j.n.katz

8/17/2019 j.n.katz

1/10

The PDF of the article you requested follows this cover page.

This is an enhanced PDF from The Journal of Bone and Joint Surgery

2007;89:2548-2550. J Bone Joint Surg Am.Jeffrey N. Katz and Elena Losina

J.N. Katz and E. Losina comment on the above letters:

This information is current as of November 6, 2007

Reprints and Permissions

Permissions] link.and click on the [Reprints and jbjs.orgarticle, or locate the article citation on

to use material from thisorder reprints or reque st permi ssionClick here to

Publisher Information

www.jbjs.org20 Pickering S treet, Needham, MA 02492-3157The Journal of Bone and Joint Surgery

http://www.jbjs.org/https://s100.copyright.com/AppDispatchServlet?PublisherName=JBJS&Publication=JBJS&Title=J.N.+Katz+and+E.+Losina+comment+on+the+above+letters%3A&PublicationDate=11/01/2007&Author=Jeffrey+N.+Katz&StartPage=2548&ContentID=89%2F11%2F2548-a&OrderBeanReset=truehttp://www.jbjs.org/https://s100.copyright.com/AppDispatchServlet?PublisherName=JBJS&Publication=JBJS&Title=J.N.+Katz+and+E.+Losina+comment+on+the+above+letters%3A&PublicationDate=11/01/2007&Author=Jeffrey+N.+Katz&StartPage=2548&ContentID=89%2F11%2F2548-a&OrderBeanReset=truehttp://www.jbjs.org/http://www.jbjs.org/http://www.jbjs.org/http://www.jbjs.org/http://www.jbjs.org/https://s100.copyright.com/AppDispatchServlet?PublisherName=JBJS&Publication=JBJS&Title=J.N.+Katz+and+E.+Losina+comment+on+the+above+letters%3A&PublicationDate=11/01/2007&Author=Jeffrey+N.+Katz&StartPage=2548&ContentID=89%2F11%2F2548-a&OrderBeanReset=true

8/17/2019 j.n.katz

2/10

Letters to The Editor

Calcaneal Osteomyelitis Causedby Exophiala jeanselmei in an Immunocompetent ChildTo The Editor:In reference to our case report entitled “Cal-caneal Osteomyelitis Caused by Exophiala jeanselmei in an Immunocompetent Child.A Case Report” (2007;89:859-62), my coau-thors and I would like to bring to the noticeof the readers of The Journal that the samecase report has been published by one of usin the Indian Journal of Medical Microbiology as an article entitled “Eumycetoma PedisDue to Exophiala jeanselmei.”1

The publication in the Indian Journalof Medical Microbiology was meant to high-light the microbiological aspects of the dis-ease as the said fungus is extremely rare. Theauthors regret any confusion this might havecaused to the readers of both articles.

Shah A. Khan, MS, MRCS(Ed)

Department of Orthopaedics, All IndiaInstitute of Medical Sciences, AnsariNagar, New Delhi 110 029, India, e-mail:[email protected] letter originally appeared, in slightly different form, on

jbjs.org. It is still available on the web site in conjunctionwith the article to which it referred.

Reference1. Capoor MR, Khanna G, Nair D, Hasan A, Rajni,Deb M, Agg arwal P. Eumycetoma pedis due t oExophiala jeanselmei. Indian J Med Microbiol.2007;25:155-7.

Navigated Total Knee ReplacementTo The Editor:We read with interest and concern the arti-cle, “Navigated Total Knee Replacement.A Meta-Analysis” (2007;89:261-9) by Bau-wens et al. We submitted a similar meta-analysis to The Journal of Bone and Joint Sur- gery over one year ago, which was appropri-ately rejected for publication because of theinclusion of data from abstracts and uncon-trolled case series. The reviewers and edi-tors also expressed concern that our findingof an advantage for navigated total knee ar-throplasty compared with conventional total

knee arthroplasty based on radiographicalignment end points needed to be balancedagainst the lack of evidence with regard todifferences in cost-effectiveness, complica-tion rates, and long-term outcomes be-tween the two procedures.

We were in the process of updatingour meta-analysis in light of more recentpublications (excluding data from abstractsand uncontrolled case series) when thestudy by Bauwens et al. was published.Having reviewed essentially the same data-base, we were perplexed by the authors’conclusion that “navigated knee replace-ment provides few advantages over conven-tional surgery on the basis of radiographicend points,” as our own meta-analysis re-vealed a significant improvement in radio-graphic end points with computer-assistednavigation.

Our concerns about the discrepanciesbetween our findings and those of Bauwenset al. prompted us to investigate their sourcedata. We contacted them, and they gra-ciously provided us with the raw data forall studies included in their meta-analysis.

On further review, we discovered multipleinaccuracies of data extraction and/or dataentry in their analysis.

In four of the studies 1-4 reviewed inthe article by Bauwens et al., the data forconventional techniques were entered intothe data set for navigated replacement foranalysis while the data for the navigated re-placements were entered into the data set forconventional techniques. We were also ableto determine errors of data extraction, dataentry, patient count, or patient group as-signment from four additional studies 5-8.One paper 9 was included and counted as

reporting mechanical axis data when thesedata were not reported in the study. A kin-ship study 10 (i.e., a study sharing overlapping

data with an already included study) was in-cluded when it should have been excluded.There were two additional studies 11,12 inwhich the numbers that we extracted wereslightly different from those in the report byBauwens et al.; we note these only as dis-crepancies (not errors) in extraction.

Our further review of their paper alsosuggested that their labeling and descrip-tion of results were misleading. Specifically,they describe their meta-analyses as thoseof “relative risk of malalignment” and labeltheir figures accordingly. In the Discussion,they state that “the available data suggestthat navigation reduces the relative risk of 3°of malalignment by 25%.” This statementis in error because their meta-analysis wasnot of the relative risk of malalignment, butrather the relative risk of alignment (i.e., thechance that a patient has alignment after theprocedure). It would, therefore, have beenaccurate for them to state that conventionaltotal knee arthroplasty decreases the relativechance of alignment by 25%. When misfit,instead of fit, is the outcome of choice, theresults are quite different from those re-

ported by Bauwens et al. Correctly stated,the risk of malalignment with conventionalreplacement is appropriately three timesthat with computer-assisted surgery.

In conclusion, our findings of dataextraction and entry errors cause us to chal-lenge the conclusions in the article regardingthe meta-analysis of radiographic endpoints following conventional comparedwith navigated knee replacement surgery.A correct data analysis demonstrates over-whelming evidence of a much lower errorrate with navigation. Reversal of some of theextracted data and misreporting of relative

risks for fit as risks of malalignment are par-tially responsible for the muted differencethat Bauwens et al. described between navi-

J Bone Joint Surg Am. 2007;89:2547-55

LETTERS TO THE EDITOR MUST BE SUBMITTED ELECTRONICALLY;INSTRUCTIONS ARE AT WWW .JBJS.ORG/LETTERS

8/17/2019 j.n.katz

3/10

T H E JO U R N A L OF B ON E & JOINT SU R G E RY · J B J S.OR GVOLUME 89-A · N U M B E R 11 · N OVEMBER 2007

LE T T E R S T O T H E E DITOR

gated and conventional total knee arthro-plasty. These errors, however, do not obviatetheir other discussion points regarding themethodological limits of the available tr ials,including a dearth of evidence on long-term

outcomes, quality of life, and costs.While we recognize and understand

the challenges inherent in performing meta-analyses, our intent is to bring these errorsto the attention of the readers of The Journal to correct any erroneous impression thatthis work may have left with the readership.

J. Bohannon Mason, MDThomas Fehring, MD

Kyle Fahrbach, PhD

Corresponding author: J. Bohannon Mason,MD, OrthoCarolina Hip and Knee Center,1915 Randolph Road, Charlotte, NC 28207,e-mail: [email protected]

Disclosure: In support of their research for orpreparation of this work, one or more of theauthors received, in any one year, outsidefunding or grants in excess of $10,000 fromDePuy, and Johnson and Johnson, Warsaw, In-diana. Neither they nor a member of their im-mediate families received payments or otherbenefits or a commitment or agreement toprovide such benefits from a commercial en-tity. No commercial entity paid or directed, oragreed to pay or direct, any benefits to any re-search fund, foundation, division, center, clini-cal practice, or other charitable or nonprofitorganization with which the authors, or amember of their immediate families, are affili-ated or associated.

D. Stengel, K. Bauwens, G. Matthes, M. Wich, F. Gebhard, B. Hanson,and A. Ekkernkamp reply:We read with great interest the letter fromDr. Mason and colleagues. Since they raisedsubstantial concerns about the validity ofour findings, we carefully reviewed the dataset that formed the basis for all analyses andfigures presented in The Journal .

We reviewed the references cited byMason et al. 1-4 and found no data shift betweenthe conventional and navigated-surgerygroups. Such a shift was unlikely since theforest plots consistently showed an advan-tage for the navigated-surgery cohort.

Mason et al. also claimed that theyfound additional errors of data extractionfrom four other studies that we reviewed 5-8,but unless they are more specific in theircriticisms, we cannot respond properly.

We would refer Mason et al. to the

Materials and Methods section of our pa-per, where we stressed that the numbers ofpatients were extracted from histogramswhenever possible. This may explain mostof the differences that they noted between

their and our data sets. Additional differ-ences might be related to different handlingof the unit of interest—that is, the patient orthe knee. Bolognesi and Hofmann 9 did in-deed report the alignment of the femoraland the tibial component rather than themechanical axis. However, if navigation im-proves both femoral and tibial componentalignment, it is very likely that the resultingmechanical axis will be optimized as well.Since the observed effects were consistentwith others, we decided to include thatstudy in our analysis. We definitely identi-fied and excluded some kinship studies, but

we could not retrieve a dual publication byMielke et al.10.

When posing a null hypothesis, it isimportant to define the accepted standard ofcare. Risk ratios and other relative measuresare asymmetric. This was the reason why wealso provided risk differences, which can beused for calculating the number needed totreat. Currently, navigation is an experimen-tal add-on and may either decrease the riskof malalignment or increase the chance ofalignment. It is, however, not justified to ar-gue that conventional surgery would in-crease the relative risk of malalignment overthat associated with navigated componentplacement. With regard to health-policy de-cisions, this is a dangerous statement since itwould imply that all patients who are notoperated on with computer assistance butundergo conventional total knee arthro-plasty by an experienced surgeon are at ahigher risk of having malalignment whencompared with those who undergo totalknee arthroplasty with navigated compo-nent placement.

Importantly, our analyses and plotsshowed a significant advantage of navigatedover conventional knee replacement interms of radiographic surrogates, so we arein complete agreement with Mason et al.Yet, unless these advantages are consistentwith improved outcomes, we think that ourconclusion “Navigated knee replacementprovides few advantages over conventionalsurgery on the basis of radiographic endpoints” is valid.

Finally, we regret that Mason et al.,after receiving our data set (the sending ofwhich shows our openness and willingness

to engage in scientific debate), did not con-tact us again to compare both data sets andto discuss, explore, and resolve any possibledifferences jointly before submitting a Letterto the Editor challenging our scientific repu-

tation. We are sorry that Dr. Mason’s groupcould not publish their paper, but we aredeeply disappointed in their behavior.

Dirk Stengel, MD, PhD, MSc Kai Bauwens, MD

Gerrit Matthes, MD Michael Wich, MD

Florian Gebhard, MD, PhDBeate Hanson, MD, MPH

Axel Ekkernkamp, MD, PhD

Corresponding author: Dirk Stengel, MD,PhD, MSc, Department of Trauma and Or-thopedic Surgery, Center for Clinical Re-search, Unfallkrankenhaus Berlin, WarenerStrasse 7, 12683 Berlin, Germany, e-mail:[email protected]

J.N. Katz and E. Losinacomment on the above letters:In their meta-analysis of the effectiveness ofnavigated total knee replacement, Bauwenset al. found that navigation was associatedwith favorable results in terms of several ra-diographic parameters. The data were insuf-ficient to evaluate effects on complicationrates or functional outcomes. The articlestimulated the above letter from Mason et

al. and a letter from Gregori and Holt 13,which prompted additional letters of clarifi-cation from Bauwens et al.

Caught in the crossfire, readers mightwell ask why a meta-analysis led to such edi-torial dueling. Of note, controversy overmeta-analysis is long-standing 14. The de-bates stem in part from the methodologicalcomplexity of meta-analysis, a powerful butchallenging analytic technique that permitspooling of estimates across studies. We willdiscuss a few of the many methodologicalcomplexities of meta-analysis to put the cor-respondence about navigated total knee re-placement in perspective.

Why Pool? Meta-Analysis Compared with Traditional Literature Review If pooling raises so many questions, whybother to pool estimates quantitativelyacross studies? In many reviews, the authorssimply array the findings of separate studiesin evidence tables without attempting tosynthesize them quantitatively into singleestimates of effect. A key rationale for pool-

8/17/2019 j.n.katz

4/10



ing is that the available evidence may consistof small studies that show positive (or nega-tive) effects but lack power to establish theassociations with significance. Pooling thesesmaller studies may avoid false-negative re-

sults due to Type-II error.A useful example of this application

of meta-analysis was provided by Felson andAnderson in a meta-analysis of the effect ofcytotoxic therapy and corticosteroids com-pared with that of corticosteroids alone forpatients with lupus nephritis 15. Prior smallstudies had suggested a beneficial effect ofcytotoxic therapy. The meta-analysis over-came the small sample sizes of the compo-nent studies and illustrated the beneficialeffect of cytotoxic therapy across studies.

Pooling also permits the investigatorto examine whether particular study charac-

teristics are associated with the principaloutcome. This technique is termed metaregression. The investigator develops a regres-sion model in which each study serves as asingle observation, contributing a single es-timate of outcome and of each covariate.The investigator can weight studies differen-tially in order to give greater importance inthe regression to those that have larger sam-ple sizes or that are of higher methodologi-cal quality. Metaregression can yield insightsabout sources of variability in outcomemeasures across studies. For example, it maybe that trial designs are associated withlarger effects and nonrandomized designs,with smaller effects, or vice versa.

Why Not Pool?Pooling the results of separate studies intosingle estimates of effect involves several as-sumptions that frequently are not satisfiedby the literature under review. Clearly, theoutcome variable must be consistent acrossstudies. This constraint poses no problemwhen the outcome is unambiguously de-fined, such as thirty-day all-cause mortalityfollowing hip replacement. However, whenstudies measure satisfaction, pain relief,functional status, and other such complexoutcome variables, the task becomes morecomplicated. These domains are often mea-sured with different tools in different stud-ies, or different cutoffs are used to definesuccess. For example, the authors of somestudies of the outcome of total knee replace-ment might use the WOMAC (WesternOntario and McMaster Universities Os-teoarthritis Index) as the principal outcomemeasure whereas others might use the SF-36

(Short Form-36) or the Knee Society Scale.Attempting to synthesize results in thesecircumstances involves essentially com-bining apples and oranges and is notadvisable. Standardization of outcome

assessment and reporting in specific fieldswould assist investigators who wish to per-form meta-analysis.

In addition, the underlying statisticalmethodology of meta-analysis assumes thateach of the studies to be synthesized repre-sents one observation from a single distribu-tion of studies. This assumption is validatedwith tests of homogeneity of the odds ratios(or other effect estimates) across studies. Ifthe group of studies to be synthesized ap-pears to emanate from a single distribution,the homogeneity criterion is met and thestudies may be synthesized in a meta-analysis.

If, on the other hand, the assumption ofhomogeneity is not met, and the studies ap-pear to be heterogeneous, then the investi-gators should be cautious about pooling.The investigators could simply choose not topool the studies quantitatively. Alternatively,the investigators might wish to perform ametaregression to identify sources of hetero-geneity. For example, it may be that higher-quality studies or a particular study design(e.g., trials) are associated with higher effectestimates.

What to Pool?A meta-analysis is essentially an observa-tional study of individual studies 16. As withall observational studies, the results are in-fluenced by the selection criteria that dic-tate which studies are included in the meta-analysis and which are excluded. An issuethat arises frequently, and was a major focusof contention about the paper by Bauwenset al., is whether to include unpublishedstudies. Excluding unpublished studies riskspublication bias, a form of selection bias inmeta-analyses that arises because positivestudies are, on the average, more likely tobe published than negative studies. How-ever, including unpublished studies thathave not passed peer review risks the in-clusion of studies with results that may notbe credible.

Another important decision iswhether to restrict the analysis to random-ized controlled trials or to include observa-tional designs. The advantage of restrictingthe analysis to randomized controlled trialsis that randomization greatly reduces therisk of selection bias in each component

study of the meta-analysis. Including obser-vational studies permits the meta-analysis tosimply propagate the biases inherent in thecomponent studies. The disadvantage ofrestricting the sample to randomized

controlled trials is that for many clinicalproblems, including navigated total kneereplacement, there are few randomizedcontrolled trials and most of the relevantliterature includes observational designs.

Returning to NavigatedTotal Knee ReplacementBauwens et al. handled most of the above-mentioned issues with sophistication. Theydecided to pool because they were concernedthat multiple underpowered studies wouldfail to establish an effect that might becomeapparent in a pooled analysis. They included

nonrandomized trials because they were notcomfortable restricting the analysis to ran-domized controlled trials. (An alternative ap-proach would be to use metaregression toexamine whether the magnitude of effect dif-fered between randomized and observationalstudies; if it did, the meta-analysis could bedone in subgroups.) The authors weighted thestudies according to sample size and quality.They used appropriate analytic techniques tolook for publication bias and, finding no evi-dence of such a bias, they restricted the analy-sis to published studies. In addition to statingthe results of these analyses of publication bias,displaying the graphical evidence would havebeen helpful to readers.

Bauwens et al. concluded that thestudies that they wished to synthesize wereheterogeneous. Having established heteroge-neity, the authors could have simply decidednot to pool the studies at all. Alternatively,they could have developed a metaregressionmodel, which would have been useful inidentifying and ultimately controlling forsources of heterogeneity. They could havestratified according to such characteristicsand tested whether the stratified meta-analysis would have yielded less heteroge-neity. The authors did indeed perform ametaregression, but they did not use it toidentify strata in which studies were morehomogeneous, as discussed here. By docu-menting heterogeneity and not doing any-thing about it, the authors in a sense made adiagnosis without offering a remedy.

Data Sharing Synthesizing the results of various studies isultimately a collaborative activity. The in-

8/17/2019 j.n.katz

5/10



vestigator will often wish to contact otherscientists who have access to original trialdata or who themselves have attempted adata synthesis. These collaborations canhelp move the field forward. In fact, the Na-

tional Institutes of Health (NIH) and otherresearch sponsors have developed specificprovisions for facilitating data sharing in or-der to best leverage the precious data gar-nered in NIH-funded studies. In this regard,we were particularly impressed by the will-ingness of Bauwens et al. to share their dataand we were disappointed that Mason et al.chose to communicate their observations ina letter to The Journal without discussing thefindings with the original authors. Readers,and ultimately patients, were not served wellby this failure to behave collaboratively.

Concluding RemarksThe meta-analysis by Bauwens et al.prompted questions about selection of stud-ies, choice of common outcome measuresacross studies, assessment and managementof heterogeneity, interpretation of results,and approaches to collaboration. The les-sons learned from these studies of navigatedtotal knee replacement are that investigatorsshould make individual studies as definitiveas possible by using the most rigorous de-signs feasible, powering studies adequately,and using standardized measures of out-come. Pooling is a powerful method for ag-gregating information across studies, but itis ultimately a collaborative effort. Leadersin the field should designate standard mea-sures of outcome to facilitate pooling, andinvestigators should work collaborativelywith one another so that data synthesesmove the field forward, bringing quality andvalue to patients.

Jeffrey N. Katz, MD, MSc Elena Losina, PhD

Corresponding author: Jeffrey N. Katz, MD,MSc, Orthopaedic and Arthritis Center forOutcomes Research, Brigham and Women’sHospital, 75 Francis Street, PBB-B3, Boston,MA 02115, e-mail: [email protected]

Disclosure: The authors did not receive anyoutside funding or grants in support of theirresearch for or preparation of this work. Nei-ther they nor a member of their immediatefamilies received payments or other benefits ora commitment or agreement to provide suchbenefits from a commercial entity. No com-mercial entity paid or directed, or agreed topay or direct, any benefits to any research fund,

foundation, division, center, clinical practice,or other charitable or nonprofit organizationwith which the authors, or a member of theirimmediate families, are affiliated or associated.

These letters originally appeared, in slightly different form,on jbjs.org. They are still available on the web site in conjunc-tion with the article to which they refer.

References1. Bäthis H, Perlick L, Tingart M, Lüring C, Zura-kowski D, Grifka J. Alignment in total knee ar thro-plasty. A comparison of computer-assisted surgerywith the conventional technique. J Bone Joint Surg Br.2004;86:682-7.

2. Perlick L, Bäthis H, Lerch K, Lüring C, Tingart M,Grifka J. [Navigated implantation of total knee en-doprostheses in secondary knee osteoarthritis ofrheumatoid arthritis patients as compared with con-ventional technique]. Z Rheumatol. 2004;63:140-6.German.

3. Saragaglia D, Picard F, Chaussard C, MontbarbonE, Leitner F, Cinquin P. [Comput er-assisted kneearthroplasty: comparison with a conventional proce-dure. Results of 50 cases in a prospective random-ized study]. Rev Chir Orthop Reparatrice Appar Mot.2001;87:18-28. French.

4. Sparmann M, Wolke B, Czupalla H, Banzer D, ZinkA. Positioning of total knee arthroplasty with and with-out navigation support. A prospective, randomisedstudy. J Bone Joint Surg Br. 2003;85:830-5.

5. Chauhan SK, Scott RG, Breidahl W, Beaver RJ.Computer-assisted knee ar throplasty versus aconventional jig-based technique. A randomised,prospective trial. J Bone Joint Surg Br. 2004;86:372-7.

6. Confalonieri N, Manzotti A, Pullen C, Ragone V.Computer-assisted technique versus intramedullaryand extramedullary alignment systems in total kneereplacement: a radiological comparison. Acta OrthopBelg. 2005;71:703-9.

7. Kim SJ, MacDonald M, Hernandez J, Wixson RL.Computer assisted navigation in total knee arthro-plasty: improved coronal alignment. J Arthroplasty.2005;20(7 Suppl 3):123-31.

8. Perlick L, Bäthis H, Tingart M, Perlick C, Grifka J.Navigation in total-knee arthroplasty: CT based im-plantation compared with the conventional technique.Acta Orthop Scand. 2004;75:464-70.

9. Bolognesi M, Hofmann A. Computer navigationversus standard instrumentation for TKA: a single-surgeon experience. Clin Or thop Relat Res. 2005;440:162-9.

10. Mielke RK, Clemens U, Jens JH, Kershally S.[Navigation in knee endoprosthesis implantation—preliminary experiences and prospective compara-tive study with conventional implantation technique].Z Orthop Ihre Grenzgeb. 2001:139:109-16. German.

11. Anderson KC, Buehler KC, Markel DC. Computerassisted navigation in total knee arthroplasty: com-parison with conventional methods. J Arthroplasty.2005;20(7 Suppl 3):132-8.

12. Haaker RG, Stockheim M, Kamp M, Proff G,Breitenfelder J, Ottersbach A. Computer-assisted nav-igation increases precision of component placementin total knee ar throplasty. Clin Or thop Relat Res.2005;433:152-9.

13. Gregori A, Holt G. Letter regarding Navigatedtotal knee arthroplasty. A meta-analysis. (2007;89:261-269). J Bone Joint Surg Am. epub 2007Mar 27. http://www.ejbjs.org/cgi/eletters/89/2/261#31862.

14. Goodman SN. Have you ever meta-analysis youdidn't like? Ann Intern Med. 1991;114:244-6.

15. Felson DT, Anderson J. Evidence for the superi-ority of immunosuppressive drugs and prednisoneover prednisone alone in lupus nephritis. Resultsof a pooled analysis. New Engl J Med. 1984;311:1528-33.

16. Kaizar EE. Metaanalyses are observational stud-ies: how lack of randomization impacts analysis. AmJ Gastroenterol. 2005;100:1233-6.

Can Vitamin C Prevent ComplexRegional Pain Syndrome inPatients with Wrist Fractures?To The Editor:In the article “Can Vitamin C Prevent Com-plex Regional Pain Syndrome in Patientswith Wrist Fractures? A Randomized, Con-trolled, Multicenter Dose-Response Study”

(2007;89:1424-31), Zollinger et al. studiedthe prophylactic effect of vitamin C on theprevalence of complex regional pain syn-drome in 416 patients with a wrist fracture.They concluded that vitamin C is indeed ef-fective, and they recommended giving 500mg of vitamin C daily for fifty days to eachpatient with a wrist fracture to prevent com-plex regional pain syndrome.

Some limitations of this study men-tioned in the article include a large selectionbias (416 of 2137 eligible patients were en-rolled) and a low event rate due to an unex-pected low prevalence of complex regionalpain syndrome (4.2% compared with 22%in the authors’ previous study 1). This meansthat only eighteen patients (eight of the328 in the treatment group and ten of theninety-nine in the placebo group) fulfilledthe criteria for complex regional pain syn-drome. In one patient with fractures of bothwrists, complex regional pain syndrome de-veloped on one side, where the fractureturned out to be badly reduced, and theother side healed without complications.This example reveals dramatically how thisstudy demonstrates a strong confounder: al-though the number of fractures needing re-duction was equal in both groups, thequality of the reduction was not mentioned.

Open reduction and internal fixationof wrist fractures generally achieves a betterreduction than closed reduction with appli-cation of a cast. Retrospective studies of sur-gically treated wrist fractures have thereforedemonstrated a lower incidence rate ofcomplex regional pain syndrome, of around3.5%2. To my knowledge, no prospectivestudy has ever demonstrated an association

8/17/2019 j.n.katz

6/10



between the incidence of complex regionalpain syndrome and the quality of reduc-tion, but pain syndromes in general occurmore frequently when fractures are not ade-quately reduced.

Much scientific effort has been putin attempts to achieve prophylaxis andtreatment for complex regional pain syn-drome with pharmacological means, butthese efforts did not result in any clinicalrecommendations 3. Conservative physicaltherapy has provided some benefit for pa-tients with complex regional pain syn-drome 4. Since the introduction of functionaland time-contingent “pain-exposure” phy-sical therapy in children with complex re-gional pain syndrome by Sherry et al. in19995, more reports on this approach areto be expected for adult patients as well.

A difference is therefore to be ex-pected between patients with complex re-gional pain syndrome who are treated bya physical therapist and those who are not.The use of any form of physical therapy isnot mentioned in this paper, introducinganother possible confounder. This papertherefore does not provide support for theeffectiveness of vitamin C in preventingcomplex regional pain syndrome.

Jan Paul M. Frölke, MD, PhD

University Medical Center St. Radboud, P.O.Box 9101, 6900 HB Nijmegen, The Nether-lands, e-mail: [email protected]

Disclosure: The author did not receive anyoutside funding or grants in support of his re-search for or preparation of this work. Neitherhe nor a member of his immediate family re-ceived payments or other benefits or a com-mitment or agreement to provide such benefitsfrom a commercial entity. No commercial en-tity paid or directed, or agreed to pay or direct,any benefits to any research fund, foundation,division, center, clinical practice, or othercharitable or nonprofit organization withwhich the author, or a member of his immedi-ate family, is affiliated or associated.

P.E. Zollinger, W.E. Tuinebreijer,R.S. Breederveld, and R.W. Kreis reply:We read the letter of our colleague, Dr.Frölke, with great interest. First, on the ba-sis of our study, we believe that vitamin Cdoes prevent complex regional pain syn-drome. Unfortunately, most of Dr. Frölke’scomments do not apply to our study.

The number of enrolled patients inour study in relation to the number of eligi-

ble patients was mentioned in the Discus-sion of our article. The quality of reductionwas studied in this paper and in our paper inLancet 1 as well. In both studies, there was norelationship between the occurrence of

complex regional pain syndrome and theneed to undergo fracture reduction. More-over, the quality of reduction did not influ-ence the chance of complex regional painsyndrome developing. We performed thecurrent study because, to our knowledge,there have been no published studies since19991 that either confirm or refute our origi-nal findings.

To our knowledge, no prospectivestudy has ever demonstrated an associationbetween the prevalence of complex regionalpain syndrome and the quality of reduc-tion. Retrospective studies do not have the

level of evidence that is needed. Dr. Frölkemakes a misjudgment by citing the article byArora et al. 2. Arora et al. found that, of 114patients followed for one year, five had type-I complex regional pain syndrome and threehad type-II complex regional pain syn-drome. Thus, the prevalence of type-I com-plex regional pain syndrome in their study is4.39% (not 3.5% as stated in Dr. Frölke’s let-ter) and is higher than our overall preva-lence of 4.2%; it stands in contrast with the2.4% for all of our patients treated with vita-min C. The difference is even more strikingwhen the 4.39% rate is compared with theprevalence of only 1.8% in our group receiv-ing 500 mg of vitamin C and 1.7% in thegroup receiving 1500 mg.

Why the articles by Rowbotham 3, Oer-lemans et al. 4, and Sherry et al. 5 are cited isunclear to us. Our study is about the possibleprevention of complex regional pain syn-drome after a wrist fracture in adults treatedwith a prophylactic dose of vitamin C andnot about the therapy for complex regionalpain syndrome itself. The end point of ourstudy was defined as the presence of complexregional pain syndrome at any time withinone year after the fracture (see the Study De-sign section). The article by Rowbotham 3 deals with pharmacotherapy in patients withcomplex regional pain syndrome.

The article by Oerlemans et al. 4 is avery well-respected trial comparing adjuvantphysical therapy with occupational therapyfor patients with complex regional pain syn-drome. Here lies the difference with our frac-ture patients. If we had treated our patientswith physical therapy as well, we would havecreated our own confounding factor. Skep-

tics would have challenged our conclusionsand pointed to the positive effect of the physi-cal therapy rather than to the effect of vita-min C, as Dr. Frölke does now.

When complex regional pain syn-

drome develops in patients who have sus-tained a wrist fracture, it is of course treatedwith physical therapy and medication, ifnecessary 6. The article by Sherry et al. 5 dealswith the outcome in children with complexregional pain syndrome after exercise ther-apy. However, we believe that complex re-gional pain syndrome in children is acompletely different entity than complexregional pain syndrome in adults, and sois the approach to its treatment. This wasconfirmed by Wilder et al. 7, who remindedus that, in children, complex regional painsyndrome most often involves the lower ex-

tremity (87% [sixty-one] of seventy cases),which is in contrast to the situation inadults, who have more upper-extremitycomplex regional pain syndromes. Thetherapie used by Sherry et al. 5 consisted ofaerobic functionally directed exercises, hy-drotherapy, and desensitization. Whichtherapy achieved the desired outcome? Canit get more confounding than this?

Paul E. Zollinger, MDW.E. Tuinebreijer, MD, PhD, MSc, MA

R.S. Breederveld, MD, PhDR.W. Kreis, MD, PhD

Corresponding author: Paul E. Zollinger, MD,Department of Orthopaedic Surgery, Zieken-huis Rivierenland, President Kennedylaan 1,4002 WP Tiel, The Netherlands, e-mail:[email protected]


References1. Zollinger PE, Tuinebreijer WE, Kreis RW, Breeder-veld RS. Effect of vitamin C on frequency of reflexsympathetic dystrophy in wrist fractures: a random-ized trial. Lancet. 1999;354:2025-8.

2. Arora R, Lutz M, Hennerbichler A, Krappinger D, Es-pen D, Gabl M. Complications following internal fixa-tion of unstable distal radius fracture with a palmar

locking-plate. J Orthop Trauma. 2007;21:316-22.3. Rowbotham MC. Pharmacologic management ofcomplex regional pain syndrome. Clin J Pain.2006;22:425-9.

4. Oerlemans HM, Oostendorp RA, de Boo T, GorisRJ. Pain and reduced mobility in complex regionalpain syndrome I: outcome of a prospective ran-domised controlled clinical trial of adjuvant physicaltherapy versus occupational therapy. Pain.1999;83:77-83.

5. Sherry DD, Wallace CA, Kelley C, Kidder M, SappL. Short- and long-term outcomes of children withcomplex regional pain syndrome type I treated withexercise therapy. Clin J Pain. 1999;15:218-23.

8/17/2019 j.n.katz

7/10



6. The Netherlands Society of Rehabilitation Special-ists. Guidelines: Complex regional pain syndrometype I. 2006. http://www.posttraumatischedystro-fie.nl/pdf/CRPS_I_Guidelines.pdf.

7. Wilder RT, Berde CB, Wolohan M, Vieyra MA,Masek BJ, Micheli LJ. Reflex sympathetic dystrophyin children. Clinical characteristics and follow-upof seventy patients. J Bone Joint Surg Am. 1992;74:910-9.

Exposure to Direct and ScatterRadiation with Use ofMini-C-Arm FluoroscopyTo The Editor:We commend Giordano et al. on their excel-lent work in quantifying the risk of radia-tion when using a mini-C-arm fluoroscopyunit, as reported in their study entitled“Exposure to Direct and Scatter Radiation

with Use of Mini-C-Arm Fluoroscopy”(2007;89:948-52). Their methodology, how-ever, does not accommodate for the mea-surement of increased radiation exposurewhen the C-arm is used in the conventionalmethod, with the image intensifier verti-cally above the radiation source 1. Nor does itestimate what the exposure dose would beimmediately level to the receiver. Their data,however, remain of value to advance theoverall safety of fluoroscopy in theater.

In our as yet unpublished survey ofmore than seventy-five orthopaedic traineesand theater staff in the United Kingdom, we

found that the majority had poor workingknowledge of conventional image intensi-fier usage and surprisingly little insight intoionizing radiation protection issues. Al-though most orthopaedic trainees in theUnited Kingdom do not “push the button,”they do “guide” the radiographer and super-vise the surgical assistant and theater staff.Therefore, the patient, surgical teams, andtheater staff may be at r isk of exposure. Withappropriate training of surgeons, the mini-C-arm may be adopted more widely in theNational Health Service, thereby releasingoverburdened radiographers from theater

while increasing throughput and safety intheater, as alluded to by White 2. However, webelieve that this can only occur once the re-cently disbanded ionizing radiation protec-tion course has been reinstigated.

Narlaka Jayasekera, MRCSRichard Roach, FRCS(Orth)

Corresponding author: Narlaka Jayasekera,Department of Orthopaedics, Princess RoyalHospital, Telford, Shropshire TF1 6TF, UnitedKingdom, e-mail: [email protected]

Disclosures: The authors did not receive anyoutside funding or grants in support of theirresearch for or preparation of this work. Nei-ther they nor a member of their immediatefamilies received payments or other benefits ora commitment or agreement to provide suchbenefits from a commercial entity. No com-mercial entity paid or directed, or agreed topay or direct, any benefits to any research fund,foundation, division, center, clinical practice,or other charitable or nonprofit organizationwith which the authors, or a member of theirimmediate families, are affiliated or associated.

J.F. Baumhauer andB.D. Giordano reply:We appreciate the comments of Mr. Jayasek-era and Mr. Roach and acknowledge thatour methodology does not reflect a number

of conventional techniques that have beenemployed in the past during the routine useof mobile C-arm fluoroscopy.

In our paper, we make note of severaldose-reducing measures that have beenstudied over the years and have enabledmobile C-arm operators to produce high-quality images while optimizing the overallsafety to the patient and operating roomstaff. These measures include minimizingexposure time, reducing exposure factors,manipulating the x-ray beam with collima-tion, maximizing distance from the beam,using protective shielding, and imaging withthe C-arm in an inverted orientation relativeto the specimen.

Positioning the phantom limb di-rectly on the platform of the image intensi-fier increases the distance from the radiationsource to the specimen, subsequently reduc-ing the amount of scatter produced. Al-though many of these measures have beenstudied with use of a standard large C-armunit, the literature regarding similar param-eters with the mini C-arm unit is limited. Inour experimental design, we attempted tocreate a “best-case scenario” by utilizingknown dose-reducing techniques to quan-tify radiation exposure just as a surgeonwould likely strive to achieve in a true op-erating room setting.

With regard to the second portionof the correspondents’ comments, we pointout that at positions of 15 and 25 cm from afocal point on the phantom hand, we foundminimal radiation exposure (1 to 2 mrem) asmeasured with our dosimeters. These mea-surements were made in the plane of the im-age intensifier. In contrast, when the radiation

dosimeter was placed directly in the phantomhand, substantial exposure levels (181 to 272mrem) were recorded. We did not collect datapoints between these two locations.

We concur with Jayasekera and Roach

that many orthopaedic trainees and, for thatmatter, a great number of mini or large C-arm operators, have a poor understanding ofthe science behind image intensifier usage.This may lead them to grossly underestimatethe potential for high-dose radiation expo-sure if these mobile fluoroscopy units are notused judiciously and with proper intent.

A common error made by novicetrainees is the use of the mini C-arm to im-age larger body parts such as the tibia, fe-mur, humerus, elbow, or shoulder. As thetissue density and cross sectional area of theimaging subject increase, technique factors

automatically adjust, in the normal mode,to produce an image with optimal penetra-tion and visual quality. To accommodate forthe increased tissue density of a larger bodypart, technique factors increase by a sub-stantial margin, leading to a much higherradiation exposure rate than may have beenencountered when using a large C-arm.

We appreciate the interest in our pa-per and strive to advance science safety withthe commonly used fluoroscopy units.

Judith F. Baumhauer, MDBrian D. Giordano, MD

Corresponding author: Judith F. Baumhauer,MD, Division of Foot and Ankle Surgery, Uni-versity of Rochester Medical Center, 601 Elm-wood Avenue, Box 665, Rochester, NY 14642,e-mail: [email protected]


References1. Tremains MR, Georgiadis GM, Dennis MJ. Radia-tion exposure with use of the inverted-C-arm tech-nique in upper-extremity surgery. J Bone Joint SurgAm. 2001;83:674-8.

2. White SP. Effect of int roduction of mini-C-arm imageintensifier in orthopaedic theatre. Ann R Coll SurgEngl. 2005;87:53-4.

Integrity of the Lateral FemoralWall in Intertrochanteric HipFractures: An ImportantPredictor of a ReoperationTo The Editor:The article “Integrity of the Lateral FemoralWall in Intertrochanteric Hip Fractures: AnImportant Predictor of a Reoperation,”

8/17/2019 j.n.katz

8/10



(2007;89:470-5), by Palm et al., is particu-larly important because it confirms previousreports on the critical role played by thelateral wall in the reconstruction of pertro-chanteric hip fractures 1-3. While devices such

as the dynamic hip screw and sliding hipscrew have been considered the gold stan-dard in the treatment of pertrochanteric hipfractures for fifty years, this type of iatro-genic complication has been reported onlyrecently 1; thus, I would like to offer someobservations.

The lateral wall exists in conjunctionwith a pertrochanteric hip fracture; it doesnot exist, as an anatomical structure, in anormal intact femur. It is important to dis-tinguish between those fractures where thelateral wall does not exist preoperatively andthose where it does exist preoperatively and

is fractured either intraoperatively or post-operatively. The former have already beendefined in the “Fracture and DislocationCompendium,” where, in fact, the term lat-eral wall is not used 4. This classification sys-tem does distinguish types 31-A1 and 31-A2fractures, which are defined as pertrochan-teric fractures, from a type 31-A3, which isdefined as an intertrochanteric fracture. It isunfortunate that the authors do not useboth terms. Rather, they use only the termintertrochanteric fracture , which may lead tomisunderstanding and confusion. On theother hand, the iatrogenically fractured lat-eral wall, occurring during or following asurgical procedure, converts a pertrochan-teric A1 or A2 fracture into an intertrochan-teric A3 fracture and is certainly differentand deserves special attention. The clear dis-tinction between the two did not emergefrom the paper.

Because of the nature of this compli-cation, it has been considered to be a dis-tinct entity: the pantrochanteric fracture 5.

Once a fracture of the lateral wall isrecognized as an iatrogenic complication,and the events leading to the fracture areunderstood, a reevaluation of the situationis indicated. First, new definitions are neces-sary. It is important to distinguish betweenfracture collapse, the outcome of fracturingthe lateral wall (an adverse postoperativeevent), and controlled fracture impaction(a desirable postoperative event). This haspreviously been defined together with otherrelevant definitions 1 and could have been re-ferred to by the authors.

Careful definition will not onlycontribute to better understanding of the

postoperative radiograph, and hence thepatient's condition, but will also facilitatedecision-making in the postoperative reha-bilitation period, e.g., the type of weight-bearing to be instituted.

In addition, when it is possible to at-tribute the collapse to certain proceduresand/or devices, this should enable us to setnew surgical standards designed specificallyto avoid this kind of complication.

Yechiel Gotfried, MD, MS

Bnai Zion Medical Center, 47 GolombStreet, P.O.B. 4940, Haifa 31048, Israel.E-mail: [email protected]

Disclosure: The author did not receive anyoutside funding or grants in support of his re-search for or preparation of this work. Theauthor, or a member of his immediate family,

received, in any one year, payments or otherbenefits in excess of $10,000 or a commit-ment or agreement to provide such benefitsfrom a commercial entity (Orthofix, Inc.). Nocommercial entity paid or directed, or agreedto pay or direct, any benefits to any researchfund, foundation, division, center, clinicalpractice, or other charitable or nonprofit or-ganization with which the author, or a mem-ber of his immediate family, is affiliated orassociated.

H. Palm, S. Jacobsen, S. Sonne-Holm,and P. Gebuhr reply:We appreciate the interest by Dr. Gotfried inour recent article and are delighted that hefinds our study to be particularly important.In a large number of patients, our studydoes, in fact, confirm previous reports of theimportance of the integrity of the lateralfemoral wall, including the fact that a frac-ture of the lateral femoral wall is most oftenan iatrogenic complication.

Dr. Gotfried raises good questionsregarding the nomenclature used in the ar-ticle. The general nomenclature for thesefractures is quite confusing. As the termstrochanteric, pertrochanteric, pantrochan-teric, and intertrochanteric, etc., are oftenmixed up, we also find it highly relevant toachieve international consensus on thismatter. In our article, we simply used theterm intertrochanteric for all type 31-Afractures, in part, because we found thatDr. Gotfried also previously did this 1, al-though not in a later article 2 referred to inour study. We now agree that using theterms pertrochanteric for the type 31-A1and 31-A2 fractures and intertrochanteric

only for the type 31-A3 fractures wouldhave been more precise. On the other hand,we still find that we enable the reader todistinguish between the fracture types byusing the AO/OTA classification numbers,

including the very important subtypes inthe text and tables, and by showing an il-lustrating diagram.

We agree that new definitions of bio-mechanical complications are necessaryand that the knowledge that the lateralfemoral wall is an iatrogenic complicationcould contribute to a better understandingof the treatment of these fractures. We cur-rently treat type 31-A1 and 31-A2.1 frac-tures with a sliding hip screw fixed to alateral plate and type 31-A3 fractures witha sliding hip screw fixed to an intramedul-lary nail.

As a third of the 31-A2.2 and 31-A2.3 fractures in our study were convertedto 31-A3 fractures, we now also treat thesefractures using the sliding hip screw fixedto an intramedullary nail. In the future,perhaps other systems designed specifi-cally to avoid a perioperative fracture of thelateral femoral wall 1 might prove to be su-perior to treat these specific fracture sub-groups. To date, it has not been feasible tocategorize fractures into all of the AO/OTAsubgroups as this demands very largegroups of patients.

Henrik Palm, MDSteffen Jacobsen, MD

Stig Sonne-Holm, MD, DMSc Peter Gebuhr, MD

Corresponding author: Henrik Palm, MD,Department of Orthopaedic Surgery, Copen-hagen University Hospital of Hvidovre, Kette-gaard Alle 30, DK-2650 Hvidovre, Denmark,e-mail: [email protected]


References1. Gotfried Y. Percutaneous compression plating ofintertrochanteric hip fractures. J Orthop Trauma.2000;14:490-5.

2. Gotfried Y. The lateral trochanteric wall: a keyelement in the reconstruction of unstable pertro-chanteric hip fractures. Clin Orthop Relat Res.2004;425:82-6.

3. Im GI, Shin YW, Song YJ. Potentially unstable inter-trochanteric fractures. J Orthop Trauma. 2005;19:5-9.

4. Fracture and dislocation compendium. Ortho-paedic Trauma Association Committee for Codingand Classification. J Ort hop Trauma. 1996;10 Suppl1: v-ix, 1-154.

5. Gotfried Y. Pantrochanteric hip fracture: an entity.J Bone Joint Surg Br. (Suppl III) 2000;82:235.

8/17/2019 j.n.katz

9/10



Comparison of the Vastus-Splittingand Median Parapatellar Approachesfor Primary Total Knee Arthroplasty:A Prospective, Randomized Study.Surgical Technique

To The Editor:The otherwise excellent article, “Compari-son of the Vastus-Splitting and MedianParapatellar Approaches for Primary TotalKnee Arthroplasty: A Prospective, Random-ized Study. Surgical Technique” (2007;89Suppl 2 Part 1:80-92), by Kelly et al., wasmarred by an error in the legend to Figure1. The axial radiograph of the knee wasmislabeled as “a Merchant radiograph ofthe patella.”

The shape and appearance of thedista part of the femur on the radiographdemonstrates that it is really a Settegast

view. This technique requires the knee to beacutely flexed well beyond 90°, drawing thepatella, which might otherwise be severelysubluxated laterally at the trochlear level,into the intercondylar space to articulatewith the distal, or weight-bearing, surfaceof the femoral condyles.

Conversely, the “Merchant” axialview radiograph is exposed with bothknees flexed no more than 45°, showingthe patella’s true relationship to thetrochlea 1,2.

This may seem to be a minor point,but if the surgeon is not aware that thepatella is subluxated laterally prior to sur-gery, he or she may not take sufficient mea-sures to correct that subluxation duringsurgery. Many postoperative patellofemo-ral complications can be avoided if the sur-geon is aware of this problem beforesurgery.

Alan C. Merchant, MD

Stanford University, 124 Marvin Avenue,Los Altos, CA 94022, e-mail: [email protected]

Disclosure: The author did not receive anyoutside funding or grants in support of hisresearch for or preparation of this work.Neither he nor a member of his immediatefamily received payments or other benefits ora commitment or agreement to provide suchbenefits from a commercial entity. No com-mercial entity paid or directed, or agreed topay or direct, any benefits to any researchfund, foundation, division, center, clinicalpractice, or other charitable or nonprofit or-ganization with which the author, or a mem-ber of his immediate family, is aff iliated orassociated.

V.D. Pellegrini Jr., M.J. Kelly, M.N. Rumi, M. Kothari, K.J. Bailey,W.M. Parrish, and M.A. Parentis reply:We thank Dr. Merchant for correctly identi-fying our error as it relates to patellofemoral

imaging of the knee. We concur with hiscomments and, indeed, customarily per-form patellofemoral imaging with the kneein 30° of flexion to more sensitively identifylateral subluxation of the patella. The patel-lar view presented in our paper does not re-flect our usual practice.

We appreciate Dr. Merchant’s effortsin bringing this inadvertent misrepresenta-tion to our attention as well as that of thereadership of The Journal.

Vincent D. Pellegrini Jr., MD Matthew J. Kelly, MD

Mustasim N. Rumi, MD

Milind Kothari, DOKatrina J. Bailey, PT

William M. Parrish, MD Michael A. Parentis, MD

Corresponding author: Vincent D. PellegriniJr., MD, Department of Orthopaedics, Univer-sity of Maryland School of Medicine, 22 SouthGreene Street, Suite S 11 B, Baltimore, MD21201, e-mail: [email protected]


References1. Merchant AC, Mercer RL, Jacobsen RH, Cool CR.

Roentgenographic analysis of patellofemoral congru-ence. J Bone Joint Surg Am. 1974;56:1391–6.

2. Merchant AC. Patellofemoral imaging. Clin OrthopRelat Res. 2001;389:15–21.

Cost-Effectiveness of Extended-Duration Antithrombotic ProphylaxisAfter Total Hip ArthroplastyTo The Editor:We read with interest the recent paper “TheCost-Effectiveness of Extended-DurationAntithrombotic Prophylaxis After Total HipArthroplasty” (2007;89:819-28), by Skedgelet al., regarding economic decision-making,with reference to extended thrombopro-phylaxis after total hip arthroplasty. Theauthors refer to a study by Lapidus et al. 1,who stated that 38.4% of patients receivinglow-molecular-weight heparin required acommunity nurse for administration. Forcost-effectiveness, the number requiring acommunity nurse must be

8/17/2019 j.n.katz

10/10



heparin could meet a threshold of $50,000per quality-adjusted life year gained withhome care rates of

j.n.katz

Documents