Murad Alam1, David A Barzilai, David A Wrone. 1. Section of Cutaneous and Aesthetic Surgery, Department of Dermatology, Northwestern University, Chicago, Illinois 60611, USA. murad@alam.com
Abstract
BACKGROUND: Many new devices and therapeutic interventions are continually introduced in cutaneous surgery. The efficacy of these new techniques must be compared with that of preexisting standards so that patients can be appropriately counseled. OBJECTIVES: The purpose of this article is to (1) review methods for estimating sample size and power, (2) estimate the range of sample sizes sufficient to ensure that true differences are not missed in clinical trials of new procedural dermatologic therapies, and (3) consider the reasons why the sample size may be too small in procedural dermatology trials and how this problem can be addressed. METHODS: (1) Selective review of textbooks and other relevant literature, presentation of a brief tutorial describing sample size and power determination for therapeutic clinical trials comparing two groups with continuous outcomes variables; (2) implementation of standard formulae and assumptions to estimate sample size in cutaneous surgery therapeutic trials. RESULTS: Assuming that one group receives a standard surgical intervention and another group undergoes a new technique, to identify a moderate difference in efficacy between groups, at least 50 to 200 subjects will need to be enrolled if conventional strategies are used to reduce the likelihood of finding a difference that does not really exist (Type I error), as well as the likelihood of missing a true difference (Type II error). CONCLUSION: By face validity, it is apparent that most efficacy comparisons in procedural dermatology have low sample size and a concomitant risk of failing to detect actual differences between therapeutic arms. Owing to the limitations that restrict surgeons from frequently performing large randomized controlled trials in procedural dermatology, meta-analyses may be needed to pool the results of smaller studies. When it is critically important that differences between groups be accurately identified, dermatologic surgeons may consider eschewing smaller trials in favor of collaborating on larger trials with an adequate sample size.
BACKGROUND: Many new devices and therapeutic interventions are continually introduced in cutaneous surgery. The efficacy of these new techniques must be compared with that of preexisting standards so that patients can be appropriately counseled. OBJECTIVES: The purpose of this article is to (1) review methods for estimating sample size and power, (2) estimate the range of sample sizes sufficient to ensure that true differences are not missed in clinical trials of new procedural dermatologic therapies, and (3) consider the reasons why the sample size may be too small in procedural dermatology trials and how this problem can be addressed. METHODS: (1) Selective review of textbooks and other relevant literature, presentation of a brief tutorial describing sample size and power determination for therapeutic clinical trials comparing two groups with continuous outcomes variables; (2) implementation of standard formulae and assumptions to estimate sample size in cutaneous surgery therapeutic trials. RESULTS: Assuming that one group receives a standard surgical intervention and another group undergoes a new technique, to identify a moderate difference in efficacy between groups, at least 50 to 200 subjects will need to be enrolled if conventional strategies are used to reduce the likelihood of finding a difference that does not really exist (Type I error), as well as the likelihood of missing a true difference (Type II error). CONCLUSION: By face validity, it is apparent that most efficacy comparisons in procedural dermatology have low sample size and a concomitant risk of failing to detect actual differences between therapeutic arms. Owing to the limitations that restrict surgeons from frequently performing large randomized controlled trials in procedural dermatology, meta-analyses may be needed to pool the results of smaller studies. When it is critically important that differences between groups be accurately identified, dermatologic surgeons may consider eschewing smaller trials in favor of collaborating on larger trials with an adequate sample size.