The Effect of Stay-at-Home Orders on COVID-19 Cases in Europe
Arianna Fallahian, Hiroko Kobayashi, Brooke Shuey, Gloria Strong
PSYC 500 final project
Embedded Files
Introduction
66.7 million cases. 1.5 million deaths. Though these may just sound like statistics, each of these numbers represents a human life that was affected by the coronavirus. Declared as a pandemic on March 11th 2020, the COVID-19 outbreak has claimed this many lives in just eight months. In the United States, cases are at an all-time high. Though researchers are hard at work creating a vaccine, its widespread distribution could take months. While the United States' response to the pandemic was quite shaky, other countries have taken remarkable measures to slow down the spread of the virus and by analyzing such measures we can learn more about their effectiveness as a way to prepare ourselves for the future. This investigation of other countries' responses to the pandemic is also useful in organizing new pandemic task forces to prevent pandemics altogether in the future or to severely minimize their impact globally. Considering the deadly nature of this virus and the damage it has done in less than a year it is important to gather as many resources as possible and investigate the success rates of various response measures around the world to successfully defeat COVID-19. This investigation uses two datasets from the European Centre for Disease Prevention and Control that were obtained and analyzed to determine the impact of certain response measures, specifically stay at home orders, taken among five European nations.
Since the United States has become polarized by the government's decision to enforce a stay-at-home order, we look to Europe as a model.
Does the stay at home order really work? What difference does it make? How has it helped other countries?
The assumption is "Well, of course it will reduce the spread."
However, using science and statistical data to support this could sway certain individuals to reconsider admonishing the stay-at-home order.
According to a study in the American Journal of Infection Control that looked at 42 states and Washington D.C. after they issued stay-at-home orders, the implementation of this response measure led to a 58% decline overall in the community infection rate. This study was performed by researchers from Johns Hopkins Bloomberg School of Public Health and data was obtained from state government websites. Researchers found that the community infection rate dropped from a 12% daily increase in cases to a 5% daily increase, indicating a shift from total number of cases doubling every 5-6 days to doubling every 14 days. Although this study provided useful information on the role of stay-at-home orders in "flattening the curve," it also has several limitations. The evolving availability and use of COVID-19 testing especially during early months of the pandemic may have played a large role in any increase or decrease seen in the infection rate. As a result, further investigation is necessary with updated COVID-19 data considering the progress that has been made thus far in both testing and response measures.
Data Curation and Ethics
Data Curation
Data Curation
We used two data sets, taken from the European Centre for Disease Prevention and Control.(https://www.ecdc.europa.eu/en)
ECDC COVID-19 Geographic Distribution Worldwide
Provides daily records of countries’ COVID-19 cases, deaths, and 2-week infection rate per 100,000 people from 2020
ECDC European Response Measures
Provides information about Europe’s response measures taken to deal with the COVID-19 pandemic on a national level, including implementation dates and stop dates
We chose this data because they offered the most in-depth look at the COVID-19 response measures across the world. Understanding the effects these response measures had, specifically a stay-at-home order, can help us apply effective policies in the United States. We focused our
Data Ethics
Data Ethics
When dealing with medical data, it is crucial to follow HIPAA ( Health Insurance Portability and Accountability Act of 1996) which demands that all persons' personal identifying information is kept secure. Therefore, no identifying information was collected, except for race and age, which is considered a demographic, not personal identifying information.
We chose to use a .eu source, because those are direct sources of the European data.
Both sets were open source and deal with populations instead of individuals.
Data was obtained from health authorities worldwide and is screened by up to 500 sources everyday.
The data collected was from 196 countries and only cases in deaths reported from national and regional competent authorities were obtained.
We determine this was ethically collected and we will employ an ethical approach to analyzing these data sets.
Hypothesis
Hypothesis
Our hypothesis was that the slope of COVID-19 cases would be smaller during the stay-at-home orders compared to before an after the orders were in place. We analyzed this trend among France, the United Kingdom, Spain, Italy, and Cyprus.
Data Sets
Data Sets
The two pictures below give a brief view of the information the data sets contain.
ECDC COVID-19 Geographic Distribution Worldwide
Shows in depth COVID-19 stats for each country worldwide including: daily deaths, population, and cumulative number for 14 days of cases per 100,000.
ECDC European Response Measures
Shows the start and stop date of each response measure per European country.
Data Preparation and Exploration
Data Types, Shapes, and Sizes
Data Types, Shapes, and Sizes
Data Type
Size and Shape
ECDC data set
ECDC data set
Summary: ECDC data set has a vast amount of information (over 670,000 elements of data!) that informs us of:
Daily records of countries’ COVID-19 cases
Deaths
2-week infection rate per 100,000 people in 2020
Data Type
Size and Shape
Response data set
Response data set
Summary: Response data set has a smaller, more concise type of information that informs us of:
Country
Type of Response to Pandemic
Start date for response implementation
Stop date for response implementation
Renaming, Combining, and Creating Dataframes
Renaming, Combining, and Creating Dataframes
We started off by looking at the countries which implemented a stay-at-home order, then isolated 5 countries from the list.
The response dates were recoded into datatype datetime in order to obtain the correct format to use when splitting datasets.
The United Kingdom
The United Kingdom
Creating a new dataset
Creating a new dataset
First, all UK cases were isolated from the ECDC dataset to create a new data frame containing only UK COVID-19 cases
Based on the start and end date the Response dataset provided for the stay-at-home order in the UK, a new column was created in the UK dataset to categorize all cases as before, during or after the order.
Similar steps were taken to create data frames for the other 4 countries.
France
Cyprus
Italy
Spain
Data Visualization and Descriptive Statistics
Data Visualization and Descriptive Statistics
The United Kingdom
The United Kingdom
A stay-at-home order was implemented March 24th through June 9th. During that time, we see cases are kept low. There was an increase in COVID-19 related deaths, this may be due to hospital understaffing, a shortage of PPE, and hospital overcrowding in the beginning months. It's only after the order was lifted that we see a sharp increase in deaths and cases rise in an exponential fashion.
France
France
A stay-at-home order was implemented March 17th through June 11th. During that time, we see cases are kept low. There was an increase in COVID-19 related deaths, this may be due to hospital understaffing, a shortage of PPE, and hospital overcrowding in the beginning months. It's only after the order was lifted that we see a sharp increase in deaths and cases rise in an exponential fashion. Due to this increase, France implemented a second stay-at-home order on October 28th.
Italy
Italy
A stay-at-home order was implemented March 10th through June 4th. During that time, we see cases are kept low. There was an increase in COVID-19 related deaths, this may be due to hospital understaffing, a shortage of PPE, and hospital overcrowding in the beginning months. It's only after the order was lifted that we see a sharp increase in deaths and cases rise in an exponential fashion.
Cyprus
Cyprus
Cyprus implemented a stay-at-home order from March 24th to June 3rd. We see deaths and cases were kept to a minimum during this time and it is only after the order is lifted that an exponential growth in cases can be seen. Cyprus does not follow the trend set forth by the other countries, this may be due to its low population.
Spain
Spain
A stay-at-home order was implemented March 14th through June 3rd. During that time, we see cases are kept low. There was an increase in COVID-19 related deaths, this may be due to hospital understaffing, a shortage of PPE, and hospital overcrowding in the beginning months. It's only after the order was lifted that we see a sharp increase in deaths and cases rise in an exponential fashion. We also see a negative data point in the "deaths" graph. This data point cannot be explained and did not affect future analysis.
Overall, we see the five countries followed a very similar trend in all three categories: positive cases, COVID-19 related deaths, and cumulative number for 14 days of COVID-19 cases per 100,000. It is only months after the stay-at-home order that all three categories are showing an exponential growth. A normality analysis and a linear regression analysis will be conducted for all five countries. To address the incident of negative values observed in certain datasets, prior to other analyses all case inputs were converted to their absolute values.
Model Building and Validation
Checking Normality for Daily COVID-19 Cases: Spain
Checking Normality for Daily COVID-19 Cases: Spain
The negative half of the Normality Analysis is being ignored as we do not have negative COVID-19 cases. Focusing on the positive half of the analysis, we see the cases are rising faster than the normality analysis predicted.
The normality for daily COVID-19 cases in the remaining 4 countries exhibited a similar pattern.
The United Kingdom
Italy
France
Cyprus
Linear Regression Analysis of Stay At Home Orders
Linear Regression Analysis of Stay At Home Orders
Procedure
Procedure
- Separate dataset for each country based in stay at home order to run linear regression analysis
2. Create numpy array, convert dates to Matplotlib dates.
(Put in max and min values into np.array for theoretical regression line)2. Create numpy array, convert dates to Matplotlib dates.
3. Matplotlib dates were converted to actual dates for x axis when created scatter plot
3. Matplotlib dates were converted to actual dates for x axis when created scatter plot
4. Line of best fit was created and it was plotted
4. Line of best fit was created and it was plotted
Results
Results
United Kingdom
United Kingdom
Before
Before
The data appears to follow an exponential trend and not a linear one. This shows the exponential growth of positive COVID-19 cases in the UK that led to the stay-at-home order being implemented.
During
During
Unlike the other countries in our analysis, the UK still shows a positive correlation between time and positive cases. However, this analysis also shows how stay-at-home orders can be used to slow the spread of COVID-19.
After
After
After the stay-at-home order was released, we see the trend revert back to the exponential growth we saw before the stay-at-home order was implemented. This provides evidence to the effectivness of the COVID response measure and also provide evidence it may need to be initiated again.
France
France
Before
Before
The data appears to follow an exponential trend and not a linear one. This shows the exponential growth of positive COVID-19 cases in France that led to the stay-at-home order being implemented.
During
During
During the stay-at-home order, the correlation flips negative and we see a country wide daily decrease in positive tests. This provides evidence towards the effectiveness of the stay-at-home order in France.
After
After
After the stay-at-home order was released, we see the trend revert back to the exponential growth we saw before the stay-at-home order was implemented. This provides evidence to the effectiveness of the COVID response measure and also provide evidence it may need to be initiated again. France implemented another stay-at-home order October 28th.
Italy
Italy
Before
Before
The data appears to follow an exponential trend and not a linear one. This shows the exponential growth of positive COVID-19 cases in Italy that led to the stay-at-home order being implemented.
During
During
During the stay-at-home order, the correlation flips negative and we see a country wide daily decrease in positive tests. This provides evidence towards the effectivness of the stay-at-home order in Italy.
After
After
After the stay-at-home order was released, we see the trend revert back to the exponential growth we saw before the stay-at-home order was implemented. This provides evidence to the effectivness of the COVID response measure and also provide evidence it may need to be initiated again.
Spain
Spain
Before
Before
The data appears to follow an exponential trend and not a linear one. This shows the exponential growth of positive COVID-19 cases in Spain that led to the stay-at-home order being implemented.
During
During
During the stay-at-home order, the correlation flips negative and we see a country wide daily decrease in positive tests. This provides evidence towards the effectivness of the stay-at-home order in Spain.
After
After
After the stay-at-home order was released, we see the trend revert back to the exponential growth we saw before the stay-at-home order was implemented. This provides evidence to the effectivness of the COVID response measure and also provide evidence it may need to be initiated again.
Cyprus
Cyprus
Before
Before
A linear regression analysis performed on the positive COVID-19 cases before the stay-at-home order shows Cyprus breaking the typical pattern we have seen from the other countries. We expected to see an exponential growth; however, this analysis shows cases were still steadly increasing. One reason we may be seeing different results is due to the low population of Cyprus.
During
During
During the stay-at-home order, the correlation flips negative and we see a country wide daily decrease in positive tests. This provides evidence towards the effectivness of the stay-at-home order in Cyprus.
After
After
After the stay-at-home order was released, we see the trend revert to the exponential growth we saw in other countries before the stay-at-home order was implemented. This provides evidence to the effectivness of the COVID response measure and also provide evidence it may need to be initiated again.
Discussion
Describe informed insights
Discuss limitations and future direction.
Discuss implications of the results in terms of policy and community interventions.
Discussion
Discussion
The objective of this analysis was to understand the effects of a stay-at-home order on COVID-19 cases and to assess whether it was effective at reducing cases and deaths in order to determine its potential implementation in other countries. Data was obtained from the European Centre for Disease Prevention and Control which is open source and de-identified. Two data sets were merged in order to compare the variables of country, response measure, start date, end date, cases, and deaths. After checking the normality of the data, a linear regression analysis was performed on positive COVID-19 cases before, during, and after the stay-at-home order was implemented in European countries. As a result, the investigation was then narrowed down to five countries: Cyprus, UK, Spain, Italy, and France. These countries were chosen because they displayed a variety of population densities, leadership styles, and national GDPs, but shared similar response measures used against the spread of COVID-19. This variety was crucial to understanding the effectiveness of stay-at-home orders as a whole while taking other factors into account. Nearly all five countries showed an exponential growth in COVID-19 cases before and after the stay-at-home order was implemented.
During the order, four of the five countries saw a negative correlation between time and positive cases. The final country, the United Kingdom, showed a positive correlation remained but the rate at which people tested positive was slowed (the slope flattened out during the stay-at-home order).
Limitations
Limitations
Some limitations that existed within the data sets were heterogeneity in the listed social distancing measures and the way they were implemented among European countries. This included the levels of enforcement and specific rules and exceptions which all varied on a country-by-country basis. The data set also only included measures taken on a national scale but failed to include regional or local measures that may have been implemented beforehand. This would affect the slope of COVID-19 cases seen before and during the national stay-at-home order which was listed in the data set. Possible delays in implementation may have also occurred in various cities which would affect the slope of the cases. Lastly, some countries' data on response measures were made unavailable once they were no longer being implemented, therefore there is more information about more recent measures taken.
Other limitations in the data sets provided by the ECDC include the lack of complete information, negative values for cases and deaths, and the way the data is collected. The ECDC gathers its data from the regional and national authority reports for each country. The negative values were not explained by the ECDC nor in the data sets. It is unknown what is meant by having them. These countries also did not provide data for every day during the pandemic which can lead to some inaccurate readings however, there was enough data that trends were still preserved and a reliable analysis could be conducted. Population density was not considered as we looked at the countries as a whole. Finally, because the data depended on confirmed positive cases, there is no way to know how many people were infected and were unable to take a COVID-19 or chose to not take it. This is further emphasized by the potential for people to be carriers while remaining asymptomatic and thus, unknowingly spreading the virus. This limitation is compounded by the high rate of false negatives the test had in the first few months.
Despite the present limitations, the data analysis supports the assumption that the stay-at-home orders were effective at reducing COVID-19 cases and could be an effective strategy for other countries to use, such as in the United States.
Suggestions and Implications
Suggestions and Implications
More data analysis is needed to see if the stay-at-home orders would be helpful if implemented on a state-by-state basis if it is not implemented on a federal level in the United States.
Another direction for the study on the effectiveness of stay-at-home orders would be an in-depth look at how it affects different population densities instead of countries as a whole.
The study of the effectiveness of stay-at-home orders has the potential to influence other countries' approaches to future pandemics as well as to ultimately defeating COVID-19. The information obtained will not only help researchers understand the nature of the virus and the way it spreads but also help policymakers implement effective measures in response to pandemics. Around the world countries can build upon this knowledge in order to prevent any future outbreaks.
Page updated
Google Sites
Report abuse