This was a two center observational longitudinal study and patients were followed from admission to completion of surgical intervention observing for findings on eFAST, X-ray and surgical intervention.
The study was conducted at the accident and emergency (A&E) and the radiology departments of Kampala International University teaching hospital (KIU-TH) and Mbarara regional referral hospital (MRRH) in south-western Uganda.
All patients with traumatic chest injuries who attended the A&E and radiology departments of KIU-TH and MRRH during the one year period from 1st May 2021 to 30th April 2022 were considered for the study.
Sample size estimation
Daniel’s formula for determining the sample size was used . According to the study on epidemiology of motorcycle injuries presenting to Uganda’s national referral hospital, Mulago; traumatic chest injuries contributed to 34.7% of all trauma injuries . Using the formula N = Z2P (1 − P)/d2; where N = the sample size, Z = Score corresponding to 95% of confidence interval which is 1.96, P = Proportion of chest injured patients which is 34.7%; (1 − P) = 65.3%, and E = margin error rate set at 5%; the sample size N = 348. To increase the internal validity of study and catering for non-responders, the calculated sample was increased by 10% giving an estimated sample size of 383.
Adjusting sample size to the finite population, Sample size N = ns = (1 + (ns − 1)/n). Where N = adjusted population size, ns = estimated sample size, n = population under study = 142 (based on the hospital data registry), N = 104. Therefore, a sample size of 104 participants with chest injuries was considered for study duration of 12 months.
Consecutive recruitment method was used to enroll all eligible participants until the required sample size was realized. 69(66.3% of the sample size) was recruited from MRRH and the remaining 35(33.7%) from KIU-TH using proportionate sampling.
All patients with chest injuries who consented were recruited in the study.
Unstable patients that had thoracostomy before chest X-ray, those with massive hemothorax, cardiac tamponade and patients with documented evidence of pleural effusion prior to the trauma event were excluded from the study.
Training of research assistants
Five surgery residents including the principal investigator participated in the study and were trained in the use of point of care ultrasound. The team was availed with the “point of care ultrasound in resource limited environments” (PURE) model manual , which consists of core concepts of emergency ultrasound such as focused assessment for trauma, with aim to establish competence in knowledge related to the indication of the scan, image acquisition, interpretation and integration of findings into patient management. Later the team attended a four weeks’ intensive practicum on use of eFAST in chest and abdominal trauma evaluation and using data collection tools.
The PURE model involves use of the electronic “point of care” ultrasonography (POCUS) manual , didactic lectures embedded with videos, followed by practical sessions and knowledge retention assessment test . This training module has been validated in Kenya in similar settings  and is accredited by the African Federation for Emergency Medicine . The principal trainers were qualified radiologists from Uganda who were experienced in the use of FAST and eFAST. The training was facilitated by the investigator and trainees. Because of the concerns on learning curve, the investigator and research assistants continued to work under supervision of qualified sonographers and radiologists throughout the study period.
Data collecting tools
Data for this study was collected using investigator administered checklist. The key variables of interest included demographics, injury mechanisms and patterns, presence or absence of haemothorax, haemo-pneumothorax, nature of surgical intervention and findings on ultrasound, CXR and Tube thoracotomy. The two investigative techniques eFAST and CXR were compared with the findings on tube thoracostomy. The findings at tube thoracostomy were used as the surrogate gold standard to confirm if the investigations correctly detected the haemothorax.
Data collection procedure
After attending to and excluding the life threatening airway emergencies in the primary survey, the researchers explained the study and its purpose to the participants in order to obtain an informed consent document with a signature or thumb print. However in the event of suspected massive haemothorax or tension pneumothorax, eFAST was performed as part of primary survey and intervention made immediately before the administration of the questionnaire. A pretested check list of parameters of interest was used by the investigator with his data collection team at the radiology and accident and emergency departments. ATLS principles were used in initial assessment and management with eFAST as an adjunct in primary survey. A complete history, physical examination and imaging assessment of the chest, followed by chest X-ray which is the surrogate diagnostic standard in our setting was done. The findings for both investigations were recorded on the data tool. Two portable hand held ultrasound systems (Mindray DP-6600 FL, USA) were used in this study, one for each site. The device’s manufacturer has indicated eFAST as one of its uses  and has been validated in previous studies  in addition to being suitable in rural areas where there could be electricity blackouts .
This procedure was performed in accordance with Taylor and O’Rourke . The patient was asked to remove clothing and other objects such as jewelry that could interfere with the scan. Patient was positioned on examination bed either lying on back, or side or sitting up with arms raised with hands clasped around the neck depending on level of consciousness. Ultrasound gel was placed on area of chest to undergo examination. Using a transducer, ultrasound waves were sent from area being examined reflected off structures and analyzed by the ultrasound machine that created an image on the screen. The images generated were stored digitally. Patients were at times asked to cough or shift position or sniff for clarity of chest structures.
Chest X-ray procedure
This was carried out in a radiology certified room with fixed X-ray machines. The patient was asked to undress, remove jewelry, stand (PA view) OR lie (lateral decubitus view) next to a cassette that recorded images for processing. For severely injured patients and suspected spinal injury patients, the X-ray tube and the image receptor were positioned, rather than the patient or the part to avoid the risk worsening the patient’s condition. Patient was instructed to roll shoulders forward, withhold breath, and stay still while image was being taken. The image was recorded on computer and printed on film for interpretation.
Chest tube insertion procedure
This was done in the accidents and emergency department using the aseptic technique in the triangle of safety under local anesthesia according to the method described by Datta et al. . All patients who had a hemothorax volume of greater than 300 mls had chest tube insertion since this volume is associated with a retained hemothorax if not drained . Patients who had respiratory distress also underwent drainage irrespective of the volume quantified at sonography. Patients who had a volume less than 300 mls at sonography, but later deteriorated also underwent chest tube insertion.
The questionnaire was pretested at Ishaka Adventist Hospital to check whether it could extract the desired information on variables of interest and necessary changes were made. The investigator and trained research assistants (residents) collected the data. For every 5th patient, eFAST was assessed by a qualified radiologist. Where two radiologists did not agree on the radiological findings, the decision of an independent third radiologist was to be considered final. Data was checked for completeness at the end of definitive surgical intervention. The data was analyzed with the guidance of a biostatistician.
Data analysis and presentation
Data was analyzed using SPSS version 22.0. Univariate analysis for continuous variables was summarized using mean and standard deviation, whereas proportions and percentages were computed for categorical variables and presented as frequency tables. The detection rates of haemothorax were computed individually for both CXR and eFAST with reference to the findings on chest tube drainage. The sensitivity, specificity, positive predictive value, negative predictive value and accuracy were calculated using the cross tabulation procedure and the corresponding chi-squire P values determined taking thoracostomy findings as the gold standard. P value of less or equal to 0.05 was considered significant for correlation between the detection rates of hemothorax by the investigation assessed and tube thoracostomy findings. The receiver operator characteristic curve (ROC) with the corresponding area under the curve (AUC) were used to compare the efficacy of the two investigations taking thoracostomy as the gold standard.