Welcome!

The Center for Investigative Reporting has led the nation's coverage on massive backlogs and delays in processing veterans' disability claims by the Department of Veterans Affairs. We've developed a database -- which is periodically updated -- that includes various measurements of the amount of time each of the country's 58 regional office is taking to adjudicate pending claims. The data comes from internal documents leaked to CIR, documents obtained through Freedom of Information requests and publicly accessible documents hosted by The Department of Veterans Affairs.

If you use this data in any way, please credit The Center for Investigative Reporting in your product with a link back to our website at cironline.org

Questions

If you still have questions after reading this, email devdata@cironline.org

How to get the data

There are a few ways you can get ahold of the data:

1. Directly accessing the API (see more about how to do this in in the URLs and schema section). This data is availble in JSON and JSONP formats.

curl http://vetsapi.apps.cironline.org/api/data/?format=json

1. Download the spreadsheets hosted on Amazon s3. See section on CSV data

http://vbl-media.s3.amazonaws.com/data/[CITY-SLUG]-[FIELD-TYPE-SLUG].csv

1. Use our code to create your own apps. We've created a few Backbone.js collections and models that you can use. An example can be found here

URLs and schema

The data is organized as three different models: A time-series data model, a city model (technically a regional office, more later) representing the location of the data model, and field type model to differentiate between the different data types.

The exact fields follow, '*' indicates exceptions for this field type, please see data notes:

class FieldType(models.Model):
    '''
    10 field types exist
                    name                 |                slug                 
    -------------------------------------+-------------------------------------
     Completed Claims                    | completed-claims
     Average Processing Time             | average-processing-time
     Appealed Claims                     | appealed-claims
     Claims Completed per FTE            | claims-completed-per-fte
     Employees on duty                   | employees-on-duty
     Claims pending at least one year    | claims-pending-at-least-one-year
     Claims received                     | claims-received
     *Claims received average wait        | claims-received-average-wait
     *Pending Claim                       | pending-claim
     Claims Pending >= 125 Days          | claims-pending-125-days
    '''

    name = models.CharField(max_length=255)
    aspire_title = models.CharField(max_length=255)
    slug = AutoSlugField(populate_from=('name',))

url = http://vetsapi.apps.cironline.org/api/field-type/

class TimeSeriesData(models.Model):
    city = models.ForeignKey('collector.City')
    field_type = models.ForeignKey('collector.FieldType')
    value = models.FloatField()
    date = models.DateField()
    created = models.DateTimeField()

url = http://vetsapi.apps.cironline.org/api/data/

class City(models.Model):
    '''
    NOTE: Cities represent regional offices
    Generally speaking, there is one regional office per state
    serving all veterans.
    This is not always the case. Large states like
    California and Texas have multiple regional offices.
    If you'd like to look up the regional office that serves your area
    go to: http://www2.va.gov/directory/guide/home.asp?isflash=1
    '''
    name = models.CharField(max_length=255)
    slug = AutoSlugField(populate_from=('name',))

url = http://vetsapi.apps.cironline.org/api/city/

Querying the API and examples

Note: we use Tastypie to expose our data sets. This tutorial is brief and you can find more information about querying the data in Tastypie's documentation.

Accessing any of the above URLs will return a paginated list of data. Those lists can be filtered by adding parameters to the end of the url. For instance, maybe you'd like to see data points concerning the average time a veteran waits for a response from the VA, in your console type:

curl 'http://vetsapi.apps.cironline.org/api/data/?format=json&field_type__slug=average-processing-time'

Which will return the 20 most recent average processing time data points. To get the next 20 items issue the following command:

curl 'http://vetsapi.apps.cironline.org/api/data/?format=json&field_type__slug=average-processing-time&offset=20'

Or maybe you'd like see the average processing time for the regional office (city) in Portland, OR:

curl 'http://vetsapi.apps.cironline.org/api/data/?format=json&field_type__slug=average-processing-time&city__slug=portland'

You can programatically access the API as follows:

Python using the requests library and simplejson

import requests
import simplejson

response = requests.get('http://vetsapi.apps.cironline.org/api/city/')
cities = simplejson.loads(response.content)
for city in cities['objects']:
    response = requests.get('http://vetsapi.apps.cironline.org/api/data/?city__slug=%s' % city['slug'])
    data = simplejson.loads(response.content)
    print data

For examples in Javascript, please see index.html

CSV data

While the API provides more flexible access to the data it can be slow. For more reliable, fast access to the data we baked it all into CSV files and stored them on S3. Access to the data is on a per city, per field type basis much like the api though the format differs slightly. For instance, to get a list of all cities go to the following url:

http://vbl-media.s3.amazonaws.com/data/cities.csv

Now, all the spreadsheets are segmented by city for each of the 14 field types. Following the example above, if you wanted to view all data points by field type 'average processing time' for the regional office in Portland, OR., look at the following url:

http://vbl-media.s3.amazonaws.com/data/portland-average-processing-time.csv

The downside to using CSV data is no aggregate file for all data points about 'average processing time', or any other field type. To obtain the entire data set for a given field type, iterate over the list of cities requesting each file for the given field type.

As noted, the url to access CSV documents follows this format:

http://vbl-media.s3.amazonaws.com/data/[CITY-SLUG]-[FIELD-TYPE-SLUG].csv

We've created two special Backbone.js classes to help you query and parse CSV data in your web browser. See them here

Data updates

Fields accessible via publicly accessible web sites are updated weekly and monthly. The VA does not always publish data updates when they are expected so these intervals cannot be guaranteed.

• Pending claims: updates weekly
• Claims older than 125 days: updates weekly
• Average processing time: updates monthly

All other fields were obtained through Freedom of Information Act requests or through internal documents turned over to CIR and verified by the VA. Updates will be published as we receive and validate them.

Links

http://www.vba.va.gov/REPORTS/mmwr/index.asp

http://www.app.hospitalcompare.va.gov/index.cfm?org=vha

http://cironline.org/veterans

http://cironline.org/reports/map-where-veterans-backlog-worst-3792

Data notes

*'Pending Claims' and 'claims received average wait' are referred to as 'Veterans waiting on a disability claim' and 'Average wait for new claims," respectively.

• *Veterans waiting on a disability claim = Pending Claims: The number of veterans waiting for a response from the VA for compensation for a disease, injury or illness linked to service in the military. Nationally, this number also includes about 10,000 survivors and other family members seeking compensation related to service-related injuries and diseases.
• Average processing time: The average number of days a veteran waits for a decision from the VA.
• *Average wait for new claims = Claims received average wait: The average number of days a veteran filing a claim for the first time waits for a response from the VA.
• Average time to decide an appeal: The average number of days a veteran waits for a response from the VA if they were denied their original claim and had to appeal.
• Completed claims: The number of claims processed by the VA by month.
• Claims received: The number of claims received by the VA by month.
• Claims completed per FTE: The number of claims processed per VA claims employee over the course of a year.
• Employees on duty: Number of claims staff working at the Veterans Service Center at each VA regional office.
• Claims pending >= one year: The number of unprocessed claims at least a year old, including appeals.