A. SUMMARY This dataset includes COVID-19 tests by resident neighborhood and specimen collection date (the day the test was collected). Specifically, this dataset includes tests of San Francisco residents who listed a San Francisco home address at the time of testing. These resident addresses were then geo-located and mapped to neighborhoods. The resident address associated with each test is hand-entered and susceptible to errors, therefore neighborhood data should be interpreted as an approximation, not a precise nor comprehensive total.

In recent months, about 5% of tests are missing addresses and therefore cannot be included in any neighborhood totals. In earlier months, more tests were missing address data. Because of this high percentage of tests missing resident address data, this neighborhood testing data for March, April, and May should be interpreted with caution (see below)

Percentage of tests missing address information, by month in 2020 Mar - 33.6% Apr - 25.9% May - 11.1% Jun - 7.2% Jul - 5.8% Aug - 5.4% Sep - 5.1% Oct (Oct 1-12) - 5.1%

To protect the privacy of residents, the City does not disclose the number of tests in neighborhoods with resident populations of fewer than 1,000 people. These neighborhoods are omitted from the data (they include Golden Gate Park, John McLaren Park, and Lands End).

Tests for residents that listed a Skilled Nursing Facility as their home address are not included in this neighborhood-level testing data. Skilled Nursing Facilities have required and repeated testing of residents, which would change neighborhood trends and not reflect the broader neighborhood's testing data.

This data was de-duplicated by individual and date, so if a person gets tested multiple times on different dates, all tests will be included in this dataset (on the day each test was collected).

The total number of positive test results is not equal to the total number of COVID-19 cases in San Francisco. During this investigation, some test results are found to be for persons living outside of San Francisco and some people in San Francisco may be tested multiple times (which is common). To see the number of new confirmed cases by neighborhood, reference this map: https://sf.gov/data/covid-19-case-maps#new-cases-maps

B. HOW THE DATASET IS CREATED COVID-19 laboratory test data is based on electronic laboratory test reports. Deduplication, quality assurance measures and other data verification processes maximize accuracy of laboratory test information. All testing data is then geo-coded by resident address. Then data is aggregated by analysis neighborhood and specimen collection date.

Data are prepared by close of business Monday through Saturday for public display.

C. UPDATE PROCESS Updates automatically at 05:00 Pacific Time each day. Redundant runs are scheduled at 07:00 and 09:00 in case of pipeline failure.

D. HOW TO USE THIS DATASET San Francisco population estimates for geographic regions can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS).

Due to the high degree of variation in the time needed to complete tests by different labs there is a delay in this reporting. On March 24 the Health Officer ordered all labs in the City to report complete COVID-19 testing information to the local and state health departments.

In order to track trends over time, a data user can analyze this data by "specimen_collection_date".

Calculating Percent Positivity: The positivity rate is the percentage of tests that return a positive result for COVID-19 (positive tests divided by the sum of positive and negative tests). Indeterminate results, which could not conclusively determine whether COVID-19 virus was present, are not included in the calculation of pe

Analysis of ‘Covid-19 Tests by Race Ethnicity and Date’ provided by Analyst-2 (analyst-2.ai), based on source dataset retrieved from https://catalog.data.gov/dataset/68410b4b-052f-4ce3-8d0c-873b5664f1a4 on 27 January 2022.

--- Dataset description provided by original source is as follows ---

Note: As of April 16, 2021, this dataset will update daily with a five-day data lag.

A. SUMMARY This dataset includes San Francisco COVID-19 tests by race/ ethnicity and date. For each day, this dataset represents the daily count of tests collected by race/ethnicity, and how many of those were positive, negative, and indeterminate. Tests in this dataset include all tests collected from San Francisco residents who listed a San Francisco home address at the time of testing, and tests that were collected in San Francisco but had a missing home address. Data are based on information collected at the time of testing.

For recent data, about 25-30% of tests are missing race/ ethnicity information. Tests where the race/ ethnicity of the patient is unknown are included in the dataset under the "Unknown" category.

This data was de-duplicated by individual and date, so if a person gets tested multiple times on different dates, all tests will be included in this dataset (on the day each test was collected).

The total number of positive test results is not equal to the total number of COVID-19 cases in San Francisco. Each positive test result is investigated. During this investigation, some test results are found to be for persons living outside of San Francisco and some people in San Francisco may be tested multiple times. In both cases, these results are not included in San Francisco’s total COVID-19 case count. To track the number of cases by race/ ethnicity, see this dashboard: https://data.sfgov.org/stories/s/w6za-6st8

B. HOW THE DATASET IS CREATED COVID-19 laboratory test data is based on electronic laboratory test reports. Deduplication, quality assurance measures and other data verification processes maximize accuracy of laboratory test information.

C. UPDATE PROCESS Updates automatically at 05:00 Pacific Time each day. Redundant runs are scheduled at 07:00 and 09:00 in case of pipeline failure.

D. HOW TO USE THIS DATASET Due to the high degree of variation in the time needed to complete tests by different labs there is a delay in this reporting. On March 24 the Health Officer ordered all labs in the City to report complete COVID-19 testing information to the local and state health departments.

In order to track trends over time, a data user can analyze this data by "specimen_collection_date".

Calculating Percent Positivity: The positivity rate is the percentage of tests that return a positive result for COVID-19 (positive tests divided by the sum of positive and negative tests). Indeterminate results, which could not conclusively determine whether COVID-19 virus was present, are not included in the calculation of percent positive. When there are fewer than 20 positives tests for a given race/ethnicity and time period, the positivity rate is not calculated for the public tracker because rates of small test counts are less reliable.

Calculating Testing Rates: To calculate the testing rate per 10,000 residents, divide the total number of tests collected (positive, negative, and indeterminate results) for the specified race/ ethnicity by the total number of residents who identify as that race/ ethnicity (according to the 2018 5-year estimates from the American Community Survey), then multiply by 10,000. When there are fewer than 20 total tests for a given race/ethnicity and time period, the testing rate is not calculated for the public tracker because rates of small test counts are less reliable.

Read more about how this data is updated and validated daily: https://data.sfgov.org/stories/s/nudz-9tg2

There are two other datasets related to tests: 1. COVID-19 Tests 2. <a href="https://data.sfgov.org/dataset/Covid-19-Testing-by

--- Original source retains full ownership of the source dataset ---

On 6/28/2023, data on cases by vaccination status will be archived and will no longer update.

A. SUMMARY This dataset represents San Francisco COVID-19 positive confirmed cases by vaccination status over time, starting January 1, 2021. Cases are included on the date the positive test was collected (the specimen collection date). Cases are counted in three categories: (1) all cases; (2) unvaccinated cases; and (3) completed primary series cases.

1. All cases: Includes cases among all San Francisco residents regardless of vaccination status.

2. Unvaccinated cases: Cases are considered unvaccinated if their positive COVID-19 test was before receiving any vaccine. Cases that are not matched to a COVID-19 vaccination record are considered unvaccinated.

3. Completed primary series cases: Cases are considered completed primary series if their positive COVID-19 test was 14 days or more after they received their 2nd dose in a 2-dose COVID-19 series or the single dose of a 1-dose vaccine. These are also called “breakthrough cases.”

On September 12, 2021, a new case definition of COVID-19 was introduced that includes criteria for enumerating new infections after previous probable or confirmed infections (also known as reinfections). A reinfection is defined as a confirmed positive PCR lab test more than 90 days after a positive PCR or antigen test. The first reinfection case was identified on December 7, 2021.

Data is lagged by eight days, meaning the most recent specimen collection date included is eight days prior to today. All data updates daily as more information becomes available.

B. HOW THE DATASET IS CREATED Case information is based on confirmed positive laboratory tests reported to the City. The City then completes quality assurance and other data verification processes. Vaccination data comes from the California Immunization Registry (CAIR2). The California Department of Public Health runs CAIR2. Individual-level case and vaccination data are matched to identify cases by vaccination status in this dataset. Case records are matched to vaccine records using first name, last name, date of birth, phone number, and email address.

We include vaccination records from all nine Bay Area counties in order to improve matching rates. This allows us to identify breakthrough cases among people who moved to the City from other Bay Area counties after completing their vaccine series. Only cases among San Francisco residents are included.

C. UPDATE PROCESS Updates automatically at 08:00 AM Pacific Time each day.

D. HOW TO USE THIS DATASET Total San Francisco population estimates can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS). To identify total San Francisco population estimates, filter the view on “demographic_category_label” = “all ages”.

Population estimates by vaccination status are derived from our publicly reported vaccination counts, which can be found at COVID-19 Vaccinations Given to SF Residents Over Time.

The dataset includes new cases, 7-day average new cases, new case rates, 7-day average new case rates, percent of total cases, and 7-day average percent of total cases for each vaccination category.

New cases are the count of cases where the positive tests were collected on that specific specimen collection date. The 7-day rolling average shows the trend in new cases. The rolling average is calculated by averaging the new cases for a particular day with the prior 6 days.

New case rates are the count of new cases per 100,000 residents in each vaccination status group. The 7-day rolling average shows the trend in case rates. The rolling average is calculated by averaging the case rate for a part

A. SUMMARY Medical provider confirmed COVID-19 cases and confirmed COVID-19 related deaths in San Francisco, CA aggregated by several different geographic areas and normalized by 2016-2020 American Community Survey (ACS) 5-year estimates for population data to calculate rate per 10,000 residents.

On September 12, 2021, a new case definition of COVID-19 was introduced that includes criteria for enumerating new infections after previous probable or confirmed infections (also known as reinfections). A reinfection is defined as a confirmed positive PCR lab test more than 90 days after a positive PCR or antigen test. The first reinfection case was identified on December 7, 2021.

Cases and deaths are both mapped to the residence of the individual, not to where they were infected or died. For example, if one was infected in San Francisco at work but lives in the East Bay, those are not counted as SF Cases or if one dies in Zuckerberg San Francisco General but is from another county, that is also not counted in this dataset.

Dataset is cumulative and covers cases going back to 3/2/2020 when testing began.

Geographic areas summarized are: 1. Analysis Neighborhoods 2. Census Tracts 3. Census Zip Code Tabulation Areas

B. HOW THE DATASET IS CREATED Addresses from medical data are geocoded by the San Francisco Department of Public Health (SFDPH). Those addresses are spatially joined to the geographic areas. Counts are generated based on the number of address points that match each geographic area. The 2016-2020 American Community Survey (ACS) population estimates provided by the Census are used to create a rate which is equal to ([count] / [acs_population]) * 10000) representing the number of cases per 10,000 residents.

C. UPDATE PROCESS Geographic analysis is scripted by SFDPH staff and synced to this dataset daily at 7:30 Pacific Time.

D. HOW TO USE THIS DATASET San Francisco population estimates for geographic regions can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS).

Privacy rules in effect To protect privacy, certain rules are in effect: 1. Case counts greater than 0 and less than 10 are dropped - these will be null (blank) values 2. Death counts greater than 0 and less than 10 are dropped - these will be null (blank) values 3. Cases and deaths dropped altogether for areas where acs_population < 1000

Rate suppression in effect where counts lower than 20 Rates are not calculated unless the case count is greater than or equal to 20. Rates are generally unstable at small numbers, so we avoid calculating them directly. We advise you to apply the same approach as this is best practice in epidemiology.

A note on Census ZIP Code Tabulation Areas (ZCTAs) ZIP Code Tabulation Areas are special boundaries created by the U.S. Census based on ZIP Codes developed by the USPS. They are not, however, the same thing. ZCTAs are areal representations of routes. Read how the Census develops ZCTAs on their website.

Row included for Citywide case counts, incidence rate, and deaths A single row is included that has the Citywide case counts and incidence rate. This can be used for comparisons. Citywide will capture all cases regardless of address quality. While some cases cannot be mapped to sub-areas like Census Tracts, ongoing data quality efforts result in improved mapping on a rolling basis.

E. CHANGE LOG

• 9/11/2023 - data on COVID-19 cases and deaths summarized by geography are no longer being updated. This data is currently through 9/6/2023 and will not include any new data after this date.
• 4/6/2023 - the State implemented system updates to improve the integrity of historical data.
• 2/21/2023 - system updates to improve reliability and accuracy of cases data were implemented.
• 1/31/2023 - updated “acs_population” column to reflect the 2020 Census Bureau American Community Survey (ACS) San Francisco Population estimates.
• 1/31/2023 - implemented system updates to streamline and improve our geo-coded data, resulting in small shifts in our case and death data by geography.
• 1/31/2023 - renamed column “last_updated_at” to “data_as_of”.
• 2/23/2022 - the New Cases Map dashboard began pulling from this dataset. To access Cases by Geography Over Time, please refer to this dataset.
• 1/22/2022 - system updates to improve timeliness and accuracy of cases and deaths data were implemented.
• 7/15/2022 - reinfections added to cases dataset. See section SUMMARY for more information on how reinfections are identified.
• 4/16/2021 - dataset updated to refresh with a five-day data lag.

Analysis of ‘Covid-19 Tests’ provided by Analyst-2 (analyst-2.ai), based on source dataset retrieved from https://catalog.data.gov/dataset/1115e734-6897-40e9-95ff-4274b5058ebf on 11 February 2022.

--- Dataset description provided by original source is as follows ---

Note: As of April 16, 2021, this dataset will update daily with a five-day data lag.

A. SUMMARY Case information on COVID-19 Laboratory testing. This data includes a daily count of test results reported, and how many of those were positive, negative, and indeterminate. Reported tests include tests with a positive, negative or indeterminate result. Indeterminate results, which could not conclusively determine whether COVID-19 virus was present, are not included in the calculation of percent positive. Testing for the novel coronavirus is available through commercial, clinical, and hospital laboratories, as well as the SFDPH Public Health Laboratory.

Tests are de-duplicated by an individual and date. This means that if a person gets tested multiple times on different dates in the last 30 days, all of those individual tests will be included in this data as individual tests (on each specimen collection date).

Total positive test results is not equal to the total number of COVID-19 cases in San Francisco. Each positive test result is investigated. During this investigation, some test results are found to be for persons living outside of San Francisco and some are duplicates of previously received positive results. These are not included in the total San Francisco COVID-19 case count. Additionally, investigation of positive test results might not be completed on the day of receipt; new cases will be added to the total case count after full investigation and verification.

B. HOW THE DATASET IS CREATED Laboratory test volume and positivity for COVID-19 is based on electronic laboratory test reports. Deduplication, quality assurance measures and other data verification processes maximize accuracy of laboratory test information.

C. UPDATE PROCESS Updates automatically at 05:00 Pacific Time each day. Redundant runs are scheduled at 07:00 and 09:00 in case of pipeline failure.

D. HOW TO USE THIS DATASET Due to the high degree of variation in the time needed to complete tests by different labs there is a delay in this reporting. On March 24 the Health Officer ordered all labs in the City to report complete COVID-19 testing information to the local and state health departments. In order to track trends over time, a data user can analyze this data by "result_date" and see how the count of reported results and positivity rate have changed over time.

--- Original source retains full ownership of the source dataset ---

A. SUMMARY This dataset includes San Francisco COVID-19 tests by race/ethnicity and by date. This dataset represents the daily count of tests collected, and the breakdown of test results (positive, negative, or indeterminate). Tests in this dataset include all those collected from persons who listed San Francisco as their home address at the time of testing. It also includes tests that were collected by San Francisco providers for persons who were missing a locating address. This dataset does not include tests for residents listing a locating address outside of San Francisco, even if they were tested in San Francisco.

The data were de-duplicated by individual and date, so if a person gets tested multiple times on different dates, all tests will be included in this dataset (on the day each test was collected). If a person tested multiple times on the same date, only one test is included from that date. When there are multiple tests on the same date, a positive result, if one exists, will always be selected as the record for the person. If a PCR and antigen test are taken on the same day, the PCR test will supersede. If a person tests multiple times on the same day and the results are all the same (e.g. all negative or all positive) then the first test done is selected as the record for the person.

The total number of positive test results is not equal to the total number of COVID-19 cases in San Francisco.

When a person gets tested for COVID-19, they may be asked to report information about themselves. One piece of information that might be requested is a person's race and ethnicity. These data are often incomplete in the laboratory and provider reports of the test results sent to the health department. The data can be missing or incomplete for several possible reasons:

• The person was not asked about their race and ethnicity.
• The person was asked, but refused to answer.
• The person answered, but the testing provider did not include the person's answers in the reports.
• The testing provider reported the person's answers in a format that could not be used by the health department.

For any of these reasons, a person's race/ethnicity will be recorded in the dataset as “Unknown.”

B. NOTE ON RACE/ETHNICITY The different values for Race/Ethnicity in this dataset are "Asian;" "Black or African American;" "Hispanic or Latino/a, all races;" "American Indian or Alaska Native;" "Native Hawaiian or Other Pacific Islander;" "White;" "Multi-racial;" "Other;" and “Unknown."

The Race/Ethnicity categorization increases data clarity by emulating the methodology used by the U.S. Census in the American Community Survey. Specifically, persons who identify as "Asian," "Black or African American," "American Indian or Alaska Native," "Native Hawaiian or Other Pacific Islander," "White," "Multi-racial," or "Other" do NOT include any person who identified as Hispanic/Latino at any time in their testing reports that either (1) identified them as SF residents or (2) as someone who tested without a locating address by an SF provider. All persons across all races who identify as Hispanic/Latino are recorded as “"Hispanic or Latino/a, all races." This categorization increases data accuracy by correcting the way “Other” persons were counted. Previously, when a person reported “Other” for Race/Ethnicity, they would be recorded “Unknown.” Under the new categorization, they are counted as “Other” and are distinct from “Unknown.”

If a person records their race/ethnicity as “Asian,” “Black or African American,” “American Indian or Alaska Native,” “Native Hawaiian or Other Pacific Islander,” “White,” or “Other” for their first COVID-19 test, then this data will not change—even if a different race/ethnicity is reported for this person for any future COVID-19 test. There are two exceptions to this rule. The first exception is if a person’s race/ethnicity value is reported as “Unknown” on their first test and then on a subsequent test they report “Asian;” "Black or African American;" "Hispanic or Latino/a, all races;" "American Indian or Alaska Native;" "Native Hawaiian or Other Pacific Islander;" or "White”, then this subsequent reported race/ethnicity will overwrite the previous recording of “Unknown”. If a person has only ever selected “Unknown” as their race/ethnicity, then it will be recorded as “Unknown.” This change provides more specific and actionable data on who is tested in San Francisco.

The second exception is if a person ever marks “Hispanic or Latino/a, all races” for race/ethnicity then this choice will always overwrite any previous or future response. This is because it is an overarching category that can include any and all other races and is mutually exclusive with the other responses.

A person's race/ethnicity will be recorded as “Multi-racial” if they select two or more values among the following choices: “Asian,” “Black or African American,” “American Indian or Alaska Native,” “Native Hawaiian or Other Pacific Islander,” “White,” or “Other.” If a person selects a combination of two or more race/ethnicity answers that includes “Hispanic or Latino/a, all races” then they will still be recorded as “Hispanic or Latino/a, all races”—not as “Multi-racial.”

C. HOW THE DATASET IS CREATED COVID-19 laboratory test data is based on electronic laboratory test reports. Deduplication, quality assurance measures and other data verification processes maximize accuracy of laboratory test information.

D. UPDATE PROCESS Updates automatically at 5:00AM Pacific Time each day. Redundant runs are scheduled at 7:00AM and 9:00AM in case of pipeline failure.

E. HOW TO USE THIS DATASET San Francisco population estimates for race/ethnicity can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS).

Due to the high degree of variation in the time needed to complete tests by different labs there is a delay in this reporting. On March 24, 2020 the Health Officer ordered all labs in the City to report complete COVID-19 testing information to the local and state health departments.

In order to track trends over time, a user can analyze this data by sorting or filtering by the "specimen_collection_date" field.

Calculating Percent Positivity: The positivity rate is the percentage of tests that return a positive result for COVID-19 (positive tests divided by the sum of positive and negative tests). Indeterminate results, which could not conclusively determine whether COVID-19 virus was present, are not included in the calculation of percent positive. When there are fewer than 20 positives tests for a given race/ethnicity and time period, the positivity rate is not calculated for the public tracker because rates of small test counts are less reliable.

Calculating Testing Rates: To calculate the testing rate per 10,000 residents, divide the total number of tests collected (positive, negative, and indeterminate results) for the specified race/ethnicity by the total number of residents who identify as that race/ethnicity (according to the 2016-2020 American Community Survey (ACS) population estimate), then multiply by 10,000. When there are fewer than 20 total tests for a given race/ethnicity and time period, the testing rate is not calculated for the public tracker because rates of small test counts are less reliable.

Read more about how this data is updated and validated daily: https://sf.gov/information/covid-19-data-questions

F. CHANGE LOG

• 1/12/2024 - This dataset will stop updating as of 1/12/2024
• 6/21/2023 - A small number of additional COVID-19 testing records were released as part of our ongoing data cleaning efforts. An update to the race or ethnicity designation among a subset of testing records was simultaneously released.
• 1/31/2023 - updated “population_estimate” column to reflect the 2020 Census Bureau American Community Survey (ACS) San Francisco Population estimates.
• 1/31/2023 - renamed column “last_updated_at” to “data_as_of”.
• 3/23/2022 - ‘Native American’ changed to ‘American Indian or Alaska Native’ to align with the census.
• 2/10/2022 - race/ethnicity categorization was changed. See section NOTE ON RACE/ETHNICITY for additional information.
• 4/16/2021 - dataset updated to refresh with a five-day data lag.

A. SUMMARY Case information on COVID-19 Laboratory testing. This data includes a daily count of test results reported, and how many of those were positive, negative, and indeterminate. Reported tests include tests with a positive, negative or indeterminate result. Indeterminate results, which could not conclusively determine whether COVID-19 virus was present, are not included in the calculation of percent positive. Testing for the novel coronavirus is available through commercial, clinical, and hospital laboratories, as well as the SFDPH Public Health Laboratory. Tests are de-duplicated by an individual and date. This means that if a person gets tested multiple times on different dates in the last 30 days, all of those individual tests will be included in this data as individual tests (on each specimen collection date). Total positive test results is not equal to the total number of COVID-19 cases in San Francisco. B. HOW THE DATASET IS CREATED Laboratory test volume and positivity for COVID-19 is based on electronic laboratory test reports. Deduplication, quality assurance measures and other data verification processes maximize accuracy of laboratory test information. C. UPDATE PROCESS Updates automatically at 05:00 Pacific Time each day. A redundant run is scheduled at 09:00 in case of pipeline failure. D. HOW TO USE THIS DATASET Due to the high degree of variation in the time needed to complete tests by different labs there is a delay in this reporting. On March 24 the Health Officer ordered all labs in the City to report complete COVID-19 testing information to the local and state health departments. In order to track trends over time, a data user can analyze this data by "result_date" and see how the count of reported results and positivity rate have changed over time. E. CHANGE LOG 4/10/2024 - An issue with our testing data was identified and corrected leading to a small increase in testing records over time. 6/21/2023 - A small number of additional COVID-19 testing records were released as part of our ongoing data cleaning efforts. 1/31/2023 - renamed column “last_updated_at” to “data_as_of”. 1/31/2023 - added columns “cumulative_tests”, “cumulative_positive_tests”, “cumulative_negative_tests”, “cumulative_indeterminate_tests”. 4/16/2021 - dataset updated to refresh with a five-day data lag.

A. SUMMARY This dataset contains COVID-19 positive confirmed cases aggregated by several different geographic areas and by day. COVID-19 cases are mapped to the residence of the individual and shown on the date the positive test was collected. In addition, 2016-2020 American Community Survey (ACS) population estimates are included to calculate the cumulative rate per 10,000 residents. Dataset covers cases going back to 3/2/2020 when testing began. This data may not be immediately available for recently reported cases and data will change to reflect as information becomes available. Data updated daily. Geographic areas summarized are: 1. Analysis Neighborhoods 2. Census Tracts 3. Census Zip Code Tabulation Areas B. HOW THE DATASET IS CREATED Addresses from the COVID-19 case data are geocoded by the San Francisco Department of Public Health (SFDPH). Those addresses are spatially joined to the geographic areas. Counts are generated based on the number of address points that match each geographic area for a given date. The 2016-2020 American Community Survey (ACS) population estimates provided by the Census are used to create a cumulative rate which is equal to ([cumulative count up to that date] / [acs_population]) * 10000) representing the number of total cases per 10,000 residents (as of the specified date). COVID-19 case data undergo quality assurance and other data verification processes and are continually updated to maximize completeness and accuracy of information. This means data may change for previous days as information is updated. C. UPDATE PROCESS Geographic analysis is scripted by SFDPH staff and synced to this dataset daily at 05:00 Pacific Time. D. HOW TO USE THIS DATASET San Francisco population estimates for geographic regions can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS). This dataset can be used to track the spread of COVID-19 throughout the city, in a variety of geographic areas. Note that the new cases column in the data represents the number of new cases confirmed in a certain area on the specified day, while the cumulative cases column is the cumulative total of cases in a certain area as of the specified date. Privacy rules in effect To protect privacy, certain rules are in effect: 1. Any area with a cumulative case count less than 10 are dropped for all days the cumulative count was less than 10. These will be null values. 2. Once an area has a cumulative case count of 10 or greater, that area will have a new row of case data every day following. 3. Cases are dropped altogether for areas where acs_population < 1000 4. Deaths data are not included in this dataset for privacy reasons. The low COVID-19 death rate in San Francisco, along with other publicly available information on deaths, means that deaths data by geography and day is too granular and potentially risky. Read more in our privacy guidelines Rate suppression in effect where counts lower than 20 Rates are not calculated unless the cumulative case count is greater than or equal to 20. Rates are generally unstable at small numbers, so we avoid calculating them directly. We advise you to apply the same approach as this is best practice in epidemiology. A note on Census ZIP Code Tabulation Areas (ZCTAs) ZIP Code Tabulation Areas are spec

Foodborne illness is prevented by inspection and surveillance conducted by health departments across America. Appropriate restaurant behavior is enforced and monitored via public health inspections. However, surveillance coverage provided by state and local health departments is insufficient in preventing the rising number of foodborne illness outbreaks. To address this need for improved surveillance coverage we conducted a supplementary form of public health surveillance using social media data: Yelp.com restaurant reviews in the city of San Francisco. Yelp is a social media site where users post reviews and rate restaurants they have personally visited. Presence of keywords related to health code regulations and foodborne illness symptoms, number of restaurant reviews, number of Yelp stars, and restaurant price range were included in a model predicting a restaurant’s likelihood of health code violation measured by the assigned San Francisco public health code rating. For a list of major health code violations see (S1 Table). We built the predictive model using 71,360 Yelp reviews of restaurants in the San Francisco Bay Area. The predictive model was able to predict health code violations in 78% of the restaurants receiving serious citations in our pilot study of 440 restaurants. Training and validation data sets each pulled data from 220 restaurants in San Francisco. Keyword analysis of free text within Yelp not only improved detection of high-risk restaurants, but it also served to identify specific risk factors related to health code violation. To further validate our model we applied the model generated in our pilot study to Yelp data from 1,542 restaurants in San Francisco. The model achieved 91% sensitivity 74% specificity, area under the receiver operator curve of 98%, and positive predictive value of 29% (given a substandard health code rating prevalence of 10%). When our model was applied to restaurant reviews in New York City we achieved 74% sensitivity, 54% specificity, area under the receiver operator curve of 77%, and positive predictive value of 25% (given a prevalence of 12%). Model accuracy improved when reviews ranked highest by Yelp were utilized. Our results indicate that public health surveillance can be improved by using social media data to identify restaurants at high risk for health code violation. Additionally, using highly ranked Yelp reviews improves predictive power and limits the number of reviews needed to generate prediction. Use of this approach as an adjunct to current risk ranking of restaurants prior to inspection may enhance detection of those restaurants participating in high risk practices that may have gone previously undetected. This model represents a step forward in the integration of social media into meaningful public health interventions.

Analysis of ‘COVID-19 Cases by Population Characteristics Over Time’ provided by Analyst-2 (analyst-2.ai), based on source dataset retrieved from https://catalog.data.gov/dataset/a3291d85-0076-43c5-a59c-df49480cdc6d on 13 February 2022.

--- Dataset description provided by original source is as follows ---

Note: On January 22, 2022, system updates to improve the timeliness and accuracy of San Francisco COVID-19 cases and deaths data were implemented. You might see some fluctuations in historic data as a result of this change. Due to the changes, starting on January 22, 2022, the number of new cases reported daily will be higher than under the old system as cases that would have taken longer to process will be reported earlier.

A. SUMMARY This dataset shows San Francisco COVID-19 cases by population characteristics and by specimen collection date. Cases are included on the date the positive test was collected.

Population characteristics are subgroups, or demographic cross-sections, like age, race, or gender. The City tracks how cases have been distributed among different subgroups. This information can reveal trends and disparities among groups.

Data is lagged by five days, meaning the most recent specimen collection date included is 5 days prior to today. Tests take time to process and report, so more recent data is less reliable.

B. HOW THE DATASET IS CREATED Data on the population characteristics of COVID-19 cases and deaths are from: * Case interviews * Laboratories * Medical providers

These multiple streams of data are merged, deduplicated, and undergo data verification processes. This data may not be immediately available for recently reported cases because of the time needed to process tests and validate cases. Daily case totals on previous days may increase or decrease. Learn more.

Data are continually updated to maximize completeness of information and reporting on San Francisco residents with COVID-19.

Data notes on each population characteristic type is listed below.

Race/ethnicity * We include all race/ethnicity categories that are collected for COVID-19 cases. * The population estimates for the "Other" or “Multi-racial” groups should be considered with caution. The Census definition is likely not exactly aligned with how the City collects this data. For that reason, we do not recommend calculating population rates for these groups.

Sexual orientation * Sexual orientation data is collected from individuals who are 18 years old or older. These individuals can choose whether to provide this information during case interviews. Learn more about our data collection guidelines. * The City began asking for this information on April 28, 2020.

Gender * The City collects information on gender identity using these guidelines.

Comorbidities * Underlying conditions are reported when a person has one or more underlying health conditions at the time of diagnosis or death.

Transmission type * Information on transmission of COVID-19 is based on case interviews with individuals who have a confirmed positive test. Individuals are asked if they have been in close contact with a known COVID-19 case. If they answer yes, transmission category is recorded as contact with a known case. If they report no contact with a known case, transmission category is recorded as community transmission. If the case is not interviewed or was not asked the question, they are counted as unknown.

Homelessness Persons are identified as homeless based on several data sources: * self-reported living situation
* the location at the time of testing * Department of Public Health homelessness and health databases * Residents in Single-Room Occupancy hotels are not included in these figures.
These methods serve as an estimate of persons experiencing homelessness. They may not meet other homelessness definitions.

Skilled Nursing Facility (SNF) occupancy * A Skilled Nursing

--- Original source retains full ownership of the source dataset ---

A. SUMMARY Medical provider confirmed COVID-19 cases and confirmed COVID-19 related deaths in San Francisco, CA aggregated by several different geographic areas and normalized by 2016-2020 American Community Survey (ACS) 5-year estimates for population data to calculate rate per 10,000 residents. On September 12, 2021, a new case definition of COVID-19 was introduced that includes criteria for enumerating new infections after previous probable or confirmed infections (also known as reinfections). A reinfection is defined as a confirmed positive PCR lab test more than 90 days after a positive PCR or antigen test. The first reinfection case was identified on December 7, 2021. Cases and deaths are both mapped to the residence of the individual, not to where they were infected or died. For example, if one was infected in San Francisco at work but lives in the East Bay, those are not counted as SF Cases or if one dies in Zuckerberg San Francisco General but is from another county, that is also not counted in this dataset. Dataset is cumulative and covers cases going back to 3/2/2020 when testing began. Geographic areas summarized are: 1. Analysis Neighborhoods 2. Census Tracts 3. Census Zip Code Tabulation Areas B. HOW THE DATASET IS CREATED Addresses from medical data are geocoded by the San Francisco Department of Public Health (SFDPH). Those addresses are spatially joined to the geographic areas. Counts are generated based on the number of address points that match each geographic area. The 2016-2020 American Community Survey (ACS) population estimates provided by the Census are used to create a rate which is equal to ([count] / [acs_population]) * 10000) representing the number of cases per 10,000 residents. C. UPDATE PROCESS Geographic analysis is scripted by SFDPH staff and synced to this dataset daily at 7:30 Pacific Time. D. HOW TO USE THIS DATASET San Francisco population estimates for geographic regions can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS). Privacy rules in effect To protect privacy, certain rules are in effect: 1. Case counts greater than 0 and less than 10 are dropped - these will be null (blank) values 2. Death counts greater than 0 and less than 10 are dropped - these will be null (blank) values 3. Cases and deaths dropped altogether for areas where acs_population < 1000 Rate suppression in effect where counts lower than 20 Rates are not calculated unless the case count is greater than or equal to 20. Rates are generally unstable at small numbers, so we avoid calculating them directly. We advise you to apply the same approach as this is best practice in epidemiology. A note on Census ZIP Code Tabulation Areas (ZCTAs) ZIP Code Tabulation Areas are special boundaries created by the U.S. Census based on ZIP Codes developed by the USPS. They are not, however, the same thing. ZCTAs are areal representations of routes. Read how the Census develops ZCTAs on their website. Row included for Citywide case counts, incidence rate, and deaths A single row is included that has the Citywide case counts and incidence rate. This can be used for comparisons. Citywide will capture all cases regardless of address quality. While some cases cannot be mapped to sub-areas like Census Tracts, ongo

Individual shelter participation per event and total census population. This data is shown in Fig 2B. Percentage of tests taken. Individual datapoints shown in Fig 2A. (XLSX)

A. SUMMARY This archived dataset includes data for population characteristics that are no longer being reported publicly. The date on which each population characteristic type was archived can be found in the field “data_loaded_at”.

B. HOW THE DATASET IS CREATED Data on the population characteristics of COVID-19 cases are from:  * Case interviews  * Laboratories  * Medical providers    These multiple streams of data are merged, deduplicated, and undergo data verification processes.  

Race/ethnicity * We include all race/ethnicity categories that are collected for COVID-19 cases. * The population estimates for the "Other" or “Multi-racial” groups should be considered with caution. The Census definition is likely not exactly aligned with how the City collects this data. For that reason, we do not recommend calculating population rates for these groups.

Gender * The City collects information on gender identity using these guidelines.

Skilled Nursing Facility (SNF) occupancy * A Skilled Nursing Facility (SNF) is a type of long-term care facility that provides care to individuals, generally in their 60s and older, who need functional assistance in their daily lives.  * This dataset includes data for COVID-19 cases reported in Skilled Nursing Facilities (SNFs) through 12/31/2022, archived on 1/5/2023. These data were identified where “Characteristic_Type” = ‘Skilled Nursing Facility Occupancy’.

Sexual orientation * The City began asking adults 18 years old or older for their sexual orientation identification during case interviews as of April 28, 2020. Sexual orientation data prior to this date is unavailable. * The City doesn’t collect or report information about sexual orientation for persons under 12 years of age. * Case investigation interviews transitioned to the California Department of Public Health, Virtual Assistant information gathering beginning December 2021. The Virtual Assistant is only sent to adults who are 18+ years old. https://www.sfdph.org/dph/files/PoliciesProcedures/COM9_SexualOrientationGuidelines.pdf">Learn more about our data collection guidelines pertaining to sexual orientation.

Comorbidities * Underlying conditions are reported when a person has one or more underlying health conditions at the time of diagnosis or death.

Homelessness Persons are identified as homeless based on several data sources: * self-reported living situation * the location at the time of testing * Department of Public Health homelessness and health databases * Residents in Single-Room Occupancy hotels are not included in these figures. These methods serve as an estimate of persons experiencing homelessness. They may not meet other homelessness definitions.

Single Room Occupancy (SRO) tenancy * SRO buildings are defined by the San Francisco Housing Code as having six or more "residential guest rooms" which may be attached to shared bathrooms, kitchens, and living spaces. * The details of a person's living arrangements are verified during case interviews.

Transmission Type * Information on transmission of COVID-19 is based on case interviews with individuals who have a confirmed positive test. Individuals are asked if they have been in close contact with a known COVID-19 case. If they answer yes, transmission category is recorded as contact with a known case. If they report no contact with a known case, transmission category is recorded as community transmission. If the case is not interviewed or was not asked the question, they are counted as unknown.

C. UPDATE PROCESS This dataset has been archived and will no longer update as of 9/11/2023.

D. HOW TO USE THIS DATASET Population estimates are only available for age groups and race/ethnicity categories. San Francisco population estimates for race/ethnicity and age groups can be found in a view based on the San Francisco Population and Demographic Census dataset. These population estimates are from the 2016-2020 5-year American Community Survey (ACS).

This dataset includes many different types of characteristics. Filter the “Characteristic Type” column to explore a topic area. Then, the “Characteristic Group” column shows each group or category within that topic area and the number of cases on each date.

New cases are the count of cases within that characteristic group where the positive tests were collected on that specific specimen collection date. Cumulative cases are the running total of all San Francisco cases in that characteristic group up to the specimen collection date listed.

This data may not be immediately available for recently reported cases. Data updates as more information becomes available.

To explore data on the total number of cases, use the ARCHIVED: COVID-19 Cases Over Time dataset.

E. CHANGE LOG

• 9/11/2023 - data on COVID-19 cases by population characteristics over time are no longer being updated. The date on which each population characteristic type was archived can be found in the field “data_loaded_at”.
• 6/6/2023 - data on cases by transmission type have been removed. See section ARCHIVED DATA for more detail.
• 5/16/2023 - data on cases by sexual orientation, comorbidities, homelessness, and single room occupancy have been removed. See section ARCHIVED DATA for more detail.
• 4/6/2023 - the State implemented system updates to improve the integrity of historical data.
• 2/21/2023 - system updates to improve reliability and accuracy of cases data were implemented.
• 1/31/2023 - updated “population_estimate” column to reflect the 2020 Census Bureau American Community Survey (ACS) San Francisco Population estimates.
• 1/5/2023 - data on SNF cases removed. See section ARCHIVED DATA for more detail.
• 3/23/2022 - ‘Native American’ changed to ‘American Indian or Alaska Native’ to align with the census.
• 1/22/2022 - system updates to improve timeliness and accuracy of cases and deaths data were implemented.
• 7/15/2022 - reinfections added to cases dataset. See section SUMMARY for more information on how reinfections are identified.

As of 9/12/2024, we will begin reporting on hospitalization data again using a new San Francisco specific dataset. Updated data can be accessed here.

On 5/1/2024, hospitalization data reporting will change from mandatory to optional for all hospitals nationwide. We will be pausing the refresh of the underlying data beginning 5/2/2024.

A. SUMMARY Count of COVID+ patients admitted to the hospital. Patients who are hospitalized and test positive for COVID-19 may be admitted to an acute care bed (a regular hospital bed), or an intensive care unit (ICU) bed. This data shows the daily total count of COVID+ patients in these two bed types, and the data reflects totals from all San Francisco Hospitals.

B. HOW THE DATASET IS CREATED Hospital information is based on admission data reported to the National Healthcare Safety Network (NHSN) and provided by the California Department of Public Health (CDPH).

C. UPDATE PROCESS Updates automatically every week.

D. HOW TO USE THIS DATASET Each record represents how many people were hospitalized on the date recorded in either an ICU bed or acute care bed (shown as Med/Surg under DPHCategory field).

The dataset shown here includes all San Francisco hospitals and updates weekly with data for the past Sunday-Saturday as information is collected and verified. Data may change as more current information becomes available.

E. CHANGE LOG

• 9/12/2024 -Hospitalization data are now being tracked through a new source and are available here.
• 5/1/2024 - hospitalization data reporting to the National Healthcare Safety Network (NHSN) changed from mandatory to optional for all hospitals nationwide. We will be pausing the refresh of the underlying data beginning 5/2/2024.
• 12/14/2023 – added column “hospitalreportingpct” to indicate the percentage of hospitals who submitted data on each report date.
• 8/7/2023 - In response to the end of the federal public health emergency on 5/11/2023 the California Hospital Association (CHA) stopped the collection and dissemination of COVID-19 hospitalization data. In alignment with the California Department of Public Health (CDPH), hospitalization data from 5/11/2023 onward are being pulled from the National Healthcare Safety Network (NHSN). The NHSN data is updated weekly and does not include information on COVID suspected (PUI) patients.
• 4/9/2021 - dataset updated daily with a four-day data lag.