Are the Real-World Data available in Canada suitable for drug reimbursement decisions?

Nigel Rawson | CHP | 24 SEP 2019

Much enthusiasm exists for the use of real-world data (RWD) sources for drug reimbursement and pricing decisions in Canada. However, a recent report providing insight into the views of various stakeholders about the use of these resources for decision-making demonstrated that the participants had several misperceptions about RWD. These misperceptions are likely the result of the RWD resources currently available in Canada being predominantly administrative health care utilization data.

Researchers in Saskatchewan and Manitoba were innovative pioneers in the use of administrative data for health services research in the 1970s and for pharmacoepidemiology in the 1980s and 1990s. Later, researchers in other provinces also used their local administrative data for research purposes. However, Canada has failed to capitalize on its lead. Other nations have moved forward by combining administrative data with electronic medical records (EMRs), laboratory test data and, in some cases, social, psychological and genetic information to develop more comprehensive RWD.

Despite billions of dollars being spent on the Canada Health Infoway, a not-for-profit government corporation created to foster and accelerate the implementation of a pan-Canadian EMR network, Canada has been slow to introduce large-scale EMR systems, and few provinces have complete laboratory, screening and other test information that are fully linkable with their administrative data or other resources such as registries. This slowness to develop more advanced resources is perhaps not surprising when many physicians, hospitals and pharmacies still communicate about patients using faxes.

The existing RWD available in Canadian provinces are frequently challenging to access and use. Examples of access difficulties are a lack of a central point of data procurement in some provinces, data use restricted to local accredited academics, and lengthy time required to obtain administrative and ethics approvals. In addition, differences exist in how data are provided; some provinces deliver a tested dataset, while others provide a data dump that necessitates much effort by the end user to organize into a useable format. Moreover, there is a lack of consistent data standards across different provinces; for example, the primary hospital discharge diagnosis in Quebec is the most important condition encountered by the patient during hospitalization, which may not be equivalent to the most responsible hospital diagnosis (the one requiring the greatest consumption of resources) used in Saskatchewan.

Additional limitations in utilizing administrative health care data include the need to assess the validity of diagnoses in each use because diagnostic validity is not of high importance when services are recorded, and the restrictive clinical criteria that patients must fulfill to obtain access to many drugs in government programs that often vary between provinces, which can lead to biased results. These issues also make analyses based on amalgamated data from different provinces questionable because combining data from resources with differing recording criteria does not eliminate the variation – more likely, it accentuates differences resulting in misleading outcomes.

Canada has long had a need for better information on the effectiveness of its health care system, but progress towards this goal is hindered by the country’s fragmented health care system. Moreover, the siloed approach of separate branches within each provincial government supporting different health care services results in little evaluation of the impact of each branch’s expenditure on the others. For example, an increase in spending on a high-cost innovative drug is frequently seen solely as a problem for the budget of the pharmacy service branch, even if it reduces the need for hospitalizations resulting in a saving for the hospital service branch that exceeds the additional medication expenditure.

Canada’s provincial governments have generally failed to invest in improving and extending their data beyond the information recorded to manage their health care delivery budgets because they have little incentive to do so. Administrative health care utilization data mainly record diagnoses, procedures performed and dispensed prescriptions, which indicate that services were required. They do not record whether a patient benefited from a medicine or treatment. Administrative data have a role in drug safety research, but they are an inappropriate resource with which to evaluate the effectiveness of new therapies. The lack of repeat prescriptions or further visits to the physician does not necessarily mean that the patient’s condition improved. The patient may have given up on the treatment because it was not working or it caused unpleasant side effects and tried something else. Alternatively, an absence of data could mean that the patient moved away or, more extremely, that they died suddenly. An assessment of the effectiveness of new therapies requires linked access to both administrative and clinical outcome data.

When clinical data are in electronic form in Canada, they tend to be in small-scale EMR systems. The few large EMR systems that do exist are generally not linkable with administrative data and access is often restricted to specific researchers. Clinical data may also be available in registry systems, some of which have been highly successful (e.g. Cystic Fibrosis Canada has a registry that includes virtually all Canadians with this disorder that provides valuable information about sufferers), but again they are rarely linkable with government administrative data.

In June 2019, the federal government’s Advisory Council on the Implementation of National Pharmacare submitted its final report on how a national public drug insurance system should be introduced in Canada. One of the Advisory Council’s recommendations is the collection and analysis of real-world evidence to help inform formulary decision-making, especially about the benefits and risks of drugs for rare disorders.

However, except in unusual circumstances, no Canadian RWD about a drug’s use exist when a reimbursement recommendation is made about the drug and negotiations regarding its price are undertaken. Moreover, public payers and the Canadian Agency for Drugs and Technologies in Health (CADTH), which provides recommendations to the governments that fund and manage it regarding whether medications should be reimbursed by the public payers, continue to want to see randomized clinical trial results in health technology assessment submissions, even those for drugs for rare disorders where such trials are difficult to accomplish. CADTH and the public payers regard results from randomized clinical trials more highly because they are committed supporters of evidence hierarchies that place RWD at a lower level of reliability, regardless of the circumstances.

Unless RWD analyses are available from other countries and accepted by CADTH, reimbursement recommendations will most likely continue to be made on “cost-effectiveness” assessments based on the results of efficacy data derived from randomized clinical trials and the manufacturer’s anticipated list price. From July 2020, maximum prices for new drugs are also going to be based on these assessments. Although constituting the best available information at the time, efficacy is an inadequate estimate of effectiveness and the list price does not represent the discounted price offered by the manufacturer to the provincial drug programs. Only once a drug is in everyday clinical use in Canada can genuine cost-effectiveness analyses based on RWD be performed.

The Advisory Council’s recommendation for the use of RWD to inform formulary decision-making may be apposite, but the pertinent question is, are valid and reliable evaluations of the effectiveness of medicines possible with the RWD currently available? Even if the present resources were considered to be adequate for this purpose, the numerous issues related to accessing and using these data create a significant challenge, if not a total blockage, for putting this recommendation into action. The situation is only likely to change if the federal, provincial and territorial governments genuinely want improved drug reimbursement and pricing decision-making (as well as better safety evidence), become more cognizant of the problems of the inadequacy of their data resources for valid analyses, and invest in developing and expanding their RWD assets.

Nigel Rawson | CHP | 24 SEP 2019