The Need to Recognize the Difference Between a Quality Register and a Randomized Controlled Trial
To the Editor:
In the pursuit of new data on thrombolysis with recombinant tissue plasminogen activator (rtPA) in patients with acute ischemic stroke, clarity is needed on what information can be obtained from different methods; ie, observational methods, like a quality register, and experimental methods, like a randomized controlled trial (RCT).
Apart from treatment in a stroke unit that benefits all stroke patients, thrombolytic therapy for acute ischemic stroke is at present the only medical treatment available, and is by far the most promising.1 To date 2955 patients have been randomized into trials with intravenous rtPA.2 Systematic reviews of the trials showed that, despite the hazards of intracranial hemorrhage and early death, there is scope for benefit from thrombolysis up to 6 hours.2,3⇓ Based on the first positive thrombolysis trials in 1995,2,3⇓ treatment with rtPA has been licensed in USA and Canada for use within 3 hours of stroke onset for a selected group of patients. Since 2002 there is a provisional license in the European Union for treatment within 3 hours in even more selected patients, 80 years of age or younger.4 The licensing will be renewed after 3 years if (1) a further randomized trial (ECASS 3; in patients at 3 to 4 hours after onset) is performed by the manufacturer; and (2) safety (as recorded in the quality register SITS-MOST) is satisfactory among patients treated within the license. These conditions highlight the need to appreciate that an RCT must have adequate statistical power to give reliable answers3 and whether a treatment register can give valid answers to safety issues.5
The difference between a quality register and an RCT seems important to elucidate. Quality registers are based on registration of data on patients treated in ordinary clinical practice according to criteria derived from completed randomized trials. A quality register gives the means to identify rare side effects and to monitor the uptake of a new treatment. In theory, it may be of interest to compare findings in a quality register to findings in earlier trials (historical controls). However, such comparisons have very limited scientific value because the lack of randomization will inevitably introduce bias.5 By necessity there will be a long delay between the trials and the analysis of the quality register. Hence, there is no way of adjusting for case mix or for the many immeasurable changes in treatment that may have occurred over time. For example, today a much larger proportion of patients are already on antiplatelet treatment, and refinement of diagnostic tools may influence the type of patients who are entered in the register. An example of the effect of changes over time is the difference between the placebo-treated patients entered in ECASS I2,3⇓ and ECASS II2,3⇓ where, in the latter trial, the course of disease was much better despite similar inclusion criteria.
Despite the positive effect of thrombolysis shown in the systematic reviews, there are prevailing, definite, and important uncertainties (Table). We believe that these questions can only be answered validly and reliably by large RCTs, and not by observational studies and quality registers.
Sandercock P, Berge E, Dennis M, Hand P, Kwan J, Lewis S, Lindley R, Neilson A, Thomas B, Wardlaw J. A systematic review of the effectiveness, cost-effectiveness and barriers to the implementation of thrombolytic and neuroprotective treatment in the NHS. Health Technol Assess. 2002; 6: 1–112.
Wardlaw JM, del Zoppo G, Yamaguchi T, Berge E. Thrombolysis for acute ischaemic stroke (Cochrane Review). In: The Cochrane Library, Issue 3, 2003. Oxford, UK: Update Software.
Deeks JJ, Dinnes J, D’Amico R, Sowden AJ, Sakarovitch C, Song F, Petticrew M, Altman DG; International Stroke Trial Collaborative Group; European Carotid Surgery Trial Collaborative Group. Evaluating nonrandomised intervention studies. Health Technol Assess. 2003;7:iii–x, 1–173. http://www.ncchta.org