An analysis of emergency care delays experienced by traumatic brain injury patients presenting to a regional referral hospital in a low-income country.

BACKGROUND: Trauma is a leading cause of death and disability worldwide. In low- and middle-income countries (LMICs), trauma patients have a higher risk of experiencing delays to care due to limited hospital resources and difficulties in reaching a health facility. Reducing delays to care is an effective method for improving trauma outcomes. However, few studies have investigated the variety of care delays experienced by trauma patients in LMICs. The objective of this study was to describe the prevalence of pre- and in-hospital delays to care, and their association with poor outcomes among trauma patients in a low-income setting.
METHODS: We used a prospective traumatic brain injury (TBI) registry from Kilimanjaro Christian Medical Center in Moshi, Tanzania to model nine unique delays to care. Multiple regression was used to identify delays significantly associated with poor in-hospital outcomes.
RESULTS: Our analysis included 3209 TBI patients. The most common delay from injury occurrence to hospital arrival was 1.1 to 4.0 hours (31.9%). Most patients were evaluated by a physician within 15.0 minutes of arrival (69.2%). Nearly all severely injured patients needed and did not receive a brain computed tomography scan (95.0%). A majority of severely injured patients needed and did not receive oxygen (80.8%). Predictors of a poor outcome included delays to lab tests, fluids, oxygen, and non-TBI surgery.
CONCLUSIONS: Time to care data is informative, easy to collect, and available in any setting. Our time to care data revealed significant constraints to non-personnel related hospital resources. Severely injured patients with the greatest need for care lacked access to medical imaging, oxygen, and surgery. Insights from our study and future studies will help optimize resource allocation in low-income hospitals thereby reducing delays to care and improving trauma outcomes in LMICs.

Introduction

An estimated 69 million people worldwide experience a traumatic brain injury (TBI) each year [1]. The distribution of this burden is biased towards low- and middle-income countries (LMICs) which endure 90% of global injury related deaths and three times more TBI cases than high-income countries [1-3]. Among LMICs the greatest burden of TBI exists in the World Health Organization (WHO) Africa Region, which accounts for 15.9% and 11.5% of LMIC and global TBIs respectively [1]. Furthermore, predictive models have estimated a future annual TBI incidence of up to 14 million cases in Africa by the year 2050 [4].

The notable presence of TBI in African countries is likely attributable to excessive road traffic incidents precipitated by increased motor vehicle use in combination with poor compliance to road safety practices and lagging infrastructural development [5-7]. This is particularly true in sub-Saharan Africa where the cumulative incidence of TBIs due to road traffic incidents is 156 per 100,000 people; significantly higher than the global value of 106 per 100,000 people [2]. However, in addition to enduring a high TBI incidence, patients who sustain a TBI in sub-Saharan Africa have disproportionately worse outcomes in comparison to patients in high income countries [8-12]. The cause of this disparity may result largely from differences in the quality of prehospital and in-hospital care [8, 13, 14]

One widely recognized measure of hospital care quality, particularly for emergency departments (EDs), is time to care [15-18]. In high-income countries (HICs) delays to care within hospital EDs are frequent and have been shown to result in worse outcomes, including higher mortality rates [19-27]. However, information from LMICs regarding the extent of ED delays and their impact on patient outcomes is sparse [28-34]. Moreover, such data is lacking for TBI patients presenting to EDs in sub-Saharan Africa. There is a need to investigate the relationship between ED care delays and in-hospital outcomes among TBI patients in sub-Saharan Africa if we are to improve trauma outcomes in this region of the world.

Tanzania is a low-income country in East sub-Saharan Africa with a large TBI burden. The TBI prevalence ranges from 21% in urban areas to 34% in rural areas and mortality rates have reached upwards of 30% for patients presenting to large referral hospitals including Muhimbili National Hospital in Dar es Salaam and Bugando Medical Centre in Mwanza [35-37]. Although Tanzania has recently started an emergency medicine residency program and a trauma care quality improvement system, the country does not yet have national guidelines for emergency and critical care [38]. Consequently, many hospitals lack staff trained in the triage and management of critically injured patients [38]. Trauma patients in low-income settings like Tanzania may therefore experience longer delays to care and thus worse outcomes when compared to trauma patients in HICs. The objective of this study was to describe the prevalence of pre- and in-hospital care delays and their association with poor outcomes among trauma patients presenting to an emergency department (ED) in a low-income hospital.

Methods

Study design

This study is a secondary analysis of a TBI patient registry established at Kilimanjaro Christian Medical Center (KCMC) in Moshi, Tanzania. The registry prospectively enrolled patients presenting to the KCMC ED for treatment of their acute TBI (<24 hours since injury occurrence) from May 2013 to August 2017 [12]. The registry includes information on demographics, injury status, vital signs, treatment, time of treatment, injury management, and injury outcomes.

Study setting

KCMC is a tertiary referral hospital in Moshi, Tanzania serving over 15 million people [39]. Nearly 1000 TBI patients present to the ED each year with about 33% requiring admission to the intensive care unit (ICU) [12]. The mortality rate of patients presenting to KCMC with severe TBI is close to 50% [12]. The hospital contains 630 patient beds and employs 1300 staff. However, the ED houses only six adult patient beds and three physicians yet receives an average of 80 to 100 patients a day [39]. TBI surgeries are limited to burr holes, craniotomies, and craniectomies conducted by trained general surgeons, as there are no neurosurgeons stationed at KCMC. The decision to operate on a TBI patient and what kind of operation to perform is ultimately based on the judgement of a general surgeon trained in neurosurgical procedures. Alternatively, a patient or patient’s family may choose to forgo surgery if the cost cannot be covered.

Study participants

Inclusion into the TBI registry was restricted to adult patients (≥18 years) seeking care for acute TBI who survived to evaluation by an ED physician. Patients who presented with follow-up and non-acute injuries or those who died prior to physician evaluation were not eligible for inclusion. All patients who were enrolled in the registry were included in our analysis.

Study variables

Variables selected for inclusion into this analysis were age, gender, mechanism of injury, alcohol use, respiratory rate, systolic blood pressure, oxygen saturation, heart rate, Glasgow coma score (GCS), transfer status, and time to care. Transfer status was categorized as transferred or not transferred to KCMC from a previous hospital. Patients with missing prehospital course information were categorized as having an unknown transfer status. Time to care variables included time from (1) injury to hospital arrival, (2) hospital arrival to physician arrival, (3) physician arrival to lab tests sent to the laboratory, (4) physician arrival to chest or skull x ray, (5) physician arrival to brain computed tomography (CT) scan, (6) physician arrival to fluids administered, (7) physician arrival to oxygen administered, (8) physician arrival to TBI surgery, and (9) physician arrival to non-TBI surgery. The chronology of these time intervals is depicted in Fig 1. The outcome variable used in this analysis was Glasgow outcome score (GOS) dichotomized as good (4 or 5) and poor (1, 2, or 3). Among patients who survived, GOS was measured at hospital discharge. Hospital discharge was our last point of contact with patients included in our registry. We report the mean time from injury to hospital discharge as well as the mean time from injury to death.

Fig 1

Chronological sequence of time intervals.

Each arrow represents one of our nine time to care variables.

Time standards

The South African Triage Scale (SATS) is a protocol that was designed to improve patient triage in resource limited settings around South Africa [40]. Since its development, the SATS has been validated and proven reliable in both urban and rural South African hospitals as well as in numerous LMICs [41-47]. Based on time to care standards outlined in the SATS, we considered the following time intervals to categorize our nine time to care variables: 0.0–1.0 hours, 1.1–4.0 hours, 4.1–12.0 hours, and greater than 12.0 hours.

Missing time data

Systolic blood pressure and oxygen saturation were used to identify patients with hypotension or hypoxia on arrival. Binary variables were created for hypotension (<100 mmHg systolic blood pressure) and hypoxia (<92% pulse oxygen). Using these binary variables and GCS, we created two additional levels for three of our time to care variables: (1) patient did not receive the procedure and needed the procedure, and (2) patient did not receive the procedure and did not need the procedure. These two categories were used to classify patients when data was missing for time to brain CT scan, time to fluids, and time to oxygen.

We defined needing a brain CT scan as presenting with a GCS less than 13. Needing fluids was defined as being hypotensive. Needing oxygen was defined as having hypoxia or a GCS less than 8. For patients who received oxygen, the availability of time data was insufficient for analysis. We therefore categorized all oxygen recipients as having received oxygen. For time to TBI surgery and time to non-TBI surgery, we had no means by which to identify a patient’s need for surgery. As a result, if a patient had missing data for these two variables and if they did not receive surgery, then he or she was classified as having not received surgery.

Data analysis

Imputation

Variables with more than 20% of observations missing were excluded from our final analysis. Two variables were removed during this process: (1) time from physician arrival to mannitol administration, and (2) time from physician arrival to surgeon arrival. For the remaining variables, we used multiple imputation by chained equations to impute missing values. This was achieved through the mice package in R Language for Statistical Computing. All data analysis was performed using R software.

Descriptive statistics

We computed descriptive statistics for all variables stratified by our GOS outcome metric. A majority of patients in our registry who experienced a poor GOS died. Consequently, we do not differentiate our results by mortality in addition to differentiating by good and poor GOS. For continuous variables we report the mean and standard deviation. For categorical variables we report frequencies and percentages. For continuous variables, t-tests were used to identify significant differences (α<0.05) in means between the good and poor outcome groups. For categorical variables, chi-square or Fisher’s exact tests were used to identify significant differences (α<0.05) in the distribution of observations between the good and poor outcome groups.

Inferential statistics

We used logistic regression to model GOS as a function of time to care. We subsequently adjusted our models for age, gender, mechanism of injury, respiratory rate, systolic blood pressure, oxygen saturation, heart rate, and GCS. For all models we report odds ratios and 95% confidence intervals (CIs). For all models, a good outcome (GOS 4–5) was used as the reference level for our outcome variable. For time to care variables, the lowest time interval was used as the reference level. In the case of time to oxygen, there was insufficient time data to form useful time interval levels. As a result, having received oxygen was used as the reference level.

Ethics

This study was approved by the KCMC Ethics Committee and the National Institute of Medical Research, Tanzania as well as by the Duke University Institutional Review Board. Participant consent was waived as this study is a secondary analysis of a de-identified patient registry.

Results

Sociodemographic, injury, and clinical characteristics

Our sample included 3209 patients. Patients were predominantly male (82.2%) and had a mean age of 32.1 years (±16.5). Most patients suffered TBI due to a road traffic incident (67.7%). With regard to GCS, most patients were classified as mild (78.4%) followed by severe (12.5%) and moderate (9.1%). Mean vital signs of our sample are presented in Table 1. Many patients were transferred to KCMC from a neighboring hospital (21.3%). Overall, 2848 (88.8%) and 361 (11.2%) patients experienced a good and poor outcome respectively. Of the 361 patients who experienced a poor outcome, 326 patients (90.3%) died. The mean time from injury to death was 3.6 days (SD: 3.5 days). Among the remaining 2883 patients who survived, the mean time from injury to hospital discharge was 4.8 days (SD: 4.6 days).

Table 1

Demographic and clinical characteristics and crude association with a poor outcome.

Variable	Total	Good Outcome	Poor Outcome	p-Value	Odds Ratio (95% CI)	p-Value
Age, Mean (SD)	32.1 (16.5)	31.5 (16.3)	36.3 (17.7)	<0.001	1.02 (1.01–1.02)	<0.001
Sex, N (%)
Female	572 (17.8)	516 (90.2)	56 (9.8)		ref
Male	2637 (82.2)	2332 (88.4)	305 (11.6)	0.243	1.21 (0.90–1.64)	0.224
Mechanism of Injury, N (%)
Assault	462 (14.4)	434 (93.9)	28 (6.1)		ref
Road Traffic Injury	2171 (67.7)	1911 (88.0)	260 (12.0)		2.11 (1.43–3.22)	<0.001
Fall	347 (10.8)	293 (84.4)	54 (15.6)		2.86 (1.78–4.67)	<0.001
Other	229 (7.1)	210 (91.7)	19 (8.3)	<0.001	1.40 (0.76–2.55)	0.274
Glasgow Coma Score, N (%)
Mild	2517 (78.4)	2438 (96.9)	79 (3.1)		ref
Moderate	291 (9.1)	225 (77.3)	66 (22.7)		9.05 (6.34–12.90)	<0.001
Severe	401 (12.5)	185 (46.1)	216 (53.9)	<0.001	36.03 (26.87–48.78)	<0.001
Respiratory Rate, Mean (SD)	21.9 (5.0)	21.7 (4.8)	23.0 (6.5)	0.020	1.04 (1.02–1.05)	<0.001
Systolic Blood Pressure, Mean (SD)	121.7 (21.6)	121.6 (20.5)	122.5 (28.6)	<0.001	1.00 (1.00–1.01)	0.478
Pulse Oxygen, Mean (SD)	95.4 (7.5)	96.2 (5.6)	88.8 (14.3)	0.006	0.92 (0.91–0.93)	<0.001
Heart Rate, Mean (SD)	88.0 (18.1)	87.5 (17.1)	91.9 (24.1)	<0.001	1.01 (1.01–1.02)	<0.001
Transfer Status, N (%)
Not Transferred	370 (11.5)	341 (92.2)	29 (7.8)		ref
Transferred	684 (21.3)	608 (88.9)	79 (11.1)		1.47 (0.95, 2.33)	0.092
Unknown	2155 (67.2)	1899 (88.1)	256 (11.9)	0.069	1.59 (1.08, 2.41)	0.024

Time to arrival

The most common wait time from injury occurrence to hospital arrival was 1.1 to 4.0 hours (31.9%) followed by 12.1 hours or more (31.7%), 0.0 to 1.0 hours (18.9%), and 4.1 to 12.0 hours (17.5%) (Table 2). Delays from injury occurrence to hospital arrival were not different between mild, moderate, and severe GCS patients (p>0.05) (Fig 2). However, of patients with a known transfer status, most who arrived more than 4.0 hours after injury occurrence were transferred from a neighboring hospital (85.5%) (Fig 3). In addition, those with longer hospital arrival delays tended to be transferred patients (p<0.001). We found no significant associations between time to arrival and in-hospital outcome in our adjusted regression model (Fig 4).

Table 2

Delays to care and crude association with a poor outcome.

Variable		Total	Good Outcome	Poor Outcome	p-Value	Odds Ratio (95% CI)	p-Value
Time to Arrival, N (%)	0.0–1.0 h	607 (18.9)	536 (87.6)	71 (12.4)		ref
	1.1–4.0 h	1023 (31.9)	921 (88.8)	102 (11.2)		0.84 (0.61–1.16)	0.274
	4.1–12.0 h	562 (17.5)	497 (89.0)	65 (11.0)		0.99 (0.69–1.41)	0.944
	>12.0 h	1017 (31.7)	894 (89.3)	123 (10.7)	0.450	1.04 (0.76–1.42)	0.811
Time to Physician, N (%)	0.0–15.0 m	2220 (69.2)	1935 (87.0)	285 (13.0)		ref
	15.1–30.0 m	609 (19.0)	553 (91.6)	56 (8.4)		0.69 (0.50–0.92)	0.015
	30.1–45.0 m	132 (4.1)	123 (93.1)	9 (6.9)		0.50 (0.23–0.93)	0.046
	>45.0 m	248 (7.8)	237 (94.8)	11 (5.2)	<0.001	0.32 (0.16–0.56)	<0.001
Time to Labs Sent, N (%)	0.0–1.0 h	1554 (48.4)	1340 (86.1)	214 (13.9)		ref
	1.1–4.0 h	1111 (34.6)	1010 (91.5)	101 (8.5)		0.63 (0.49–0.80)	<0.001
	4.1–12.0 h	347 (10.8)	323 (92.6)	24 (7.4)		0.47 (0.29–0.71)	<0.001
	>12.0 h	197 (6.1)	175 (87.3)	22 (12.7)	<0.001	0.79 (0.48–1.23)	0.314
Time to X-ray, N (%)	0.0–1.0 h	1647 (51.3)	1427 (86.7)	220 (13.3)		ref
	1.1–4.0 h	1234 (38.5)	1123 (90.4)	111 (9.6)		0.64 (0.50–0.81)	<0.001
	>4.0 h	328 (10.2)	298 (93.0)	30 (7.0)	<0.001	0.65 (0.43–0.96)	0.037
Time to Brain CT, N (%)	0.0–1.0 h	39 (1.2)	31 (79.5)	8 (20.5)		ref
	1.1–4.0 h	55 (1.7)	45 (81.8)	10 (18.2)		0.86 (0.31–2.49)	0.777
	>4.0 h	51 (1.6)	45 (88.2)	6 (11.8)		0.52 (0.16–1.63)	0.262
	Not Received, Needed	778 (24.2)	503 (64.7)	275 (35.3)		2.12 (1.01–5.01)	0.063
	Not Received, Not Needed	2286 (71.2)	2224 (97.3)	62 (2.7)	<0.001	0.12 (0.05–0.26)	<0.001
Time to Fluids, N (%)	0.0–1.0 h	821 (25.6)	681 (82.9)	140 (17.1)		ref
	1.1–4.0 h	105 (3.3)	90 (85.7)	15 (14.3)		0.81 (0.44–1.40)	0.475
	>4.0 h	128 (4.0)	112 (87.5)	16 (12.5)		0.69 (0.39–1.18)	0.198
	Not Received, Needed	150 (4.7)	127 (84.7)	23 (15.3)		0.88 (0.53–1.40)	0.605
	Not Received, Not Needed	2005 (62.5)	1838 (91.7)	167 (8.3)	<0.001	0.44 (0.35–0.56)	<0.001
Time to Oxygen, N (%)	Received	104 (3.2)	33 (31.7)	71 (68.3)		ref
	Not Received, Needed	513 (16.0)	341 (66.5)	172 (33.5)		0.23 (0.15, 0.37)	<0.001
	Not Received, Not Needed	2592 (80.8)	2474 (95.4)	118 (4.6)	<0.001	0.02 (0.01, 0.03)	<0.001
Time to TBI Surgery, N (%)	0.0–12.0 h	304 (9.5)	266 (87.3)	38 (12.7)		ref
	>12.0 h	397 (12.4)	332 (83.9)	65 (16.1)		1.37 (0.90–2.13)	0.152
	Not Received	2508 (78.2)	2250 (89.7)	258 (10.3)	0.002	0.80 (0.56–1.17)	0.234
Time to Non-TBI Surgery, N (%)	0.0–12.0 h	164 (5.1)	146 (90.3)	18 (9.7)		ref
	>12.0 h	194 (6.0)	178 (91.8)	16 (8.2)		0.73 (0.36–1.48)	0.382
	Not Received	2851 (88.8)	2524 (88.5)	327 (11.5)	0.414	1.05 (0.65–1.80)	0.847

Fig 2

Proportion of mild, moderate, and severe GCS patients in each level of time to care variables.

R = Received. NR = Not Received. N = Needed. NN = Not Needed.

Fig 3

Proportion of non-transfers and transfers who experienced delays to hospital arrival following injury occurrence.

Fig 4

Results of adjusted logistic regression between time to care variables and in-hospital outcome.

Odds ratios show the odds of a poor outcome. R = Received. NR = Not Received. N = Needed. NN = Not Needed.

Time to physician evaluation

Upon hospital arrival, most patients waited 0.0 to 15.0 minutes to be evaluated by a physician in the ED (69.2%), followed by 15.1 to 30.0 minutes (19.0%), more than 45.0 minutes (7.8%), and 30.1 to 45.0 minutes (4.1%) (Table 2). A majority of mild (66.9%), moderate (71.1%), and severe (81.8%) GCS patients waited 15.0 minutes or less to be evaluated by an ED physician (Fig 2). Nearly all patients were evaluated within 45.0 minutes (92.2%). We found no significant associations between time to physician evaluation and in-hospital outcome in our adjusted regression model (Fig 4).

Time to diagnostics

Following physician evaluation, most patients waited 0.0 to 1.0 hours to receive lab tests (48.4%) (Table 2). A majority of severe GCS patients (56.1%) received lab tests within 1.0 hours of physician evaluation whereas a majority of mild (52.0%) and moderate (53.0%) GCS patients received lab tests more than 1.0 hours after evaluation. In our adjusted regression model, receiving lab tests 1.1 to 4.0 hours after physician evaluation was protective against a poor outcome in comparison to those receiving lab tests within 1.0 hours of evaluation (OR: 0.61; 95% CI: 0.43, 0.86) (Fig 4).

A majority of patients waited 0.0 to 1.0 hours after physician evaluation to receive an x-ray (51.3) (Table 2). Most severe GCS patients (61.3%) received an x-ray within 1.0 hours of evaluation whereas most mild (50.4%) and moderate (53.2%) GCS patients received an x-ray at least 1.1 hours after evaluation (Fig 2). We found no significant associations between time to x-ray and in-hospital outcome in our adjusted regression model (Fig 4).

Few patients in our sample received a brain CT scan (4.5%) (Table 2). Most patients did not need and did not receive a CT scan (71.2%). However, a large proportion of patients needed and did not receive a CT scan (24.2%). A majority of severe (95.0%) and moderate (92.1%) GCS patients did not receive a CT scan when needed (Fig 2). We found no significant associations between time to brain CT scan and in-hospital outcome in our adjusted regression model (Fig 4).

Time to treatments

One-fourth of patients in our sample received fluids within 1.0 hours of physician evaluation (25.6%). This includes 23.4% of mild, 30.9% of moderate, and 35.2% of severe GCS patients (Fig 2). However, a majority of patients did not need and did not receive fluids (62.5%) including most mild (65.3%), moderate (53.3%), and severe (51.6%) GCS patients. In our adjusted regression model, not needing and not receiving fluids was protective against a poor outcome in comparison to those who received fluids within 1.0 hours of evaluation (OR: 0.62; 95% CI: 0.44, 0.86) (Fig 4).

Only 3.2% of patients received oxygen (Table 2). Most of these patients had a severe GCS (74.0%) (Fig 2). A majority of patients did not need and did not receive oxygen (80.8%), including 93.4% of mild, 83.2% of moderate, and 0.0% of severe patients. However, 16.0% of patients needed and did not receive oxygen. This includes 6.1% of mild, 12.0% of moderate, and 80.8% of severe GCS patients. In our adjusted regression model, not needing and not receiving oxygen was protective against a poor outcome in comparison to patients who received oxygen (OR: 0.37; 95% CI: 0.18, 0.77) (Fig 4). Needing and not receiving oxygen was also protective against a poor outcome (OR: 0.40; 95% CI: 0.23, 0.70).

Most patients did not receive a TBI (78.2%) or non-TBI (88.8%) surgery (Table 2). TBI surgery recipients included 19.5% of mild, 28.5% of moderate, and 33.4% of severe GCS patients (Fig 2). Non-TBI surgery recipients included 11.1% of mild, 12.0% of moderate, and 9.0% of severe GCS patients. In our adjusted regression model, waiting more than 12.0 hours after physician evaluation for a non-TBI surgery was protective against a poor outcome in comparison to patients who waited 12.0 hours or less (OR: 0.32; 95% CI: 013, 0.79) (Fig 4).

Discussion

To our knowledge, this study is the first to investigate the epidemiology of pre- and in-hospital care delays among trauma patients in sub-Saharan Africa. Understanding the epidemiology of care delays experienced by ED patients is an important step in improving emergency care quality in low-resource hospitals. Our findings from a low-income national referral hospital in Tanzania offer insight into the patterns and consequences of care delays endured by trauma patients in LMICs. First, the status of pre-hospital care and hospital transfer systems in low-income settings may conceal differences between mild, moderate, and severely injured patients with regard to hospital arrival delays and their effect on trauma outcomes. Second, in-hospital delays experienced by trauma patients in low-income settings may be driven exclusively by constraints to non-personnel related resources rather than personnel related resources. Third, the easily accessible nature of time to care data makes it a potential tool for improving real time resource allocation, and thus trauma care, in low-resource hospitals.

Delays to hospital arrival

An estimated 91.3% of Africa’s population lacks access to emergency medical services, including prehospital care [48]. Consequently, individuals who sustain TBI must acquire transport through relatives, friends, bystanders, or commercial drivers [49]. Thus, severely injured patients are expected to experience longer delays to hospital arrival because they have greater difficulty arranging for transport. However, in our study, delays to hospital arrival were not significantly different between mild, moderate, and severe GCS patients (p = 0.768). We have two possible explanations for this result. First, as suggested by the literature, severely injured patients are more likely to succumb to airway compromise, respiratory failure, or uncontrolled hemorrhaging without prompt prehospital care [13]. A significant number of severely injured patients in countries without prehospital services may therefore die before reaching a hospital [50]. Second, a large percentage of patients in our registry were transferred to KCMC from other hospitals (Table 1). A sub-analysis of our data evaluating transfer status and time to hospital arrival revealed that patients with longer arrival delays tend to be transfers (p<0.001). Transferred patients must be able to survive transport and may therefore comprise primarily less severe presentations depending on geographical and contextual factors [51]. Thus, even severely injured patients who do reach their nearest hospital may not appear in our registry as we are a regional referral hospital.

Access to hospital resources

According to SATS, the ideal time from hospital arrival to physician evaluation for emergency, very urgent, and urgent trauma cases is immediately, less than 10 minutes, and less than 1.0 hours respectively [52]. The evaluation of nearly all severe GCS patients in our sample was within the timeframe recommended by SATS for emergency and very urgent cases; this demonstrates efficient triage prioritizing the critically injured. A comprehensive evaluation of 364 EDs in the United States revealed an average evaluation wait time of 31.8 minutes for cases classified as emergent (see in 0.0 to 14.0 minutes) [20]. Collectively, these high resource EDs failed to achieve national target wait times for their most time-sensitive patients. Nearly 70.0% of all patients in our sample (including 81.8% of all severely injured patients) were evaluated in 0.0 to 15.0 minutes, suggesting shorter wait times than what has been observed in the United States. Factors affecting patient inflow, notably ED overcrowding and triaging, are therefore unlikely causes of poor in-hospital outcomes among our sample.

Compared to personnel associated resources, our health system was more limited by non-personnel related resources as seen in other LMIC hospitals and which has been associated with poor outcomes [53, 54]. In our sample, 92.0% of moderate and 95.0% of severe GCS patients in our sample needed but did not receive a brain CT scan (Fig 2). Patients must be in a stable condition to undergo transport for medical imaging. Patient stability may explain the disparity we observed between mild and moderate/severe cases. However, in general, patient access to medical imaging is limited across countries in sub-Saharan Africa. In 2012, 100% of Kenya’s radiologists were located only in cities and major towns leaving the rural population at a disadvantage [55]. In 2011, only 56.0% of urban patients and 13.0% of rural patients in Uganda who required medical imaging were imaged [56]. In 2014, an evaluation of registered radiological equipment in Tanzania estimated 5.7 general radiography units per one million people; significantly lower than the WHO recommended 20 units per one million people [57]. Similar conditions restricting access to medical imaging may exist in other sub-Saharan African countries. The causes of low medical imaging access may include inequitable access to medical education among rural communities, a lack of radiological training interventions, limited human resources, and low health insurance coverage [55, 58]. Although we found no significant association between CT scan access and in-hospital outcome, the deficiency of CT scans for those most in need among our sample and the documented shortage of diagnostic imaging procedures across other sub-Saharan African countries warrant further investigation into the need for medical imaging equipment in sub-Saharan Africa and its impact on patient outcomes.

Nearly 81.0% of severe GCS patients in our sample needed but did not receive oxygen (Fig 2). Hypoxia has long been known to worsen TBI outcomes [59]. However, patients in our sample who needed and did not receive oxygen had lower odds of experiencing a poor in-hospital outcome (OR: 0.40; 95% CI: 0.23, 0.70). We believe that in the setting of limited monitoring capacity, oxygen may have been administered on the basis of visual rather than diagnostic indications, decreasing the likelihood of oxygen administration to patients with mild or visually undetectable hypoxia. In addition, for those who received oxygen, oxygen saturation may have been recorded into our registry either before or after oxygen administration. Consequently, we are unable to disaggregate non-hypoxic from hypoxic patients among those who received oxygen. As a result of these limitations, our reported associations between oxygen receipt and outcome status may not reflect the true relationship between these two variables. The need to address supplemental oxygen shortages in LMIC hospitals has been well documented [60, 61]. However, the impact of oxygen administration delays on TBI patient outcomes in LMICs has not been established and would benefit from continued investigations using more focused registries.

Surgical data

Only 21.8% of patients in our sample received a TBI surgery, including 28.5% of moderate and 33.4% of severe GCS patients (Fig 2). Observational studies conducted in neighboring regions of Tanzania have found even higher proportions of surgery recipients among moderate and severe TBI patients [36, 37]. Similar studies in Rwanda and Uganda have also documented higher rates of surgery among TBI patient cohorts [11, 62]. It is possible that we observed a lower provision of surgery among our sample because patients in our registry did not need surgery. Alternatively, the provision of surgery at KCMC may not be optimized such that those most in need are the ones receiving surgical treatment. Unfortunately, limitations to our data prevent us from determining a patient’s need for surgery and therefore which of these two possibilities is more likely. Regardless, the optimization of surgical provision remains an important aspect of TBI care both in Tanzania and in LMICs globally [63-65].

Tanzania has recently implemented reforms to build on its low national neurosurgical capacity [66]. In the meantime, innovative solutions are required to optimize surgical care. Most notably, prognostic models designed to improve triage through the prediction of TBI patient outcomes can assist in allocating surgical care to those who will most likely benefit [67]. The implementation of TBI registries in LMIC hospitals provides an opportunity to harness time to care data with the aim of optimizing the prediction performance of such innovative tools. Although the relationships presented in this study would benefit from further epidemiological investigations, time to care data is easy to collect and should be considered during the development of clinical decision-making technologies. Care delays could prove to be powerful predictors of TBI outcomes.

Limitations

Interpretation of our study’s results is not complete without careful consideration of limitations. First, the TBI registry used for analysis likely underrepresents the most severe injury cases who are more likely to die before reaching a hospital and are therefore less likely to appear in a hospital registry. While care delays may have the greatest impact on this patient population, our registry only captured patients who were well enough to reach the hospital. Since we would expect care delays to have the greatest impact on severely injured patients, we believe that a survivor bias inherent in our study would drive our measures of association towards the null.

Second, our variable time to hospital arrival was based on self-reported or family-reported information regarding the time of injury occurrence. This form of information bias may have led to exposure misclassification with regard to our variable time to hospital arrival. However, because all time information was recorded in the registry prior to ascertainment of patient outcomes, exposure misclassification was independent of patient outcome status. This form of non-differential exposure misclassification would bias our measure of association between time to hospital arrival and outcome status towards the null. Thus, we would expect the true association between hospital arrival delays and patient in-hospital outcome status to be stronger than what is reported in our study.

Finally, our analysis only considered patients presenting to the KCMC ED in Moshi, Tanzania. Given that KCMC is a regional referral hospital and the third largest hospital in the country, the quality of care provided at KCMC is likely preferable to care offered at most other hospitals in the Kilimanjaro region. Care delays may therefore have a larger impact on patients in these other hospitals if care quality is a mediator between care delays and patient outcomes.

Conclusion

Reducing delays to care is an important component of improving outcomes for all trauma patients. This study provides insight into the epidemiology of care delays among trauma patients presenting to a low-income tertiary referral hospital in sub-Saharan Africa. There is a paucity of literature on the subject of care delays in LMICs. Future studies should continue investigating the association between care delays and trauma patient outcomes in low-resource settings. Time to care data is easy to collect as it requires no tools and only two data points for a given patient: time of event A and time of event B. Time to care data collection may therefore be effortlessly integrated into hospital registries in any context.

25 Jun 2020

PONE-D-20-14993

An Analysis of Emergency Care Delays Experienced by Traumatic Brain Injury Patients Presenting to a Regional Referral Hospital in a Low-Income Country

PLOS ONE

Dear Dr. Staton,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please, clarify the points addressed b reviewer 1 to improve your manuscript.

Please submit your revised manuscript by Aug 09 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Hans-Peter Simmen, M.D., Professor of Surgery

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please provide additional details regarding participant consent.

In the ethics statement in the Methods and online submission information, please ensure that you have specified (i) whether consent was informed and (ii) what type you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed).

If your study included minors, state whether you obtained consent from parents or guardians.

If the need for consent was waived by the ethics committee, please include this information.

3. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Dear authors,

I have read your work with interest since the topic is of enormous importance. I think the manuscript is well-written. The compact title reflects the manuscript well. The introduction is clear and with sufficient references.

While reading your draft the following questions came to my mind:

A) Most importantly I would like you to clarify the timepoint at which you measured the outcome of your patients. How long was the time from injury to measuring the GOS on average with SD? The time to follow-up is not mentioned at all! Please adapt this aspects of your draft accordingly!

B) Regarding the reported results (line 197 ff): Please do not only differentiate between good and poor results regarding the GOS but also report on the mortality rates.

C) In Table 1 and Figure 3 you mention the possibility of an unknown Transfer Status-> please clarify.

D) In line 284 to 289 you state "Most patients did not receive a TBI (78.2%) or non-TBI (88.8%) surgery (Table 2). TBI surgery recipients included 19.5% of mild, 28.5% of moderate, and 33.4% of severe GCS patients (Figure 2). Non-TBI surgery recipients included 11.1% of mild, 12.0% of moderate, and 9.0% of severe GCS patients. In our adjusted regression model, waiting more than 12.0 hours after physician evaluation for a non-TBI surgery was protective against a poor outcome in comparison to patients who waited 12.0 hours or less (OR: 0.32; 95% CI: 013, 0.79)". What kind of TBI surgeries where performed based on what kind of decision making process? What did the injury pattern of the non-TBI surgery recipients look like and what kind of surgical interventions was performed?

E) In line 365 and 366 you state "However, for those who received oxygen,

366 oxygen saturation may have been recorded into our registry either before or after administration." If you are not able to say whether the oxygen saturation was measured before or after administration, how reliable is this information overall?

Reviewer #2: The authors report a retrospective analysis of acute TBI patients from a prospective patient registry of 3209 cases from the time period 2013 - 2017. The report is well written, the analyses are sound and the reported findings are interesting, original and important.

The major finding is, that improvement of TBI outcome (of those patients who live to see a physician) requires improvement of non-personnel related resources (availability of neuroimaging, oxygen and surgery).

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Lennart Stieglitz

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

18 Sep 2020

July 20, 2020

To the editors and reviewers of PLOS ONE,

Thank you for the opportunity to edit and revise our manuscript. We have tried to address each reviewer comment below.

Reviewer #1

Reviewer Comment: Dear authors, I have read your work with interest since the topic is of enormous importance. I think the manuscript is well-written. The compact title reflects the manuscript well. The introduction is clear and with sufficient references. While reading your draft the following questions came to my mind:

Reply: Thank you very much for this comment.

Reviewer Comment: A) Most importantly I would like you to clarify the timepoint at which you measured the outcome of your patients. How long was the time from injury to measuring the GOS on average with SD? The time to follow-up is not mentioned at all! Please adapt this aspects of your draft accordingly!

Reply: Thank you for this excellent point. GOS was measured at the time of hospital discharge. The mean time from injury to hospital discharge was 4.8 days (SD: 4.6 days). The mean time from injury to death was 3.6 days (SD: 3.5 days). This information has been included in the methods and results section of our revised manuscript.

Our dataset does not capture follow-up information as our TBI registry was part of a study designed to define areas for quality improvement in the initial management of TBI patients at KCMC. Our last point of contact with patients included in the registry was hospital discharge. This was clarified in the methods section of our revised manuscript.

Reviewer Comment: B) Regarding the reported results (line 197 ff): Please do not only differentiate between good and poor results regarding the GOS but also report on the mortality rates.

Reply: Thank you for this comment. Among the 361 patients who experienced a poor outcome, 326 patients (90.3%) died. Consequently, results differentiating between those who survived and those who died are not significantly different from results differentiating between those who experienced a good outcome and those who experienced a poor outcome. We therefore chose only to differentiate between good and poor outcomes. We have included in the results section of our revised manuscript a sentence stating the number of deaths in the poor outcome group as well as our reason for reporting results as such in the methods section of our revised manuscript.

Reviewer Comment: C) In Table 1 and Figure 3 you mention the possibility of an unknown Transfer Status-> please clarify.

Reply: Thank you for this comment. Our variable Transfer Status was categorized as transferred from a previous hospital or not transferred from a previous hospital. Patients with missing prehospital course information were categorized as having an unknown transfer status, as we could not determine whether or not these patients were transferred to KCMC from a previous hospital. An explanation of the unknown transfer status category has been included in the methods section of our revised manuscript.

Reviewer Comment: D) In line 284 to 289 you state "Most patients did not receive a TBI (78.2%) or non-TBI (88.8%) surgery (Table 2). TBI surgery recipients included 19.5% of mild, 28.5% of moderate, and 33.4% of severe GCS patients (Figure 2). Non-TBI surgery recipients included 11.1% of mild, 12.0% of moderate, and 9.0% of severe GCS patients. In our adjusted regression model, waiting more than 12.0 hours after physician evaluation for a non-TBI surgery was protective against a poor outcome in comparison to patients who waited 12.0 hours or less (OR: 0.32; 95% CI: 013, 0.79)". What kind of TBI surgeries where performed based on what kind of decision making process? What did the injury pattern of the non-TBI surgery recipients look like and what kind of surgical interventions was performed?

Reply: Thank you for this comment. TBI surgeries are limited to burr holes, craniotomies, and craniectomies conducted by trained general surgeons as there are no neurosurgeons stationed at KCMC. However, our registry does not include information on the specific type of TBI surgery received by a patient. The decision to operate on a TBI patient and what kind of operation to perform is ultimately based on the judgement of a general surgeon trained in neurosurgical procedures. Alternatively, a patient or patient’s family may choose to forgo surgery if the cost cannot be covered. Again, our registry does not include information on the decision process used to inform the provision of surgery for a given patient. This information has been included in the methods section of the revised manuscript.

With regard to non-TBI surgeries, our registry does not include the types of operations performed on non-TBI surgery recipients nor does it include the injury patterns of non-TBI surgery recipients. Based on other injury registries currently established at KCMC, we do know that non-TBI surgeries include primarily otolaryngology, general, and orthopedic operations. However, we do not have specific details of these procedures and these procedures may not represent the operations received by non-TBI surgery recipients in our dataset. Thus, we chose not to include this information in our manuscript.

Reviewer Comment: E) In line 365 and 366 you state "However, for those who received oxygen, oxygen saturation may have been recorded into our registry either before or after administration." If you are not able to say whether the oxygen saturation was measured before or after administration, how reliable is this information overall?

Reply: Thank you for this comment. Among patients who received oxygen, we cannot be certain whether oxygen saturation was measured before or after oxygen administration. Although this limitation makes our oxygen saturation data non-optimal for these patients, it has the effect of underestimating potential hypoxia, and it may help explain why we observed an unintuitive association between oxygen receipt and a worse outcome status. For example, a patient with a poor outcome may have presented with hypoxia, received oxygen, and then had their oxygen saturation measured. Thus, this hypoxic patient would appear in our registry as non-hypoxic. We have included a statement in the discussion that addresses the reliability of our results regarding our reported associations between oxygen receipt and outcome status.

Reviewer #2

Reviewer Comment: The authors report a retrospective analysis of acute TBI patients from a prospective patient registry of 3209 cases from the time period 2013 - 2017. The report is well written, the analyses are sound and the reported findings are interesting, original and important. The major finding is, that improvement of TBI outcome (of those patients who live to see a physician) requires improvement of non-personnel related resources (availability of neuroimaging, oxygen and surgery).

Reply: Thank you very much for this comment.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

29 Sep 2020

An analysis of emergency care delays experienced by traumatic brain injury patients presenting to a regional referral hospital in a low-income country

PONE-D-20-14993R1

Dear Dr. Staton,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Hans-Peter Simmen, M.D., Professor of Surgery

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

2 Oct 2020

PONE-D-20-14993R1

An analysis of emergency care delays experienced by traumatic brain injury patients presenting to a regional referral hospital in a low-income country

Dear Dr. Staton:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Hans-Peter Simmen

Academic Editor

PLOS ONE