Using Cutaneous Functional Units (CFUs) to Understand Burn Characteristics Associated With Range of Motion at Hospital Discharge in Adult Burn Survivors

Abstract

Background

Burn injuries often lead to the development of burn scar contractures (BSC) and subsequent motion loss, which significantly impact function and quality of life for burn survivors. Understanding the relationship between burn characteristics and motion outcomes is critical to improving patient care and outcomes. We used a Cutaneous Functional Units (CFU) framework to investigate acute burn features at or near the time of injury associated with a joint-level range of motion at hospital discharge.

Methods

This study is an exploratory analysis of data collected as part of a previous prospective multicenter cohort study (N = 307 patients, 7441 joint motions). CFU features, including burn extent, skin grafting, burn location (proximal/ distal to joint crease), and adjacent CFU involvement, were evaluated while controlling for demographic and clinical variables. Fractional regression models evaluated associations between CFU burn features and range of motion.

Results

A higher percentage of CFU-burned and CFU-grafted, as well as advanced age, were significantly associated with reduced range of motion (P < .01). Burn extent in the proximal CFU areas also had a significant negative relationship with range of motion (P < .01). Total body surface area (TBSA) burned and TBSA-grafted were not associated with motion outcomes. Based on this exploratory analysis, a prediction model for BSC risk is proposed for further testing.

Conclusion

Our findings highlight the significance of burn characteristics within CFUs as key factors influencing motion outcomes. The CFU framework provides a standardized and functionally relevant approach to evaluating the localized impact of burn characteristics on joint motion, addressing a critical gap in the understanding of motion loss following burn injuries.

INTRODUCTION

Body and limb motions are fundamental to daily life, providing the freedom for self-care, functional mobility, and recreational, work, and school activities. The formation of a contracture due to scarring after burn injury often results in restricted motion, affecting 38%-80% of burn survivors at the time of hospital discharge.^1,2 Burn scar contracture (BSC) is commonly defined as an “impairment caused by the replacement of skin with pathological scar tissue of insufficient extensibility and length resulting in a loss of motion or tissue alignment of an associated joint or anatomical structure.”^3(p547) Motion loss due to BSC can lead to functional impairment and emotional distress for survivors^4–6 and can impact a person’s quality of life, psychosocial health, and ability to work.^7–9 Improved understanding of factors related to the development of BSCs could enhance prevention efforts and improve patient outcomes.

Early and effective treatments after a burn injury preserve motion and prevent BSCs from occurring.^2,10–12 However, the incidence of BSCs at hospital discharge is not declining despite medical and surgical advances.^1,2,13 This may, in part, be related to how motion or loss of motion is assessed in the burn-injured population. Typically, human motion is evaluated using osteokinematics, which involves studying the angular movement of bones within planes, or arthokinematics, which focuses on more detailed movements within the joints. Offering a neologism using a comparable portmanteau, cutaneokinematics, therefore, considers the biomechanical ability of the skin to elongate and accommodate movement, which is often overlooked when evaluating motion.¹⁴ This aspect of motion becomes particularly relevant when the skin is impaired, such as in the case of a burn injury.

Cutaneokinematics offers a new paradigm for evaluating the loss of motion after a burn injury. Cutaneous Functional Units (CFUs) are fields of skin that functionally contribute to a range of motion and, collectively, they create a useful framework for applying the concept of cutaneokinematics.¹⁵ CFUs were first described by Richard et al.¹⁵ who used double exposure photography with noninjured adult males to evaluate areas of skin that accommodate motion at nearby joints. This study defined nine CFU areas and discovered that most skin movements occurred close to the joint being tested. However, skin movement also occurred quite distant from the joint, such as over the pelvis during neck extension when a person is lying supine. Subsequent research used a computerized wound mapping program and applied hierarchical decomposition to define 182 CFUs.¹⁶ Exploring motion loss after burn injury through the lens of CFUs could enhance our understanding of the cutaneokinematics associated with burn-scarred skin and its effect on BSC formation.^16,17 A graphical video abstract demonstrates the concept of cutaneokinematics for torso skin movement associated with shoulder abduction in uninjured skin, immature scars, and mature scars (Supplementary Video Abstract).

Studies have shown that when scars replace normal skin during the repair of burn wounds, the tissue is weaker, stiffer, and less extensible.^18–21 These changes can affect the ability of the scar to elongate or move to accommodate motion at nearby joints. Only a few studies have used a CFU framework to evaluate BSCs, and the results of those studies demonstrate a relationship between more scarring in the associated CFU and less range of motion recovery²² or more severe motion limitations.²³ These previous studies investigated limited body areas and only examined the percentage of burn injury within the CFU. Other body areas and burn characteristics that may influence scarring and outcome, such as depth of injury, skin grafting, location within the CFU, and adjacent CFU involvement, still need to be investigated.

To address this gap, the purpose of this study was to evaluate the relationship between multiple burn features within CFUs and joint range of motion for 86 body motions. Using a CFU framework, our specific objective was to quantify which acute burn characteristics at or near the time of injury were associated with a joint-level range of motion measured at hospital discharge. We also explored the utility of using a CFU framework to better understand motion loss after a burn injury and propose a prediction model. Additionally, we provide detailed descriptions and coding of the CFU hierarchical framework to be used as a reference in future research to facilitate more consistent application of CFUs in burn care and rehabilitation.

METHODS

Study design and data source

This study is a secondary analysis of data collected from a multicenter, prospective longitudinal cohort study of adult patients hospitalized for burn injury. Funded by the United States Department of Defense, the “Burn Patient Acuity Demographics, Scar Contractures, and Rehabilitation Treatment Time Related to Patient Outcomes” study (ACT)²⁴ enrolled patients receiving burn care at 13 different American Burn Association (ABA) Verified Burn Centers between September 2010 and December 2013. The dataset for this analysis was obtained from the ABA upon written request.²⁵ The ACT dataset includes demographic, injury, treatment, and outcome data for 307 adult burn survivors.²⁴ Care during the ACT cohort study was provided per the standards at each burn center.

Protection of human subjects

The primary ACT study was reviewed and approved by the US Army Medical Research and Materiel Command Institutional Review Board, and subjects provided informed consent according to the approved protocol.²⁴ For this secondary analysis of de-identified data, the University of California, Davis Institutional Review Board determined the study did not meet the definition of human subject research.

Eligibility criteria

Figure 1 shows the eligibility criteria for this secondary analysis, which were applied at the level of joint motion (eg, elbow flexion, wrist extension, etc.). The full dataset includes N = 17 704 joint motions nested within 307 patients. When the range of motion was collected during the ACT study, investigators reported conditions contributing to the motion deficit when measured at discharge. For this analysis, joint motions affected by a concomitant (eg, fracture) or preinjury (eg, arthritis) condition were excluded. However, if the patient contributed data for other joint motions affected by the burn but not a concomitant or preburn condition, those observations were retained for analysis.

Figure 1.

Flow diagram of exclusion criteria applied at the level of the joint motion.

In the dataset, finger motion was measured in two ways, isolated, representing one joint motion with the immediate proximal and distal joints positioned in the opposite direction as the joint being measured, and composite, representing the same joint motions but with the immediate proximal and distal joint positioned in the same direction as the joint being measured. Since these measurements are duplicate measurements of the same joint under two different movement conditions, all composite motions were excluded because the isolated motions most closely represent the measurement conditions of the other body joint motions in the dataset.

Some thumb CFUs were not coded to the correct motion in the burn diagram output and were excluded from the analysis. Mouth-opening data was excluded because no normal references were provided from the primary ACT study Standard Operating Procedures (SOP), and neck rotation and lateral flexion were excluded because the burn diagram did not capture a related CFU area. After excluding ineligible motions and missing data, the final sample for this secondary analysis included n = 7441 joint motions nested within 307 patients.

Measurement

Definitions and levels of measurement for all the variables are described in detail in Table 1.

Table 1.

Definition of Predictor Variables and Coding for Study Analyses

Variable	Area for analysis	Measurement defined	Level of measurement
Predictors
CFU-Burned	Primary CFU	Amount of any depth burn within the primary CFU.a	Continuous proportion (0%-100%)
CFU-Grafted	Primary CFU	Amount of any type of skin grafting within the primary CFU.a	Continuous proportion (0%-100%)
CFU-Depth	Primary CFU	Amount of full-thickness burn within the primary CFU.a	Continuous proportion (0%-100%)
CFU-Proximal	Subdivision of primary CFU	Amount of burn within the proximal portion of the primary CFU.b	Continuous proportion (0%-100%)
CFU-Distal	Subdivision of primary CFU	Amount of burn within the distal portion of the primary CFU.c	Continuous proportion (0%-100%)
CFU-Adjacent	Adjacent CFU	Amount of burn within the adjacent CFU.d	Continuous proportion (0%-100%)
Outcome
Range of motion	Passive joint range of motion measured at hospital discharge and linked to primary CFU	Proportion of available range of motion to expected normal range of motion- converted from degrees to proportions.	Continuous proportion (0-1)
Control variables
TBSA-Burned	Whole body	Percentage of the body that sustained any depth of burn injury was converted to a proportion for analysis.	Continuous percentage (0%-100%)
TBSA-Grafted	Whole body	Percentage of the body that received any type of skin grafting was converted to a proportion for analysis.	Continuous percentage (0%-100%)
For ease of interpretation, TBSA and skin grafting predictors were converted to units of 20 percentage points.
Age	Patient-level age	Age of the patient upon hospital admission	Continuous
Sex	Patient-level gender	Reported gender of the patient. Options: male, female, other, unknown	Categorical: - Female - Male - Missing, other, unknown
Race	Patient-level race	Reported primary race of the patient. Options: African American, American Indian or Alaskan Native, Asian, Native Hawaiian or other Pacific Islander, Not Reported, Unknown, White	Categorical: - White - Non-White - Missing, not reported, or unknown
Comorbidity	Patient-level presence of comorbidity	Reported at least one of the defined preexisting medical conditions: diabetes, lupus, circulatory disorder, hypertension, renal insufficiency, hepatitis, HIV/AIDS, cancer, seizure history, pulmonary diagnosis, dementia	Dichotomous: - Not present - Present
Movement Code	Individual joint motion	86 different coded motions from the original ACT dataset paired with a designated CFU	Fixed effects for 85 motions
Burn Center	Burn center where the patient received care	13 different burn centers participating in the study and individually coded	Fixed effects for 12 burn centers

^aPrimary CFU is defined as the largest defined area of skin closest to the skin crease where motion is being assessed per the CFU hierarchy.

^bProximal CFU is defined as the portion of the primary CFU closest to the joint crease if the primary CFU has a designated subdivision defined per the CFU hierarchy.

^cDistal CFU is defined as the portion of the primary CFU farthest from the joint crease if the primary CFU has a designated subdivision defined per the CFU hierarchy.

^dAdjacent CFU is defined as the smallest distinct skin unit immediately anatomically proximal to the primary CFU per the CFU hierarchy.

Burn injury characteristics

The primary independent variables of interest for this study were five CFU burn injury features derived from the original dataset: (1) percentage of burn injury within the primary CFU (CFU-Burned), (2) percentage of skin grafting within the primary CFU (CFU-Grafted), (3) extent of full-thickness burn within the CFU (CFU-Depth), (4) location of burn within the CFU, using subdivisions of areas proximal or distal to the joint crease (CFU-Proximal, CFU-Distal), and (5) percentage of burn within the adjacent CFU (CFU-Adjacent). Data for the CFU characteristics were quantified using a web-based electronic body diagram called Surface Area Graphic Evaluation (SAGE).²⁶ The computerized SAGE diagram was developed based on the Lund and Browder burn diagram and in 2015, CFUs were integrated.^16,26 Richard et al. modified the SAGE diagram with the capability to identify 182 distinct CFUs and 401 total nested CFU codes.¹⁶ A SAGE diagram was completed for each ACT study patient at the following time points: (1) on post-admission day 5, (2) 3 weeks after admission, and (3) following each skin graft procedure. Data from a final diagram reflecting all burn characteristics was used for this analysis.

Range of motion

The primary outcome variable for this study was a joint range of motion measured at hospital discharge and analyzed at the level of the joint for each motion (eg, knee flexion range of motion, knee extension range of motion). During the ACT study, patients’ passive range of motion was measured using standard goniometry methods and recorded in degrees.²⁷ For this secondary analysis, the range of motion data was converted from degrees to a proportion of the expected normal motion for that particular joint motion. For example, if the measured range of motion for elbow flexion is 120 °C, and the typical value is 140 °C, the proportion is 0.86. This allowed for different joint motions to be compared on the same scale. Specifically, we used the recorded available range of motion divided by the expected normal range of motion, according to the ACT study SOP training manual based on published typical values for normal range of motion.^24,27 This is provided as Supplementary Appendix A. We evaluated 86 motions measured at hospital discharge (including multiple finger joints and both sides of the body).

Control variables

Control variables were determined based on previous research and availability in the ACT database.^{13,23,28–30} These included total body surface area burned (TBSA-Burned), percentage of body level skin grafting (TBSA-Grafted), age, sex, race/ethnicity, and presence of medical comorbidities (Table 1).

CFU coding

For every joint motion, a “primary” CFU was identified based on the hierarchical decomposition of codes and defined as the largest area of skin closest to the skin crease or joint where most skin movement is believed to occur. These areas were coded using previous research,¹⁵ the ACT hierarchical diagram,¹⁶ and for any areas that had uncertainty from previous publications, discussions with the principal investigator of ACT (personal communication with Reg Richard MS, PT, BT-C, February-April, 2023). If the primary CFU area was linked to more than one motion (eg, anterior forearm skin is linked to wrist extension and supination), then the code was expanded within the statistical software and recoded using a unique code associated with each motion. The primary CFUs were coded and matched to their associated joint motions. If a primary CFU contained nested CFU codes (subdivisions), the numeric codes for those subdivided areas were used to determine the burn location as proximal or distal to the joint crease (different from anatomically proximal and distal) within the primary CFU. Although CFUs are coded as distinct areas of the body for this study, it is important to remember that skin is continuous in nature. Therefore, to account for injury near but outside of the identified primary CFU, we also defined an “adjacent” CFU. These were the smallest distinct units of skin anatomically proximal to the primary CFU, which may have been a primary or subdivision of the CFUs for the adjacent motion. The codes and linked movements are offered in a separate publication in table and diagram format for reference and consistent use in future research.

Data cleaning

Data preparation and analysis were conducted using Stata SE version 18.0 (StataCorp, College Station, TX, USA). After all movement codes were linked with CFU codes, any duplicate or missing data was excluded (Figure 1). Data were typically missing in situations where range of motion values were available for CFU areas not affected by the burn, or nested CFU areas were not linked to a motion. In the case of duplicates, the most complete record was maintained. Range of motion data for each movement code was transformed to a proportion of the expected normal. Joint-level and patient-level data were merged, organized, and checked for completeness and structure. Data were examined for outliers. Consistency and logical checks were performed on the data to ensure data integrity.

Statistical analyses

Descriptive statistics, specifically measures of central tendency (mean and median) and measures of variability (standard deviation and interquartile ranges) are reported to summarize patient and joint-level data. Frequency distributions describe the occurrence of patients by sex, race/ethnicity, mechanisms of injury, and presence of skin grafts or comorbidities.

The primary analysis sought to explore which burn features (ie, CFU-Burned, CFU-Grafted, CFU-Depth) within the associated primary CFUs were associated with joint range of motion at hospital discharge while controlling for TBSA-Burned, TBSA-Grafted, age, sex and race, burn center, and individual joint motion. Variables represented as percentage points (TBSA and CFU) were recoded into units of 20 percentage points for ease of clinical interpretation. Subanalyses were conducted to determine if additional burn features (ie, CFU-Proximal, CFU-Distal, and CFU-Adjacent) were related to a mean range of motion outcome. An additional subanalysis was conducted to repeat the primary analysis in a subset of observations excluding edema and open wounds as temporary conditions noted to contribute to a range of motion deficits. Observations limited by pain were retained in the subanalysis with the rationale that pain at discharge, even if temporary, may resolve more slowly and have an enduring impact on motion. Furthermore, literature supports that pain is associated with lower physical functioning over time suggesting a lasting impact on motion.³¹

Fractional response models were used to regress joint range of motion (as a proportion of expected normal) onto CFU burn characteristics, controlling for covariates. Fractional regression was chosen because it can be used to model variables that take on values within a bounded range.³² Unlike linear regression, predicted values for fractional regression are guaranteed to remain within the bounds of 0 and 1, but unlike beta regression, they may include the bounds of 0 and 1.^33–35 Fractional regression can also help avoid biased estimates of the regression coefficients which can occur when standard linear regression models are applied to bounded data.³⁶ This statistical approach is a quasi-likelihood estimation method that is robust to the misspecification of some aspects of the underlying distribution.³⁴ Significance was set at P ≤ .05.

The logit link function was used because of its comparable mathematical simplicity and familiar log odds interpretation. The results are interpreted with an effect size corresponding to the odds ratio but instead of the ratio applying to an odds, it applies to a range of motion index. The range of motion index is defined as a ratio with the conditional expected value of the proportion of the range of motion in the numerator and the complement of this (1 minus the conditional expected value) in the denominator. For simplicity, we will refer to our effect size estimates as odds ratios (OR). Since one model was being estimated for all motions at once, the issue of different amounts of normally available motion by joint needed to be resolved (eg, the ankle dorsiflexion has an average reported normal value of 20 °C of motion compared to shoulder flexion with 180 °C). Degrees of motion were rescaled to a proportion of the maximum so that for all joints, the outcome ranged from 0 (no movement) to 1 (full movement).

Patients contributed more than one motion to the dataset if burned in multiple CFUs. Multiple joint motions from the same person are unlikely to be conditionally independent.³⁷ Similarly, each burn center enrolled multiple patients. Therefore, a robust sandwich estimator accounted for these cluster effects (ie, motions nested within patients and patients nested within centers) when estimating standard errors and test statistics.³³ Both joint motion and burn center were included as fixed effects in this exploratory model to account for any explainable variation in mean levels of the outcome between parts of the body or participating burn centers.

From this exploratory model, we used statistical and clinical justification to recommend a parsimonious final model. Statistical considerations included comparing a small set of candidate models using varying combinations of predictor variables and goodness-of-fit metrics. The Stata “margins” command was used to estimate and interpret the predictive margins and marginal effects based on the final fitted model.

RESULTS

Data from 307 patients and 7441 joint motions were analyzed after exclusion criteria were applied (Figure 1). Demographic and clinical characteristics of the sample are shown in Table 2. The mean (SD) age of the patients was 43.59 years (±17.01, range 18-87) years. Median TBSA-Burned was 8.19% (IQR: 11.31, range 0.50%-67.22%) and length of hospital stay (LOS) was 14 days (IQR: 12, range 4-146 days). Most patients were male (70.68%), White (81.76%), and most commonly burned by flame injury (71.34%). At least one medical comorbidity was noted upon admission for 29.97% of patients. Most patients underwent skin grafting (79.48%), and of those, the median TBSA-Grafted was 4.85% (IQR: 7.00, range 0.22%-54.40%).

Table 2.

Demographic and Clinical Characteristics of the Sample

Patient-level characteristics	Subjects (307)
	Mean (SD, range)
Age (years)	43.59 (17.01, 18-87)
	Median (IQR, range)
TBSA-Burned (%)	8.19 (11.31, 0.50-67.22)
TBSA-Grafted (%)	4.85 (7.00, 0.22-54.40)
LOS (days)	14 (12, 4-146)
BMI on admission	27.38 (7.64, 15.19-49.01)
Number of CFU-burned per patient	19 (27, 1-86)
Number of joint motions per patient with limited ROM	8 (15, 0-64)
	Frequency n (%)
Sex
Female	90 (29.32)
Male	217 (70.68)
Race
White	251 (81.76)
Non-White	32 (10.42)
Missing/Unknown/Not reported	24 (7.82)
Burn etiology
Flame	219 (71.34)
Tar/grease/oil	35 (11.40)
Hot Liquid	28 (9.12)
Contact	16 (5.21)
Friction	6 (1.95)
Chemical	2 (0.65)
Hot Gas	1 (0.33)
Underwent at least one skin graft procedure	244 (79.48)
Had at least one medical comorbidity	92 (29.97)
Joint-level characteristics	Joint motions (7441)
	Mean (SD, range)
Amount of CFU area burned (CFU-Burned) (%)	67.27 (35.28,0.01-100)
Amount of CFU area skin grafted (CFU-Grafted (%)	60.71 (36.34, 2-100)
Percent of available motiona (%)	67.78 (24.37, 0-99.41)

^aFor motions with less than 100% (n = 3595).

Medians reported if kurtosis >3 and skewness >~0.

BMI, body mass index; CFU, cutaneous functional units; IQR, interquartile range; LOS, length of hospital stay; ROM, range of motion; TBSA, total body surface area.

Medical comorbidities noted upon admission include cancer, circulatory disorder, hypertension, pulmonary diagnosis, diabetes, dementia, renal insufficiency, hepatitis, seizure history, and lupus.

Non-White includes African American, Asian, American Indian/Alaskan Native, Native Hawaiian/ Pacific Islander.

Patients had a median of 19 (IQR: 27, range 1-86) primary CFU areas burned and 8 (IQR: 15, range 0-64) motions with a limited range of motion. CFU and joint-level data are also included in Table 2. The mean (SD) extent of burn within the CFUs was 67.27% (35.28%, range 0.01%-100%). For subjects who had skin grafting, the mean (SD) extent of skin grafting within the CFU was 60.71% (36.34%, range 2.00%-100%). For all subjects and observations (joint motions), the mean (SD) percentage of available motion at hospital discharge was 84.50% (23.35%, range 0%-100%). Just under half of the observations (48.11%) demonstrated less than normal range of motion. For those motions, the mean (SD) percentage range of motion available was 67.78% (24.37%, range 0-99.41) of expected normal motion.

Table 3 presents the results from the full fractional regression model for the main analysis. A higher proportion of CFU-Burned, CFU-Grafted, and higher age was significantly associated with motion loss at hospital discharge. For every unit change (20 percentage points) in the extent of burn within the CFU, the proportion of the range of motion decreases significantly (OR = 0.88, 95% CI: 0.83-0.93, P < .01) at hospital discharge. Similar results were found for a unit change in the percentage of skin grafting within the CFU (OR = 0.92, 95% CI 0.87-0.97, P < .01) and for every year increase in age (OR = 0.98, 95% CI 0.98-0.99, P < .01). TBSA-Burned, TBSA-Grafted, sex, race, or presence of comorbidity were not significantly associated with a range of motion outcome at hospital discharge. The percentage of deep-thickness burn within the CFU was also not associated with a range of motion outcome. The marginal effects for 10%, 25%, 50%, 75%, and 90% of CFU-Burned and CFU-Grafted by age are depicted in Figure 2a and b using the final model described below for visualization of the magnitude and direction of the relationships.

Table 3.

Fractional Regression Explanatory Model to Explore Proof of Concept Regarding the Relationships of the Variables.

	Number of observations				7437
	Prob > chi 2				0.00
Proportion of ROM	Odds ratio	Robust std. err.	P > |z|	[95% conf. interval]
TBSA-Burneda	1.28	0.22	0.15	0.92	1.79
TBSA-Grafteda	0.63	0.15	0.06	0.40	1.01
CFU-Burneda	0.88	0.02	0.00	0.83	0.93
CFU-Grafteda	0.92	0.03	0.00	0.87	0.97
CFU-Deptha	0.98	0.06	0.68	0.87	1.10
Sex (male reference)	1.01	0.14	0.95	0.78	1.32
Comorbidity	1.05	0.14	0.70	0.82	1.37
Age	0.98	0.00	0.00	0.98	0.99
Race (White reference)
Non-White	0.95	0.19	0.79	0.64	1.40
Missing	0.78	0.12	0.08	0.59	1.03
_cons	58.61	22.43	0.00	27.69	124.07

Results: odds ratios (interpreted as a range of motion index) for the proportion of the range of motion at discharge per unit in the predictor. Robust standard errors, statistical significance (P < .05), and 95% confidence intervals are also reported.

^aVariables represented as percentage points were converted into units of 20 percentage points for ease of clinical interpretation.

Movement code and burn center were included in the model as fixed effects (ORs not shown).

CFU, cutaneous functional units; CI, confidence interval; ROM, range of motion; SE, standard errors; TBSA, total body surface area.

Figure 2.

(a) Margins graph of the proportion of joint motion by CFU-Burned and age. (b) Margins graph of the proportion of joint motion by CFU-Grafted and age.

Prediction model development

CFU predictors that significantly contributed to explaining the variance in the range of motion in the primary analysis were CFU-Burned and CFU-Grafted. Age was also statistically significant and retained in the model. TBSA-Burned and TBSA-Grafted were not significant predictors but should be included in future research due to the frequency and ease of collecting the data and their theoretical importance established in previous research. Sex and race also did not significantly improve model performance in this sample; however, they are retained because they have theoretical and empirical importance despite exhibiting mixed associations in other burn research. There may be a genetic predisposition for scarring related to race, which supports the need for a more personalized approach to identifying BSC risk.³⁸ For the purposes of investigation in burn care, Fitzpatrick skin types may better capture the traits linked with scarring than race and should be considered as an alternative in future studies.³⁹ CFU-Depth did not improve model performance. Including the presence of at least one comorbidity also did not improve model performance and was removed for the final model. Therefore, considering these statistical and clinical justifications, based on 7437 observations, the most parsimonious model included CFU-Burned, CFU-Grafted, TBSA-Burned, TBSA-Grafted, sex, race, age, and joint motion and burn center (when appropriate) retained as fixed effects (Table 4).

Table 4.

Final Model Using Fractional Regression Including Predictors With Statistical and Clinical Justifications for Retention.

Final model	Number of observations				7437
	Prob > chi 2				0.00
Proportion of ROM	Odds ratio	Robust SE	P > |z|	95% CI
TBSA-Burneda	1.07	0.20	0.74	0.74	1.55
TBSA-Grafteda	0.84	0.23	0.51	0.49	1.42
CFU-Burneda	0.86	0.02	0.00	0.82	0.91
CFU-Grafteda	0.92	0.03	0.01	0.87	0.98
Sex (male reference)	0.99	0.14	0.96	0.76	1.30
Age	0.98	0.00	0.00	0.98	0.99
Race (White reference)
Non-White	0.86	0.23	0.57	0.50	1.46
Missing	0.76	0.23	0.07	0.56	1.02
_cons	68.53	25.89	0.00	32.68	143.70

Odds ratios (interpreted as a range of motion index) for the proportion of the range of motion at discharge per unit in the predictor. Robust standard errors, statistical significance (P < .05), and 95% confidence intervals are also reported.

^aVariables represented as percentage points were converted into units of 20 percentage points for ease of clinical interpretation.

Movement code and burn center were included in the model as fixed effects (ORs not shown).

CFU, cutaneous functional units; CI, confidence interval; ROM, range of motion; SE, standard errors; TBSA, total body surface area.

Subanalyses

A subanalysis using the exploratory model was conducted to evaluate the impact of burn location within the primary CFU (n = 2927 observations). Subunits (CFU-Proximal and CFU-Distal) were added to the model and CFU-Burned was removed, and again 20 percentage points were used for the units. This analysis found a significant negative association between the range of motion and burn within the proximal area of the CFU (OR = 0.86, 95% CI 0.81-0.92, P < .01) but no relationship with the proportion of burn within the distal area of the CFU (OR = 1.02, 95% CI 0.96-1.08, P = .71). An additional subanalysis of the association of CFU-Adjacent was conducted. Adding CFU-Adjacent to the original full model resulted in 6015 observations. No significant association was found between the extent of burn injury within the adjacent CFU and range of motion (OR = 0.98, 95% CI 0.93-1.02, P = .30).

Removing the observations from the dataset that were reported to have edema and open wounds, assuming that edema and open wounds are temporary limiters of motion, a subanalysis was run using the same variables as the primary analysis. There were 6589 observations left and fractional regression demonstrated that CFU-Burned (OR = 0.86, 95% CI 0.80-0.92, P < .01) and CFU-Grafted (OR = 0.91, 95% CI 0.85-0.98, P < .01) remained significantly associated with proportion of range of motion. Age (OR = 0.98, 95% CI 0.97-0.99, P < .01) also retained a significant relationship. All other variables showed no significant relationship to range of motion in this subset.

DISCUSSION

This study examined the relationship between acute burn characteristics and joint-level range of motion outcomes at hospital discharge using a CFU framework. Our findings demonstrated that the extent of burn injury and skin grafting within a CFU and burn extent within the proximal location of the CFU were significantly associated with motion outcomes at the related joints. These key findings reveal the importance of location-specific evaluation of burn factors when evaluating motion problems after a burn injury. The results provide valuable insights and direction for clinical care and future research.

CFU-specific features and motion outcomes

Our study found that a greater extent of CFU-Burned (ie, burn injury of any depth within the CFU) was significantly associated with less range of motion at hospital discharge. This finding is consistent with the limited studies that have investigated the topic. For example, our group previously studied children with burn injuries and reported a significant negative correlation between the percentage of CFU scarred and maximal shoulder abduction and flexion range of motion as measured with goniometry and 3D motion analysis within 6 months of injury.²² Another study by Lensing et al. used the ACT dataset to evaluate motion limitation as a categorical outcome of BSC severity and reported that the percentage of CFU-burned independently predicted moderate-severe limitations in range of motion.²³ Using the same dataset (ACT) but including joint motions of the hands, our study built upon these findings by representing range of motion as a continuous proportional variable, allowing for a more nuanced analysis.

Our study also identified novel associations for CFU-Grafted (ie, extent of skin grafting or skin substitute within the CFU) and CFU-Proximal (ie, extent of burn within the proximal location of the CFU relative to the joint crease) with motion outcomes. Skin grafts have less pliability than normal skin, thus theoretically reducing the ability for tissue within CFUs to elongate and permit motion at the associated joints.^18–21 Our study confirmed there is a link between CFU-Grafted and reduced motion, emphasizing the need for further investigation of the impact of skin grafts on motion. Specifically, studies need to explore if there is a minimal threshold of skin grafting within the CFU that influence motion and the types or thicknesses of skin grafts that have an impact.^20,40,41

The significant impact of burn injuries within the proximal portion of the CFU, but not the distal portion, aligns with previous studies suggesting skin movement occurs predominantly near joint creases^30,42 While our study quantified proximal and distal areas of a CFU using a reproducible framework, Schouten et al. previously reported that burns “across” a joint had more impact on joint limitation than burns “adjacent” to a joint.¹⁶ The CFU areas we used started at one joint crease and ended at the next joint crease, so a direct comparison to areas considered “across” and “adjacent” by Schouten and colleagues cannot be made. However, both studies highlight the need to quantify burn location in a standardized manner, and CFUs not only offer a consistent framework for this but have the advantage that they are also linked to specific joint motions.

It is challenging to define the true borders of skin areas that move and elongate to accommodate motion given that skin is a continuous structure. To date, the boundaries of only nine CFUs have been empirically tested, while the others have been defined as best as possible based on theory and clinical observations.^15,16 Our findings that the extent of burn within the proximal, but not distal CFU relative to the joint crease, is statistically associated with motion provide valuable information as the borders of CFUs are further investigated and iterated. The current CFU framework offers well-defined, coded segments of skin based on functionally relevant areas. To advance the science of cutaneokinematics, researchers need to consistently reference and build upon common and well-defined frameworks as evidence evolves. To support this goal, we have provided a detailed description of the movements and CFU codes used in this study and recommendations for additional CFUs as a separate reference for other authors.⁴³ Our findings and this resource can be used to help guide ongoing studies that aim to redefine or iterate the CFU borders and overall CFU hierarchy as knowledge evolves.

CFU-Depth and CFU-Adjacent were not significantly associated with motion outcomes. CFU-Depth was not included in our final model; however, in settings where skin grafting is not as readily available, depth of injury should be considered and may have a stronger association with motion outcomes due to increased scarring in burns allowed to heal secondarily.^28,44 The lack of association found with CFU-Adjacent (ie, the extent of burn of any depth in the neighboring CFU) may be a reflection of the relatively low TBSA in this sample, resulting in fewer adjacent CFUs having burns or less burn extent within them. This finding is consistent with another study that also had a sample with low median TBSA burns.³⁰ In studies evaluating larger burns, CFU-Adjacent may exert more of an effect on motion which is an important area for further investigation.

A unique feature of this dataset was that it included conditions that contributed to range of motion deficits when discharge motion was measured. Although the long-term impact of factors that limit the range of motion at discharge has not been extensively studied, we conducted an exploratory subanalysis excluding temporary motion limiters (ie, edema and open wounds). This subanalysis confirmed the robustness of our findings in that CFU-Burned and CFU-Grafted remained significantly associated with motion loss.

Patient-level characteristics and motion outcomes

Age emerged as having a significant negative association with range of motion. The proportion of available joint motion decreases with age and as the extent of burn and grafting in the designated CFU increases (Figure 2). Holding age constant, range of motion decreases with increasing burn and grafting to the CFU. Likewise, holding CFU-Burned and CFU-Grafted constant, increasing age is associated with less range of motion. This agrees with one large-scale cohort study¹³ but conflicts with other studies that report lower BSC prevalence in older adults.^28,45,46 It is known that older patients who sustain burn injury have unique medical and surgical needs,⁴⁷ and our findings suggest they may also need special rehabilitation attention to prevent motion loss after injury.

Interestingly, TBSA-Burned and TBSA-Grafted did not correlate with range of motion outcomes at hospital discharge. This finding contrasts with previous studies that have shown these patient-level burn characteristics are associated with the development,^29,48 severity,^13,49 and number of BSCs.^13,50 However, the unit of analysis for these studies was the patient, while our unit of analysis was joint motion which may explain the discrepancy. The CFU framework offers a localized and more granular assessment of burn characteristics within specific body areas. This approach provides joint-specific insights compared to TBSA which is global measures of the extent of burn and skin grafting.⁵¹ This more focused approach is particularly useful when evaluating motion since some joints are more prone to BSC than others,^{23,29,30,48,52} and have been linked to specific functional problems.^6,53,54 Therefore, evaluating risk factors using CFUs may offer a more relevant and precise way of predicting BSC and subsequent functional problems.

Limitations

This study does have limitations; most notably, the range of motion data was only available at hospital discharge, which limits conclusions about long-term outcomes. Previous studies suggest that BSC rates improve post-discharge, and the optimal time to assess BSC as a more permanent condition still needs to be established.^1,30,55 The data analyzed were collected over a decade ago, which may not reflect current outcomes. However, burn rehabilitation practices have changed minimally during this period, and the dataset remains the most extensive one available for CFUs. Another possible limitation is that joint motion was analyzed as a fixed effect across all motions to maximize statistical power, but this approach may overlook differences between specific areas of the body regarding the risk of motion loss.^29,48,52,55 Additionally, burn centers were included as a fixed effect, limiting generalizability beyond the 13 participating burn centers. These burn centers, though diverse in care models and geographically distributed across one high-income country, were not racially diverse, nor do they represent children. Since burns are a global issue with disparities in care affecting all age groups,^56,57 similar studies should include more diverse populations and be conducted in resource-limited settings, as risk factors for motion loss may differ for different samples of patients.

Despite these limitations, a major strength of the study is the extensive CFU data available and the rich outcome and treatment data in the dataset used for analysis. Exploring motion problems within a CFU framework allowed for a more focused evaluation of the impact of burn injury on joint motion in specific body areas. This study demonstrated the value of using CFUs, and future work should build upon these findings.

Implications for future research

This study established a significant relationship between CFU-related burn features and motion, emphasizing the value of a localized assessment. Based on exploratory model building, we proposed a prediction model for identifying the risks of BSC. This model needs further validation and refinement in this sample and in broader, more diverse samples of burn survivors to ensure its robustness and generalizability.

CFUs offer a standardized and functionally relevant method for evaluating burn injury location, which can be used to evaluate other patient outcomes, guide clinical decision-making, and advance the quantification of burn injuries in a more personalized and meaningful manner. Motion limitations are not always indicative of functional problems,^53,54 and future research should extend beyond motion outcomes and explore links between CFU burn characteristics and broader outcomes such as physical function and quality of life.

Our study adopted a global approach, examining all joints together to investigate the impact on motion when the specified CFU area is burned or grafted. An important next step is evaluating specific joint motions to identify injury patterns and individual body areas at high risk for BSC. This type of research could guide clinical decision-making and inform resource utilization, particularly in cases involving extensive burns or in resource-constrained settings. Additionally, CFU data could be instrumental in guiding the development of tailored surgical and nonsurgical interventions to prevent or correct BSCs, which play a critical role in long-term patient outcomes.^12,56–58 Investigating how modifiable factors, such as variations in treatments and services, influence the relationship between CFU burn features and outcomes could lead to more efficient and effective care strategies.

Lastly, to maximize the clinical utility of CFUs, it is essential to simplify their integration into practice. The inherent complexity of CFUs and their numeric coding system highlight the need for advanced tools and burn models that can capture, quantify, and analyze CFU data efficiently. Automated data capture systems could facilitate this process, enabling real-time, personalized care planning focusing on specific joint motions at risk. The current CFU framework¹⁶ and its detailed description used in this study can serve as a foundation resource for future research.⁴³ Researchers and developers are encouraged to use this framework as a collective starting point for refining CFUs, improving their usability, and advancing the science of burn care. By leveraging the CFU framework, future studies have the potential to transform burn injury assessment and treatment, paving the way for more individualized, meaningful, and impactful patient care.

CONCLUSION

This study offers valuable new insights into the relationship between burn features within a CFU framework and joint-level range of motion outcomes at hospital discharge. Using this framework to quantify burn characteristics, we performed exploratory statistical modeling to evaluate predictors that had not been previously investigated. This addressed a critical gap in the literature linking burn features within specific functional skin areas to range of motion outcomes. The findings reveal that the extent of burn injury and skin grafting within a CFU, as well as the burn extent in the proximal location of the CFU, are critical factors influencing motion outcomes. Our results highlight the importance of a localized and standardized approach to evaluating burn injuries and reinforce the utility of the CFU framework as a functionally relevant and clinically valuable tool for understanding motion loss.

Supplementary material

Supplementary material is available at Journal of Burn Care & Research online.

Funding

None declared.

Conflict of interest statement

None declared.