The Effect of Strict State Measures on the Epidemiologic Curve of COVID-19 Infection in the Context of a Developing Country: A Simulation from Jordan

COVID-19 has posed an unprecedented global public health threat and caused a significant number of severe cases that necessitated long hospitalization and overwhelmed health services in the most affected countries. In response, governments initiated a series of non-pharmaceutical interventions (NPIs) that led to severe economic and social impacts. The effect of these intervention measures on the spread of the COVID-19 pandemic are not well investigated within developing country settings. This study simulated the trajectories of the COVID-19 pandemic curve in Jordan between February and May and assessed the effect of Jordan’s strict NPI measures on the spread of COVID-19. A modified susceptible, exposed, infected, and recovered (SEIR) epidemic model was utilized. The compartments in the proposed model categorized the Jordanian population into six deterministic compartments: suspected, exposed, infectious pre-symptomatic, infectious with mild symptoms, infectious with moderate to severe symptoms, and recovered. The GLEAMviz client simulator was used to run the simulation model. Epidemic curves were plotted for estimated COVID-19 cases in the simulation model, and compared against the reported cases. The simulation model estimated the highest number of total daily new COVID-19 cases, in the pre-symptomatic compartmental state, to be 65 cases, with an epidemic curve growing to its peak in 49 days and terminating in a duration of 83 days, and a total simulated cumulative case count of 1048 cases. The curve representing the number of actual reported cases in Jordan showed a good pattern compatibility to that in the mild and moderate to severe compartmental states. The reproduction number under the NPIs was reduced from 5.6 to less than one. NPIs in Jordan seem to be effective in controlling the COVID-19 epidemic and reducing the reproduction rate. Early strict intervention measures showed evidence of containing and suppressing the disease.


11
While COVID-19 can cause severe illness and death, many uncertainties exist. The full extent of 12 the pandemic, especially in developing countries, the full clinical spectrum of illness, including 13 the prevalence of mildly symptomatic cases (4), and the true case fatality rates (6), are not truly 14 known. With 81% of infected cases developing only mild symptoms of COVID-19, it was 15 suggested that many infected individuals with mild symptoms may not seek testing (7). This 16 adds to the uncertainty of , especially in the developing countries with limited 17 testing and treating capabilities, and may make the true case count be as much as 10 times higher 18 than reported (9). Projecting case count, therefore, is an essential tool for public health response 19 measures and health system management. 20 Globally, two vital non-pharmaceutical interventions (NPIs) strategies have been identified to 21 control the spread of an epidemic; mitigation and suppression. The former focuses on slowing the 22 spread of the disease, but not necessarily stopping it, by reducing the healthcare demand peak and 23 by protecting at risk groups. Suppression, on the other hand, focuses on reversing the epidemic 1 growth, reducing case numbers to low levels and maintaining that situation indefinitely (10, 11) 2 In developed countries these measures have been effective in controlling the spread of COVID-19 3 (10, 12, 13). Such effect has been assessed using mathematical modeling that simulated the spread 4 of SARS-CoV-2 infection across the population and shaped control measures that might mitigate 5 future transmission (10, 12-21). One outcome of such simulation is the predicted epidemic curve 6 representing the number of infections caused by the virus over time. Using a set of parameters, 7 such simulation measures the impact of different interventions that can directly affect the predicted 8 epidemic curve (21). Mathematical modelling, therefore, presented itself as a powerful tool for 9 understanding transmission of COVID-19 and exploring different scenarios. Still, using such 10 modeling from developing countries, where healthcare systems are relatively weak, protective 11 equipment are scarce, and poor testing and treatment capacity exist, is controversial (7, 22, 23).

12
The Hashemite Kingdom of Jordan, a country in the Middle East region, initiated, on February 13 27, its national response to COVID-19 by banning non-Jordanian travelers from high-risk 14 countries from entering Jordan. On March 2, the first COVID-19 case was reported for a national 15 arriving from Italy. In the same week, Jordan initiated a quarantine for arrivals from selected 16 European countries. On March 15, a total of 12 new cases were reported, and all educational 17 institutions, tourism sites, cafes, and restaurants were ordered closed. All arriving passengers 18 (N=5,050) were then handled as suspected cases and immediately quarantined. Jordan then 19 prohibited travel between governorates, suspended all flights, closed borders, suspended public 20 transportation, closed commercial complexes, suspended non-emergency medical services, 21 closed public and private sectors, implemented stay-at-home policy, and prohibited public, 22 social, and religious events. Jordan then declared national lockdown, a state of emergency, and 23 imposed a curfew. During the early couple of days of the curfew, a complete nationwide 1 lockdown banned people from leaving their homes. Citizens were then allowed five specific days 2 to locally move, walk, and neighborhood grocery stores were allowed to open between 10 AM 3 and 6 PM. Driving was not allowed and moving between administrative geographic boundaries 4 was permitted under emergency circumstances.

5
The number of newly reported COVID-19 cases in Jordan fluctuated between three cases and 42 6 daily cases (mean number of daily reported cases was 15 cases). As of May 1, the number of  The current research will also advance our knowledge about COVID-19 in developing countries 21 and the effect of publicized responses implemented with widespread adherence and support in 22 Jordan.

1
A modified Susceptible, Exposed, Infected, and Recovered (SEIR) epidemic model (24) to 2 simulate the spread of COVID-19 in Jordan was utilized. SEIR model simulates the infectious 3 disease spread assuming no births, deaths or introduction of new individuals occurred. As such, 4 each individual is initially assigned to each of the following disease states (deterministic 5 compartments): susceptible (S), exposed (E), infectious (I) or recovered (R). The deterministic 6 compartments in the SEIR model are fairly sophisticated quantitative mathematical models yet are 7 easily run utilizing public data and known disease characteristics (24). We have modified the 8 standard SEIR model by adding compartmental states that reflect the compartmental population 9 and research needs. Our modified model categorized Jordan population into six deterministic 10 compartments; susceptible, exposed, infectious pre-symptomatic (representing the total number of 11 infections in Jordan), infectious with mild symptoms (i.e. not needing hospitalization), infectious 12 with moderate to severe symptoms (i.e. needing hospitalization), and recovered.

13
In designing the modified simulation model, we assumed that an exposed individual may become 14 infectious pre-symptomatic who then may progress to recovered, or progress to either mild or 15 moderate to severe symptomatic individual, both of whom may then progress to recovered. The 16 following brief shows the compartmental states applied in our study:   The modified model predicts the number of simulated COVID-19 cases by each compartmental 1 state in Jordan. It also has the potential to distinguish hidden (asymptomatic or mild, not seeking 2 hospital care) from identified infected cases needing hospitalization (moderate to severe cases).

3
Indeed, standard SEIR models are estimated assuming that all infected people are reported. Such 4 an assumption for the novel COVID-19 pandemic is largely unreasonable, as many infected 5 people show no or mild symptoms and as the testing procedure is not available in mass, many 6 remain undetected (25). The model also accounts for hospitalization of moderate to severe cases 7 by adjusting the contact rate. It is assumed that such cases will be detected and quarantined 8 within healthcare setting as they will be seeking medical services. Hence, their contact rates will 9 decrease tremendously.    Beta (β): is the contact rate which describes the spread of disease in the community. 4 Since Jordan culture is homogeneous, and people follow traditional forms for 5 greeting, we have set the standard contact rate (β) to 0.37 (16,28,29). To reflect the  (in days) and the time to the peak (in days). Each S1 curve was also fitted against the reported 20 daily number of cases.  Figure 1 presents the number of daily new COVID-19 cases in the pre-symptomatic 2 compartmental state, simulated under the S1 and S2 using the same scale. S1 curve is 3 demonstrated as a "baby" curve under S2 curve that started after February 1 and ended before 4 April 20. The simulation model, under S1, predicted that on March 20 the highest number of 5 daily new cases in the pre-symptomatic compartmental state will be 65 cases, after which, the 6 number of simulated daily new cases started to decrease. By April 24, the predicted daily new 7 cases leveled at zero. Considering that the simulation was set to start on February 1, and the NPIs 8 commenced on March 17, it took the epidemic curve 49 days to grow to its peak and the total 9 duration of the epidemic curve was predicted at 83 days. The cumulative number of cases was 10 predicted at 1,048. For the hypothetical scenario of no-action (S2), the epidemic took a total of 11 147 days to reach its peak of 238,142 daily new cases by June 27, and the cumulative number of 12 cases reached about 9.5 million around December 1.

13
The simulated daily new mild COVID-19 cases under S1 reached its peak on March 21 with 36 14 cases and a total duration of 49 days (Figure 2). After which, the simulated daily new mild case 15 count started to decrease and reached, on April 27, zero daily new cases (total duration of the 16 epidemic curve was 87 days).

17
As seen in Figure 3, the simulated daily new moderate to severe cases, under S1, reached a 18 maximum number on March 24 with a total of 46 cases (a total of 53 days). The number then 19 decreased to zero cases on April 27 (total number of days for the epidemic was 87 days).

20
In Figure 4, we plotted the actual reported daily new cases in Jordan against the simulated cases 21 in our model (S1). The curves representing the simulated number of daily new 22 in both the mild and moderate to severe compartmental states, had good pattern compatibility 23 with those depicting the number of reported cases in Jordan, with a peak of new cases on March 1 24. 2 Under S1, the simulated cumulative recovery was 1,044 cases by June 30. Out of the total 3 cumulative cases, 695 cases were in the moderate to severe compartmental state, i.e. needing 4 hospital care, while 795 were in the mild compartmental state, i.e. mostly hidden cases within the 5 community. As well, based on the S1 model, the simulated reproduction number (R0) for 6 COVID-19 after implementing NPIs in Jordan was estimated at 0.9.

7
Further comparisons between the S1 and S2 simulated models are presented in the The simulated model suggests that Jordan has presented a unique strategy that allowed 17 "snuffing" the COVID-19 pandemic at an early stage and, supposedly, resuming normal life 18 seemingly more quickly than many others. However, herd immunity has not been secured and a 19 second wave remains a concern. This is especially evident given that after about 10 days of zero 20 new cases in late April and early May, a new cluster of cases emerged and a new epidemic curve 21 started. These cases were traced to a truck driver who tested negative at the border in late April, 22 and then was admitted to the hospital as a COVID-19 case on May 8. So far, the second wave 1 has produced a daily case count of about 15 cases between May 8 and June 20. 2 Strict NPI measures implemented in Jordan, which lasted for more than six weeks, reduced 3 COVID-19 transmission and likely reduced the reproduction number to less than one. A similar 4 discussion was presented for the UK, for example (13), where, in the absence of control 5 measures, the epidemic would quickly overwhelm the healthcare system. A combination of 6 moderate interventions (school closures, shielding of older groups and self-isolation) was 7 predicted to be unlikely to prevent the epidemic that would far exceed available ICU capacity in 8 the UK. More intensive lockdown-type measures, however, predicted an effective protection of 9 the healthcare system from being overwhelmed. Of importance, lockdown scenario for the UK 10 effectively reduced R0 near or below one (13).

11
Our results are critical not only for public health decision makers, but also for risk households or local communities. The assumption of a universal contact rate used in the 23 proposed model was, however, adjusted for all cases with moderate to severe clinical 1 manifestations. Considering that these cases are most likely to be detected within healthcare 2 settings and be hospitalized, we reduced their contact rate to its minimum to overcome this 3 limitation. 4 A combination of NPIs with isolation and contact tracing were reported to present a synergistic 5 effect that increased the prospect of containment of . Knowing that Jordan has 6 implemented strict contact tracing and isolation of contacts limits our ability to clearly compare 7 the actual reported numbers to those presented under S1. Until detailed information about cases 8 identified via contact tracing and isolation are made available, the presented model (S1) is the 9 only available method to meet the objective of the current study. As well, the numbers presented 10 under S2 seemed to be high values as it assumed no prevention and control measures were 11 implemented. Their interpretation, therefore, should be limited to comparison with S1 and should 12 be seen as mostly hypothetical.

13
Until today, COVID-19 models presented from developing countries are scarce and not fully 14 investigated especially where disease suppression is considered as the main strategy to combat 15 the epidemic and reduce the potential impact of cases on the healthcare systems.   Authors' contributions: KK conceived the idea, prepared initial data for simulation, and drafted 8 the initial manuscript. BA ran the simulation models, and prepared the initial methods and