Alberto Mangano

No adjustment vs. adjustment formula as input weight for propofol target-controlled infusion in morbidly obese patients.

Background and objective The purpose of this prospective, randomized, double-blind study was to determine the predictive performance of target-controlled infusions of propofol in morbidly obese patients using the ‘Marsh’ pharmacokinetic parameter set. Methods Twenty-four patients (ASA II or III, age 25–62 years, BMI 35.5–61.7) were randomly allocated to receive propofol target-controlled infusion based on a weight adjustment formula (group adjusted) or without adjustment [group total body weight (TBW)]. Anaesthesia was induced by a propofol-targeted concentration of 6 μg ml−1 that was subsequently adapted to maintain stable bispectral index values ranging between 40 and 50. Arterial blood samples were collected before the start of the infusion and every 15 min thereafter to determine the predictive performances. Results There were no statistically significant differences between the groups with regard to performance errors, divergence and wobble. Results are presented as median (interquartiles). Median performance error and median absolute performance error were −31.7 (−35.9, −19.4) and 31.7% (20.2, 35.9) for group adjusted and −16.3 (−26.3, 2.2) and 20.6% (14.8, 26.9) for group TBW, respectively. Wobble median value was 7.4% (3.8, 8.4) for group adjusted and 8.2% (7.0, 9.6) for group TBW. As for wobble and divergence, no statistically significant differences were found between groups. Conclusion Weight adjustment causes a clinically unacceptable performance bias, which is not corrected when TBW is used as an input to the ‘Marsh’ model. It is, therefore, advisable to administer propofol to morbidly obese patients by titration to targeted processed-EEG values.

Subtotal scalp reconstruction after traumatic avulsion: a technical note.

We present the case of a 19-year-old female who suffered traumatic avulsion of the entire scalp while working with textile industrial machinery and present a pertinent review of the anatomy of the scalp.

Beneficial Effects of Humidified, Warmed Carbon Dioxide Insufflation During Laparoscopic Bariatric Surgery: A Randomized Clinical Trial. What if Sample Size Calculation Made Difference?

Sir: We read with interest the article published by Dr. Savel and colleagues [1]. While the authors are to be commended for their aim to conduct a prospective, randomized double-blind trial about the effect of warm and humidified carbon dioxide insufflation on postoperative pain and morphine requirements, there seems to be some issues with the study design. In particular, even though this trial appears to be adequately designed and analyzed, it does not seem to be adequately powered. In fact, in order for a particular finding to be claimed as significant (or not), the study should be adequately powered. The authors state their primary endpoints were visual analogic pain scores (VAS) and morphine requirements. They refer to an older article [2] but, unfortunately, seem to have failed sample size calculation. Nowhere in this reference article is the variability (SD) in morphine requirement at 24 h 10 mg. Considering a more reasonable variability (SD) of 30 mg and considering a clinically significant difference of 10 mg, a total of 193 patients per group would have been required (considering the “standard” α error of 0.05 and a power of 0.8) in this study. As for VAS, the authors did not perform any sample size calculation, and therefore, it is almost impossible to perform a pre-study sample size calculation. However, assuming a mean VAS value of 2.5±1.8 at 24 h [3] and considering a reduction of 1.8 clinically significant, more than 30 patients per group would have been required. It is also possible to perform a post hoc sample size calculation or power analysis on the statistical tests they have used from their data. In particular, considering a mean VAS value of 3.8±1.7 in the control group (Table 2), and a difference of 1.3, their test had a power of only 0.52 to detect such difference (considering an α error of 0.05). Appropriate sample size calculation and power is mandatory to draw adequate conclusions. In this case, the authors state no differences in opioid requirements and postoperative pain exists between the 2 groups. Are we sure it is really so? We hope our observations are useful to other authors planning surgical trials. “Editor’s Note: “Dr. Savel was invited to respond to this Letter-to-the Editor and declined.”

Continuous Infusion of Intraperitoneal Bupivacaine After Laparoscopic Surgery: A Randomized Controlled Trial—What about Statistical Power and Analysis?

Sir: We read with interest the article published by Dr. Sherwinter and colleagues [1]. While the authors are to be commended for their aim to conduct a prospective, randomized, double-blind trial about the effect of intraperitoneal local anesthetic infusion on postoperative pain and opioid requirements in obesity surgery, there seems to be some issues with both the design of the study and the statistical analysis. As for study design, the major issue is that this study does not seem to be adequately powered. In fact, in order for a particular finding to be claimed as significant (or not), the study must have enough power. The authors state that the aim of this study is to evaluate visual analogic pain scores (VAS) and morphine requirements after intraperitoneal infusion of bupivacaine. As for morphine, they refer to a study which evaluates the effect of warmed and humidified carbon dioxide insufflation in laparoscopic surgery [2]. Assuming a standard deviation of 25 (and not 10 as the authors wrongly assume), more than 130 patients per group would have been required (considering the “standard” α error of 0.05 and a power of 0.8) to detect an 11-mg difference in morphine requirement at 48 h. Moreover, it is interesting that the authors choose a study where patient-controlled analgesia was provided via an infusion pump, whereas they decided to perform a nurse-based administration of morphine. Maybe they could have chosen another study for sample size calculation? As for VAS, the authors did not perform any sample size calculation, and therefore, it is almost impossible to perform a pre-study sample size calculation. However, it is possible to perform a post hoc sample size calculation or power analysis on the statistical tests they have used from their data. In particular, considering a mean VAS value of 3.5±2.4 in the control group and a difference of 1.7, their test had a power of only 0.47 to detect such difference (considering an α error of 0.05). Another critical point is the statistical analysis. Despite its existence for many years, the importance of nonlinear mixed-effects modeling in pain studies has been recently reasserted [3]. Nowhere is the importance of a method which accounts for both interindividual variability (e.g., unique sensitivity to pain) able to characterize the time course of pain in different patients and intraindividual variability (various sources of “noise”) as critical as in pain OBES SURG (2009) 19:817–818 DOI 10.1007/s11695-009-9826-6

Clinical investigation of the novel iron‐chelating agent, CP94, to enhance topical photodynamic therapy of nodular basal cell carcinoma: when statistics make a difference

licular ostia. The diagnosis of cicatricial alopecia was suggested by scalp dermoscopy that showed diminished hair follicle density with loss of follicular ostia and absence of signs of inflammation or traction. The absence of yellow dots or perifollicular keratosis and erythema excluded alopecia areata and frontal fibrosing alopecia. Our series confirms Goldberg’s opinion that CMA can occur in the absence of traction and points out the importance of dermoscopy in the diagnosis of hair disorders. This noninvasive technique allows immediate differential diagnosis between CMA and other diseases affecting the scalp margin, such as alopecia areata and frontal fibrosing alopecia.

A Double-Blind, Randomized, Placebo-Controlled, Two-Dose Comparative Study of Botulinum Toxin Type A for Treating Glabellar Lines in Japanese Subjects: What If Sample Size and Statistical Tests Mattered?

We read with interest the article published by Harii and Kawashima [1]. While the authors are to be commended for investigating the effect of different doses of botulinum toxin type A (BTX-A) in Japanese subjects, there are several issues with both the design of the study and the statistical analysis used. With respect to the study design, despite the statement from the authors that it was a prospective, randomized, double-blind, placebo-controlled study, it does not appear to be so. There was no hypothesis to test; it appears to be a pilot study. Even if the authors wanted to test the hypothesis that halving the dose of BTXA (from 20 to 10 U) would not result in a significant decrease in a patient’s response rate, the sample size calculation is still missing. In fact, in order for a particular finding to be claimed as significant (or not), the study must have enough power. The authors do not state anything about the expected change in response rate values and therefore it is almost impossible to perform a prestudy sample size calculation. However, it is possible to perform a post-hoc sample size calculation or power analysis on the statistical tests they have used from their data. In particular, considering a rough response rate of 90% in the 20-U group, a 10% clinically significant difference in response rate, and the Bonferroni’s correction for multiple comparisons, more than 160 patients per group would have been required (considering the ‘‘standard’’ a error of 0.05 and a power of 0.8). Moreover, the randomization method is not clearly explained. Another critical point is the statistical analysis. The authors state that the t test and ANOVA were used for continuous variables. This may not be the correct way to go. Are all the variables normally distributed? If not, nonparametric tests should be used. Moreover, the authors state they used a nonparametric test on categorical variables. Why? Could a categorical variable be non-normally distributed? While this study is interesting for plastic and cosmetic surgeons and dermatologists, appropriate statistical analysis of data from trials is fundamental to draw adequate conclusions. We recommend that all authors perform the right sample size calculation before beginning any type of prospective, randomized, controlled trial and use the correct statistical tests afterward. This is the only way to convince readers that what they claim to be significant is really so. We hope our suggestions will be useful to other authors who will be doing similar studies in the future.

Dipyrone Increases the Blood Flow of Arterial Dorsal Skin Flaps: Are Prospective Studies Always Properly Designed?

We read with interest the article published by Gulmez et al. [1]. While the authors are to be commended for their goal of conducting a prospective, randomized, double-blind trial about the effect of dipyrone on blood flow and necrosis (i.e., flap survival), there are several issues with respect to the design of the study and the statistical analysis used. With respect to study design, the major issue is that there is no reference to any sample size calculation. In fact, in order for a particular finding to be claimed as significant (or not), the study must have enough power. In this particular case, the expected change in blood flow should be stated and from that the sample size should be calculated (e.g., considering the ‘‘standard’’ a error of 0.05 and a power of 0.8). Moreover, there is no reference to the randomization method. Another critical point we want to bring up is the statistical analysis used. The authors state that ‘‘Student’s ttest at a p level of significance less than 0.05 was used to compare the percentage of necrotic area and blood flow changes from baseline for each time point measure.’’ This may not be the correct way to go. Are all the variables normally distributed? If not, nonparametric tests should be used. Moreover, this study involves both tests between groups and tests within the same individuals (i.e., repeated measurements on blood flow and fraction of necrosis). Therefore, close attention should be paid to whether use of a t test for unpaired or paired data is appropriate and, similarly, to the use of either a Wilcoxon rank-sum test or signed-rank test. While this study is extremely interesting for plastic surgeons, appropriate statistical analysis of data from trials is crucial for adequate conclusions. In addition, we have recently shown that the anesthetic technique used affects regional blood flow in spine surgery [2]. Because most patients undergoing plastic surgery procedures are under general anesthesia, it would be interesting to investigate the effect of different anesthetic techniques on flap vascularization and viability. We hope our suggestions will be useful to other authors who will be involved in similar studies in the future.