Statistical Assumptions in Orthopaedic Literature: Are Study Findings at Risk?

Background As orthopaedic surgery becomes more evidence-based, the need for rigorous research has increased. This results in more complex studies that employ more sophisticated statistical analysis, often some form of regression. These statistical techniques require the data to meet certain assumptions for the findings to be considered valid. The purpose of this study is to determine the common regression techniques employed in the orthopaedic surgery literature, and demonstrate how often the assumptions of regression analyses are met and reported. Methods Studies published in the Journal of Bone & Joint Surgery (JBJS) in 2017 and 2018 were reviewed. Commentaries, editorials, and systematic reviews were excluded. The statistical analyses performed in each study were documented. When regression analyses were utilized, the article was reviewed for evidence that the necessary assumptions underlying the statistical methodology were assessed and met. Results From the 470 studies that were reviewed, the most common statistical test reported was the independent-samples t-test (n=215, 45.7%). Also, 201 studies (42.8%) implemented some form of regression analysis. The most common regression was a logistic regression (n= 106). None of the 201 studies using regression analysis reported meeting all of the necessary assumptions to appropriately use a regression test. Conclusion Many recent studies published in JBJS depended on regression analyses to reach their conclusions, but none fully reported the necessary assumptions of these tests. Orthopaedic surgery journals should be more transparent in reporting the methodology of statistical tests, and readers must beware of possible gaps in statistical methodology and critically evaluate the studies' findings.

Improving outcomes after hip fracture at a safety net hospital with a standardised hip fracture protocol.

INTRODUCTION: standardised protocols for the care of geriatric hip fractures demonstrate improved patient outcomes with decreased cost. The purpose of this study is to evaluate outcomes of a standardised hip fracture protocol at an urban safety-net hospital. METHODS: All trauma patients presenting to our urban safety-net hospital are included in a trauma database and inpatient outcomes recorded. A hip fracture protocol was introduced at our institution in 2015, which depended on admission to a monitored setting due to the absence of a geriatric co-management service. The database was queried to identify patients surgically treated for a geriatric hip fracture in the 3 years prior to protocol implementation (2012-2014) and patients treated in the 3 years following protocol implementation (2016-2018). Demographics, time to surgery, inpatient complications, and length of stay were compared between groups. RESULTS: A total of 633 patients treated operatively for isolated hip fractures were identified, 262 patients in the 2012-2014 pre-protocol cohort, and 371 patients in the 2016-2018 protocol cohort. Following implementation of a hip fracture protocol the number of patients admitted to a surgical service increased from 198 (76%) to 348 (94%, p < 0.005) with the number of patients being admitted to a monitored setting increasing from 40 (15%) to 83 (22%, p = 0.026). The time to surgery was reduced to 2.75 days (p = 0.054). The complication rate fell from 23% to 4% (p < 0.0005). Length of stay was significantly reduced from 13.2 days to 12 days (p = 0.045). CONCLUSIONS: A hip fracture protocol including admission to a monitored setting can be effectively implemented at an urban safety-net hospital where geriatric co-management is not available. This resulted in a decrease in complications and length of stay. Additional interventions are required to decrease average time to surgery below 36 hours.

Radiographic Humerus Union Measurement (RHUM) Demonstrates High Inter- and Intraobserver Reliability in Assessing Humeral Shaft Fracture Healing.

BACKGROUND: Orthopedic surgeons use radiographs to determine degrees of fracture healing, guide progression of clinical care, and assist in determining weight bearing and removal of immobilization. However, no gold standard exists to determine the progression of healing of humeral shaft fractures treated non-operatively. PURPOSE: The purpose of this study was to determine whether a scale comparable to the modified Radiographic Union Score for Tibial (RUST) fractures applied to non-operatively treated humeral shaft fractures can increase interobserver reliability in determining fracture healing. METHODS: A retrospective review was undertaken by three orthopedic traumatologists and one musculoskeletal radiologist, who evaluated 50 sets of anteroposterior and lateral radiographs, presented at random, of non-operatively treated humeral shaft fractures at various stages of healing from 17 patients. The radiographs were scored using a modified RUST scale called the Radiographic Humerus Union Measurement (RHUM). Observers were blinded to the time from injury. After a 4-week washout period, observers again scored the same radiographs. Observers classified each fracture as either healed or not healed based on the combination of radiographs. Inter- and intraobserver reliability of the RHUM were determined using an intraclass correlation coefficient (ICC). Interobserver reliability of determining a healed fracture was calculated using Cohen's kappa (κ) statistics. A receiver operator characteristic curve was conducted to determine the RHUM score predictive of a fracture being considered healed. RESULTS: ICC demonstrated almost perfect interobserver reliability (ICC, 0.838; ICC 95% CI, 0.765 to 0.896) and intraobserver reliability (ICC range, 0.822 to 0.948) of the RHUM. κ demonstrated substantial agreement between observers in considering a fracture healed (κ = 0.647). Receiver operating characteristic (ROC) curve demonstrated that a RHUM of 10 or higher is an excellent predictor of the observer considering the fracture healed (area under the ROC curve = 0.946, specificity = 0.957, 95% CI specificity, 0.916 to 0.979). CONCLUSIONS: This cortical scoring system has excellent interobserver reliability in humeral shaft fractures treated non-operatively. Consistent with previous cortical scoring systems, a RHUM score of 10 or above can be considered radiographically healed.

Scoring of radiographic cortical healing with the radiographic humerus union measurement predicts union in humeral shaft fractures.

PURPOSE: The purpose of this study is to determine if the radiographic humerus union measurement (RHUM) is predictive of union in humeral shaft fractures treated nonoperatively. METHODS: All patients with long bone fracture nonunion presenting to a single surgeon were enrolled in a prospective registry. This registry was queried to identify patients with humeral shaft fractures treated nonoperatively and developed nonunion. The nonunion cohort was matched to a three to one gender- and age-matched control group that were treated nonoperatively for a humeral shaft fracture and achieved union. Two fellowship-trained orthopedic traumatologists blinded to eventual union scored radiographs obtained 12 weeks after injury using the RHUM. A binomial logistic regression determined the effect of the RHUM on the likelihood of developing union. RESULTS: Nine patients with humeral shaft fractures treated nonoperatively with radiographs 12 weeks after injury that developed nonunion were identified. These patients were matched to 27 controls. Logistic regression demonstrated the RHUM was a significant predictor of healing 12 weeks after humeral shaft fracture treated nonoperatively (p = 0.014, odds ratio 9.434, 95% CI for OR 1.586-56.098). All patients with RHUM below 7 went on to nonunion. All patients with RHUM above 8 healed. Three of seven patients (43%) with RHUM of 7 or 8 healed. CONCLUSION: The RHUM demonstrated an increased likelihood of achieving union 12 weeks after injury. Orthopedic surgeons can counsel patients that fractures with RHUM scores of 6 or below are in danger of developing nonunion and can target interventions appropriately.

Second physician review of radiographs after wrist and ankle reductions offers limited utility to clinical management.

PURPOSE: The purpose of this study is to determine the clinical utility of second-physician review of radiographs obtained after reduction of distal radius and ankle fractures. METHODS: Fifty consecutive ankle and distal radius fractures requiring reduction were reviewed. The time from post-reduction radiograph to second-physician interpretation was obtained. The second-physicians' interpretation was evaluated for clinically influential information. Patients requiring a repeat reduction were identified, and the timing of the repeat reduction radiograph was compared with the timing of the second-physician interpretation of the initial post-reduction radiograph. RESULTS: The mean time of second-physician interpretation for post reduction ankle radiographs was 6 h and 47 min (range 4 min to 43 h and 3 min). Eleven of 50 (22%) interpretations of post reduction ankle radiographs commented on acceptability of reduction. The mean time of second-physician interpretation for post reduction distal radius radiographs was 5 h and 34 min (range 8 min to 22 h and 59 min). Seven of 50 (14%) interpretations of post reduction distal radius radiographs commented on acceptability of reduction. Three distal radius (6%) and 8 ankle fractures (16%) required repeat reduction. Repeat reductions were completed in 10/11 cases (91%) before the second-physician review of the initial post reduction radiograph was obtained. In only 1 case of repeat reduction was the second-physician review of the post reduction radiograph available before repeat reduction was attempted. CONCLUSION: The timing and quality of second-physician review of post-reduction radiographs offers little utility to the clinical management of ankle and distal radius fractures.

Function and Knee Range of Motion Plateau Six Months following Lateral Tibial Plateau Fractures.

The purpose of this study is to determine when functional outcome no longer improves following tibial plateau fracture. A patient series of operatively treated tibial plateau fractures was reviewed. Patients were evaluated using the short musculoskeletal function assessment (SMFA), range of motion (ROM) assessment, and pain levels at visual analog scale (VAS) at 3, 6, and 12 months postoperatively. Fractures were classified by the Schatzker's classification using preoperative imaging. The case series was divided into two groups based on fracture patterns. Friedman's tests were conducted to determine if there were differences in SMFA, ROM, or VAS throughout the postoperative course. A total of 117 patients with tibial plateau fractures treated operatively, with complete follow-up and without complication, were identified. Seventy-seven patients (65.8%) sustained lateral tibial plateau fractures (Schatzker's I-III). Friedman's test demonstrated significant differences in SMFA (p < 0.0005) and ROM (p < 0.0005) at the three time points. Post hoc analysis demonstrated a significant difference in SMFA (p < 0.0005) and ROM (p = 0.003) between 3 and 6 months postoperatively but no significant difference in either metric between 6 and 12 months postoperatively. Friedman's test demonstrated no significant difference in VAS postoperatively (p = 0.210). Forty patients (34.2%) sustained medial or bicondylar tibial plateau fractures (Schatzker's IV-VI). Friedman's test demonstrated significant differences in SMFA (p < 0.0005) and ROM (p < 0.0005) at the three time points. Post hoc analysis demonstrated a strong trend toward significance in SMFA between 3 and 6 months postoperatively (p = 0.088), and demonstrated a significant difference between 6 and 12 months postoperatively (p = 0.013). ROM was found to be significantly different between 3 and 6 months postoperatively (p = 0.010), but no difference was found between 6 and 12 months postoperatively (p = 0.929). Friedman's test demonstrated no significant difference in VAS postoperatively (p = 0.941). In this cohort, no significant difference in function, ROM, or pain level exists between 6 and 12 months after treatment of lateral tibial plateau fractures. However, there are significant improvements in function for at least 1 year following medial or bicondylar tibial plateau fractures.

Radiographic union score for tibia fractures predicts success with operative treatment of tibial nonunion.

BACKGROUND: The purpose of this study is to evaluate the ability of preoperative and postoperative radiographic union scores for tibia fractures (RUST) to predict treatment success of tibia fracture nonunion. MATERIALS AND METHODS: Patients presenting for operative treatment of tibia fracture nonunion were enrolled in a prospective data registry. Enrolled patients were followed at regular intervals for 12 months. Preoperative and 12 week postoperative radiographs were reviewed and scored using the RUST criteria. Postoperative time to union was determined by clinical and radiographic measures. Multivariate regressions were conducted to predict time to union using preoperative and postoperative RUST while controlling for treatment method. Receiver operating characteristic (ROC) curve was conducted to determine the accuracy of preoperative RUST in predicting failure of treatment. RESULTS: Sixty-eight patients with aseptic tibia fracture nonunion treated operatively were identified. Sixty-one patients achieved union. Mean preoperative RUST was 7.5 (SD 1.4). Mean postoperative RUST was 9.2 (SD 1.4). Multivariate linear regressions demonstrated that preoperative (p = 0.043) and postoperative (p = 0.007) RUST are significant predictors of time to union after tibia fracture nonunion surgery. ROC curve demonstrated preoperative RUST below 7 was a good predictor of developing persistent tibia fracture nonunion (AUC = 0.83, Sensitivity = 1.000, Specificity = 0.745). CONCLUSIONS: RUST preoperatively and postoperatively predicts outcome after nonunion surgery. RUST can be used as part of the complete clinical picture to shape patient expectations and guide treatment.

Is Wikipedia a complete and accurate source for musculoskeletal anatomy?

PURPOSE: Wikipedia is a popular online encyclopedia generating over 5.4 billion visits per month, and it is also a common resource for the general public and professionals for medical information. The goal of this study is to determine the accuracy and completeness of Wikipedia as a resource for musculoskeletal anatomy. METHODS: The origin, insertion, innervation, and function of all muscles of the upper and lower extremities as detailed on Wikipedia was compared to the available corresponding information in Grant's Atlas of Anatomy (14th edition). Entries were scored for both accuracy and completeness. Descriptive statistics were calculated and associations between and within entries for accuracy and completeness were assessed by McNemar's tests. Information on Wikipedia's references was also collected. RESULTS: Overall, data on Wikipedia was 97.6% complete and 98.8% accurate when compared to Grant's Atlas of Anatomy. 78.6% of all entries were fully complete and accurate, with 15.3% of entries containing one error and 6.1% containing two errors. There were no associations between or within entries' accuracy and completeness. Only 62% of references from Wikipedia included were from academic sources. CONCLUSIONS: Musculoskeletal anatomy entries on Wikipedia are imperfect; they have inaccurate and missing information. Furthermore, a considerable proportion of references cited in entries are from poorly identified sources. While Wikipedia is an easily accessible resource for a large number of people and much of the anatomic information is appropriate, it cannot be considered to be an equivalent resource when compared to anatomic texts.

Predictors of Patient Reported Pain After Lower Extremity Nonunion Surgery: The Nicotine Effect.

BACKGROUND: Nonunion of long bone fractures is a serious complication for many patients leading to considerable morbidity. The purpose of this study is to elucidate factors affecting continued pain following long bone nonunion surgery and offer better pain control advice to patients. METHODS: Patients presenting to our institutions for operative treatment of long bone fracture nonunion were enrolled in a prospective data registry. Enrolled patients were followed at regular intervals for 12 months using the Short Musculoskeletal Function Assessment (SMFA), visual analog scale (VAS), physical examination, and radiographic examination. The registry was reviewed to identify patients with a tibial or femoral nonunion that went on to union with complete follow up. Univariate analyses were conducted to identify patient characteristics associated with postoperative pain. Identified patient factors with univariate p-values <0.1 were included in multivariate linear regression models in order to identify risk factors for pain 3 months, 6 months, and 12 months after nonunion surgery. RESULTS: Ninety-one patients with tibial or femoral nonunion who went on to union and had complete follow-up were identified. A Friedman test revealed mean pain score decreased significantly by 3 months postoperatively (p<0.0005). Univariate analyses demonstrated age (p=0.016), days from injury to nonunion surgery at our institution (p=0.067), smoking status (p<0.0005), wound status at time of injury (p=0.085), anesthesia (p=0.045), and nonunion location in the bone (p=0.047) were associated with postoperative pain in at least one time point postoperatively. These were included in multivariate models that revealed nonunion location (p=0.035) was predictive of pain 3 months postoperatively, smoking status was predictive of pain 3 months (p=0.012) and 6 months (p<0.0005) postoperatively, and days from injury to nonunion surgery at our institution was predictive of pain 6 months (p=0.024) and 12 months (p=0.004) postoperatively. CONCLUSION: Healed patients have improved pain levels after lower extremity nonunion surgery. Orthopedic surgeons should stress smoking cessation programs and minimize delay to nonunion surgery, in order to maximize pain relief in this patient cohort.