An algorithm to estimate the real time secondary infections in sub-urban bus travel: COVID-19 epidemic experience at Chennai Metropolitan city India

Abstract

Globalization, global climatic changes, and human behavior pose threats to highly pathogenic avian influenza (HPAI) virus spillover from animals to human. Current SARS-CoV2 transmission continues in several countries despite drastic reduction in COVID-19 cases following world-wide containment measures including RNA vaccines. China reimposed lockdown in November 2022 following the surge in commercial hubs. Urban population density and intracity travel in over-crowded public transport play crucial roles in early transition to an exponential phase of the epidemic in metro-cities. Based on the SARS-CoV2 transmission during the lockdown period in Chennai metro-city, we developed an algorithm that mimics a real-time scenario of passengers boarding and deboarding at each bus-stop on a trip of 36.1 km in 21G bus service in Chennai city to understand the pattern of secondary infections on a daily basis. The algorithm was simulated to estimate R0, and the COVID-19 secondary infections was estimated for each bus trip. Results showed that the R0 depended on the boarding and deboarding of the infected individuals at various bus stops. R0 varied from 0 to 1.04, each trip generated 5–9 secondary infections and four bus stops as potential locations for a higher transmission level. More than 80% of the working population in metro-cities depends on unorganized sectors, and separate mitigation strategies must be in place for successful epidemic containment. The developed algorithm has significant public health relevance and can be utilized to draw necessary containment plans in near future in the event of new COVID-19 wave or any other similar epidemic.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13337-022-00804-9.

Introduction

It has been well established that the human-to-human transmission of respiratory viruses occurs via aerosol droplets when an infected person sneezes, coughs or speaks. Fomites are a significant source of droplet transmission as the virus can persist on surfaces for up to 96 hours [14]. The SARS-CoV-2 infection caused clusters of severe acute respiratory illness in China and was linked to 32% ICU admissions and 15% mortality [11]. It is estimated that the median incubation period is 5.1 days, and more than 97% of the infected persons developed symptoms within 11.5 days [23]. Initially, the prevalence of infection was lower in children less than 15 years of age, higher in male adults between the age of 34 and 59 years. Also, the morbidity in terms of clinical manifestations, severity by lung involvement and the mortality were related to the age of the infected persons, the existing co-morbidities and their treatments [3, 10]. Subsequently, the transmission of Delta and Omicron variants changed this trend over a period of 2 years and now, every age group is susceptible for the infection. However, the mortality was drastically reduced in several countries and the reduction in mortality is attributed to several factors including the vaccination.

Globalization and wide international travel across continents increased the risk of pandemics, primarily viral pathogens. During the last two decades, WHO has perceived several thousand potential signals for the multi-countries spread of several pathogens and four viral pathogens, H1N1, SARS-CoV-1, MERS, and Ebola, which exhibited real pandemic threat [26–29]. The current SARS-CoV-2 pandemic continues to be a public health threat globally even after vaccine intervention following emergency approval. In addition, cases reported during the spread of COVID-19 Delta, and Omicron variants demonstrated the impact of urbanization in the epidemic spread in several countries. The amalgamation of several peripheral areas with the metro-cities, high population density, unorganized housing in slums and travel to these cities for employment, in particular for the construction works and Micro–Small–Medium-Enterprises (MSME), are attributed to the speedy and the wide-spread of the infections within and adjacent areas of these cities. A multi-disciplinary approach that looked into the fundamental urban factors revealed that population density, multiple vehicle subways, and solid waste management significantly influence COVID-19 cases. Also, the distance from the epicenter strongly influenced epidemic transmission [17]. There are huge expansions of cities world-wide, and it is anticipated that the expansion will continue further for another two or three decades. In India, during the second wave from March to September 2021, when the Delta variant was the prominent strain, the Western and Northern parts of the country reported a high number of cases. After a low profile of cases reported for nearly 3 months, India reported exponential transmission at the end of December 2021. During this period, the highest number of cases were reported from the Western and Southern parts of the country [31]. During this period, several countries experienced exponential transmission despite the vaccination drive. Now, several countries have achieved the required vaccine coverage and therefore relaxed the travel restrictions and safety measures such as social distancing and facial mask. But, the pandemic is not over, there are rebound of cases in China and as on October 3, 2022 about 615 million confirmed COVID-19 cases and 6.5 million related deaths were reported globally [25].

The authors had earlier shown an exponential transmission mostly clustering in and around metro-cities, despite the intervention of the country-wide lockdown for more than 60 days from the midnight of 24th March 2020 [1]. Millions of people use public transport such as buses and trains every day and are often overcrowded in peak hours when the working population commutes between home and work sites. During the current COVID-19 pandemic outbreak, people are advised to maintain social distancing of 1 to 2 m. However, following these guidelines in public mass transportation systems is difficult. Apart from aerosol transmission, the probability of transmission due to fomites is likely to be high in public transport modes as windows, handrails, tickets, handles, panels, doors, and seats are likely to be contaminated with infective agents [14]. Secondary infections through the public transport play a significant role in COVID-19 transmission and the size of the epidemic in metro-cities.

Computational modelling can play a significant role in predicting the growth of epidemics, analyzing the transmission dynamics and to study the effect of control strategies. Numerical simulation incorporating the environmental factors in SEIR model showed that the rate of environmental contamination influences the number of COVID-19 cases [22]. Several countries, including the United States, have experienced aggregation of COVID-19 cases in thickly populated cities such as New York and California. We have shown that data-driven now-casting is preferable rather than fore-costing for metropolitan cities like Chennai to prevent the transition from pre-exponential to exponential phase by implementing city-based containment strategies [15]. Although COVID-19 prediction has been made for the cities [24], there is lack of reports to assess the high possibility of secondary infections while using the public transport for intra-city travel on daily basis.

The authors are fascinated to understand the factors involved in the transition from pre-epidemic to epidemic phase, in particular, the urban transmission of viral epidemics, short-term prediction models for nowcasting, and the promotion of alternate strategies to control epidemics in urban settings. In the present study, we attempted to identify the factors that compromise the safety of the urban population while utilizing the day-to-day public mass transportation facilities. Considering the randomness in contact with positive COVID-19 cases during urban and sub-urban bus travel, we preferred the Stochastic modelling approach to study the transmission of SARS-CoV-2 in Chennai. The model can be extended to other metropolitan, and medium size cities as similar population density, mass transportation needs, and social behaviors are witnessed in these urban establishments. Before we declare the termination of the current COVID-19 pandemic, highly pathogenic avian influenza (HPAI) virus detections were reported in 33 EU/EEA countries and the UK from December 2021 to March 2022, small spillover to humans was also reported in the UK, China, and Cambodia [7]. In the case of a specific correct mutation, there is a high possibility of transmission to the human population [18], primarily through the occupational spread. Therefore, it is desirable to explore the possible transmission potential of the epidemic viruses based on the rich experiences that we gained through the COVID-19 pandemic.

Methodology

Study site

The COVID-19 epidemiological profile in India indicates that the distribution of cases was high in metro cities like Delhi, Mumbai, and Kolkata and other thickly populated cities like Pune, Ahmadabad, Trivandrum, and Bangalore. These cities attract migration of the population for employment and education. Based on the reliable information on transportation in these cities and COVID-19 periodic data updates, we considered the city of Chennai in the southern part of India for assessing and predicting secondary infections during intra-city day-to-day travel. Chennai is the capital city of Tamil Nadu and one of the four metropolitan cities of India. It has an area of over 178.20 sq.Km and a total reported population of 4,646,732 in the year 2011. Ason 2017, there were about 1.82 million workers in Chennai metropolitan area, and there were more than 50,000 Micro–Small–Medium–Enterprises (MSME). However, less than 2% of the working population of Chennai was employed in MSME. While 10.9% of the workers were categorized as marginal workers, 86.4% were working in unorganized sectors. Chennai is connected with three other metro-cities Delhi, Mumbai, Kolkata, by Golden Quadrilateral Super-highways for promoting business, agriculture, education, employment, and regional culture. CMBT (Chennai Mofussil Bus Terminus) is the largest bus terminus in Asia, and it links all intercity buses from Chennai to other districts in Tamil Nadu and neighbouring states. Metropolitan Transport Corporation (MTC) of Chennai is the major public transport service in Chennai, and as on July 2022, it offered a fleet of 3448 buses, operated 3233 services from 31 depots on 604 routes, and carried about 2,88,200 passengers every day [9]. These fleets form the backbone of the transport for the working population as bus services are available to all the nooks and corners of this large metropolitan city.

To control COVID-19 transmission, the public health experts advocated operating the city buses only at 50% capacity (20 passengers with two crew members per bus per trip) to maintain social distancing and minimize secondary infections during the travel. Therefore, it was necessary to know the development of secondary infections despite maintaining the social distance by operating bus services at 50% capacity. In the present study, we considered the 21-G Tambaram to Broadway route to simulate the COVID-19 infection scenario as this is one of the major city bus-route in Chennai (Fig. 1). This route has around 40 bus-stops, and in each bus-stop, the number of passengers boarding and departing the bus was randomly generated with 15 passengers travelling from Tambaram to Broadway, and at each stop, a maximum of 5 passengers board or depart from the bus. Considering the pre-lockdown utilization of mass transportation scenario on this route, we assumed that each bus would be allowed a maximum of 20 commuters throughout the trip. The other assumption was that amidst 22 commuters on board (20 passengers and 2 crew members), at least one of the members was SARS-CoV-2 infected at the initial point. The developed simulation model also allows infected individuals to enter and exit the bus at subsequent bus stops randomly. The developed model utilizes a stochastic approach for simulating the travelling passengers; hence, the model creates more realistic scenarios of the mobility of the passengers. For a single day, 25 trips were considered, and the number of simulated secondary COVID-19 infections was computed for each bus stop for each trip. Further, the total number of secondary infections predicted among the passengers taking this bus route of 36.1 km was estimated.

Fig. 1.

a Four States and four metropolitan cities viz., Mumbai, Chennai, Delhi, and Ahmedabad with high COVID-19 case load, b Chennai Metropolitan intercity 21G Bus Route round trip terminals from Tambaram to Broadway of 36.1 km with 40 stops in between the start and end terminals

Model algorithm

The developed model is described in the form of an algorithm as expressed below (Algorithm 1). The algorithm was implemented and executed using Anaconda Python, in which a virtual environment was created.

Table 1 shows typical simulated trip details, including the bus-stops, travel time between each bus-stop, number of passengers entering the bus at each stop, the number of infected individuals entering and exiting the bus at each stop, the number of susceptible at each stop, the estimated basic reproduction number (R0) in the bus during the journey and the number of secondary infections.

Table 1.

Details of a typical simulated 21G bus trip from Tambaram to Broadway for the journey time between 4.40 and 6.26 PM

Bus Stop	Timing	Arraival	Arraival_inf	Depature_inf	Infected	Susceptible	R0	Sec_Infected
Tambaram	4:40	22	1	0	1	21	0.0000	0
Kadaperi	4:42	3	0	0	1	21	0.0000	0
Chromepet Government Hospital	4:48	3	1	0	2	20	0.0020	0
Chromepet MIT Flyover	4:50	5	0	1	1	21	0.0010	1
Chromepet	4:52	5	0	0	1	21	0.0016	0
Oberoi Hotel	4:54	3	0	0	1	21	0.0024	0
Ponds Company	4:56	5	0	0	1	21	0.0034	0
Pallavaram	5:00	4	0	0	1	21	0.0061	0
Army Camp	5:04	3	0	0	1	21	0.0097	0
Tirusulam	5:06	2	0	0	1	21	0.0119	0
Meeanampakkam	5:12	5	0	0	1	21	0.0199	0
Old Airport	5:16	3	0	0	1	21	0.0263	0
Junction of Palavanthangal & GST	5:17	4	0	0	1	21	0.0281	0
Alandur Cement Road	5:18	5	1	0	2	20	0.1169	0
Alandur	5:19	2	0	0	2	20	0.1242	0
St. Thomas Mount Head Post Office	5:22	5	0	1	1	21	0.0378	1
Prnaipalai	5:24	1	0	0	1	21	0.0422	0
Chellammal College	5:32	3	0	0	1	21	0.0617	0
Saidapet Court/ Taluka Office Road	5:34	2	0	0	1	21	0.0671	0
Chinna Malai	5:35	1	0	0	1	21	0.0699	0
Anna University	5:38	4	0	0	1	21	0.0787	0
Gandhi Mandapam (Guindy Park)	5:41	4	0	0	1	21	0.0879	0
Kotturpuram	5:47	1	0	0	1	21	0.1079	0
Nandanam (Kotturpuram Bridge)	5:49	3	0	0	1	21	0.1149	0
Adyar Gate (Park Sheraton)	5:51	1	0	0	1	21	0.1222	0
Kaliyappa Hospital	5:53	5	0	0	1	21	0.1297	0
Ramakrishna Madam	5:57	5	1	0	2	20	0.5486	0
Mylapore Tank	5:59	1	0	0	2	20	0.5782	0
Luz Church	6:00	4	0	1	1	21	0.1576	2
Thiruvalluvar Statue	6:02	4	0	0	1	21	0.1661	0
Yellow Pages (VM Street)	6:04	1	0	0	1	21	0.1747	0
Kalyani Hospital	6:05	3	0	0	1	21	0.1791	0
D.G.P Office	6:07	2	0	0	1	21	0.1880	0
Queen Marys College	6:08	4	0	0	1	21	0.1926	0
Kannagi tatue	6:12	2	0	0	1	21	0.2113	0
Marina Beach	6:14	4	0	0	1	21	0.2209	0
Anna Square	6:16	2	0	0	1	21	0.2307	0
Island Ground	6:20	1	0	0	1	21	0.2510	0
Secretariat	6:22	5	0	0	1	21	0.2614	0
R.B.I Parrys	6:23	1	1	0	2	20	0.9837	0
Broadway	6:26	0	0	2	2	20	1.0405	3

Results

The main expected outcome of this exercise was to estimate the number of secondary infections while intracity travel in one of the longest bus routes in a metro-city during the course of the COVID-19 epidemic, that can be extrapolated to similar epidemics in future.

The variation of the total number of susceptible and infected individuals travelling on a bus through the course of the journey from Tambaram to Broadway for a single simulated trip is shown in Fig. 2. Since the model utilizes a stochastic approach for simulating the journey, the variation in the total number of susceptible and primary infected individuals is observed to be random in nature, representing real-time scenarios.

Fig. 2.

The variation in the number of initial infected individuals and the number of susceptible populations travelling in the 21G bus from the start to end (Tambaram to Broadway) for a single trip of 106 min, simulated using the developed model

The reproductive number (R0) was estimated taking into consideration the duration of the travel, current time at each bus stop, total infected, the total number of passengers on board, and volume of the breathing space inside the bus. The changes in the estimated R0 in the bus are shown in Fig. 3 as a function of time for a typical trip starting from Tambaram at 04.40 pm and ending at Broadway at 06:26 pm. We considered this particular trip for the estimation of secondary infections as it represents the peak traffic involving several sectors of the population, including college students. It is seen that the R0 increases progressively over the duration of the trip for a single index case and is higher at the bus-stops after the entry of the additional primary cases. At the end of the journey, due to the higher exposure time and accumulation of primary infected individuals over the course of the journey, the R0 reaches a value of 1.0405.

Fig. 3.

The variation in the estimated reproductive number (R0) in the bus, shown as a function of time, during 21G single trip from Tambaram at 04.40 pm and ending at Broadway at 06:26 pm

Figure 4 depicts the total number of secondary infections for each trip from Tambaram to Broadway for a total of 25 trips in a day. It is seen that the number of secondary infections varies from 5 to 9 for each trip, and hence a single bus-route in a metropolitan city can cause a significant number of secondary infections even for the 50% occupancy advocated by the public health experts to maintain social distance.

Fig. 4.

The total number of estimated secondary infections in each simulated 21G bus trip from the start to end (Tambaram to Broadway) for 25 trips in a day. Each bar indicates the number of secondary infections in each trip

In addition, the simulated secondary infections for a total of 25 trips in a single day 21G bus travel for each bus stop is shown in Fig. 5. Results demonstrated that the total number of secondary infections generated at each bus-stop varies randomly. It is observed that the highest number of simulated secondary infections was observed at the destination terminus (Broadway) due to the prolonged exposures of the susceptible passengers with multiple primary cases entering at various bus stops. The study also picked up the potential locations (bus stops) for the higher number of secondary infections that can be attributed to the number of passengers at these bus stops due to the local population density and daily routine aggregation of the population for the purposes of the industrial or IT sector employment or education. Further, the day-wise total number of simulated secondary infections for 25 trips/day is presented for a period of 30 days in Fig. 6. It appears that a significant number of secondary infections are possible in a single-day bus journey in Chennai city when the exposure time is high, as seen in 21G bus route.

Fig. 5.

The variations and the total secondary infections in a single day for 25 simulated trips at each bus-stop in the 21G route from start to end (Tambaram to Broadway). The colour in the bar indicates the number of simulated secondary infections in each trip from Tambaram to Broadway

Fig. 6.

The variations and the total number of estimated secondary infections for 25 simulated trips per day in 21 G bus travel. Each colour in the bar indicates the number of simulated secondary infections in each trip from the start to end (Tambaram to Broadway)

Discussion

In the present study, we analyzed the potential secondary infections during the travel in the city bus for one of the most popular and also the longest bus service routes that covers the major south-to-north stretch of the Chennai metro-city in India. The results revealed that for the simulation of 25 trips of a single day, even for the occupancy limited to 50% capacity in each trip (20 passengers and two crew members), the R0 reached 1.04 at the destination terminus for the random entries of a variable number of index cases. It was also observed that each trip generated 5–9 secondary SARS-CoV-2 infections for a 1-day simulation. Secondary COVID-19 infection predicted by the stochastic modelling approach for 750 trips of a single bus-route for a period of 1 month indicates that the epidemic can extend to an exponential phase in a short period of 2–4 weeks. This is evident even for the urban bus services, with 50% occupancy advocated to contain the epidemic.

Predictions of the epidemic size for a country or a particular geographical location were made during earlier pandemics, such as influenza, SARS, and MERS, with the assumption that the individuals in the population under the study are homogeneous with respect to susceptibility, contact status, transmission and the road to recovery. A similar approach has been taken up by many investigators for the national-level predictions for the recent COVID-19 pandemic. However, a close look into the spread of this contagion indicates that most of the transmissions during the initial and the exponential phase of the epidemic are intensified in urban locations, in particular, the metropolitan cities that share the same characteristics of population density, economic activities, and overcrowding in low socioeconomic zones. Corburn et al. (2017) showed that successful containment of the COVID-19 epidemic in a country depends on the management of the urban population below the poverty line, urban health and delivery, sustaining the supply chain during the lockdown period, community participation and long-term benefits of the settlement population [5]. Mathematical modelling and the sensitivity analysis with six parameters and nine parameters showed that SARS-CoV-2 tends to follow oscillatory dynamics in India, size of the epidemic, social distancing along with timely hospitalization and case isolation are critical parameters in SARS-CoV2 epidemic containment. In addition, early intervention is expected to reduce the intensity of the epidemic peak [12, 13]. Extensive numerical analysis has shown that combination of two or more non-pharmacological interventions are likely to be more effective in epidemic control [19].

Deterministic and stochastic models are widely applied to understand the transmission dynamics of the epidemic and to suggest intervention strategies. For understating and the predictions of the epidemic in large populations at the country level, the application of a deterministic model is considered appropriate [4, 21]. In such scenarios, the estimation of the basic reproduction number (R0) will indicate whether the epidemic will take off irrespective of the initial condition in that large population. However, it has been well documented that for small localized population-based outcomes, the deterministic model breaks down [2]. Unlike in workplaces such as industries, in the scenario of bus travel, contact with infected persons follows a highly random phenomenon. In the present study, the investigators developed an algorithm using python language based on a stochastic approach and applied it with the objective of predicting the SARS-CoV-2 secondary infections during bus travel in Chennai metro city.

Experiences from several COVID-19-affected countries indicate that in the absence of an effective vaccine, mobility restricted only to the essential commodities and social distancing at the early phase of the transmission itself paved the way for the successful containment of the epidemic, as seen from the success story of Greece. On the other hand, Italy applied the containment strategy six weeks after the epidemic peak, which resulted in the failure of the containment. Predictions based on the ecological niche model (ENM) attributed a higher number COVID-19 positives cases to the population density and number of shopping malls per kilometer in addition to the number of bus stops in that area [20]. Singapore, a small country comparable in population size to the Delhi metro-city in India, contained the local SARS-CoV-2 transmission by strict implementation of the ‘stay at home strategy’ in the country [30].

It has been observed that the density of the urban population increased the scope for rapid COVID-19 transmission in India [6]. Similar trend was observed in the USA, and population density was found to be a critical parameter for the initial R0 [16]. Following the implementation of vaccination, there was a reduction in transmission of the alpha variant than the Delta variant. Following the spread of Omicron variant (B.1.1.529), even an effective vaccination drive with 50–70% population coverage for vaccination could not contain COVID-19 transmission [8]. It was first reported in South Africa during the first week of December 2021 and spread rapidly throughout Europe, UK, and USA, and currently, it is the major strain circulation globally despite worldwide COVID-19 vaccine coverage. India implemented the countrywide vaccine campaign by March 2021, and by 27th December 2022, when the third wave started to peak, 42.05% population received full (2 doses) vaccination. India maintained a low profile of cases from 27 October 2021 to 28 December 2021. However, on 31st December 2021, in 3 days, there was a two-fold rise, and in another week, on 07th January 2022, there was a sixfold rise [31]. This exponential transmission is attributed to the breakthrough infection by the spread of Omicron variant, which has been proven to be highly transmissible. Several countries have released travel restrictions and slowly started removing the safety norms in open areas. In addition, there is a growing trend for vaccine refusal, and soon free COVID-19 vaccine strategy will be withdrawn from the public health system.

Conclusions

Post-vaccination COVID-19 epidemiological profile with Omicron variant indicates low hospitalization and low case-fatality rate. While a few epidemiologists are optimistic that we have reached a ‘Pandemic end game’ following the declining trend in test positivity, WHO warns that it is premature for such a conclusion and the possibility of emerging virulent strains cannot be over-ruled. Considering the rise in flu case reports from April 2022 in flu transmission zones, WHO advocates “Integrated Surveillance:” for flu and COVID-19. The results of the current prediction indicate that random contacts during 10–96 min bus travel in a single day can result in five to ninefold secondary infections even with the restricted bus occupancy of 50%. A large working population that utilizes the city bus services are likely to be affected much more compared to the population that utilize their own vehicle for intracity transport on daily basis. Available data from Chennai metro-city indicates that more than 80% of the working population is engaged in unorganized sectors. Therefore, separate mitigation strategies must be in place for all the metro cities that report a large proportion of the cases. These strategies are possible only with community participation and effective media support. A large population who had received the second dose of vaccination more than 6 months back is likely to be going through the waning phase of the vaccination effect. In large countries like India, flu vaccination is optional, and it is not available free of cost in the public health system. In the absence of separate mitigation strategies for metro-cities, the emergence of any virulent variant similar to the Delta variant or the emergence of a flu epidemic will result in unbearable strain on the health system and a long-term impact on the economic status of the working population and the MSMEs of the country.

Supplementary Information

Below is the link to the electronic supplementary material.

Author’s contributions

LDB conceived the idea, developed the design and prepared the final manuscript, GRA, KK Developed the algorithm, tested, modified and finalized the program and revised the manuscript. BA collected the references and prepared the first draft. AK revised the manuscript and accorded the final administrative approval.

Funding

No separate funding was received for the study and the manuscript submission.

Declarations

Conflict of interest

No competing interests for any of the authors.