1. Drownings avoided with life jackets: model development
Presentation by Dr. L. Daniel Maxim

2. Disclaimer
Contents of brief represent potentially useful ideas—work in progress (WIP)—not necessarily a finished product
The contents of this briefing have not yet been approved by competent USCG authority
Your ideas welcome
Written working paper available

3. Outline of presentation
Need for valid model
Model development
Numerical illustration for open motorboats
Parameter estimates
Uncertainty
The way forward

4. Need for model
Generally accepted that life jackets can avoid drownings, but only if worn…
Limited success with voluntary initiatives has prompted USCG to consider options for mandatory life jacket wear
Regulatory proposals must be supported by valid analyses and may be subject to challenge

5. …but only if worn
Many accidents involve immersion without the opportunity to don a lifejacket—or the opportunity to retrieve one from the boat!
ROGUE_WAVE_CLAIMS_CAPTAINS_LIFE_112210_SP.

6. Why care about drownings?
Single largest cause of US recreational boating fatalities
Drownings account for ~ 70% of US recreational boating fatalities (2008 data shown at right)

7. Drownings trends: slight downtrend, but still very important

8. Model antecedents
Similar to those developed and routinely employed by NHTSA for motorcycle helmets, seat belts, and air bags
Life jacket model reflects similarities and differences from helmets

9. Model development

10. Model structure: logical possibilities
Number of potential drownings
Wear life jacket
Do not drown
Drown
Don’t wear life jacket

11. Logical possibilities: focus on wearers
Number of potential drownings
Wear life jacket
Do not drown
Drown

12. Logical possibilities: focus on wearers
Generally accepted that lifejackets prevent drownings, so the probability of survival given immersion is enhanced
Still, lifejackets do not guarantee that drownings are avoided, for reasons shown following

13. Why might people drown who wear a life jacket?
Illustrative reasons
The life jacket worn was not the right size, the right type for the sea conditions, was old, worn out, broken (e.g., malfunction of CO2 system), torn, or worn improperly
The vessel capsized and the wearer was trapped inside because egress was impossible or the life jacket actually prevented egress
The victim drowned from wave splash (mouth immersions) before rescue
The boater received an injury or impairment that, by itself, didn't kill the boater, but was severe enough to prevent the boater from doing those things necessary to keep the boater’s face out of the water or prevent what are termed "mouth immersions" from restricting the victim’s airway over time

14. Improper wearing of life jacket

15. Improper wearing of life jacket

16. How captured in model
Define g as probability that person who wears life jacket does not drown
Therefore, 1 – g is probability that person who wears life jacket does drown
Probability g difficult to estimate empirically, but presumably “relatively high”
NTSB estimated that 85% of those who drowned would have survived if wearing life jacket—by this estimate, g = 0.85

17. Logical possibilities: non wearers
Number of potential drownings
Don’t wear life jacket
Do not drown
Drown

18. Why might people survive if not wearing life jacket?
Reasons
The boater was able to find some other object (e.g., stayed with the boat, found a life ring) that was able to be used for floatation
The potentially fatal mishap occurred in an area sufficiently close to the shore that the boater was able to swim to safety
The potentially fatal mishap occurred in sufficiently close proximity to one or more other vessels that they were able to rescue the potential victim
Continued emphasis on swimming skills important

19. Survival without life jacket

20. How captured in model
Define k as probability that person who does not wear a life jacket yet does not drown
Therefore, 1 – k is probability that person who does not wear a life jacket and does drown
Probability k more difficult to estimate directly, but presumably lower than g
Estimation of k illustrated in later slides

21. Logical possibilities: focus on number of potential drownings

22. Number of potential drownings, F
Not known, a priori
But drownings result from boat occupants entering the water as a result of capsizing, falls overboard, and flooding/swamping
Not all potential fatalities reported because some are rescued or are able to effect self rescue and may not be involved in reportable accident
Therefore, F not measured in BARD and needs to be estimated by indirect means (see other slides)

23. BARD 2008 Drownings
According to data from BARD in 2008, capsizing, falls overboard, and flooding or swamping, accounted for 400 (78.4%) of the 510 reported drowning fatalities

24. Capsizing, falls overboard, and flooding or swamping

25. Number of potential drownings, F
Notation
Number of potential drownings, F
Wear life jacket
Do not drown
Drown
Don’t wear life jacket g S1 u D1
1 - g k S2
1 - u D2
1 - k

26. Data
BARD provides annual data on number who drowned wearing life jackets (D1) and number who drowned not wearing life jackets (D2)
JSI provides data on wear rates (u)
NTSB and others provide estimates of g
How can these inputs be used to estimate drownings avoided?

27. Numerical example: open motorboats

28. Open motorboats—why?
Drownings on open motorboats accounted for 252 (49%) of the 510 reported drownings in 2008
Including the four boat types with the highest number of reported drownings—open motorboats, canoes, rowboats, and kayaks—the total drownings in 2008 among these types were 391, 76.7% of the total drownings in that year

29. Sequence of calculations
Number of potential drownings, F
Wear life jacket
Do not drown
Drown
Don’t wear life jacket
93.3
0.85
0.179
521.4 14
0.15

30. Number of potential drownings, F
Filling in the blanks
Number of potential drownings, F
Wear life jacket
Do not drown
Drown
Don’t wear life jacket
93.3
0.85 79
0.179
F = 521.4 14
0.15
0.449
192
0.821
236
0.551
428.1

31. What if: wear rate u had been higher?
If u were to equal 1 (100% wear), then D1 would equal 78, and 252 – 78 = 174 drownings would be avoided
Potential drownings avoided can be calculated for any assumed u
Easy calculation using spreadsheet software

32. Uncertainty analysis of wear rate
Wear rate with mandatory PFD uncertain, but likely to be less than 100%
Drownings avoided at u = 0.7 (USACE estimate) ≅ 109 annually
Estimate uncertain, but clearly much greater than drownings avoided under present Strategic Plan

33. Parameter estimates: detailed model

34. NTSB authoritative and applies to overall US data
What about g or k?
Selected estimates g k
Source
0.93 NR
Lunetta et al. (1998) Finland
0.90
Oregon State Marine Board
0.85
NTSB (1993)
0.75
MCA (UK) also MSA (NZ)
0.74
Browne et al. (2003) New York
0.68
0.50
Australia (1992 – 1998)
0.55
0.42
Cummings et al. (2010)
0.38
0.12
Hudson and Conway (2003) fishermen in Alaska
NTSB authoritative and applies to overall US data

35. NTSB Study Analysis of data from 18 states; ≅ 52% of fatalities
Considered 281 drownings where boater not wearing PFD
Experts concluded 85% might have survived if wearing PFD
Provisionally recommended for use with detailed model

36. The way forward

37. Technical Apply to multi-year data sets
Apply to other types of craft (i.e., canoes, kayaks, and rowboats)
Additional research on some technical issues
Rewrite working paper for less technical audience?
Thing to avoid—micro analysis of data

38. Some technical work already done
Figure at left shows D1 and D2 values for open motorboats for five-year period
Numbers relatively stable
Use of five year averages changes estimates slightly

39. Some technical work already done
Excluding boats with unknown length, 86.4% of 2008 drownings occurred on boats ≤ 21 ft long
Therefore, focus of effort should be on smaller craft
Type probably better cut than length, but correlated

40. Drownings “the big four”

41. Focus

42. Technical analysis: recommendations
Address “big four” (open motorboats, canoes, kayaks, and rowboats) using simplest model, but focus on open motorboats
Avoid further disaggregation (e.g., time of day, season) because of sample size issues
Need to ensure that correct values of u are used from JSI report
Preliminary answers on next slide assuming u = 0.7

43. Illustrative results Quantity Open Motorboat Canoe Kayak Rowboat Total
Drownings
Avoided
Base
case D1
115 34 68 10 D2
1323
340
106
214 D
1438
374
174
224
Years 6
D per year
239.7
62.3
29.0
37.3 u
0.177
0.312
0.739
0.371
New u1
0.7
New D1
456.09
76.28
18.87
New D2
481.97
148.26
102.07
New D
938.06
224.54
120.93
156.34
37.42
20.16
Drownings avoided
83.32
24.91
17.18
125.41
(34%)
Illustrative results

44. Status Model and calculations reviewed and deemed appropriate by NTSB
Work group formed to consider way forward including Dr. Linda Quan, NTSB, and Coast Guard personnel
Working paper written and slight revisions made as result of input from group

45. Way forward This analysis preliminary and limited to “science” issues
If mandatory life jacket wear to be considered, then a cost-benefit analysis will be required
The benefit side of the equation obviously includes lives saved, but may include other benefits (e.g., near drownings avoided)

46. Way forward
Proposed regulation needs to be justified using content and format in accord with OMB Circular A-4 (among other things)
Other issues need to be addressed, e.g.,
What are likely compliance rates?
Need to limit potential lives saved to “waters subject to the jurisdiction of the United States”
What costs are appropriate to include?

47. Your ideas?
Write in this space