P2Y12 Inhibitor Utilization for Acute Coronary Syndrome in the Elderly: A Provincial Analysis From 2008 to 2018

Background Guidelines recommend acetylsalicylic acid (ASA) and ticagrelor following acute coronary syndrome (ACS), but appropriate prescription practices lag. We analyzed the impact of government medication approval, national guideline updates, and publicly funded drug coverage plans on P2Y12 inhibitor utilization. Methods Accessing provincial databases, we obtained data for elderly ACS patients in Ontario, Canada between 2008 and 2018. Using an interrupted-time series, we evaluated types of P2Y12 inhibitors prescribed at discharge, and changes to their utilization following ticagrelor’s national approval by Health Canada, national antiplatelet therapy guidelines (by the Canadian Cardiovascular Society (CCS)), and ticagrelor's coverage by a publicly funded medication plan. Results We included 114,142 patients (49.4%-PCI and 7.7%-CABG). Proportion of PCI patients utilizing P2Y12 inhibitors increased from 73.4% to 86.9% (p<0.001) and 11.4% to 46.5% (p<0.001) for CABG patients. Among PCI patients, clopidogrel utilization declined monthly after 2010 national guidelines were published (0.7%; p<0.0001) and within the rst month after ticagrelor’s national approval by Health Canada (5.3%; p=0.03). Among PCI patients, ticagrelor utilization increased within the rst month (24.5%; p<0.0001) and continued increasing monthly (0.4%; p<0.0001) after its coverage by a publicly funded medication plan. Among CABG patients, ticagrelor’s coverage was associated with a monthly increase in its utilization (0.2%; p<0.0001). Among PCI patients, clopidogrel utilization declined within the rst month (6.1%; p=0.003) and ticagrelor utilization increased monthly (0.3%; p=0.05) after 2012 CCS guidelines. Conclusion National guideline updates and drug coverage by a publicly funded medication plan signicantly improved P2Y12 inhibitor utilization. Barriers to appropriate antiplatelet therapy in the surgical population must be explored. Abstract Database (DAD) and Same-Day Surgery Database (SDS) from Canadian for Health Information (CIHI), National Ambulatory Care Reporting System (NACRS), and Registered Persons Database (RPDB). These databases, linked via unique encoded identiers analyzed at ICES, were used to analyse health outcomes as individuals can be tracked over time.(18–21) Available data on admission for ACS, patient characteristics, and management strategies were captured retrospectively from the CCN (CorHealth) database.


Introduction
Based on large, multi-center randomized controlled trials (RCTs), guidelines recommend dual antiplatelet therapy (DAPT) with acetylsalicylic acid (ASA) and a P2Y12 inhibitor (speci cally, ticagrelor) for 12 months after an acute coronary syndrome (ACS), regardless of management strategy: medical management, percutaneous intervention (PCI), or coronary artery bypass grafting (CABG).(1-10) (Fig. 1) Despite these recommendations and available evidence, appropriate DAPT use remains variable. For example, the ACS Re ective Program -a multicenter national observational registry conducted from 2011 to 2013 -found that clopidogrel remained the most common prescribed P2Y12 inhibitor across Canada at a rate of 82%, compared to only 9% for ticagrelor. Further, the prospective Canadian Observational Antiplatelet Study (COAPT) demonstrated that despite RCTs and guideline recommendations, only a minority of ACS patients were discharged on ticagrelor (11.1%) or prasugrel (5.7%), compared to clopidogrel, after PCI. (11,12) DAPT is particularly underused in patients undergoing CABG. In a cohort of patients with ACS, CABG was an independent predictor for DAPT underutilization (odds ratio [OR] 0.09, 95%CI 0.05-0.14). (13)(14)(15) Suboptimal evidence implementation is not unique to antiplatelet therapy and ACS patients. Glynn et al. reviewed random patient samples from 12 metropolitans in the United States (U.S.) and found that less than 55% received guideline-recommended care. (16) More speci c to cardiovascular health, a review of care gaps (de ned as under-utilization or under-dosing of proven treatments) found that only 34-60% of eligible outpatients in Europe, Canada and U.S. received appropriate medications for heart failure, coronary artery disease or atrial brillation. (17) As such, implementation science strategies are paramount in improving healthcare delivery and quality of care.
To address the knowledge-implementation gap around antiplatelet therapy and ACS, we used provincial administrative data to examine P2Y12 inhibitor utilization after discharge of elderly ACS patients. We aimed to evaluate whether the preponderance of clopidogrel prescriptions among patients managed medically or with PCI still stands, and if CABG patients saw an increase in DAPT utilization. We also sought to determine how prescription practices varied between patients managed with PCI, CABG or medically, and how -if at all -knowledge publication, government approval, guideline recommendations and publicly funded medication plan coverage impacted practice patterns.

Study Population
We included all patients 65 years of age or older who presented with an ACS and were subsequently referred for coronary angiogram in the province of Ontario, Canada from October 1, 2008 (chosen as such due to availability of complete Cardiac Care Network (CCN) datasets from then onwards) to March 31, 2018. Ontario includes 20 cardiac centers, which provide care to 14.5 million inhabitants. Apart from universal healthcare, the provincial government provides a publicly-funded medication plan for those 65 years or older (Ontario Drugs Bene ts (ODB)).

Databases and Cohort De nition
We obtained data on baseline characteristics and prescriptions lled from ve databases linked at the Institute for Clinical Evaluation Sciences (ICES) -CorHealth, Ontario Drug Bene t (ODB) database, Discharge Abstract Database (DAD) and Same-Day Surgery Database (SDS) from Canadian Institute for Health Information (CIHI), National Ambulatory Care Reporting System (NACRS), and Registered Persons Database (RPDB). These databases, linked via unique encoded identi ers analyzed at ICES, were used to analyse health outcomes as individuals can be tracked over time. (18)(19)(20)(21) Available data on admission for ACS, patient characteristics, and management strategies were captured retrospectively from the CCN (CorHealth) database.
Using variables from CCN, we applied the following cohort exclusions: aged younger than 65 years or older than 105 years (which was derived from RPDB), did not have a coronary angiogram within 3 months of ACS, primary reason for cardiac surgery referral was not coronary disease, died prior to discharge from facility performing coronary angiogram, normal anatomy on angiogram, or nonobstructive coronary artery disease.

Outcomes
We tracked the number of P2Y12 inhibitor prescriptions lled within 14 days (per Ontario Primary Care Medication Reconciliation Guide, medications should be reconciled within 14 days from hospital discharge), based on treatment strategy. Eligible agents included ticagrelor, clopidogrel and prasugrel. We assumed that patients taking either one of these antiplatelet agents were also taking ASA, since it is available over the counter and cannot be tracked by ODB.
Our outcome of interest was change in prescription utilization at speci c time points, decided by discussion among the authors and feedback from reviewers; approval of ticagrelor by Health Canada in

Statistical Analysis
We created an interrupted time series with the dataset aggregated on a monthly basis. The dependent variable was the proportion of patients dispensed a P2Y12 inhibitor within 14 days of hospital discharge. Our autoregressive model included the following covariates: 1. Intervention (medical, PCI or CABG) -binary indicator denoting 0 during pre-intervention period and 1 during post-intervention period; 2. time (in months) -ordinal indicator denoting months since start of study; 3. time since intervention (in months) -ordinal indicator denoting months since time interruption (i.e. implementation of intervention);

autoregressive processes (if applicable).
We also incorporated lag periods in our interrupted time series; they included the month of intervention and three months after (to allow for uptake of intervention into clinical practice). All data points enclosed in the lag period were excluded from analysis.
To build the model, we initially added 13 autoregressive parameters because the datasets were aggregated by month − 13 processes allow the models to detect and adjust for seasonality. In a backward stepwise fashion, autoregressive parameters were removed until all autoregressive parameters remained statistically signi cant. Presence of autoregression (or autocorrelation) was assessed using Durbin-Watson statistic. (22) Presence of seasonality (stationarity) was assessed using Dickey-Fuller statistic. (23) The output and data analysis for this study was generated using SAS software.

Ethics approval
The study was approved by the Hamilton Integrated Research Ethics Board (HiREB #4196). ICES is an independent, non-pro t research institute whose legal status under Ontario's health information privacy law allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement.

Results
Population Characteristics Figure 1 represents the study ow diagram. We identi ed a total of 114,142 elderly patients and present their baseline characteristics in Table 1 proportion presenting with congestive heart failure (CHF), chronic obstructive pulmonary obstruction (COPD), cerebrovascular disease (CVD), and peripheral vascular disease (PVD). Further, medically managed patients were more likely to already be on antithrombotic medications before their ACS. At the beginning of our accrual period, the proportion of ACS patients prescribed P2Y12 inhibitors at discharge was 73.4% in the PCI group, 11.4% in the CABG group, and 31.4% in the medically managed group. As of March 2018, the proportion had increased to 86.9% (p < 0.001) for patients undergoing PCI, 46.5% (p < 0.001) for those undergoing CABG, and 32.3% (p = 0.775) for those managed medically. The baseline characteristics, including medical co-morbidities and previous antithrombotic use, were measured when patient presented with their ACS.
While we evaluated prasugrel utilization independently, the number of patients discharged on prasugrel was extremely low, and we were unable to conduct a robust analysis; the models did not converge and could not be interpreted.

PCI Patients
The publication of the 2010 CCS Antiplatelet Therapy Guidelines in May 2011 was associated with a signi cant decline of 6.1% (p = 0.003) in utilization of all available P2Y12 inhibitors within the rst month. The trend continued, but not signi cantly, in the months after. Individual clopidogrel utilization also declined within the rst month after guideline updates, albeit not signi cantly. Clopidogrel utilization, however, declined signi cantly by 0.7% (p < 0.0001) every month thereafter.

CABG Patients
The publication of 2010 CCS Antiplatelet Therapy Guidelines in May 2011 was associated with a decline in utilization of all available P2Y12 inhibitors within the rst month and every month thereafter, however this was not signi cant. Clopidogrel utilization demonstrated similar trends, which were also not signi cant.

Medically Managed Patients
The publication of 2010 CCS Antiplatelet Therapy Guidelines was associated with a signi cant decline of 3.8% (p = 0.002) in utilization of all available P2Y12 inhibitors within the rst month, followed by a signi cant decline of 0.2% (p = 0.002) every month after. Individually, the guideline updates also impacted clopidogrel utilization within the rst month and were associated with a signi cant decline of 2.5% (p = 0.02), followed by a signi cant decline of 0.3% (p < 0.0001) every month after.
P2Y12 Inhibitor Utilization and Ticagrelor Health Canada Approval (Fig. 3) We could not evaluate the impact of ticagrelor's Health Canada approval on ticagrelor prescriptions itself as it was not approved by ODB yet.

PCI Patients
The approval of ticagrelor by Health Canada in June 2011 was associated with a signi cant decline of 7.3% (p < 0.002) in utilization of all available P2Y12 inhibitors within the rst month, and while the trend continued monthly thereafter, it was not signi cant. Utilization of clopidogrel declined by 5.3% (p = 0.03) within the rst month, and continued to decline every month thereafter, but not signi cantly.

CABG Patients
The approval of ticagrelor by Health Canada in June 2011 demonstrated a decline in utilization of all available P2Y12 inhibitors, but this was not signi cant. The trend continued every month thereafter, without reaching statistical signi cance. Clopidogrel utilization showed similar trends, which also failed to reach statistical signi cance.

Medically Managed Patients
The approval of ticagrelor by Health Canada in June 2011 was associated with a signi cant decline of 4.3% (p < 0.001) in utilization of all available P2Y12 inhibitors within the rst month and continued to signi cantly decline monthly thereafter by 0.2% (p < 0.001). Clopidogrel utilization also declined by 3.2% (p = 0.003) within the rst month and continued to signi cantly decline every month thereafter by 0.3% (p < 0.001).
P2Y12 Inhibitor Utilization and Coverage of Ticagrelor by the Publicly Funded Medication Plan (Fig. 3)

PCI Patients
Ticagrelor's coverage by the publicly funded medication plan in August 2013 was associated with an increase of 11.4% (p < 0.0001) in utilization of all available P2Y12 inhibitors within the rst month after coverage began, and while the trend continued every month after, it was not signi cant. The approval was associated with a signi cant decline of 14% (p < 0.0001) in clopidogrel utilization, followed by a decline of 0.2% (p = 0.03) every month thereafter. Meanwhile, ticagrelor utilization increased by 24.5% (p < 0.0001) within the rst month and continued to increase by 0.4% every month (p < 0.0001) thereafter.

CABG Patients
Ticagrelor's coverage by the publicly funded medication plan in August 2013 was not associated with a signi cant change in utilization of all available P2Y12 inhibitors, or clopidogrel and ticagrelor individually, within the rst month. However, after, there was a signi cant increase of 0.3% (p = 0.0001) every month in utilization of all available P2Y12 inhibitors. With the coverage, clopidogrel utilization also increased every month after by 0.2% (p = 0.04). Meanwhile, ticagrelor utilization also increased by 0.2% (p < 0.0001) every month after its approval.

Medically Managed Patients
Ticagrelor's approval by the publicly funded medication plan in August 2013 was not associated with a signi cant impact on utilization of all available P2Y12 inhibitors. Individually, the approval was associated with a signi cant decline of 0.1% (p = 0.005) in clopidogrel utilization every month after.
Meanwhile, ticagrelor utilization demonstrated a signi cant increase of 2.2% (p < 0.0001) within the rst month after its approval and continued to signi cantly increase by 0.12% (p < 0.0001) every month after.

PCI Patients
The publication of 2012 CCS Antiplatelet Therapy Guidelines in August 2013 was associated with a signi cant increase of 7.4% (p = 0.0004) in utilization of all available P2Y12 inhibitors within the rst month after publication. The trend continued in the subsequent months but was not signi cant. The updated guidelines were associated with a decline of 6.1% (p = 0.003) in clopidogrel utilization within the rst month after publication, followed by a signi cant decline of 0.4% (p = 0.002) every month thereafter. Meanwhile, ticagrelor utilization demonstrated an increase within the rst month, albeit not signi cant.

CABG Patients
With the publication of 2012 CCS Antiplatelet Therapy Guidelines in August 2013, utilization of all available P2Y12 inhibitors decreased by 5.2% (p = 0.05) within the rst month but did not demonstrate a signi cant trend every month thereafter. While clopidogrel utilization also declined within the rst month, and monthly afterwards, the change was not signi cant. Similarly, while ticagrelor utilization increased every month after guideline publication, the change was not signi cant.

Medically Managed Patients
The publication of 2012 CCS Antiplatelet Therapy Guidelines was not associated with a signi cant change in utilization of all available P2Y12 inhibitors within the rst month or after. Individually, the guideline updates were also not associated with a signi cant change in clopidogrel utilization. After the 2012 guideline updates, ticagrelor utilization, however, demonstrated a signi cant increase of 1.7% (p < 0.0001) within the rst month. This trend continued every month after but was not signi cant.

Discussion
We analysed P2Y12 inhibitor utilization data on 114,142 elderly ACS patients, and identi ed three key evidence-practice gaps. First, the uptake of evidence is slow and although the proportion of patients receiving P2Y12 inhibitors after an ACS has improved in the last decade, the proportion of patients receiving guideline-directed therapy with ticagrelor remains low. Second, DAPT use is suboptimal for patients who undergo CABG. Third, and perhaps most importantly, policies have the biggest in uence on changing practice. Drug coverage by a publicly funded medication plan and guideline updates had the largest impact on appropriate medication utilization.
DAPT has proven bene t for ACS patients undergoing medical management, PCI or CABG.(6,7) An RCT of 18,624 patients randomized to ASA and ticagrelor demonstrated that the latter reduced the primary composite outcome of vascular mortality, myocardial ischemia (MI) or cerebrovascular accidents (CVA) -9.8% with ASA and ticagrelor vs 11.75% with ASA and clopidogrel (HR 0.84; 95%CI 0.77-0.92) at 12months.(6) Previously, Yusuf et al. compared DAPT with ASA and clopidogrel with ASA monotherapy and demonstrated that a composite of death from cardiovascular causes, non-fatal MI or stroke occurred in 9.3% of patients receiving DAPT and in 11.4% of patients receiving ASA only (RR 0.80; 95% CI 0.72-0.90; p < 0.001). (7) The results of CABG subgroups from both above trials were also consistent with above ndings, demonstrating a reduction in major adverse cardiovascular events (MACE) when using more potent antiplatelet therapy regimens. (24,25)  Of the 83% of patients who were discharged on any antiplatelet therapy, 57% were prescribed ASA and clopidogrel and only 0.1% of patients were discharged on ASA and ticagrelor. (30) DAPT is even further underutilized in ACS patients who undergo CABG compared to PCI. In a retrospective study of 8,939 ACS patients in Australia, CABG was an independent predictor for DAPT underutilization; OR 0.09, 95%CI 0.05-0.14.(13) A recent survey evaluating the practice patterns of 75 Canadian cardiac surgeons around postoperative antiplatelet management found that only 45% of cardiac surgeons would restart DAPT in patients who suffered a recent ACS. Interestingly, they were more likely to initiate DAPT in patients who had previous stents to vessels that were not bypassed, required endarterectomy, or suffered a peri-operative MI. Respondents (81%) were more concerned with preventing bleeding than recurrent ischemic events. (31) While underutilization of ticagrelor post-ACS in our cohort may appear concerning, we must note that our population is elderly, with a mean age over 74, and use of clopidogrel may be justi ed. Recently, the POPular AGE Trial (Clopidogrel versus ticagrelor or prasugrel in patients aged 70 years or older with non-ST-elevation acute coronary syndrome) demonstrated that ticagrelor was prematurely discontinued in 47% of patients (due to bleeding and dyspnea) compared to clopidogrel, which was only discontinued in 22% of patients. The authors also concluded that clopidogrel was an appropriate alternative to ticagrelor among ACS patients aged 70 years or older.(32) Within Canada, the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease registry evaluated clopidogrel versus ticagrelor prescriptions and outcomes in ACS patients who were discharged after a PCI. In their cohort of 11,185 PCI patients, compared with clopidogrel, ticagrelor was not associated with lower risk of MACE (aHR 0.97; 95% CI 0.85-1.10), but was associated with an increased risk of major bleeding (aHR 1.51; 95% CI 1.29-1.78) and dyspnea (aHR 1.98; 95% CI 1.47-2.65).(29) These ndings introduce a nuance to bleeding and dyspneic risk with ticagrelor, which seems particularly pronounced among the elderly, potentially explaining why clopidogrel remains so widely used in our cohort.
Additionally, it should be noted that evidence for use of DAPT after CABG has signi cant limitations -RCTs are small or are sub-studies of larger RCTs. Furthermore, there is signi cant population heterogeneity within the RCTs on whether patients underwent off-pump CABG (OPCAB) or CABG with cardiopulmonary bypass. In fact, a large majority of patients randomized in the two trials demonstrating DAPT superiority over ASA monotherapy underwent OPCAB. (33,34) The available evidence does not evaluate the use of DAPT in patients undergoing multiple arterial grafting, which are associated with signi cantly better patency rates and clinical outcomes compared with vein grafts. and the effect DAPT has on the latter may not apply to arterial grafts. (35,36) Given the limitation of existing evidence, physicians and surgeons are more concerned with post-procedural bleeding risk, lack of dedicated robust evidence in CABG patients.
Given the above ndings and real-world data, the discordance between guidelines and practice is likely multifactorial. The underutilization of ticagrelor may be due to some of the barriers to physician adherence discussed by Cabana et al. Speci cally, with ever-expanding research, physicians may nd it challenging to keep abreast with updated guidelines or apply them due to previous practice inertia. While lack of familiarity with ticagrelor (pharmacology and evidence around it) or external barriers (perceived issues with costs and drug coverage) may play a role, the most likely factors are likely due to lack of outcome expectancy (belief in e cacy) and lack of agreement (belief in safety, especially fear of increased bleeding, and belief in strength of recommendations). (37) In fact, the fear of post-procedural bleeding may also explain why all P2Y12 inhibitor utilization among CABG patients declined signi cantly after the 2012 CCS Antiplatelet Therapy Guidelines were published. Despite strong recommendations for ASA and ticagrelor, surgeons were likely not convinced of the e cacy and safety in this population. (11) Our observation that provincial coverage increases prescription practice is aligned with the results of other studies and highlights the importance of health bene ts and provincial formulary coverage. In 2011, Quebec became the rst province in Canada to cover ticagrelor and prasugrel after an ACS; on discharge, Quebec showed increased use of ticagrelor and prasugrel (37.1%) compared to rest of Canada (19.2%). (11) In 2008, Jackevicius et al. published on the impact of ODB coverage and clopidogrel use among patients who underwent PCI after an acute MI in Ontario; demonstrating an increase from 35% utilization before coverage to 88% after coverage in the rst 30 days after discharge (p < 0.001). (38) Though adherence to guideline-directed use of DAPT following ACS will never reach 100%, targeted knowledge translation strategies are required to improve antiplatelet management for ACS patients in Ontario, and around the world. As per the practical knowledge translation framework, a national survey to understand barriers towards appropriate antiplatelet prescription is required. (39) First, this would help address the underutilization of ticagrelor as compared to clopidogrel, and would aim to address the subset of ACS patients undergoing CABG, where surgeons are least adherent to prescribing guideline-directed DAPT.
Increased prescription compliance in this population could close the practice gap and improve secondary prevention for these patients. Lastly, knowledge translation strategies should be developed during the medication study period (the pre-trial results stage) to expedite uptake if the study is positive.

Limitations
Our study has several limitations. First, we included patients who had a coronary angiogram within three months after an ACS. Given this, there is a potential for survival bias in our cohort assembly since some patients may have died before undergoing an angiogram and may represent a population with different prescription patterns. While we discuss P2Y12 inhibitor prescription practices in our study, it should be noted that patient compliance is a signi cant limitation of our study. That is, our data capture is dependent on patients lling their prescriptions, and not whether physicians prescribed the appropriate medication. Similarly, not all patients are compliant with over the counter ASA, and the inability to capture how many patients were taking this also remains a limitation. Last, 24.5% of medically managed patients, 20.7% of PCI patients, and 13.3% of CABG patients were on P2Y12 inhibitors before presenting with an ACS. Presumably, if these patients were continued on the same, they may not have required a prescription re ll within the 14-days after discharge, leading to a miscount of actual P2Y12 inhibitor utilization. Lastly, a signi cant number of patients were on more potent anticoagulants (either a vitamin K antagonist or a direct oral anticoagulant), and may present a higher risk for bleeding, contraindicating the addition of a P2Y12 inhibitor, especially given the elderly age of our cohort. Furthermore, the study design was such that it did not capture patients under the age of 65, nor capture prescriptions not covered by ODB. Due to this, we did not capture the prescription utilization of younger ACS patients. And we can only make assumptions about ticagrelor utilization rates prior to ODB coverage based on the decline demonstrated in utilization of other P2Y12 inhibitors. For instance, the drop in clopidogrel and all available P2Y12 inhibitor prescriptions within the month after ticagrelor's Health Canada approval and the 2010 CCS Antiplatelet Therapy Guideline publication likely re ects an increase in out-of-pocket (paid-for by patients) ticagrelor prescriptions. Furthermore, the use of "Brilinta Savings Card", which allowed some patients to pay as low as $5 for a 30 day supply of ticagrelor after Health Canada approval, along with samples from drug companies, most likely led to the decline in utilization of all available P2Y12 inhibitors, as more patients switched to ticagrelor. However, we cannot con rm this hypothesis.

Conclusion
Our study shows an overall increase in utilization of all available P2Y12 inhibitors for elderly ACS patients compared to previous Canadian national registries. Despite guideline recommendations and de nitive evidence, gaps exist in management of ACS patients for which we must identify barriers to inform future knowledge translation interventions. Policy changes appear to be most effective at moving clinicians towards guideline-directed medical therapy.

Declarations Ethics Approval and Consent to Participate
The study was approved by the Hamilton Integrated Research Ethics Board (HiREB #4196).
No individual patient consent to participate was required.

Consent for Publication
Not applicable.

Availability of Data and Materials
The data that support the ndings of this study are available from ICES, but restrictions apply to the availability of these data, which were used under licensure for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of ICES.
Note: ICES is an independent, non-pro t research institute whose legal status under Ontario's health information privacy law allows it to collect and analyze health care and demographic data, without consent, for health system evaluation and improvement.

Competing Interests
Dr Shamir Mehta is a consultant for AstraZeneca and receives grant support from them.
All the other authors do not have any competing interests; nancial or otherwise.

Funding
This study was conducted with grant support awarded to Dr Saurabh Gupta by Hamilton Health Sciences -NIF-17425. Additionally, this study was supported by ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC).
Author Contributions SG made substantial contributions to the conception and design of the work; acquisition and interpretation of data; drafted the work and substantially revised it.
EBC made substantial contributions to the conception and design of the work; and substantially revised it.
AE made substantial contributions to interpretation of data; drafted the work and substantially revised it. CE made substantial contributions to interpretation of data; drafted the work and substantially revised it.
AB and NW made substantial contributions to drafted the work and substantially revised it.
JC made substantial contributions to data acquisition, analysis and interpretation of data. SM, JDS made substantial contributions to design of the work and substantially revised it.
RW made substantial contributions to contributions to the conception and design of the work; substantially revised it; and supervised all efforts.