Reducing ambulance response times in Jakarta – Indonesia: Discrete Event Simulation

Abstract

Background: Ambulance services respond to emergency health situations and provide emergency medical care, stabilisation, and transport to definitive care. These services should be able to respond to an emergency health situation in an effective and timely manner to improve patient outcomes. Ambulance services in Jakarta – Indonesia, a lower-middle income country in South-east Asia, face many challenges such as vast geographical areas, traffic, inadequate numbers of ambulances, fragmented services, low public awareness, and unknown demand for the service. In fact, only 9% of patients attending hospital emergency departments in Jakarta use an ambulance as the mode of transport, and these patients have the highest transport response time compared to other modes of transport such as a private car, taxi, motorcycle or public transport. Ambulance response time in Jakarta is estimated at 37 minutes, which is way higher than the 10 minutes in the US or 7 minutes in Japan. There is therefore an imperative need to reduce ambulance response time in Jakarta in order to improve patient outcomes such as survival. One important way to reduce response time is to reduce travel time by positioning ambulance posts in neighbourhoods without a nearby ambulance post. Aim: The aim of this study was therefore to describe the demand for ambulance services in Jakarta – Indonesia, to describe the response to this demand, and to use a computer simulation of the system to assess whether creating new ambulance posts and reducing ambulance travel time will reduce ambulance response time. Methodology: The demand for ambulance services and the emergency response in Jakarta was simulated using Discrete Event Simulation (DES). DES is a simulation of stochastic processes by modelling a system whose state may change only at a discrete point in time. The ambulance service processes (discrete events) that were simulated include the arrival of a call, a decision to dispatch an ambulance for the call, the time the ambulance leaves the post, and the time the ambulance arrives at the site. The parameters that were used for the model were estimated from 2019 data that were obtained from the Emergency Ambulance Services 119 in Jakarta, and GIS data (geocode of ambulance posts and road travel distance from an ambulance post to a neighbourhood) that were curled from Google maps. The arrival of calls was modelled as a non-homogeneous Poisson process that depends on the time of the day, the day of the week, while the municipality of the call, the probability of dispatching an ambulance, the dispatch time, and the response time were modelled as empirical distributions from the 2019 data. After verifying the simulation model by comparing 100 runs of the simulation to the 2019 historical data, we simulated the effect of creating 21 new ambulance posts (in neighbourhoods that were at least 5 km away from the nearest ambulance post) on the ambulance response time in Jakarta. Results: During the period January 1st to December 31st, 2019, a total of 56,571 calls for emergency services were received from neighbourhoods in Jakarta. There were about 180 calls per day during weekdays and about 110 calls per day during the weekends. Also, the number of calls was highest between 9 am and 2 pm (11 calls per hour) and lowest overnight between 11 pm and 5 am (1 – 2 calls per hour). Of the calls received, 24% were from Central Jakarta, 23% from West Jakarta, 21% from East Jakarta, 17% from South Jakarta, 13% from North Jakarta, and less than 1% from the Thousand Islands. Also, 19% of the calls were for primary medical evacuations while 81% were for secondary medical evaluations. Ambulances were dispatched for 87% of these calls from 170 ambulance posts around Jakarta. The median dispatch time was 99.3 minutes [IQ