Differences in trauma care between district and regional hospitals and impact of a trauma intake form with decision support prompts in Ghana: A stepped-wedge cluster randomized trial

Abstract

Background:

We sought to determine the achievement of key performance indicators (KPIs) of initial trauma care at district (first-level) and regional (second-level) hospitals in Ghana and to assess the effectiveness of a standardized trauma intake form (TIF) to improve care.

Methods:

A stepped-wedge cluster randomized trial was performed with direct observations of trauma management before and after introducing the TIF at emergency units of eight hospitals for 17.5 months. Differences in KPIs were assessed using multivariable logistic regression and generalized linear mixed regression.

Results:

Management of 4,077 patients was observed; 30% at regional and 70% at district hospitals. Eight of 20 KPIs were performed significantly more often at regional hospitals. TIF improved care at both levels. Fourteen KPIs improved significantly at district and eight KPIs improved significantly at regional hospitals. After TIF, regional hospitals still performed better, with 18 KPIs being performed significantly more often than district hospitals. After TIF, all KPIs were performed in >90% of patients at regional hospitals. Examples of KPIs for which regional performed better than district hospitals after TIF included: assessment for oxygen saturation (83% vs. 98%) and evaluation for intra-abdominal bleeding (82% vs. 99%, all p< 0.001). Mortality decreased among seriously injured patients (Injury Severity Score ≥ 9) at both district (15% before vs. 8% after, p=0.04) and regional (23% vs. 7%, p=0.004) hospitals.

Conclusions:

TIF improved care and lowered mortality at both hospital levels, but KPIs remained lower at district hospitals. Further measures are needed to improve initial trauma care at this level.

1. Introduction

Injury is a leading cause of death globally, accounting for 4.4 million deaths annually, most of which are in low- and middle-income countries (LMICs) [1]. In addition to primary prevention, strengthening care of the injured is an important component of efforts to lower the injury burden [2]. There is especially a need for better organization and planning to improve the quality of care delivered. Use of the World Health Organization’s (WHO) Trauma Care Checklist (TCC) resulted in notable improvements in key performance indicators (KPIs) of initial trauma care for all patients and lowered mortality among severely injured in tertiary hospitals in nine countries [3].

Ensuring the quality of trauma care is necessary at all hospital levels. Measures to assure quality may need to be implemented differently at different levels. Differing resources and capabilities for trauma care at different hospital levels have been well documented in many LMICs [4–8]. Planning for such differing levels of resources is an important component of trauma system planning in many countries [9, 10]. The effect of these variations on achievement of KPIs at different hospital levels has not been adequately addressed.

The first aim of this study was to understand how the process of initial trauma care and achievement of KPIs differed between district (first-level) compared with regional (second-level) hospitals in Ghana. The second aim was to determine whether a quality improvement initiative (the trauma intake form, with built-in clinical decision support prompts) impacted care differently at these different hospital levels. Results of the trial for the overall patient population (all hospitals combined) have been previously reported [11]. The current study focuses on differences in care at different hospital levels.

2. Methods

2.1. Setting

Ghana is a lower-middle income country with a population of 33 million [12]. Trauma care is provided at district (first-level), regional (second-level or referral), and tertiary hospitals. At district hospitals, basic trauma care is provided by medical officers (general, non-specialist doctors), nurses, and/or physician assistants, and sometimes by specialists. Regional hospitals have specialists (e.g., general or orthopedic surgeons) while tertiary hospitals provide advanced trauma care [13]. Essential trauma care items and services such as chest tubes, diagnostic imaging, and major general surgery are frequently lacking at district hospitals, with slight improvement in availability at regional hospitals. Ghanaian hospitals lack routine quality improvement processes for monitoring trauma care [5].

2.2. Study Design

We performed a stepped-wedge, cluster randomized trial with a quality improvement intervention (trauma intake form (TIF)) to improve initial assessment and care provided by emergency unit (EU) health service providers (EHSP) in six district and two regional hospitals, purposively selected for their adequate volume of injured patients (≥75 patients per month). The TIF was designed as a checklist that provides real-time clinical decision support prompts to promote adherence to the ABCDE approach to initial trauma care [14]. Items on the TIF were adapted from the WHO TCC and context-appropriate audit filters for non-tertiary hospitals in LMICs [3, 15], and represented KPIs for initial trauma care.

2.3. Study procedures

Details of study procedures, including data collection methods, hospital sampling, and power estimation have been previously described [11, 16]. Briefly, trained research assistants (RAs) observed and documented initial assessment and care of injured patients at EUs of study hospitals for an initial period of 3.5 months. The TIF was then introduced to 2 nearby hospitals, selected by simple random sampling, after training the EHSPs on its use. Two nearby hospitals were randomized together for ease of study management. This was followed by sequentially introducing the TIF to randomly selected groups of 2 hospitals every 3.5 months until study completion at 17.5 months. RAs were posted at each EU in rotating 8-hour shifts to record completion of KPIs (e.g., checking vital signs, airway assessment), or lack thereof, using an observation form derived from the TIF.

Direct observations were done without interaction with patients or EHSPs and complemented with medical records. KPIs were considered as performed if RAs directly observed EHSPs perform them or if they were documented in the medical record or both. Where performing a given KPI was not required (e.g., abdominal examination in an isolated hand injury), RAs recorded the KPI as performed if the medical record indicated that it was not required.

2.5. Data analysis

Analyses were performed with Stata 17 (StataCorp, USA). Hospitals were grouped according to level of care (district vs. regional). Primary outcomes were performance of KPIs and documentation of data in medical records. “Important clinical data documented” was defined as documentation of all of patient age, sex, injury type, injury intent, injury mechanism, consciousness level (AVPU), heart rate, and blood pressure at EU arrival. Abbreviated Injury Scale (AIS) scores were assigned to each injury and overall Injury Severity Score (ISS) was calculated from these [17]. Serious injury was defined as ISS≥9 (including moderate, severe, and higher categories) [17, 18]. Secondary outcomes were complications (any of pneumonia, wound infection, or deep vein thrombosis) and in-hospital mortality.

Analyses included only patients with complete primary outcome data. Characteristics of patients managed at different hospital levels were compared with Chi-square. Differences in outcomes between patients managed at different levels of care were determined with multivariable logistic regression, adjusted for hospital cluster and significantly different variables in bivariate analysis (i.e., patient age and injury severity). Differences in outcomes at the different levels of care before vs. after TIF introduction were estimated using generalized linear mixed regression [19, 20], adjusted for time period and hospital cluster and significantly different variables in bivariate analysis (i.e., sex, injury mechanism, and injury severity at regional hospitals). Data were analyzed on intention-to-treat basis. Patients managed after TIF introduction were considered in the intervention arm regardless of whether the TIF was applied in their management.

2.6. Ethics

The study was approved by Kwame Nkrumah University of Science and Technology’s IRB (CHRPE/AP/142/20). EHSPs provided written informed consent to be observed. The study was pre-registered at clinicaltrials.gov (NCT04547192).

3. Results

There were 4,158 patients assessed for eligibility into the study. After excluding non-trauma patients and those with incomplete primary outcome data, 4,077 patients were included for analysis (Figure 1).

Figure 1.

Trial CONSORT flow diagram

3.1. Patient characteristics

There were 2,835 (70%) injured patients initially managed at district hospitals. They had a mean age of 28±16 (SD) years compared to 31±15 (SD) years for those managed at regional hospitals, (p<0.001). District hospitals managed less seriously injured patients (6%) than regional hospitals (15%), (p< 0.001) (Table 1). Sex, mechanism, and intent of injury were similar between the two hospital groups.

Table 1.

Characteristics of injured patients presenting to emergency units (EU) of select Ghanaian district and regional hospitals (N=4,077)

All patients (n=4,077)
	District hospital (n=2,835)		Regional hospital (n=1,242)
	N	(%)	N	(%)	p-value*
Sex
Male	2,079	(73)	931	(75)	0.366
Female	736	(26)	307	(25)
Missing	20	(0.7)	4	(0.3)
Age, Mean (SD), Years	28	(16)	31	(15)	<0.001
Mechanism of injury
Blunt	2,157	(76)	970	(78)	0.187
Penetrating	611	(22)	239	(19)
Burns	63	(2)	33	(3)
Missing	4	(0.1)	0	(0)
Intent
Unintentional	2,508	(88)	1,107	(89)	0.063
Intentional	279	(10)	134	(11)
Unknown/Missing	48	(2)	1	(0.1)
Seriously injured	172	(6)	189	(15)	<0.001
Patients presenting to district hospitals (n=2,835)
	Before TIF introduction (n=1,586)		After TIF introduction (n=1,249)
	N	(%)	N	(%)	p-value*
Sex
Male	1,151	(73)	928	(74)	0.631
Female	415	(26)	321	(26)
Missing	20	(1)	0	(0)
Age, Mean (SD), Years	28	(16)	29	(16)	0.345
Mechanism of injury
Blunt	1,189	(75)	968	(78)	0.094
Penetrating	363	(23)	248	(20)
Burns	31	(2)	32	(3)
Missing	3	(0.2)	1	(0.1)
Intent
Unintentional	1,427	(90)	1,081	(87)	0.605
Intentional	150	(9)	129	(10)
Unknown/Missing	9	(0.6)	39	(3)
Seriously injured	84	(5)	88	(7)	0.054
Referred to higher level of care	215	(14)	182	(15)	0.439
Patients presenting to regional hospitals (n=1,242)
	Before TIF introduction (n=481)		After TIF introduction (n=761)
	N	(%)	N	(%)	p-value*
Sex
Male	379	(79)	522	(73)	0.020
Female	102	(21)	205	(27)
Missing	0	(0)	4	(0.5)
Age, Mean (SD), Years	31	(15)	31	(15)	0.899
Mechanism of injury
Blunt	394	(82)	576	(76)	0.035
Penetrating	77	(16)	162	(21)
Burns	10	(2)	23	(3)
Intent
Unintentional	427	(89)	680	(89)	0.442
Intentional	53	(11)	81	(11)
Unknown/Missing	1	(0.2)	0	(0)
Seriously injured	91	(19)	98	(13)	0.004
Referred to higher level of care	17	(4)	31	(4)	0.631

Statistical tests done excluding missing data.

p-values in bold denotes statistical significance.

EU: Emergency Unit

Seriously injured – Injury Severity Score ≥9.

3.2. Initial management of injury: district vs. regional hospitals at baseline

KPIs were at lower baseline levels at district compared to regional hospitals. Before TIF introduction, eight of 20 KPIs were performed less often at district hospitals than at regional hospitals. Examples include assessment for oxygen saturation (45% district vs. 88% regional, p=0.001) and evaluation for intra-abdominal bleeding (22% vs. 83%, p=0.024). Documentation of important clinical data was also performed less often at district hospitals (40% vs. 92%, p<0.001) (Table 2). Consideration for analgesics was, however, higher at district hospitals (96% vs. 92%, p=0.009). Before TIF introduction, nine of 20 KPIs were performed less than 50% of the time at district hospitals, compared to only 2 of 20 KPIs being performed less than 50% of the time at regional hospitals (Table 2).

Table 2.

Assessment and management of injured patients presenting to emergency unit (EU) of select Ghanaian district and regional hospitals before TIF introduction (N=2,067)

	District hospital (n=1,586)		Regional hospital (n=481)
	N	(%)	N	(%)	AOR	p-value
Triage and monitoring
Mobility at EU arrival assessed	1,328	(84)	479	(99.6)	0.02	<0.001
Respiratory rate at EU arrival assessed	723	(46)	471	(98)	0.02	<0.001
Temperature at EU arrival assessed	1,167	(74)	477	(99.2)	0.02	<0.001
Oxygen saturation level at EU assessed	706	(45)	422	(88)	0.13	0.001
Primary assessment and actions
Airway assessed	1,128	(71)	378	(79)	0.78	0.865
Chest examined	889	(56)	395	(82)	0.32	0.377
Intravenous line placed	723	(46)	370	(77)	0.31	0.004
External bleeding checked for and controlled	826	(52)	401	(83)	0.21	0.254
Intra-abdominal bleeding evaluateda	350	(22)	397	(83)	0.06	0.024
Pelvic fracture evaluated	288	(18)	344	(72)	0.10	0.050
All distal pulses checked	349	(22)	350	(73)	0.12	0.129
Fluid and/or blood requirement considered	657	(41)	333	(69)	0.40	0.139
Spine immobilized for RTI or fall victims (n=1,106 district, 347 regional)b	65	(6)	185	(53)	0.06	0.003
Splinting of fractures considered (n=224 district, 120 regional)	140	(63)	64	(53)	1.53	0.399
Physical examination findings recorded	1,484	(94)	476	(99)	0.18	0.164
Alcohol on breath assessedc	291	(18)	140	(29)	0.61	0.364
Total burn surface area estimated (n=32 district, 9 regional)	18	(56)	3	(33)	7.41	0.142
Tetanus considered for bites, burns, lacerations, and abrasions (n= 1,271 district, 383 regional)	906	(71)	335	(87)	0.36	0.306
Antibiotic considered (e.g., open fracture)	1,271	(80)	437	(91)	0.38	0.411
Analgesics considered	1,529	(96)	443	(92)	2.35	0.009
Documentation
Blood pressure	926	(58)	451	(94)	0.10	0.001
Heart rate	955	(60)	474	(99)	0.02	<0.001
Important clinical data documentedd	639	(40)	444	(92)	0.06	<0.001
Complications
Any complication	4	(0.3)	20	(4)	0.06	0.044
Mortality	22	(1)	25	(5)	0.62	0.452
Mortality among seriously injured (n=84 district; n=91 regional)	13	(15)	21	(23)	0.55	0.386

Key performance indicators listed in the table represent elements of the Trauma Intake Form

EU– Emergency Unit; RTI – Road Traffic Injury; EHSP – Emergency Unit Health Service Provider; AOR – Adjusted Odds Ratio (adjusted for patient age and injury severity score, and hospital (cluster)).

p-values in bold denotes statistical significance.

^aInternal abdominal bleeding ruled out by any one of: abdominal exam, ultrasound, X Ray, computerized tomography;

^bSpine immobilization implies use of either cervical collar or backboard;

^cAssessed by smelling the patient’s breathe;

^dImportant clinical data documented - all of the following: patient sex, patient age, mechanism of injury, intent of injury, heart rate at EU arrival, blood pressure at EU arrival, consciousness level at EU arrival, and injury type

3.3. Impact of TIF on KPIs of initial management of injury

The TIF was used in 969 (78%) district hospital patients and 751 (99%) regional hospital patients (p<0.001). Patient demographics at district hospitals were similar before and after TIF introduction. At regional hospitals, however, there were higher proportions of males, blunt injury mechanisms, and seriously injured patients before TIF introduction (Table 1). Improvements in 14 out of 20 KPIs were recorded at district hospitals after TIF introduction (Tables 3 and 5). Examples include airway assessment (71% to 98%, p=0.020) and chest examination (56% to 95%, p=0.002), which improved by ≥10% with compliance of ≥90% after TIF introduction. KPIs which improved ≥10% with 80%-89% compliance after TIF introduction included evaluation for intra-abdominal bleeding (22% to 82%, p=0.001) and tetanus consideration for open wounds (71% to 81%, p<0.010). Some KPIs had >10% improvement but with compliance after TIF introduction still <80%, such as assessment for pelvic fracture (18% to 79%, p<0.001). Regional hospitals recorded significant improvements in eight of 20 KPIs after TIF introduction (Tables 4 and 5). These include KPIs which improved by ≥10% with compliance of ≥90% after TIF introduction such as airway assessment (79% to 99%, p<0.001) and chest examination (82% to 99.9%, p<0.001). After TIF, all KPIs were achieved in over 90% of patients at regional hospitals.

Table 3.

Comparison of assessment and management of injured patients presenting to emergency unit (EU) of select Ghanaian district hospitals before vs. after TIF introduction (n=2,835)

	Before TIF introduction (n=1,586)		After TIF introduction (n=1,249)
	N	(%)	N	(%)	AOR	p-value
Triage and monitoring
Mobility at EU arrival assessed	1,328	(84)	1,138	(91)	2.10	0.254
Respiratory rate at EU arrival assessed	723	(46)	1,033	(83)	3.51	0.072
Temperature at EU arrival assessed	1,167	(74)	1,122	(90)	1.81	0.228
Oxygen saturation level at EU assessed	706	(45)	1038	(83)	3.99	0.003
Primary assessment and actions
Airway assessed	1,128	(71)	1,221	(98)	19.99	0.020
Chest examined	889	(56)	1,185	(95)	32.65	0.002
Intravenous line placed	723	(46)	1,143	(92)	62.23	<0.001
External bleeding checked for and controlled	826	(52)	1,118	(90)	13.95	0.030
Intra-abdominal bleeding evaluateda	350	(22)	1,020	(82)	64.38	0.001
Pelvic fracture evaluated	288	(18)	988	(79)	86.20	<0.001
All distal pulses checked	349	(22)	1,097	(88)	375.04	<0.001
Fluid and/or blood requirement considered	657	(41)	1,105	(88)	25.72	<0.001
Spine immobilized for RTI or fall victims (n=1,106 before, 907 after)b	65	(6)	723	(80)	546.65	<0.001
Splinting of fractures considered (n=224 before, 182 after)	140	(63)	165	(91)	3.26	0.015
Physical examination findings recorded	1,484	(94)	1,207	(97)	0.59	0.625
Alcohol on breath assessedc	291	(18)	672	(54)	117.02	<0.001
Total burn surface area estimated (n=32 before, 29 after)	18	(56)	14	(48)	43.41	0.457
Tetanus considered for bites, burns, lacerations, and abrasions (n=1,271 before, 999 after)	906	(71)	812	(81)	5.66	0.010
Antibiotic considered (e.g., open fracture)	1,271	(80)	1,128	(90)	4.33	0.056
Analgesics considered	1,529	(96)	1,239	(99)	6.30	0.002
Documentation
Blood pressure	926	(58)	960	(77)	1.73	0.054
Heart rate	955	(60)	1,095	(88)	2.20	0.073
Important clinical data documentedd	639	(40)	847	(68)	2.58	0.009
Complications
Any complication	4	(0.3)	6	(0.5)	7.33	0.237
Mortality	22	(1)	13	(1)	0.59	0.322
Mortality among seriously injured (n=84 before; n=88 after)	13	(15)	7	(8)	0.25	0.040

Key performance indicators listed in the table represent elements of the Trauma Intake Form

EU– Emergency Unit; RTI – Road Traffic Injury; EHSP – Emergency Unit Health Service Provider; AOR – Adjusted Odds Ratio (adjusted for time period and hospital (cluster)); Seriously injured – Injury Severity Score ≥9.

p-values in bold denotes statistical significance.

^aInternal abdominal bleeding ruled out by any one of: abdominal exam, ultrasound, X Ray, computerized tomography;

^bSpine immobilization implies use of either cervical collar or backboard;

^cAssessed by smelling the patient’s breathe;

^dImportant clinical data documented - all of the following: patient sex, patient age, mechanism of injury, intent of injury, heart rate at EU arrival, blood pressure at EU arrival, consciousness level at EU arrival, and injury type.

Table 5.

Summary of improvements in KPIs after TIF introduction among injured patients presenting at emergency unit (EU) of district and regional hospitals

	Key Performance Indicator
	District hospitals	Regional hospitals
≥10% improved AND had ≥90% compliance after TIF	Airway assessed *	Airway assessed *
	Chest examined *	Chest examined *
	iv line placed *	iv line placed *
	External bleeding checked and controlled *	External bleeding checked and controlled *
	Splinting of fractures considered *	Intraabdominal bleeding evaluated *
	Temperature at EU arrival	Alcohol on breath assessed *
	Antibiotics considered	Total burn surface area estimated *
		Pelvic fracture evaluated
		All distal pulses checked
		Fluid and /or blood requirement considered
		Tetanus considered for bites, burns, lacerations, and abrasions
		Spine immobilized for RTI or fall victims
		Splinting of fractures considered
		Oxygen saturation at EU arrival
≥10% improved AND had 80-89% compliance after TIF	Oxygen saturation at EU arrival *
	Intraabdominal bleeding evaluated *
	All distal pulses checked *
	Fluid and /or blood requirement considered *
	Spine immobilized for RTI or fall victims *
	Tetanus considered for bites, burns, lacerations, and abrasions *
	Heart rate documented
>10% improved BUT still <80% compliance after TIF	Pelvic fracture evaluated *
	Alcohol on breath assessed *
	Important clinical data documented *
	Blood pressure documented
Already > 90% compliance (before and after)	Analgesics considered *	Analgesics considered *
	Physical examination findings recorded	Respiratory rate at EU arrival
		Mobility at EU arrival
		Physical examination findings recorded
		Temperature at EU arrival
		Heart rate documented
		Antibiotics considered
		Important clinical data documented

^*Improvement in key performance indicator is statistically significant (p< 0.05)

Table 4.

Comparison of assessment and management of injured patients presenting to emergency unit (EU) of select Ghanaian regional hospitals before vs. after TIF introduction (n=1,242)

	Before TIF introduction (n=481)		After TIF introduction (n=761)
	N	(%)	N	(%)	AOR	p-value
Triage and monitoring
Mobility at EU arrival assessed	479	(99.6)	756	(99.3)	0.63	0.713e
Respiratory rate at EU arrival assessed	471	(98)	754	(99)	2.29	0.130e
Temperature at EU arrival assessed	477	(99)	751	(99)	1.59	0.584e
Oxygen saturation level at EU assessed	422	(88)	747	(98)	0.90	0.942
Primary assessment and actions
Airway assessed	378	(79)	757	(99)	51.57	<0.001
Chest examined	395	(82)	760	(99.9)	165.47	<0.001
Intravenous line placed	370	(77)	754	(99)	207.28	<0.001
External bleeding checked for and controlled	401	(83)	760	(99.9)	37.28	0.009
Intra-abdominal bleeding evaluateda	397	(83)	756	(99)	21.62	0.002
Pelvic fracture evaluated	344	(72)	749	(98)	0.32	0.434
All distal pulses checked	350	(73)	754	(99)	1.97	0.768
Fluid and/or blood requirement considered	333	(69)	755	(99)	9.20	0.121
Spine immobilized for RTI or fall victims (n=347 before, 529 after)b	185	(53)	514	(97)	0.83	0.850
Splinting of fractures considered (n=120 before, 134 after)	64	(53)	128	(96)	5.06	0.080
Physical examination findings recorded	476	(99)	760	(99.8)	3.33	0.416
Alcohol on breath assessedc	140	(29)	728	(96)	24.41	<0.001
Total burn surface area estimated (n=9 before, 23 after)	3	(33)	23	(100)	-	<0.001 e
Tetanus considered for bites, burns, lacerations, and abrasions (n=383 before, 637 after)	335	(87)	622	(98)	1.16	0.857
Antibiotic considered (e.g., open fracture)	437	(91)	751	(99)	1.94	0.615
Analgesics considered	443	(92)	761	(100)	-	<0.001 e
Documentation
Blood pressure	451	(94)	698	(92)	8.27	0.012
Heart rate	474	(99)	754	(99)		0.416
Important clinical data documentedd	444	(92)	695	(91)	2.64	0.135
Complications
Any complication	20	(4)	0	(0)	-	<0.001 e
Mortality	25	(5)	9	(1)	0.37	0.438
Mortality among seriously injured (n=91 before; n=98 after)	21	(23)	7	(7)	0.26	0.004 e

Key performance indicators listed in the table represent elements of the Trauma Intake Form

EU– Emergency Unit; RTI – Road Traffic Injury; EHSP – Emergency Unit Health Service Provider; AOR – Adjusted Odds Ratio (adjusted for time period, patient sex, injury mechanism, and injury severity score, and hospital (cluster)).

p-values in bold denotes statistical significance.

^aInternal abdominal bleeding ruled out by any one of: abdominal exam, ultrasound, X Ray, computerized tomography;

^bSpine immobilization implies use of either cervical collar or backboard;

^cAssessed by smelling the patient’s breathe;

^dImportant clinical data documented - all of the following: patient sex, patient age, mechanism of injury, intent of injury, heart rate at EU arrival, blood pressure at EU arrival, consciousness level at EU arrival, and injury type.

^eBivariate analysis using Fisher’s exact test presented, due to one or more cells with zero values or non-convergence of logistic regression model; calculation of odds ratios not possible when one of cells contains zero values.

3.4. Status of initial management of injury after TIF: district vs. regional hospitals in after period

After TIF introduction, district hospitals still performed KPIs less often than regional hospitals, with 18 of 20 KPIs being performed significantly less often at district hospitals. Examples include assessment for oxygen saturation (83% vs. 98%, p< 0.001) and evaluation for intra-abdominal bleeding (82% vs. 99%, p<0.001). In addition, documentation of important clinical data improved at district hospitals but remained lower than at regional hospitals (68% vs. 91%, p=0.016) (Table 6). KPIs were at lower baseline levels (before TIF) at district compared to regional hospitals (Table 2), and hence had further to increase. Lower starting points for district hospitals could partly explain why they did not achieve similar levels of KPI performance after TIF.

Table 6.

Assessment and management of injured patients presenting to emergency unit (EU) of select Ghanaian district and regional hospitals after TIF implementation (N=2,010)

	District hospital (n=1,249)		Regional hospital (n=761)
	N	(%)	N	(%)	AOR	p-value
Triage and monitoring
Mobility at EU arrival assessed	1,138	(91)	756	(99)	0.01	<0.001
Respiratory rate at EU arrival assessed	1,033	(83)	754	(99)	0.02	<0.001
Temperature at EU arrival assessed	1,122	(90)	751	(99)	0.07	<0.001
Oxygen saturation level at EU assessed	1,038	(83)	747	(98)	0.07	<0.001
Primary assessment and actions
Airway assessed	1,221	(98)	757	(99)	0.27	0.124
Chest examined	1,185	(95)	760	(99.9)	0.03	<0.001
Intravenous line placed	1,143	(92)	754	(99)	0.11	0.001
External bleeding checked for and controlled	1,118	(90)	760	(99.9)	0.01	<0.001
Intra-abdominal bleeding evaluateda	1,020	(82)	756	(99)	0.03	<0.001
Pelvic fracture evaluated	988	(79)	749	(98)	0.07	0.001
All distal pulses checked	1,097	(88)	754	(99)	0.08	<0.001
Fluid and/or blood requirement considered	1,105	(88)	755	(99)	0.07	<0.001
Spine immobilized for RTI or fall victims (n=907 district, 529 regional)b	723	(80)	514	(97)	0.11	0.002
Splinting of fractures considered (n=182 district, 134 regional)	165	(91)	128	(96)	0.46	0.234
Physical examination findings recorded	1,207	(97)	760	(99.9)	0.04	<0.001
Alcohol on breath assessedc	672	(54)	728	(96)	0.05	0.003
Total burn surface area estimated (n=29 district, 23 regional)	14	(48)	23	(100)	-	<0.001 e
Tetanus considered for bites, burns, lacerations, and abrasions (n=999 district, 637 regional)	812	(81)	622	(98)	0.11	0.015
Antibiotic considered (e.g., open fracture)	1,128	(90)	751	(99)	0.13	0.033
Analgesics considered	1,239	(99)	761	(100)	-	<0.001 e
Documentation
Blood pressure	960	(77)	698	(92)	0.29	0.074
Heart rate	1,095	(88)	754	(99)	0.03	<0.001
Important clinical data documentedd	847	(68)	695	(91)	0.20	0.016
Complications
Any complication	6	(0.5)	0	(0)	-	0.089e
Mortality	13	(1)	9	(1)	1.36	0.200
Mortality among seriously injured (n=88 district; n=98 regional)	7	(8)	7	(7)	1.11	0.736

Key performance indicators listed in the table represent elements of the Trauma Intake Form

EU– Emergency Unit; RTI – Road Traffic Injury; EHSP – Emergency Unit Health Service Provider; AOR – Adjusted Odds Ratio (adjusted for patient age and injury severity score, and hospital (cluster)).

p-values in bold denotes statistical significance.

^aInternal abdominal bleeding ruled out by any one of: abdominal exam, ultrasound, X Ray, computerized tomography;

^bSpine immobilization implies use of either cervical collar or backboard;

^cAssessed by smelling the patient’s breathe;

^dImportant clinical data documented - all of the following: patient sex, patient age, mechanism of injury, intent of injury, heart rate at EU arrival, blood pressure at EU arrival, consciousness level at EU arrival, and injury type;

^eBivariate analysis using Fisher’s exact test presented, due to one or more cells with zero values; calculation of odds ratios not possible when one of cells contains zero values.

3.5. Referral patterns

Similar percentages of patients were referred to higher levels of care from district hospitals in the before (n=215, 14%) and after (n=182, 15%; p=0.439) periods. Similar percentages were referred from regional hospitals in the before (n=17, 4%) and after (n=31, 4%; p=0.631) periods. The percentages referred were higher at district compared with regional hospitals in both the before (14% vs. 4%; p<0.001) and after (15% vs. 4%, p<0.001) periods.

3.6. Mortality and complications

There was a reduction in overall complication rate among regional hospital patients (4% before TIF vs. 0% after TIF, p<0.001), but not among district hospital patients (0.3% before vs. 0.5% after, p=0.24) (Tables 3 and 4). Mortality did not change for the overall patient cohort at either hospital level. But there were decreases in mortality among seriously injured patients (ISS≥9) at both district (15% vs. 8%, p=0.04) and regional (23% vs. 7%, p=0.004) hospitals (Tables 3 and 4).

Discussion

We sought to understand how KPIs of trauma care differed between district and regional hospitals and to determine whether the TIF could improve care at both hospital levels. The study showed higher KPI achievement at regional hospitals at baseline. TIF had a strong effect, with increases in many KPIs, and with decreased mortality among seriously injured patients at both levels. KPI achievement remained higher at regional hospitals after TIF, indicating room for improvement at district hospitals.

This study provides evidence on the effectiveness of checklists and other memory prompts to improve trauma care, for which evidence is scarce in LMICs. The WHO Trauma Care Checklist (WHO TCC) improved KPIs for trauma care and decreased mortality for severely injured patients at tertiary hospitals in nine countries, mostly LMICs [3]. The current study modified the WHO TCC to make it more applicable at non-tertiary hospitals. This included building memory prompts into medical records for use by individual providers rather than trauma teams, as was intended for WHO TCC. Trauma teams are rarely available in non-tertiary hospitals in LMICs and care is more commonly provided by smaller numbers or even individual providers.

The TIF improved KPIs of care at both levels. Notably, all KPIs were achieved for over 90% of patients at regional hospitals after TIF. Hence, simple interventions like TIF can bring hospitals that are already doing moderately well to high levels of functioning, despite considerable resource restrictions. Improvements at district hospitals were also notable. In the before period, many KPIs were achieved in only 10-30% of district hospital patients.

Many KPIs did improve (e.g. pulse oximetry, abdominal evaluation), though mostly below 90%.

The differences between district and regional hospitals are somewhat expected, given the higher number of specialists at regional hospitals. Nonetheless, most initial trauma care at both levels is provided by nurses and junior general doctors. Specialists mainly get involved after initial care. Also, other than pulse oximetry, resources required to achieve KPIs are minimal and are mainly physical examination. In theory, there is no reason why initial trauma care could not be delivered equally well at both levels. Resource restrictions may impair overall care, however, and thus may deter care providers from performing and documenting KPIs. Despite the substantial gains with TIF at district hospitals, further work is needed. Future improvements could include reinforcing use of TIF or similar memory prompts. The more complete achievement of KPIs at regional hospitals after TIF may be explained by the higher percent implementation of TIF (99%) at these hospitals compared to district hospitals (78%).

This study also focuses attention on second-level hospitals as key components of LMIC trauma systems. Trauma and surgical care have been relatively under-emphasized and under-studied at this intermediate level [21]. The global surgery movement has emphasized district (first-level) hospitals [22, 23]. Most trauma research in most countries derives from tertiary hospitals. Prior research showed that only 9% of all operations performed in Ghana were performed at regional hospitals [24], indicating they are relatively under-utilized. Given that the simple TIF intervention brought KPIs at regional hospitals to above 90%, the current study demonstrates the high quality of care that can be delivered at this level and indicates that further developing and promoting this level is warranted.

This study has several limitations. First, there may have been a Hawthorne effect with the introduction of observers. This would likely be the same at both district and regional hospitals and would not influence comparisons of these levels. Second, ISS is likely under-assessed at LMIC hospitals compared with high-income countries, given lesser availability of imaging. This might affect ISS assessment between district and regional hospitals, but is unlikely to affect ISS assessment in the before vs. after periods within each hospital level. Third, mortality was likely under-counted as we only detected mortality for patients who died at study hospitals and were not able to obtain information on survival of referred patients. This would under-count mortality more at district hospitals where there were more referrals. However, it would not affect the finding of before vs. after mortality reduction, as referral percentages were similar in before vs. after periods within each hospital level. Fourth, complications were based on information recorded in medical records and were likely under-assessed.

Conclusions

This study demonstrated the effectiveness of a standardized trauma intake form to improve initial trauma care for all patients and to lower mortality among seriously injured patients at both district and regional hospitals. At regional hospitals, all KPIs of trauma care were achieved in over 90% of patients after the intervention. KPIs were achieved to a lesser extent at district hospitals, both before and after the intervention, indicating the need for further actions to strengthen care at this level. Similar interventions using memory prompts to improve initial trauma care should be promoted widely at similar LMIC non-tertiary hospitals.