Since March 2020, the COVID-19 pandemic has impacted the population of San Francisco, causing over 36,000 cases and 500 deaths. However, the virus has had uneven impacts across different communities, disproportionately affecting racial and ethnic minority groups. Now, with the widespread distribution of COVID-19 vaccines from Pfizer, Moderna, and Johnson & Johnson and the possibility of herd immunity in reach, the city is beginning its recovery.
In this project, I will assess the distribution of the COVID-19 vaccine in San Francisco, focusing on the vaccine rollout for underserved and vulnerable communities and the impact the vaccines have had on the city as a whole. In order to understand these implications, I hope address three main questions:
How has the availability of vaccines in San Francisco impacted COVID-19 cases and deaths?
What does the rollout of COVID-19 vaccines look like for different communities? Are there disparities between populations?
Are vaccination sites accessible, especially to underserved populations?
COVID-19 Overview
In order to understand the impact of the vaccines, I will first examine the COVID-19 pandemic in San Francisco, looking at overall case rates and hospitalizations throughout the city. From San Francisco's Open Data Portal, we can see that the number of COVID-19 cases peaked in early January this year at 560 new cases a day, and has been on a steady decline since. Currently, the number sits at around 30 new cases daily, with a weekly average of around 27 new cases a day.
Chart 1: Daily and Weekly Averaged COVID-19 Cases in San Francisco
However, the distribution of COVID-19 cases across San Francisco is very uneven, with census tracts in the Southeast regions of the city suffering at much higher rates of around 1,000 cases per 10,000 residents, while the Western neighborhoods are at around 200 cases per 10,000 residents, illustrated in the map below. As follows, deaths from COVID-19 are also more common in the Southeast neighborhoods. These differences in COVID-19 in different census tracts rates are also reflective of income disparities in the city, which can be used as a social determinant of health.
Additionally, from this New York Times map of various ICU locations and their capacities, it is clear that the virus is still exhausting available resources, with the Kaiser Permanente hospital at around 90% capacity and the Zuckerberg General Hospital at around 80% capacity.
Overall, while COVID-19 cases have decreased, San Francisco is still working to contain the virus, but there are still many disparities in the impact of the pandemic on different neighborhoods and communities in the city. In particular, it seems that the census tracts and neighborhoods with lower income and larger populations of racial and ethnic minority groups have suffered more greatly from the pandemic.
Vaccine Rollout
Jasper Harris gets his COVID-19 vaccine at Laguna Honda; Source: SF.GOV
In this section, I will analyze the success of the COVID-19 vaccine rollout in San Francisco and draw conclusions about how the distribution of vaccines has impacted the rates of COVID-19 cases and deaths in order to illustrate the importance of vaccines in reducing the impact of the virus.
As vaccines have become widely available in the last few months, San Francisco has followed a multi-phase rollout plan in order to vaccinate our most vulnerable populations first. Some notable dates from the plan that should be kept in mind when analyzing the charts include:
Phase 1A: Frontline healthcare workers
Phase 1B, February 24th, 2021: Ages 65 and over and other essential workers
March 15th, 2021: Ages 16 to 64 with qualifying underlying health conditions
April 1st, 2021: General population, ages 50 and over
April 15th, 2021: General population, ages 16 and over
Data and Methodology
To examine the effects of the vaccine rollout, I used data from the San Francisco Open Data Portal to create charts displaying the number of each type of shot given and the rates of vaccinations, cases, and deaths for COVID-19. I decided to include visualizations for the rate of vaccinations at a city level and at a neighborhood level because having a more granular breakdown will be useful for understanding the vaccinated populations later on.
It should be noted that the data in chart 2 and chart 3 were accessed April 30th, 2021, while the data used for map 3 was access May 10th, 2021, so there may be minor disparities in the total numbers and rates of vaccinations. However, these differences should not change our overall analysis of the data.
Ideally, map 3 would have used a census tract breakdown for the vaccination data in order to make better comparisons to the COVID-19 case rates in map 1, but this data was unavailable; instead, I used breakdowns by Analysis Neighborhoods. Additionally, the Lincoln Park, McLaren Park, and Golden Gate Park neighborhoods were not found in the data, and the neighborhood-level values do not include around 2,000 people vaccinated in San Francisco that did not report a valid address. The rate and population values are also based off the 2019 5-year estimates from ACS data, so population totals may be different from current numbers.
Vaccination Numbers
Looking at the cumulative numbers of vaccinations given in San Francisco, we can see that San Francisco is at 50% complete vaccination, meaning over 400,000 people have had 1 shot of Johnson & Johnson or 2 shots of Pfizer or Moderna, and 73% have received at least one shot. These rates are pretty similar to California, where around 45% of the population have been fully vaccinated, and ahead of the United States as a whole, where only 32% of the population are fully vaccinated and 46% have received at least one dose.
From the map below, we can see that San Francisco has started its vaccine rollout with the average percentage of the neighborhood population vaccinated sitting at 62.4%, which is lower than the citywide vaccination rate of 73%. This means that there must be areas that have especially low vaccination rates and large disparities between the neighborhoods. The neighborhood with the lowest vaccination rate seems to be Lakeshore, in the Southwest, with only 34.8% of their population receiving at least one dose, while Japantown has the highest rate at 75.8% and Mission Bay and the Financial District following close behind at 74.6%, 73.8%, respectively.
Map 3: Rate of Vaccinations (At Least One Dose) in San Francisco Neighborhoods
Comparing map 1 and map 3, it looks like the Southwest Lakeshore region has relatively high COVID-19 case rates, with census tract case rates ranging from 50 to 466 per 10,000 residents and the average case rate of the four census tracts in the neighborhood being around 245 per 10,000 residents, but has low vaccination rates. Similarly, the Presidio and Marina areas share the same problem with high case rates but low vaccination rates.
Additionally, while the Mission and Bayview Hunters Point neighborhoods have relatively high vaccination rates of around 65%, they are some of the most heavily impacted regions in map 1 with average case rates in the thousands per 10,000 residents.
Vaccination Impacts
When analyzing the rates of vaccinations, cases, and deaths for the entire San Francisco population, we notice that the rate of COVID-19 cases has begun to plateau, especially as vaccines have been distributed and completed. Similarly, with the increase in vaccine administrations, death rates have also gone down. Because this chart uses cumulative case, death, and vaccination numbers over the total population of San Francisco to calculate the rate, a flat line means that there are no increases, a positive sign for the effectiveness of the vaccine, which is what we can begin to see with cases and deaths.
Chart 3: Rate of Vaccinations, Cases, and Deaths in San Francisco
So, from this section, it seems that the rollout of vaccines has had a really great effect on the San Francisco population as a whole. And, in conjunction with stay-at-home orders, the COVID-19 vaccine has greatly minimized cases and deaths. This downward trend is still ongoing with more of the population getting vaccinated each day. However, it is clear that vaccine rollout is uneven between neighborhoods with some reaching 75% while others struggle at 35%. It is important for the San Francisco leadership to address these disparities to ensure that all qualifying populations will be able to get vaccinated, no matter where they reside, as we have demonstrated that vaccines are helpful in reducing COVID-19 related cases and deaths.
In this section, I will continue to build off my previous analysis in order to assess the equity of vaccine distribution by exploring the disparities in vaccinations for different demographics, including race and ethnicity and age, and further exploring the populations of some of the previously identified neighborhoods with high numbers of COVID-19 cases but low vaccination rates, including the Lakeshore, Mission, and Bayview Hunters Point neighborhoods. Using this, I hope to evaluate the success of the vaccine rollouts for underserved and vulnerable populations in San Francisco.
Data and Methodology
In order to understand the equity of vaccine rollout, I continued to use COVID case, death, and vaccine data from the San Francisco Open Data Portal. Additionally, I also used data from the 2019 American Community Survey 5-year estimates to better understand the demographics of San Francisco and assess the disparities in vaccine distribution.
For chart 4, I compiled different datasets into one chart to make comparisons much easier. In order to reduce visual clutter, I simplified the categories into Black or African American, Hispanic or Latino, Asian, White, and Other. The Other category includes data for American Indian or Alaska Native, Native Hawaiian and Other Pacific Islander, Multi-Racial, Other, and Unknown race groups as these populations made up a much smaller portion of San Francisco. Unknown was a label used in the cases, deaths, and vaccines datasets and made up 6.44% of cases, 3.9% of deaths, and 4.69% of vaccinations.
For chart 5, I followed the same procedure as chart 4, but instead of using race and ethnicity, I looked at different age groups. However, unlike chart 4, where I was able to include the number of vaccines for each population in the same chart, I created a separate chart 6 for vaccinations by age. This was because the categorization of age was different between datasets.
Race and Ethnicity
In graphing the proportion of COVID-19 cases, deaths, vaccinations, and population for each race and ethnicity, I found that there were large disparities in the impact of the virus between the different groups. Some of which include:
The Hispanic and Latino population, in blue, has been disproportionately affected by the pandemic. They make up only 15% of San Francisco's population, yet have comprised of 40% of cases and 20% of deaths.
Black and African American populations, in red, face similar inequities at 6% of cases, 7.6% of deaths, and 5% of the total population.
Although only 18% of cases are from Asian populations, in yellow, they represent 37% of the deaths, while making up 34% of the total population.
Because we know that there are no biological differences in COVID-19 risk by race and all races have been engaging in the same prevention methods, we can see that COVID-19 is disproportionately impacting communities of color due to institutionalized racism and structural inequities.
Chart 4: Race and Ethnicity as a Share of COVID-19 Cases, Deaths, Vaccinations, and Population
Additionally, the distribution of vaccines among these racial and ethnic groups is also inequitable. Black and African American populations are being underserved, making up only 3.6% of vaccines but 5% of the population. Similarly, Hispanic and Latino populations are also undervaccinated, making up 12.7% of vaccines but 15% of the population.
However, we must also acknowledge that the Other category is particularly large and includes populations with unknown racial and ethnic identities, which could also change our perception on the equity of vaccine distribution.
Age
Looking at age as a share of cases, deaths, and vaccinations, we can see that populations ages 18-39 seem to have higher numbers of cases compared to their total populations, but the majority of deaths due to COVID-19 have been from the population aged 60+. From this, we understand that the impact and lethality of the virus is much more extreme for older populations.
Chart 5: Age as a Share of COVID-19 Cases, Deaths, and Population
In comparing the number of vaccinated people in each age group to their total population counts, we can see that vaccination across ages has been pretty equitable, even despite the challenges older populations may face in trying to schedule appointments online.
From this, we can conclude that there are still disparities in the distribution of vaccines to different demographics in San Francisco. In particular, minorities and communities of color have been underserved, receiving lower proportions of doses while facing the heaviest impacts of the virus. However, the city has done a good job of supporting people of all ages, including the more vulnerable elderly population, in receiving their vaccines.
Accessibility of Vaccination Locations
In this last section, I will explore the accessibility of vaccination sites in order to better understand what may be causing the disparities we see in vaccinations for different racial and ethnic minorities. I will assess whether the services offered at various locations are adequate enough to support the needs of various populations throughout San Francisco and if the physical locations are accessible to all populations.
Data and Methodology
To do this, I will use data from the San Francisco Open Data Portal to create maps of the vaccination sites. With these maps and diagrams, I will be able to evaluate vaccination deserts in the city. However, these maps are isolated to only present data about San Francisco, so it is important to keep in mind that this does not paint an entire picture about what vaccination sites look like in the surrounding region.
For map 4, I created a map of the different locations and included filters for wheelchair accessibility and languages served. This data was taken directly from the dataset of vaccination sites provided by the city, but seems pretty incomplete, with many unknown values for the locations. I chose to include wheelchair accessibility as a filter because I think it is an important consideration as older populations are more likely to face such mobility barriers. However, this is also a very limited evaluation of wheelchair accessibility as it only focuses on that particular location and not the accessibility of the surrounding area for wheelchairs, which could be a problem in San Francisco neighborhoods with more hills. I also chose to include the languages served filter because San Francisco has a really diverse population of immigrants who may not prefer to communicate in or understand English, so it is important to provide services that they can use too.
For map 5, I mapped the same dataset, but rather than focusing on the different features of each vaccination site, I focused on the physical locations. For this map, I drew 4 circles around each point, with increasing radii of 0.25 miles for each. I chose a radius of 0.25 miles because it is reasonable to assume that most people would be willing to walk 0.25 miles to a vaccination site, as that is a common assumption made in the first-mile-last-mile public transit problem. I chose to end the circles at a 1 mile radius because in food deserts, low access is defined as living over 1 mile away from a grocery store, so I thought the same could apply to vaccination deserts.
For chart 7, I used data from SF Planning's Neighborhood Profile. This document uses 2012-2016 ACS data, so it might be slightly outdated.
Impedance
Using map 4 and its filters, we can analyze the potential roadblocks that people may face when attempting to get vaccinated. While this data looks incomplete, we can still use it to garner a basic understanding of the services that vaccination sites offer to the San Francisco population to reduce the difficulties of getting a vaccine.
From the map, around 70% of locations are definitely wheelchair accessible, but the locations are spread out throughout the city, so it is likely that someone who needs wheelchair accessibility will be within reasonable distance of a location that can support their needs.
However, given the limited data about languages served, it seems like less than 10 vaccination sites, out of 95 total in San Francisco, offer additional language support aside from English. Additionally, these locations are mostly clustered on the East side of the city. For such a diverse city, the lack of language services at vaccination sites is concerning, as there are many racial and ethnic groups living in the city who do not speak English as their first language and struggle to understand.
Map 4: Services Offered at COVID-19 Vaccination Sites in San Francisco
In map 5, we see that most San Francisco communities live greater than 1 mile away from a vaccination site, which can be a challenge for those who travel by foot. We also notice that there are fewer vaccination sites in the South and larger areas of vaccination deserts in these neighborhoods. This can pose challenges to the people living in these areas as it will be harder for them to get their shots.
Map 5: COVID-19 Vaccination Deserts in San Francisco
In order to better understand who may be living in the previously mentioned vaccination deserts, map 6 overlays the different San Francisco neighborhoods with the vaccine locations. While most residential neighborhoods contain at least 1 vaccination site, it is often located far from many of the residents living in the neighborhood. For example, in the South of Market neighborhood, all vaccination sites are located on the boundaries of the region, and populations living in the Northwest part of Sunset/Parkside have to travel much further to get a vaccine. Additionally, there are still neighborhoods that do not have a vaccination site within its boundaries. These include:
Visitacion Valley
Twin Peaks
North Beach
Treasure Island
Seacliff
Presidio
Map 6: COVID-19 Vaccination Sites in San Francisco Neighborhoods
To understand who is more likely to live in vaccination deserts, I explored the demographics of the Visitacion Valley neighborhood.
Compared to the city of San Francisco, Visitacion Valley has a much higher minority population with nearly double the percentage of Asian, Black, and Latino populations. Additionally, there are slightly more Spanish and Asian Language speaking households. Median household income in this neighborhood was also lower than San Francisco's overall median household income, $54,745 and $88,643, respectively, and 15% of the population are in poverty.
Overall, it seems that San Francisco has tried to spread out vaccination sites throughout the city, placing additional locations in some of the more vulnerable communities and providing resources for vulnerable populations. However, they have also missed many neighborhoods and populations that would benefit from additional vaccination sites, such as Visitacion Valley in the Southeast. While I cannot definitively link the vaccine disparities by race to the location of vaccination sites, it seems like the populations in vaccine deserts are more likely to be lower income minorities and communities of color.
Conclusion
Looking at the big picture, the city of San Francisco has made great efforts to make vaccine distribution more equitable through the targeted rollouts and strategic placement of vaccination sites. However, from my analysis, it seems that there are still communities in San Francisco that are being underserved, creating disparities in vaccine access and increasing the impact of the virus on these populations.
We have seen drastic benefits of COVID-19 vaccines on the spread and fatality of the virus. These impacts have been reflected in the trends of cases and deaths over time.
Given that COVID-19 vaccines have been effective, it is important for the city to make them widely available. For now, it seems that there are still disparities in vaccine rollouts for racial and ethnic minorities, but not for different age groups.
While the city has made efforts to make vaccination locations accessible, it is necessary to increase resources and support for underserved communities through services such as language support, wheelchair accessibility, and working with local organizations. Additionally, many neighborhoods could benefit from additional vaccination sites in more central locations to allow for easier access.
Overall, I would say that their efforts have been commendable.
Ethical Concerns
The disparities faced by communities of color during COVID-19 have revealed the underlying institutionalized racism and structural inequities in our health system. They have experienced higher rates of COVID-19 infection, hospitalization, and death.
With the widespread availability of and access to COVID-19 data, we, as data scientists, must ensure that we are advocating for fair access to health. And while it may be exciting to work with such meaningful data, we must also realize that there are people behind each data point whose lives may depend on policy decisions driven by our data analysis.
