Epidemiologic studies have demonstrated that over

Rate of contralateral hip fracture after dynamic hip screw vs intramedullary nail for treatment of pertrochanteric hip fractures Jonathan D. Hughes, MD, Justin H. Bartley, MD, Kindyle L. Brennan, PhD, PT, Yolanda Munoz Maldonado, PhD, Michael L. Brennan, MD, and Christopher D. Chaput, MD A retrospective, comparative study was performed reviewing the electronic medical records and digital radiographs of patients who underwent treatment for intertrochanteric and pertrochanteric hip fractures with either a hip screw and side plate (HSSP) or intramedullary nail. A total of 430 patients were treated with HSSP, and 725 were managed with a cephalomedullary nail (CMN). Of these, 103 sustained a contralateral hip fracture. Fixation technique was not associated with a significant difference in the rate of contralateral fracture. Among the patients with a contralateral fracture, the median time to contralateral fracture was 119.28 months following HSSP and 81.97 months following CMN. Bisphosphonate use was found to be a significant predictor of contralateral fracture for all patients, but when matching using propensity scores, its use was found to be insignificant. In conclusion, there was no difference in the rate of subsequent contralateral hip fracture when comparing HSSP with CMN. Additionally, the time to second surgery between the two treatment modalities was found to be statistically insignificant. It is unclear if bisphosphonate use increased the odds of having a contralateral fracture, regardless of the surgical intervention. The difference in the bisphosphonate effect using propensity score matching suggests that the results may be due to confounding variables and bias. Epidemiologic studies have demonstrated that over 1.6 million US adults sustain a hip fracture yearly, and that number is expected to increase over the next 5 years (1). Nearly 14% of Americans die within 6 months of the initial fracture, while 24% die within 12 months (2 4). Additionally, 2% to 12% of patients with a hip fracture sustain a contralateral fracture (5 12). Souder et al (12) indicated an increased risk of contralateral hip fracture after initial treatment for a femoral neck fracture with closed reduction and percutaneous pinning, as compared with arthroplasty. Zlowodzki et al (13) demonstrated that cannulated hip screw constructs for femoral neck fractures are more prone to shortening, which may decrease tension on abductors and increase the risk for falls and subsequent hip fractures. Because the type of surgical treatment affects the rate of subsequent fractures following femoral neck fracture, it is reasonable to believe that it would also affect the rate of contralateral fractures following intertrochanteric and pertrochanteric (IT/PT) fractures, which are generally treated with either hip screw and side plate constructs (HSSP) or cepha- 268 lomedullary nails (CMN). The advantages, disadvantages, and outcomes following HSSP compared with CMN for IT/PT fractures have been the subject of multiple studies; however, to our knowledge, no study has compared HSSP with CMN in regards to rate of subsequent contralateral hip fracture. This study aimed to determine whether the type of surgical fixation (HSSP vs CMN) for initial proximal femoral fractures (excluding neck) influenced the rate of contralateral hip fracture and then assessed whether bisphosphonate use, diabetes mellitus, and smoking affected the rate of contralateral proximal femur fractures for each treatment modality. METHODS This study was a retrospective, comparative chart review of two cohorts of patients within a single health care system (Level 3 evidence). This study was approved by our hospital s institutional review board prior to initiation. All patients who underwent treatment of an IT/PT proximal femur fracture (AO-Müller/Orthopaedic Trauma Association [AO/OTA] classification of A1, A2, or A3) with an HSSP or CMN between January 1, 2001, and March 23, 2013, within our single level 1 trauma health care system were identified. Patients who had malignant disease, sustained any fracture other than an IT/PT hip fracture of a native hip, were treated with any procedure other than HSSP or CMN, or sustained the fracture as a result of high-energy trauma were excluded from this study. High-energy trauma was defined as an injury resulting from any mechanism other than a fall from standing or seated height. In addition, we excluded patients with sequential bilateral hip fractures whose first fracture was not managed at our facility. We were the treating facility of all hip fractures included in this retrospective review. In total, 1157 medical records met inclusion criteria and were reviewed. All patients were managed by orthopedic senior staff surgeons. The orthopedic implants utilized were the Stryker Omega Dynamic Hip Screw (West Chester, PA) and the Stryker Gamma Intramedullary Locking Nail (Kalamazoo, MI). Fracture pattern From the Department of Orthopedic Surgery, Scott & White Medical Center, Temple, Texas. Corresponding author: Jonathan D. Hughes, MD, Scott & White Medical Center, 2401 S. 31st Street, Temple, TX 76508 (e-mail: Jonathan.Hughes@BSWHealth.org). Proc (Bayl Univ Med Cent) 2017;30(3):268 272

based on Evans classification and the AO/OTA classification, as well as individual surgeon experience, influenced the method of treatment (14, 15). Most AO/OTA 31-A1 and stable AO/ OTA 31-A2 fractures are treated with an HSSP at our facility, while unstable AO/OTA 31-A2 fractures are treated with a CMN. Lastly, AO/OTA 31-A3 fractures are treated with a CMN. Studies have defined unstable fractures as those with extension into the subtrochanteric region, comminution of the lateral wall, comminution of the posterior-medial cortex, and reverse obliquity types (16, 17). Postoperatively, patients were weight-bearing as tolerated and began working with physical therapists immediately. Patients were discharged home with home health or were discharged to a skilled nursing facility, nursing home, or rehabilitation center, depending on physical therapy recommendations and the amount of assistance available at home. International Statistical Classification of Diseases and Related Health Problems (ICD-9) codes were utilized to identify patients with unspecified trochanteric fractures (ICD-9 820.20) and intertrochanteric fractures (ICD-9 820.21). The medical records were reviewed and data collected including medical record number, date of birth, gender, fracture type, surgery type, mechanism of injury, surgery date, surgeon involved, bisphosphonate use, diabetic status, and smoking status. For all those with subsequent proximal femur fractures, we recorded the time that elapsed from initial to subsequent fractures. Digital radiographs were then reviewed for each fracture to confirm fracture type (pertrochanteric, intertrochanteric, subtrochanteric), note surgery performed (HSSP, CMN), and classify the fracture types according to the Evans and AO/OTA classification system for proximal femur fractures. After radiographic review, the subtrochanteric femur fractures were excluded from the data set. Secondary variables were collected from the patient chart: age, sex, cigarette smoking, bisphosphonate use, and history of diabetes. Smoking status was classified based on a scale of 1 to 5: Grade 1 were patients who had never smoked or smoked <10 pack-years, Grade 2 were former smokers of 10 to 50 pack-years, Grade 3 were former smokers of >50 pack-years, Grade 4 were current smokers of <1 pack per day, and Grade 5 were current smokers of >1 pack per day. Significant smoking history was defined as Grade 2 or higher. Descriptive statistics are reported as mean (standard deviation) or median (range) for continuous variables. Categorical variables are described as counts and percentages. Outcome variables were compared by contralateral fracture and fixation methods. Chi-square tests were used to compare two independent nominal variables. A logistic regression model was fit to the data. A P value of 0.25 was used to select variables for the model, and the final model selection used the methods of stepwise, forward, and backward selection. Profile likelihood methods were used to calculate the confidence intervals of the odds ratios. Due to the retrospective nature of the data, a conditional logistic regression model using one-to-one propensity score matching was performed to adjust for differences between the fixation groups. Patients were matched according to a propensity score calculated using a logistic regression model on type of fixation as a response, with gender, diabetes status, fracture type, bisphosphonate use, first fracture classification, and first fracture side used as covariates. The number of digits to match in the probabilities was 0.001. A log-rank test was used for comparison of time to second fracture by fixation technique. A level of 0.05 was considered statistically significant for all tests. The software used was SAS/STAT, Version 9.4 of the SAS System for Windows, and StatXact version 10.1. No external funding sources were used for this retrospective review. RESULTS Following review of all medical records and radiographs, 1157 patients met inclusion criteria. Two of these patients had subtrochanteric extension of their fracture and were excluded from the analysis for a total of 1155 subjects. Of these, 103 (8.9%) patients had contralateral fractures. The study population included 841 (72.8%) females and 314 (27.2%) males with an average age of 82.4 (10.3) years at the time of initial hip fracture (Table 1). Four hundred and thirty (37.2%) were managed with an HSSP and 725 (62.8%) were managed with a CMN. Demographic data for these patients are listed in Table 2. Table 1. Patient characteristics by occurrence of a second fracture Second fracture No Yes P Variable (N = 1052) (N = 103) value Female 760 (72%) 81 (79%) 0.20 0.22 Age at first fracture (years) 83.9 (35.1 109.0) 79.5 (51.6 108.0) Diabetes mellitus 247 (24%) 23 (22%) 0.90 Never smoked/<10 pack-years 716 (68%) 70 (68%) 0.98 Current smoker <1 pack-years 56 (5%) 6 (6%) Current smoker >1 pack-years 29 (3%) 2 (2%) 10 50 pack-years 148 (14%) 16 (16%) >50 pack-years 102 (10%) 9 (9%) Bisphosphonate yes 217 (21%) 32 (31%) 0.02 First fracture characteristics Fracture type Pertrochanteric 880 (84%) 84 (82%) 0.58 Intertrochanteric 172 (16%) 19 (18%) Fixation type 0.11 Cephalomedullary nail 668 (64%) 57 (55%) Hip plate and side screw 384 (36%) 46 (45%) A1 311 (32%) 39 (42%) 0.13 A2 569 (58%) 45 (48%) A3 107 (11%) 9 (10%) July 2017 Treatment of pertrochanteric hip fractures 269

Table 2. Patient characteristics by type of fixation method Variable HSSP (N = 430) CMN (N = 725) P value Female 302 (70%) 539 (74%) 0.13 0.03 Age at first fracture (years) 82.3 (51.6 109.7) 83.9 (35.1 109.1) Diabetes mellitus 101 (24%) 169 (23%) 1.0 Never smoked/<10 pack-years 297 (69%) 489 (68%) 0.96 Current smoker <1 pack-years 23 (5%) 39 (5%) Current smoker >1 pack-years 10 (2%) 21 (3%) 10 50 pack-years 58 (14%) 106 (15%) >50 pack-years 42 (10%) 69 (10%) Bisphosphonate yes 84 (20%) 165 (23%) 0.21 First fracture characteristics Fracture type Pertrochanteric 362 (84%) 602 (83%) 0.62 Intertrochanteric 68 (16%) 123 (17%) 0.62 A1 259 (71%) 91 (13%) <0.001 A2 103 (28%) 511 (72%) A3 5 (1%) 111 (16%) HSSP indicates hip screw and side plate; CMN, cephalomedullary nail. The percentage of contralateral hip fractures in both HSSP and CMN groups was 8.9%, with 46 patients (10.7%) in the HSSP group and 57 patients (7.9%) in the CMN group suffering a contralateral proximal femur fracture (P = 0.11). Among the patients with a contralateral fracture, the median time to contralateral fracture following HSSP was 119.28 (60.98 141.54) months and for CMN, 81.97 (65.57 110.30) months. A logistic regression model was statistically significant (P = 0.014), with only bisphosphonate use as a signifi cant factor for predicting a contralateral fracture (P = 0.01; Tables 3 and 4). Residual diagnostics indicated no violations to the assumptions in the model, and the Hosmer-Lemeshow test pointed to an adequate fit to the data. However, the c statistic (c = 0.58) indicated that the model was just slightly better than chance (0.5) in predicting the outcome. This may suggest that, although the model fits well and bisphosphonate use is a predictor of a contralateral fracture, there are unknown factors that could increase the prediction of a contralateral fracture. A conditional logistic regression model using bisphosphonate and type of fixation using propensity scores for matching the patients was not significant (P = 0.2). Selection of variables was performed using stepwise, forward, and backward elimination. None of the methods chose a variable in the matched analysis. Bisphosphonate and type of fixation were added to the model for comparison with the unmatched logistic regression. The differences in the models suggest that the results may be due to Table 3. Coefficient estimators for the final unmatched logistic regression model DF Parameter estimate SE P value OR 95% CI for OR Intercept 1 2.2 0.11 <0.0001 1.026 Bisphosphonate 1 0.28 0.11 0.01 1.8 1.1 2.7 use yes Fixation method cephalomedullary nail 1 0.18 0.10 0.08 0.7 0.5 1.1 OR indicates odd ratio; CI, confidence interval; SE, standard error. Table 4. Univariate logistic regression analysis with second fracture as an outcome Variable Parameter OR Gender Female (reference) Male OR 95% CI 2.41 0.17 0.7 0.4 1.1 P value 0.17 Age (numerical) 0.01 0.99 0.97 1.0 0.22 Diabetes mellitus No (reference) Yes 0.03 1.1 0.7 1.8 0.79 Never smoked or <10 packs/ year (reference) 10 50 packs/year >50 packs/year Current smoker of 1 pack/year Current smoker of >1 pack/year Bisphosphonates No (reference) Yes Surgery type Hip plate and side screw ( reference) Intramedullary nail Side Right (reference) Left A1 (reference) A2 A3 2.4 0.15 0.05 0.14 0.3 1.1 0.9 1.1 0.7 0.6 1.9 0.4 1.8 0.4 2.4 0.1 2.4 2.2 0.3 1.7 1.1 2.7 0.2 0.7 0.5 1.1 0.8 0.9 0.6 1.3 2.4 0.2 0.1 0.6 0.3 OR indicates odd ratio; CI, confidence interval. 0.4 0.99 0.3 1.4 0.97 0.02 0.10 0.44 0.12 confounding variables and bias. All regression diagnoses were checked and model assumptions satisfied. Comparing characteristics of patients with type of surgery (HSSP vs CMN), age and fracture classification were statistically significant (P = 0.03 and < 0.0001, respectively). 270 Baylor University Medical Center Proceedings Volume 30, Number 3

DISCUSSION To our knowledge, no previous studies have specifi cally compared methods of intertrochanteric and pertrochanteric hip fracture fixation and the associated risk of subsequent contralateral hip fracture. Souder et al (12) recently reported a significantly increased risk of subsequent contralateral hip fracture after treatment of a femoral neck fracture with closed reduction percutaneous pinning in comparison with hip arthroplasty. The purpose of this study was to report the rate of contralateral hip fracture after an initial IT/PT fracture treated with HSSP compared with CMN. One might expect this study to produce outcomes similar to those of Souder et al (12), especially with the large number of patients and a similar database of patients. However, our results indicated no difference in the rate of contralateral hip fractures when treating IT/PT proximal femur fractures with HSSP compared with CMN. Regardless of type of fixation used in these circumstances, the patient had an 8.9% chance (with a 95% confidence interval of 7.3 to 10.7) of suffering a subsequent contralateral hip fracture. We found bisphosphonate use to be associated with having a subsequent contralateral hip fracture (P = 0.02), but this result does not hold in a propensity score match analysis. Eighty-four patients in the HSSP group and 165 patients in the CMN group were taking bisphosphonates for a diagnosis of osteoporosis. The odds for a second fracture were 1.7 times greater for patients taking bisphosphonates compared with those not taking bisphosphonates. Diabetes mellitus and smoking history did not demonstrate a statistically significant association with subsequent hip fracture after the index fracture (P = 0.79 and 0.97, respectively). There were more diabetic patients in the CMN group than in the HSSP group, but this difference was not statistically significant (P = 1.0). This may have elevated the rate of contralateral hip fractures in this group, but our analysis did not validate this assumption. Our data indicate that when utilizing appropriate surgical indications in the setting of AO/OTA 31-A1 and A2 IT/PT fractures, there was no difference in the rate of contralateral hip fractures regardless of method of fixation. A recent literature review conducted by Kaplan et al (18) recommended fixing unstable intertrochanteric hip fractures with a CMN due to evidence that intramedullary devices aid in early mobilization and return of ambulatory function, presumably by maintaining a better reduction and not allowing the shortening and decreased offset seen with HSSP fixation (19, 20). Kaplan does not stand alone, as several other prospective randomized trials have supported a better preservation of reduction when nails are used (21 23). The two surgical groups, however, were not entirely comparable; 31-A1 fractures were treated more often with HSSP, while 31-A2 fractures were treated more often with CMN. Our results do not suggest that highly unstable fractures treated with either HSSP or CMN will have a similar contralateral fracture rate. Even when CMN nails are used to fix more complex and potentially unstable fracture patterns, a similar rate of contralateral fracture is seen compared with HSSP when utilized in less severe fracture patterns. There was no difference in the distribution of age in patients that had a second fracture. Current reports also indicate that the older the patients, the more likely they are to sustain both an index fracture and a subsequent hip fracture (24, 25). Overall, the median age of our cohort at the time of their initial hip fracture was 82.4 years for both the HSSP and CMN group; the median age of the cohort that sustained a contralateral fracture at the time of their initial hip fracture was 81.8 years. The average age of a contralateral fracture in 103 patients was 83.6 years. These data are in line with current literature that reports an average age of 77 ± 6.7 years and 80 ± 6.4 years at the time of initial and second fracture, respectively (11). Both treatment groups experienced a similar median length of time until they suffered a contralateral hip fracture (HSSP was 9.9 years and CMN was 6.8 years; P = 0.62). A literature review found a wide range of data with intervals from initial to subsequent fracture between 2 and 7+ years (11, 26). Souder et al (12) noted a 21.2- to 24.9-month interval between an initial femoral neck fracture and the subsequent hip fracture. More research into defining a tighter interval (and possibly exploring reasons for such varied intervals) may be warranted. The limitations of our study include its retrospective nature and surgeon bias. All retrospective studies are limited by their ability to retrieve data and control variables. Furthermore, the sample size is often diminished due to missing data points. Surgeon bias such as varied experience in interpretation of radiographs and familiarity with surgical techniques also likely affected how the fractures were classified and which implant the patient received. The scope of this study did not include a number of variables, which potentially affected our findings. A prospective study would most likely be required to get accurate data on mental status, ambulatory status, degree of osteoporosis, and current control of diabetes throughout the period of study, all of which would lead to more accurate data and a better understanding of confounding variables. To summarize, our data showed no difference in the rates of contralateral hip fracture when comparing HSSP with CMN. Our results did suggest that patient-related variables such as bisphosphonate use are associated with an increase in the rate of contralateral hip fracture after initial IT/PT fracture; however, these are complex topics and further research would be needed to definitively make a statement on these variables. The factors of implant choice (HSSP vs CMN) and smoking history did not appear to affect the rate of contralateral hip fracture after IT/PT fracture. 1. Hung WW, Morrison RS. Hip fracture: a complex illness among complex patients. Ann Intern Med 2011;155(4):267 268. 2. Jacobs JJ, Andersson GB, Bell JE, Weinstein SL, Dormans JP, Gnatz SM, Lane N, Puzas JE, St. Clair EW, Yelin EH. The Burden of Musculoskeletal Diseases in the United States: Prevalence, Societal and Economic Cost. Rosemont, IL: American Academy of Orthopaedic Surgeons, 2008. 3. Hannan EL, Magaziner J, Wang JJ, Eastwood EA, Silberzweig SB, Gilbert M, Morrison RS, McLaughlin MA, Orosz GM, Siu AL. Mortality and locomotion 6 months after hospitalization for hip fracture: risk factors and risk-adjusted hospital outcomes. JAMA 2001;285(21):2736 2742. July 2017 Treatment of pertrochanteric hip fractures 271

4. Lu-Yao GL, Baron JA, Barrett JA, Fisher ES. Treatment and survival among elderly Americans with hip fractures: a population-based study. Am J Public Health 1994;84(8):1287 1291. 5. Boston DA. Bilateral fractures of the femoral neck. Injury 1982;14(3):207 210. 6. Chiu KY, Pun WK, Luk KD, Chow SP. Sequential fractures of both hips in elderly patients a prospective study. J Trauma 1992;32(5):584 587. 7. Dinah AF. Sequential hip fractures in elderly patients. Injury 2002; 33(5):393 394. 8. Lönnroos E, Kautiainen H, Karppi P, Hartikainen S, Kiviranta I, Sulkava R. Incidence of second hip fractures. A population-based study. Osteoporos Int 2007;18(9):1279 1285. 9. Nymark T, Lauritsen JM, Ovesen O, Röck ND, Jeune B. Short time-frame from first to second hip fracture in the Funen County Hip Fracture Study. Osteoporos Int 2006;17(9):1353 1357. 10. Schrøder HM, Petersen KK, Erlandsen M. Occurrence and incidence of the second hip fracture. Clin Orthop Relat Res 1993;289:166 169. 11. Shabat S, Gepstein R, Mann G, Kish B, Fredman B, Nyska M. The second hip fracture an analysis of 84 elderly patients. J Orthop Trauma 2003;17(9):613 617. 12. Souder CD, Brennan ML, Brennan KL, Song J, Williams J, Chaput C. The rate of contralateral proximal femoral fracture following closed reduction and percutaneous pinning compared with arthroplasty for the treatment of femoral neck fractures. J Bone Joint Surg Am 2012;94(5):418 425. 13. Zlowodzki M, Ayeni O, Petrisor BA, Bhandari M. Femoral neck shortening after fracture fixation with multiple cancellous screws: incidence and effect on function. J Trauma 2008;64(1):163 169. 14. Evans EM. The treatment of trochanteric fractures of the femur. J Bone Joint Surg Br 1949;31B(2):190 203. 15. Marsh JL, Slongo TF, Agel J, Broderick JS, Creevey W, DeCoster TA, Prokuski L, Sirkin MS, Ziran B, Henley B, Audige L. Fracture and dislocation classification compendium 2007: Orthopaedic Trauma Association classification, database and outcomes committee. J Orthop Trauma 2007; 21(10 Suppl):S1 S133. 16. Dhanwal DK, Dennison EM, Harvey NC, Cooper C. Epidemiology of hip fracture: worldwide geographic variation. Indian J Orthop 2011;45(1):15 22. 17. Nikitovic M, Wodchis WP, Krahn MD, Cadarette SM. Direct health-care costs attributed to hip fractures among seniors: a matched cohort study. Osteoporos Int 2013;24(2):659 669. 18. Kaplan K, Miyamoto R, Levine BR, Egol KA, Zuckerman JD. Surgical management of hip fractures: an evidence-based review of the literature. II: intertrochanteric fractures. J Am Acad Orthop Surg 2008;16(11):665 673. 19. Nuber S, Schönweiss T, Rüter A. [Stabilisation of unstable trochanteric femoral fractures. Dynamic hip screw (DHS) with trochanteric stabilisation plate vs. proximal femur nail (PFN)]. Unfallchirurg 2003;106(1):39 47. 20. Pajarinen J, Lindahl J, Michelsson O, Savolainen V, Hirvensalo E. Pertrochanteric femoral fractures treated with a dynamic hip screw or a proximal femoral nail. A randomised study comparing post-operative rehabilitation. J Bone Joint Surg Br 2005;87(1):76 81. 21. Adams CI, Robinson CM, Court-Brown CM, McQueen MM. Prospective randomized controlled trial of an intramedullary nail versus dynamic screw and plate for intertrochanteric fractures of the femur. J Orthop Trauma 2001;15(6):394 400. 22. Ahrengart L, Törnkvist H, Fornander P, Thomgren KG, Pasanen L, Wahlstrom P, Honkonin S, Lindgren U. A randomized study of the compression hip screw and Gamma nail in 426 fractures. Clin Orthop Relat Res 2002;401:209 222. 23. Miedel R, Ponzer S, Törnkvist H, Söderqvist A, Tidermark J. The standard Gamma nail or the Medoff sliding plate for unstable trochanteric and subtrochanteric fractures. A randomised, controlled trial. J Bone Joint Surg Br 2005;87(1):68 75. 24. Hardy DC, Descamps PY, Krallis P, Fabeck L, Smets P, Bertens CL, Delince PE. Use of an intramedullary hip-screw compared with a compression hip-screw with a plate for intertrochanteric femoral fractures. A prospective, randomized study of one hundred patients. J Bone Joint Surg Am 1998;80(5):618 630. 25. Hagino H, Sawaguchi T, Endo N, Ito Y, Nakano T, Watanabe Y. The risk of a second hip fracture in patients after their first hip fracture. Calcif Tissue Int 2012;90(1):14 21. 26. Fukushima T, Sudo A, Uchida A. Bilateral hip fractures. J Orthop Sci 2006;11(5):435 438. 272 Baylor University Medical Center Proceedings Volume 30, Number 3