Letter by Antonopoulos et al Regarding Article, "Carotid Artery Stenting Versus Carotid Endarterectomy: A Comprehensive Meta-Analysis of Short-Term and Long-Term Outcomes"
Stroke welcomes Letters to the Editor and will publish them, if suitable, as space permits. Letters must reference a Stroke published-ahead-of-print article or an article printed within the past 3 weeks. The maximum length is 750 words including no more than 5 references and 3 authors. Please submit letters typed double-spaced. Letters may be shortened or edited. Include a completed copyright transfer agreement form (available online at The conclusions of the meta-analysis recently published in
... cently published in your distinguished journal by Economopoulos et al 1 regarding the comparison of carotid artery stenting (CAS) and carotid endarterectomy (CEA) seem fairly robust. The authors reported a thoughtfully designed and executed meta-analysis, one that synthesizes the results from the literature and provides useful clinical information in a debatable field. Interestingly enough, the authors have found that both short-and long-term stroke and death or stroke were more frequently observed in CAS, compared with CEA, whereas the opposite was true for short-term myocardial infarction and cranial nerve injury. To expand the clinical implications of this meta-analysis, we have calculated a clinically and epidemiologically meaningful index that is commonly used in assessing the effectiveness of a healthcare intervention and communicating results among studies: number needed to treat/number needed to harm (NNT/NNH). Data extraction from eligible studies, as presented in Supplemental Table VI of the original article, 1 included study name, number of patients, and number of events for short-term and long-term events. Event rates for CAS and CEA among all studied outcomes were calculated, and the NNT/NNH with 95% CI was estimated post hoc, 2 based on statistically significant odds ratios, using "metannt" command (STATA Corp.). The NNT was equal to 1/(CEA group event rateϪCAS group event rate). In case of a negative group event rate difference (CEAϪ CAS), we presented NNH, as the CAS intervention was harmful. In our case, NNT indicated the number of patients that would have to undergo CAS to prevent 1 morbidity-or mortality-related event. Our analysis based on the data by Economopoulos et al, 1 suggested that myocardial infarction (NNTϭ112; 95% CI, 83-270) and cranial nerve injury (NNTϭ20; 95% CI, 19-22) were in favor of CAS, whereas stroke (NNHϭ51; 95% CI, 30-116) and death or stroke (NNHϭ45; 95% CI, 28-95) were in favor of CEA, among short-term outcomes. With respect to long-term outcomes, stroke (NNHϭ37; 95% CI, 22-101) and death or stroke (NNHϭ35; 95% CI, 19-156) were in favor of CEA (Table) . Although many epidemiological indices have been used for assessing a treatment effect, it seems more meaningful to use NNTs/NNHs for clinical decision making. 3 NNTs/NNHs summarize the effect of treatment in terms of the number of patients a clinician needs to treat/harm with a particular therapy to expect to prevent/cause 1 adverse event. 3-5 These measures improve the reporting of effect estimates and enable better interpretation of the results, as they facilitate interpretation in terms of patients treated rather than the arguably less-intuitive probabilities. 5 In addition, reporting of NNTs/NNHs gives a comprehensive way of understanding how much effort is needed to prevent/cause 1 event. 3 As a consequence, they depict the amount of clinical effort needed to prevent the same event in patients with other disorders or the same event among patients being treated with other comparable measures. On the contrary, concerns have been raised whether NNTs/NNHs associated with an intervention are different for periods of time longer than that studied in the original studies. 4 An additional issue concerns whether these indices estimated with patients with 1 baseline risk are consistent if applied to a patient with a different risk. 3, 4 In conclusion, although potential limitations might exist, NNT/NNH is a clinically useful measure of treatment efficacy and a comprehensive way of communicating results that should be estimated when performing a meta-analysis.