Impact of Adding a Decision Aid to Patient Education in Adults with Asthma: A Randomized Clinical Trial

Background Not providing adequate patient education interventions to asthma patients remains a major care gap. To help asthma patients and caregivers discuss inhaled controller medication use, our team has previously developed a decision aid (DA). We sought to assess whether adding this DA to education interventions improved knowledge, decisional conflict, and asthma control among adults with asthma. Methods A parallel clinical trial (NCT02516449). We recruited adults with asthma, aged 18 to 65 years, prescribed inhaled controller medication to optimize asthma control. Educators randomly allocated participants either to the education + DA or to the education group. At baseline and two-month follow-up, we measured asthma knowledge (primary outcome) with a validated self-administered questionnaire (score –37 to +37). Secondary outcomes included decisional conflict and asthma control. Blinded assessors collected data. Between the two time points, the within- and between-group changes were estimated by generalized linear mixed models. Results Fifty-one participants (response rate: 53%; age: 44 ± 13 years; women: n = 32) were randomized either to the education + DA group (n = 26) or to the education group (n = 25), and included in statistical analyses. Between baseline and follow-up, mean [95% CI] knowledge scores increased from 21.5 [19.9–23.2] to 25.1 [23.1–27.0] in the education + DA group (P = 0.0002) and from 24.0 [22.3–25.7] to 26.0 [24.0–28.0] in the education group (P = 0.0298). In both of the groups, decisional conflict and asthma control improved. There were no differences between groups. Conclusions Education improved knowledge, decisional conflict, and asthma control whether the DA was added or not.

importance of physicians' adherence to asthma prescribing guidelines. Inhaled corticosteroids (ICS), as regular controller therapy, have been shown to effectively reduce asthma symptoms and improve, among other things, health-related quality of life (3). Long-acting beta 2 -agonists (LABAs), leukotriene receptor antagonists (LTRAs), and, in severe asthma, oral corticosteroids could be required as add-on therapy if asthma control is not achieved with the use of ICS alone (3). Treating acute asthma symptoms relies on reliever therapy, such as inhaled fast-acting beta 2 -agonists (FABAs). These, however, should be used at low-dose and low-frequency (3).
Several reasons may explain uncontrolled asthma. Physicians' non-adherence to asthma prescribing guidelines may be one of them. On the other hand, patients' lack of knowledge about the disease and its treatments may affect self-management (1). Moreover, inadequate environmental control at home, school, or work may lead to uncontrolled asthma, since atopy, tobacco smoke, and respiratory tract infections worsen symptoms (3). In addition, age is a predictor of the level of asthma control, since comorbidities increase with it (1). New evidences suggest that obesity may also be associated with uncontrolled asthma (1).
Other reasons may explain suboptimal asthma control. The level of asthma control may be overor underestimated if an objective measurement of the patients' airflow obstruction is not undertaken (11). Moreover, many patients with asthma misunderstand their condition or the side-effects of pharmacotherapy (4), which highlights the need for educational interventions (11). Furthermore, both the insufficient use of controller medication and the overuse of reliever therapy, which underscore non-adherence, lead to uncontrolled asthma [1].

Achieving adherence to pharmacotherapy by lessening decisional conflict with the use of shared decision making aids
The appropriate use of asthma drugs is a critical issue. In asthma, patients' knowledge of medication and perception of health and drug costs have been shown as predictors of the appropriate use of pharmacotherapy (19). Having consulted a specialist for asthma seems also to modulate adherence to treatments (19). Age has also been shown as a predictor of compliance with medication (20).
A recent Cochrane review has highlighted interventions that aimed to improve adherence to medication in asthma (21). However, no intervention included in that Cochrane review relied on shared decision making, even though the approach has been shown to improve knowledge (22), which is a predictor of the appropriate use of pharmacotherapy (19). Four characteristics define shared decision making in a context of treatment (23): 1) caregiver's and patient's involvement; 2) values and expectations, relevant and evidence-based information; 3) discussion; and 4) agreement.
Decision aids are frequently used in shared decision making. They are defined as "evidencebased tools to prepare people to participate in making specific and deliberated choices among healthcare options in ways they prefer. They supplement (not replace) clinician's counselling" (24). Decision aids are educational (23). They provide detailed and evidence-based information on the disease and its treatment options so that patients make informed decisions (24). They present risks, benefits and uncertainties of each treatment option in order to ensure that patients fully understand the ins and outs of the decision to be taken (24), which is easier when numbers, rather than words, are used to express probabilities (22). In addition, decision aids are designed to help patients clarify their values regarding the expected benefits of a treatment or its potentially undesirable consequences, so that patients' decisions are value-based (24). Consequently, decision aids lessen decisional conflict (22), which, otherwise, increases with the lack of support, resources, skills, self-confidence, and knowledge about treatment options, but also with social pressure, patients' unrealistic expectations, unclear values, and perception of others (25).
If shared decision making lowers decisional conflict (22), it also aims to optimize pharmacotherapy, bringing patients to adequately use their medication, so that they do not over-or underuse their drugs. In order to evaluate the effect of shared decision making on adherence to medication and health outcomes in asthma, Wilson and colleagues have enrolled 612 patients with poorly controlled asthma in a randomized controlled trial (26). Their study aimed to compare a shared decision making group to a clinician decision making group. After a 1-year follow-up, shared decision making has been shown to improve adherence to controller medication, lower the use of rescue medication, enhance quality of life and likelihood of asthma control, and reduce the number of medical visits. To our knowledge, this is the only study that has assessed shared decision making in asthma.
Even though shared decision making has been shown to lessen decisional conflict among patients who otherwise felt uninformed or unclear about their values (22), the question remains whether it lowers decisional conflict among patients with asthma. Moreover, the intervention assessed by Wilson and colleagues did not rely on shared decision making aids, but rather on the negotiation of decisions. In addition, although over 500 patient decision aids have already been developed worldwide (27), to our knowledge, none of those has been intended yet for asthma management.
The Ottawa Decision Support Framework (ODSF) is intended for the development of patient decision aids (24). It uses a three-step decision process: 1) identifying decision support needs, 2) developing, and 3) assessing decision aids. The framework has been developed according to psychosocial theories. It is based on the assumption that satisfied decisional needs lead to a better decision quality, which, in turn, enhances health behaviours and outcomes. According to the ODSF, improving decision quality relies on the use of decision aids.
Des Cormiers and colleagues have taken the first steps to develop shared decision making aids in asthma, based on the ODSF. Indeed, they carried out a decisional conflict evaluation among 50 patients with asthma and a needs assessment among 15 others (Des Cormiers A, Legare F, Boulet LP. Assessment of Decisional Conflict in Asthma Management. In preparation. & Des Cormiers A, Legare F, Boulet LP. Needs Assessment in Asthma Management. In preparation.) Those 15 patients were recruited at the ICUPQ Outpatient Asthma Clinic or at the IUCPQ Research Center, and were asked to fill a French adapted version of the Population Needs Assessment Questionnaire (28). As a result of this cross-sectional study, Des Cormiers and colleagues identified which particular decisions regarding asthma management are problematic and for which it would be worth developing shared decision making aids. Consequently, since we believe that the acceptance of medication leads to the appropriate use of pharmacotherapy, and since the latter is known to improve asthma control, we aim, in the present study, to develop and then assess shared decision making aids in asthma.

Objectives
Based on a linear knowledge translation process, which makes the assumption that the appropriate use of shared decision making aids leads to medication acceptance and adherence, which, in turn, enhance asthma control, we aim to achieve the following objectives.

Primary objectives
 To develop shared decision making aids in asthma.  Among adults with mild to severe asthma, to assess the effect of shared decision making aids on: o Asthma knowledge; o Patients' decisional conflict.

Secondary objectives
 Among adults with mild to severe asthma, to assess the effect of shared decision making aids on: o Compliance with recommended use of pharmacotherapy; o Asthma control.

Hypothesis
We hypothesize that our shared decision making aids will enhance knowledge and lessen decisional conflict. In addition, we believe that they may also improve compliance with recommended use of pharmacotherapy, by increasing maintenance drug use. Consequently, we consider that our shared decision making aids may lead to a better asthma control.

Methods
The aids will be developed, and then assessed.

Developing shared decision making aids
The shared decision making aids will be developed according to the International Patient Decision Aid Standards (IPDAS), a set of criteria that are internationally approved, and on which decision aids can be quality rated (27). The first aid to be developed will be about medication intake.

Developing prototypes of shared decision making aids
The aids will be designed in order to present a brief description of asthma and a summary of the current guidelines regarding asthma management. In addition, they will provide evidence-based information, such as probabilities of risks, benefits, and uncertainties associated with each treatment option, whenever possible (24). The aids will be designed in order to help patients clarify and express what matters the most to them, and to guide and facilitate discussion and communication of facts and values between patients and caregivers (24;27).

Gathering feedback on the aids and improving the aids
Once the aids will be designed, 2 health professionals (general practitioner, respirologist, or asthma educator) and 5 patients with asthma will be asked to provide feedback on the prototypes of the shared decision making aids. Patients will be recruited from the IUCPQ Outpatient Asthma Clinic by the asthma education nurse or from the IUCPQ Research Center by research nurses. Patients with a current diagnosis of asthma in their medical report will be eligible (see Inclusion criteria below). Relevant information on the study objectives will be given to patients interested to participate in the study by the study coordinator. Enrolled patients will sign an informed consent before participating in the study. The study will have been approved by the institutional ethics committee. According to the IPDAS criteria, none of the enrolled patients or health professionals reviewing the aids can further be involved in their field testing (27).
Enrolled patients and health professionals will have to fill an adapted version of the OHRI Acceptability Questionnaire (29). The questionnaire will have been given to them by the study coordinator. Feedback on the content, complexity, structure, schematisation, and length of the information provided in the aids will be sought, and suitability for decision making will be assessed. Once feedback will be gathered, the aids will be refined in order to meet end-users' needs. Afterward, 3 other health professionals and 5 other patients will fill the OHRI Acceptability Questionnaire to provide more feedback on the modified prototypes of the shared decision making aids. Once again, the aids will be refined according to their comments. Further feedback will be gathered if needed, until saturation is reached. Then, the aids will be evaluated.

Assessing shared decision making aids
A pilot study designed as a prospective randomized controlled trial will compare two groups of adults with mild to severe asthma, one in which decision aids are used (experimental group) and another in which the aids will not be used (control group). The effect of the aids on asthma knowledge, decisional conflict, compliance with recommended use of pharmacotherapy, and asthma control will be evaluated, because we make the assumption that the acceptance of medication may lead to the appropriate use of pharmacotherapy, which may, in turn, improve asthma control.

Evaluation of the aids
In order to recruit patients with mild to severe asthma in the study, medical reports of patients attending the IUCPQ Outpatient Asthma Clinic or having participated in studies at the Research Center will be screened by the study coordinator. Patients who will meet the eligibility criteria (see Inclusion criteria below) will be contacted by the asthma nurse at the Outpatient Asthma Clinic prior to their meeting with her, or by the study coordinator if patients are recruited at the research center. Relevant information on the study objectives will be given by the study coordinator to eligible patients who will be interested to participate in the study. Those who will agree to participate in the study will sign an informed consent. The study will have been approved by the institutional ethics committee. Patients who will have given feedback on the aids in step 2 will not be enrolled, according to the IPDAS criteria.
The baseline visit (visit 1A) will take place at the IUCPQ Research Center and Outpatient Asthma Clinic. With the help of the study coordinator/asthma nurse, asthma control will be assessed, using the Asthma Control Scoring System (30). The study coordinator will evaluate patients' compliance with pharmacotherapy, using the Compliance Interview Questionnaire. In addition, the Decisional Conflict Scale (18) and the Asthma Knowledge Questionnaire (31) will be filled by patients. In order to control for covariates, if relevant, data on enrolled patients' health (age, height, weight, atopy, respiratory tract infections) and environmental factors will be gathered by the study coordinator, using the IUCPQ Asthma Consultation Form. For the same reason, enrolled patients' data on asthma knowledge at baseline will also be used. Visit 1A should last about 45 minutes.
After having filled the questionnaires, the study coordinator will give patients sealed envelopes in order to randomize them in the intervention or control group (see Randomization (sequence concealment) below). Consequently, patients will get, or not, the shared decision making aids, depending on the group they belong to (visit 1B). Those who will belong to the experimental group will have to pass through the steps of the aids before meeting the asthma nurse. Those belonging to the control group will skip this stage and therefore will meet the nurse directly, as in usual care. In both groups, there will be one meeting with the nurse per patient.
After a 2-month follow-up, patients will come to the IUCPQ Research Center for a 50-minute appointment (visit 2). With the help of the study coordinator, they will be asked to fill the same questionnaires as in the baseline visit.

Target population
Men and women, between 18 and 65 years, with mild to severe asthma will be recruited. They must be able to read in order to adequately use the shared decision making aids, and understand the ins and outs of their treatment options.

Selection of subjects
Patients will be recruited among those attending the IUCPQ Outpatient Asthma Clinic or Research Center. Such convenience sampling will facilitate the enrolment of patients with mild to severe asthma. The inclusion criteria will be assessed through the patients' medical reports.

Inclusion criteria
 Eligible patients, aged 18 to 65 years, with a current diagnosis of mild to severe asthma will be recruited.
o Asthma diagnosis will be made upon one of the following criteria (3):  FEV 1 increases by at least 12% after the use of a bronchodilator (and a minimum ≥200 mL);  Current asthma symptoms and a positive methacholine challenge test (20% fall in FEV 1 after inhalation of a provocative dose of methacholine 16 mg/mL);  A respirologist's current diagnosis of asthma found in the patient's medical report.
 Severity of asthma will be assessed according to prescribed pharmacotherapy, as suggested by current guidelines (1;32): o Patients with mild asthma using SABA as well as low doses (≤250 mcg/day beclomethasone or equivalent) of inhaled glucocorticosteroids; o Patients with moderate asthma using SABA as well as low to moderate doses (250mcg/day< dose ≤500 mcg/day beclomethasone or equivalent) of inhaled glucocorticosteroids, with or without additional therapy (LABA or LTRA); o Patients with severe asthma using high doses (>500 mcg/day beclomethasone or equivalent) of inhaled glucocorticosteroids, and additional pharmacotherapy (LABA, LTRA, or Prednisone), and SABA.

Exclusion criteria
 Patients having taken part in an asthma educational program at the IUCPQ Outpatient Asthma Clinic in the last 6 months;  Patients aged 40 years or older with prebronchodilator FEV 1 <80% of predicted value (in order to ensure that patients with chronic obstructive pulmonary disease are excluded from the study);  Smoking >10 pack-years.

Sample size
Since the study is a pilot, 50 patients are intended to be recruited to participate in the randomized control trial. Sample size calculations will be made during the study, based on gathered data. If feasible, more than 50 patients may be enrolled in order to get sufficient statistical power.

Randomization (sequence generation)
Using statistical software, a statistician will generate a blocked randomization list that will be used to allocate patients to either the intervention or the control group. Blocks of four will be used.

Randomization (sequence concealment)
Following the randomization list, 50 numbered opaque sealed and equally weighted envelopes will be given to the study coordinator. Each envelope will contain either the decision making aids (experimental group) or blank pages (control group). Allocation concealment will be ensured since the study coordinator will have to give sequentially the sealed envelope to the enrolled patient who, once alone, will have to open it. Therefore, the study coordinator will not know in which group patients belong to.

Blinding
The person administering the questionnaires (study coordinator) will be blinded to the group assignment as will be the persons analysing the data (study coordinator and statistician).

Study intervention
In this study, the intervention is the use of shared decision making aids, which will be compared to usual care (not using the aids).

Shared decision making aids
The shared decision making aids that will be used in the intervention group will be the latest version of those developed in the previous stage of this study (developing shared decision making aids). Patients in the experimental group will receive the aids and will be asked to go through them prior to their meeting with the asthma education nurse at the IUCPQ Outpatient Asthma Clinic. During this meeting, the asthma nurse will review patient's written action plan. Individualized education and counselling will also be provided. The length of the meeting will be evaluated.

Usual care
In the control group, decision aids will not be used. As in usual care, patients meet directly the asthma nurse, who will review their written action plan, and provide individualized education and counselling. The length of the meeting will be evaluated.

Operational Definitions of Dependent Variables & Measurements
In this study, the dependent variables, or health outcomes, are asthma knowledge, decisional conflict, compliance with recommended use of pharmacotherapy, and asthma control. Covariates are asthma knowledge, age, body mass index, atopy, environmental triggers, and respiratory tract infections.

Asthma knowledge questionnaire (primary outcome)
Knowledge of asthma will be assessed by the Asthma Knowledge Questionnaire (31). The questionnaire will be filled by patients at baseline and after a 2-month follow-up. The French questionnaire comprises 37 items (true, false, don't know answers), evaluating 4 domains of asthma knowledge: biomedical, asthma severity, general knowledge and treatment (31). Developed in France, the original questionnaire had 38 items, but the eleventh has been withdrawn because it was not relevant in Quebec. The total knowledge score equals the good answers (rated 1) minus the wrong answers (rated -1). Don't know answers are rated 0. For each item, a negative score is indicative of asthma misunderstanding, a null score of a knowledge gap, and a positive score of gained asthma knowledge. The total knowledge score ranges from -37 to 37; thus, asthma knowledge will be modeled as a continuous variable. Psychometric properties of the questionnaire have been tested and showed good reliability and reproducibility (31).

Decisional conflict (primary outcome)
Decisional conflict will be assessed through a French version of OHRI Decisional Conflict Scale (18). The questionnaire will be filled at baseline and after a 2-month follow-up. Sixteen items evaluate 5 dimensions of decisional conflict (informed, values clarity, support, uncertainty, effective decision) using a Likert scale. Patients' responses range from 0 (strongly agree) to 4 (strongly disagree). Patients' responses are summed, divided by 16, and multiplied by 25. Consequently, the decisional conflict score ranges from 0 (low decisional conflict) to 100 (high decisional conflict). A score greater than 37.5 suggests that there is a meaningful decisional conflict, but a score that is less than 25 is not associated with it (33). The variable will nevertheless be analysed as a continuous one. The reliability of the OHRI Decisional Conflict Scale (English low literacy version) has been evaluated using a Cronbach's α, which value was above the 0.7 threshold (34). Test-retest coefficient also supports its reliability (35). The scale has been shown responsive to change, because it is able to detect an effect size of 0.2 to 0.3 in studies that compare an intervention group to a control group (33).

Compliance with treatments (secondary outcome)
Compliance with pharmacotherapy in the last week will be evaluated. An interview questionnaire based on the 2003 Asthma Consensus Conference guidelines (36) and adapted to the 2012 Asthma Consensus Conference guidelines (3) will be administered at baseline and after a 2-month follow-up. It has already been used to describe the appropriate use of asthma drugs among patients with moderate to severe asthma (19). The questionnaire is composed of a set of 11 hierarchical criteria; there were originally 12, but one of them is not relevant anymore. The set of criteria evaluates, in the last week, the use and the frequency of use of controller/rescue medication, and add-on therapy (if relevant). In order to be classified as compliant, patients need to meet all the 11 criteria (19), which set patients who are very compliant apart from others (37).
The questionnaire also allows compliance to maintenance/rescue/add-on medication to be expressed as a percent score ([doses taken/doses prescribed] × 100). In that case, a score that is ≥80% but ≤100% is indicative of compliance (38). The variable will nevertheless be analysed as a continuous one.

Asthma Control (Secondary Outcome)
Asthma control will be quantified using the Asthma Control Scoring System (ACSS) (39), which is based on Canadian Asthma Consensus Guidelines. The ACSS will be administered at baseline and after a 2-month follow-up. Two types of parameters are assessed by the ACSS: the frequency of symptoms in the last week (clinical parameter) and the percentage of predicted value of FEV 1 (physiological parameter). FEV 1 will be measured according to the ATS criteria (40) and predicted values will be obtained from IUCPQ. Both clinical and physiological parameters will be expressed as a percent score and the mean of those scores will give the global asthma control score. If a score ≥80% is indicative of asthma control, the variable will however be analysed as a continuous one. Discriminative properties of the ACSS have been evaluated (30). The reliability of the instrument has been assessed through internal consistency testing and test-retest, and has been shown to be acceptable. Moreover, correlation with other instruments evaluated construct validity, which has been shown to be good. Furthermore, evaluative properties of the ACSS were assessed: responsiveness indexes supported its sensitivity, which means that the instrument detects changes over time and between groups.

Asthma knowledge (covariate)
Since the knowledge of asthma is also a predictor of decisional conflict and compliance with treatments, the values of the Asthma Knowledge Questionnaire (31), which will have been filled by patients at baseline, will be used. As a covariate, asthma knowledge will be modeled as a continuous variable.

Predictors of Asthma Control (Covariates)
Age (in years) and body mass index for adults (weight (kg) / [height (m)] 2 or weight (lb) / [height (in)] 2 x 703) will be assessed. In addition, inflammatory factors associated with asthma control will be measured. These factors are atopy (self-reported allergy; self-reported exposure to allergen), environmental triggers (smoking or exposure to second hand smoke), and respiratory tract infections in the last month. All those will be assessed through the IUCPQ Asthma Consultation Form. Covariate measurements will be undertaken at baseline. Questions regarding respiratory tract infections in the last month will also be asked after a 2-month followup. Factors associated with asthma control will be modeled as continuous or categorical variables.

Socio-demographic variables
Socio-demographic variables (gender, level of education -using the categories of the Institut de la statistique du Québec) and the year of asthma diagnosis will be measured in order to better describe the study population.

Data analysis
Descriptive analyses will be performed in order to describe the study population: continuous variables will be described as means or medians, depending on normality of distributions, and categorical variables as proportions.
Generalised linear models will be used to compare the intervention and control groups on the dependent variables (41). For asthma knowledge, decisional conflict, compliance with maintenance/rescue/add-on pharmacotherapy, and asthma control, a normal distribution and an identity link will be used, since all 4 variables are expressed as continuous ones. Assumptions of normality, and homoscedasticity will also be checked. A log transformation could be used to correct departure from normality, if relevant. For compliance with pharmacotherapy, a binomial distribution and a log link will be used as this variable is expressed as a binary one. In models, effects for group (intervention and control), time (baseline and 2-month follow-up) and groupby-time interaction will be included. The effect of the shared decision making aids on health outcomes will then be estimated by the interaction term between the group and time. Mean differences (continuous variables) and prevalence ratios (dichotomous variable) have been chosen as effect measures. In addition, generalized estimating equations will be used in order to take repeated measures into account.
If covariates (asthma knowledge, age, body mass index, atopy, environmental triggers, and respiratory tract infections) are not distributed evenly between groups, we will adjust for them in analyses. Predictors of health outcomes will be incorporated into the model as confounders if they result in a >10% change in the effect measure, following a descending procedure. Exploratory subgroup analyses according to asthma severity will be conducted since pharmacotherapy and compliance with treatments are getting more complex with the severity of the disease. Statistical interaction of body mass index will also be exploratory assessed since this covariate is hypothesized to modify the effectiveness of intervention aiming at improving asthma control (42). In addition, exploratory statistical interaction of age, education and length of the patient's meeting with the asthma nurse will be evaluated since we hypothesize that the effect measure may vary according to the level of measurement of these variables. Regarding adherence to medication, sensitivity analyses will be carried out according to different cut-offs, that will range from 60 to 80% of inhaled prescribed doses.
SAS version 9.3 (SAS Institute Inc., Cary, USA) will be used to perform the statistical analyses, using PROC GENMOD. A 2-tailed p-value less than 0.05 will be considered indicative of statistical significance. Since the study is a pilot, power calculation will be made a posteriori. Alternatively, the study size is based on the number of patients that may realistically be recruited at the IUCPQ Research Center or Outpatient Asthma Clinic according to the timeframe. Nevertheless, sample size calculations will be made during the study, based on gathered data. If feasible, more patients may be enrolled in order to get sufficient statistical power.

Timeframe
Passing through the stages of this study will last 1 year. The intervention itself will be 2 months long, including baseline visit, follow-up, and outcome measurements.

Utility Perspectives
The appropriate use of SDM aids may enhance knowledge of asthma and lessen decisional conflict, which, in turn, may improve compliance with recommended use of pharmacotherapy and asthma control. Moreover, this study may provide patients and health professionals with a new and efficient approach to make informed, evidence-based and value-based decisions in the context of asthma management.
The study conclusions will be the subject of a Master's thesis. They are also intended to be published in 2 peer-review articles. Moreover, gathered data on patients' compliance with pharmacotherapy are expected to be reanalysed in order to compare compliance to prescribed pharmacotherapy and compliance to self-reported medication. Patients and health professionals' perception of the usability of the aids in an Outpatient Asthma Clinic is also expected to be further assessed in focus groups. In addition, patients' data on asthma knowledge may be used in order to validate a French Canadian Asthma Knowledge Questionnaire. Patients' data on decisional conflict and adherence may be used to further evaluate the concordance between patients' preferences and implemented options, according to the IPDAS criteria.

Bias
Convenience sampling and loss to follow-up may induce selection bias in this pilot randomised controlled trial. In order to avoid loss to follow-up, visit 2 will be scheduled at the end of visit 1. In addition, we will phone enrolled patients in the week preceding visit 2 in order to ensure that they remember their appointment at the IUCPQ Research Center.
Self-reported questionnaires are at risk of nondifferential misclassification that may weaken the effect of the aids on health issues. Although validated questionnaires are used in this study, they remain mostly subjective, and may induce information bias.
Furthermore, adherence to treatments is a complex phenomenon which has been linked to 200 different covariates (43). In addition, social pressure and lack of resources/ self-confidence are known to worsen decisional conflict (25) but will not be assessed in this study. Asthma control is also a complex phenomenon for which every single predictor will not have been assessed. Thus, residual confounding effect is possible.
Recent knowledge translation theories suggest that the knowledge translation process is more multidirectional than linear (44). Since our knowledge translation model is linear, it could not be excluded that external factors, such as health literacy and numeracy skills, may play a role in the success or failure of our intervention.

Generalizability
Assuming that internal validity is not threatened, the results of this study should be generalized to patients with mild to severe asthma, to those who are followed by a physician and by an asthma educator in a tertiary care center in Quebec City, and to those who are motivated by using shared decision making aids, and willing to do so.

Research Utility
If the effect of the aids on health outcomes is shown to be significantly positive, the aids may eventually be implemented in the Quebec health system, particularly through the Quebec Asthma and Chronic Obstructive Pulmonary Disease Network, and thereby may become key elements in the management of asthma.

Level of education
University degree X X Proportion Years with asthma X X Mean (standard d