in kilometres) from a major medical community, exceptional circumstances, doctor/patient ratio, distance from a major population centre, and size of the community. This scoring system is similar to that proposed by Leduc.[22]

The number of specialists and the number of family physicians working in each community each year was obtained from British Columbia Medical Directory information (1979-80 to 1998-99).[23] Every physician listed in each directory for each community was tabulated on a spread sheet. A cohort of 1979 family physicians/general practitioners was identified. These physicians were from 78 communities. The total number of family physicians listed in each community was tabulated, and the total number of physicians staying more than 9 years was also tabulated. A physician could be counted more than once—every time he or she moved to another community—since a new community retention rate would have to be calculated.

For each of the 78 communities in the study, a physician retention variable was computed as the ratio of the number of physicians with 10 or more years’ stay in the community divided by the total number of physicians in the community, and then multiplied by 100 to express it as a percentage. With this new derived variable as a response variable, multivariate regression analysis was used to assess the combined effect of the study variables (i.e., community population, number of specialists, NIA score, Fraser Institute school score, ski-hill proximity, latitude, percentage of Aboriginal peoples). The effect of each of the five "community-size" variables in combination with community population was also assessed. A series of five multiple regression models were then fit with community population plus each of the other five variables to determine whether the five community-size variables were truly independent of community population.

Results
Long-term family physician retention (the percentage of physicians listed for 10, 15, or 20 or more years) increases sharply once a community population exceeds 7000 (Table 1). The percentage of family physicians listed per community for 10 or more years also increases with increasing number of specialists, with increasing Fraser Institute school rating score, with proximity to a ski hill, with a more southerly location, with a decreasing Northern and Isolation Allowance score, and with a decreasing percent of Aboriginal population.

Multivariate analysis revealed that all seven variables are associated with physician retention. The strongest single predictor is the NIA score; the higher the score, the lower the percent retention. Stepwise regression modeling revealed that once the information about the NIA score is included in the model, none of the other variables adds significant predictive information about physician retention.

Multiple regression analysis reveals that when community population is taken into account, ski-hill proximity and the number of specialists do not provide significant additional predictive information, but latitude, percentage Aboriginal, and Fraser Institute school ratings do have additional predictive information. That is, the effects of latitude, percentage Aboriginal, and Fraser Institute school ratings are independent of community size.

Discussion
It can be argued that the ultimate expression of a physician’s community satisfaction is whether he or she chooses to stay and work in that community. If this is true, one would expect to find relationships between medical directory information and various community characteristics. Our data suggest long-term family physician retention is associated with a wide range of community factors; namely, the work situation, the quality of schools, recreational opportunities, climate, degree of isolation, and cultural milieu.

The data clearly show that long-term family physician retention rates dramatically increase once a community population exceeds 7000. Family physician long-term retention rates are also higher if there is a specialist in a community, if the community is less isolated, if there is a ski hill nearby, and if the high school has a high academic rating (Table 1).

The strongest single predictor of long-term physician retention was NIA score, which seems to confirm the validity of the NIA scoring system.

Multiple regression analysis also reveals that community population size confounds the ski-hill proximity and the number of specialists variables. Having specialists to consult is not the only work-related benefit associated with working in a larger community.

In larger communities there are more doctors to share call, cover vacations and educational leaves, and provide professional support—things that should decrease job stress and improve job satisfaction. Having a hospital facility in a community means that there is a place to admit, observe, and treat patients under controlled conditions—additional factors that make a family physician’s work easier and more rewarding.[5,7]

The association between increasing percentage of Aboriginal population and decreasing long-term family physician retention is independent of community population size. Perhaps this reflects the fact that looking after Aboriginal people is more stressful than looking after non-native people.

This is because Aboriginal people, per capita, have more health problems—both physical and psychosocial—than non-Aboriginal people.[14,24,25] Perhaps it reflects the fact that understanding complex Aboriginal cultural issues takes time that some physicians may not be willing or have the ability to invest.[26] Our data also reveal that good schooling and latitude are independent factors associated with long-term physician retention.

We believe that the information in this study can be used by health-care planners and hospital administrators to generate hypotheses, design studies, and evaluate how successful their communities are at retaining physicians. For example, our data show that 25% to 30%10-year physician retention is about as good as any large community can expect; communities with fewer than 7000 people can expect 10-year physician retention rates of around 10%. Communities with physicians retention rates exceeding these values should consider themselves fortunate.

Conclusion
Medical directory data can be used to quantitatively examine relationships between some community variables and long-term physician retention rates.

Acknowledgments
We would like to thank Amy Thommasen for inputting data, constructing initial graphs, and summarizing the initial data sets. This research project was supported by the Lloyd Jones Collins Foundation and by the North American Primary Care Research Group.

Dr Thommasen is a professor in the Faculty of Community Health at the University of Northern BC and is currently doing locums throughout rural BC. Dr Berkowitz is a staff statistician for the UBC Department of Family Practice. Dr Grzybowski is the director of research in the UBC Department of Family Practice and, after many years of practising medicine in the Queen Charlotte Islands, now has a clinic in Vancouver.