Rosiglitazone use and associated adverse event rates in Canada between 2004 and 2010
© Rawson and Terres; licensee BioMed Central Ltd. 2013
Received: 9 July 2012
Accepted: 25 February 2013
Published: 5 March 2013
We examined the change in the use of rosiglitazone-containing products (RCPs) Canada-wide between 2004 and 2010 and whether the rates of adverse events in association with RCP therapy in Canadian patients changed in this period to better understand the real world use of RCP medications and as part of a regulatory commitment by GlaxoSmithKline to Health Canada to assess whether there was an impact of a risk communication on cardiac safety.
RCP utilization data were obtained from IMS Brogan’s longitudinal de-identified patient database (known as LRx) that tracks prescription activity using store-based data collection from pharmacies in all Canadian provinces. Adverse events (AEs), serious adverse events (SAEs) and cardiac AEs associated with RCP use in Canadian patients between April 2004 and December 2010 were identified from GlaxoSmithKline’s AE database and, using the LRx data, rates per 100,000 patients were estimated.
A total of 239,184 patients were identified as having received at least one RCP prescription between 2004 and 2010 from the LRx. After excluding those with inconsistent gender or age, only one RCP prescription at the pharmacy, a prescription from a pharmacy that had not consistently reported for the past six years or an unreasonably high number of prescriptions, 180,936 patients remained for the analysis. The number of reports identified from the AE database that occurred between April 2004 and December 2010 was 1,037. The average monthly rates of AEs, SAEs and cardiac AEs decreased by 57%, 43% and 4%, respectively, between the observed periods, April 2004-October 2007 and November 2007-December 2010.
The findings of this analysis demonstrate a significant decrease in RCP use in Canada following a meta-analysis publication suggesting harm, which has been maintained. It is not possible to disentangle whether the continuing decline can be attributed to the meta-analysis, the changes in prescribing guidelines, media attention or a combination of some or all of these factors.
Following the publication in May 2007 of a meta-analysis that suggested an increased risk of myocardial infarction associated with rosiglitazone-containing products (RCPs), a dramatic decrease in the use of RCPs has been demonstrated in the United States and Europe[2–5]. In Canada, two reports from Ontario and British Columbia showed a similar decline in RCP use in those provinces[6, 7]. However, neither examined whether there was an effect associated with a risk communication on cardiac safety issued by Health Canada in November 2007, one component of which was that rosiglitazone “is no longer approved as monotherapy for type 2 diabetes.” There was also a reminder that RCPs are not indicated for use as a component of a triple therapy (i.e. in combination with metformin and a sulfonylurea) diabetes regimen.
To better understand the real world use of RCP medications and as part of a regulatory commitment by GlaxoSmithKline (GSK) to Health Canada to assess whether there was an impact of the risk communication across Canada in terms of decreasing use of RCPs as monotherapy or as part of triple or triple-plus therapy, we examined the change in the use of RCPs Canada-wide between 2004 and 2010. The commitment also required an evaluation of the rates of adverse events (AEs), especially cardiac AEs, in association with RCP therapy in Canadian patients before and after the release of the risk communication.
The only cross-Canada resource for drug utilization commercially available is IMS Brogan’s longitudinal de-identified patient database (known as LRx), which tracks prescription activity using store-based data collection from pharmacies in all Canadian provinces. LRx covers 63% of prescriptions nationally and is recognized as a critical national information source.
Patients dispensed prescriptions for RCPs, other oral anti-diabetics or insulin between 2004 and 2010 were identified from LRx and those dispensed RCPs alone or in combination retained by IMS Brogan. Since LRx only began in 2004 and data from the first three months are misleading because they demonstrate the building of the database rather than real utilization, the number of RCP patients per month was calculated from April 2004 onwards. Since it is the usual way that IMS Brogan provides data and because a national picture was required, proportional allocation procedures established by IMS were used to provide an estimate of overall national RCP use and as monotherapy, in combination with another oral anti-diabetic or insulin (dual therapy), or in combination with multiple other anti-diabetic products (triple-plus therapy) in the Canadian population of 34.5 million. These data were provided to GSK.
AEs, serious adverse events (SAEs) and cardiac AEs associated with RCP use in Canadian patients between April 2004 and December 2010 were identified from GSK’s AE database and, using the LRx data, rates per 100,000 patients were estimated. The AE database has been approved by and information from it is shared with regulatory agencies. It has been shown to be more comprehensive than both the US Food and Drug Administration and the World Health Organization adverse reaction databases for GSK’s products. Nevertheless, since the data come from voluntary reporting, they are undoubtedly under-reported. In addition, the AE data did not specify each patient’s RCP treatment regimen.
Since the data used in this analysis were de-identified, ethics approval and patient consent were not required. Moreover, because the analysis was focused on the impact of the Health Canada risk communication on a change in prescribing requirements and cardiac AEs, information on fractures or other AEs was not collected.
A total of 239,184 patients were identified as having received at least one RCP prescription between 2004 and 2010 from the LRx. After excluding those with inconsistent gender or age (11,868), only one RCP prescription at the pharmacy (9,669), a prescription from a pharmacy that had not consistently reported for the past six years (35,347), or an unreasonably high number of prescriptions (1,364), 180,936 patients remained. These exclusions remove a significant number of patients, but we chose to be confident in the data that we used rather than unsure about the reliability of part of the data.
Estimated average monthly rates of adverse events per 100,000 patients in two time periods
April 2004 - October 2007
November 2007 - December 2010
Serious adverse events
Cardiac adverse events
The dispensing of RCPs across Canada decreased substantially from the peak in May 2007 to October 2007 and continued to decline after the November 2007 risk communication through to December 2010. These results are consistent with other analyses in Canada[6, 7] and with studies in the United States and Europe[2–5]. However, due to the limited information available, it was not possible to assess whether the risk communication had any additional impact on an already decreasing utilization pattern.
The largest decrease was in the monotherapy group and the smallest in the triple-plus therapy group. The former is consistent with the change in prescribing guidelines. The latter is not, but it should be recognized that patients on triple-plus therapy are likely to be further along the diabetes disease progression pathway and to require multiple drugs to maintain glycemic control. In such patients, it may have been extremely difficult to find an effective product to replace the RCP. Access to data on the use of other anti-diabetic drugs may have provided greater insight into whether the risk communication impacted therapy.
There was no evidence of an increased rate of adverse events following the publication of the Health Canada risk communication, which might have been anticipated since patients potentially at risk may have been the ones who discontinued RCP therapy. In fact, there was an apparent decrease in the rate of all AEs and serious AEs. The rate of cardiac AEs was already low so that a similar decline was not likely to occur. Despite the comprehensiveness of the GSK AE database, the estimated rates of adverse events are impacted by small numbers, potentially due to under-reporting, and by the relatively crude but best available method used to calculate them, which is a significant limitation of this analysis. The lack of information on each patient’s RCP treatment regimen in the AE data was a further limitation.
The findings of this analysis demonstrate a major decrease in RCP use in Canada following the meta-analysis publication, which has been maintained. It is not possible to disentangle whether the continuing decline can be attributed to the meta-analysis, the changes in prescribing guidelines, media attention or a combination of some or all of these factors.
NSBR was an epidemiologist and JART is Director, Specialty Care in Medical Affairs, GlaxoSmithKline Inc., Mississauga, Ontario, Canada. NSBR left GlaxoSmithKline Inc. in March 2012 and is now President, Eastlake Research Group, Oakville, Ontario, Canada.
We thank the staff of IMS Brogan for extracting and providing the data for this analysis. The views expressed are those of the authors and do not necessarily represent those of GlaxoSmithKline Inc., IMS Brogan, or any other organization. Funding for the data purchase was provided by GlaxoSmithKline Inc.
- Nissen SE, Wolski K: Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med. 2007, 356: 2457-2471. 10.1056/NEJMoa072761.PubMedView ArticleGoogle Scholar
- Stewart KA, Natzke BM, Williams T, Granger E, Casscells SW, Croghan TW: Temporal trends in anti-diabetes drug use in tricare following safety warnings in 2007 about rosiglitazone. Pharmacoepidemiol Drug Saf. 2009, 18: 1048-1052. 10.1002/pds.1819.PubMedView ArticleGoogle Scholar
- Starner CI, Schafer JA, Heaton AH, Gleason PP: Rosiglitazone and pioglitazone utilization from January 2007 through May 2008 associated with five risk-warning events. J Manag Care Pharm. 2008, 14: 523-531.PubMedGoogle Scholar
- Cohen A, Rabbani A, Shah N, Alexander GC: Changes in glitazone use among office-based physicians in the US, 2003–2009. Diabetes Care. 2010, 33: 823-825. 10.2337/dc09-1834.PubMedPubMed CentralView ArticleGoogle Scholar
- Ruiter R, Visser LE, van Herk-Sukel MPP, Geelhoed-Duijvestijn PH, de Bie S, Straus SMJM, Mol PGM, Romio SA, Herings RMC, Stricker BHC: Prescribing of rosiglitazone and pioglitazone following safety signals: analysis of trends in dispensing patterns in the Netherlands from 1998 to 2008. Drug Saf. 2012, 35: 471-480. 10.2165/11596950-000000000-00000.PubMedView ArticleGoogle Scholar
- Shah BR, Juurlink DN, Austin PC, Mamdani MM: New use of rosiglitazone decreased following publication of a meta-analysis suggesting harm. Diab Med. 2008, 25: 871-874. 10.1111/j.1464-5491.2008.02462.x.View ArticleGoogle Scholar
- Morrow RL, Carney G, Wright JM, Bassett K, Sutherland J, Dormuth CR: Impact of rosiglitazone meta-analysis on use of glucose-lowering medications. Open Med. 2010, 4: E50-PubMedPubMed CentralGoogle Scholar
- Health Canada endorsed important safety information on rosiglitazone. 2007, Ottawa: Health Canada, http://www.hc-sc.gc.ca/dhp-mps/medeff/advisories-avis/prof/_2007/avandia_hpc-cps_5-eng.php (accessed: 18 Dec 2012)
- IMS Brogan. Longitudinal (LRx) insights (part of Lifelink). 2012, Mississauga, ON: IMS Health Inc, http://imsbrogansolutions.com/main.php?t=services&i=15 (accessed: 18 Dec 2012)
- Morgan SG, Raymond C, Mooney D, Martin D: The Canadian Rx Atlas. 2008, Vancouver: Centre for Health Services and Policy Research, 2Google Scholar
- Hammond IW, Gibbs TG, Seifert HA, Rich DS: Database size and power to detect safety signals in pharmacovigilance. Expert Opin Drug Saf. 2007, 6: 713-721. 10.1517/14740318.104.22.1683.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.