Antibiotic dispensation rates among participants in community-driven health research projects in Arctic Canada

Background Community-driven projects that aim to address public concerns about health risks from H. pylori infection in Indigenous Arctic communities (estimated H. pylori prevalence = 64%) show frequent failure of treatment to eliminate the bacterium. Among project participants, treatment effectiveness is reduced by antibiotic resistance of infecting H. pylori strains, which in turn, is associated with frequent exposure to antibiotics used to treat other infections. This analysis compares antibiotic dispensation rates in Canadian Arctic communities to rates in urban and rural populations in Alberta, a southern Canadian province. Methods Project staff collected antibiotic exposure histories for 297 participants enrolled during 2007–2012 in Aklavik, Tuktoyaktuk, and Fort McPherson in the Northwest Territories, and Old Crow, Yukon. Medical chart reviews collected data on systemic antibiotic dispensations for the 5-year period before enrolment for each participant. Antibiotic dispensation data for urban Edmonton, Alberta (average population ~ 860,000) and rural northern Alberta (average population ~ 450,000) during 2010–2013 were obtained from the Alberta Government Interactive Health Data Application. Results Antibiotic dispensation rates, estimated as dispensations/person-years (95% confidence interval) were: in Arctic communities, 0.89 (0.84, 0.94); in Edmonton, 0.55 (0.55, 0.56); in rural northern Alberta, 0.63 (0.62, 0.63). Antibiotic dispensation rates were higher in women and older age groups in all regions. In all regions, the highest dispensation rates occurred for β-lactam and macrolide antibiotic classes. Conclusions These results show more frequent antibiotic dispensation in Arctic communities relative to an urban and rural southern Canadian population.


Background
Antibiotic-resistant bacteria are a major public health concern because they limit the effectiveness of therapeutic options available for the treatment of bacterial infections. Much evidence suggests that exposure to antibiotics leads to antibiotic-resistant bacterial infections.
In particular, consistent evidence shows an association between frequent exposure to antibiotics for the treatment of unrelated bacterial infections and the prevalence of antibiotic-resistant Helicobacter pylori infection [1][2][3][4][5][6].
The Canadian North Helicobacter pylori (CANHelp) Working Group links northern Canadian communities, health care providers and regional health officials with University of Alberta investigators to conduct community-driven research aimed at addressing concerns about health risks from H. pylori infection; the ultimate goal is to inform public health policy pertaining to control of H. pylori infection [7]. In Canada, northern Indigenous communities are disproportionately burdened by H. pylori infection compared to multi-ethnic populations in southern Canada [8][9][10]. Available data show stomach cancer rates to be elevated in Indigenous Arctic populations as well, [11][12][13] an occurrence of great concern to affected communities and their health care providers.
H. pylori is a bacterium that colonizes the lining of the stomach and/or duodenum, [14] where it is nearly always accompanied by gastritis. This infection often persists indefinitely; in persistent cases, it increases the risk of peptic ulcer disease and stomach cancer [15][16][17][18]. Treatment to eliminate H. pylori infection has been observed to improve peptic ulcer healing, decrease peptic ulcer recurrence and reduce the risk of stomach cancer [19,20]. Frequent treatment failure in populations with high prevalence of H. pylori infection is a major impediment to the development of effective H. pylori control strategies for such populations [21]. CANHelp community projects collected data on participants' exposure to antibiotics as part of inquiry aimed at identifying causes of treatment failure.
Few studies have collected information on antibiotic exposure in geographically defined communities, even though it is known that the frequency of antibiotic use varies by geographic region, as well as other factors including age and sex [22][23][24]. Geographic differences in prescribing practices or non-prescription access to antibiotics likely contribute to differences in the occurrence of antibiotic-resistant infections across geographic regions [25,26].
The aims of the current study are to describe antibiotic dispensation rates, as a measure of the frequency of overall exposure to systemic antibiotics, by community, sex, age and antibiotic class, in Canadian Arctic communities and compare these rates to those of urban and rural outpatient populations in the southern Canadian province of Alberta. These results will shed light on whether increased exposure to antibiotics is a potential explanation for reduced effectiveness of antibiotics in Indigenous Arctic populations, thus making them vulnerable to poor infection outcomes.  [39,40]. Each community project, guided by a planning committee comprising community leaders, included several components: surveys to collect risk factor data; non-invasive screening for H. pylori infection; endoscopy with gastric biopsy for pathological and microbiological assessment; treatment; and ongoing knowledge exchange activities. Diverse outreach strategies sought to encourage all community residents to participate. Project staff obtained informed consent from participants 17 years of age or older, parental consent for children under 17 years of age, and assent from 7 to 16year-old participants.

This analysis includes participants in community
The lead author (KW) collected chart review data for her MSc thesis research on factors associated with antibiotic-resistant H. pylori infection; [41] specifically, she collected antibiotic exposure history pertaining to CANHelp community project participants who fulfilled either or both of the following criteria for inclusion in this analysis: 1) available H. pylori isolates cultured from gastric biopsies and tested for antibiotic susceptibility; 2) completed treatment to eliminate H. pylori infection and were tested after treatment to assess post-treatment infection status. We collected each participant's antibiotic exposure history from their medical chart, housed in community health centres, using a chart review tool to record systemic antibiotics prescribed for any reason during the 5 years before the participant's enrolment date. This 5-year exposure period spanned different calendar years across communities because community projects were launched sequentially: our chart review McPherson (a few participants enrolled after the launch year, so their antibiotic exposure period began slightly later). Data collected from medical charts included: demographic factors; frequency of antibiotic prescriptions; types of antibiotics prescribed; and the reason for each prescription. An antibiotic prescription was defined as a single prescription of at least one systemic antibiotic regardless of the dose, dosing frequency or duration of the prescription.
As a measure of antibiotic use frequency, we estimated antibiotic dispensation rates, defining dispensation as the process by which a prescribed medication is given to a patient for whom a prescription was written. We used prescriptions as proxies for dispensations because medical charts noted prescriptions rather than dispensations, and because prescriptions given to patients at health centres in participating northern communities are dispensed routinely from locally stocked medication by health centre staff at the time of prescription; thus, in this study population, it can be assumed that each prescription noted in a medical chart was dispensed.
We expressed dispensation rates as the average of the number of antibiotic courses dispensed per person per year during each participant's 5-year review period. We calculated these rates by dividing the total number of systemic antibiotic courses prescribed for all participants during the 5-year review period by the product of the number of participants and the sum of the number of years reviewed in the medical chart of each participant. We estimated antibiotic dispensation rates by community (Aklavik, Old Crow, Fort McPherson, Tuktoyaktuk), sex, age (categorized in 20-year age groups), and antibiotic class (β-lactams, macrolides, nitroimidazoles, nitrofurans, fluoroquinolones, tetracyclines, rifamycins).
To put the antibiotic dispensation rates observed in the western Arctic communities in perspective, we compared them to rates observed in Alberta, a western Canadian province located directly south of the Northwest Territories. We estimated antibiotic dispensation rates for the outpatient populations of urban Edmonton (Alberta's capital) and rural northern Alberta using the online Alberta Government Interactive Health Data Application (IHDA), [42] which incorporates data from the following sources: the Pharmaceutical Information Network (PIN) database; the Alberta Health Care Insurance Plan Adjusted Mid-Year Population Registry Files; and the Alberta Health and Wellness Postal Code Translation File. The PIN captures all drug dispensation events occurring in community pharmacies across Alberta. Aggregate PIN data are available through the IHDA by geography, age group, sex, year, and antibiotic class. We restricted estimates of the Edmonton antibiotic dispensation rate to the population residing within the city limits (subzones Z4.1-Z4.4) and the rural northern Alberta rate to residents of the Alberta North Zone (subzones Z5.1-Z5.5); when we accessed data for both regions in March 2017, the only available data covered 2010 through 2013.
We estimated the outpatient antibiotic dispensation rates by dividing the number of antibiotic courses dispensed by the sum of the population during each year from 2010 through 2013. For statistical comparison, we used an estimation approach, presenting 95% confidence intervals (CI) for all estimated rates, rather than statistical significance testing, following best practice in statistical methods for epidemiology [43,44]. To compare the Arctic community population to each of the two Alberta populations, we estimated rate differences and 95% CIs.

Results
We collected antibiotic exposure histories from medical charts of CANHelp community project participants: 164 from Aklavik; 67 from Old Crow; 52 from Fort McPherson; and 14 from Tuktoyaktuk. Table 1 presents the demographic characteristics of the study population.
Among the 297 participants, the median number of systemic antibiotic prescriptions during the 5-year review period was 3 (IQR, 1-6; range, 0-38; mean, 4.3; SD, 4.6). Table 2 presents estimated antibiotic dispensation rates in CANHelp community project participants and the outpatient populations of Edmonton and rural northern Alberta. The overall antibiotic dispensation rate was notably higher in the CANHelp community project study population (8.9 courses per 10 person-years) than in northern rural Alberta (6.3 courses per 10 person-years) or Edmonton (5.5 courses per 10 person-years). The estimated antibiotic dispensation rate in western Arctic communities was 3.3 (95% CI: 2.9, 3.8) courses per 10 personyears higher than the Edmonton outpatient population rate and 2.6 (95% CI: 21, 31) courses per 10 person-years higher than the rural northern Alberta outpatient rate. Northern Alberta is a large geographic region with two small cities and vast sparsely populated areas. Table 3 presents estimated antibiotic dispensation rates for sub-zones of this region. Of note, the two cities have antibiotic dispensation rates slightly lower than Edmonton's, while rates in the sparely populations zones range from 5.9 to 7.3 per 10 person-years, all substantially lower than the estimated rates in the Arctic communities. Table 4 presents estimated antibiotic dispensation rates for the Arctic communities and the outpatient populations of Edmonton and rural northern Alberta by sex, age and antibiotic class. Across all populations, the estimated dispensation rate was higher in woman than in men, the highest rates occurred in senior adults, and β-lactam antibiotics were dispensed at a substantially higher rate than other antibiotic classes. Stratification by age group and sex shows that dispensation rates were substantially higher in the Arctic communities within each age group and in both sexes. Of note, compared to the Alberta populations, dispensation rates in the Arctic communities were similar for β-lactams and macrolides, lower for fluoroquinolones and tetracyclines, higher for nitrofurans, and substantially higher for nitroimidazoles.

Discussion
Evidence from this study suggests that antibiotics are dispensed more frequently to western Arctic Canadians than to either urban or rural residents of the southwestern Canadian province of Alberta, independently of age and sex. This study showed additionally that antibiotics were dispensed more frequently to women relative to men and to the elderly relative to other age groups in each of the 3 settings investigated. Dispensation rates for the most frequently dispensed antibiotic classes, βlactams and macrolides, was similar in the 3 populations, while fluoroquinolones and tetracyclines were dispensed less frequently and nitrofurans and nitroimidazoles were dispensed more frequently in the Arctic communities.
In the present study, the dispensation rate for systemic antibiotic prescriptions among participants in CANHelp Working Group community projects was 8.9 prescriptions per 10 person-years; an even higher antibiotic prescription rate of 15 per 10 person-years was estimated for an Alaska Native population in 2003 [2]. The urban antibiotic dispensation rates we estimated for Edmonton and the two small northern Alberta cities ranged from 5.2-5.5 per 10 person-years; a slightly lower antibiotic dispensation rate of 4.4 per 10 person-years was estimated from a 1992 study of United States physician practices [45]. More frequent use of antibiotics in rural and remote Arctic regions of North America relative to urban settings in this region is compatible with diverse explanations including higher incidence of infectious diseases, more limited access to diagnostic technology used to confirm bacterial infections leading to more frequent dispensation of antibiotics without a confirmed diagnosis, and different prescription practices resulting from more frequent provision of health care by nurses in phone consultation with physicians rather than by physicians directly [46,47].
Variation in the use of antibiotics for the treatment of common bacterial infections may contribute to variation  in antibiotic-resistant infection across geographic regions and sociodemographic groups [25,26]. For example, the frequency of clarithromycin-resistant H. pylori infection has been reported to be higher in children than in adults, with evidence suggesting that this contrast is due to more frequent use of macrolide antibiotics in children for the treatment of respiratory tract infections [1,48]. The frequency of metronidazole-resistant H. pylori infection has been reported to be higher in women than in men, [49] perhaps due to widespread use of nitroimidazole antibiotics for the treatment of gynaecological infections [2,[49][50][51][52][53][54]. For example, in a study conducted in the United Kingdom, nitroimidazole antibiotics were prescribed more frequently to women than to men, and the prevalence of metronidazole-resistant H. pylori infection was also higher in women relative to men [4]. Similarly, reports of higher prevalence of tetracyclineresistant H. pylori infection in women relative to men may be due to more frequent use of tetracycline in women for the treatment of urogenital infections [55].
Higher prevalence of fluoroquinolone-resistant H. pylori infection in women relative to men and in youth relative to adults may be associated with more frequent use of fluoroquinolone antibiotics for the treatment of urogenital and respiratory tract infections, respectively [55].
An important limitation of the present analysis is the potential underestimation of antibiotic use among participants in the Arctic Canadian community H. pylori projects. Participants' medical charts only capture antibiotics dispensed at their community health centre. Many residents of the project communities spend time away from home and may receive health care in other locations. Most residents of northern Canadian communities, however, receive most of their health care at their community health centre; it is, therefore, likely that most antibiotics dispensed during the review period for this study were captured by our chart review. The data sources used for the Alberta populations would similarly miss prescriptions dispensed outside Alberta; however, given that Albertans are all entitled to health care provided by the provincial government, they are not likely to seek outpatient care for infections outside the province.
Another limitation pertains to our differential strategy for capturing dispensation events across territorial and provincial populations. In the northern communities included here, prescription events equate with dispensations. In Alberta, however, primary nonadherence is possible: an individual may be prescribed a drug but may not fill the prescription, precluding its dispensation  [56,57]. While this is unlikely to impact our ability to assess differences in dispensation patterns, it does limit our ability to investigate plausible explanations for results. Similarly, we were unable to assess whether a person who was dispensed an antibiotic used the medication as prescribed. As a result, we are restricted to reporting differences in potential exposure to antibiotics across these populations.
In this study, the time periods for which antibiotic dispensation rates were estimated differ in each of the Arctic communities (2002-2007, 2005-2010, 2006-2011, 2007-2012) due to the sequential conduct of these projects. These periods overlap to different degrees with the time period captured by the publicly accessible Alberta data, which was 2010-2013 at the time of data analysis. We would not, however, expect substantial changes in antibiotic dispensation during 2002 to 2013 based on studies of outpatient populations in Ontario, Canada and in the United States, which suggest that antibiotic dispensation rates remained relatively stable across age groups during this time period [58,59].

Conclusions
This study reveals more frequent dispensation of antibiotics in Arctic Canada relative to urban and rural populations in southern Canada. These results suggest that increased exposure to antibiotics is a potential explanation for reduced effectiveness of antibiotics in Indigenous Arctic populations for treating infections such as H. pylori infection. More generally, Indigenous Arctic populations may be particularly vulnerable to antibioticresistant infections associated with frequent exposure to antibiotics. The evidence generated by this analysis will be useful for developing strategies aimed at reducing health disparities arising from inequitable health care in Indigenous Arctic populations relative to other North American populations.