Burning questions for PF&R

[futurechris]

Someone else's pun.

Feel free to add comments with any domain knowledge questions you would like answered by our PF&R rep Mark Whitaker.

Once we get this block answered, the answers should be moved to the wiki and a new ticket should be created for the next round of questions.

[hannahkane]

What are the sources of calls? Do they all come in via 911? How is it determined that a call goes to the Fire Dept?

[sky-t]

In the inctimes table, some incident_ids have repeated timedesc_ids. Why do these occur? For instance, incident_id 1102989 has 3 timedesc_id labeled 3. Sometimes the timestamps are the same among all the rows and other times, they are different. Knowing why they occur will help us understand how to treat these rows.

[markawhitaker]

@hannahkane Almost all calls come through 9-1-1. The dispatchers at the Bureau of Emergency Communications (BOEC) are trained to determine whether the call should go to police, fire, or ambulance or some combination of those. There are some incidents where a patient may just show up at the fire station or where the engine may be out driving around and spot a fire or car accident. Also after initial dispatch from BOEC, PF&R may dispatch additional resources to the scene if necessary. There is a field in the Incident table with a code for how the call was received--it's not one we use much, but I can provide more details if there's interest

[markawhitaker]

@sky-t The simplest answer is that some incidents have more than one responder. So if both a fire engine and an ambulance are dispatched to a scene, then there will be a dispatch time for the engine and for the ambulance, each in a separate row. The specific apparatus for each row is identifid by the responder_id field in the inctimes table. If the responder_id field is blank that refers to an action/timestamp that is performed by BOEC. I'll try to prepare some more notes on inctimes and some of the other tables for tomorrow night's meeting. If I get that done, I'll post them here or whatever the appropriate place might be

[hannahkane]

What is the difference between TimeDesc_ID 0 and TimeDesc_ID 1? Why do some entries not have a 1?

[futurechris]

Augmenting this with a bit from Mark @ 1/25 meeting. I asked about situations we'd observed in the data where two separate inctimes were very distant, e.g. first one was in September 2015, second was January 2016. Mark explained that this would depend on the responder unit, but could be explained by a follow-up investigation, etc. Or, could be erroneous data. Case-by-case.

We may want to do some exploration and decide what to do with these rows.

[futurechris]

What data does the public education dept. have about # of homes with a smoke and/or carbon monoxide alarm?

[markawhitaker]

@futurechris Every time there is a structure fire, the fire report is supposed to include a section on whether or not the structure had a smoke detector and how it performed in the fire. That is the only data we collect on smoke detectors according to our Fire Marshal. I believe this data contributed to this article: http://www.oregonlive.com/portland/index.ssf/2015/04/smoke_alarms_missing_in_30_of.html

I've uploaded all the data related to structure fires, detectors, and associated injuries to the Google Drive in a new folder called Smoke Detector Data that should allow someone to create something similar to the Oregonian article above. There could be interesting findings about specific neighborhoods that regularly don't have smoke alarms or types of buildings that don't. I'm not sure if there are enough observations to draw conclusions (we don't have that many fires), but I have not spent much time with this data.

[futurechris]

@markawhitaker, we are grabbing census data for the following categories, to include in the FMA dashboard:

• Sex, Age, Race, Language, Household Income, Types of health insurance

Would you also be interested in any of these?

• Occupation (broad categories)
• Geographical mobility (whether they’ve moved in the past year)
• Travel time to work (can be broken down by mode of transportation, too)
• Mode of transportation (to work)
• Time leaving home to go to work
• Number of children in household
• Households with one or more non-relatives
• Households with people over 60
• Educational attainment for the population 25 years and over
• Limited English speaking households
• Poverty status
• Per capita income
• Veteran status
• Food stamps / SNAP
• Vacancy status
• Number of rooms
• Units in structure
• Year structure built

[markawhitaker]

Others of interest would include: geographical mobility, households with people over 60, educational attainment, limited english speaking households, poverty status, and food stamps. Year structure built might also be interesting from a fire prevention stand point, but I'm thinking the City might have better data on that than the Census. Thanks!

[hannahkane]

@markawhitaker - Cat's here and she's wondering if we have any data about when an incident involves a death?

[hannahkane]

@markawhitaker - another question - do we track whether the incident involves children?

[markawhitaker]

@hannahkane We keep track of when an incident involves a fire death. We do not have data on EMS incidents and deaths.

We also collect ages for each patient involved in an EMS call.

[markawhitaker]

@hannahkane To follow up, if you're interested in patient ages and/or fire deaths, I can provide tables that link them to the IncidentID.

[hannahkane]

thanks, @markawhitaker. Let me ask the team about it tonight and we'll get back to you.

[hannahkane]

@markawhitaker - if it's relatively simple to get that data, we'd like to take a look at it.

[markawhitaker]

@hannahkane I'm putting an Excel file on the google drive with two tabs:

1) Fire Deaths: Incident_ID and FireDeaths. Pretty straight forward with the Incident ID and the number of fire deaths at that incident. There were 37 fire deaths in Portland between 2010 and 2016, and there are 36 rows here because one incident had two fire deaths.

2) Patient Age: Incident_ID, Age, and Gender. Some Incident IDs will appear more than once, indicating that the incident had multiple patients. These records only show those incidents for which a full patient care report was filled out so it does not necessarily represent every medical incident that PF&R responded to. It includes all patient ages for incidents from CY 2010 through CY 2016. There are some null values, representing patient care reports where the age field was left blank.

[hannahkane]

thanks, @markawhitaker!

[hannahkane]

hey @markawhitaker - we're wondering what "good intent" calls are. can you help us out?

[hannahkane]

Also, we'd like to know what "service calls," "natural conditions," and "mutual aid response" are. Thank you!

[futurechris]

@markawhitaker Relatedly, we're wondering how PF&R defines medical vs. not. We split calls into classes and "medical aid / rescue calls" are pretty distinct, but do you lump "good intent", "false calls", "service calls", "fire/explosion", and "hazardous conditions" under "fire"?

I.e., when you say "X% of our incidents are actually medical calls", do you then say "100-X% are fire", or ... How do you present that?

We're trying to compute statistics for fire vs. medical, and we're not sure how to split them up. I imagine PF&R would prefer stats for each category, but in terms of presenting to the public, we've been thinking "fire vs. medical" is a good split... Just not sure how to define that split.

Any input would be helpful, thanks!

[futurechris]

@markawhitaker Another question: What is the distinction between 'Battalion Chief' and 'Chief', in unit types?

We know there are 4 BCs on duty at any time, but haven't been able to get an understanding of what role 'Chiefs' play.

[markawhitaker]

@hannahkane Good intent calls generally cover those calls where it turns out that there is not really an emergency. Some examples include: arriving and not finding a victim or incident; controlled, legal burns; steam or gas mistaken for smoke; an expected hazmat incident where it turns out the material is harmless.

Service calls cover a lot of the services that firefighters provide outside of fire, medical, rescue, and hazmat. Some of the common ones include: water problems such as a water main break, street flooding, or basement flooding; animal rescue; assisting police; unauthorized burning; lockouts; and even ring or jewelry removal.

Natural conditions: These should be rare, but are responses to severe weather and natural disasters: windstorms, earthquakes, lightning strike, other severe weather. These get tricky because often there is another condition present as well. For example, if a windstorm knocks down a power line: a downed power line should be a hazardous condition, but it could also be classified as a natural condition since its cause was the windstorm. I'm not sure how consistent we are on these.

Mutual Aid: When PF&R leaves our service area to assist a neighboring jurisdiction. Once again, there is inconsistency here. If PF&R goes to help Vancouver fight a commercial fire, it could be recorded as mutual aid, but it could also just be recorded as a commercial fire.

[markawhitaker]

@futurechris For classifying calls, we generally provide three categories: fire, ems/rescue, and other. So the other category would include the hazmat, good intent, false calls, etc. Depending on whether you're basing it on initial dispatch or situation found, the percentage breakdowns differ.

On dispatch, initial type code the split is roughly: 13% fire, 78% medical/rescue, 9% other.

Based on situation found, it's more like 4% fire, 63% medical/rescue, and 33% other.

[markawhitaker]

@futurechris the BC is part of the on-duty staffing and is largely expected to take over incident command on large incidents. For example, all fires are supposed to have a BC on scene and the BC is considered part of the effective response force.

Chief responses are much more variable, I generally drop them from the analysis. On very large incidents, they'll set up incident command and expect all chiefs to be there: on-duty deputy chief, Fire Chief, Fire Marshal, Emergency Operations Chief, and Training Chief. Other calls may require the Safety Chief. Response time isn't usually important for these types of chiefs.