Provider Factors Associated with Intramedullary Nail Use for Intertrochanteric Hip Fractures

Provider Factors Associated with IntramedullaryNail Use for Intertrochanteric Hip FracturesBy Mary L. Forte, PhD, DC, Beth A. Virnig, PhD, MPH, Lynn E. Eberly, PhD, Marc F. Swiontkowski, MD,

Roger Feldman, PhD, Mohit Bhandari, MD, MSc, FRCSC, and Robert L. Kane, MD

Investigation performed at the University of Minnesota, Minneapolis, Minnesota

Background: Intramedullary nails provide no clear outcomes benefit in the majority of patients with intertrochanteric hipfracture, yet their use in the United States continues to increase. Non-patient factors that are associated with intra-medullary nail use among Medicare patients have not been examined. The goal of this study was to identify the surgeonand hospital characteristics that were associated with the use of intramedullary nails compared with plate-and-screwdevices among elderly Medicare patients with intertrochanteric hip fractures.

Methods: Medicare beneficiaries who were sixty-five years of age or older and underwent inpatient surgery to treat anintertrochanteric femoral fracture with use of an intramedullary nail or a plate-and-screw device were identified from theUnited States Medicare files for 2000 to 2002. Surgeon and hospital characteristics from the Medicare provider en-rollment files were merged with the claims. Generalized linear mixed models with fixed and random effects modeled theassociation between surgeon and hospital factors and intramedullary nail use (compared with plate and screws), con-trolling for patient age, sex, and race; subtrochanteric fracture; Charlson comorbidity score; nursing home residence; andMedicaid-administered assistance. The adjusted odds ratios of receiving an intramedullary nail by year, surgeon, andhospital factors are reported.

Results: There were 192,365 claims for surgery to treat an intertrochanteric hip fracture that met the inclusion criteriaand matched with surgeon and hospital information. There were 15,091 surgeons who performed intertrochanteric hipfracture surgeries in Medicare patients in 3480 hospitals between March 1, 2000, and December 31, 2002. Thesurgeon factors associated with intramedullary nail use include younger surgeon age (less than forty-five years old), anosteopathy degree, and operating at more than one hospital. The hospital factors associated with intramedullary nail useinclude a higher volume of intertrochanteric hip fracture surgeries, teaching hospital status, and having resident as-sistance during surgery. Surgeon factors improved the model fit more than hospital factors.

Conclusions: The use of intramedullary nails was strongly associated with early-career surgeons and surgeon trainingprograms. Our findings suggest that orthopaedic faculty at teaching hospitals and younger surgeons may be selectingorthopaedic implants on the basis of factors other than clinical outcomes evidence. We expect that intramedullary nailuse will continue to increase as long as new surgeons are preferentially trained in intramedullary nailing procedures andsurgeon reimbursement remains insulated from the treating hospital’s burden of their choices for higher cost devicesunder the Medicare payment system.

Level of Evidence: Prognostic Level II. See Instructions to Authors for a complete description of levels of evidence.

A commentary by Saam Morshed, MD, MPH,

and Kevin J. Bozic, MD, MBA, is available atwww.jbjs.org/commentary and as supplementalmaterial to the online version of this article.

Disclosure: In support of their research for or preparation of this work, one or more of the authors received, in any one year, outside funding or grants inexcess of $10,000 from the Centers for Medicare and Medicaid Services to the Research Data Assistance Center (contract 500-01-0043). In addition,one or more of the authors or a member of his or her immediate family received, in any one year, payments or other benefits in excess of $10,000 or acommitment or agreement to provide such benefits from commercial entities (United Health Group, Medtronic, and Lewin and Associates) and paymentsor other benefits of less than $10,000 or a commitment or agreement to provide such benefits from a commercial entity (SCAN Health Plan).

1105

COPYRIGHT � 2010 BY THE JOURNAL OF BONE AND JOINT SURGERY, INCORPORATED

J Bone Joint Surg Am. 2010;92:1105-14 d doi:10.2106/JBJS.I.00295

More than 280,000 hip fractures occur annually in theUnited States, and most affect elderly individuals1.Geriatric hip fractures are associated with high

morbidity and mortality2-8. Intertrochanteric hip fracturesaccount for nearly half (47%) of all hip fractures in elderlyindividuals9-12.

The majority of intertrochanteric hip fractures aretreated surgically with one of two types of internal fixationdevices: a plate with screws or an intramedullary nail. Thereare no outcomes benefits for intramedullary nails comparedwith plate and screws in the treatment of stable intertro-chanteric hip fractures13-17, which account for up to 70% ofsuch fractures18-21. Intramedullary nails are preferred by manysurgeons for patients with intertrochanteric hip fractures withsubtrochanteric extension and isolated subtrochanteric frac-tures22-24, although definitive outcomes support for the use ofintramedullary nail devices in unstable fractures has not beenestablished16,17,24-27.

Since 1992, Medicare has paid surgeons more to useintramedullary nail devices than a plate and screws because ofthe higher assigned relative value units (RVUs) for intramed-ullary nailing procedures. In contrast, hospitals are paid underthe same diagnosis-related group for both the intramedullarynail and plate-and-screw procedures, although the intramed-ullary nail devices cost hospitals more than plate-and-screwimplants13,28,29. Despite equivalent outcomes and higher implant-related complication rates16,17,25,30,31, intramedullary nail use inthe United States has continued to increase and is now thedominant mode of internal fixation in the treatment of pa-tients with an intertrochanteric hip fracture28,32,33. Addition-ally, substantial geographic variation in intramedullary nail usehas been identified among Medicare patients with an inter-trochanteric hip fracture, which was not explained by patient-related factors21,28.

Non-patient-related factors that are associated with in-creased intramedullary nail use among Medicare patients havenot been examined. The goal of this study was to identify thesurgeon and hospital factors that were associated with intra-medullary nail use among Medicare patients with an inter-trochanteric hip fracture treated with internal fixation during2000 to 2002, controlling for patient factors. We hypothesizedthat experienced orthopaedic surgeons and surgeons inteaching hospitals were most likely to use an intramedullarynail for intertrochanteric hip fracture treatment during 2000 to2002.

Materials and MethodsData Sources and Patient Selection

Medicare claims and enrollment data from March 1, 2000,through December 31, 2002, were used to identify

Medicare beneficiaries who were sixty-five years of age orolder, were enrolled in Parts A and B, were not enrolled inmanaged care plans, and underwent inpatient surgery to treatan intertrochanteric hip fracture with use of either an intra-medullary nail or a plate-and-screw device. We used 2000 to2002 as the study period because intramedullary nail devices

were in widespread use in the United States and did not un-dergo design modifications from 1998 through 2002.

Patients were included if there was a specific, five-digitICD-9 (International Classification of Diseases, Ninth Revi-sion34) diagnosis code for a pertrochanteric femoral fracture(820.2x to 820.3x) on either the hospital or the physician claimfor a particular surgical procedure, the corresponding claim inthe other provider file had one of four ICD-9 diagnosis codes(820.2x, 820.3x, 820.8, or 820.9), and the patient was treatedwith either an intramedullary nail or plate-and-screw deviceby CPT (Current Procedural Terminology35) codes 27245 or27244, respectively.

Patient data were obtained from the Medicare ProviderAnalysis and Review hospital claim files, the Medicare Part B(physician) claim files, and the Denominator (enrollment)files. Surgeon and hospital characteristics were determinedfrom the Medicare Physician Identification and EligibilityRegistry file and the Medicare Provider of Services facility file,and merged with the claims with use of the unique Medicareprovider numbers.

Surgery dates were matched by patient identifier acrossthe hospital and physician claims within a ±7-day time frame36.For patients with more than one intertrochanteric hip fractureduring the study period, only the first surgical claim was re-tained. Since our interest was in index, low-energy hip frac-tures in elderly patients, we excluded patients whose hospitalclaim indicated surgery for complications or revisions ofprevious hip surgery, metastatic or associated cancers, infec-tion, bilateral fracture, or high-energy trauma. Additional caseselection details have been previously reported21.

The study was reviewed by the University of Minnesotainstitutional review board and was determined to be exempt.

VariablesThe dichotomous outcome of interest was the type of implantused, either an intramedullary nail or plate and screws. Theprocedure code on the surgeon’s claim enabled us to determinethe specific internal fixation device used in each patient. Theunit of analysis was the patient. Surgeon and hospital char-acteristics were used as predictors while controlling for patientfactors. A year indicator was also included to account forchanges in intramedullary nail use over time.

Each surgeon’s volume of intertrochanteric hip fracturesurgeries in Medicare patients was calculated as the totalnumber of internal fixation procedures performed per surgeonprovider number between March 1, 2000, and December 31,2002. Surgeon volume was categorized into quartiles to facil-itate comparison. As an estimate of surgical experience, eachsurgeon’s age in years was calculated from his or her date ofbirth in the Medicare Physician Identification and EligibilityRegistry file to December 31, 2001, and was then divided intofive-year age categories. Orthopaedic board certification status(yes or no) and professional degree (MD or DO [doctor ofosteopathy]) were determined from the Medicare PhysicianIdentification and Eligibility Registry file, while practicestructure (group or other) was determined from both the

1106

TH E J O U R N A L O F B O N E & JO I N T SU R G E RY d J B J S . O R G

VO LU M E 92-A d NU M B E R 5 d M AY 2010PR O V I D E R FAC T O R S A S S O C I AT E D W I T H IN T R A M E D U L L A RY NA I L US E

F O R IN T E R T R O C H A N T E R I C HI P F R AC T U R E S

Medicare Physician Identification and Eligibility Registry fileand the surgeons’ claims. The number of hospitals where eachsurgeon surgically treated patients with an intertrochanterichip fracture (number of case hospitals) was determined fromthe number of unique hospital identifiers per surgeon identi-fier (1, 2, 3, and ‡4) in the claims.

Each hospital’s volume of intertrochanteric hip fracturesurgeries in Medicare patients was calculated as the totalnumber of Medicare patients who had internal fixation for anintertrochanteric hip fracture per hospital identifier during thethirty-four-month study period. Each hospital’s case volumewas then grouped into a four-level categorical variable byvolume quartiles. The type of hospital ownership (for profit,nonprofit, or government) and teaching status (nonteaching,teaching without any resident assistance in cases, or teachingwith resident assistance in one case or more) were determinedfrom the Provider of Services file, and medical resident caseinvolvement under the direction of a teaching physician wasdetermined by the presence of a –GC modifier on the proce-dure code from the surgeon’s Part B claim35,37.

Our models contained patient factors that potentiallyinfluence device choice, including age (sixty-five to seventy-four years, seventy-five to eighty-four years, and eighty-fiveyears or more), sex, race (white, black, or other), residence in anursing home before the fracture (yes or no), subtrochantericfracture, and state Medicaid-administered assistance. Weconsidered patients to have resided in a nursing home at thetime of fracture if we identified at least one Medicare Part Bprovider claim that took place in a nursing home within twomonths preceding the date of hip surgery using the Place ofService code from any provider’s Part B claim38. To control fordifferences in socioeconomic status, patients who receivedstate Medicaid-administered assistance during their hospitalstay were identified from monthly enrollment indicators in theDenominator file. We then created a four-level categoricalvariable by cross-classifying preadmission residence (nursinghome or other) with state Medicaid-administered assistancestatus (yes or no) to avoid potential bias from nursing homepatients in the regression coefficient for state Medicaid-administered assistance. The adapted Charlson comorbidityscores (0, 1, 2, or ‡3)39,40 were excluded from our final modelsbecause of the lack of significance in predicting device choicein all preliminary models, which was consistent with priorwork on patient factors that used device choice as theoutcome21.

Separate models were estimated by calendar year andthen pooled for the final analyses because of the consistency ofour findings over time. We excluded cases with surgical datesprior to March 1, 2000, to allow for the two-month nursingfacility claim look-back, which resulted in a thirty-four-monthanalytic file.

Statistical AnalysesWe used generalized linear mixed models with fixed and ran-dom effects to model the association between surgeon andhospital factors and the outcome of device choice (intramed-

ullary nail or plate and screws), while controlling for patientfactors. Random effects models are typically used to studyvariation, rather than simply control for it. These models canaccount for the correlated nature of the data and the imperfectnesting of surgeons within hospitals41,42. Patients in this studywere perfectly nested within surgeons, but surgeons oftenoperated at several hospitals, with just over half of the surgeonsoperating at only one hospital during the study period. Like-lihood ratio tests were used to compare nested models41. Non-nested models were compared with use of information criteria,which penalizes the log likelihood for the number of modelparameters (Akaike information criterion; AIC) or subjects(Bayesian information criterion; BIC), with smaller valuesindicating better fit of the model to the data41.

We considered three types of models: fixed (means)without random intercepts, random hospital intercepts, andrandom surgeon intercepts. We examined four groups ofpredictors, each with and without random surgeon or hospitalintercepts, for a total of twelve models, including an intercept-only model; patient factors with year; patient and surgeonfactors with year; and patient, surgeon, and hospital factorswith year, for which Akaike and Bayesian information criteriamodel fit estimates are reported. Analyses were conducted withuse of the SAS statistical program, NLMIXED procedure(version 9.1 for Windows; SAS Institute, Cary, North Caro-lina). Coefficient estimates, standard errors, odds ratios, and95% confidence intervals for the odds ratios are reported forthe regression results. Descriptive statistics are reported as thepercentage, mean, or median.

Source of FundingThe Centers for Medicare and Medicaid Services provided thedata for this study but did not play a role in the investigation.

ResultsSample Description

Between March 1, 2000, and December 31, 2002, 192,365Medicare claims for intertrochanteric hip fracture sur-

geries met our inclusion criteria and matched with completesurgeon and hospital information (98.2% matched). Inter-trochanteric hip fracture surgeries with internal fixation wereperformed by 15,091 surgeons in 3480 United States hospitalsfor fee-for-service Medicare beneficiaries during that timeperiod. The proportion of patients with an intertrochanterichip fracture who received an intramedullary nail comparedwith a plate and screws increased from 8.1% of the patients in2000 to 17.1% in 2002.

The provider characteristics are described in Table I. Themedian age among surgeons was forty-six years. Over thethirty-four-month study period, the median surgeon volumeof intertrochanteric hip fractures was ten cases (average, 3.4cases per year); 7.8% of surgeons operated on only one patientduring the study. Surgeons with an MD degree composed95.4% of the providers, and surgeons were almost exclusivelyself-identified as orthopaedic surgeons (98.4%). Nearly two-thirds (64.9%) of surgeons were orthopaedic board certified;

1107




64.2% practiced in a group practice. Just over half of surgeons(55%) operated at only one hospital, while 14.2% of surgeonsoperated at three or more hospitals during the study.

The median number of Medicare patients treated for anintertrochanteric hip fracture per hospital was forty-one overthirty-four months, or approximately fourteen patients peryear (Table I). Eighty hospitals (2.3%) treated only one patientwho had a hip fracture with internal fixation during the studyperiod. Two-thirds (66.9%) of the case hospitals were non-profit facilities. The majority of hospitals (69.1%) were notteaching facilities. Only 150 hospitals (4.3%) were teachinghospitals that had resident involvement in surgical cases, andan additional 26.6% of the hospitals were identified as teachinghospitals without resident assistance during hip fracturesurgeries.

The patients were, on the average, eighty-four years old(Table II); 77.2% were female and 94.6% were white. Sub-trochanteric fractures were identified in 10.6% of the patients.One-fifth (20.1%) of the patients with a hip fracture wereadmitted from a nursing home. The majority (80.8%) of pa-tients did not receive state Medicaid-administered assistance.

Surgeon Random Intercepts ModelThe results from the best-fitting model, the surgeon randomintercepts model with the full set of predictors, are shown inTable III. This regression model shows the average effect ofeach predictor across surgeons, but additionally it allows eachsurgeon to have their own intercept, or baseline probability ofintramedullary nail use. The odds of a patient with an inter-trochanteric hip fracture receiving an intramedullary nail de-vice compared with a plate and screws in 2001 were 1.77 timeshigher than in 2000, and nearly five times higher in 2002compared with 2000 (odds ratio, 4.68; 95% confidence inter-val, 4.40 to 4.97).

There was an inverse relationship between the age of thesurgeon and intramedullary nail use. After controlling forpatient and other provider factors, surgeons under the age offorty-five years had higher odds of intramedullary nail usecompared with surgeons in the median age group (forty-five toforty-nine years old), with the highest comparative odds ofintramedullary nail use among the youngest group of surgeons(less than thirty-five years old) (odds ratio, 5.13; 95% confi-dence interval, 3.87 to 6.81). Surgeons who were fifty years orolder had significantly lower odds of selecting an intramed-ullary nail than did surgeons who were forty-five to forty-nineyears old; however, the standard errors for surgeons who werefifty years or older were slightly larger than for those who wereless than forty-five years old, indicating larger variability indevice choice among the older surgeons.

Surgeons’ volume of patients with an intertrochanterichip fracture was not significantly associated with device se-lection. The odds of receiving an intramedullary nail deviceamong patients of surgeons who performed four or feweroperations over the thirty-four months were not significantlydifferent from the odds of receiving an intramedullary nailamong patients of the highest-volume surgeons (eighteen

TABLE I Surgeon and Hospital Characteristics Among Medicare

Patients Treated for Intertrochanteric Hip Fracture

from 2000 to 2002*

VariablePercentage of

Providers

Surgeon factors for 15,091 surgeons

Surgeon age

<35 yr 8.57

35-39 yr 17.13

40-44 yr 17.94

45-49 yr 18.02

50-54 yr 14.44

55-59 yr 12.02

60-64 yr 7.96

‡65 yr 3.92

No. of case hospitals†

1 54.98

2 30.83

3 9.97

‡4 4.22

Surgeon case volume*

1-4 24.62

5-10 26.60

11-17 22.83

‡18 26.05

Professional degree

Medical doctor (MD) 95.44

Doctor of osteopathy (DO) 4.56

Orthopaedic board certification

Board certified 64.92

Not board certified 35.08

Practice structure

Group practice 64.24

Other 35.76

Hospital factors for 3480 hospitals

Teaching status

Nonteaching 69.08

Teaching—no residentassistance with case(s)

26.61

Teaching—resident assistancewith case(s)

4.31

Hospital case volume*

1-17 25.66

18-41 24.68

42-78 24.94

‡79 24.71

Type of ownership

Nonprofit 66.90

For profit 14.77

Government 18.48

*Patients treated with internal fixation from March 1, 2000, throughDecember 31, 2002. †The number of hospitals at which each surgeonperformed internal fixation to treat Medicare fee-for-service patients withan intertrochanteric hip fracture during the study period.

1108




patients or more). Although not significant, there was a U-shaped trend in the odds ratios of intramedullary nail use bysurgeon volume categories, with lower values among mid-volume surgeons relative to the highest and lowest-volumeproviders.

The odds ratios of intramedullary nail use increased asthe number of hospitals where a surgeon performed suchprocedures increased. Patients of surgeons who performed hipfracture surgeries at four or more hospitals had more thandouble the odds of receiving an intramedullary nail device thandid patients of surgeons who operated at only one hospital,controlling for other factors (odds ratio, 2.4; 95% confidenceinterval, 1.76 to 3.38).

After controlling for surgeon age and other patient andprovider factors, patients treated by doctors of osteopathy hadtwice the odds of receiving an intramedullary nail device thandid patients of medical doctors.

Orthopaedic board certification and practice structurewere not significantly associated with intramedullary nail use.A provider specialty indicator variable (orthopaedic or other)was omitted from the final models because it was not signifi-

cantly related to device choice in all preliminary models amongthis almost exclusively orthopaedic provider group.

Hospitals’ volume of Medicare patients with an inter-trochanteric hip fracture was positively associated with intra-medullary nail use (Table III). Compared with the highestvolume hospitals (seventy-nine operations or more overall andcorresponding to twenty-eight operations or more per year),patients surgically treated in hospitals that performed less thanthe median volume of forty-one operations during the studyhad 16% to 17% lower odds of receiving an intramedullarynail, controlling for other factors. Hospitals that performed justabove the median number of operations (third volume quartile)were equally as likely to use intramedullary nails as hospitals inthe highest volume quartile for patients with a hip fracture.

Patients treated at teaching hospitals that did not haveresident assistance during surgeries had 12% higher odds ofreceiving an intramedullary nail for the fracture than patientstreated at nonteaching facilities. Moreover, patients who hadthe surgery performed at a teaching hospital with residentassistance during surgery had 58% higher odds of receiving anintramedullary nail device than patients treated at nonteachinghospitals, controlling for patient and other provider factors.Intramedullary nail use did not significantly differ by the typeof hospital ownership.

Patient CovariatesPatients with a subtrochanteric fracture compared with otherpertrochanteric fractures had substantially higher odds of re-ceiving an intramedullary nail (odds ratio, 17.51; 95% confi-dence interval, 16.47 to 18.61; see Appendix), although onlyone-third of patients with a subtrochanteric fracture receivedan intramedullary nail during this period. Consistent withprevious work21, the influence of other patient factors on in-tramedullary nail use was generally modest and tended tocontradict the device use expectations based on claim factorsthat correlate with clinical factors, such as low bone density43,44.

Model ComparisonsOur findings are highly consistent across all three models(Table III and Appendix). Regardless of how we modeled theprovider-device association, we arrived at the same conclu-sions regarding the influence of various provider predictors onintramedullary nail use. Although the magnitude of each fac-tor’s influence differed across models, the direction of the in-fluence of each predictor on intramedullary nail use essentiallydid not change, regardless of the model used. The odds ratiosfor the influence of the surgeon’s age, his or her professionaldegree, the number of hospitals where the surgeon operated,the hospital’s teaching status, and the hospital’s case volumewere particularly stable across all models. The effect of thesurgeon’s case volume was similar across the three models, butit was not a significant predictor of device choice in the modelwith surgeon random intercepts, suggesting a great deal ofvariation in intramedullary nail use among providers withinthe surgeon volume strata. The odds ratios of intramedullarynail use, particularly among young surgeons (less than forty

TABLE II Characteristics of Medicare Patients with an

Intertrochanteric Hip Fracture Treated with

Internal Fixation from 2000 to 2002*

Patient Factor Percentage of Patients

Age

65-74 yr 12.31

75-84 yr 39.43

‡85 yr 48.27

Sex

Male 22.84

Female 77.16

Race

White 94.57

Black 3.15

Other 2.29

Type of fracture

Subtrochanteric 10.59

Other pertrochanteric 89.41

Residence and stateassistance status

Non-nursing home,without state assistance

68.27

Nursing home,without state assistance

12.51

Non-nursing home,with state assistance

11.68

Nursing home,with state assistance

7.54

*Data on 192,365 fee-for-service cases between March 1, 2000,and December 31, 2002.

1109




TABLE III Surgeon and Hospital Factors Associated with Intramedullary Nail Use Compared with Plate and Screws for Treatment of

Medicare Patients with an Intertrochanteric Hip Fracture: Surgeon Random Intercepts Model*

Odds Ratio† 95% Confidence Interval for Odds Ratio

Surgeon factors

Surgeon age

<35 yr 5.13‡ 3.87-6.81

35-39 yr 2.58‡ 2.05-3.24

40-44 yr 1.40 1.17-1.75

45-49 yr§ 1.00 –

50-54 yr 0.79 0.62-0.99

55-59 yr 0.61 0.47-0.79

60-64 yr 0.70 0.52-0.94

‡65 yr 0.64 0.42-0.96

No. of case hospitals#

1§ 1.00 –

2 1.21 1.03-1.41

3 1.47‡ 1.17-1.84

‡4 2.44‡ 1.76-3.38

Surgeon case volume**

1-4 1.23 0.99-1.52

5-10 0.88 0.74-1.06

11-17 0.85 0.71-1.03

‡18§ 1.00 –

Professional degree

Medical doctor (MD)§ 1.00 –

Doctor of osteopathy (DO) 2.06‡ 1.51-2.81

Orthopaedic board certification

Board certified§ 1.00 –

Not board certified 1.13 0.97-1.31

Practice structure

Group practice§ 1.00 –

Other 0.99 0.87-1.14

Hospital factors

Teaching status

Nonteaching§ 1.00 –

Teaching—no resident assistance with case(s) 1.12 1.03-1.22

Teaching—with resident assistance with case(s) 1.58‡ 1.26-2.00

Hospital case volume#

1-17 0.83 0.70-0.98

18-41 0.84 0.75-0.94

42-78 0.94 0.86-1.02

‡79§ 1.00 –

Type of ownership

Nonprofit§ 1.00 –

For profit 1.10 0.99-1.23

Government 1.05 0.93-1.19

Year

2000§ 1.00 –

2001 1.77‡ 1.67-1.88

2002 4.68‡ 4.40-4.97

*Controlling for patient age, sex, race, nursing home residence, subtrochanteric fracture, and Medicaid assistance. †A value in boldface indicates a significantdifference (p < 0.05). ‡P < 0.0001. §Reference level. #The number of hospitals at which each surgeon performed internal fixation to treat Medicare fee-for-servicepatients with an intertrochanteric hip fracture during the study period.**Medicare patients with intertrochanteric hip fracture treated with intramedullary nail or plate and screws between March 1, 2000, and December 31,2002.

1110




years old), showed dramatic increases with use of the surgeonrandom intercepts model.

Model FitThe models with either surgeon or hospital random interceptshad significantly better fit than the models without randomeffects (p < 0.0001 for likelihood ratio tests; results not shown).The surgeon-specific random intercepts models better fit thedata (lower information criteria) than the hospital-specificrandom intercepts models did. The best-fitting model con-tained all patient, surgeon, and hospital predictors withsurgeon random intercepts (see Appendix). However, afterinclusion of the surgeon-specific intercepts, the addition ofthe hospital predictor variables only minimally improved themodel fit.

Discussion

We found a strong association between surgeon factors,teaching hospital status (with and without resident in-

volvement in surgery), and hospital volume and intramedul-lary nail use in the treatment of intertrochanteric hip fracturein Medicare patients during 2000 to 2002. Surgeon factorsaccounted for more of the variation in intramedullary nail usethan the hospital factors we examined.

Younger surgeons were the primary adopters of the in-tramedullary nailing procedures, likely reflecting higher in-tramedullary nail use in surgeries performed at the teachinghospitals that trained them. Given the seventeen-year durationof higher RVUs for intramedullary nailing procedures largelybecause of the work RVU component, we expected the odds ofintramedullary nail use to be higher among mid-career, board-certified surgeons in group practices45,46 who had attained somedegree of experience and surgical proficiency that would en-able them to master a new procedure that Medicare estimatedwould require more planning and technical skills, greater judg-ment, and a longer time to perform than plate-and-screw pro-cedures for intertrochanteric hip fracture. Experienced teachingsurgeons who mentored and educated residents first had toadopt intramedullary nailing procedures in order to trainyoung surgeons in the technique. By the time of this study,however, we found the highest intramedullary nail use amongthe youngest surgeons, presumably with the least experience,implying that our study took place somewhere during mid-diffusion of intramedullary nail technology into hip fracturetreatment. Our findings suggest that younger surgeons con-tinue to perform the procedures that they become familiarwith in training, while older surgeons, who may be morecomfortable or proficient with plate-and-screw procedures,were less likely to change to the newer intramedullary naildevices during 2000 to 2002.

Assuming that both surgical skill and judgment improvewith experience and given the intramedullary nail adoptionpattern found among the youngest and least experiencedsurgeons, these findings challenge Medicare’s estimate through2008 that intramedullary nailing procedures require superiorsurgical skills and surgeon judgment. Our finding of high in-

tramedullary nail use among young surgeons is consistent withrecent patterns of intramedullary nail use noted in the AmericanBoard of Orthopaedic Surgery data28.

The precise mechanisms that account for our finding ofhigher intramedullary nail use among osteopathic surgeonscompared with surgeons with a medical degree in all modelssuggest that there are additional, unmeasured factors that dif-ferentially affect decisions by osteopathic surgeons in their im-plant choices for patients with an intertrochanteric hip fracture.

After controlling for patient and other provider factors,we found higher odds ratios of intramedullary nail use amongsurgeons who operated at more than one hospital, particularlyamong surgeons who operated at four or more hospitals.Compared with surgeons who operated at one hospital, thosewho operated at four or more hospitals were younger (a me-dian age of forty-two years compared with forty-seven years),had a higher average case volume (seventeen compared withseven cases), and more often held a degree in osteopathycompared with a medical degree (10.2% compared with3.7%), but the difference persisted even when surgeon age andother factors were controlled in our models. Lower use ofintramedullary nails among surgeons who operated at onehospital may reflect their greater cognizance of the facility costsincurred by their device choices within their host hospitals,rather than their slightly higher median age.

One-third of the hospitals in this study were staffed by atleast one surgeon who operated at four or more hospitals, andthese hospitals tended to be moderate-to-high volume, non-government facilities. A greater proportion of the hospitalsused by surgeons who operated at four or more hospitals wereteaching facilities, and these teaching hospitals had the sameproportion of cases with surgical resident involvement com-pared with hospitals that did not have cases by surgeons whooperated at four or more hospitals. These findings, combinedwith higher intramedullary nail use in high-volume hospitals,suggest that factors other than orthopaedic surgeon coverageor resident availability, such as faster procedure times28,47, maybe a factor in intramedullary nail use that allows more rapidoperating-room turnover in high-volume hospitals, thereby atleast partially offsetting the higher cost of intramedullary naildevices to these hospitals24,28,29.

The modestly higher use of intramedullary nails amongteaching hospitals compared with nonteaching hospitals re-flects their earlier adoption of new technology in general46,48.There are several possible explanations for earlier adoption oftechnology among teaching hospitals. Academic surgeons atteaching hospitals may feel responsible for educating surgeontrainees in state-of-the-art techniques. Academic surgeons mayalso be more likely to work closely with manufacturers todesign and test new devices49,50. The orthopaedic device in-dustry may provide surgeons in teaching hospitals with in-tensive marketing and operating-room attention for selectdevices in the hopes of training a new cohort of surgeons inthese procedures. Although teaching hospitals may treat pa-tients with more severe case-mix profiles than nonteachingfacilities do, the substantially higher odds of receiving an in-

1111




tramedullary nail device in hospitals with resident involvementin cases, compared with those without resident involvement,suggest that there are differences in patient selection for op-erations in which resident involvement is allowed or otherintramedullary nail-related incentives that uniquely affectresident-supervising surgeons that we were not able to accountfor in this analysis. Ironically, both younger surgeons andteaching surgeons who are expected to have greater under-standing of evidence-based medicine principles than older andnonteaching surgeons51 were most likely to select intramed-ullary nail implants to treat intertrochanteric femoral fracturesin this study.

Two important strengths of this claims-based study arethe large number of patients with a hip fracture for whom wewere able to evaluate device-related provider patterns and theinclusion of multiple patient and provider variables to predictthe device choice outcome.

We acknowledge several important limitations of ourstudy. Our analysis incorporated only one random effect at atime, which limited our ability to analyze provider patterns withhigher degrees of nesting. However, given the stability of thedirection of the findings across the models examined, we do notexpect that alternative models would change the overall findings.

The patients in this study were in fee-for-service Medi-care only; full provider volumes, especially in areas of highenrollment in managed care, are not reflected. However, giventhe insignificant effect of surgeon volume on intramedullarynail use, potential underclassification of surgeon volume inseveral states that had high enrollment in Medicare managedcare in 2000 to 2002 would most likely not change our overallsurgeon volume findings. We would expect any misclassifica-tion of hospital volume to enhance or not substantially changeour hospital volume findings, given the positive effect of vol-ume on intramedullary nail use. We would not expect sub-trochanteric fracture rates or fracture patterns that may impactdevice selection to differ by the type of insurance.

We were unable to account for other factors that mayinfluence the surgeon’s device choice for Medicare patientswith an intertrochanteric hip fracture, including specific fracturepatterns, procedure speed and ease, marketing and related deviceindustry influences, clinical patient features, and the surgeon’sopinion about which fracture patterns are unstable, none ofwhich are available in claims-based studies. In the presence ofpreferential training in intramedullary nailing procedures, wedo not expect that inclusion of data on fracture patterns wouldchange our findings. We do not expect systematic variation inunstable fractures that would favor newly trained surgeons orteaching and high-volume hospitals, particularly for what isessentially a nonelective surgical condition. Additionally, there isno consensus among surgeons as to which intertrochantericfracture patterns constitute unstable fractures, particularly withinthe AO/OTA 31-A2 subtypes52. Thus, the decision to label afracture as unstable relaxes the clinical criteria necessary to justifythe selection of an intramedullary nail device.

Practice structure and orthopaedic board-certificationstatus were not significantly associated with intramedullary nail

use when other surgeon factors were included in the models.Board-certification status is likely underidentified in the Medi-care provider enrollment file, since it is not required enrollmentinformation for Medicare providers53,54, may reflect initial pro-vider enrollment prior to board examinations, and may only beupdated for practice site changes. Neither variable affectsMedicare payments to surgeons, and the validity of the boardcertification and group practice identifiers in the MedicarePhysician Identification and Eligibility Registry file specificallyamong orthopaedic surgeons have not been examined54.

Finally, although this study used Medicare patient infor-mation derived from claims for case selection, excellent con-cordance has been shown between the Medicare physician andhospital claim files in the identification of hip fracture diagnosesand type of implant used36. However, this device validation wasreported only for femoral neck fractures for which only oneCPT code is used to report two distinct procedures36. In contrastto femoral neck fractures, internal fixation procedures for in-tertrochanteric hip fractures are CPT code-specific to bothfracture site and device (27244 and 27245) in the physicianclaim files and both diagnosis and ICD-9-CM procedure codein the hospital files, which enhances the validity of our casedetermination for this exclusively claims-based study36,55.

In conclusion, the use of intramedullary nail devices wasstrongly associated with early-career surgeons and hospitalswith surgeon training programs. Despite the emphasis onevidence-based medicine, our findings suggest that orthopae-dic faculty at teaching institutions, surgeons at high-volumehospitals, and younger surgeons are tending to select ortho-paedic implants that are more expensive but not more effec-tive. We expect that intramedullary nail use will continue toincrease as long as new surgeons are preferentially trained inintramedullary nailing procedures and surgeon reimburse-ment remains insulated from the treating hospital’s burden oftheir choice of higher cost devices under Medicare’s diagnosis-related group payment system. We recommend further anal-ysis to determine how these device-related provider patternsaffect outcomes for Medicare patients after surgery for inter-trochanteric hip fracture.

AppendixTables showing the influence of patient factors on in-tramedullary nail use, the influence of provider factors

using alternate models, and the relative improvement of modelfit from successive models are available with the electronicversion of this article on our web site at jbjs.org (go to thearticle citation and click on ‘‘Supporting Data’’). n

Mary L. Forte, PhD, DCDepartment of Orthopaedics,University of Maryland School of Medicine,22 South Greene Street, Suite S11B,Baltimore, MD 21201.E-mail address: [email protected]

1112




Beth A. Virnig, PhD, MPHRoger Feldman, PhDRobert L. Kane, MDDivision of Health Policy and Management, School of Public Health,University of Minnesota, 420 Delaware Street S.E.,Minneapolis, MN 55455

Lynn E. Eberly, PhDDivision of Biostatistics, School of Public Health,University of Minnesota, MMC 303,420 Delaware Street S.E., Minneapolis, MN 55455

Marc F. Swiontkowski, MDDepartment of Orthopaedic Surgery,University of Minnesota,2450 Riverside Avenue South,Minneapolis, MN 55454

Mohit Bhandari, MD, MSc, FRCSCDivision of Orthopedic Surgery,Department of Surgery, McMaster University,293 Wellington Street North, Suite 110,Hamilton, ON L8L 2X2, Canada

References

1. Karanicolas PJ, Bhandari M, Walter SD, Heels-Ansdell D, Sanders D, SchemitschE, Guyatt GH. Interobserver reliability of classification systems to rate the quality offemoral neck fracture reduction. J Orthop Trauma. 2009;23:408-12.

2. Robbins JA, Biggs ML, Cauley J. Adjusted mortality after hip fracture:from the cardiovascular health study. J Am Geriatr Soc. 2006;54:1885-91.

3. Roche JJ, Wenn RT, Sahota O, Moran CG. Effect of comorbidities and postop-erative complications on mortality after hip fracture in elderly people: prospectiveobservational cohort study. BMJ. 2005;331:1374.

4. Magaziner J, Simonsick EM, Kashner TM, Hebel JR, Kenzora JE. Survival expe-rience of aged hip fracture patients. Am J Public Health. 1989;79:274-8.

5. Richmond J, Aharonoff GB, Zuckerman JD, Koval KJ. Mortality risk after hipfracture. 2003. J Orthop Trauma. 2003;17(8 Suppl):S2-5.

6. Miller CW. Survival and ambulation following hip fracture. J Bone Joint Surg Am.1978;60:930-4.

7. Holmberg S, Conradi P, Kalen R, Thorngren KG. Mortality after cervical hipfracture. 3002 patients followed for 6 years. Acta Orthop Scand. 1986;57:8-11.

8. Kenzora JE, McCarthy RE, Lowell JD, Sledge CB. Hip fracture mortality. Relationto age, treatment, preoperative illness, time of surgery, and complications. ClinOrthop Relat Res. 1984;186:45-56.

9. Koval KJ, Aharonoff GB, Rokito AS, Lyon T, Zuckerman JD. Patients with femoralneck and intertrochanteric fractures. Are they the same? Clin Orthop Relat Res.1996;330:166-72.

10. Michelson JD, Myers A, Jinnah R, Cox Q, Van Natta M. Epidemiology of hipfractures among the elderly. Risk factors for fracture type. Clin Orthop Relat Res.1995;311:129-35.

11. Endo Y, Aharonoff GB, Zuckerman JD, Egol KA, Koval KJ. Gender differences inpatients with hip fracture: a greater risk of morbidity and mortality in men. J OrthopTrauma. 2005;19:29-35.

12. Zuckerman JD. Hip fracture. N Engl J Med. 1996;334:1519-25.

13. Saudan M, Lubbeke A, Sadowski C, Riand N, Stern R, Hoffmeyer P. Per-trochanteric fractures: is there an advantage to an intramedullary nail?: a random-ized, prospective study of 206 patients comparing the dynamic hip screw andproximal femoral nail. J Orthop Trauma. 2002;16:386-93.

14. Adams CI, Robinson CM, Court-Brown CM, McQueen MM. Prospective ran-domized controlled trial of an intramedullary nail versus dynamic screw and plate forintertrochanteric fractures of the femur. J Orthop Trauma. 2001;15:394-400.

15. Lorich DG, Geller DS, Nielson JH. Osteoporotic pertrochanteric hip fractures.Management and current controversies. J Bone Joint Surg Am. 2004;86:398-410.

16. Parker MJ, Handoll H. Gamma and other cephalocondylic intramedullary nailsversus extramedullary implants for extracapsular hip fractures in adults. CochraneDatabase Syst Rev. 2005;19:CD000093.

17. Parker MJ, Handoll HH. Gamma and other cephalocondylic intramedullary nailsversus extramedullary implants for extracapsular hip fractures in adults. CochraneDatabase Syst Rev. 2008;3:CD000093.

18. Utrilla AL, Reig JS, Munoz FM, Tufanisco CB. Trochanteric gamma nail andcompression hip screw for trochanteric fractures: a randomized, prospective,comparative study in 210 elderly patients with a new design of the gamma nail. JOrthop Trauma. 2005;19:229-33.

19. Hoffman CW, Lynskey TG. Intertrochanteric fractures of the femur: a random-ized prospective comparison of the Gamma nail and the Ambi hip screw. Aust N Z JSurg. 1996;66:151-5.

20. Radford PJ, Needoff M, Webb JK. A prospective randomised comparison ofthe dynamic hip screw and the gamma locking nail. J Bone Joint Surg Br. 1993;75:789-93.

21. Forte ML, Virnig BA, Kane RL, Durham S, Bhandari M, Feldman R, SwiontkowskiMF. Geographic variation in device use for intertrochanteric hip fractures. J BoneJoint Surg Am. 2008;90:691-9.

22. Sadowski C, Lubbeke A, Saudan M, Riand N, Stern R, Hoffmeyer P. Treatmentof reverse oblique and transverse intertrochanteric fractures with use of an intra-medullary nail or a 95� screw-plate: a prospective, randomized study. J Bone JointSurg Am. 2002;84:372-81.

23. Miedel R, Ponzer S, Tornkvist H, Soderqvist A, Tidermark J. The standardGamma nail or the Medoff sliding plate for unstable trochanteric and subtrochan-teric fractures. A randomised, controlled trial. J Bone Joint Surg Br. 2005;87:68-75.

24. Kaplan K, Miyamoto R, Levine BR, Egol KA, Zuckerman JD. Surgical manage-ment of hip fractures: an evidence-based review of the literature. II: intertrochantericfractures. J Am Acad Orthop Surg. 2008;16:665-73.

25. Parker MJ, Handoll HH. Gamma and other cephalocondylic intramedullary nailsversus extramedullary implants for extracapsular hip fractures. Cochrane DatabaseSyst Rev. 2002(4):CD000093.

26. Saarenpaa I, Heikkinen T, Jalovaara P. Treatment of subtrochanteric fractures.A comparison of the Gamma nail and the dynamic hip screw: short-term outcome in58 patients. Int Orthop. 2007;31:65-70.

27. Robinson CM, Houshian S, Khan LA. Trochanteric-entry long cephalomedullarynailing of subtrochanteric fractures caused by low-energy trauma. J Bone Joint SurgAm. 2005;87:2217-26.

28. Anglen JO, Weinstein JN; American Board of Orthopaedic Surgery ResearchCommittee. Nail or plate fixation of intertrochanteric hip fractures: changing patternof practice. A review of the American Board of Orthopaedic Surgery Database. J BoneJoint Surg Am. 2008;90:700-7.

29. Kyle RF. SpotCheck: Intertrochanteric fracture treatment. AAOS Now. 2009Jan. http://www.aaos.org/news/aaosnow/jan09/clinical2.asp. Accessed 2009Aug 25.

30. Butt MS, Krikler SJ, Nafie S, Ali MS. Comparison of dynamic hip screw andgamma nail: a prospective, randomized, controlled trial. Injury. 1995;26:615-18.

31. Aros B, Tosteson AN, Gottlieb DJ, Koval KJ. Is a sliding hip screw or IM nail thepreferred implant for intertrochanteric fracture fixation? Clin Orthop Relat Res.2008;466:2827-32.

32. Centers for Medicare and Medicaid Services. Part B physician/supplier nationaldata - CY 2007: top 200 level 1 Current Procedural Terminology (HCPCS/CPT) codes,ranked by charges. 2008; http://www.cms.hhs.gov/MedicareFeeforSvcPartsAB/Downloads/Level1CHARG07.pdf?agree=yes&next=Accept. Accessed 2010 Jan 28.

33. Centers for Medicare and Medicaid Services. Physician/supplier proceduresummary data file, 2004. Baltimore: Center for Medicare and Medicaid Services;2005.

34. Hart AC, Hopkins CA, editors. 2002 ICD-9-CM expert for hospitals. 6th ed. WestValley City, UT: St. Anthony/Medicode; 2001.

35. American Medical Association. CPT 2000: current procedural terminology.Chicago: American Medical Association Press; 2000.

36. Baron JA, Lu-Yao G, Barrett J, McLerran D, Fisher ES. Internal validation ofMedicare claims data. Epidemiology. 1994;5:541-4.

37. Department of Health and Human Services, Centers for Medicare and MedicaidServices. Medicare carriers manual, part 3-claims process. Transmittal 1723. 2001

1113




Sep 26. http://www.cms.hhs.gov/transmittals/downloads/R1723B3.pdf. Ac-cessed 2008 Aug 4.

38. Office of the Federal Register, National Archives and Records Administration.Code of Federal Regulations, Title 42, Chapter IV: Centers for Medicare and Med-icaid Services. Department of Health and Human Services, Part 483–Requirementsfor States and Long Term Care Facilities: § 483.40(c)(1) Physician services. 2003;http://edocket.access.gpo.gov/cfr_2003/octqtr/pdf/42cfr483.40.pdf. Accessed2010 Jan 27.

39. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifyingprognostic comorbidity in longitudinal studies: development and validation. J ChronicDis. 1987;40:373-83.

40. Romano PS, Roos LL, Jollis JG. Adapting a clinical comorbidity index for usewith ICD-9-CM administrative data: differing perspectives. J Clin Epidemiol. 1993;46:1075-9, 81-90.

41. Molenberghs G, Verbeke G. Models for discrete longitudinal data. New York:Springer; 2006.

42. Littell RC, Milliken GA, Stroup WW, Wolfinger RD, Schabenberger O. SAS� formixed models. 2nd ed. Cary, NC: SAS Institute, Inc; 2006.

43. Kellie SE, Brody JA. Sex-specific and race-specific hip fracture rates. Am J PublicHealth. 1990;80:326-8.

44. Fang J, Freeman R, Jeganathan R, Alderman MH. Variations in hip fracturehospitalization rates among different race/ethnicity groups in New York City. EthnDis. 2004;14:280-4.

45. Freiman MP. The rate of adoption of new procedures among physicians.The impact of specialty and practice characteristics. Med Care. 1985;23:939-45.

46. Escarce JJ. Externalities in hospitals and physician adoption of a new surgicaltechnology: an exploratory analysis. J Health Econ. 1996;15:715-34.

47. Ziran BH, Heckman DS, Olarte CM, Chou K, Baranick J. Intramedullary hipscrew versus standard compression hip screw: early postoperative rehabilitationcomparisons. Orthopedics. 2009;32:83.

48. Sloan FA, Valvona J, Perrin JM, Adamache KW. Diffusion of surgical technology.An exploratory study. J Health Econ. 1986;5:31-61.

49. Department of Health and Human Services. "Examining the relationship betweenthe medical device industry and physicians": Testimony of Gregory E. Demske, As-sistant Inspector General for Legal Affairs, Office of Inspector General, Department ofHealth and Human Services. 2008 Feb 27. Hearing before the Senate Special Com-mittee on Aging, United States Senate. http://www.oig.hhs.gov/testimony/docs/2008/demske_testimony022708.pdf. Accessed 2008 Nov 29.

50. United States Department of Justice. Five companies in hip and knee replace-ment industry avoid prosecution by agreeing to compliance rules and monitoring.2007. http://www.usdoj.gov/usao/nj/press/files/pdffiles/hips0927.rel.pdf. Ac-cessed 2008 Nov 29.

51. Poolman RW, Sierevelt IN, Farrokhyar F, Mazel JA, Blankevoort L, Bhandari M.Perceptions and competence in evidence-based medicine: are surgeons gettingbetter? A questionnaire survey of members of the Dutch Orthopaedic Association.J Bone Joint Surg Am. 2007;89:206-15.

52. Butler M, Forte ML, Kane RL, Joglekar S, Duval SJ, Swiontkowski M, Wilt T.Treatment of common hip fractures. Evidence report/technology assessment no.184 (Prepared by the Minnesota Evidence-based Practice Center under Contract No.HHSA 290 2007 10064 1). AHRQ Publication No. 09-E013. Rockville, MD: Agencyfor Healthcare Research and Quality; 2009 Aug.

53. Center for Medicare and Medicaid Services. Medicare enrollment applicationfor physicians and non-physician practitioners. 2008. http://www.cms.hhs.gov/cmsforms/downloads/cms855i.pdf. Accessed 2008 May 5.

54. Baldwin LM, Adamache W, Klabunde CN, Kenward K, Dahlman C, Warren LJ.Linking physician characteristics and Medicare claims data: issues in dataavailability, quality, and measurement. Med Care. 2002;40(8 Suppl):IV-82-95.

55. Fisher ES, Baron JA, Malenka DJ, Barrett JA, Kniffin WD, Whaley FS, Bubolz TA.Hip fracture incidence and mortality in New England. Epidemiology. 1991;2:116-22.

1114




Provider Factors Associated with Intramedullary Nail Use for Intertrochanteric Hip Fractures

Documents

Transcript of Provider Factors Associated with Intramedullary Nail Use for Intertrochanteric Hip Fractures