Measuring andUnderstanding Treatment

Effectiveness in Hand Surgery Sophocles H. Voineskos, MD, MSca,b,*,Christopher J. Coroneos, MDa,b,Achilleas Thoma, MD, MSc, FRCSCa,b,c,Mohit Bhandari, MD, PhD, FRCSCc,d

KEYWORDS

� Hand surgery � Treatment outcome � Comparative effectiveness research � Relative risk � P-value� Confidence interval � Minimal clinically important difference � Number needed to treat

KEY POINTS

� The randomizedcontrolled trial is the referencestandard studydesignwhencomparing2 therapeuticinterventions.

� Before interpreting study results, the methodology of the trial must be assessed to ensure therewere no errors in design or execution that might invalidate the results.

� Absolute and relative measures of treatment effectiveness assist in the interpretation of studyresults.

� Valid and reliable patient-important outcomemeasures in hand surgery exist and should be used toassess outcomes.

� Study findings should be interpreted within the clinical context, and preference, of the individualpatient being treated.

INTRODUCTION

In 1992, evidence-based medicine (EBM) wasintroduced as a shift in medical paradigms.1 Thedemand for a rational decision-making processin health care has increased the importance of us-ing EBM to enhance the traditional skills of clinicalpractice. Today, rational application of EBM prin-ciples to hand surgery research is not only

Conflicts of Interest: The authors state that they have nowere received for the preparation of this article.a Division of Plastic and Reconstructive Surgery, DepartmMain Street West, Hamilton, Ontario L8N 3Z5, CanadMcMaster University, 202-39 Charlton Avenue East, Hamilical Epidemiology and Biostatistics, McMaster University H4K1, Canada; d Division of Orthopaedic Surgery, DeparBuilding, 293 Wellington Street North, Hamilton, Ontari* Corresponding author. Division of Plastic and ReconsUniversity, HSC 4E12, 1200 Main Street West, Hamilton,E-mail address: sophocles.voineskos@medportal.ca

Hand Clin 30 (2014) 285–292http://dx.doi.org/10.1016/j.hcl.2014.04.0080749-0712/14/$ – see front matter � 2014 Elsevier Inc. All

recommended, but expected if surgeons are toidentify and appraise best evidence.2

Evidence-based hand surgery posits that clini-cians should be using the best evidence availablewhen managing their patients, even if the quality ofthat evidence is limited.3 In-depth backgroundknowledge and sound clinical decision makingare more important than ever, as evidence of

conflicts of interest or financial disclosures. No funds

ent of Surgery, McMaster University, HSC 4E12, 1200a; b Surgical Outcomes Research Centre (SOURCE),ton, Ontario L8N 1Y3, Canada; c Department of Clin-SC 2C, 1280 Main Street West, Hamilton, Ontario L8Stment of Surgery, McMaster University, Well-Healtho L8L 8E7, Canadatructive Surgery, Department of Surgery, McMasterOntario L8N 3Z5.

rights reserved. hand.th

eclinics.com

Voineskos et al286

treatment effectiveness does not automaticallyimply that the treatment should be administered.Applying EBM principles for an intervention aids

surgeons in choosing the most appropriate surgi-cal procedure for a particular patient. Surgeonsneed the tools to assess the strength and useful-ness of the available evidence and to evaluatethe measurement and treatment effectiveness ofan outcome. This process will allow us to defendtherapeutic interventions based on available evi-dence and not anecdote.4

In a randomized controlled trial (RCT), individ-uals are randomly allocated to treatment groupsand followed to determine the effect of the thera-peutic intervention on one or more outcomes.The advantage of a methodologically sound RCTis the ability to balance both known and unknownprognostic factors between the 2 comparisongroups.3

Conversely, in a nonrandomized (observational)study, the deliberate choice of treatment of eachpatient implies that the observed outcome maybe a result of differences among the patients un-dergoing treatment, rather than the treatment it-self.4 The potential roles of observational studiesin hand surgery have been discussed previously.5

With an appropriate research question, and recog-nition of inherent limitations, the observationalstudy can be used effectively in hand surgery.5

This article uses a practical example to demon-strate how evidence-based surgery concepts areused to interpret and apply the results of an RCTin hand surgery.

CLINICAL SCENARIO

Mr W. White is a 57-year-old man with Dupuytrendisease. On examination, he has a metacarpopha-langeal joint contracture of 35� of his ring fingerand an easily palpable palmar cord over the ringmetacarpal. The patient states that a family mem-ber, who is in the medical field, provided him with arelevant publication to pass along to you and rec-ommended he request a limited fasciectomy (LF)treatment of his Dupuytren contracture. He pro-duces a printed copy of the following RCT byvan Rijssen and colleagues6: “Five-year results ofa randomized clinical trial on treatment in Dupuyt-ren disease percutaneous needle fasciotomyversus limited fasciectomy.”The publication provided to you describes an

RCT including 111 patients, with 115 affectedhands, with a minimal passive extension deficitof 30� and the existence of a clearly definedpalmar cord. Participants were randomly allocatedto percutaneous needle fasciotomy (PNF) or LF.The primary outcome was disease recurrence.6

ASSESSING THE METHODOLOGY OF THETRIAL

This article focuses on interpreting treatmenteffectiveness and applying the results of the trialto your patient. However, before using resultsfrom an RCT, the methodology of the trial mustbe assessed to ensure the trial’s validity. Themeth-odological assessment of this RCT has been sum-marized in Box 1. The detailed guide to answer thequestion “Are the results valid?” for a studycomparing therapeutic interventions can be foundin “Users’ Guides to theMedical Literature”3 and inthe “Users’ Guide to the Surgical Literature”.7

Furthermore, a discussion on specific challengesfor RCTmethodology in hand surgery is available.8

IMPORTANT CONCEPTS IN INTERPRETINGTREATMENT EFFECTIVENESSHow Large was the Treatment Effect?

The primary outcome in van Rijssen and associ-ates’ RCT6 is recurrence of disease. This outcomeis dichotomous, in which the patient either has anevent or he does not.6

In the van Rijssen and associates’ RCT,6 23.3%of the LF group developed recurrence (with one ofthese patients developing extension of disease). Inthe PNF group, 84.9% of patients developedrecurrence.6 These results can be expressed asa risk difference, relative risk (RR), or relative riskreduction (RRR) (Box 2).The absolute difference is also known as abso-

lute risk reduction (ARR) or risk difference. TheARR is the difference between the proportion whohad recurrence in the LF group (0.233) and the pro-portion who had recurrence in the PNF group(0.849), or PNF � LF 5 0.849 � 0.233 5 0.616, or62%. This measure of effect uses absolute, ratherthan relative, terms to compare the proportion ofpatients who experienced recurrence.When weighing benefit and harm, concept of a

number needed to treat (NNT) is used. The NNTis the number of patients a surgeon must treat tocreate one positive outcome (see section Are theLikely Treatment Benefits worth the PotentialHarm? for further discussion of NNT).Another way to express the impact of the treat-

ment is through an RR. The RR represents the riskof recurrence among patients who received LFrelative to those who received PNF, or LF/PNF 50.233/0.849 5 0.274, or 27%. To interpret thisRR, the risk of recurrence when undergoing LF isabout one-quarter that of PNF.When using a dichotomous outcome, the most

commonly reported measure of a treatment effectis the complement of the RR, which is the RRR. It

Box 1Key features of the RCT: LF versus PNF

Question Answer Details

Did intervention and control groups start with the same prognosis?Were patients randomized? Yes Performed by drawing of lotsWas randomization concealed? Partially Assignment envelopes were

numbered, but no mention ofbeing sealed or opaque

Were patients in the studygroups similar with respect toknown prognostic factors?

Yes Baseline characteristics displayedfor reader

Was prognostic balance maintained as the study progressed?Were patients blinded? Probably not Not mentioned. PNF occurred in

“outpatient treatment room”.LF occurred in “surgicaltheater”

Were outcome assessorsblinded?

No Measurements taken by thesurgeon

Were surgeons blinded? No Not possibleDid the investigators take intoconsideration the learningcurve?

Unknown Information is not included in themanuscript

Were the groups prognostically balanced at the study’s completion?Was follow-up complete? No, but had 5 year outcome 121 patients were randomized,

93 patients were included inthe final analysis. This trial hada follow-up rate of 76.8% for a5 year outcome

Were patients analyzed in thegroups to which they wererandomized?

Yes

Was the trial stopped early? No

Data from van Rijssen AL, ter Linden H, Werker PM. Five-year results of a randomized clinical trial on treatment inDupuytren’s disease: percutaneous needle fasciotomy versus limited fasciectomy. Plast Reconstr Surg2012;129(2):467–77.

Treatment Effectiveness in Hand Surgery 287

is expressed as a percent: RRR 5 (1 � LF/PNF) �100%5 (1� 0.274)� 100%5 76.2%, or 76%. AnRRR of 76%means that LF reduced the risk of revi-sion surgery by 76% relative to that occurringamong patients who underwent PNF. The greaterthe RRR, the more effective is the intervention.

The results section of the van Rijssen and asso-ciates’ RCT6 also displays the RR of recurrenceover time in a survival analysis (the Kaplan-Meiersurvival estimate). The weighted RR over the entirestudy is called a hazard ratio, which represents theinstantaneous risk over the study time period.

The discussed relative measures of treatmenteffect (RR and RRR) are commonly used. How-ever, it is important to differentiate between the ab-solute measure (ARR), and the relative measures(RR,RRR). Although not the case in the van Rijssenand associates’ RCT,6 there are occasions whensimply stating an RRR can be deceptive.

For example, if the recurrence rate of Dupuytrendisease when performing a PNF is 50% and whenperforming an LF is 25%, then the RRR is calcu-lated to be 50%, which is impressive.

However, if the recurrence rate of Dupuytrendisease was rare, for example 2% in the PNFgroup and 1% in the LF group, the RRR is also50%. In this example, simply interpreting the re-sults with a relative measure of treatment effectis misleading. The corresponding ARR of recur-rence of disease of 1% is no longer as excitingor likely to change your clinical practice.

How Precise was the Estimate of TreatmentEffect?

The true risk reduction of a treatment of a popu-lation is unknown. The risk reduction of Dupuyt-ren recurrence (see Box 2) provided by the

Box 2Understanding treatment effectiveness in thevan Rijssen and associates’ RCT

Exposure

Outcome at 5 Years

RecurrenceNoRecurrence Total

LF 10a 33 43PNF 8 45 53

PNF risk: 8/53 5 84.9%.LF risk: 10/43 5 23.3%.Absolute risk reduction or risk difference (ARR) 5

PNF � LF 5 84.9% � 23.3% 5 61.6%, or 62%.Relative risk (RR) 5 LF/PNF 5 (10/43)/(8/53) �

100% 5 27.4% or 27%.Relative risk reduction (RRR) 5 [1 � (LF/PNF)] �

100% 5 1 � 27.4% 5 72.6% or 73%.Number Needed to Treat (NNT)5 1/ARR5 1/0.6165

1.6, or 2.a In the LF group, one of the 10 patients with recur-

rence had extension of the disease.

Data from van Rijssen AL, ter Linden H, Werker PM.Five-year results of a randomized clinical trial on treat-ment in Dupuytren’s disease: percutaneous needle fas-ciotomy versus limited fasciectomy. Plast ReconstrSurg 2012;129(2):467–77.

Voineskos et al288

results of the van Rijssen and associates’ RCT6

is a point estimate based on data from a sampleof patients recruited from the generalpopulation.The P-value is often used by investigators to

determine statistical significance, with a conven-tional threshold of 0.05. A P-value of .05 repre-sents a probability of 1 in 20 that the observedeffect is due to chance alone, when there is norelationship between the treatment and outcome.The van Rijssen and associates’ RCT6 reports a

difference in recurrence rate between the 2 surgi-cal interventions of P<.001. This result is inter-preted as a statistically significant differencebetween the 2 groups; the recurrence rate of Du-puytren disease after 5 years in the PNF group issignificantly higher than in the LF group.Unfortunately, there are limitations to the

concept of a P-value. Dichotomy is introduced inthe form of labeling a study as “positive” or “nega-tive” when a P-value is used.3 The drawback of the“yes/no” dichotomy is obvious when considering atrial outcome with a P-value of .049. This trialwould be considered a “positive” trial, but howdifferent is this result from 0.05? or a “negative”trial that might not receive as much attention witha P-value of .051? In many cases, the differencein these examples can be attributed to a singleevent, demonstrating the potential fragility of theanalysis this P-value represents. Furthermore, the

P-value is unable to measure the strength of anassociation.2

A confidence interval (CI) provides the range ofvalues within which, given the trial data, the true ef-fectmight actually lie.9 TheCI defines the range thatincludes the true risk reduction for a given probabil-ity, assuming theRCTwasmethodologically sound.Conventionally, a 95% CI is used and can be in-

terpreted as the range of values the investigatorsbelieve, with 95% certainty, that the true popula-tion value lies.9 Therefore, this also means theybelieve there is a 5% chance that the calculatedCI did not encompass the true treatment effect.Unfortunately, the CI is not provided around the

treatment effect in the van Rijssen and associates’RCT.6AlthoughaCIof “1.597 to2.628” is referencedby the investigators after reporting the P-value, it isnot clear to the reader what this CI represents.In general, as sample size increases in a trial, the

number of outcome events increases, and there isa greater likelihood that the treatment effectobserved in the trial converges with the true treat-ment effect.3 Therefore, more precision (narrowerCIs) can result from larger sample sizes.CIs can help avoid the dichotomy of considering

a trial “positive” or “negative”. A CI can demon-strate the magnitude and direction of a treatmenteffect and help to determine clinical significance.In a “positive” trial, the CI can tell you if the sam-

ple size was adequate and results are definitive, ornondefinitive and further trials are required. Like-wise, in a “negative” trial, the CI can be used todetermine whether the sample size was adequateand the trial results definitely negative, or nondefi-nitive and further trials are required.3

Are the Results Clinically Significant?

A trial can demonstrate a statistically significant dif-ference between treatment groups, yet this differ-ence may not be of clinical significance to thepatient. Owing to the precision gained fromextremely large sample sizes, a statistically signifi-cant result may in fact represent a clinically trivialchange in the patient’s outcome. Again using thehypothetical example mentioned earlier where therecurrence rates for PNF and LF were rare anddiffered by only 1%, if the study was conducted inmany thousands of patients and yielded statisti-cally significant results, it indicates lower recur-rence in the LF group. Would you change yourclinical practice and abandonPNFentirely becauseof the “significant” difference in recurrence rates?Judgment of the appropriateness of the primary

outcome chosen is also required. The dichotomousprimary outcome, recurrence of Dupuytren dis-ease, is a patient-important outcome.As discussed

Box 3Inclusion and exclusion criteria of the RCT: LFversus PNF

Inclusion Criteria

1. A flexion contracture of at least 30� in themetacarpophalangeal, proximal interpha-langeal, or distal interphalangeal joints

2. A clearly defined pathologic cord in thepalmar fascia

3. Willingness to participate in this trial

Exclusion Criteria

1. Patients with post-surgical recurrence orextension of the disease

Treatment Effectiveness in Hand Surgery 289

earlier, the change in ARR for a dichotomousoutcome should be substantial. A large change inARR, 62%, (see Box 2) is demonstrated in the vanRijssen and associates’ RCT.6 The result of the pri-mary outcome is therefore clinically significant.

A continuous primary outcome can provide richinformation. Theminimal clinically important differ-ence (MCID) is a measure of a clinically important,or relevant, change in health10 in a continuousoutcome. The MCID is the smallest difference inoutcome score that patients perceive as beneficialenough that would mandate, in the absence ofside effects or excessive cost, a change in the pa-tient’s health care management.3,9,10

Effect size is ameasure of the strength of the rela-tionship between the intervention and theoutcome.11 It is calculated by taking the differencein mean scores between the 2 intervention groupsand dividing by the standard deviation (SD) of thescores in the control group (or the pooled SD ofthe treatment and control groups) (Effect Size 5(mtreatment� mcontrol)/SD).

3 Cohen provided a generalrule of thumbwhen interpreting themagnitude of aneffect size, approximately 0.2 being a small change,approximately 0.5 being a moderate change, andapproximately 0.8 being a large change.12 Normansuggested that a reasonable estimate of an MCIDcorresponds to an effect size of 0.5.13,14

In the publication of the van Rijssen and associ-ates’ 6-week RCT results,15 the disabilities of thearm, shoulder and hand (DASH) questionnaire isused as secondary outcome. The DASH is a vali-dated, patient-reported outcome measure. It is aquestionnaire that consists of 30 items and isscored from 0 to 100. In this scale, a lower scoremeans the patient experiences less disability. Ingeneral, for hand function, the DASH has an MCIDof 10.16–18 This means it would take a decrease inDASH score of 10 for a patient to perceive a benefitfrom their treatment. The value of the MCID is notsharedwith the reader in the 6-week RCT results.15

Therefore, although the reader is told that “DASHscores of both groups differed significantly at alltime points after treatment,” the reader cannot besure whether this difference in DASH score wasclinically significant for the patient.

2. Patients who were not allowed to stop tak-ing their anticoagulants

3. Patients generally unfit to have surgery

4. Patients who were not willing to participatein this study or had a specific treatment wish

Data from van Rijssen AL, Gerbrandy FS, ter Linden H,et al. A comparison of the direct outcomes of percuta-neous needle fasciotomy and limited fasciectomy forDupuytren’s disease: a 6-week follow-up study.J Hand Surg Am 2006;31:717–25.

IMPORTANT CONCEPTS IN APPLYINGRESULTS TO PATIENT CARECan the Results be Applied to My Patient?

The patient sitting in your office will often havedifferent characteristics than the patients whowere enrolled in the trial. Your patient may be olderor younger, with or without comorbidity or otherdemographic factors that would have excludedthem from participation in the trial.

If inclusion and exclusion criteria are clearly re-ported in the trial manuscript and your patientmeets the inclusion criteria, without violating theexclusion criteria, you can confidently apply the re-sults to your patient. The authors’ patient, MrWhite, fits the inclusion criteria and would nothave been excluded from this trial (Box 3).

However, what if the patient sitting in your officedoes not meet the study inclusion criteria? The re-sults of a trial would likely still apply to your patientif, for example, he/she has a comorbidity, or is4 years too old to have been included in the trial.A good approach to decide whether the resultscan be applied to your patient is to ask whetherthere is a compelling reason the results shouldnot apply to your patient.3

On the other hand, treatments, especially surgi-cal interventions, may not be uniformly effective.Because the results of an RCT are an estimate ofthe average treatment effect, applying thisaverage effect means the surgeon may exposethe patient who may only benefit slightly from thesurgery, to the full cost or risk of the intervention.4

Were All Clinically Important OutcomesConsidered?

Treatments are indicated when they provideimportant benefits.3 Hand surgeons should useappropriate techniques to measure relevant

Voineskos et al290

endpoints. This objective should be balanced withthe potential of burdening patients with several as-sessments and questionnaires.Physical tests and biomechanical measure-

ments (ie, range of motion, pinch strength, gripstrength etc.) are the most commonly reportedoutcome measures in hand surgery.19 Althoughobjective and reproducible measurements canbe obtained, these outcomes may not reflect thetrue benefit of a treatment to the patient. Whatthe surgeon might view as a considerableimprovement in grip strength may not correspondto improved hand function from the patient’sperspective.20

Both patients and clinicians should use a mea-surement that demonstrates the intervention im-proves the outcome that is important to thepatient, patient-important outcomes. The interestin patient-important outcomes, and the shift awayfrom objective, physical measures, has lead to anincrease in health-related quality of life (HRQL)questionnaires.19 These patient-reported ques-tionnaires are either broad, general health ques-tionnaires or region-specific and disease-specificquestionnaires. Some examples of validatedpatient-reported questionnaires in hand surgeryare listed in Box 4.When examined closely, the primary outcome

of the van Rijssen and associates’ RCT6 is infor-mation from a continuous biomechanical mea-surement (degrees of joint contracture) abridgedinto a dichotomous outcome of recurrence(contracture >30� or <30�). Using a validated scalesuch as the DASH,21 QuickDASH, MHQ,22 or themore recent Health Utilities Index (HUI, a generic

Box 4Commonly used patient-reported outcome instrume

Instrument Relevant Heal

Michigan Hand Questionnaire(MHQ)

Various healthhand disord

Disabilities of the Arm, Shoulder andHand Questionnaire (DASH)

Upper extremiand sympto

Quick Disabilities of the Arm,Shoulder and Hand Questionnaire(QuickDASH)

Upper extremiand sympto

Carpal Tunnel Questionnaire (BostonQuestionnaire)

Severity of symfunctional scarpal tunne

Data from Refs.21–23

quality of life scale from which utilities can becalculated)24 could have been more useful in inter-preting the benefit of the procedures. The MHQand the HUI-3 have been shown to be reliableand valid tools when assessing patients with Du-puytren disease.25 A new scale, specific to Du-puytren contracture will be useful for future handsurgery trials. The Unite Rhumatologique des Af-fections de la Main (URAM) is a practical 9-itemscale with a scoring system of 0 to 45, which hasalso been validated for Dupuytren disease.26 Thisscale was not available when van Rijssen and col-leagues started their study.

Are the Likely Treatment Benefits Worth thePotential Harm?

The final step in interpreting the effectiveness of atreatment is to decide whether the benefits of theintervention are worth the effort and risk that youand the patient will invest in the treatment andpost-operative recovery. As discussed earlier, alarge RRR may appear impressive, although theimpact on your patient could be minimal. Theconcept of NNT can be used to examine the abso-lute impact of the intervention on your practice.The NNT is the inverse of the ARR (NNT5 1/ARR).In the van Rijssen and associates’ RCT (van Rijs-

sen 2012), for the primary outcome of 5-yearrecurrence, the NNT 5 1/0.616 5 1.6, or 2. Thissuggests that for every 2 patients treated withLF, as opposed to PNF, the surgeon can preventone episode of recurrence.Infections were rare in the van Rijssen and asso-

ciates’ RCT.6 When evaluated at 1 week, the

nts in hand surgery

th Condition Scoring of Scale

states related toers

� 37-item scale

� Total score 0–100

ty physical functionms

� 30-item scale

� Total score 0–100

ty physical functionms

� 11-item scale

� Total score 0–100

ptoms, andtatus related tol syndrome

� Symptom severity scale,11-items

� Functional severity scale,8-itmes

Treatment Effectiveness in Hand Surgery 291

infection rate for the LF group was 1.8% and in thePNF group was 0%. The absolute risk differencefor infection at 1 week was ARR 5 1.8% �0% 5 1.8%. The number needed to harm (NNH)(where NNH is the number of interventions neededto harm one additional patient) would be NNH5 1/ARR 5 1/0.018 5 55.6, or 56 patients.

Therefore, for every 100 patients we considertreating with LF, we would expect to prevent 50episodes of recurrence (NNT 5 2, 100/2 5 50),at the cost of having 2 infections (NNH 5 56,100/56 5 1.79, or 2). When considering this infor-mation, the utility of LF in Dupuytren disease be-comes more certain. Understanding the conceptof NNT and NNH for relevant events helps the sur-geon and patient weigh the benefits against poten-tial harms associated with the intervention.

Are the Likely Treatment Benefits Worth thePotential Costs?

Hand surgeons are now expected both to use thebest available evidence to guide clinical decisionsand to ensure that the treatment will be cost-effective. Collecting cost data simultaneouslywhile executing an RCT allows investigators topiggyback an economic evaluation to the trial.27

Important steps in performing an economicanalysis include deciding which costs (direct or in-direct) and consequences (a single outcome likerecurrence or a patient-important outcome mea-sure like the DASH) to be measured. Furthermore,a decision must be made on which economicperspective to take: the Patient, a Hospital/ClinicalPractice, a Third Party Payer (like Medicare or theNational Health Service), or Society. Guidelines onconducting and interpreting the results of an eco-nomic evaluation in plastic and hand surgery areavailable.27–29

An economic evaluation was not performed inthe van Rijssen and associates’ RCT.6 Only 95economic evaluations comparing 2 plastic surgeryinterventions have been published between 1986and 2012; 79% of these evaluations were foundto be only partial economic evaluations (cost ana-lyses).30 In a systematic review of hand surgerystudies, Chung and colleagues31 identified a spe-cific need for more research on economic analysis.Therefore, in an era of diminishing health care re-sources, there is an opportunity to incorporateeconomic evaluations into surgical clinicalresearch and contain health care costs.

SUMMARY

Results of surgical interventions are viewed ashighly variable due to known and unknown patientfactors, as well as differences in preoperative care,

post-operative care, and surgeon technique andexperience. When interpreting the results of a trial,it is important to perform amethodical assessmentto determine whether a study is providing valid re-sults and how these results can be applied in yourpatient’s clinical context. Although cliniciansperform this general process when discussingthe risks and benefits of a therapy with their pa-tient, the ability to interpret and understand thestrength of the available evidence is essential.This process will help surgeons incorporateevidence-based surgery into their practice andimprove patient care.

REFERENCES

1. Evidence-Based Medicine Working Group. Evi-

dence-based medicine. A new approach to the

teaching of medicine. JAMA 1992;268(17):2420–5.

2. Waljee J, Larson BP, Chung KC. Measuring treat-

ment effectiveness: a guide to incorporating the

principles of evidence-based medicine. Plast Re-

constr Surg 2012;130(6):1382–94.

3. Guyatt G, Rennie D, Meade M, et al. Users’ guides

to the medical literature: a manual for evidence-

based clinical practice. 2nd edition. New York:

McGraw-Hill; 2008.

4. Bhandari M, Haynes RH. How to appraise the effec-

tiveness of treatment. World J Surg 2005;29:570–5.

5. Graham B. Strategies for nonrandomized clinical

research in hand surgery. Clin Plast Surg 2005;32:

529–36.

6. van Rijssen AL, ter Linden H, Werker PM. Five-year

results of a randomized clinical trial on treatment in

Dupuytren’s disease: percutaneous needle fasciot-

omy versus limited fasciectomy. Plast Reconstr

Surg 2012;129(2):467–77.

7. Thoma A, Farrokhyar F, Bhandari M, et al. Users’

guide to the surgical literature how to assess a ran-

domized controlled trial in surgery. Can J Surg 2004;

47(3):200–8.

8. Thoma A. Challenges in creating a good random-

ized controlled trial in hand surgery. Clin Plast

Surg 2005;32:563–73.

9. Cadeddu M, Farrokhyar F, Levis C, et al, for the

Evidence-Based Surgery Working Group. Users’

guide to the surgical literature: understanding confi-

dence intervals. Can J Surg 2012;55(3):207–11.

10. Jaeschke R, Singer J, Guyatt GH. Ascertaining the

minimal clinically important difference. Control Clin

Trials 1989;10:407–15.

11. Cadeddu M, Farrokhyar F, Thoma A, et al, for the

Evidence-Based Surgery Working Group. Users’

guide to the surgical literature: how to assess power

and sample size. Can J Surg 2008;51(6):476–82.

12. Cohen J. Statistical power analysis in the behaviou-

ral sciences. Hillsdale (NJ): Erlbaum; 1988.

Voineskos et al292

13. Norman GR, Sridhar FG, Guyatt GH, et al. Relation

of distribution and anchor-based approaches in

interpretation of changes in health-related quality

of life. Med Care 2001;39(10):1039–47.

14. Norman GR, Sloan JA, Wyrwich KW. Interpretation of

changes in health-related quality of life: the remark-

able universality of half a standard deviation. Med

Care 2003;41(5):582–92.

15. van Rijssen AL, Gerbrandy FS, ter Linden H, et al.

A comparison of the direct outcomes of percuta-

neous needle fasciotomy and limited fasciectomy

for Dupuytren’s disease: a 6-week follow-up study.

J Hand Surg Am 2006;31:717–25.

16. Beaton DE, Davis AM, Hudak P, et al. The DASH

(disabilities of the arm, shoulder and hand) outcome

measure: what do we know about it now? Br J Hand

Ther 2001;6(4):109–18.

17. Hunsaker FG, Cioffi DA, Amadio PC, et al. The Amer-

ican academy of orthopaedic surgeons outcomes

instruments: normative values from the general pop-

ulation. J Bone Joint Surg Am 2002;84(2):208–15.

18. Beaton DE, van Eerd D, Smith P, et al. Minimal change

is sensitive, less specific to recovery: a diagnostic

testing approach to interpretability. J Clin Epidemiol

2011;64(5):487–96.

19. Alderman AK, Chung KC. Measuring outcomes in

hand surgery. Clin Plast Surg 2008;35:239–50.

20. Giladi AM, Chung KC. Measuring outcomes in hand

surgery. Clin Plast Surg 2013;40:313–22.

21. Hudak PL, Amadio PC, Bombardier C. Development

of an upper extremity outcome measure: the DASH

(Disabilities of the Arm, Shoulder, and Head). Am J

Ind Med 1996;29:602–8.

22. Chung KC, Pillsbury MS, Walters MR, et al. Reli-

ability and validity testing of the Michigan Hand Out-

comes Questionnaire. J Hand Surg Am 1998;23(4):

575–87.

23. Levine DW, Simmons BP, Koris MJ, et al. A self-

administered questionnaire for the assessment of

severity of symptoms and functional status in carpal

tunnel syndrome. J Bone Joint Surg Am 1993;

75(11):1585–92.

24. Furlong WJ, Feeny DH, Torrance GW, et al. The

Health Utilities Index (HUI) system for assessing

health-related quality of life in clinical studies. Ann

Med 2001;33(5):375–84.

25. Thoma A, Kaur MN, Ignacy TA, et al. Psychometric

properties of health related quality of life instruments

in patients undergoing palmar fasciectomy for Du-

puytren’s disease: a prospective study. Hand

2014;9(2):166–74.

26. Beaudreuil J, Allard A, Zerkak D, et al, URAM Study

Group. Unite Rhumatologique des Affections de la

Main (URAM) scale: development and validation of

a tool to assess Dupuytren’s disease-specific

disability. Arthritis Care Res (Hoboken) 2011;

63(10):1448–55.

27. Thoma A, Sprague S, Tandan V. Users’ guide to the

surgical literature: how to use an article on economic

analysis. Can J Surg 2001;44(5):347–54.

28. Thoma A, Strumas N, Rockwell G, et al. The use of

cost-effectiveness analysis in plastic surgery clinical

research. Clin Plast Surg 2008;35:2850296.

29. Kotsis SV, Chung KC. Fundamental principles of

conducting a surgery economic analysis study. Plast

Reconstr Surg 2010;125:727–35.

30. Ziolkowski NI, Voineskos SH, Ignacy TA, et al. Sys-

tematic review of economic evaluations in plastic

surgery [systematic review]. Plast Reconstr Surg

2013;132(1):191–203.

31. Chung KC, Burns PB, Sears ED. Outcomes

research in hand surgery: where have we been

and where should we go? J Hand Surg Am 2006;

31:1373–9.