Abstract
Objective:
The purpose of this study is to develop a scoring system that stratifies burn patients at the time of hospital admission according to risk of developing heterotopic ossification (HO).
Summary of Background Data:
HO in burns is an uncommon but severely debilitating problem with a poorly understood mechanism and no fully effective prophylactic measures.
Methods:
Data were obtained from the Burn Model System National Database from 1994 to 2010 (n = 3693). The primary outcome is diagnosis of HO at hospital discharge. Logistic regression analysis was used to determine significant demographic and medical predictors of HO. A risk scoring system was created in which point values were assigned to predictive factors and final risk score is correlated with the percent risk of developing HO. The model was internally and externally validated.
Results:
The mean age of the subjects is 42.5 ± 16.0 years, the mean total body surface area (TBSA) burned is 18.5 ± 16.4%, and the population is 74.9% male. TBSA and the need for grafting of the arm, head/neck, and trunk were significant predictors of HO development (P < 0.01). A 13-point risk scoring system was developed using these significant predictors. The model c-statistic is 0.92. The risk scoring system demonstrated evidence of internal and external validity. An online calculator was developed to facilitate translation of knowledge to practice and research.
Conclusions:
This HO risk scoring system identifies high-risk burn patients suitable for diagnostic testing and interventional HO prophylaxis trials.
Keywords: burn, complication, heterotopic ossification, outcome, range of motion, risk calculator, risk score, risk scoring system
Heterotopic ossification (HO), the development of abnormal bone in the soft tissue, is an uncommon but severely debilitating complication following burn injury. It is most often associated with major trauma (eg, traumatic brain, spinal cord, blast, or burn injuries) or orthopedic surgery (eg, hip arthroplasty, elbow surgery, and amputation). The symptoms of HO include pain, decreased range of motion, and joint contracture. These complications have negative effects on mobility and quality of life.1 In burn patients, the incidence of HO varies between 0.2% and 5.6%,2–4 and is most frequent in patients with larger burns.3 Treatment for HO in the burn population is currently inadequate, as there are no established early diagnostic or prophylactic interventions. Surgical excision is the only evidence-based intervention and is implemented after HO diagnosis. Although surgery removes the heterotopic bone, the joint contracture often persists and range of motion is rarely completely restored.5,6 Persistent range of motion deficits are likely due to a combination of issues, including soft tissue scarring and residual heterotopic bone. As currently available treatments for HO are of unknown or partial efficacy, it is imperative to understand the mechanism of HO development to design prophylactic interventions. Current prophylaxis options, including nonsteroidal anti-inflammatory drugs (NSAIDS), bisphosphonates, and radiation therapy, involve risks, particularly in critically ill burn patients. Nonetheless, few studies have evaluated prophylaxis regimens in the burn population and those studies are limited by sample size, retrospective design, and negative results.7,8 Given the relative infrequency of HO in burns and the paucity of studies examining prophylactic interventions, there is a need to identify those at a high risk for HO development to foster investigation of diagnostic and preventative measures.
Recent developments in the pathophysiology of HO are driving the need for improved ability to identify those at highest risk of HO development for design of clinical trials. A burn injury causes inflammation in the affected tissue thereby attracting myeloid cells and lymphocytes that catalyze the release of cytokines. Bone morphogenetic proteins, tumor necrosis factors, and interleukin-6 cause mesenchymal stem cell migration and differentiation.9,10 Inflammatory signals increase osteogenic differentiation of mesenchymal stem cells potentially resulting in heterotopic bone formation.10 Although the mechanism of HO development in thermal burns in not fully understood, recent animal model studies have identified mechanistic pathways of the HO cascade.9 New work has demonstrated the role of early hypoxic signaling and hypoxia inducible factor 1-alpha in HO development as well as the potential use of inhibitors of this pathway in its prevention.11–13 Regardless of the targeted mechanism, prophylactic efforts focused on those at highest risk mitigate the deleterious effects of developing treatments such as impaired bone growth, wound healing, and bleeding risk.14
Although previous studies have identified predictors for HO,1,3,4 a formalized risk scoring system has not been developed. Furthermore, prior investigation of risk factors has not focused on indicators known at admission, which is most relevant for identifying a population suitable for implementation of an early intervention. Thus, the aim of this study is to develop an admission HO risk scoring system for the burn population. A risk stratification system will foster the development of novel diagnostic technologies and prophylactic interventions as well as facilitate mechanistic research.
PATIENTS AND METHODS
Data Source and Study Design
This study uses a retrospective review of prospectively collected data from the Burn Injury Model System National Database (BMS Database). The BMS Database is a multicenter, longitudinal database funded by the National Institute on Disability, Independent Living, and Rehabilitation Research since 1993. It includes demographic, injury, and outcome data and currently includes 4 centers in the United States.15
The current BMS Database enrollment criteria for patients 18 years or older is as follows:
Autografting surgery for wound closure AND
18 to 64 years of age with a burn injury ≥20% total body surface area OR
≥65 years of age with a burn injury ≥10% total body surface area OR
≥18 years of age with a burn injury to their face/neck, hands, or feet OR
≥18 years of age with a high-voltage electrical burn injury
Modifications were made to the inclusion criteria over time. Details of the inclusion criteria, data collection process, and data collection sites can be found at http://burndata.washington.edu/.
Subjects with burn dates between 1994 and 2010 were included. Patients less than 18 years in age, deceased at discharge, or with primary diagnoses other than thermal burn (eg, frostbite, radiation, hydrofluoric acid, abrasions, and skin diseases) were excluded from analysis.
Study Variables
Both demographic and medical data of the study population available at time of admission to acute care were examined. Demographic data include age, sex, and race. Medical data include total body surface area (TBSA) burned, location of graft, inhalation injury, burn etiology, history of other medical problems, and location of burn. Although grafting is not necessarily performed on admission, eventual autografting was used as a surrogate for the presence of full or deep-partial thickness injury. The authors determined that a trained burn surgeon would be able to estimate the need for autografting at admission. The outcome is the presence of HO at any anatomical location confirmed radiographically. The HO diagnosis variable is abstracted from the medical record at time of discharge as part of the BMS Database acquisition process.
Data Analysis
Descriptive statistics were calculated for demographic and medical data. Logistic regression analysis was used to determine significant demographic and medical predictors of HO. A P value <0.05 was considered statistically significant. Multiple imputation was used for missing data according to established methodology.16,17 Estimates from 100 imputed datasets were combined using Rubin rules. This method is preferable to excluding missing data in instances when the missing data are not missing completely at random. Statistical analyses were performed using StataCorp. 2015 (StataCorp LP, Stata Statistical Software: Release 14; College Station, TX).
Risk Scoring System Development
A point system for determining risk of HO development was calculated using predictor variables available at admission (described above). This system was developed using the Framingham Heart Study methodology.18 The following steps were used to build the risk scoring system: (1) the model was estimated; (2) the risk factors were organized into categories and reference values were determined for each category; (3) a base category was chosen for each risk factor; (4) the distance from base category was determined for each other category based on the regression coefficient; (5) the constant was set; (6) points associated with each category were determined using the formula provided by the Framingham Heart Study methodology18; and (7) the risks associated with point totals based on the regression were determined. The final result is an admission risk scoring system for HO development by hospital discharge in the burn population.
Model Validation
The HO risk scoring system model was externally validated. Of note, the BMS Database has been externally validated,19 demonstrating its representativeness of the national burn population. Applying the risk scoring system model to another set of burn patients confirms the model’s generalizability to the larger burn population. An external database that included consecutive adult (≥18 years of age) patients with a burn diagnosis admitted to Massachusetts General Hospital (MGH) between January 2000 and January 2010 (MGH External Database) was used. Patients who died, had a nonburn wound, or received acute care elsewhere were excluded from the MGH External Database. Of note, MGH patients began enrollment in the BMS Database in October 2012; therefore, the MGH External Database and BMS Database do not exhibit overlapping data. The 2 databases exhibit some differences. The BMS Database excludes patients with small burns to noncritical areas, whereas the MGH External Database includes all burns that required hospital admission, regardless of burn size and critical area involvement. As a result, the BMS Database is expected to demonstrate a higher incidence of HO than the MGH External Database because many lower risk HO patients (those with small burns to noncritical areas) were excluded from the BMS Database.
The following steps were used to examine the model’s external validity: (1) fit the model developed using the BMS Database to the MGH External Database; (2) compare the discrimination, the area under the receiver operating characteristic curve (c-statistic), to that of the BMS Database model; (3) compare the calibration; (4) because an ideally calibrated model contains a constant of 0 and a slope of 1 as per prior methodologies,20–22 if either the constant is not 0 or the slope is not 1, use the slope and the difference in prevalence to update the coefficients to generate a more valid (ie, generalizable) model; (5) ensure updating of the coefficients resets the constant and slope to 0 and 1, respectively; and (6) compare the c-statistic of the revised model to the original model.
RESULTS
Patient Characteristics
Of the 6071 burn patients who met the eligibility requirements between 1994 and 2010, 1962 were under age 18, 367 were deceased at discharge, and 49 had a primary diagnoses other than thermal burn. A total of 3693 subjects were included for analysis (Fig. 1).
FIGURE 1.
Flow chart of study population based on the inclusion and exclusion criteria.
Of the 3693 eligible subjects, 2758 contained data on presence of HO at discharge. Of these 2758 patients, 98 (3.6%) had developed HO at time of discharge. The mean age of the population was 42.5 ± 16.0 years, the mean TBSA was 18.5 ± 16.4%, and the population was 74.9% male. Details of the demographic and medical characteristics of the BMS Database study population are reported in Table 1. The following variables had more than 5% missing data: burn etiology (12.5%), TBSA % (11.8%), arm burn (12.05%), head/neck burn (12.3%), trunk burn (12.3%), leg burn (12.2%), hand burn (12.3%), perineum burn (22.3%), foot burn (12.2%), arm graft (12.8%), head/neck graft (13.0%), trunk graft (12.8%), leg graft (12.8%), hand graft (12.9%), perineum graft (26.1%), foot graft (13.2%), inhalation injury (12.8%), other medical problems (14.57%), and HO (25.3%).
TABLE 1.
Demographic and Medical Characteristics of the Study Population
| Variable | Total | HO | No HO | P |
|---|---|---|---|---|
| No. of subjects | 3693 | 98 | 2660 | — |
| Age, mean (SD) | 42.5 (16.0) | 41.5 (13.3) | 42.5 (15.9) | 0.572 |
| Male, n (%) | 2766 (74.9) | 81 (82.6) | 2017 (75.8) | 0.119 |
| Race/Ethnicity, n (%) | 0.491 | |||
| Caucasian | 2229 (68.7) | 72 (73.5) | 1859 (70.2) | — |
| Black | 484 (14.9) | 7 (7.1) | 340 (12.8) | — |
| Hispanic | 369 (11.4) | 11 (11.2) | 303 (11.4) | — |
| Asian | 63 (1.9) | 3 (3.1) | 56 (2.1) | — |
| Other | 100 (3.1) | 5 (5.1) | 89 (3.3) | — |
| Primary burn etiology, n (%) | <0.001 | |||
| Fire | 2123 (65.7) | 88 (89.8) | 1738 (65.9) | — |
| Contact | 808 (25.0) | 4 (4.1) | 654 (24.8) | — |
| Electric | 207 (6.4) | 3 (3.1) | 178 (6.7) | — |
| Other | 94 (2.9) | 3 (3.1) | 69 (2.6) | — |
| TBSA %, mean (SD) | 18.5 (16.4) | 48.5 (18.1) | 17.6 (14.9) | <0.001 |
| Burn location, n (%) | — | |||
| Arm | 2196 (67.6) | 97 (99.0) | 1777 (66.9) | <0.001 |
| Head/Neck | 1564 (48.3) | 87 (88.8) | 1348 (50.9) | <0.001 |
| Trunk | 1806 (55.7) | 95 (96.9) | 1455 (54.8) | <0.001 |
| Leg | 1665 (45.1) | 66 (67.3) | 1326 (49.9) | <0.001 |
| Hand | 2213 (68.3) | 92 (93.9) | 1817 (68.5) | <0.001 |
| Perineum | 279 (7.6) | 24 (26.4) | 201 (8.6) | <0.001 |
| Foot | 878 (27.4) | 29 (29.9) | 700 (26.4) | 0.133 |
| Graft location, n (%) | — | |||
| Arm | 1551 (48.2) | 96 (98.0) | 1244 (47.0) | <0.001 |
| Head/Neck | 363 (11.3) | 46 (47.4) | 274 (10.4) | <0.001 |
| Trunk | 921 (28.6) | 81 (82.7) | 722 (27.3) | <0.001 |
| Leg | 1193 (32.3) | 56 (57.7) | 955 (36.0) | <0.001 |
| Hand | 1394 (37.7) | 74 (77.1) | 1156 (43.7) | <0.001 |
| Perineum | 100 (3.7) | 12 (15.2) | 71 (3.2) | <0.001 |
| Foot | 620 (19.3) | 10 (10.4) | 513 (19.4) | 0.046 |
| Inhalation injury, n (%) | 356 (11.1) | 33 (33.7) | 266 (10.0) | <0.001 |
| Other medical problems, n (%) | 1103 (35.0) | 30 (30.9) | 899 (35.0) | 0.540 |
The Total column contains all the data available for the total analysis group. The HO and No HO columns stratify the group of subjects that had data on the presence of HO.
Comparison of mean and median and examination of skewness and kurtosis indicated a normal distribution for continuous variables.
SD indicates standard deviation; TBSA, total body surface area burned.
Data Analysis
HO data were not recorded in 935 BMS Database subjects (25.32%). Using multiple imputation, estimates from 100 datasets were combined using Rubin’s rules. The results of the regression analysis show that an increase in risk of HO is strongly associated with an increase in TBSA burned, as well as arm, head/neck, and trunk burns that required grafting. For every 1% increase in TBSA up to a TBSA of 30%, the risk of HO increases by 13% (OR, 1.135; P < 0.005). For every 1% increase in TBSA above 30%, the risk of HO increases by 4.3% (OR, 1.043; P < 0.001). Patients who required an arm graft (P < 0.006), head/neck graft (P < 0.001), and trunk graft (P < 0.006) exhibited a 5.064, 2.716, and 2.396 times the risk of developing HO, respectively (Table 2).
TABLE 2.
Logistic Regression Analysis Examining Predictors of Development of Heterotopic Ossification
| Indicator | Odds Ratio | P | 95% Confidence Interval |
|---|---|---|---|
| TBSA % | |||
| 1% increases in TBSA up to 30% | 1.14 | 0.004 | 1.04–1.24 |
| 1% increases in TBSA above 30% | 1.04 | 0.000 | 1.02–1.06 |
| Arm graft | 5.06 | 0.005 | 1.63–15.77 |
| Head/Neck graft | 2.72 | 0.000 | 1.71–4.33 |
| Trunk graft | 2.40 | 0.005 | 1.30–4.42 |
TBSA indicates total body surface area burned.
Risk Score System Development
From these predictors, the HO risk scoring system was developed (Table 3). The HO risk scoring system includes 4 variables: TBSA categories of 10% increments, and 3 different graft locations: arm, head/neck, and trunk. The predictive power of this model is strong, as indicated by a model c-statistic of 0.92. The risk scoring system also demonstrates graphical evidence of validity in predicting HO formation (Fig. 2).
TABLE 3.
Admission Scoring System for Assessing HO Risk
| Indicator | Points |
|---|---|
| TBSA (%) | |
| 0–10 | 0 |
| 11–20 | 2 |
| 21–30 | 4 |
| 31–40 | 6 |
| 41–60 | 7 |
| 61–80 | 8 |
| 81–90 | 9 |
| >90 | 10 |
| Arm graft | |
| No | 0 |
| Yes | 1 |
| Head/Neck graft | |
| No | 0 |
| Yes | 1 |
| Trunk graft | |
| No | 0 |
| Yes | 1 |
Points are added to produce final risk score. TBSA indicates total body surface area burned.
FIGURE 2.
Risk at admission of developing HO by risk score. The x-axis is each possible point sum with the count for each point sum below it. The bar in the middle of each box represents the median predicted risk, and the top and bottom of the box are the 75th and 25th percentiles of the distribution, respectively. The whiskers cover 1.5 times the difference between the 25th and 75th percentiles, and the dots represent values that are outside of this range. The horizontal line indicates the prevalence of HO among the study population (3.6%).
Model Validation
The MGH External Database included 1381 patients. In this database, 26 subjects (1.9%) developed HO. As described previously, the lower incidence of HO in this dataset is expected and attributed to the difference in inclusion criteria of the 2 databases. The population exhibited a mean age of 39.9 years, a mean TBSA of 10.8%, and was 72.5% male. The MGH External Database population is consistent with the national population of adults hospitalized for acute burn injury.23
The calibration model was estimated by computing the linear predictor of the MGH External Database with reference to the model from the original BMS Database. The c-statistics and the calibrations of the 2 models were compared. The slope and constant were updated so that they would be as close to 0 and 1, respectively, as possible. This updating of the coefficients resulted in a constant of 0.041 and a slope of 1.0. The c-statistics of the BMS Database model (0.919) and the MGH External Database model (0.957) were very similar, thus the updating was successful. In addition to minor changes in the intercept and slope, the updating resulted in a decrease in the number of possible points from 14 to 13. The number of points associated with an arm graft changed from 2 to 1.
DISCUSSION
This is the first paper to develop an HO risk scoring system in the burn population based on admission characteristics. Burn size and deep burns at specific locations (arm, head/neck, and trunk) are significant predictors of HO development. This scoring system will help stratify patients at the time of admission for HO risk thereby advancing the pursuit of diagnostic and prophylactic measures as well as our understanding of HO pathophysiology.
HO risk stratification will help identify a population appropriate for early diagnostic imaging. It is most efficacious and safe to employ diagnostic imaging for those at a higher risk of ectopic bone formation to avoid unnecessary testing. Selective diagnostic testing of at-risk subpopulation allows for earlier diagnosis. Further, novel diagnostic modalities, such as Spect computed tomography (CT), Raman spectroscopy, and near infrared imaging,24–26 are best tested in higher risk populations.
Establishing a subpopulation at a high risk for HO will help future investigators evaluate potential prophylactic interventions. There is sparse research on the efficacy of prophylaxis in the burn population; in addition, prophylactic options with demonstrated efficacy in other populations exhibit risks. NSAIDs, used in the spinal cord injury population, not only inhibit osteogenic differentiation of progenitor cells27 but also increase the risk of long-bone nonunion, gastrointestinal bleeding, and renal failure. Another prophylactic option, bisphosphonates, carry risk of gastrointestinal erosion, osteonecrosis, and hypocalcemia.28 Radiation therapy, with efficacy in the hip arthroplasty population,29 is associated with an increased risk of malignancy and requires patient compliance. These prophylactic options are not well studied in the burn population because of associated risks coupled with the low incidence of HO. This risk scoring system will help identify burn patients at highest risk for entrance into clinical trials for prophylactic interventions.
In addition, this risk scoring system will facilitate exploration of the underlying HO mechanism and further our understanding of the shared pathophysiology of postinjury HO in other populations. HO is a similarly devastating complication of brain injury, spinal cord injury, blast injury, hip arthroplasty, elbow surgery, and amputation. The pathophysiology is not well understood in any of these populations; therefore, identification of high-risk groups is needed across all populations to increase our understanding of HO. This study represents a first major step in this direction. Our knowledge of the pathophysiology of HO is primarily informed by animal studies because of the difficulty of studying mechanistic pathways in a low incidence problem in humans. Recent animal data support the role of inflammation in HO development. Inflammation after burn causes a release of cytokines that spur osteogenic differentiation.9,10 Larger burns are at an increased risk of HO development. Therefore, TBSA could be an indicator of the level of inflammatory response. A risk stratification system has the potential to foster mechanistic human studies by identifying those at highest risk of HO formation.
A risk scoring system is an objective way to identify a segment of the population that may benefit from greater monitoring, diagnostic testing, prophylaxis, and potentially treatment. For example, the Framingham Heart Study algorithms have been used by clinicians and researchers to help determine treatment guidelines.30 Other point-based risk scoring systems are available to the public on the internet and mobile applications. The authors have created an online burn HO risk calculator to foster its use in clinical and research settings (www.spauldingrehab.org/HOburncalculator).
In deciding a threshold for determining high risk for HO development, the authors recommend weighing the risk of the intervention against the risk of HO development. For example, one may consider targeting all those with above average risk of HO development with a minimal risk intervention such as gentle passive range of motion. In contrast, for a prophylactic intervention with a higher risk, such as NSAIDs or radiation, one may decide to target those at very high risk of HO. This model will aid researchers in determining subpopulations with HO risk commensurate with the risk of the prophylactic or diagnostic intervention. Furthermore, lessons from this calculator can be used to help develop similar calculators for other at risk populations such as trauma, hip replacement, spinal cord, and traumatic brain injury patients.
Although this study employs a retrospective design, it is the largest cohort of burn patients ever studied for the assessment of HO risk. Investigations of low prevalence diseases, such as HO in burns, commonly employ case-control study design due to challenges in recruiting large, prospective cohorts. Prior studies examining HO risk factors were limited by single-center design and small sample size. Although prior research identified predictors for the development of HO, this is the first risk scoring system for HO in burns with ability to assist in making clinical decisions and intervention guidelines. This study did not account for prophylactic measures received by study subjects; however, prophylactic interventions do not have established efficacy in the burn population and these data are not contained in the BMS Database. The anatomical location of HO is not collected in the BMS Database. Nonetheless, the goal of this study was to predict the development of HO regardless of site and most proposed prophylactic interventions are systemic.
CONCLUSIONS
This 13-point risk scoring system stratifies burn patients by risk of HO development at time of admission. The risk scoring system will aid future investigation of HO physiology and prophylaxis by identifying high-risk subpopulations.
ACKNOWLEDGMENT
Authors would like to thank Jacqueline Bailey for reviewing medical records.
External funding for this study was provided by grants NIDILRR #90DP0035; NIH/National Institute of General Medical Sciences Grant K08GM109105-0; Plastic Surgery Foundation National Endowment Award; the Association for Academic Surgery Roslyn Award; American Association for the Surgery of Trauma Research & Education Foundation Scholarship; DOD: W81XWH-14-DMRDP-CRMRP-NMSIRA and American Association of Plastic Surgery Research Fellowship. This work was also partially supported by DOD work units W81XWH-14-2-0010 and 602115HP.3720.001.A1014.
The contents of this manuscript were developed under a grant from the National Institute on Disability, Independent Living, and Rehabilitation Research, NIDILRR grant number 90DP0035. NIDILRR is a Center within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The contents of this manuscript do not necessarily represent the policy of NIDILRR, ACL, HHS, and you should not assume endorsement by the Federal Government.
Footnotes
This is a multicenter longitudinal database study utilizing the Burn Model System National Database.
No reprints will be available.
This work was prepared as part of their official duties. Title 17 U.S.C. §105 provides that “Copyright protection under this title is not available for any work of the United States Government.” Title 17 U.S.C §101 defined a US Government work as a work prepared by a military service member or employees of the United States Government as part of that person’s official duties. The opinions or assertions contained in this paper are the private views of the authors and are not to be construed as reflecting the views, policy, or positions of the Department of the Navy, Department of Defense nor the United States Government.
The authors report no conflicts of interest.
REFERENCES
- 1.Medina A, Shankowsky H, Savaryn B, et al. Characterization of heterotopic ossification in burn patients. J Burn Care Res. 2014;35:251–256. [DOI] [PubMed] [Google Scholar]
- 2.Chen HC, Yang JY, Chuang SS, et al. Heterotopic ossification in burns: our experience and literature reviews. Burns. 2009;35:857–862. [DOI] [PubMed] [Google Scholar]
- 3.Levi B, Jayakumar P, Giladi A, et al. Risk factors for the development of heterotopic ossification in seriously burned adults. J Trauma Acute Care Surg. 2015;79:870–876. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Orchard GR, Paratz JD, Blot S, et al. Risk factors in hospitalized patients with burn injuries for developing heterotopic ossification: a retrospective analysis. J Burn Care Res. 2014;36:465–470. [DOI] [PubMed] [Google Scholar]
- 5.Hunt JL, Arnoldo BD, Kowalske K, et al. Heterotopic ossification revisited: a 21-year surgical experience. J Burn Care Res. 2006;27:535–540. [DOI] [PubMed] [Google Scholar]
- 6.Tsionos I, Leclercq C, Rochet JM. Heterotopic ossification of the elbow in patients with burns. Results after early excision. J Bone Joint Surg Br. 2004;86:396–403. [DOI] [PubMed] [Google Scholar]
- 7.Shafer DM, Bay C, Caruso DM, et al. The use of eidronate disodium in the prevention of heterotopic ossification in burn patients. Burns. 2008;34:355–360. [DOI] [PubMed] [Google Scholar]
- 8.Crawford CM, Varghese G, Mani MM, et al. Heterotopic ossification: are range of motion exercise contraindicated? J Burn Care Res. 1986;7:323–327. [DOI] [PubMed] [Google Scholar]
- 9.Peterson JR, De La Rosa S, Sun H, et al. Burn injury enhances bone formation in heterotopic ossification model. Ann Surg. 2014;259:993–998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ramirez DM, Ramirez MR, Reginato AM, et al. Molecular and cellular mechanisms of heterotopic ossification. Histol Histopathol. 2014;29:1281–1285. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Agarwal S, Loder S, Brownley C, et al. Inhibition of Hif1(prevents both trauma-induced and genetic heterotopic ossification. Proc Natl Acad Sci U S A. 2016;113:E338–E347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Agarwal S, Loder S, Li J, et al. Diminished chondrogenesis and enhanced osteoclastogenesis in leptin-deficient diabetic mice (ob/ob) impair pathologic, trauma-induced heterotopic ossification. Stem Cells Dev. 2015;24:2864–2872. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Loder S, Agarwal S, Sorkin M, et al. Lymphatic contribution to the cellular niche in heterotopic ossification. Ann Surg. 2016. [Epub ahead of print]. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ranganathan K, Loder S, Agarwal S, et al. Heterotopic ossification: basic-science principles and clinical correlates. J Bone Joint Surg Am. 2015;97:1101–1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Klein MB, Lezotte DC, Heltshe S, et al. Functional and psychosocial outcomes of older adults after burn injury: results from a multicenter database of severe burn injury. J Burn Care Res. 2011;32:66–78. [DOI] [PubMed] [Google Scholar]
- 16.Newgard CD, Lewis RJ. Missing data. JAMA. 2015;314:940–941. [DOI] [PubMed] [Google Scholar]
- 17.Sterne J, White I, Carlin J, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. Brit Med J. 2009;338:b2393. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sullivan LM, Massaro JM, D’Agostino RB. Presentation of multivariate data for clinical use: the Framingham Study risk score functions. Stat Med. 2004;23:1631–1660. [DOI] [PubMed] [Google Scholar]
- 19.Lezotte DC, Hills RA, Heltshe SL, et al. Assets and liabilities of the Burn Model System data model: a comparison with the National Burn Registry. Arch Phys Med Rehabil. 2007;88(12 Suppl 2):S7–S17. [DOI] [PubMed] [Google Scholar]
- 20.Steyerberg EW. Clinical Prediction Models: A Practical Approach to Development, Validation, and Updating (Statistics for Biology and Health). New York, NY: Springer; 2010. [Google Scholar]
- 21.Moons KG, Kengne AP, Grobbee DE, et al. Risk prediction models: II. External validation, model updating, and impact assessment. Heart. 2012;98:691–698. [DOI] [PubMed] [Google Scholar]
- 22.Steyerberg EW, Vergouwe Y. Towards better clinical prediction models: seven steps for development and an ABCD for validation. Eur Heart J. 2014;35:1925–1931. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.American Burn Association. National Burn Repository 2015 Report Version 11.0. Chicago, IL: American Burn Association; 2015. [Google Scholar]
- 24.Peterson JR, Okagbare PI, De La Rosa S, et al. Early detection of burn induced heterotopic ossification using transcutaneous Raman spectroscopy. Bone. 2013;54:28–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Perosky JE, Peterson JR, Eboda ON, et al. Early detection of heterotopic ossification using near-infrared optical imaging reveals dynamic turnover and progression of mineralization following Achilles tenotomy and burn injury. J Orthop Res. 2014;32:1416–1423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Brownley RC, Agarwal S, Loder S, et al. Characterization of heterotopic ossification using radiographic imaging: evidence for a paradigm shift. PLoS One. 2015;10:e0141432. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Teasell RW, Mehta S, Aubut JL, et al. A systematic review of the therapeutic interventions for heterotopic ossification after spinal cord injury. Spinal Cord. 2010;48:512–521. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sullivan MP, Torres SJ, Mehta S, et al. Heterotopic ossification after central nervous system trauma. Bone Joint Res. 2013;2:51–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Ayers DC, Pellegrini VD, Evarts CM. Prevention of heterotopic ossification in high-risk patients by radiation therapy. Clin Orthop Relat Res. 1991;263: 87–93. [PubMed] [Google Scholar]
- 30.National Cholesterol Education Program. (NCEP) Expert Panel. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Arch Intern Med. 2002;6:284. [Google Scholar]