A Mixed-methods Study to Assess Interrater Reliability and Nurse Perception of the Braden Scale in a Tertiary Acute Care Setting

Research regarding the reliability of the Braden Scale and nurses’ perspectives on the instrument for predicting pressure ulcer (PU) risk in acute care settings is limited. A mixed-methods study was conducted in a tertiary acute care facility to examine interrater reliability (IRR) of the Braden Scale and its subscales, and a qualitative survey using semi-structured interviews was conducted among nurses caring for patients in acute care units to gain nurse perspective regarding scale usability.Data were extracted from a previous retrospective, randomized, controlled trial involving adult patients with compromised mobility receiving care in a tertiary acute care hospital in Canada. One-way, intraclass correlation coefficients (ICCs) were calculated on item and total scores, and kappa statistics were used to determine reliability of categorizing patients on their risk. Interview results were categorized by common themes. Reliability was assessed on 64 patients, where nurses and research staff independently assessed enrolled participants at baseline and after 72 hours using the Braden Scale as it appeared on an electronic medical record. IRR for the total score was high (ICC = 0.807). The friction and shear item had the lowest reliability (ICC = 0.266). Reliability of categorizing patients’ level of risk had moderate agreement (κ = 0.408). Three (3) major and 12 subthemes emerged from the 14 nurse interviews; nurses were aware of the scale’s purpose but were uncertain of its effectiveness, some items were difficult to rate, and questions were raised as to whether using the scale enhanced patient care. Aspects identified by nurses to enhance usability included: 1) changes to the electronic version (incorporating the scale into daily assessment documents with readily available item descriptions), 2) additional training, and 3) easily available resource material to improve reliability and usability of scale. These findings need to be considered when using the Braden Scale in clinical practice. Further study of the value of the total Braden Scale and its subscales is warranted.

Pressure ulcers (PUs) are a common health problem that has a significant impact on morbidity, mortality, and health care utilization and cost.^1-4 As shown in cross-sectional surveys and reliability studies,⁵ factors such as limited mobility, malnutrition, increasing age, and excessive moisture may contribute to PU development. A survey⁴ and a reliability study⁶ suggested that identifying patients at increased risk for PUs can lead to targeted interventions to prevent their occurrence; a systematic review⁷ has shown prevention rather than treatment offers greater hope of a cost-effective way to deal with PUs. 

The Braden Scale is a widely used instrument for predicting PU risk.^8,9 It consists of 6 items (sensory perception, moisture, activity, mobility, nutrition, and friction and shear) that evaluate factors contributing to the development of PUs.^5,8 Descriptions are provided for each item’s subscore category; items are ranked on a scale of 1 to 4 (from greatest to lowest risk), with the exception of the friction and shear item where the scale is 1 to 3.⁸ The individual subscores then are summed to give a total score that corresponds to the patient’s level of risk for PUs. Although the original Braden Scale has defined risk ranges for no risk (19–23), mild (15–18), moderate (13–14), high (10–12), and very high risk (<9), a variety of different categories and score ranges have been used and tested in clinical practice to define risk.^10-17 Risk ranges may differ in clinical practice depending on the facility using the Braden scale. A retrospective quantitative analysis¹⁰ was conducted of Braden scores of patients (N = 2625) admitted in an acute care facility to compare a modified 3-group risk classification with the traditional 5-group classification. The investigators categorized hospitalized patients as being at mild (total Braden score ≥17), moderate (12–16), and high risk (≤11) for developing a PU and found no significant differences between the 3-group and 5-group risk classifications.

As noted by Kottner and Dassen¹⁸ in their literature review, several studies have reported on the validity of the Braden Scale; however, research regarding the interrater reliability (IRR) in terms of intraclass correlation coefficients (ICC) is limited and only a few studies evaluated the reliability of item subscores. This review examined the methods (using Pearson’s product-moment correlation, Cohen’s kappa, overall percentage of agreement) previously used to analyze the IRR of items and total Braden scores and found inconsistencies between the values calculated. Kottner and Dassen recommended using ICC in future studies on IRR of Braden scores. Satisfactory IRR, defined as the concordance of ratings between multiple raters for the same patient, is essential for ensuring the reproducibility of ratings.^18,19 Although the IRR of the Braden Scale has been examined in IRR studies in long-term care⁵ (N = 152), home care²⁰ (N = 691), and intensive care unit²¹ (N = 45) settings, its reliability on general medical and surgical units has not been evaluated. This is an area needing further study, because PUs are common among hospitalized patients.²¹ An observational study²¹ noted IRR research is especially lacking in assessing the scoring of individual items. 

Cross-sectional survey studies in Sweden (N = 85),²² Belgium (N = 553),²³ and Ethiopia (N = 248)²⁴ assessed nurses’ knowledge on PU prevention strategies and found general knowledge was inadequate. A cross-sectional (N =553) study by Beeckman et al²³ suggested nurse attitude toward PU prevention influences the implementation of preventive measures. The situation in Canadian hospitals is unknown because research about nurse perspective and particularly their attitude toward the Braden scale is lacking. A better understanding of nurse perspective regarding the Braden Scale could lead to greater use of the instrument in the implementation of PU prevention strategies. 

The overall aims of this study were to examine the IRR of the item and total scores of the Braden scale in a real-life clinical setting and to determine the perspectives of nurses toward the Braden scale and its use in an acute care setting. 

IRR.
Study setting and population. This report uses data from a substudy of a randomized, controlled trial (RCT) (see NCT02325388 at ClinicalTrials.gov for more details) on the efficacy of continuous pressure imaging technology at reducing interface pressure of patients who are at risk of PUs in an acute hospital setting. The RCT was conducted in a tertiary care teaching hospital in Calgary, Alberta, Canada. Participants received care from internal medicine, neurology, spinal injury, or intensive care units and were recruited based on the parent RCT²⁵: participants were either gender, at least 18 years of age, and required assistance or were completely dependent for bed mobility on the “Bed” components in the de Morton Mobility Index, a 15 item unidimensional measure of mobility.²⁶ Participants who voluntarily withdrew from the study, were transferred, were discharged or died within 72 hours of enrollment, or lacked Braden scoring during the study period were excluded from this report. The Conjoint Health Research Ethics Board at the University of Calgary approved the study (REB13-0794). 

The substudy served as an independent examination of the use of the Braden Scale in clinical practice. Braden scores for each enrolled participant were completed for the RCT by 1 of 2 research staff members (one of which was a research nurse) at baseline and after 72 hours. Other members of the research team (N = 13), which consisted of wound care clinicians and researchers, trained the research staff in the administration of the Braden Scale. Unit nurses who worked on internal medicine, neurology, spinal injury, and intensive care units during the study and used the Braden Scale regularly for patient assessments independently determined Braden scores per institutional policy requirement to assess risk for all patients on the units. Because the Braden Scale was used on a regular basis by nurses as a required assessment, and because the study goal was to examine its reliability in a real-life clinical setting, additional training on its administration was not provided by the facility’s research team before the start of this substudy. Unit nurses were trained by the unit nurse educators during their new staff orientation, which was separate from, and not associated with, the RCT. 

The Braden Scale used by nursing staff to chart their assessments was part of the institutional electronic medical record (EMR) system used at this facility. This version included the 6 Braden Scale items along with the rating options identified by number and heading for scores; it did not include full descriptions for each item or scoring options. A printed version of the scale included the 6 items and rating options and full descriptions of each. Both the EMR and paper resources of the Braden Scale had been instituted on the units well before the start of the study and were not initiated by the research team. Because the authors’ main objective was to examine scale usability in a real-life clinical setting as it was being used before the RCT, any differences between the EMR and the paper resources of the scale were not reconciled before the start of the substudy. 

Data collection, sample size determination, and statistical analysis. Data were retrospectively extracted from the EMR of participants enrolled between December 2014 and May 2015. The data collected included participant age, gender, and Charlson comorbidity index score^.27 The Charlson comorbidity index is a “weighted index that takes into account the number and the seriousness of comorbid disease (that is used to) predict risk of death from comorbid disease.”²⁷ A high index predicts an increased cumulative mortality associated to comorbid disease. Braden scores (item and overall) recorded by the research staff and unit nurses on the same participants were collected. The Braden ratings recorded by unit nurses closest to these baseline and 72-hour assessments were used for comparisons in this substudy. All pairs of comparisons were completed for scores done within 1 to 2 days of each other. This led to a total of 2 comparisons (baseline and 72-hour assessment) for each participant. If a comparison was not completed within this time frame, it was not included in the analysis. Braden scores recorded by the research staff were stored on a password-protected electronic database (REDCap, Research Electronic Data Capture) that was separate from the EMR the nurses used for charting. The presence of PUs also was assessed clinically by the research team during a head-to-toe skin assessment and stored on the REDCap electronic database. 

The study population consisted of 93 consecutive participants, and a total of 103 comparisons were made. ICC (1, k) (one-way random model, average measures) and kappa statistics were calculated using IBM SPSS software, version 22 (IBM, Amonk, NY) to determine the IRR of the Braden ratings. Because the Braden Scale is a unique instrument with continuous numerical scores that also can be categorized into risk categories, both calculations provide insight on the reliability of the scale. The ICC is best calculated by treating the scores as continuous variables, while kappa estimates treat the scores as categorical variables.^28,29 ICC was used to determine the IRR for the item and total scores, where each comparison was analyzed as a single independent entry for the given assessment time point (baseline or 72 hours). Kappa statistics were used to determine IRR for categorizing patients being at risk for a PU. Individual total Braden scores assessed at baseline or 72 hours were first categorized using risk ranges before being analyzed using kappa statistics. The 3 risk categories used for kappa testing were mild (total Braden score ≥17), moderate (12–16), and high risk (≤11), which was reflective of the risk ranges used at this facility.¹⁰ This report focuses on ICC calculations because they account for the degree of disagreement for IRR estimates and provide a better representation of the overall IRR for the Braden Scale.^18,29 ICC compares the variability of different ratings of the same patient to the total variation of all ratings of all patients.¹⁸ However, the kappa values from this study could lay the groundwork for future comparisons.¹⁸ ICC and kappa values were interpreted separately and not compared to each other because the calculation methods differed.^18,29 

Nurse perspectives on the Braden Scale. 
Participants. Volunteer male and female registered nurses (RNs) who worked on the internal medicine, spinal injury, and intensive care units during the study and used the Braden Scale participated in the interviews. 

Interviews. Participants were interviewed one-on-one in a quiet conference room on their patient care unit. These interviews were conducted by a research assistant who was not involved in the quantitative data collection in June 2015. The identity of participants was protected by an impartial interviewer interested in their opinions and experiences. The participants had no previous contact with the interviewer before the interviews. All participants gave written informed consent and were offered the opportunity to contact the interviewer after the interviews to add to their answers or to ask questions regarding the study. The interviews followed a semi-structured format that explored the participants’ understanding and knowledge of the Braden Scale and their views on the significance, effectiveness, and usability of the tool guide (see Table 1). The interviews lasted 15 to 20 minutes and were held during the nurses’ break times. Interviews were audiorecorded with field notes made after each interview. The interviewer transcribed the recordings verbatim. Participants were recruited and interviewed until the interviewer was satisfied each unit was well represented and that data saturation was reached. Field notes and transcripts were referred to and used in the analysis to assist in the interpretation of the data. 

Qualitative data analysis. Qualitative data obtained from interview field notes and transcripts were analyzed thematically. Each transcript was analyzed line-by-line for meaningful themes common among all interviews. The interviewer independently coded 14 interview transcripts using NVivo Software Version 10, 2012 (QSL International Pty Ltd, Burlington, MA) to generate the initial set of themes and subthemes, which were later discussed with another research team member until consensus was reached to form the final set of themes. Participants did not contribute to data analysis and interpretation. 

Quantitative results. 
Sample. The initial sample size of 93 participants was reduced to 64 following 29 exclusions due to voluntarily withdrawal from the study (1), unit transfers (3), death within 72 hours of enrollment (5), or not having Braden score determinations done during the enrollment period (20) (see Figure 1). A total of 103 Braden score comparisons with either 1 or 2 observations were made per participant. Of the sample, 30 participants (47%) had a preexisting PU at the start of the study; the majority were Stage 1 (11 skin sites) and Stage 2 (26 skin sites) PUs. Mean participant age was 67.68 years (SD = 17.45) and median age 71 years (range 24–92). Men were slightly more predominant (38; 59.4%). Mean Charlson comorbidity index score was 3.27 (SD = 2.85). The most prevalent admitting medical diagnoses observed in this population were cerebrovascular diseases (22, 34.4%), hemiplegia or paraplegia (20, 31.3%), and/or diabetes mellitus with or without complications (14, 21.9%). 

The mean total Braden score recorded was 14.03  ± 2.54. The research staff recorded an average total score of 14.25 ± 2.38, and the unit nurses had a mean total score of 13.80 ± 2.68. Of the 103 comparisons made, research staff and unit nurses recorded the same total score for 70 assessments. The unit nurses recorded a higher risk score compared to the research staff for 20 assessments. For 13 assessments, the research staff recorded a higher risk score compared to the unit nurses. 

IRR. IRR of item and total scores for individual units and overall are shown in Table 2. Overall, ICC scores showed excellent agreement for total (0.807) and nutrition (0.722) scores; good agreement for sensory perception (0.726), moisture (0.659), activity (0.595) and mobility (0.691); and poor agreement for friction and shear (0.266). Comparisons across individual units showed a good deal of variability on the ICC values. However, good to excellent agreement was achieved on the overall score for all units. The ICC for Braden scores of participants with no PUs at the start of the study showed excellent agreement (0.871), but scores for persons with PUs at entry demonstrated a fair agreement (0.533). Kappa values for categorizing the level of risk of PU development showed moderate agreement overall (κ = 0.408). 

Qualitative results. Table 3A and 3B summarizes the common themes and subthemes identified in the analysis of 14 nursing staff interviews. Thematic analysis revealed concerns on the scale utility and administration, barriers and challenges faced by nurses, and the benefits of using the assessment tool. The friction and shear item was identified by nurses as a difficult item to rate due to its subjectivity. Nurses suggested changes to the scale presentation on the EMR and recommended additional training sessions on the interpretation of scale items and scores.

In this study, the IRR of the Braden Scale was assessed in an acute medical care setting where scores recorded by research staff and the unit nurses were compared. To the authors’ knowledge, this study is one of the first to examine the IRR of item and total scores for electronic versions of the Braden Scale. The clinical nurses in this study were not provided additional training on the Braden Scale in order to study the IRR in a real-life setting. Semi-structured interviews were conducted with unit nurses to understand their perspectives on the utility and usability of the scale. The ICC of the total score for all units (0.807) indicated a high level of agreement on the scores recorded by research staff and the unit nurses for the same patient. This finding was comparable to previous studies evaluating ICC for the total Braden score in an ICU setting.^9,21 The kappa values for categorizing the patients’ level of risk were moderate in agreement overall (κ = 0.408). Although the degrees of agreement suggested by the ICC and kappa values differ, it is important to note the methods of calculation are different and the values should not be compared to each other.^18,29 Kappa statistics can be interpreted as the degree of observed agreement that exceeds the expected agreement by chance alone.¹⁸ The researchers chose to calculate the kappa for the 3 Braden Scale risk categories to provide some insight for clinicians on the IRR of classifying patients as mild, moderate, or high risk of PU development because this was the intended use of the instrument in practice. However, kappa only considers the categories of exact agreement and is subject to less flexibility compared to ICC. 

The ICC compares the variability of different ratings of the same patient to the total variation of all ratings, which provides more flexibility in calculations.^18,29 Additionally, Kottner and Dassen¹⁸ recommended the use of ICC for calculating the degree of IRR for individual and total scores over any other method of calculation. For this reason, the authors focused on ICC to analyze the IRR of individual and total scores, while the kappa calculations were used as additional insight on the reliability of categorizing risk patients. 

With regard to item subscores, the friction and shear item had the lowest ICC value (0.266). This is in contrast to a previous study⁶ in an acute care setting where the friction and shear item had 1 of the highest ICC values with a strong degree of agreement. The difference in findings could be due to variations among the sample population (the current study sample had a higher risk of PU formation because of the restricted mobility inclusion criteria of the RCT). This was evident in this study; 47% of the participants had new or preexisting PUs. Thus, it is conceivable this subpopulation of patients may potentially be more prone to IRR issues compared to patients at lower risk. 

From the interviews in this study, nurses identified friction and shear as a difficult item to rate. This was attributed to the subjective wording used to describe the item subscores and a disconnect between titles and specific descriptions. Friction and shear was an item requiring improvement in scoring reliability. Previous studies in home care²⁰ and intensive care settings²¹ have identified sensory perception, activity, nutrition, and moisture as items with the lowest ICC values.⁶ The ICC values for sensory perception, activity, nutrition, and moisture were comparable to what was found in previous studies.²¹ 

Despite the high ICC correlation for the total score, the Braden Scale will not be clinically useful without staff buy-in, the access to use the scale, or a standardized action plan to implement changes in response to the risk assessment. As a result, it is important to understand the usability and compliance of the scale as used in clinical practice. Qualitative data from the interviews with nurses in this study provided insight on their perspectives toward the Braden Scale and its usability. The thematic analysis was based on nurses’ personal experiences with the Braden Scale. A number of the nurses viewed the Braden Scale as a mandatory assessment tool administered on a set time schedule determined by unit policy. 

With regard to the EMR version of the Braden Scale used in this study, the assessment tool was found in a separate flow sheet that differed from the daily assessment sections. Nurses indicated this was inconvenient and prevented them from remembering to chart the Braden scores. Furthermore, the EMR version of the scale includes only the rating headings and numbers for item scoring without the full descriptions. Although nurses were advised by the unit nurse educators to use the paper versions of the original scale provided by the units for the item descriptions, the interviewed nurses indicated these resources were often difficult to find. Through the authors’ observation of scale use in a real-life clinical setting, it is interesting to note that while the nursing staff was mainly using an abridged version of the scale, the IRR of the scale was not compromised (total score ICC = 0.807). However, having the descriptions easily accessible to the nurses to review could have assisted in the rating process by reducing the ambiguity in item and rating interpretations. The interviewed nurses also suggested that incorporating the full descriptions to the EMR version would be helpful. 

Originally, the Braden Scale was designed and tested to be used in its entirety.³¹ Bergstrom³¹ advised against any alterations to the scale because these modifications could add to the subjectivity of the tool. The authors support Bergstrom’s request and suggest all electronic versions of the Braden Scale be consistent with the original paper version, despite having additional paper resources available for reference. Based on the current findings, when implementing an electronic version of the Braden Scale, scale items and descriptions should be readily available.

Several nurses identified friction and shear as a difficult item to rate, which was consistent with the current quantitative findings. Sensory perception, moisture, mobility, and nutrition were items that also posed challenges. Nurses felt the wording used in item titles created difficulties in scoring, which was consistent with the notion that the Braden Scale incorporates subjective wording.³² According to the survey study (N = 102) by Choi et al,³³ the subjective nature of the Braden Scale may lead to difficulties that could affect scale reliability. To compensate for these difficulties, nurses indicated they would choose higher risk score as a precaution. However, this overestimation was not done by all nurses and could have led to discrepancies in the ratings recorded. 

Previous studies have suggested training and experience may be factors underlying discrepancies in Braden Scale ratings. This appears to be particularly true for new staff.⁶ In a cross-sectional study (N = 201), Nuru et al²⁴ reported nurses exposed to formal PU prevention training tended to be more knowledgeable than persons lacking this training. The study⁴ assessing the knowledge, attitudes, and practice regarding general PU prevention of nurses in Sweden found nurses were more knowledgeable if they had frequent training sessions. Although information on years of experience was not collected during the interviews, the nurses had a single inservice education session (separate from the RCT) organized by their corresponding units, along with the training provided by the new staff orientation. The inservice sessions included training on the purpose of the Braden Scale, risk factors of PU, and the full descriptions of each item and rating. As a result, nurses were generally knowledgeable about the purpose of the Braden Scale. However, some of the specific details from the education sessions were difficult to recall because the current study was conducted several months or years later. 

A number of the nurses interviewed questioned the utility of the Braden Scale because research attesting to its validity was not provided in the training session. Another issue was the uncertainty on how to interpret and utilize the Braden scores in the care of patients. As a result, some of the nurses suggested frequent refresher courses regarding the Braden Scale would be beneficial. 

Regarding attitudes toward PU prevention, Kallman and Suserud⁴ suggested nurses were convinced their own clinical judgment was superior to any PU risk assessment tool,⁴ which was also the case in this study. A majority of the nurses interviewed believed the concepts of the Braden Scale existed in standard care and preventative measures were already being implemented. They indicated this was more effective than the Braden Scale in identifying patients at risk and preventing PUs. Nurses also stated other forms of communication, such as verbal and written reports between shifts, worked better in guiding efforts directed at PU prevention compared to the Braden Scale. The nurses interviewed appeared to have a predetermined notion of the patient’s risk for PU development regardless of the scale and emphasized the patient’s level of mobility was the greatest indicator of risk. Although nurses may have an indifferent view toward the use of the Braden Scale, not using this instrument or some other validated tool for PU risk identification could have serious implications.^4,33 Without a formal assessment tool in place, PUs would not be assessed in a standardized manner and at-risk patients may be missed. This could lead to the development of avoidable PUs, because key factors in prevention may be overlooked. 

A majority of the nurses interviewed found the largest benefit of the routine use of the Braden Scale was it served as a reminder of PUs and the importance of identifying factors that predisposed to their development. Nurses also indicated the scale provided PU prevention guidance to new and less experienced staff. It may be possible to capitalize on this in marshalling support for additional training on the utility of the Braden Scale in identifying at-risk patients who would benefit from specific PU prevention measures. Other positive aspects of the Braden Scale nurses identified included its ease of use and the utility of the total sum score in quantifying risk. 

Further research regarding IRR of the Braden Scale would benefit from a multisite design enrolling a larger number of participants where concurrent scores on consecutive patients are compared. Additionally, comprehensive studies comparing units within acute care settings could provide a better understanding of what leads to observed differences between these units.

This study may be limited by the small sample size and the different time points of assessment completed by the research team and nursing staff due to the retrospective nature of the study. Any discrepancies between the EMR and the printed version of the Braden scale may be a limitation. Additionally, the study is based on results obtained from 1 hospital, albeit different units in that facility. Researchers also wanted to examine the differences between units in terms of IRR but were limited by the small and inconsistent sample sizes across the 4 different units. However, the authors believe the IRR comparison between units is a potentially fruitful area for future research because they suspect IRR varies from unit-to-unit based on factors such as unit culture and leadership as suggested by the quantitative pilot study by Marchionni et al.³⁴ 

The IRR values in terms of ICC obtained in this study indicated Braden scores in a real-life clinical setting have a high level of agreement comparable to what was found in literature, with the exception of the friction and shear item. Further training regarding the Braden Scale, particularly regarding friction and shear, may be beneficial in improving score reliability. Overall, nurses believed the use of the Braden Scale was based on unit policy. Subjective wording, challenges with the usability of the EMR version, and inaccessible resources led to difficulties in using the scale. In addition to personal experiences with the Braden Scale, training and unfamiliarity of scale validity may have led nurses to believe use of the scale was redundant to the care they were providing. Study findings can be used to direct efforts to enhance the usability of the scale in preventing the PUs. Nurses indicated the Braden Scale serves as a reminder of the factors that predispose to PU development and it provided guidance on assessment. Recommendations to improve its usability would include attention to how it is presented in electronic versions, ensuring the inclusion of item descriptions and ready access to resource material; and further training on the validity, administration, and interpretation of the Braden Scale. Based on these findings, the factors identified in the study need to be considered when using the Braden Scale in clinical practice. By addressing these issues, effective implementation of the instrument may be achieved. 

The study was funded by the Alberta Innovates Health Solutions Collaborative Research and Innovation Opportunities team grant (number 20130152).

Dr. Ho is an Associate Professor, Cumming School of Medicine, University of Calgary; and Section Chief, Physical Medicine and Rehabilitation, Department of Clinical Neurosciences, University of Calgary and Alberta Health Services at the Foothills Medical Centre, Calgary, AB, Canada. Ms. Cheung is a Research Assistant, W21C Research and Innovation Centre, Cumming School of Medicine, University of Calgary. Ms. Southern is a Senior Statistician, O’Brien Institute of Public Health; and W21C Research and Innovation Centre, Cumming School of Medicine, University of Calgary. Ms. Ocampo is a Research Associate; and Dr. Kaufman is a Strategic Programs Manager, W21C Research and Innovation Centre, Cumming School of Medicine, University of Calgary. Dr. Hogan is the Chair of the Brenda Strafford Foundation in Geriatric Medicine, University of Calgary. Dr. Baylis is an Executive Co-Leader, W21C Research and Innovation Centre; and a Section member, Department of Medicine, Division of General Internal Medicine, Cumming School of Medicine, University of Calgary. Dr. Conly is the Medical Director, W21C Research and Innovation Centre, Cumming School of Medicine; Professor, Departments of Medicine and Infection Prevention and Control, Alberta Health Services; and a faculty member, O’Brien Institute for Public Health, University of Calgary. Dr. Stelfox is the Scientific Director of Alberta Health Services’ Critical Care Strategic Clinical Network; Associate Professor in the Departments of Critical Care Medicine, Medicine and Community Health Sciences, University of Calgary; and a faculty member, O’Brien Institute for Public Health, Cumming School of Medicine, University of Calgary. Dr. Ghali is the Scientific Director, O’Brien Institute for Public Health, Foothills Medical Centre, University of Calgary; Co-Leader, W21C Research and Innovation Centre; and a Professor, Departments of Community Health Sciences and General Internal Medicine, Cumming School of Medicine.

Please address correspondence to: Chester H. Ho, Division of Physical Medicine and Rehabilitation, Department of Clinical Neurosciences, Foothills Hospital, University of Calgary, 1403- 29th Street NW, Calgary, AB, Canada T2N 2T9; email: chester.ho@albertahealthservices.ca.