Design
A before-after single-arm pilot study involving a cohort of traffic police officers was initiated in June 2018. The study aimed to evaluate the impact of an educational program on knowledge, perceived skills confidence, and utilization of trained skills in postcrash first aid (PFA). The study protocol was registered retrospectively at the Registry for International Development Impact Evaluation with ID number: RIDIE-STUDY-ID-5bb71e0ed1e89.
Setting
The present study was conducted in Dar es Salaam Region, Tanzania. The region, which has an area of 1,590 km2, is the location of the city of Dar es Salaam which is a major commercial seaport and Tanzania’s largest city, with an estimated population of more than 5.7 million (14). Dar es Salaam Region was selected as the setting for the present study because, according to a 2016 report by the Tanzania Police Force and the National Bureau of Statistics, it has the nation’s highest number of road traffic incidents, accounting for more than a third of all such incidents per year (15). Almost all major road intersections and road crash hotspots in Dar es Salaam are lined with traffic police posts for observation of road safety. Currently there are more than 70 traffic police posts with a distance from one post to another ranging between one to one and half kilometre apart along the road. Presence of traffic police posts and communication system may facilitate availability of traffic police in the event of a crash.
Participants
Traffic police officers were recruited to participate in the present study for three main reasons: (1) First aid provision is part of their job description; (2) they are authority figures and they command the crash scene; and (3) they are readily available at the crash scene. Details on participant background characteristics such as age, sex, educational background, and work experiences are provided in Table 1.
Sample Size
Using the online Statistics for Psychologists program (AICBT Ltd., https://www.ai-therapy.com/psychology-statistics/terms/), we estimated that sample size should be 128 participants to detect an effect size (Cohen’s d) of 0.25 for trauma first aid knowledge as primary outcome, at an alpha error rate of .05 (two-tailed) and a beta error rate of .20. An additional 10% was added to account for loss to follow-up; thus, the final sample size was 141 participants. An effect size of 0.25 was chosen on the basis of empirical distributions of effect sizes from comparable educational intervention studies that considered it of practical significance (16).
PFA Program
The PFA educational program focused on imparting basic knowledge and skills to traffic police on managing injury victims at the scene and on the way to the hospital. The course program was developed in accordance with WHO guidelines on essential knowledge, skills, equipment, and supplies for the provision of basic first aid (8). Three experienced emergency and trauma care teachers facilitated the course. Topics covered during the training included scene survey, provider safety, and initial assessment and care of the injury victim. Initially, lectures and discussions on topics were conducted in groups of about 27 participants per session. Afterward, participants were divided into groups of about 8 to 10 for practical training. Role-play and mannequins were used to impart practical skills. The course content was covered in two days (a total of 16 hours of session time). At the end of the training, the traffic police were given leaflets on basic steps in managing injury victims for use as a reference.
Data Collection and Procedures
Permission for conducting PFA training and data collection for evaluating the training was granted by the office of Inspector General of Police. The recruitment process was left under the mandate of researchers who ensured rights of officers to participate or decline participation were maintained without interference from police management.
A list of 340 police officers from a previous database regarding knowledge, self-reported practice, and attitudes of traffic police officers was used to proportionately select officers from jurisdictional areas of Kinondoni, Ilala and Temeke (13). A simple random sampling by drawing a YES/NO tickets from the box was used. 141 police officers who drew YES tickets were invited to participate in the study, whereof 135 officers attended the training and completed the before and after questionnaires. The reason for those not attending and completing the training was mainly due to sickness, attending emergencies and being assigned other tasks. Before training sessions, selected officers were informed about the purpose and duration of the educational program and then were asked for their consent to participate.
Data were collected before and immediately after the educational program by questionnaire. Six months after the training, the police officers were contacted by telephone to identify their work locations; the researchers then physically visited the identified locations and asked the officers to complete a follow-up questionnaire. An updated version of a self-administered questionnaire from a previous study was used in the present study for data collection (13).
The process through which officers were recruited and follow-up was executed is summarized in Figure 1.
Questionnaire
The questionnaire was developed on the basis of WHO guidelines on essential knowledge and skills for provision of basic first aid (8). A Swahili version already existed and was tested for reliability; the alpha value of .74 for the knowledge domain indicated acceptable reliability of the scale (13). We modified the existing Swahili version of the questionnaire by adding more knowledge items and developed new questions on assessing perceived skills confidence, utilization of trained skills, and training experience. Two experts in emergency and trauma training in Tanzania assessed the entire questionnaire for content validity. Face validity was ensured by piloting the questionnaire to 15 traffic police officers who did not take part in the main study. Minor structural changes such as sequencing questions according to systematic trauma assessment were made. The questionnaire consisted of 34 items; baseline characteristics of participants such as sex, age, educational level, work experience, previous first aid training, and number of injured persons cared for were included. Cronbach's alpha coefficients for reliability of the test in all domains (knowledge, perceived skills confidence, and training experience) in the questionnaire ranged from .70 to .94 when measured before, immediately after, and six months after training. The values indicate that the items within each domain were consistent in measuring the same attributes (17). Details of respective outcome measures in the questionnaire are described below.
Outcome Measures
Three primary outcomes were measured: knowledge of PFA, perceived PFA skills confidence, and PFA skills utilization.
Knowledge of PFA. The same eight multiple choice questions assessing participants' knowledge of PFA were asked before, immediately after, and six months after PFA training. Questions involved condition given priority to care in case of multiple injuries, assessment of consciousness, airway management, breathing assessment, control of external bleeding, care of fracture of extremities, positioning of victim, and head and neck immobilization. For each knowledge question, a correct response was scored one and an incorrect response was scored zero. The item scores were summed up and divided by the number of items to get an average score, which was then converted into a percentage score.
Perceived PFA skills confidence. The same six questions assessing participants' perceived PFA skills confidence were asked before, immediately after, and six months after PFA training. The questions were about management of compromised airway, breathing assessment, control of external bleeding, care of fractures of extremities, positioning victim, and head and neck immobilization. Questions were answered on a 5-point (1 = very low, 2 = low, 3 = moderate, 4 = high, and 5 = very high). Item scores were summed up and divided by the number of items to get an average score.
Utilization of trained PFA skills. Six questions on managing compromised airway, breathing assessment, control of external bleeding, care of fracture of extremities, positioning victim, and head and neck immobilization were answered on a 4-point Likert scale six months after the intervention (1 = never, 2 = sometimes, 3 = often, 4 = always). Percentages were calculated to determine utilization of skills based on Likert scale categories.
One secondary outcome, PFA training experience, was measured. Five questions on overall PFA training experience were answered on a 4-point Likert scale (1 = strongly disagree, 2 = disagree, 3 = agree, 4 = strongly agree).
Respective Likert scales are common in measuring participants’ responses in different contexts. They were chosen based on literature review (18). For details on overall questions, see Additional file 1: PFA questionnaire.
Data Analysis
To perform data analysis, we used the Statistical Package for Social Sciences (SPSS), version 24. Demographic characteristics were reported as means, standard deviations, counts, and percentages.
Paired t tests were used to compare differences in participants’ PFA knowledge scores before training (T0), immediately after (T1), and 6 months following completion of training (T2). A repeated-measures analysis of variance (RM-ANOVA) was used to compare differences in scores within subjects using the measurement points (T0, T1 and T2). A Bonferroni adjustment was made for multiple comparisons to control for familywise error rates; with this correction, we set p < .017 as our significance level (19). We used McNemar’s test to compare individual PFA knowledge items scores (correct and incorrect responses) and the measurement points (T0, T1 and T2) for association.
Regarding perceived PFA skills confidence, since data from Likert scales is ordinal and tends to be non-normally distributed, a nonparametric test (Wilcoxon sign rank test) should ideally be conducted and reported. Considering that in the present study the nonparametric test and the parametric test (paired t test) provided almost the same conclusion about the change in perceived skills confidence, we report the means, which present the results better than the medians (20). Spearman rank-correlation analysis was used for the assessment of the interrelationships among outcome measures and baseline characteristics such as age, sex, educational level, and work experience.