Development and evaluation of deep learning algorithms for assessment of acute burns and the need for surgery.

Assessment of burn extent and depth are critical and require very specialized diagnosis. Automated image-based algorithms could assist in performing wound detection and classification. We aimed to develop two deep-learning algorithms that respectively identify burns, and classify whether they require surgery. An additional aim assessed the performances in different Fitzpatrick skin types. Annotated burn (n = 1105) and background (n = 536) images were collected. Using a commercially available platform for deep learning algorithms, two models were trained and validated on 70% of the images and tested on the remaining 30%. Accuracy was measured for each image using the percentage of wound area correctly identified and F1 scores for the wound identifier; and area under the receiver operating characteristic (AUC) curve, sensitivity, and specificity for the wound classifier. The wound identifier algorithm detected an average of 87.2% of the wound areas accurately in the test set. For the wound classifier algorithm, the AUC was 0.885. The wound identifier algorithm was more accurate in patients with darker skin types; the wound classifier was more accurate in patients with lighter skin types. To conclude, image-based algorithms can support the assessment of acute burns with relatively good accuracy although larger and different datasets are needed.

Sociodemographic Patterns of Pediatric Patients in Specialized Burn Care in Sweden.

Trauma is a leading cause of mortality in children. Burns affect children disproportionally. Although burn incidence and mortality are decreasing, differences in the risk depend on socioeconomic status. The present study aimed to investigate the sociodemographic patterns of pediatric patients (0-17 years) managed at the two burn centers in Sweden, Uppsala, and Linköping, between 2010 and 2020. Method: This retrospective register-based study used hospital records from the two burn centers combined with information from Statistics Sweden plus data regarding number of asylum seekers from the Swedish Migrations Agency. Choropleth maps representing the patients' geographical distribution were created. Information about income levels per geographic area was added. A Wilcoxon signed-rank test was performed to investigate differences in median income levels between the areas where the patients lived, related to Sweden's median income. Results: The study included 2455 patients. Most of the children aged below 5 years (76%) and were boys (60%). The mean percentage of total skin area was 4.2%. There was no significant increment or decrease in the incidence of pediatric burns during the study. Most patients with recorded zip codes lived in areas with an income level below the national median (n = 1974, 83%). Children with asylum status were over-represented compared with residents and/or Swedish citizens. Conclusions: In Sweden, most pediatric burns occur in families that live in areas with low-income levels. Pediatric burns affect children with asylum status disproportionally compared with those who are residents in and/or citizens of Sweden. Prevention strategies should be designed and implemented to alleviate this health inequity.

Does the estimation of burn extent at admission differ from the assessment at discharge?

INTRODUCTION: Estimation of total body surface area (TBSA) burnt and burn depth are among the most central parts of acute burn assessment/treatment as they determine the level and type of care needed. Traditional methods for determining burn extent on admission often lead to inaccurate estimations, especially in paediatric or overweight patients. AIM: To compare %TBSA at admission with validated %TBSA at discharge in different patient populations to investigate if significant over- or underestimation occurs. METHOD: This retrospective observational study is based on a patient registry of all the patients (n = 863) treated at the Uppsala University Hospital's Burn Centre between 2010 and 2018. The patients were divided into subgroups based on age, gender, body mass index (BMI) and validated burn extent. The %TBSA estimated at admission was compared to the validated %TBSA in all groups separately. RESULTS: As has been published before, we also found that the %TBSA in paediatric patients was more often overestimated as were the smaller injuries, whereas larger injuries were often underestimated. BMI did not clearly affect the estimations and there was no clear difference between the genders in estimated %TBSA. CONCLUSION: Inaccurate estimations of %TBSA are common, particularly for paediatric patients and small or large injuries. We recommend a careful accurate approach when calculating %TBSA in the paediatric population to avoid over- and under-resuscitation. Increased education and training are recommended to improve accurate estimation in the future. LAY SUMMARY: The correct estimation of both extent and depth of burn is very important. This assessment guides the lever of care needed, the necessary amount of fluid resuscitation, the predicted outcome and more. It has been proven notably difficult to correct assess, especially the extent of a burn. Despite different tools as the "Rule of Nine" (body area divided into multiples of 9% body surfaces), the "Rule of Palm" (Patient's palm, fingers included, approximates 1% of body surfaces), the Lund & Browder chart (detailed, age-specific body areas) and different more technical solutions. Often inaccurate estimations are done which thus affect the treatment. This depth and extent estimation is usually performed when the patient is admitted. However, it is known that burns change appearance during the first few days of care. In our Burn Center we have also performed this estimation when the patient is discharged. At this point it is known the true extent and depth of the initial burn. In this retrospective observational study, we compared the burn extent estimated on admission with the one on discharge to investigate whether the initial assessment is accurate. This study highlights the issue of frequent inaccurate burn extent estimations, especially in subgroups as overweight patient or pediatric patients.

Want Correct Percentage of TBSA Burned? Let a Layman Do the Assessment.

Accurate determination of burn size significantly impacts both immediate management and long-term outcome. In the era of evidence-based medicine, the variability in TBSA% assessment shown by traditional methods may prove unacceptable and technology-aided systems become the "accepted standard." The objective of this study was to push this scenario to the limit by investigating the accuracy and consistency of TBSA% estimations using a computer-aided tool. Five Laymen (health care-burn management naïve people) were trained on the handling of the technology-aided assessment tool Burn Case 3D© and asked to calculate TBSA% for 18 clinical pictures of burns with different patterns and sizes. Forty-four burn Professionals (senior burn surgeons, plastic surgery residents, anesthesiologists, emergency physicians, senior registered nurses) were provided the same pictures and assessed TBSA% using traditional paper-based tools ("Rule of Palm"; "[Wallace] Rule of Nines"; "Lund and Browder chart). The Laymen's computer-aided calculations did not differ significantly (P > .05) from the senior burn surgeons' estimations in 17 of the 18 cases. However, when comparing the Laymen's TBSA% calculations with the whole group Professionals there were significant differences (P < .05) in (again) 17 of the 18 cases. Laymen's calculations were also more consistent (mean SD, 0.95%). The Professionals showed a generalized significant overestimation of TBSA% as compared with the Laymen's calculations (up to 198.5%). Innovative software provide a high potential to improve objectivity and quality of burn assessment in the future.

Assessment of incidence of severe sepsis in Sweden using different ways of abstracting International Classification of Diseases codes: difficulties with methods and interpretation of results.

OBJECTIVE: To compare three International Classification of Diseases code abstraction strategies that have previously been reported to mirror severe sepsis by examining retrospective Swedish national data from 1987 to 2005 inclusive. DESIGN: Retrospective cohort study. SETTING: Swedish hospital discharge database. PATIENTS: All hospital admissions during the period 1987 to 2005 were extracted and these patients were screened for severe sepsis using the three International Classification of Diseases code abstraction strategies, which were adapted for the Swedish version of the International Classification of Diseases. Two code abstraction strategies included both International Classification of Diseases, Ninth Revision and International Classification of Diseases, Tenth Revision codes, whereas one included International Classification of Diseases, Tenth Revision codes alone. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: The three International Classification of Diseases code abstraction strategies identified 37,990, 27,655, and 12,512 patients, respectively, with severe sepsis. The incidence increased over the years, reaching 0.35 per 1000, 0.43 per 1000, and 0.13 per 1000 inhabitants, respectively. During the International Classification of Diseases, Ninth Revision period, we found 17,096 unique patients and of these, only 2789 patients (16%) met two of the code abstraction strategy lists and 14,307 (84%) met one list. The International Classification of Diseases, Tenth Revision period included 46,979 unique patients, of whom 8% met the criteria of all three International Classification of Diseases code abstraction strategies, 7% met two, and 84% met one only. CONCLUSIONS: The three different International Classification of Diseases code abstraction strategies generated three almost separate cohorts of patients with severe sepsis. Thus, the International Classification of Diseases code abstraction strategies for recording severe sepsis in use today provides an unsatisfactory way of estimating the true incidence of severe sepsis. Further studies relating International Classification of Diseases code abstraction strategies to the American College of Chest Physicians/Society of Critical Care Medicine scores are needed.