reply: additional complications of corneal crosslinking
TRANSCRIPT
186 LETTERS
REPLY: We appreciate having another chance topoint out that corneal CXL is a clinical procedure withcomplications and failures, and we are grateful thatKymionis et al. drew attention tomore infrequent com-plications that were not mentioned in our discussion.
We fully agree that an impact on the cornea suchas CXL may initiate a recurrence of herpes keratitisand in caseswith an ocular herpes history, systemic an-timetabolite therapy is clearly indicated. Stimulation ofan inflammatory reaction such as diffuse lamellar ker-atitis after LASIK by CXL is also not a surprising sideeffect, and we agree that topical steroids should pre-cede CXL in cases of iatrogenic keratectasia.
Please keep in mind: CXL is a technically easy pro-cedure but represents a significant impact on the eyetreated.dTobias Koller, MD, Theo Seiler, MD, PhD
Left-handed residentsKim et al.1 recently published their finding of a sig-
nificantly lower incidence of posterior capsule tearand vitreous loss in cataract surgery performed byleft-handed residents. Statistical analysis was basedon the number of cases performed by left-handed sur-geons versus those performed by right-handed sur-geons (170 versus 1560). However, the number ofleft-handed surgeons in the studywas extremely small(3 left-handed compared with 33 right-handed sur-geons), which severely limits any conclusions thatcan be drawn. The authors state that ‘‘the number ofcaseswasmore indicative than the number of residentsbecause learning phacoemulsification cataract surgeryis unique in several ways, including having ‘‘a verysteep learning curve.’’ This argument is spurious.
The small number of left-handed surgeons studiedintroduces abias that cannot be overcomebyhaving in-dividuals perform more cases, not when the real ques-tion studied is whether left-handed residents have aninherent advantage in learning cataract surgery.
Articles cited by the authors regarding possiblesuperiority of left-handed cataract surgeons wereanecdotal2 or had a high nonresponse rate on the ques-tionnaire that precluded multivariate analysis.3 Cau-tion should be applied when one wishes to identifycharacteristics to be used ‘‘as potential screening toolsfor entry into residency programs.’’1 Generalizationsshould not be made on the basis of the study of 3left-handed resident surgeons.
Preston H. Blomquist, MDDallas, Texas
REFERENCES1. Kim JY, Ali R, Cremers SL, Yun S-C, Henderson BA. Incidence of
intraoperative complications in cataract surgery performed by
J CATARACT REFRACT SURG -
left-handed residents. J Cataract Refract Surg 2009; 35:1019–
1025
2. Freeman MJ, Singh J, Chell P, Barber K. Modular phakoemulsifi-
cation training adapted for a left-handed trainee. Eye 2004; 18:35–
37. Available at: http://www.nature.com/eye/journal/v18/n1/pdf/
6700528a.pdf. Accessed October 5, 2009
3. Mahmood S, von Lany H, Cole MD, Charles SJ, James CRH,
Foot B, Gouws P, Shaw S. Displacement of nuclear fragments
into the vitreous complicating phacoemulsification surgery in the
UK: incidence and risk factors. Br J Ophthalmol 2008; 92:488–
492
REPLY: We agreewith Blomquist that the number ofleft-handed surgeons in the study was small and thisdoes limit the conclusions that can be drawn. How-ever, the incidence of left-handedness among theAmerican population (11%) is similar to the percent-age of left-handed surgeons in this study (9.0%) andthe percentage of surgeries performed by left-handedsurgeons (9.8%). Because of the small number of left-handed residents, a specialized statistical model wasused instead of a conventional multivariate logisticmodel. Based on the input from the reviewers, we ex-amined the possibility of analytical problems com-pounding unwanted biases because of sparse databeing divided into 2 groups, one of which contained90% of the data and the other only 10% of the data. An-alytical tests such as Bayesian and hierarchical regres-sion methods can be used in such instances of sparsedata sets. After consideration of the various methods,we alternatively applied a multivariate logistic gener-alized estimating equation model with robust stan-dard errors that accounted for the clustering betweensubjects in the same resident, which was more suitablefor our study design and the characteristics of our datathan Bayesian and hierarchical regression methods.
We also agree with Blomquist that it is important toweigh many factors when assessing potential candi-dates for entry into residency programs and do notcondone using handedness as the important factor.The goal of this report was to highlight the interestingfinding of lower complications rates in our left-handedresidents. While this finding is based on small but sig-nificant numbers, we recommend additional prospec-tive larger studies to further evaluate this relationshipof handedness and complication rates.dJae Yong Kim,MD, PhD, Bonnie An Henderson, MD
Safety and efficacy of transepithelialcrosslinking (C3-R/CXL)
Baiocchi et al.1 measured stromal riboflavin usinga 0.1% riboflavin–20% dextran solution and concludedthere are concerns about safety and efficacy of transepi-thelial crosslinking. Their studyhas several limitations:
VOL 36, JANUARY 2009