key: cord-0831386-ztrmpvbr
authors: Suwantika, Auliya A.; Dhamanti, Inge; Suharto, Yulianto; Purba, Fredrick D.; Abdulah, Rizky
title: The cost-effectiveness of social distancing measures for mitigating the COVID-19 pandemic in a highly-populated country: A case study in Indonesia
date: 2021-12-23
journal: Travel Med Infect Dis
DOI: 10.1016/j.tmaid.2021.102245
sha: febd50367c452b4be9f5b2794408fbb1be4069f7
doc_id: 831386
cord_uid: ztrmpvbr

BACKGROUND: As one of the strategies to mitigate the COVID-19 pandemic, social distancing (SD) measures are recommended to control disease spread and reduce the attack rate. Therefore, this study aims to estimate the costs and effects of SD measures through school closures, workforce, and community contact reductions for mitigating the COVID-19 pandemic in Indonesia. METHODS: Two mitigation scenarios of SD for 1 month and continuous SD were compared with the baseline (no intervention). A modified Susceptible-Exposed-Infected-Recovered (SEIR) compartmental model accounting for disease spread during the latent period was applied by considering a 1-year time horizon. The costs of healthcare, school closures, and productivity loss due to disease as well as intervention were considered to estimate the total pandemic cost among all scenarios. RESULTS: In a comparison with the baseline, the result showed that total savings in scenarios of SD for 1 month and continuous SD was approximately $415 billion and $699 billion, respectively, while the averted deaths were 4.6 million and 8.5 million, respectively. Sensitivity analysis showed that basic reproduction number, infectious period, daily wage, incubation period, daily ICU admission cost, and case fatality rate were the most influential parameters affecting the savings and the number of averted deaths. CONCLUSIONS: SD measures through school closures, workforce, and community contact reductions were concluded to be cost-saving. Increasing the duration of social distancing tends to increase both the savings and the number of averted deaths.

Background: As one of the strategies to mitigate the COVID-19 pandemic, social distancing 25 (SD) measures are recommended to control disease spread and reduce the attack rate. 26 Therefore, this study aims to estimate the costs and effects of SD measures through school 27 closures, workforce, and community contact reductions for mitigating the COVID-19 28 pandemic in Indonesia. 29

Methods: Two mitigation scenarios of SD for 1 month and continuous SD were compared 30 with the baseline (no intervention). A modified Susceptible-Exposed-Infected-Recovered 31

(SEIR) compartmental model accounting for disease spread during the latent period was 32 applied by considering a 1-year time horizon. The costs of healthcare, school closures, and 33

productivity loss due to disease as well as intervention were considered to estimate the total 34 pandemic cost among all scenarios. 35

Results: In a comparison with the baseline, the result showed total savings in scenarios of SD 36 for 1 month and continuous SD was approximately $415 billion and $699 billion, respectively, 37

while the averted deaths were 4.6 million and 8.5 million, respectively. Sensitivity analysis 38

showed that basic reproduction number, infectious period, daily wage, incubation period, daily 39 ICU admission cost, and case fatality rate were the most influential parameters affecting the 40 savings and the number of averted deaths. 41

Conclusions: SD measures through school closures, workforce, and community contact

The general occurrence of COVID-19 is a disruptive event that increases morbidity and 49 mortality globally and causes severe economic, social, and political impacts. Since the World 50

Health Organization (WHO) has declared the pandemic status of COVID-19 on March 11, 51 2020, the threat of this disease has been a major public health concern. 1 Learning from the 52 experience of the last influenza pandemic, three major mitigation interventions were therefore 53

provided. This includes curtailing interactions between infected and uninfected populations 54 through social distancing (SD) measures, decreasing symptomatic patients' infectiousness by 55 antiviral treatment, and reducing the susceptibility of infected individuals with vaccines. 2 56

Currently, there are no specific drugs for the treatment or prevention of COVID- 19 period. 9 Four compartments were applied for the number of susceptible (S), exposed (E), 86 infected (I), and recovered (R) individuals. The model subdivides the total population size at 87 time t denoted as N(t) into susceptible S(t), exposed E(t), infected I(t), and recovered R(t). The 88 model also states the exposed persons have made contact with a contaminated individual, yet 89 they are not infectious. Moreover, its dynamics were described by a set of four equations 90

correlating to the periods of the disease's progress: 91

(2) 93 94

(3) 95 96

(4) 97 98

In the equation above, Rt is the time-varying basic reproduction number, Tinf is the infectious 102 period, and Tinc is the incubation period. 9 103

The baseline pandemic is believed to continue developing exponentially and then 104 decline naturally until all susceptible people had contracted the disease. The transmission of 105 infection was probabilistically implemented in the model between the infected and vulnerable 106

individuals that had contact with the virus. In response to the growing distribution of the 107 disease, births and natural deaths (e.g., due to chronic diseases) are also believed to be steady. 108

Therefore, selected parameters were drawn from the article, and up to this point, detailed 109 information has been provided on the mitigation strategy in the model (see Figure 1 ). 110

In the context of health economics, approximately 271.6 million of the total population 111 in Indonesia was used to simulate the impact of SD for 1 month and continuous SD as 112

alternative strategies for mitigating the COVID-19 pandemic. 10 By comparing two mitigation 113

techniques to the baseline, overall savings on pandemic expense (in US Dollars according to 114 2020 price point) and the number of prevented deaths were estimated. Additionally, univariate 115

sensitivity tests were performed to analyze various input variables' influence on the saving and 116 the number of prevented deaths. 117 118

Epidemiological parameters 119

In this study, the reproduction number (R0) was applied at 2.2 (95% CI; 1.4-3.9) as the 120

indicator of infectiousness to calculate the number of secondary infections developed by each 121 infected person. 11 Tinc and the Tinf were estimated to be 5.2 (95% CI; 4.1-7.0 days) and 2.3 days 122 (95% CI; 0-14.9 days), respectively. 11 The rates of mortality and hospitalization were 123 calculated to be 3.7% (95% CI; 3.6-3.8%) and 18.4% (95% CI; 11.0-37.6%), respectively. 12 124 We identified recovery time for minor symptoms, time to hospitalization, and time from the 125 end of incubation to death at 24.7 (95% CI; 22.9-28.1 days), 12 7 (95% CI; 4-9 days), and 21 126 days (95% CI; 17-25 days), respectively. 13 In addition, the length of hospital stay was projected 127

to be 11 days (95% CI; 7-14 days). 13 The proportion of non-ICU and ICU hospitalization was 128 estimated as 32.9% and 67.1%, respectively, from the total cases recorded. 14 Furthermore, 129 approximately 42% and 81% reductions in disease attack rates during SD for 1 month and 130 continuous SD, respectively, were applied from two previous studies that considered the 131 potential of SD measures on reducing the pandemic level. 15, 16 To investigate the impact of SD 132 measures, three major mitigation interventions applied through school closures, workforce and 133 community contact reductions were considered. It was assumed that all school children needed 134 to stay at home and prevent contact with outsiders. For workforce reduction, all workers were 135 expected to have a 50% probability of staying at home and not make contact with their co-136

workers. Finally, all individuals were expected to have 50% less contacts in the community. 137 138

Cost parameters 139

The total number of the pandemic cost was specifically estimated from the societal 140

perspective. This included direct healthcare costs, for example, outpatients, non-ICU & ICU  141 hospitalizations, as well as indirect (e.g., productivity loss due to disease and intervention) and 142 school closures' costs. The cumulative costs of all health events were estimated in the baseline 143 scenario by adding the overall cost for outpatients, as well as non-ICU and ICU 144

hospitalizations. These were determined by multiplying the average daily cost by the average 145 duration of stay for each age category. productivity loss due to disease was estimated using an average wage of $11 per day and work-153 days lost due to illness. 19 Furthermore, the numbers of work-days lost due to GP visits, 154

hospitalization and ICU admission were estimated to be 25, 36, and 34 days, respectively. 12,13 155 To estimate the loss due to intervention, the working-age population (20-59 years old) was 156 assumed to lose 50% of their productivity in both scenarios of SD for 1 month and continuous 157

SD. The cost of missed school days due to SD was estimated using an average regular tuition 158 fee of $0.23 per student for a public institution and the number of school days lost depending 159 on the length of each mitigation circumstance. 20 Furthermore, the cost of childcare was not 160 included in this study. All costs were measured in US Dollars using the Central Bank of 161

Indonesia's most recent index changes from March 2020. 21 The model's input parameters can 162 be seen in Table 1 . 163 164

In The results confirmed that the overall expenses of the pandemic was $777 billion, $362 177 billion, and $78 billion in the baseline, SD for 1 month and continuous SD, respectively. 178

Besides, the cost of healthcare, school closures and productivity loss due to disease and 179

intervention was considered to estimate the overall pandemic cost in all scenarios. Both 180 mitigation scenarios were thought to be cost-effective since these approaches were more 181 successful while still being less expensive. More specific information about the cost analysis 182

can be seen in Table 3a . 183

When compared to other costs, the loss of productivity due to disease is expected to be 184 reduced by 2% and 92% in scenarios of SD for 1 month and continuous SD, respectively. 185

Productivity loss due to intervention tended to increase by 7% and 96% in both aforementioned 186 scenarios, respectively. Furthermore, the cost of healthcare is expected to decline by 5-6% and 187 that of school closures tended to increase by 1-2% in both mitigation scenarios. More specific 188

information on the cost shift in a percentage as a result of intervention can be seen in Table 3b . 189

Sensitivity analysis showed basic reproduction number, infectious period, daily wage, 190 incubation period, daily ICU hospitalization cost and case mortality rate were the most essential 191 parameters influencing the savings (see Figure 3a ). In addition, basic reproduction number, 192 infectious period, incubation period, and case fatality rate affected the number of averted deaths 193 (see Figure 3b ).

Discussion 196

Up to now, approximately 4.26 million confirmed cases and 0.14 million deaths due 197

to COVID- 19 The scale and speed of SD measures are unprecedented globally. Nevertheless, the 218 length of maintaining tight suppression measures by a country before behavioural fatigue 219 occurs in the population remains unclear. Based on predictions that SD measures have to be in 220 place for several months, there is an urgent need to identify the possible means by which a 221 country effectively reopens schools and workplaces. 30 This study estimated the total pandemic 222 cost to be $777 billion, $362 billion and $78 billion in the baseline, as well as scenarios of SD 223 for 1 month and continuous SD, respectively. The largest contribution to the total cost in all 224 scenarios was productivity loss due to disease and intervention (92-96%). The current result 225 has a similarity with two previous studies on the cost-effectiveness of strategies for mitigating 226 an influenza pandemic in Australia. 15, 16 Even though the fact that SD seems to significantly 227 reduce the incidence rate and the peak of mortality rate, this intervention provides a really huge 228 impact on the non-medical cost. 229

In general, the results showed all mitigation scenarios were considered to be cost-saving 230 since the interventions were more effective and less costly, which are also in line with some 231 previous studies. In a comparison with the baseline, total savings in SD for 1 month and 232

continuous SD are liable to be approximately $415 billion and $699 billion, respectively. In 233 addition, increasing the SD duration increase the intervention's effectiveness. The SD for 1 234 month and continuous SD scenarios tend to cause approximately 4.6 and 8.5 million averted 235 deaths, respectively. 16, 31 ,32 A systematic review on economic evaluations of interventions 236 against influenza pandemics highlighted SD had the potential to be cost-saving. 31 Measures 237 such as the SD that decreased person-to-person contact, had the highest cost per averted death 238 because of the economic disruption caused by the measures. 32 Milne et al. also stated a rigorous 239 and sustained SD intervention is cost-effective for mitigating the pandemic. 16 Sensitivity 240 analysis results also had similarity with several previous studies, which showed basic 241 reproduction number, infectious period, daily wage, incubation period, daily ICU admission 242 cost, and case fatality rate were the most influential parameters affecting the savings and the 243 number of averted deaths for mitigating a pandemic situation. 6,33 Basic reproduction number, 244

infectious period, incubation period and case fatality rate significantly affect the peak time, 245 peak infected proportion and total attack rate. 33 Additionally, daily wage and ICU 246 hospitalization cost are strongly associated with productivity loss and treatment cost due to 247 COVID-19, respectively. 8 248 J o u r n a l P r e -p r o o f This is the first economic evaluation study for mitigating the COVID-19 pandemic in 249

Indonesia, therefore, it has several major innovative aspects. Firstly, the setting used was 250 specifically focused to be in the country. An SEIR compartmental model accounting for the 251 spread of COVID-19 during the latent period was also developed, while input parameters were 252 mostly derived from the best available data. Secondly, this study was conducted from the 253 societal perspective, which is relevant for evaluating SD measures that are considered to have 254 higher non-medical costs than pharmacological interventions. Thirdly, two mitigation 255 scenarios were developed within a hypothetical model of disease spread on the duration of SD 256

measures. This is crucial since the government has the policy to review the decision 257 periodically, and cost-effectiveness is an important criterion for prioritizing mitigation 258 strategies in a pandemic situation. The lack of reliable local data on epidemiological parameters 259 (e.g., basic reproduction number; incubation and infectious period; time to recover, 260 hospitalization and death; case fatality & hospitalization rates; and disease attack reduction) 261

was discovered to be the study's main limitation. To deal with this, data were extrapolated 262 from several published studies in China, [9] [10] [11] 

Disease Control Priorities: Improving Health and Reducing Poverty. Disease Control 303 Priorities

Critical preparedness, readiness and response actions for COVID-19

Committee of COVID-19 Handling and National Economic Recovery. COVID-19 310 monitoring data

School closures and 317 influenza: systematic review of epidemiological studies

Health outcomes 319 and costs of community mitigation strategies for an influenza pandemic in the United 320 States

Cost-Effective Strategies for Mitigating a Future 322 Influenza Pandemic with H1N1

Epidemic calculator

of novel coronavirus-infected pneumonia

Estimates of the 332 severity of coronavirus disease 2019: a model-based analysis

Imperial College COVID-19 Response Team . Impact of non-pharmaceutical 338 interventions (NPIs) to reduce COVID-19 mortality and healthcare demand

Cost-Effective Strategies for Mitigating a Future 343 Influenza Pandemic with H1N1

The Cost Effectiveness of Pandemic Influenza 345 Interventions: A Pandemic Severity Based Analysis

The Differences of Unit 347 Cost Calculation by Activity Based Costing (ABC) Method and Double Distribution 348

Method for Category 2 of Pulmonary TB Patients in Outpatient and Inpatient of 349 Lung Hospital

352 19. Statistics of Indonesia. GDP per capita

Ministry of Finance. Dana BOS tahun 2020 naik dan sudah bisac air di Januari. 2020

The Central Bank of Indonesia. Exchange rate information

Effects of weather-related social distancing 363 on city-scale transmission of respiratory viruses

The impact of transmission control measures during the first 366 50 days of the COVID-19 epidemic in China

Effect of non-pharmaceutical interventions for 369 containing the COVID-19 outbreak

Effects of school closure on 372 incidence of pandemic influenza in Alberta

Economic Evaluation of Individual School Closure 375 Strategies: The Hong Kong 2009 H1N1 Pandemic

The Effects of School 380 Closures on Influenza Outbreaks and Pandemics: Systematic Review of Simulation 23 381 Studies

Estimating the costs of school closure for 383 mitigating an influenza pandemic

School closure and management practices during coronavirus outbreaks 386 including COVID-19: a rapid systematic review

Systematic Review of Economic Evaluations of Preparedness Strategies and 390 Interventions against Influenza Pandemics

The economy-wide impact of pandemic 392 influenza on the UK: a computable general equilibrium modelling experiment

Sensitivity Analysis of an Individual-Based 395

Model for Simulation of Influenza Epidemics

Infectious time (Tinf) 2.3 days (95% CI; 0-14

Time to death from end of incubation 21 days

Length of stay (hospitalization) 11 days

Time to recover for mild cases 24

Hospitalization rate 18.4% (95% CI

Disease attack rate reduction (SD for 1 month

Disease attack rate reduction

Average daily tuition fee $0.23 (95% CI

Average daily cost of hospitalization Average

Average daily cost of ICU admission Average tariff= $219.15 (Min= $110

Work-days lost (hospitalization) 36 (95% CI; 30-42

J o u r n a l P r e -p r o o f