ABSTRACT
Patients in Nunavut rely exclusively on airplane to access medical care beyond the nursing stations in communities. This can take the form of scheduled flights for chronic and non-urgent issues or dedicated medevacs for emergencies. Each community is routinely visited by family physicians (FP) who provide in-person primary care. The frequency and duration of FP visits depends on the community size, with larger communities having longer and more frequent visits. During their visits, FPs can be called upon to assist in emergencies. This study provides a detailed portrait of the territory’s medical travels between 2012 and 2018. Contrary to our initial hypothesis, we show that the presence or absence of an FP in the community did not have a significant impact on the rates of medevacs. However, we found that the rates of non-urgent scheduled flights increased. Our findings provide in-depth information on the rates of medevacs and non-urgent travel in Nunavut. They also raise important questions for primary care in remote areas by demonstrating an increase in routine travel requirements when physicians are present in those communities. As health outcomes were not assessed, further studies are required before recommendations can be made to change the rate of FP visits.
KEYWORDS: Medevac, aeromedical evacuation, arctic, Nunavut, family practice, fly in fly out, FIFO, primary care, Inuit
Introduction
The delivery of health care to Canada’s Northern Indigenous communities is faced with a myriad of challenges including, but not limited to, remote geography, extremes of weather, a paucity of experienced health care personnel, and cultural barriers between caregivers and patients [1]. Nunavut is Canada’s largest territory by land mass and its most sparsely populated. Nunavut has a population of 38,396 (85.8% Indigenous identity) spread out over 1,836,993.78 km2 for a population density of 0.03 people/km2 [2,3]. Provision of health care is divided into three distinct geographical regions with a “hub and spoke” model of care. Iqaluit, the territorial capital, is the medical hub for the Qikiqtaaluk region and supports 11 remote communities. Each of these remote communities has a health centre staffed by registered nurses, nurse practitioners, licenced practical nurses, and advanced care paramedics. Physician and specialist services are provided using a fly-in, fly-out (FIFO) model with a combination of phone, email, and telehealth support between visits. Remote medical advice was uniformly available and accessible for all communities, although formal Telehealth service was only established in 2020. Community visits by Family Physicians (FP) are as frequent as 1 week per month for the larger communities, and as infrequent as four times per year for the smallest, most remote communities. When a patient requires urgent physician assessment/treatment, they must travel by plane to a central hub. For less urgent cases people travel by commercial flight (schedevac) and for emergencies they travel by aeromedical evacuation (medevac). The final decision to transfer a patient to Iqaluit – either by medevac or schedevac – is made by the physician on-call in the emergency room (ER) in Iqaluit after consultation with either the nursing staff or the community physician present on-site.
One major concern for this model of care is rising costs. The per capita health expenditure in Nunavut is 2.5 times the national average, with health care accounting for 30% of the territory’s GDP [2]. In 2014–15 the Nunavut Department of Health’s operations budget totalled $332 M, 1/3 of which was associated with medical transport [3]. Data from 2017 to 2018 shows the average length of stay following a scheduled transfer is 4 days, with an average cost of $1,320 and the average cost of an aeromedical evacuation is $18,122 (Dept of Health, communication to Authors). Previous studies have shown that over 30% of infants are evacuated for bronchiolitis by their second birthday and considering that 11.9% of the total population is under the age of 4 this is just one example of the large financial burden medical travel has on the system [4,5].
Policy makers need to consider a long list of complex and interconnecting factors when determining the division of health care dollars. Overall, there is insufficient evidence to guide effective and cost-efficient modelling of rural and remote healthcare [6–11]. There are very few studies comparing rates of aeromedical evacuation in remote communities when a physician is present, and those that do exist are conflicting [12,13]. The purpose of this study is to assess rates of scheduled and emergency aeromedical evacuations in the remote communities of the Qikiqtaaluk region of Nunavut and to determine if the presence of a physician alters rates. This is an important factor in healthcare cost modelling and may aid decisions to recruit physicians to remote areas.
Methods
Study data
This study was conducted using data from the Medical Travel Database from 2012 to 2018. The Medical Travel Database contains data on all medical related flights paid for by the Government of Nunavut. Very rare events such as private flights or flights covered by insurance companies for non-residents were not included in this study. This database allows the Department of Health to provide reliable estimates on number of medical flights initiated, the type of medical flight required (medevac vs. schedevac), the community in which the flight was initiated, and the reason for travel (provided as an ICD-9 diagnostic code).
All data from 1 January 2012 to 31 December 2018 were extracted from the medical travel database. A total of 246,343 cases were initially included for assessment of eligibility.
Study population
The target population for this study were Nunavummiut who required medical travel, initiated from the Qikiqtaaluk region, between 1 January 2012 and 31 December 2018. There was a total of 69,434 unique cases that met the eligibility criteria.
Inclusion and exclusion criteria
The analysis included persons from 11 communities in the Qikiqtaaluk region and was not restricted based on age, sex, or ethnicity. Patients whose medical travel was initiated from Iqaluit (n = 19,103) or whose community of initiation was missing (n = 4) were excluded from the analysis. There was a total of 50,327 eligible flights in the final analysis.
Measurements
Medical travel flight
A medical travel flight was determined by an entry into the medical travel database, where the invoice amount was greater than $0. The type of flight was determined by a variable in the database referring to medevac or schedevac flights. Escorts, usually a family member, are often present with a patient who requires medical travel and are also included in the medical travel database. These escorts are associated with the patient by their health card number. Escorts were removed from the data for analysis, and only flights for the patient were included.
Physician presence
The presence of a physician was determined by the physician schedule. This schedule indicated the community, the start date, and the end date of a physician visit. This schedule was merged with the medical travel data to ascertain for each patient whether there was a physician present in the community when their flight was initiated.
Sample size
The sample for this study was obtained from retrospective records and included all eligible cases. As such, no sample size calculations were conducted.
Statistical analysis
Statistical analysis was conducted with STATA/IC Version 14.2. Descriptive statistics were used to provide summary data on the individual sex and age of each patient, community of origin, month and year of transport, physician presence, and by type of flight for the total data set from 1 January 2012 to 31 December 2018. Chi-squared analyses were conducted to assess rates of scheduled and emergency aeromedical evacuations in the remote communities of the Qikiqtaaluk region of Nunavut to determine if the presence of a physician alters rates. The association between flight type and each variable and between physician presence and each variable was also assessed to examine the extent that a variable was a potential covariate. Further evidence of potential confounding was explored through univariate analysis and by adjusting the crude association between flights and physician presence by each potential covariates one at a time. Poisson regression was used to examine the presence of confounding by comparing the coefficients in the confounder adjusted estimates to the crude estimates. All potential confounding variables were carried forward for inclusion in multivariable Poisson regression.
A multivariable Poisson regression model was constructed in a stepwise process, adding each potential confounding variable one at a time, starting with the variables that showed the greatest changes from the crude estimates in univariate analysis. Specifically, Poisson regression was modelled by including community as a covariate. The exposure term used in the Poisson regression model was the number of days over which flights were observed. This controls for time-invariant characteristics of each community, such as general population size, remoteness, or baseline medical travel patterns, without explicitly including population size in the model. By doing so, we ensure that the estimates reflect within-community changes in medical travel rates associated with physician presence, rather than structural differences across communities. Any confounding variable that improved the fit of the model, as determined through an adjusted Wald test, remained in the final model. Age and sex were not included in the multivariable Poisson regression.
Variables were assessed for multicollinearity with one another using Pearson r correlation to examine the degree of the relationship between pairs of confounding variables. No potential confounding variables were found to be correlated with one another.
Sub-groups
One sub-group was identified for analysis based on a review of the crude analysis. This sub-group was the community of initiation. Differences in the crude association between the rate of medical flights and physician presence were detected by community. As such, the associations were presented stratified by community of initiation for the confounder adjusted analyses.
Sensitivity analysis
It was unclear whether a physician should be considered present in the community on the day they fly into or out of the community as flight times were not recorded. If the physician was arriving after normal clinic hours, they may not be present at the health centre until the following day. While the main analyses assumed the physician was present on these days, we conducted a separate analysis to evaluate the impact of excluding a physician on the days in which they fly.
Ethics approval
This study was approved by the research ethics board of Memorial University of Newfoundland (HREB #2021.160). The Nunavut Research Institute confirmed that a licence was not required as new data was not collected. The use of medical travel data was covered by a data agreement with the Government of Nunavut.
Results
Flights
Of the 246,343 flights in the medical travel database, 50,327 (20.4%) were eligible for inclusion in the analysis. Figure 1 outlines the data cleaning process and eligibility assessment. Most of the flights were schedevac (90.1%) with less than 1 in 10 flights being medevac (9.9%). A physician was not present in the community for most flights initiated (84.8%).
Figure 1.
Curation of the medical travel database.
No association was found between physician presence and type of flight initiated (p = 0.118). Most schedevac flights were taken by female patients (60.9%); however, medevacs were split evenly among male and female patients (51.0% and 49.0%, respectively) between 1 January 2012 and 31 December 2018. Tables 1 and 2 highlight the descriptive characteristics of the data used in the final analysis.
Table 1.
Sex and age distribution of the medical travel patients.
| Medevac N= (%) |
Sched-Evac N= (%) |
Total N= (%) |
p-value | |
|---|---|---|---|---|
| 4,968 (9.9) | 45,359 (90.1) | 50,327 (100.0) | ||
| Sex | ||||
| Male | 2,434 (12.5) | 17,074 (87.5) | 19,508 (100.0) | <0.001 |
| Female | 2,531 (8.4) | 27,633 (91.6) | 30,164 (100.0) | |
| Unknown* | 3 (0.5) | 652 (99.5) | 655 (100.0) | |
| Age | ||||
| 0–9 | 1,206 (14.1) | 7,343 (85.9) | 8,549 (100.0) | <0.001 |
| 10–19 | 755 (10.9) | 6,205 (89.2) | 6,960 (100.0) | |
| 20–29 | 944 (10.9) | 7,750 (89.1) | 8,694 (100.0) | |
| 30–39 | 529 (8.7) | 5,533 (91.3) | 6,062 (100.0) | |
| 40–49 | 429 (8.1) | 4,867 (91.9) | 5,296 (100.0) | |
| 50–59 | 281 (5.5) | 4,825 (94.5) | 5,106 (100.0) | |
| 60–69 | 374 (7.2) | 4,811 (92.8) | 5,185 (100.0) | |
| 70–79 | 260 (8.7) | 2,735 (91.3) | 2,995 (100.0) | |
| 80+ | 187 (21.5) | 684 (78.5) | 871 (100.0) | |
| Unknown* | 3 (0.5) | 606 (99.5) | 609 (100.0) |
*The “unknown” category for age and sex is shown for transparency but was excluded from all statistical testing (e.g. chi-squared analyses).
Table 2.
Descriptive characteristics of medical travel, by travel type, between 1 January 2012 and 31 December 2018.
| Community size (pop.) * | Medevac N= (%) |
Sched-Evac N= (%) |
Total N= (%) |
p-value | |
|---|---|---|---|---|---|
| 4,968 (9.9) | 45,359 (90.1) | 50,327 (100.0) | |||
| Community | |||||
| Arctic Bay | 868 | 313 (8.0) | 3,597 (92.0) | 3,910 (100.0) | <0.001 |
| Kinngait | 1,441 | 622 (9.3) | 6,057 (90.7) | 6,679 (100.0) | |
| Clyde River | 1,053 | 616 (11.2) | 4,888 (88.8) | 5,504 (100.0) | |
| Grise Fiord/Resolute | 342 (combined) | 144 (7.4) | 1,799 (92.6) | 1,943 (100.0) | |
| Sanirajak | 848 | 396 (10.6) | 3,341 (89.4) | 3,737 (100.0) | |
| Igloolik | 1,682 | 890 (11.3) | 6,999 (88.7) | 7,889 (100.0) | |
| Kimmirut | 389 | 180 (7.8) | 2,115 (92.2) | 2,295 (100.0) | |
| Pangnirtung | 1,481 | 733 (9.0) | 7,431 (91.0) | 8,164 (100.0) | |
| Pond Inlet | 1,617 | 860 (11.6) | 6,532 (88.4) | 7,392 (100.0) | |
| Qikiqtarjuaq | 598 | 214 (7.6) | 2,600 (92.4) | 2,814 (100.0) | |
| Year | |||||
| 2012 | 546 (8.6) | 5,826 (91.4) | 6,372 (100.0) | <0.001 | |
| 2013 | 462 (7.2) | 5,991 (92.8) | 6,453 (100.0) | ||
| 2014 | 608 (9.0) | 6,142 (91.0) | 6,750 (100.0) | ||
| 2015 | 663 (9.2) | 6,525 (90.8) | 7,188 (100.0) | ||
| 2016 | 761 (9.9) | 6,958 (90.2) | 7,719 (100.0) | ||
| 2017 | 935 (11.6) | 7,150 (88.4) | 8,085 (100.0) | ||
| 2018 | 993 (12.8) | 6,767 (87.2) | 7,760 (100.0) | ||
| Month | |||||
| January | 392 (8.7) | 4,097 (91.3) | 4,489 (100.0) | <0.001 | |
| February | 372 (9.7) | 3,473 (90.3) | 3,845 (100.0) | ||
| March | 444 (10.7) | 3,698 (89.3) | 4,142 (100.0) | ||
| April | 445 (9.9) | 4,059 (90.1) | 4,504 (100.0) | ||
| May | 484 (10.3) | 4,219 (89.7) | 4,703 (100.0) | ||
| June | 419 (9.0) | 4,221 (91.0) | 4,640 (100.0) | ||
| July | 398 (11.0) | 3,230 (89.0) | 3,628 (100.0) | ||
| August | 448 (11.0) | 3,640 (89.0) | 4,088 (100.0) | ||
| September | 447 (9.8) | 4,117 (90.2) | 4,564 (100.0) | ||
| October | 429 (9.3) | 4,168 (90.7) | 4,597 (100.0) | ||
| November | 366 (8.8) | 3,805 (91.2) | 4,171 (100.0) | ||
| December | 324 (11.0) | 2,632 (89.0) | 2,956 (100.0) | ||
| Physicians | |||||
| Present | 716 (9.4) | 6,917 (90.6) | 7,633 (100.0) | 0.118 | |
| Not present | 4,252 (10.0) | 38,442 (90.0) | 42,694 (100.0) |
*Community size is based on the 2016 census data [14]. As it was implicit in the statistical analysis, yearly variation has no impact on the modeling. Accurate annual data is also not available.
Rate of medical travel
The patients flown per day, on average, across the Qikiqtaaluk region is 19.7. Approximately 1.9 of these flights are for medevac, while 17.7 are for schedevac flights.
Pangnirtung initiates more travels than any other community, averaging 3.2 patients per day with Grise Fiord and Resolute Bay combined initiating the least, at less than 1 flight per day on average (0.76). However, Igloolik initiated more medevac flights per day, on average compared to any other community, at 0.35, or one every 3 days. In comparison, Grise Fiord and Resolute Bay on average only initiated a medevac flight once every 20 days. Table 3 shows the average rate of medical travel per day by community and travel type.
Table 3.
Average rate of medical travel, by community and flight type, 1 January 2012 to 31 December 2018.
| Community | Number of Patients flown (all)/day | Number of Medevacs/day | Number of Schedevacs/day |
|---|---|---|---|
| Arctic Bay | 1.53 | 0.12 | 1.41 |
| Kinngait | 2.61 | 0.24 | 2.37 |
| Clyde River | 2.15 | 0.24 | 1.91 |
| Grise Fiord/Resolute | 0.76 | 0.06 | 0.70 |
| Sanirajak | 1.46 | 0.15 | 1.31 |
| Igloolik | 3.09 | 0.35 | 2.74 |
| Kimmirut | 0.90 | 0.07 | 0.83 |
| Pangnirtung | 3.19 | 0.29 | 2.91 |
| Pond Inlet | 2.89 | 0.34 | 2.55 |
| Qikiqtarjuaq | 1.10 | 0.08 | 1.02 |
| Qikiqtaaluk Region | 19.68 | 1.94 | 17.74 |
Association between physician presence and medical travel
Univariate analysis
The results from univariate analysis of the association between physician presence and the rate of medical travel are shown in Table 4. The unadjusted rate ratio showed a 27% increase in the rate of flights per day per community when a physician was present, compared to when no physicians were present within the community (RR: 1.27, 95% CI: 1.24 to 1.30, p < 0.001). Similar findings were present regardless of the type of medical flight.
Table 4.
Crude association between the rate of flights and the presence of physicians, by flight type, 1 January 2012 to 31 December 2018.
| Flights | RR | 95% CI | p-value |
|---|---|---|---|
| All | 1.27 | 1.24 to 1.30 | <0.001 |
| Medevac | 1.19 | 1.10 to 1.29 | <0.001 |
| Sched-Evac | 1.28 | 1.25 to 1.31 | <0.001 |
Comparing physician present to not present.
Model building strategy
The multivariable Poisson regression model building strategy is summarised in Table 5. The final model (model 4) controlled for community, year, and month. All significantly improved the fit of the model and provided some evidence for positive confounding on the association between physician presence and rate of medical travel.
Table 5.
Regression of the rate of patient flown per day depending on the presence of physicians, confounders entered into model in forward stepwise process.
| Model # | Parameters | RR | p-value* |
|---|---|---|---|
| 1 | Crude | 1.27 | <0.001 |
| 2 | Community | 1.09 | <0.001 |
| 3 | Community, Year | 1.07 | <0.001 |
| 4 | Community, Year, Month | 1.06 | <0.001 |
* p-value to test the assumption that the current model is better than the model above it.
Multivariable analysis
After adjusting for temporal (year and month) and geographic (community) characteristics, a 5% reduction was observed in the rate of medevacs when a physician was present in a community in any given month compared to when no physicians were present. However, we are 95% confident that this observed difference in the rate of flights could have been as much as a 13% reduction or as high as a 3% increase. Therefore, there is no evidence that there is a true change in medevac rates when a physician was present compared to when no physicians were present in the community (RR = 0.95, 95% CI: 0.87 to 1.03, p = 0.214). There does appear to be strong evidence for an association between the rate of schedevac flights and physician presence, with the data showing a 7% increase in the rate of schedevac flights when a physician was present compared to when no physicians were present (RR = 1.07, 95% CI: 1.05 to 1.10, p < 0.001).
Missing data
Missing data was excluded from the analysis assessing the association between the number of flights and physician presence. The missing data represents less than one one-hundredths of a percent and therefore likely had no impact on the results. As such, multiple imputation techniques were not used.
Sub-group analysis
When stratified by community, the association between physician presence and rate of medevac flights remained significant for Clyde River and Grise Fiord/Resolute Bay. Clyde River saw a 29% reduction and Grise Fiord/Resolute Bay saw an 82% reduction in Medevac flights when a physician was present compared to when no physicians were present (Clyde River RR = 0.71, 95% CI: 0.54 to 0.94, p = 0.017; Grise Fiord/Resolute Bay RR = 0.18, 95% CI: 0.43 to 0.72, p = 0.16).
The association between physician presence and rate of schedevac flights was also significant for Clyde River and Grise Fiord/Resolute Bay, with a 13% increase and 35% decrease in rate, respectively, when a physician was present. Kinngait, Igloolik, Kimmirut, and Pangnirtung also saw increases in the rate of schedevac flights when a physician was present, while Arctic Bay saw a decrease. No significant differences were detected in Sanirajak, Pond Inlet, or Qikiqtarjuaq. The adjusted association between the rate of medical flights and physician presence, by flight type, within each community is presented in Table 6.
Table 6.
Adjusted association between the rate of flights and the presence of a physician, by flight type and community, 1 January 2012 to 31 December 2018.
| Community | RR | 95% CI | p-value* |
|---|---|---|---|
| All | |||
| Arctic Bay | 0.86 | 0.77 to 0.96 | 0.006 |
| Kinngait | 1.11 | 1.04 to 1.18 | 0.001 |
| Clyde River | 1.08 | 1.00 to 1.17 | 0.061 |
| Grise Fiord/Resolute | 0.62 | 0.50 to 0.76 | <0.001 |
| Sanirajak | 1.08 | 0.97 to 1.20 | 0.159 |
| Igloolik | 1.09 | 1.03 to 1.15 | 0.003 |
| Kimmirut | 1.34 | 1.17 to 1.54 | <0.001 |
| Pangnirtung | 1.07 | 1.01 to 1.13 | 0.020 |
| Pond Inlet | 1.05 | 0.98 to 1.12 | 0.150 |
| Qikiqtarjuaq | 1.06 | 0.94 to 1.20 | 0.358 |
| Medevac | |||
| Arctic Bay | 0.83 | 0.56 to 1.21 | 0.325 |
| Kinngait | 0.91 | 0.74 to 1.13 | 0.411 |
| Clyde River | 0.71 | 0.54 to 0.94 | 0.017 |
| Grise Fiord/Resolute | 0.18 | 0.43 to 0.72 | 0.016 |
| Sanirajak | 1.21 | 0.89 to 1.65 | 0.231 |
| Igloolik | 1.06 | 0.89 to 1.25 | 0.504 |
| Kimmirut | 1.10 | 0.65 to 1.88 | 0.720 |
| Pangnirtung | 0.83 | 0.69 to 1.02 | 0.070 |
| Pond Inlet | 1.17 | 0.98 to 1.40 | 0.082 |
| Qikiqtarjuaq | 0.92 | 0.57 to 1.49 | 0.729 |
| Sched-Evac | |||
| Arctic Bay | 0.86 | 0.77 to 0.96 | 0.010 |
| Kinngait | 1.13 | 1.06 to 1.21 | <0.001 |
| Clyde River | 1.13 | 1.04 to 1.23 | 0.005 |
| Grise Fiord/Resolute | 0.65 | 0.53 to 0.81 | <0.001 |
| Sanirajak | 1.06 | 0.95 to 1.19 | 0.277 |
| Igloolik | 1.10 | 1.03 to 1.16 | 0.003 |
| Kimmirut | 1.36 | 1.18 to 1.57 | <0.001 |
| Pangnirtung | 1.09 | 1.03 to 1.16 | 0.003 |
| Pond Inlet | 1.03 | 0.96 to 1.10 | 0.373 |
| Qikiqtarjuaq | 1.07 | 0.94 to 1.22 | 0.289 |
Adjusted for month and year; stratified by community
*The results in this table are exploratory and presented to highlight potential variation across communities. No corrections were applied for multiple comparisons. P-values should be interpreted with caution, as a small number of statistically significant findings may be due to chance alone.
Sensitivity analysis
After adjusting the travel days to physician “not-present”, the data produced stronger effects and were more likely to show a significant association between schedevac flights and physician presence. However, there were two notable exceptions: Arctic Bay and Kimmirut. In Arctic Bay, the association between schedevac flights and physician presence changed from a significant 14% decrease in rate of flights when a physician was present (RR = 0.86, 95% CI: 0.77 to 0.96) to a non-significant finding in the sensitivity analysis, indicating an 11% increase in the rate of flights when a physician was present (RR = 1.11, 95% CI: 0.98 to 1.26, p = 0.102). In Kimmirut, the initial 36% increase in rate of schedevac flights when a physician was present (RR = 1.36, 95% CI: 1.18 to 1.57, p < 0.001) was no longer present in the sensitivity analysis, showing almost no difference in the rate of schedevac flights when a physician was present. (RR = 1.01, 95% CI: 0.82 to 1.28, p = 0.924)
For medevac flights, there were no statistically significant differences in the findings between the two analyses, except for Clyde River, which no longer show any significant effect on the association between medevac flights and the presence of a physician (RR = 0.73, 95% CI: 0.52 to 1.03, p = 0.075). The results of the sensitivity analysis are presented in Table 7.
Table 7.
Adjusted association between the rate of flights and the presence of a physician (not present on flight days), by flight type and community, 1 January 2012 to 31 December 2018.
| Community | RR | 95% CI | p-value* |
|---|---|---|---|
| All | |||
| Arctic Bay | 1.09 | 0.96 to 1.23 | 0.168 |
| Kinngait | 1.24 | 1.15 to 1.33 | <0.001 |
| Clyde River | 1.20 | 1.09 to 1.32 | <0.001 |
| Grise Fiord/Resolute | 0.45 | 0.33 to 0.60 | <0.001 |
| Sanirajak | 1.21 | 1.07 to 1.37 | 0.002 |
| Igloolik | 1.14 | 1.07 to .22 | <0.001 |
| Kimmirut | 1.01 | 0.83 to 1.24 | 0.902 |
| Pangnirtung | 1.18 | 1.11 to 1.26 | <0.001 |
| Pond Inlet | 1.19 | 1.11 to 1.28 | <0.001 |
| Qikiqtarjuaq | 1.30 | 1.13 to 1.51 | <0.001 |
| Medevac | |||
| Arctic Bay | 0.86 | 0.54 to 1.37 | 0.529 |
| Kinngait | 0.88 | 0.68 to 1.14 | 0.321 |
| Clyde River | 0.73 | 0.52 to 1.03 | 0.075 |
| Grise Fiord/Resolute | 0.13 | 0.02 to 0.91 | 0.040 |
| Sanirajak | 1.37 | 0.95 to 1.96 | 0.089 |
| Igloolik | 1.05 | 0.86 to 1.28 | 0.622 |
| Kimmirut | 1.04 | 0.51 to 2.12 | 0.915 |
| Pangnirtung | 1.00 | 0.81 to 1.25 | 0.977 |
| Pond Inlet | 1.13 | 0.91 to 1.39 | 0.271 |
| Qikiqtarjuaq | 0.76 | 0.39 to 1.47 | 0.411 |
| Sched-Evac | |||
| Arctic Bay | 1.11 | 0.98 to 1.26 | 0.102 |
| Kinngait | 1.27 | 1.18 to 1.37 | <0.001 |
| Clyde River | 1.26 | 1.14 to 1.39 | <0.001 |
| Grise Fiord/Resolute | 0.47 | 0.35 to 0.63 | <0.001 |
| Sanirajak | 1.19 | 1.05 to 1.36 | 0.008 |
| Igloolik | 1.15 | 1.08 to 1.24 | <0.001 |
| Kimmirut | 1.01 | 0.82 to 1.24 | 0.924 |
| Pangnirtung | 1.20 | 1.13 to 1.28 | <0.001 |
| Pond Inlet | 1.20 | 1.11 to 1.29 | <0.001 |
| Qikiqtarjuaq | 1.35 | 1.16 to 1.57 | <0.001 |
Adjusted for month and year; stratified by community
*The results in this table are exploratory and presented to highlight potential variation across communities. No corrections were applied for multiple comparisons. P-values should be interpreted with caution, as a small number of statistically significant findings may be due to chance alone.
Discussion
The remote geography of Canada’s northern communities means there will always be some reliance on aeromedical evacuations to provide emergency and specialist care to the local populations. There should be, however, a concerted effort to reduce unnecessary transport as studies have shown that displacing people from their home communities causes increased stress and a loss of autonomy [11]. Studies have hypothesised that there are numerous factors accounting for increased medevacs including inadequate staffing, high nursing turnover, limited scope of practice of providers, risk aversion, exhaustion, burnout, and lack of trust between front-line nurses and hospital-based physicians [15,16]. Profound social, cultural, economic and historical factors are also responsible for poorer health outcomes of the mainly Inuit population in Nunavut [17–19].
Placing full-time family physicians in remote, fly-in communities has several challenges including recruitment, lack of accommodations, reduced access to professional development, and personal and professional isolation [9,20–23]. The remote communities of Nunavut, Canada, all have populations much smaller than the “critical minimum population base” of 5000 defined by Wakerman et al. as the community size required to sustain a full-time physician practice [9]. However, this population size may be smaller in Nunavut considering the high costs of medical travel and that Nunavut does not use a fee-for-service payment model.
The authors of this study hypothesised that the presence of a physician would lower the rates of aeromedical evacuation and therefore reduce overall costs. Interestingly, this study found no statistically significant difference in rates of medevac when a physician was present compared to not. This is not necessarily unexpected considering that these patients were likely sick enough to require the resources and time beyond what is available in remote community health centres. However, there was an unexpected finding that the rate of less urgent schedevacs significantly increased for most remote communities when a physician was present. In isolation, this may indicate an increase in costs for the system, but it does not factor in long-term health outcomes and costs related to them. Importantly, our data set did not include any information on health outcomes and looked only at travel rates. It is widely accepted that prevention and primary care are significantly more cost-effective than secondary treatments [24–26]. This is likely magnified in a geographical setting where specialist care requires out-of-territory travel. Thus, the higher rates of schedevacs found during visits by FPs, while potentially costlier at first look, likely corresponds to overall savings to the system in the longer-term. Beyond the cost analysis, this study should not be used to discourage or diminish community visits by FPs, as continuity of care is also widely accepted as significantly improving health outcomes [27–30].
Effect on health outcomes
Data shows that despite some of the highest health-care expenditures per person, health outcomes for Canadian Inuit lag those of non-indigenous Canadians [31,32]. The life expectancy at birth is 10 years lower in Nunavut compared to the rest of Canada [33]. These health disparities cannot be solely attributed to the FIFO model, but there is evidence that this model of care has several issues that are less-than-optimal to a community’s health [13,34–37].
When taking this data into consideration as an isolated factor, we are unable to recommend the full-time staffing of physicians in these remote communities from a pure cost-savings perspective. However, this research raises further questions including whether full-time access to a primary care physician would alter morbidity and mortality outcomes. Of note, telehealth was formally established in Nunavut in 2020, after the studied period. Studying its impact on rates of patient travels would be an interesting avenue of research.
The demand for acute care in Nunavut’s remote communities, coupled with high staffing turnover, often supersedes the provision of primary care. This leads to reduced follow-up for chronic conditions and post discharge from hospital [22,31,32,38,39]. Lack of primary care can cause increases in inpatient admissions, length of stay, and transfers to tertiary care centres [40–42]. Also, the presence of a physician can improve outcomes in major trauma [43] as well as team cohesion. Even in stable chronic conditions (e.g. epilepsy [44]), medevacs will often be necessary despite optimal control and medevacs will always be needed in response to emergencies. Finally, the effectiveness of specialist outreach in this model has been shown to be dependent on good primary care [6].
From a cohesive and integrated health care delivery team perspective, having a physician on site would allow for better division of tasks, thereby reducing staff burnout. The absence of a family physician increases stress on front-line staff, which in turn increases staff turnover [45–47]. Furthermore, relief agency nursing staff in the remote towns, and locum physicians in the hub hospital often lack the knowledge of the remote context, leading to less-than-ideal care decisions and poor communication during transport [38,48,49]. If avoidable, these represent significant unnecessary costs to the health care system as well as detrimental experiences to the patient’s wellbeing. Medical travel accounts for such a significant part of the Territorial healthcare budget. It is tempting to hypothesise that community-based physicians could not only improve health care delivery but also reduce its costs over time.
Covariates
Assessing confounding variables was limited given the variables available in the medical travel database. Each potential confounder was examined to consider associations with both medical travel and physician presence. No variables were determined to be on the causal pathway between medical travel and physician presence.
The variables assessed for potential confounding were community, year, and month. Community and year were strongly associated with both physician presence and number of flights and therefore were potential confounding variables. In the model building strategy, all three variables were included in a stepwise process and the crude rate ratio for the association between the rate of medical travel and physician presence was reduced, indicating that each variable may have been a positive confounder on this association. As shown in Tables 6 and 7, we conducted multiple subgroup comparisons by community and flight type. These were exploratory in nature and not based on pre-specified hypotheses. As such, we did not apply formal corrections for multiple comparisons. The findings should be interpreted with caution, as the increased number of comparisons raises the likelihood that some significant results may be due to chance.
Age and sex were available in the medical travel database but were not included in the assessment for confounding. The data was aggregated during the analysis and presence of a physician is an attribute of the community rather than the individual. As such, the analysis was at the community level and not at the individual level. An important aspect to consider is also the high rates of travels related to pregnancy which potentially account for the higher number of females needing medical flights [50]. This was also noted in a cohort study from the Kivalliq region [42].
Conclusion
Aeromedical evacuations are a necessary component of the fly-in, fly-out model of healthcare in Canada’s remote northern regions. This study shows that the presence of a physician in a remote, fly-in community does not alter the rate of emergency aeromedical evacuations for patients. It does, however, increase the rate of less urgent patient transfers. This raises the question of whether there is a significant reduction in time to diagnosis or treatment of certain conditions. Further studies need to be performed assessing patient outcomes and patient satisfaction under these conditions.
Acknowledgment
We wish to thank Dr Wendy Graham MD, Dr Shabnam Asghari MD PhD, Mr Thomas Heeley MASP and the entire team from the Memorial University 6for6 rural research faculty development program. Michelle Swab from the Health Sciences Library at Memorial University for her assistance with the literature search. Dr Gwen Healey Akearok PhD, Executive and Scientific Director of the Qaujigairtiit Health Research Centre for assistance with study design and research approval. Michelle Doucette, Manager of Population Health Information for her assistance with study design and setting up a data sharing agreement with the GN. The Department of Medical Affairs at the GN for providing funds for travel to attend the 6for6 programme. Finally, we also wish to thank both Reviewer one and two for their thoughtful comments which elevated our discussion.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Disclaimer
The views expressed in this study are those of the authors and are not necessarily endorsed by the Government of Nunavut. The authors declare no conflict of interest.
References
- [1].Paquin V, Sandy G, Perrault-Sullivan G, et al. Twenty “must-read” research articles for primary care providers in Nunavik: scoping study and development of an information tool. Int J Circumpolar Health. 2019. Jan;78(1):1578638. doi: 10.1080/22423982.2019.1578638 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Young TK, Chatwood S.. Health care in the north: what Canada can learn from its circumpolar neighbours. Can Med Assoc J. 2011. Feb;183(2):209–13. doi: 10.1503/cmaj.100948 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Government of Canada . Canada Health Act 2014/15 annual report. 2015. 4. Available from: https://www.canada.ca/en/health-canada/services/health-care-system/reports-publications/canada-health-act-annual-reports/report-2014-2015.htmlp.
- [4].Banerji A, Panzov V, Robinson J, et al. The cost of lower respiratory tract infections hospital admissions in the Canadian Arctic. Int J Circumpolar Health. 2013;72(1):21595. doi: 10.3402/ijch.v72i0.21595 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Creery D, Iyer P, Samson L, et al. Costs associated with infant bronchiolitis in the Baffin region of Nunavut. Int J Circumpolar Health. 2005. Feb;64(1):38–45. doi: 10.3402/ijch.v64i1.17952 [DOI] [PubMed] [Google Scholar]
- [6].Gruen R, Bailie R. Specialist clinics in remote Australian Aboriginal communities: where rock art meets rocket science. J Health Serv Res Policy. 2004. Oct;9(2):56–62. doi: 10.1258/1355819042349844 [DOI] [PubMed] [Google Scholar]
- [7].Lyle D, Saurman E, Kirby S, et al. What do evaluations tell us about implementing new models in rural and remote primary health care? Findings from a narrative analysis of seven service evaluations conducted by an Australian Centre of Research Excellence. Rural Remote Health. 2017;17(3):3926. doi: 10.22605/RRH3926 [DOI] [PubMed] [Google Scholar]
- [8].Wakerman J, Curry R, McEldowney R. Fly in/fly out health services: the panacea or the problem? Rural Remote Health. 2012. doi: 10.22605/RRH2268 [DOI] [PubMed] [Google Scholar]
- [9].Wakerman J, Humphreys JS, Wells R, et al. Primary health care delivery models in rural and remote Australia: a systematic review. BMC Health Serv Res. 2008. Dec;8(1):276. doi: 10.1186/1472-6963-8-276 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Wilson NW, Couper ID, De Vries E, et al. A critical review of interventions to redress the inequitable distribution of healthcare professionals to rural and remote areas. Rural Remote Health. 2009;9(2):1060. Available from: https://www.rrh.org.au/journal/article/1060 [PubMed] [Google Scholar]
- [11].Mitton C, Dionne F, Masucci L, et al. Innovations in health service organization and delivery in northern rural and remote regions: a review of the literature. Int J Circumpolar Health. 2011;70(5):460–472. doi: 10.3402/ijch.v70i5.17859 [DOI] [PubMed] [Google Scholar]
- [12].Haren MT, Setchell J, John DL, et al. The impacts of withdrawal and replacement of general practitioner services on aeromedical service trends: a 13-year interrupted time-series study in Tennant Creek, Northern Territory. BMC Health Serv Res. 2015. Jun;15(1):456. doi: 10.1186/s12913-015-1110-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Hussain J, Robinson A, Stebbing M, et al. More is more in remote Central Australia: more provision of primary healthcare services is associated with more acute medical evacuations and more remote telephone consultations. Rural Remote Health. 2014. Nov;14(4):2796. doi: 10.22605/RRH2796 [DOI] [PubMed] [Google Scholar]
- [14].Statistics Canada . 2016 census of population. 2016. Available from: https://www12.statcan.gc.ca/census-recensement/2016/index-eng.cfm
- [15].McDonnell L, Lavoie JG, Healey G, et al. Non-clinical determinants of medevacs in Nunavut: perspectives from northern health service providers and decision-makers. Int J Circumpolar Health. 2019. Jan;78(1):1571384. doi: 10.1080/22423982.2019.1571384 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Lee E, Gudmundson B, Lavoie JG. Returning childbirth to Inuit communities in the Canadian Arctic. Int J Circumpolar Health. 2022. Dec;81(1). doi: 10.1080/22423982.2022.2071410 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Inuit Tapiriit Kanatami. Social determinants of Inuit health in Canada. Ottawa, Canada; 2014. [Online]. Available from: https://www.itk.ca/wp-content/uploads/2016/07/ITK_Social_Determinants_Report.pdf [Google Scholar]
- [18].Truth and Reconciliation Comission of Canada . Canada’s residential schools: the Inuit and northern experience. Montréal and Kingston: McGill-Queen’s University Press; 2016. [cited 2025 Aug 8]. https://ehprnh2mwo3.exactdn.com/wp-content/uploads/2021/01/Volume_2_Inuit_and_Northern_English_Web.pdf [Google Scholar]
- [19].Office of the Auditor General of Canada . Health care services — Nunavut: report of the Auditor General of Canada to the Legislative Assembly of Nunavut. Mar 2017. Available: https://www.oag-bvg.gc.ca/internet/English/nun_201703_e_41998.html
- [20].Bourke L, Humphreys JS, Wakerman J, et al. Understanding rural and remote health: a framework for analysis in Australia. Health Place. 2012. May;18(3):496–503. doi: 10.1016/j.healthplace.2012.02.009 [DOI] [PubMed] [Google Scholar]
- [21].Buykx P, Humphreys J, Wakerman J, et al. Systematic review of effective retention incentives for health workers in rural and remote areas: towards evidence‐based policy. Australian J Rural Health. 2010. Jun;18(3):102–109. doi: 10.1111/j.1440-1584.2010.01139.x [DOI] [PubMed] [Google Scholar]
- [22].Peiris D, Wirtanen C, Hall J. Aeromedical evacuations from an East Arnhem Land community 2003–2005: the impact on a primary health care centre. Australian J Rural Health. 2006. Dec;14(6):270–274. doi: 10.1111/j.1440-1584.2006.00828.x [DOI] [PubMed] [Google Scholar]
- [23].Russell DJ, McGrail MR, Humphreys JS, et al. What factors contribute most to the retention of general practitioners in rural and remote areas? Aust J Prim Health. 2012;18(4):289. doi: 10.1071/PY11049 [DOI] [PubMed] [Google Scholar]
- [24].Commonwealth Secretariat Health Division . Cost effectiveness of primary health care. Commonw Secretariat. 1987. doi: 10.14217/9781848594234-en [DOI] [Google Scholar]
- [25].Mountain R, Kim D, Johnson KM. Budget impact analysis of adopting primary care–based case detection of chronic obstructive pulmonary disease in the Canadian general population. CMAJ Open. 2023. Nov;11(6):E1048–E1058. doi: 10.9778/cmajo.20230023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Dale SB, Ghosh A, Peikes DN, et al. Two-year costs and quality in the comprehensive primary care initiative. N Engl J Med. 2016. Jun;374(24):2345–2356. doi: 10.1056/NEJMsa1414953 [DOI] [PubMed] [Google Scholar]
- [27].Hollander MJ, Kadlec H. Financial implications of the continuity of primary care. Perm J. 2015;19(1):4–10. doi: 10.7812/TPP/14-107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Knight JC, Dowden JJ, Worrall GJ, et al. Does higher continuity of family physician care reduce hospitalizations in elderly people with diabetes? Popul Health Manag. 2009. Apr;12(2):81–86. doi: 10.1089/pop.2008.0020 [DOI] [PubMed] [Google Scholar]
- [29].Huang P-T, Kung P-T, Kuo W-Y, et al. Impact of family physician integrated care program on decreasing utilization of emergency department visit and hospital admission: a population-based retrospective cohort study. BMC Health Serv Res. 2020. Dec;20(1):470. doi: 10.1186/s12913-020-05347-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Knight JC, Mathews M, Aubrey-Bassler K. Relation between family physician retention and avoidable hospital admission in Newfoundland and Labrador: a population-based cross-sectional study. CMAJ Open. 2017. Oct;5(4):E746–E752. doi: 10.9778/cmajo.20170007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Bar-Zeev S, Kruske SG, Barclay LM, et al. Adherence to management guidelines for growth faltering and anaemia in remote dwelling Australian Aboriginal infants and barriers to health service delivery. BMC Health Serv Res. 2013;13:250. doi: 10.1186/1472-6963-13-250 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Michiel Oosterveer T, Kue Young T. Primary health care accessibility challenges in remote Indigenous communities in Canada’s north. Int J Circumpolar Health. 2015. Jan;74(1):29576. doi: 10.3402/ijch.v74.29576 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Statistics Canada . Life expectancy at birth and at age 65, by province and territory, threeyear average. Stat Canada. 2017. doi: 10.25318/1310040901eng [DOI] [Google Scholar]
- [34].Busbridge M, Smith A. Fly in/fly out health workers: a barrier to quality in health care. Rural Remote Health. 2015;15(2):3339. doi: 10.22605/RRH3339 [DOI] [PubMed] [Google Scholar]
- [35].Carey TA, Sirett D, Russell D, et al. What is the overall impact or effectiveness of visiting primary health care services in rural and remote communities in high-income countries? A systematic review. BMC Health Serv Res. 2018. Dec;18(1):476. doi: 10.1186/s12913-018-3269-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Hussain R, Maple M, Hunter S, et al. The fly-in fly-out and drive-in drive-out model of health care service provision for rural and remote Australia: benefits and disadvantages. Rural Remote Health. 2015;15(3):3068. doi: 10.22605/RRH3068 [DOI] [PubMed] [Google Scholar]
- [37].Margolis SA. Is fly in/fly out (FIFO) a viable interim solution to address remote medical workforce shortages? Rural Remote Health. 2012;12:2261. doi: 10.22605/RRH2261 [DOI] [PubMed] [Google Scholar]
- [38].Bar-Zeev S, Barclay L, Kruske S, et al. Factors affecting the quality of antenatal care provided to remote dwelling Aboriginal women in northern Australia. Midwifery. 2014;30(3):289–296. doi: 10.1016/j.midw.2013.04.009 [DOI] [PubMed] [Google Scholar]
- [39].Bennett KJ, Probst JC, Vyavaharkar M, et al. Missing the handoff: post-hospitalization follow-up care among rural Medicare beneficiaries with diabetes. Rural Remote Health. 2012;12(3):2097. doi: 10.22605/RRH2097 [DOI] [PubMed] [Google Scholar]
- [40].Lorig KR, Sobel DS, Stewart AL, et al. Evidence suggesting that a chronic disease self-management program can improve health status while reducing hospitalization. Med Care. 1999. Jan;37(1):5–14. doi: 10.1097/00005650-199901000-00003 [DOI] [PubMed] [Google Scholar]
- [41].Quilty S, Valler D, Attia J. Rural general physicians: improving access and reducing costs of health care in the bush. Australian Health Rev. 2014;38(4):420. doi: 10.1071/AH13197 [DOI] [PubMed] [Google Scholar]
- [42].Clark W, Lavoie JG, Mcdonnell L, et al. Trends in Inuit health services utilisation in Manitoba: findings from the qanuinngitsiarutiksait study. Int J Circumpolar Health. 2022. Dec;81(1). doi: 10.1080/22423982.2022.2073069 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [43].Allen O. Anthill and other injuries: a case for mobile allied health teams to remote Australia. Australian J Rural Health. 1996;4(1):33–42. doi: 10.1111/j.1440-1584.1996.tb00184.x [DOI] [PubMed] [Google Scholar]
- [44].Ng MC, Pavlova M. Status epilepticus in the Canadian Arctic: a public health imperative hidden in plain sight. Epilepsia Open. 2021. Dec;6(4):703–713. doi: 10.1002/epi4.12538 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [45].Kulig JC, Stewart N, Penz K, et al. Work setting, community attachment, and satisfaction among rural and remote nurses. Public Health Nurs. 2009. Sep;26(5):430–439. doi: 10.1111/j.1525-1446.2009.00801.x [DOI] [PubMed] [Google Scholar]
- [46].Lenthall S, Wakerman J, Opie T, et al. What stresses remote area nurses? Current knowledge and future action. Australian J Rural Health. 2009. Aug;17(4):208–213. doi: 10.1111/j.1440-1584.2009.01073.x [DOI] [PubMed] [Google Scholar]
- [47].Stewart NJ, D’Arcy C, Kosteniuk J, et al. Moving on? Predictors of intent to leave among rural and remote RNs in Canada. J Rural Health. 2011. Jan;27(1):103–113. doi: 10.1111/j.1748-0361.2010.00308.x [DOI] [PubMed] [Google Scholar]
- [48].Young SK, Young TK. Assessing clinical support and inter-professional interactions among front-line primary care providers in remote communities in northern Canada: a pilot study. Int J Circumpolar Health. 2016. Jan;75(1):32159. doi: 10.3402/ijch.v75.32159 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [49].Birch S, Minore B, Boone M, et al. The effects of nursing turnover on continuity of care in isolated First Nation communities. Can J Nurs Res. 2005;37(1):86–100. https://cjnr.archive.mcgill.ca/article/view/1928 [PubMed] [Google Scholar]
- [50].Pauktuutit Inuit Women of Canada . Improving access to sexual and reproductive health service across Inuit Nunangat: national Inuit midwifery forum report. Apr 2024. [Online]. Available from: https://pauktuutit.ca/wp-content/uploads/PAUKTUUTIT-National-Inuit-Midwifery-Forum-Report_EN.pdf