Trajectory analysis of the coronavirus pandemic and the impact of precautionary measures in the Kingdom of Bahrain

Abstract The Kingdom of Bahrain announced its first case of COVID-19 infection on February 24, 2020. Since that time, the government has imposed several restrictions such as closures, banning gatherings, and closing border crossings to limit the spread of the pandemic and relieve pressure on the healthcare system. This article provides an overview of the current trajectory of the pandemic in the Kingdom of Bahrain. In addition, the article introduces and applies a methodology to analyze the impact of the interventions and precautionary measures enforced by the government to limit the COVID-19 disease propagation. The results show that most of the enforced precautionary measures were effective in reducing the spread of the disease by a percentage ranging from 20.2% to 41.8%. A religious occasion in Bahrain—involving large gatherings—had increased the spread of the disease by 28.7%. Not enough evidence is found to conclude that reopening interventions had caused the disease to spread again.


Introduction
The world is currently experiencing a very difficult time due to the newly evolved, novel coronavirus pandemic, COVID-19, which has spread across the world, reaching 217 countries and territories (COVID-19 Coronavirus Outbreak, 2020) in just eight months and resulting in over 43 million cases of infection and more than one million deaths, and continues at a notable upward trend. The pandemic has also had impacts on the world economy and on the social lives of people around the world.
Unfortunately, preliminary studies show that the aftermath of the pandemic could last many years after the outbreak is contained. The situation was made worse by the lack of a vaccine to slow the spread of the disease until it can be eradicated, despite concerted efforts by many countries to develop an effective vaccine. Furthermore, because much remains unknown about the pandemic, the trajectory of this virus is hard to predict. Several countries have succeeded in suppressing the outbreak by adopting clinical practices and physical measures, including China, South Korea, Australia, Singapore, and others. However, there are far more countries still struggling to control the pandemic, including the United States, India, and Brazil. This implies that the impact of the outbreak could be significantly mitigated through the adoption of strict, physical non-pharmaceutical interventions (NPIs) and measures such as lockdowns, quarantines, social distancing, wearing masks, following good hygiene practices, and other isolation measures, which currently represent the only defense against COVID-19. However, maintaining such actions over prolonged periods has major economic and social consequences (e.g., bankruptcy and high unemployment rates) (Policy Brief, 2020).
Previous outbreaks have shown the world the importance of health promoting interventions, especially interventions developed with an understanding of the local culture in close collaboration with government authorities and leaders (Leach, 2008;World Health Organization, 2012).
Based on a recent study by Mitze, Kosfeld, Rode, and W€ alde (2020), the daily growth of COVID-19 infections can be reduced by 40% through the use of face masks. Rousan and Al-Najjar (2020) explored the impact of government interventions, in China. They found that the growth rate for new infections decreased from 4.43% to 2.64% after the introduction of government interventions. The growth rate for recovered cases was found to increase from 3.86% to 4.91% after the implementation of government regulations and rules (Rousan & Al-Najjar, 2020). Ferguson et al. (2020) investigated the impact of several NPIs in reducing the number of deaths and the level of demand on the healthcare system caused by COVID-19 in UK and USA. The results show that the impact of each NPI individually on the predicted number of deaths and level of healthcare demand is limited, but the impact becomes substantial when these NPIs are considered in combination. Applying NPIs at early stages of the COVID-19 pandemic in Taiwan caused the respiratory disease cases to be reduced by 50% (Hsieh, Lin, Wang, Pauleen, & Chen, 2020).
Another type of intervention occurs within the healthcare system (i.e., clinical interventions) such as increasing hospital capacity. During the 2014 Ebola outbreak in Lofa County, Liberia in West Africa, the overall number of cases was increasing by a factor of 4.7 when the number of beds was 10, and this factor decreased to 1.4 when the number of beds was increased to 40 (Funk et al., 2017). Teslya et al. (2020) explored the effect of selfimposed and government-imposed measures on the spread of COVID-19. They found that considering these measures together significantly mitigates the pandemic.
Some researchers have explored the impact of government interventions during the COVID-19 pandemic on other factors such as people's lives and activities (de Matos et al., 2020) and economy (Bonaccorsi et al., 2020;Kanitkar, 2020).
The Kingdom of Bahrain, like other countries, has suffered the impact of COVID-19 since February 2020. To slow the spread of infections, the government considered several precautionary measures including the enforcement of mask wearing, quarantining exposed individuals, instituting travel bans, closing schools and workplaces, cancelling mass gathering social events, and applying social distancing measures. These precautionary measures were aimed at reducing disease transmission and relieving pressure on the healthcare system by reducing the size and delaying the timing of the pandemic peak in the Kingdom of Bahrain.
The National Taskforce to Combat the Coronavirus in the Kingdom of Bahrain established an operating COVID-19 War Room on February 13, 2020, led by His Royal Highness the Crown Prince, to coordinate Bahrain's response to the emerging COVID-19 threat even before discovering the first case of infection in the Kingdom. The War Room's primary function is to help the Government of Bahrain to provide a wholeof-government response to the threat presented by SARS-CoV-2. The War Room coordinates the efforts of the numerous stakeholders involved in the Kingdom's response, including the COVID-19 call center, the testing and vaccination efforts conducted at the Bahrain International Exhibition Centre, admissions and emergencies, the provision of primary care at the numerous sites reserved for the national SARS-CoV-2 response, and national contact-tracing efforts. Furthermore, the team collects and collates statistics on the virus to assist senior leadership decision-making. Authorities in the Kingdom of Bahrain established quarantine centers for those suspected of infection and monitors voluntary quarantine for people in contact with patients who have been diagnosed with the disease.
In this article, an overview and discussion of the trajectory of COVID-19 in the Kingdom of Bahrain is provided in terms of statistics related to the number of daily and monthly new confirmed cases, recoveries, active cases, deaths, and tests performed. A statistical analysis is performed to evaluate the impact of precautionary measures enforced by the government and religious and social occasions that potentially involved overcrowded gatherings on the spread of the disease. The goal of this analysis was to explore whether the precautionary measures applied in the Kingdom of Bahrain were effective in containing and controlling the spread of disease and to investigate whether ignoring such precautionary measures led to an acceleration in the spread of the disease. To perform this analysis, we introduced and applied a novel methodology that considers both an indicator for the spread of the disease and the change in the number of tests.

Methodology
The actual data were collected for the number of daily new confirmed cases, recovered cases, deaths, and tests administered from the National Taskforce to Combat the Coronavirus in the Kingdom of Bahrain. The data are reported in an excel sheet, where the data for each day were reported at the end of the day and represents the actual data for that specific day. Figure 1 shows a sample of this data. We performed a descriptive statistical analysis for the data and described the current pandemic trajectory in the Kingdom of Bahrain. In this study, we considered data gathered from the beginning of the pandemic in the Kingdom of Bahrain on February 24, 2020 until the date on which we started finalizing the article, October 15, 2020.
To predict the spread of a disease, scientists use different mathematical models. These models provide estimates for the peak timing and value, end of pandemic dates, and total number of confirmed cases throughout the pandemic. These models are highly dependent on the number of daily reported confirmed cases, and therefore, the estimates may change from one day to the next. As the spread of the disease becomes more contained and controlled through the applied precautionary measures, the number of confirmed cases potentially decreases, and the estimates indicated by the prediction models improve. The Susceptible-Infected-Recovered (SIR) model (Al-Anzi, Alenizi, Al Dallal, Abookleesh, & Ullah, 2020; Hethcote, 2000) is one of the models used most frequently in the literature to estimate the future trajectory of a pandemic. Although other models, such as regression-based models (Al-Anzi et al., 2020; Almeshal, Almazrouee, Alenizi, & Alhajeri, 2020) and the Susceptible-Exposed-Infectious-Removed (SEIR) model (Godio, Pace, & Vergnano, 2020), could also be applied, they are not considered in this article because applying different modeling techniques and comparing their results is outside of the scope of this article.
In SIR modeling (Al-Anzi et al., 2020;Hethcote, 2000), in the absence of a vaccine and at any time t, the population N of a country consists of three categories: susceptible S(t), infected I(t), and recovered The rate of change for each of these factors is given as follows: where factors a and b are parameters determined by, respectively, the rate at which the susceptible cases become infected and the rate at which the infected cases become recovered. The basic reproduction number, R 0 (D'arienzo & Coniglio, 2020; Perasso, 2018) is a well-known and widely considered indicator for the spread of a disease. It is defined as the number of secondary infections caused by a single case of infection in a susceptible population. In SIR, the basic reproduction number is formally defined as: The value of R 0 depends on three factors. The first factor is the probability of infection when contact occurs between an infected and a susceptible person. The second factor is the rate of contact between an infected and a susceptible person. The third factor is the duration of infectiousness. It is important to note that precautionary measures are aimed at reducing the rate of contact between an infected and a susceptible person. Thus, when successfully applied, these measures potentially reduce the value of R 0 . The value of R 0 is a nonnegative real number where a value of "1" indicates that the spread of the disease remains as is without change, a value less than "1" indicates that the spread of the disease is speeding up, and a value greater than "1" indicates that the spread of the disease is slowing down.
To build the SIR prediction model, we used an existing MATLAB SIR modeling tool (Batista, 2020). The tool takes into account the possibility of multiple waves resulting from the summation of logistic functions (Reed & Pearl, 1927). Considering daily new confirmed cases as the only required input, the tool optimizes the model parameters by minimizing the difference between the actual and predicted number of cases.
To explore the impact of a precautionary measure enforced in the Kingdom of Bahrain on the spread of COVID-19, we calculated the value of R 0 for each of the 15 days before the date on which the precautionary measure was implemented and calculated the average, avg b , of these 15 R 0 values. In addition, we calculated the value of R 0 for the 15 days after the date on which the precautionary measure was implemented and calculated the average, aveg a , of these 15 R 0 values. To assess whether the precautionary measure was effective in slowing down the spread of the disease and the degree of this effectiveness, the percentage of change in the average R 0 values, denoted as POCR, before and after the implementation of the precautionary measure was calculated using the formula POCR ¼ (aveg aavg b ) Â 100/avg b . A positive POCR indicates that the precautionary measure caused the spread of the disease to speed up, and a negative POCR indicates that the precautionary measure caused the spread of the disease to slow down. The absolute value of POCR indicates the strength of the precautionary measure in speeding up or slowing down the spread of the disease, and therefore, it can be used as an indicator for which of the precautionary measures were more successful.
The change in the value of R 0 is mainly derived from the change in the number of daily confirmed cases, which is potentially affected by several factors. One of the key factors is degree of adherence to precautionary measures, where the value of R 0 potentially decreases as people adhere more strictly to precautionary measures. Another important factor is the number of daily tests performed. Increasing the number of tests potentially increases the number of daily confirmed cases and consequently increases the value of R 0 .
To ensure that the change in average R 0 was not mainly caused by a change in the number of tests performed, we calculated the percentage of change in the number of tests, denoted as POCT, performed for the same periods of time. We calculated the total number of tests, T b , performed during the 15 days before the intervention or occasion date and the total number of tests, T a , performed during the 15 days that followed the intervention or occasion date. The percentage of change in the daily number of tests before and after the implementation of the precautionary measure was calculated using the for- Considering the two factors, POCR and POCT, there are four possibilities to be considered in the assessment for whether the precautionary measure or occasion caused the spread of the disease to increase or decrease. The first possibility is when the value of POCR is positive and the value of POCT is negative. This case indicates that although the number of tests decreased, which typically decreases the number of positive cases, the impact from the intervention or occasion in speeding up the spread of the disease outweighed the impact from the change in the number of tests.
The second possibility is when the value of POCR is negative and the value of POCT is positive. This case indicates that although the number of tests increased, which typically increases the number of positive cases, the impact from the intervention or occasion in slowing down the spread of the disease outweighed the impact from the change in the number of tests.
The third possibility is when both POCR and POCT are positive. In this case, if the value of POCR is considerably higher than the value of POCT, then one will conclude that what potentially caused the spread of the disease to increase to that extent is the intervention or occasion, not the tests. Otherwise, the change in POCR is mainly derived by the change in POCT, and thus, the obtained values do not provide enough evidence that the intervention or occasion caused the spread of the disease to increase or decrease.
The fourth possibility is when both POCR and POCT are negative. In this case, if the absolute value of POCR is considerably higher than that of POCT, then one will conclude that what potentially caused the spread of the disease to decrease to that extent is the intervention or occasion, not the tests. Otherwise, the change in POCR is mainly derived by the change in POCT, and thus, the obtained values do not provide enough evidence that the intervention or occasion caused the spread of the disease to increase or decrease. The four possibilities and their resulting conclusions are summarized in Table 1, where the positive and negative signs indicate the increase and decrease of the value, respectively.
In this article, we considered 12 enforced government interventions and precautionary measures and three religious occasions involving gatherings. The impact of each of these precautionary measures and occasions on the spread of the disease in the Kingdom of Bahrain is explored herein.

Results and discussion
In this section, the current and future trajectories of the pandemic in the Kingdom of Bahrain are described and discussed. In addition, the impact of government interventions and precautionary measures and religious occasions involving gatherings on the spread of the disease in the Kingdom of Bahrain are statistically explored.

Current trajectory
By October 15, 2020, the total number of confirmed cases, recovered cases, active cases, and deaths due to COVID-19 in the Kingdom of Bahrain reached 76,954, 73,013, 3655, and 286, respectively. Figure 2 shows the trajectory of the total number of confirmed cases, active cases, and recovered cases throughout the pandemic in the Kingdom. Figure 3 shows the distribution of confirmed cases by month from March through September, reflecting two main peaks in new confirmed cases in June and September.
On June 8, the number of new confirmed cases reached 602, which was the peak of the first wave. The peak of the second wave occurred on September 10, when 757 new cases were reported. On October 15, the Kingdom of Bahrain was ranked as having the third highest rate of confirmed cases per capita in the world. Figure 4 shows the distribution of recovered cases by month from March through September. July featured the largest number of recovered cases. As of October 15, the percentage of confirmed cases that had been recorded as recovered cases was 94.9%, which is relatively high compared to other world countries. Figure 5 shows the distribution of deaths by month from March through September. The highest number of deaths was reported in June (68). As of October 15, the percentage of confirmed cases that had been recorded as deaths was 0.37%, which is very low compared to other world countries. Figure 6 shows that the number of tests performed each month increased gradually since the beginning of the pandemic. About ten times more tests were performed during September than during March. Increased testing helps identify more cases and contain the spread of the disease. On October 15, the Kingdom of Bahrain was ranked ninth in the world in terms of testing per capita. The change in daily positivity rates (i.e., number of positive test results divided by number of tests performed) is shown in Figure 7; the positivity rate ranged from 0.16% to 11.2%, with an average of 4.7%.
The relation between the daily number of tests performed and the daily number of confirmed cases is shown in Figure 8. The correlation is strong, with a coefficient of 0.893. We applied the linear regression technique and found that the relation between these two variables can be fitted using the following equation: number of confirmed cases ¼ À46.29 þ 0.055 Â number of tests, where R 2 ¼ 0.797.
The linear regression analysis results indicate that there was a strong relation between the number of daily random tests and the number of new confirmed cases, where more tests yielded more new cases. The equation of the regression line indicates that 5.5% of total daily tests are reported as new confirmed cases. COVID-19 pandemic in the Kingdom of Bahrain from February 24, 2020, the day the first five cases were reported, through October 15, 2020. The red curve represents the fitted and predicted trajectory obtained using SIR modeling. The curve shows that the pandemic went through two main waves. The peak of the second wave was higher than that of the first wave. The SIR model predicted that the total number of confirmed cases will reach 80,753 by February 17, 2021, which is estimated to be the date of the end of the pandemic. Based on this estimation, October 15, 2020 was the date at which the Kingdom had passed 95% of the total number of estimated cases. The prediction results indicate that Oct 22, 2020 and Nov 7, 2020 are expected to be the 97% and 99%, respectively, end-of-pandemic dates. It is important to note that these estimations change from one day to the next based on the reported daily confirmed cases. This number is affected by many factors. Some factors, such as implementing opening-up interventions for economic reasons, could delay the end-of-pandemic date, whereas other factors, such as adherence to precautionary measures and guidelines and the availability of a vaccine, could make the end-of-pandemic date earlier.

Impact of precautionary measures and occasions on the spread of the disease
To explore the impact of precautionary measures and occasions involving large gatherings on the spread of the disease in the Kingdom of Bahrain, we considered two factors: R 0 and number of tests.  Table 2 lists the main precautionary measures and occasions considered throughout the pandemic. The table consists of two parts. The first part includes 12 government interventions and precautionary measures, and the second part includes three religious occasions. For the first two occasions (i.e., Eid Al-Fitr and Eid Al Adha), relatives visit each other and family gatherings occur, whereas for the third occasion (i.e., Day of Ashoora), numerous large gatherings take place. For each intervention and occasion, we calculated the average R 0 value for each of the 15 days before the intervention or occasion date and for the 15 days after and reported them in the fourth and fifth columns of Table 2. We calculated the percentage of change in the average R 0 values from before and after each event, as reported in the fourth and fifth columns, and reported the results in the sixth column. For example, in the case of intervention PM1, which occurred on March 17, 2020, the average R 0 value for the 15 days before the intervention (March 3-17) was 2.29, while the average R 0 value for the 15 days after the intervention (March 18-April 1) was 1.56. The percentage of change in the average R 0 values is calculated as (1.56 À 2.29) Â 100/2.29 ¼ À31.8%. It is important to note that the negative sign for this percentage of change indicates that there was a decrease in the average R 0 value, which indicates that this specific intervention helped decrease the spread of the disease.
The percentage of change in the total number of tests before and after the intervention or occasion date is reported in the seventh column in Table 2. For example, in the case of intervention PM1, the total number of tests performed during the period from March 3-17, 2020 was 11,158, and the total number of tests performed from March 18-April l, 2020 was 21,541. The percentage of change, in this case, is 93.1%. This means that during the 15 days following the intervention, the number of tests performed increased substantially. Therefore, it was expected that the values of R 0 for the following 15 days would increase as well. However, contrary to that expectation, the average R 0 value decreased, indicating that the impact from intervention PM1 outweighed the impact from the change in the number of tests. This result confirms that the decrease in the average R 0 value was due to the effectiveness of the intervention in reducing the spread of the disease.
The results presented in Table 2 show that the lockdown and related precautionary interventions, including PM1, PM2, PM3, PM8, PM9, and to a lesser extent, PM4, had significant and clear impacts on limiting the spread of the disease, even though the number of daily tests increased considerably during the same periods of time. The effects of these interventions in reducing average R 0 values were greater than the effect of increased testing in increasing these values, thus proving the effectiveness of these interventions for slowing the spread of disease. The results show that the two most effective interventions for reducing the spread of the disease were PM8 and PM9. Typically, labor accommodations are crowded, and the disease spreads quickly in such an environment unless precautionary regulations are applied, as indicated by strong effects of PM8. In contrast, as expected, opening-up measures prompted by economic concerns, including PM7, PM10, and PM12, led to an increase in average R 0 values. However, this increase is less than that of the percentage of change in the number of daily tests. Therefore, the obtained results do not provide enough evidence that the opening-up interventions caused the disease to spread again. This result potentially indicates that people followed precautionary measures when the opening-up interventions were implemented.
The results for testing-related interventions, including PM5, PM6, and PM7, clearly show that introducing new testing options and methods led to significant increases in the number of tests performed, which helped in detecting and isolating more positive cases. Intervention PM11 did not show a similar clear trend, which may be due to the fact that people prefer to be tested at public health centers and hospitals rather than at private hospitals. Thus, allowing private hospitals to administer tests did not result in more people being tested. It is important to note that that the higher values for the percentage of change in average R 0 for testing related interventions PM5, PM6, and PM7 are mainly derived from the relatively larger value for the percentage of change in the number of daily tests that resulted from the implementation of new testing options.
For occasions involving gatherings, the results for occasions O1 and O2, which included small gatherings of relatives, show that the low values for the percentage of change in average R 0 values was mainly derived by similar values for the percentage of change in the average number of daily tests. This is likely due to the fact that such gatherings were very limited in size and may indicate that health precautions and guidelines were followed. As a result, this does not provide evidence that occasions O1 and O2 caused the disease to spread. In contrast, due to crowded gatherings associated with occasion O3, where health precautions and guidelines may not have been adhered to, there was a large increase in the average R 0 value that did not correspond with a large increase in daily tests. This observation indicates that this percentage of change in average R 0 values was not due to an increase of daily tests but due to negligence in following health precautions and guidelines.
As shown in Table 2, precautionary measures enforced by government and health officials were effective in limiting the spread of the disease. Some interventions and precautionary measures led to a larger reduction in the spread of the disease than others, and therefore, these interventions and precautionary measures should be re-enforced throughout the pandemic, whenever the spread of the disease substantially increases. Introducing new and varying testing methods led to a better grasp of how much the disease had spread. Finally, crowded gatherings must be avoided or strictly controlled during the pandemic.

Conclusions
As it is in most countries around the world, COVID-19 is still spreading in the Kingdom of Bahrain. The Kingdom of Bahrain has one of the highest rates of confirmed cases per capita, but it also has one of the highest percentages of recovered cases and testing rates per capita, and has a relatively low percentage of deaths. In the absence of a vaccine, the pandemic is not expected to end soon in the Kingdom of Bahrain. Since the beginning of the pandemic, the government of Bahrain has enforced several interventions and precautionary measures. The results of our statistical study show that the closurebased interventions and precautionary measures were effective in slowing down the spread of the disease in the Kingdom of Bahrain. In contrast, the results indicate that some religious occasions involving mass gatherings caused the disease to spread again. Not enough evidence is found that the opening-up interventions implemented due to economic concerns caused the disease to spread again, and this might be due to adherence to precautionary measures. It is always difficult to find the optimal balance between strictly enforcing necessary precautionary actions and avoiding a recession.