1. Lessons from one year's worth of HAZMAT
headlines
Ralph Stuart, CIH <secretary@dchas.org>
August 30, 2011
2. The Project
ā¤ Beginning February 1, 2010, I have been using Google reader to
collect newspaper headlines in English from the global press.
ā¤ As of August 20, 2011, I have collected 2696 events that google
identiļ¬ed as containing these key words:
ā¤ āhazmatā
ā¤ āchemicalā and (āļ¬reā or āexplosionā)
ā¤ ālaboratoryā and (āļ¬reā or āexplosionā or āaccidentā or āinjuryā)
ā¤ There is a lot of duplication of stories; I use the ļ¬rst one that I ļ¬nd on
a particular event.
4. Why?
ā¤ The purpose is to provide context to headline grabbing events such as
the UCLA pyrophoric incident, Texas Tech laboratory explosion, and
Yale machine shop death
5. Why?
ā¤ The purpose is to provide context to headline grabbing events such as
the UCLA pyrophoric incident, Texas Tech laboratory explosion, and
Yale machine shop death
ā¤ The core question in my mind is: Can we tell if the safety performance of
laboratories is different from other parts of the economy?
6. Why?
ā¤ The purpose is to provide context to headline grabbing events such as
the UCLA pyrophoric incident, Texas Tech laboratory explosion, and
Yale machine shop death
ā¤ The core question in my mind is: Can we tell if the safety performance of
laboratories is different from other parts of the economy?
ā¤ The intent is not statistical (because of the many ļ¬lters between an
event and the press as well as between the press and the readers), but
rather to identify stories with learning opportunities, as well as
keeping the CH&S community abreast of events and trends of
interest.
7. Why?
ā¤ The purpose is to provide context to headline grabbing events such as
the UCLA pyrophoric incident, Texas Tech laboratory explosion, and
Yale machine shop death
ā¤ The core question in my mind is: Can we tell if the safety performance of
laboratories is different from other parts of the economy?
ā¤ The intent is not statistical (because of the many ļ¬lters between an
event and the press as well as between the press and the readers), but
rather to identify stories with learning opportunities, as well as
keeping the CH&S community abreast of events and trends of
interest.
ā¤ However, some numbers provide helpful context for an individual
report.
8. How?
ā¤ The Google Newsreader service identiļ¬es news stories on the web
that contain speciļ¬c keywords
ā¤ Pinboard is a web site bookmarking service that allows you to tag the
a web site of interest with keywords for future review and sorting
ā¤ I review the stories and classify them based on:
ā¤ Location
ā¤ Economic sector (industrial, transportation, public, lab, other)
ā¤ Type of event (explosion, ļ¬re, release, discovery)
ā¤ Extent of damage (response, injury, death, follow-up)
ā¤ Primary Chemical Involved
ā¤ I exclude āwhite powderā and fuel releases during normal trafļ¬c
accidents.
12. The Electronic Advantage
ā¤ Dr. Rob Toreki, president of ilpi.com, inspired by human-based inconsistencies in the
reports, offered to automate the process starting in about June of 2010. The reļ¬nement
of this intelligence engine is ongoing.
ā¤ This was done with custom javascript programming and organizing the digests for
DCHAS-L via a cron job.
ā¤ Consistent data entry is the core advantage - suggested tags are easier to generate and
avoid misspelling
ā¤ Much easier identiļ¬cation of location of the event
ā¤ Learning as time goes by gets encoded into the system
ā¤ It takes about 15 minutes per day for someone with both hazmat and computer
experience
13. The Results:
How often do events occur?
Month Events/day
2/28/10 3.7
3/31/10 3.4
4/30/10 2.3
5/29/10 4.5
6/30/10 4.3
7/31/10 3.7
8/31/10 4.4
9/30/10 4.9
10/31/10 6.3
11/28/10 4.5
12/30/10 4.6
1/31/11 4.9
2/28/11 5.9
3/31/11 5.3
4/30/11 5.2
14. Results: Where are they reported?
Country Percent reported
United States 76%
United Kingdom 6%
India 4%
Canada 4%
Australia 3%
China 1%
New Zealand 1%
15. Results: Where are they reported?
Country Percent reported
United States 76% An example of the
āheadlinesā ļ¬lter:
United Kingdom 6%
Of 171 Death Events:
India 4% US reported 80 (46%)
India reported 27 (16%)
Canada 4%
China reported 15 (9%)
Australia 3%
China 1%
New Zealand 1%
16. Results:
What Sector was Involved?
Percent of Events
Sector
Reports
Industrial 37%
Transportation 19%
Home 17%
Other 9%
Laboratory 9%
Illegal 5%
Education 4%
17. Results: What Happened?
Type of Percent of
event total
Release 54%
Fire 23%
Explosion 16%
Discovery 7%
18. Results: How Bad Was It?
Percent
Extent
reported
Response 71%
Injury 30%
Death 9%
19. Results: What Chemicals?
Chemical Percent
other 40%
unknown_chemical 13%
petroleum 7%
meth_lab 6%
acid 6%
ammonia 4%
solvent 3%
ag_chemicals 3%
wastes 3%
chlorine 3%
explosives 3%
pool_chemicals 2%
21. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
22. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
ā¤ Thereās a reason for the regulations.
23. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
ā¤ Thereās a reason for the regulations.
ā¤ A big event can develop from a small risk.
24. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
ā¤ Thereās a reason for the regulations.
ā¤ A big event can develop from a small risk.
ā¤ Information moves in odd ways.
25. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
ā¤ Thereās a reason for the regulations.
ā¤ A big event can develop from a small risk.
ā¤ Information moves in odd ways.
ā¤ People learn from stories; DCHAS-L follow up discussion is unpredictable
26. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
ā¤ Thereās a reason for the regulations.
ā¤ A big event can develop from a small risk.
ā¤ Information moves in odd ways.
ā¤ People learn from stories; DCHAS-L follow up discussion is unpredictable
ā¤ The press isnāt great, but isnāt bad, with chemical names; itās worse with
follow up.
27. Some Lessons I Draw
ā¤ HAZMAT happens - we should learn from it.
ā¤ Thereās a reason for the regulations.
ā¤ A big event can develop from a small risk.
ā¤ Information moves in odd ways.
ā¤ People learn from stories; DCHAS-L follow up discussion is unpredictable
ā¤ The press isnāt great, but isnāt bad, with chemical names; itās worse with
follow up.
ā¤ No trends are evident over the year, but copy catting happens (meth labs,
bottle bombs or suicides) - whether this is individuals copying each other or
the press copying itself is not clear.
28. Moving Forward
ā¤ Managing free-form chemical
information electronically presents
interesting opportunities in helping
laboratory workers conduct risk
reviews
ā¤ You can use the data yourself at
http://www.pinboard.in/u:dchas
ā¤ Or contact me for an Excel
spreadsheet with the information in
it.