The Economic Burden of Asthma in Canada

Overview

The study will consist of three phases: Phase A: General population survey for estimation of the prevalence of asthma, and prospective collection of resource utilization and quality-of-life data for 12 months, Phase B: comparison of healthcare resource usage collected prospectively to the data collected using administrative data in the population recruited in Phase A, and Phase C: Economic modeling of asthma to extrapolate the findings across Canada and into the future years.

Phase A: This phase will be a prospective cohort study of randomly chosen residents of two census subdivisions of BC with a self-reported physician diagnosis of asthma. The study participants will be recruited via landline and cell phone-based random digit dialing across two regions in British Columbia: the Census Sub-Division of the city of Vancouver and the Census Division of Central Okanagan. The target area was chosen based on Census Sub-Division as it provides a clear definition of the target population for whom demographic and socioeconomic data is available through national and provincial census and surveys. The Okanagan Census Sub-Division was particularly chosen as it provides data on both urban and rural populations (18.6% rural population in 2006). According to 2006 census profile, the population of these two Census Sub-Divisions is, respectively, 578,041 and 162,276. Accordingly, about 78% of participants would be recruited from Vancouver, and 22% from Okanagan. The study will randomly sample participants from all the area codes in these regions. The target population will include household residents residing in telephone exchanges within the target Census Sub-Divisions. Random digit dialing will be geographically stratified with an associated sample weight that will approximate a probability sample of the individuals in the population. Previous studies have shown that acceptably high response rates can be achieved among Canadian participants with asthma. Random digit dialing samples have been shown to be more representative than population samples drawn using alternative methods such as telephone directories or electronic white pages. However, random digit dialing is not free of bias. Non-random non-response is an issue, and it has been shown that reducing the non-response rate may directly reduce non-response bias. Non-responders in random digit dialing may not have been contacted because the non-responders work multiple minimum-wage jobs. The investigators will place multiple calls and leave messages on answering machines. Random digit dialing is also subject to no-telephone bias - households without telephones are out of reach by random digit dialing, and the households without telephones probably differ in socioeconomic status than the overall population. However, for the urban, suburban, and semi-rural sub-populations within British Columbia, telephone coverage is almost universal. Inclusion of cell phones in random digit dialing is an attempt to mitigate such biases and increase the fraction of population covered by sampling. More importantly, the inclusion of cell phones is likely to result in inclusion of participants that might be absent from landline random digit dialing (younger and more educated groups) thus mitigating the sampling bias of random digit dialing. Cell phone random digit dialing is an active area of research and it is too soon to know with confidence what should and should not be regarded as a "Best Practice". For inclusion of cell phones in random digit dialing, the investigators will use the "Screening Approach". In this approach the interview is only conducted with participants sampled via the cell phone frame who do not have a landline, thus excluding numbers from the cell phone sample in the overlap (i.e., screening out the participants with both a cell phone and a landline). In this alternative sampling design, participants who have at least one household landline telephone and use at least one cell phone would be eligible for inclusion only from the landline frame. This decreases the bias due to higher probability of sampling from households who have both a landline and cell phone. The study will employ the Waksberg method of random digit sampling. In this method, all telephone numbers are initially split into blocks of equal size, called "primary sampling units". The primary sampling unit will be a mixture of landlines and cell phones. One randomly selected telephone number is called from each selected primary sampling unit. When a residential connection is reached in the first stage, the primary sampling unit qualifies for inclusion in the second stage. Respondents will be asked the following question: "Is there a member of the household between ages of 1 to 85 who has had asthma ever diagnosed by a physician". If the response is affirmative the study assistant will ask if the participant in question can continue the conversation directly (parents in case of children). Eligible participants (or in the case of children their parents or guardians) will be informed of the objectives of the study, verbally consented, and then screened for eligibility based on age and history of asthma. Participants who meet eligibility criteria and verbally consent over the phone will be asked to come to the study laboratory (one at Vancouver and one at Kelowna, corresponding to the two Census Sub-Divisions) and the participant will receive a detailed explanation of the study and the study consent form at that time. All participants who provide signed consent will then proceed to spirometry (simple spirometry without bronchodilator response) and study data collection. After the baseline visit, participants enter a 12-months period of follow-up with visits at months 3, 6, 9, and 12. Participants will be asked to attend one of the study centres and data on the quality of life and asthma-related resource utilization will be gathered. On the final visit, participants will undergo spirometry for objective diagnosis of asthma with possible methacholine challenge test for the participants whose asthma cannot reliability be included or excluded. Phase B: In this phase of the study, health records of participants who have participated in the phase A of the study for the entire follow-up period and the 12 months period prior to the participants' enrolment will be retrieved from the BC Linked Health Database. In addition, the investigators will identify cases of asthma in the entire British Columbia population based on the administrative data for the same period of time. The BC Linked Health Database is a longitudinal administrative health care database containing participant-specific, anonymized health data from 1985 onwards on British Columbia's 4 million residents with health insurance. Records will be matched based on participants' unique Personal Health number and hence the matching will be almost completely accurate. Data extracted from the BC Linked Health Database will include Medical Services Plan data, which encompass fee-for-service physicians, the Discharge Abstracts Database of hospital inpatient separation records, as well as records of death certificates. Prescription drug use will be determined from the BC Pharmanet database. This is a population-based prescription drug database that captures essentially all dispensing episodes by outpatients residing in the province on a prescription-by-prescription basis (regardless of funding source).. One request thought the British Columbia ministry of health (personal communication with Data Stewardship Secretariat) is required to retrieve data on the study participants as well as cases of asthma across the entire province, without additional cost for the latter group. The investigators will use this opportunity to retrieve data on cases of asthma in the entire province and compare different aspects of resource utilization between objectively verified cases of asthma and the data retrieved using a case definition algorithm on an administrative database. Such data will be used to retrospectively estimate the direct costs of asthma during the follow-up and in the 12 months prior to the enrolment in the study. The direct medical costs calculated in this way will be compared with the direct costs calculated in the prospective phase of the study. In addition, the investigators will examine the representativeness of the sample population by comparing the participant's health records with case of asthma in British Columbia identified through the BC Linked Health Database. Given the objective validation of the diagnosis of asthma at the end of the follow-up period in Phase A, the investigators will also be able to explore the sensitivity of existing case ascertainment algorithms based on administrative data at levels of severity and control. Phase C: Economic model of asthma: The data collected from participants during this study will be extrapolated to the population of British Columbia and Canada by appropriate adjustment for demographic and socioeconomic differences and reported prevalence of asthma in the country. The data will also be used to populate a Markov model of asthma to project the burden of asthma for the next 10 years. The Markov model of asthma will extend the model previously developed by Price et al. However, the investigators will model all 4 states of severity as defined by the Global Initiative in Asthma and whether at each year the participant's symptoms are controlled or uncontrolled. The Markov model will hence consist of 8 states (plus state of death) for all permutations of levels of severity and control. The model will also incorporate both the steady-state of chronic asthma and the acute state associated with an asthma exacerbation. Whenever possible, the investigators will estimate the parameters of the model from the cohort of participants in the Phase A and B of the study. This will include the initial distribution of individuals according to levels of severity and control, and the quality of life weights and annual direct and indirect costs estimated from the statistical analysis of the cohort data. The long term transition probabilities among levels of severity and control cannot reliably be estimated given the short follow-up of the participants in this study. The investigators will use values estimated from the literature for these and other components of the model.

Conditions

Interventions

Eligibility

Outcomes