Going to the Nearest Hospital vs. Designated Trauma Centre for Road Traffic Crashes: Estimating the time difference in Delhi, India

Abstract

Background:

Time to hospital after a road traffic crash (RTC) plays a vital role in determining the outcome for crash victims. In Delhi, there are 7 designated trauma centres where crash victims are typically taken, which may not be nearest hospital. We compare the transport time access (crash to hospital) depending on whether the victim is transported to a designated trauma centre or the nearest hospital.

Data and Methods:

For each RTC geocoded manually from police records, the nearest hospital and the designated trauma centre is identified using Google Maps places nearby Search API and guidelines. Travel time matrix is generated between RTC’s and identified hospitals using Google maps distance matrix API. Index accounting inter-district differences is developed.

Results and Conclusions:

The network of designated trauma centres in New Delhi are located such that they can be accessed within 45 minutes of most crashes while nearest hospital within 30 minutes. As a result the vast majority of crash victims are likely to receive timely care if they are rapidly transferred to either of these caregivers. However, for the most severely injured and time-sensitive cases, bypassing nearest hospital for trauma care, could substantially improve survival outcomes.

1. Introduction

Trauma systems for providing care to road traffic crash (RTC) victims are under-developed in India. The process of forming a robust care system for injury prevention, pre-hospital & acute hospital care and rehabilitation units to diminish the risk of permanent disability within demographic profile is still underway (Pal et al., 2014). The Planning Commission of India estimates the cost of injury to Indian society as 3% of GDP (Kumar Gupta et al., 2011). The prime concern after a RTC is to reduce the impact of injury by reaching appropriate health care facility as early as possible. An appropriate health care facility could be a hospital, defined as geographically fixed facility with acceptably trained personnel to deliver emergency medical care or it could be a specialized trauma care unit equipped with appropriate and specialized care facilities to deal with severe trauma injury which would require timely diagnosis and treatment by a multidisciplinary team of health care professionals (Kobusingye et al., 2006; Kumar Gupta et al., 2011).

Many studies have shown that the duration of time between a crash and definitive care are important for survival of victims. Observations from a Swedish study on survivability in road traffic crashes show that, from the group that sustained survivable injuries, about 12% of crash victims could have survived if they had reached the hospital earlier but reaching a “trauma care unit” would have been better, saving an additional 20% which means, 32% of survivable injuries could be saved if they were taken to a trauma care unit quickly (European Commision, 2009).The most efficient emergency services take about 30–45 minutes to take the victim from crash location to the hospital in non-urban areas (Coats and Davies, 2002; Bigdeli, Khorasani-Zavareh and Mohammadi, 2010). Timely arrival of emergency services, followed by quick transport is crucial for an efficient emergency system(Lerner, 2001; Roy et al., 2010; Radjou, Mahajan and Baliga, 2013). The first 60 minutes after the crash is generally termed as the golden hour. While the golden hour is not the perfect framework but an internationally accepted framework to highlight the importance of time. A significant amount of the literature claims that time is an important determinant of the outcomes of the road crash victims. Starting from “Critical 4 min”, during which if airway is blocked and not immediately cleared may lead to death. Next is “Platinum 10 min”, which is recommended maximum on-scene stabilization time for seriously injured persons (Rogers, Rittenhouse and Gross, 2015). For total time to receive definitive care, it is believed that if proper first aid is given to crash victim in first one hour, their chance of survival increases. Some studies endorse first 60 minutes, claimed as “Golden Hour” as the acceptable time window to access adequate treatment for accident victim while some claim its origins to be hoax (Gopala krishnan 2012; Lerner & Moscati 2001). There is still an ongoing debate on the adequate time for appropriate treatment, while international standards tend to highlight importance of time to access hospital care in terms of golden hour. Even though, the literature has substantial use of golden hour terminology, we are too far from generalizing pre-hospital time window to outcomes/survival (Rogers, Rittenhouse and Gross, 2015). Thus, while it is controversial whether the first hour after a traumatic event has a special status as a time threshold, it is widely understood that rapid transfer to a medical facility improves outcomes.

Measures to reduce travel time to hospital care after a road traffic crash include better pre hospital transportation services, and the provision of a higher number of healthcare units that are geographically distributed for effective spatial coverage (Coats and Davies, 2002; Bigdeli, Khorasani-Zavareh and Mohammadi, 2010). The total time to medical care is a combination of several time intervals: notification time, activation time, response time, on scene time and transport time. Transport time is only one component (Coats and Davies, 2002). In formal emergency response systems, components other than transport time might have a huge impact on the total time to reach hospital care (Bigdeli, Khorasani-Zavareh and Mohammadi, 2010; Corrado et al., 2017; Roy, 2017). A research study based in Iran points out that scene to hospital transport time are significantly longer for interurban incidents. The authors stress the need for more research on local needs and accessibility for city based interurban roads and allied locations (Bigdeli, Khorasani-Zavareh and Mohammadi, 2010). In India, the national trauma care capacity building guidelines recommend a level I designated trauma care unit within 750 to 800 km from locations of high crash frequency (Government of India. Ministry of Health and Family Welfare, 2015).

Globally, pre-hospital care systems either emphasize a “scoop-and-run” approach, where the goal is to rapidly transport the victim to the hospital, or “stay-and-stabilize”, which emphasizes the need to perform stabilizing interventions such as intubation and placement of intravenous access lines for fluid replacement prior to transportation of the victim. A growing literature suggests that scoop-and-run has better outcomes because performing invasive interventions in the field is difficult and results in delays(Haas & Nathens 2008; Varghese,M. 2016; Jayaraman et al. 2009). Regardless, in LMIC settings that do not have an established infrastructure to support advanced life support (ALS) in the pre-hospital setting, scoop-and-run is the only option, and minimizing time to hospital is a key strategy for improving outcomes. Currently, in India people tend to scoop and run the crash victims to nearest hospital using whatever vehicle is available instantly. In fact, scoop and run is one of the most recommended approaches in emergency transportation. In typical urban settings of urban environment, with relatively short transport times, there isn’t much evidence available to support field ALS while some suggestion of harm exists. In very rare cases ALS acts as lifesaving, but the rarity of such events and the effort required to maintain the competence is quite disproportionate. The support towards complex EMS systems has emerged with a hope for its effectiveness without much evidence (Sasser et al. 2006; Varghese,M. 2016; Nagata et al. 2011; Morrison 2015; Sanghavi et al. 2015; Haas & Nathens 2008).

Besides minimizing transport time, other key issues that affect outcome include training of first responders, and the choice of medical facility to which the victim should be transported. Providing basic training to citizens and police personnel in India still remains a disconnected link in the current structure of post-crash management which is generally taken care by fragmented bodies such as non-government organizations and others alike (Posaw, Aggarwal and Bernstein, 1998; Pal et al., 2014). A parallel discussion prevails in pre-hospital care and road safety science, which tends to differentiate between decision to take the road crash victim to appropriate/definitive care or to the nearest hospital. Effect of pre-hospital interventions or by other hospitals on life-threatening events, prior to trauma care was studied in Portugal. It revealed decreased mortality of trauma patients even after extended time to trauma care (Gomes et al., 2010).

Selection of best hospital after road traffic crash has been highly debatable. Should it be level I trauma centre or the nearest community hospital or level III/IV trauma care unit is still under discussion. Spatial factors like distance from crash location to the nearest local hospital or tertiary care trauma centre have different individual challenges in each region. Triaging recommendations given to formal Emergency Medical Services (EMS) is to identify the best hospital as per the condition of the victim and not merely take them to the nearest hospital. With proper triaging and ambulatory services, about 1/3rd lives could be saved in a year (Kobusingye et al., 2006). In India, injury victims generally end up in government hospitals for emergency care with reasonably good acute care facilities are provided by tertiary care teaching hospitals. Other institutions giving medical care include private hospitals, private medical practitioners and institutions (Pal et al., 2014).

In UK, protocol is to take patients with ISS > 15 to Major Trauma Centre (MTC), while it is observed that patients with less severe injuries (ISS≤15) may be de-prioritized compared to the major trauma patients for operations, rehabilitation resources, etc and would actually be harmed if taken to trauma centre. But deciding ISS on the field is nearly impossible, so following these guidelines stringently would then become another task for agencies to implement(Hyde et al., 2012). In Australia, no clear guidelines are defined. Trauma care is given priority over general emergency care for poly trauma cases while observations show around 20% of all trauma cases are taken to a hospital and then transferred to trauma care, which would be delayed by about 19 hours in most cases. Australian Major Trauma Care National Clinical Network is working to bring down these travel times by developing more specialized care centres and by assistance from the transport sector by acting to avoid congested travel conditions through active cooperation between the authorities, Road Policing and prioritizing emergency services to minimize any disruption to travel times related to the crash(Frith et al., 2018). WHO recommends to take injured person as soon as possible to a hospital with the right equipment and personnel to provide the needed care avoiding transfers between hospital categories. Guidelines highlight the importance of clear protocols of triaging by matching injury severity to destinations but fail to comment on their true nature(World Health Organization and Publications, 2016).In developed countries like US, regionalized civilian trauma networks have been operational since late 19^th century(Grossman, 2009). In Casablanca regionalized trauma care system is newly being developed and is in construction stage(Organization of Islamic Cooperation (COMCEC), 2016). German system comprises of 110 Level I or supraregional trauma centers believed to cater to 50% of most severely injured patients and about 200 regional trauma centers (Level II)(Oestern, Garg and Kotwal, 2013). In OIC countries, developing regional trauma centres has been associated with improving post-crash response, systems are being established and reaching directly to trauma care is becoming priority(Organization of Islamic Cooperation (COMCEC), 2016). In Sub Saharan Africa, district hospitals are directed to transfer complex and critical patients to the 4 tertiary facilities in the country(Boschini et al., 2017).

Studies recommend that when transport times are higher than 60 minutes, patients would benefit from visiting nearest hospital, however when its less than 30 minutes, its beneficial to bypass smaller hospital and visit definitive care facility (Harrington et al., 2005). A study based in Rhode Island with one designated trauma centre mentions transfer protocol to trauma centre if transportation time is less than 20 minutes and to the nearest medical facility if patients injured are beyond 20 minutes travel time from the trauma centre. Their results show that 91% of the patients who were transported directly from the scene to the trauma centre reached within 20 minutes and 96.1% within 25 minutes (Harrington et al., 2005).

While the final verdict/best practice is still undecided among academic community and policy makers, certain guidelines reflect this distributed view on above issue. Perhaps, few emergency care services are purposefully directed to take the victims to the appropriate designated healthcare facility/trauma centre (Harrington et al., 2005). Another advisory is to take the road crash victim to the nearby hospitals to be able to stabilize the patient earlier and before being transferred to higher level facilities for appropriate care (Harrington et al., 2005; World Health Organization, 2015). As per Supreme Court guidelines, India, no hospital and medical facility can deny treatment to a road crash victim. These guidelines recommend taking road crash victims to the nearest hospital. Under these directives, no hospital whatsoever can deny admission in emergency cases of road accident and women in labour. They are bound to provide primary treatment to emergency cases under current legal framework. Many victims end up going to the nearest hospital, however nearest hospital may not have the requisite facilities and they tend to get transferred to nearest designated trauma centre (Radjou, Mahajan and Baliga, 2013; Roy, 2017). We understand that if the patients were to go directly to the designated trauma centre which is equipped with facilities to deal with accident related serious injury, it might save some time of their overall recovery. As previously mentioned, poly-trauma cases need specialized care and the crash locations considered in this study are representative of such severe crashes with each having at least one fatality during crash. These cases need appropriate care as early as possible.

Post-crash access systems are rudimentary in LMIC’s like India with no significant policies being followed in relation to transportation of the injured. The access to care after road traffic crash thus becomes dependent on the informal sources of help like bystanders, relatives, police etc. The protocols defined by government agencies to take the victims to appropriate care providers i.e. designated trauma centres thus remains underserviced by general population but mostly followed by formalized services. The article aims to understand, the differential access to trauma care vs hospital care assuming that the victims are able to immediately receive transportation help. The article could aid in prioritizing policies for accessing trauma care or nearest hospital. We compare how access to trauma centre would fare as compared to nearest hospital and assess spatial disparity in regionalized trauma care system by developing dissimilarity index.

a). Objectives

In this article, we develop a methodology to evaluate post-crash access and apply it to Delhi, India. We hereby compare post-crash access time to care, i.e. time taken to reach hospital care vs trauma care from crash scene .This study focuses solely on time from the crash site to the hospital. Other components of time (i.e. calling for help, arrival of help and on scene time) are independent of hospital choice and, therefore, are not considered in this study.

The study objectives include:

• To estimate the travel time to the designated trauma centre and geographically nearest hospital from each crash location in New Delhi.

• To evaluate district-level variation in travel time to designated trauma centre and geographically nearest hospital from road traffic crash.

• To estimate the time lapse for reaching designated trauma centre instead of geographically nearest hospital after a road traffic crash.

• To formulate dissimilarity index to compare spatial access to care.

2. Data Collection and Methodology

a). Identification of road traffic crash locations

Police records have been identified as the best source of information on road traffic crashes in LMIC’s including India. While these crashes underrepresent non-fatal injury data, they quite adequately capture fatalities in cities (Pal et al., 2014; Mohan, Tiwari and Bhalla, 2015). Prior studies addressing the issue of accessibility to trauma care primarily use census districts or residential blocks as the proxy of incident locations, however, it has been observed that only 63.5% of injuries occur near residential locations(Apparicio et al., 2008; Gao et al., 2016). This incites us to look for better alternatives for identifying locations of injured. In this article we have resorted to using actual location of crash, which is obtained from police first information report for road traffic crashes which contains description of the crash location.The FIR’s have been collected for four years (from January, 2013 to December, 2016) for road traffic crashes and each location has been read and manually geocoded to obtain nearly exact road traffic crash location. The records consisted of 6738 fatal crash location addresses. Each address was geocoded manually by searching for address in Google Maps to avoid any discrepancy in location of incident. Out of these 1167 locations could not be found exactly as per given address, these locations were plotted by approximately locating crash on nearest location on road. Due to their discrepancies in recording format of addresses, it was difficult to use the automated methods of finding geo-locations, which was initially experimented with and later given up owing to inefficient results. The identified geo-locations were plotted on ArcGIS and overlapping locations were collated. Certain locations appeared to lie outside Delhi boundary, they were discarded. This led to discarding of 200 crash records. Further 200 records were discarded after identifying nearest hospital, as results returned were missing values of either travel time or distance or both, even after three iterations of fetching results for these location pairs.

b). Google Maps API

Google Maps provide a range of Application Programming Interface (API) which enable the users to fetch data from Google’s extensive database, within terms and conditions. The functionality enables users to determine the nearby places, calculate distance and approximate time to reach a specified destination point from a specified location. The command to get results requires a key, which needs billing details as per Google’s revised policies, June 2018. One key can be obtained per project, Google supports up to 10 projects per user. Each key has restricted usage of up to 2500 calls for free users. The limit varies as per the API called. The results returned are in JSON/XML format which need to be stored in required formats. For instance, a sample call of the Google API to fetch restaurant within 1500m radius of a certain address- Latitude and Longitude is: https://maps.googleapis.com/maps/api/place/nearbysearch/json?location=−33.8670522,151.1957362&radius=1500&type=restaurant&keyword=cruise&key=YOUR_API_KEY (Munir and Omair, 2015).

c). Identification of nearest hospitals and trauma centres

There are two types of health facilities included in the study: designated trauma centres and nearest hospitals.

Delhi is the capital city of India with about 1800 fatal crashes per year (Police, 2018). Due to its vast extent, it has been subdivided into nine districts, namely North, South, East, West, Central, New Delhi, South West, North West and North East (Registrar General of India. New Delhi, 2011). To manage the burden of road traffic crashes, the government of Delhi specifies a regional trauma care facility for each district within Delhi (Table 2). The guidelines suggest government services and citizens to visit these centres of care in case of medical emergencies (District wise list of hospitals, 2018). The trauma centre and district boundaries are shown in Figure 2. The identified definitive care units are taken as the trauma centre destinations for each district.

Table 2:

District wise area and their respective designated trauma centre

District	Area (km2)	Designated Trauma Centre
North-West	445	Baba Saheb Ambedkar Hospital
South-West	436	Safdarjung Hospital
South	255	Jai Prakash Narayan Apex Trauma Centre (JPNATC)
West	123	Deen Dayal Upadhyay Hospital (DDU)
North	69.3	Sushruta Trauma Centre
East	65.5	Guru Teg Bahadur Hospital (GTB)
North-East	62.8	Guru Teg Bahadur Hospital (GTB)
New Delhi	38.6	Jai Prakash Narayan Apex Trauma Centre (JPNATC)
Central	15.2	Ram Manohar Lohia Hospital (RML)

Figure 2:

District Map of Delhi

Nearest hospital in this study has been identified as the geographically nearest hospital to the crash location. Google maps API identifies these facilities with 90% accuracy in this study (Table1). Rest 10% cases are either health facilities other than hospitals or hospitals like specialized maternity care units, ayurveda centres, veterinary hospitals, dental and eye centres or paediatric care units. In cases, where the trauma centre might be the nearest hospital to the crash location, they would overlap in the travel time result estimates.

Table 1:

Descriptive analysis of the nearest hospitals obtained from Google Maps API

Type of Facility	Count	%
Hospital	557	90.13
Other	29	4.69
Clinic	14	2.27
Eye Hospital	8	1.29
Maternity Care	7	1.13
Paediatric Hospital	3	0.49
Total	618	100

The nearest hospital to all identified crash locations were identified using Google Maps Places API (Google, Mountain View, California, USA). The data was fetched with keyword = hospital and type = hospital. To automate and expedite the process of getting the travel time estimates, a project in Java was developed, which was further run in Eclipse Jee Oxygen. The standard version of API gives basic results in the form of name of the hospital, place id and geometry (Latitude and Longitude). These fields were recorded and matrix representing nearest hospital was generated for all the crash locations. API returns nearest hospital ranked by distance, of which the nearest hospital name, its place id and geometry (Latitude-Longitude) were extracted from the results into a Microsoft Excel spreadsheet.

d). O-D travel time matrix

Taking crash location as the origin and nearest hospital/ trauma centre as the destination, travel time and distance was found for each pair using Google Maps Distance Matrix API (Google, Mountain View, California, USA). Both the API were embedded in the Java project.

As discussed, the literature identifies the difference in opinion on nearest versus appropriate healthcare facility selection in case of health emergency. To estimate the actual travel time differences between trauma centre and nearest facility, the Google Maps Distance Matrix API (Google, Mountain View, California, USA) is used.

The basis of travel time estimation employed in this study is Google Maps API. Effect of congestion is automatically accounted as google returns travel time on network in real-time traffic. Google has a well-defined framework which is followed to estimate the travel time, the accuracy of the estimated results is based on predictive analytics on geo-tracking of their real time users which has been lately established to be fairly accurate. Studies assessed the accuracy of three estimation methods (linear arc distance, Google maps, and ArcGIS network analyst) against observed transport times in a large cohort of pre-hospital patient transports, they found that of these three methods, Google maps API results were closest to the actual results for 86.6% of the cases while other methods were 79%(linear arc estimates) and 81.3%(ArcGIS estimates) accurate(Wallace et al., 2014). We have tried our best to control for consistency in the time of calling the API. The results obtained have been repeated for a week and averaged out to obtain the final dataset used for analysis. It shows a representative trend towards the travel time access to the health care facilities(Shaw et al., 2017). There might be slight differences in the travel time estimated in a particular duration owing to variation in the time of fetching the results, which has been diminished by averaging out the results.

To find the reproducibility of these results we tried to find out the coverage of google maps API in all the LMIC’s. World bank’s list of LMIC’s(World bank, 2019) is compared with the countries where google maps API is available for its all attributes (Google Maps, 2019) and it was found that out of 138 LMIC’s , google maps API is not available in 22 countries. For other countries, Google Maps API is operational for geocoding but traffic layer is not available in following 10 countries. Thus, these methods have a replication possibility in at least 106 LMIC’s some degree of reliability in these countries as mentioned in previous section. The comparison of access differential of travel times does not get affected much due to their simultaneous measurement for both nearest hospital and trauma centre.

e). Dissimilarity Index

An index is developed to assess the differences among districts with respect to golden hour, called Dissimilarity Index. In terms guide of proportions of crashes and their access time, most districts seem comparable. However to develop a method to understand if there is any substantial difference among the spatial access of districts in terms of travel time to trauma centre or nearest hospital, we formulated a Dissimilarity Index. International standards identify 60 minutes as the time within which if victims reach a hospital, their chances of survival and recovery are high. To assess the difference in regionalized care among the districts, an index is formulated with respect to 60 minutes.

$$DissimilarityIndexwithrespecttofirst60minutesaftercrash=(\sum (x_{ij}/60)/N)$$ (1)

Where, x_ij = travel time from crash locations to the assigned destination (trauma centre or nearest hospital); N = total number of crash locations; lesser the value of this index shows better overall access to hospitals with respect to 60 minutes for that district

3. Results

a). RTC’s distribution per district in Delhi

Overall there are 6738 fatal RTC’s reported in Delhi from Jan 2013 to Dec 2016. Out of these only 6338 could be accurately geo-located. In each of nine districts in Delhi: North, South, East, West, Central, New Delhi, South West, North West and North East, road crash burden aggregated for four years is 707, 930, 647, 1057, 269, 361, 1002, 999, 366. While South, West, South West, North West have highest fatalities; North East, New Delhi and Central Delhi are the least affected by fatalities in road crashes (Figure 3).

Figure 3:

District wise distribution of road traffic crashes

b). Travel time estimates to trauma centre

In the nine districts, there are seven trauma centres as identified by Delhi government, the plot for trauma centre and district boundaries are shown in (Figure 2). Considering each crash location in each district as origin and designated trauma centre as the destination, travel time was estimated using Google Distance Matrix API. The estimated results (Table 3, Figure 4) show that 2.7 % of RTC victims are within 0 to 5 minutes from the designated trauma centre, 20% are within 5 to 15 minutes, 45.3 % are within 15 to 30 minutes, 22.3% are within 30 to 45 minutes, 10.3% are within 45 to 60 minutes, while only 2% are beyond 60 minutes. It is observed that while in North West, East, North and Central districts, about 70 to 80% of road crash locations are within 30 minutes distance, for West, Southwest, Northeast and New Delhi, these proportions fall down to 35 to 55 %.

Table 3:

Percentage Travel time distribution of RTCs to Trauma Centre (Tc) and Nearest Hospital (Nh) and difference (Tc-Nh)

Time In Min		< 0	0.1–5	5.1–15	15.1–30	30.1–45	45.1–60	>60
	*Tc	0	1.2	18.3	32.3	20.6	27.2	0.5
West	**Nh	0	44.6	50.7	4.6	0.1	0	0
	***Diff	1	5.8	26.1	27.5	31	8.4	0.1
South West	Tc	0	1.2	10.8	37	28.3	13.4	9.3
	Nh	0	28.9	58.5	12.3	0.3	0	0
	Diff	2	11.5	27.4	25	20.1	8.8	5.4
South	Tc	0	0	2	35.6	48.7	11.2	2.5
	Nh	0	36.7	55.6	7.6	0.1	0	0
	Diff	2.2	2.3	11	51.2	26.7	6.8	0
North West	Tc	0	2.9	17.2	58.8	19.5	0.9	0.7
	Nh	0	44	51.4	4.6	0	0	0
	Diff	1.5	8.9	39.9	40.5	8.2	0.7	0.2
North East	Tc	0	0.8	38.8	22.7	30.3	7.1	0.3
	Nh	0	47	50.8	2.2	0	0	0
	Diff	4.6	15.6	34.7	13.9	28.7	2.2	0.3
North	Tc	0	12	43.2	40.5	4.4	0	0
	Nh	0	44.7	52.5	2.8	0	0	0
	Diff	20.3	19.8	36.4	23.4	0.1	0	0
New Delhi	Tc	0	0.3	27.2	33	14.7	24.7	0.3
	Nh	0	39.6	57.9	2.5	0	0	0
	Diff	8.3	4.4	42.4	9.4	27.7	7.5	0.3
East	Tc	0	3.4	20.6	64.6	11	0.3	0.2
	Nh	0	45.8	48.5	5.7	0	0	0
	Diff	2.3	10.5	45.4	40.7	0.9	0	0.2
Central	Tc	0	3	36.1	61	0	0	0
	Nh	0	44.2	51.3	4.5	0	0	0
	Diff	2.2	13.8	68.8	15.2	0	0	0
Overall	Tc	0	2.7	20	42.6	22.3	10.3	2.1
	Nh	0	40.8	53.2	5.9	0.1	0	0
	Diff	4.4	9.5	32.6	31.2	16.9	4.5	1

^*Tc: Trauma Centre

^**Nh: Nearest Hospital,

^***Diff: Difference

Figure 4:

Proportion of RTC’s within travel time (in minutes) to Trauma Centre (Tc) and Nearest Hospital (Nh) and Difference (Tc-Nh)

c). Travel time estimates to nearest hospital

Considering each crash location in each district as origin and the identified nearest hospital as the destination, travel time was further estimated using Google Distance Matrix API. The identified nearest hospitals were checked for their appropriateness by manually searching each unit and finding out if it actually is a hospital or other type of facility. About 90% of these are general medicine hospitals and other 10% are other types of hospitals and facilities (Table 4). The spatial distribution of these identified hospitals is plotted (Figure 5). The overlapping hospitals were collated and shown as single unit in spatial representation (Figure 5). The estimated results (Table 3, Figure 4) show that 40.8 % of RTC victims are within 0 to 5 minutes from the nearest hospital, 53.2% are within 5 to 15 minutes, 5.9 % are within 15 to 30 minutes, 0.1% are within 30 to 45 minutes. Interestingly, within Delhi, none of the fatal crash locations in the past four years was more than 45 minutes from a hospital. A hospital presence was observed within a 30 minute drive from the crash location for about 70 to 90% of the cases. This reveals that given the road crash victims have access to some form of mode of transportation after crash, the possibility to reach a nearby hospital within 30 minutes is quite high.

Table 4:

Dissimilarity Index between districts in Delhi

Dissimilarity Index
Time	Overall Index =Sum/N	West	South West	South	North West	North East	North	New Delhi	East	Central
#tcgh	0.45	0.52	0.57	0.57	0.39	0.42	0.25	0.47	0.36	0.28
##nhgh	0.12	0.11	0.15	0.13	0.11	0.10	0.11	0.12	0.12	0.12
###diff_gh	0.33	0.41	0.42	0.44	0.28	0.32	0.14	0.36	0.24	0.17

^#tcgh: Trauma centre with respect to Golden hour

^##nhgh: Nearest hospital with respect to Golden hour

^###diff_gh: (Trauma centre minus nearest hospital) with respect to Golden hour

Figure 5:

Spatial Distribution of Identified Nearest Hospitals for Crash Locations

d). Travel time estimate comparison between trauma centre and nearest hospital

One of the objectives of this study is to develop methodology to assess policy of bypassing nearest hospital for appropriate care in trauma centres. To find this, the travel time for trauma centre as the destination and nearest hospital as the destination was subtracted for each crash and the proportions of cases that vary within the time difference slabs were identified for each district separately and combined for an overall estimate. It was observed that (Table 3) 9.5% cases are within 5 minutes difference from a trauma centre versus nearest hospital. For 32.6% cases this difference ranged between 5 to 15 minutes, for 31.2% between 15 to 30 minutes, for 16.9% between 30 to 45 minutes. For about 5% of the cases this difference was beyond 45 minutes, where accessing the nearest hospital could be a better choice

e). Travel time estimation and its dependence on mode of transport

In the above results it appears that most crash locations are accessible to a hospital within first 60 minutes. However, this is applicable only if the mode of transport is available right at the time of crash and at the site of crash. In most countries, it is observed that road crash victims are brought by non-ambulatory transportation modes such as taxis, three wheelers (autos), private vehicles etc (Demetriades et al., 2005; Prakashy et al., 2013). In Delhi, India, recent estimates from an urban university hospital observe that of the 34.4% patients had come directly to the hospital after a road traffic crash, came by regular taxis/auto rickshaws-13.5%, private cars-12.4% and police vans-9.3% are the predominant modes of transport used after a road traffic crash (Roy, 2017). Considering that vehicles like three wheelers (autos), taxis are available on site or some public person takes the initiative to help in his private vehicle, these estimates might then fare well amid international standard of 60 minutes. If otherwise, help needs to be called for in the form of ambulance/police or relatives, then we need to account time taken for someone to initiate the call for help and then the response time taken by the help to reach the site. These estimates then might considerably change depending on the availability and response of the called modes of transport for transportation of road crash victims.

f). Dissimilarity Index

Lesser the value of this index shows better overall access to hospitals with respect to 60 minutes for that district (Table 4). The values of index are relatively higher for trauma centre as the destination. Two districts: Central (0.28) and North (0.24) fare better and are comparable. They are closely followed by East (0.35) and North West (0.38). The other districts such as West, South West and North West have higher values showing comparably higher travel time to reach trauma care. As far as nearest hospitals are concerned, index shows that almost all the districts are consistently comparable. Although, mere presence of hospitals may not prove their competence to treat or handle road traffic crash victims (Government of India. Ministry of Health and Family Welfare, 2015).

4. Discussion

a). Travel time estimates to trauma centre

Accounting one trauma centre at a time, it was observed that in North West, East, North and Central districts, about 70 to 80% of road crash locations are within 30 minutes distance, for West, Southwest, Northeast and New Delhi, these values fall down to 35 to 55%. Although, these trauma centres have been designated to cater to a certain region of Delhi, there is no hard evidence which establishes this direction. There is a possibility of distribution of patient load among these trauma centres as well, which is fairly indicated by these numbers which are on lower side as compared to regions with only one trauma centre catering to the population. Also these variations in locations and their reach can explain spatial variations in traffic mortality rates in areas with different emergency services accessibility, to signify this, further investigation would be required (Bentham, 1986; Durkin et al., 2005; Zwerling et al., 2005; Li et al., 2008; Sánchez-Mangas et al., 2010).

b). Travel time estimates to nearest hospital

Having a standardized system of centre designation for injury care as per injury burden of a region could aid in reducing its effects. In our estimated results about 99% lie within 30 minutes’ drive from a nearby hospital. Few other results estimate benefits for patients in visiting nearest hospital if the travel time to trauma care exceeds 30 minutes (Bentham, 1986; Durkin et al., 2005; Harrington et al., 2005; Zwerling et al., 2005; Li et al., 2008; Sánchez-Mangas et al., 2010). This might instinctively drive us to recommend the trauma centre as the ideal destination for all the crashes but we need to carefully account for other factors like the resources, capacity and existing load on individual trauma centre before making this as policy decision. Current trauma care guidelines do not give clear protocols for transportation of injured in Delhi. Better triaging systems and clear protocols for redirecting to proper care facilities allows for effective use of limited resources (Sasser et al., 2005; Cameron et al., 2008; Government of India. Ministry of Health and Family Welfare, 2015). Setting up this type of system in low and middle income countries like India could help in management of trauma care systems. For countries lacking formal pre-hospital care systems, simple steps like issuing advice lines and providing layperson education programmes could help in establishing better community wide triaging systems (World Health Organization and Publications, 2016).

c). Travel time estimate comparison between trauma centre and nearest hospital

For about 5% of the cases, this difference in travel time between trauma centre and nearest hospital was beyond 45 minutes, where accessing the nearest hospital could be a sensible choice. Observations show survival of 32% of survivable injuries if the primary destination is the trauma centre (European Commision, 2009). Nearest hospital versus trauma centre though in theory seems insightful, in practice it might be difficult to identify the nearest hospital and nearest trauma centre for a layperson transport to the hospital if digital aid or locations are not known to the people who end up transporting the victims to the hospitals. In a similar study of patient transportation by Chicago Fire Department (CFD) the travel time from the scene to the hospital (transport time) also was three minutes longer in the group with chose to bypass nearest hospital (7 ± 3 vs. 4 ± 2 minutes, P < .005), they infer that the urban use of hospital bypass would not reduce trauma patient survival in those who arrive at the trauma centre with serious injury (Sloan et al., 1989). International guidelines by developed and developing countries such as Australia, US, Europe, New Zealand, South Africa, Casablanca, Germany are either already in place or moving towards trauma care infrastructures and assigning specialized care to critical patients without transfers. These guidelines base themselves on defining injury severity, which is very difficult in case of poly trauma cases of road traffic injuries, as many times severe injuries are internal and are cannot be visibly judged. If such polices are recommended further we might save on transfer time or rather patients might actually loose important minutes to access care which could be crucial to their outcomes.

d). Dissimilarity Index

Absence of an appropriate triaging and first responder system in most low and middle income countries like India, lead to regional differences of patient distribution across health care facilities. Studies indicate that gaps within systems and inefficiencies of different stakeholders and care givers leads to death of about a third patients and only about a fifth are able to receive medical care within first 60 minutes after road crash (Bigdeli, Khorasani-Zavareh and Mohammadi, 2010; Newgard et al., 2010; Clark, Winchell and Betensky, 2013; Dinh et al., 2013; Vanderschuren and McKune, 2015). To find the regional difference between trauma care districts, this study estimates an index with respect to first 60 minutes after road crash. In our estimates, index values are relatively higher for cases destined to trauma centre. As far as nearest hospitals are concerned, index shows that almost all the districts are consistently comparable. The estimation does show some mild differences between these districts. Comparable districts being Central (0.28) and North (0.24), closely followed by East (0.35) and North West (0.38). Then there are other districts such as West, South West and North West which have higher values showing comparably higher travel time to reach trauma care. The distinction could be attributable to differences within districts. Delhi, being a million plus city with extensive unaccounted burden of road traffic injuries, it is difficult to completely rely on this index for addressing spatial dissonance in care infrastructure (Joshipura et al., 2003; Roy et al., 2016).The current data recording system is incapable of recording this distribution of patients across different healthcare facilities (Fitzgerald et al., 2006; Kumar et al., 2008).In light of absence of any data to check the accountability of this result, this index gives us a wise estimate and a starting point for system improvement.

5. Conclusions

The differential accessibility of hospitals and trauma centre in Delhi may be useful to Delhi policy analysts for strengthening existing healthcare system. Also, the method for evaluating regional trauma care systems and calculating physical accessibility has been described and demonstrated to be useful. As the policies in LMIC’s are moving towards regionalized trauma care systems, these methods could be applied to other countries that have minimum information of crash locations, as other inputs like hospital detection and travel time evaluation are based on open source methods.

On the surface, there appears to be adequacy of hospitals and trauma centres in Delhi based on travel time access. Spatial disparity of travel time access (Dissimilarity Index) to care is observed between individual districts, suggesting starting point for systemic improvements. Within the state, 90% of the crash locations could access trauma care within 45 minutes. Further investigation needs to be done for individual trauma centre’s capacity to handle the overall daily inflow of trauma patients, which might induce the need to strengthen other hospitals to manage trauma cases. 99.97% have a nearby hospital within 30 minutes of travel time. The study could not account for available pre-hospital facilities for transportation, which would further affect the overall travel time. Also, mere presence of hospitals might not signify their capabilities to be able to treat these patients. There is a need of auditing the available facilities for their operational characteristics and rating the trauma centres for their true capabilities to deal with poly trauma cases.

The study delivers potentially useful information that can be used in health service delivery planning and assessment. Accessibility analysis using road travel times produces more accurate descriptions and offers better evaluation of the existing system. The study has a scope of replicability in other LMIC’s, with the knowledge of basic structure of the hospitals’ locations, policies available and crash location data.

6. Assumptions and Limitations

We are assuming that the trauma centre has the facilities for appropriate care and that the nearest hospitals has the ability to stabilize the victim but may or may not have the facilities needed for definitive care. Also, trauma care facilities would be capable of better triaging the victims, owing to their expertise and experience in dealing with trauma victims. Other major assumption, is that Google maps API is able to identify the nearest hospitals correctly, while this might have certain degree of error (Table 1).

The delay might be in the response time for the transportation service to arrive or for some help for transportation to be available on the crash location after a road traffic crash(Fitch, 2005).To be able to assess the true nature of total time to a hospital (including response time and travel time) need is to examine further, which is beyond the scope of current study.

Availability of mode of transport at the instant of crash is discretionary. It varies with many factors like socio-demographic profile, region, time of the day and general travel patterns, bystander behaviour etc. Full time availability of a vehicle to transport at the crash site was assumed in the study.

Figure 1:

Flowchart of study implementation