Escolar Documentos
Profissional Documentos
Cultura Documentos
2014–10:44am] [1–7]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/MSLJ/Vol00000/140015/APPFile/SG-MSLJ140015.3d (MSL) [PREPRINTER stage]
Technical report
John Searle1
Abstract
A dilution model is widely used to link blood alcohol concentration and the quantity of alcohol consumed. Whilst some
authors use the total body water formulation of that model, others use the Widmark Factor formulation. A paper by
Forrest gave a table of example values of the Widmark Factor and Barbour, based on Forrest’s work and using Forrest’s
computer program, subsequently presented Forrest’s results by way of a chart. Whilst the results of Forrest and Barbour
are often used interchangeably, there is a significant difference between them on the factors for women. This paper
examines the source of the unexpected discrepancy. It is essential to quote an error range, in blood alcohol concen-
tration calculations, for the results. The extent of that error range was investigated by Gullberg who also employed the
Widmark Factor formulation. Gullberg concluded that when reporting a calculated blood alcohol concentration, a
coefficient of variation of 21% should be applied. Similarly, Gullberg concluded that when calculating the volume of
drink, a coefficient of variation of 12½% should be applied. The present paper derives and publishes the formulae for
calculating this coefficient of variation. It is then shown that Gullberg’s conclusions are mistaken: the coefficient of
variation is not some fixed percentage but must be calculated in each case.
Keywords
alcohol, calculations, error, uncertainty, Widmark
Searle 3
One may therefore write: Figure 1 shows the two sets of results. Forrest9
gave a mathematical relationship which may be sim-
100 vzad plified to the following form:
C¼ bt
rM
Widmark Factor for men r ¼ 1.0181–0.01213 BMI
This formula calculates the BAC from a past his- Widmark Factor for women r ¼ 0.9367–0.01240 BMI
tory of alcohol consumption, a form which may be
called the Forward Widmark calculation. The for- For men, the tabulated examples published by Forrest
mula may of course be re-arranged to make v the and the chart by Barbour follow closely this simple
subject and so calculate, from a measured level of relationship. For women, the charts of Barbour
blood alcohol B, the volume of drink consumed. follow the relationship but the table by Forrest does
That may be called the Reverse Widmark calculation. not. Forrest’s examples, for women, appear to be
Widmark gave examples of both directions of erroneous.
calculation. Zuba et al.5 comment that the procedure developed
by Forrest is practical and appears to encompass the
current state of knowledge relating to upgrading
The Widmark Factor Widmark’s equation. That appears to be the case,
The Widmark Factor, denoted by r, is not a simple but one must work from the simple mathematical
constant but depends on anthropometric parameters. expression of Forrest’s results and not from the
The influence of such parameters was explored by table of examples he gave.
Forrest, who found that gender and BMI were the
most important.
Uncertainty of the calculated result
Other parameters such as age and stature3 have
been suggested, and BMI has its limitations in char- With the BAC formula, as with any mathematical
acterising body build.8 However, Forrest’s results are formula, errors in the input parameters will produce
widely used. Forrest published examples of what the an error in the calculated result. That error can be
average factor would be, for men and for women, at estimated, by the method of error propagation, from
different levels of BMI. Those examples, and interpol- the contribution of each input parameter.
ations between them, are often used in calculations Suppose in general terms that a result y is to be
presented in Court. calculated from a formula
Barbour subsequently obtained from Forrest the
computer program which had been used to calculate y ¼ f ðx1 , x2 . . . . . . :xn Þ
the Widmark Factor from the BMI. Barbour then ran
the program to obtain extensive results, which he pub- where the first input variable x1 is subject to an error
lished in the form of two charts, one for men and one of standard deviation S1, the second input variable x2
for women. When those charts are applied to the is subject to an error of standard deviation S2 and so
examples Forrest gave, it is found that the two on. If those input errors are normally distributed then
authors agree entirely on the results for men, but for they will, according to the method of error propaga-
women the results differ: tion, create in y an error which has a standard
This discrepancy for women should not exist. The deviation of:
matter is of practical importance since, for a woman
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
of high BMI, it can lead to a difference of 25% or
Sy ¼ ½@y=@x1 2 S21 þ ½@y=@x2 2 S22 þ þ ½@y=@xn 2 S2n
more in the estimation of BAC.
0.90
0.80
Barbour and Forrest (men)
Widmark Factor
Barbour (women)
0.70
X
X
0.60 Forrest (women) X
0.50
15 20 25 BMI 30
Figure 1. Barbour’s and Forrest’s results for the Widmark Factor. Note: BMI ¼ Weight in kilograms/Square of height in metres.
Searle 5
ev, the uncertainty in the volume drunk, is also likely (ev ¼ 0.05) with an ABV of 4.0 0.12% (ez ¼ 0.03).
to be zero in these circumstances. In a real-life event, It is soon calculated that the mass of alcohol in the
the duration of drinking and the volume drunk may drink is 3550 0.040 0.789 ¼ 112 grams.
both have significant uncertainty and estimates of It is also soon found that the value of Co is
those uncertainties must be made from the
circumstances. 100 112
¼ 188 mg%:
Other coefficients of variation concern the param- 0:73 81:6
eters relating to alcohol dilution and elimination, that
is er the accuracy to which Widmark’s Factor can be The rate of elimination is 14.8 mg%/hour so that
determined, ez the accuracy of manufacturer’s values after 5 hours, when all the alcohol has been absorbed
of ABV and eb the accuracy of the assumed rate of and some has been eliminated, the calculated BAC
elimination. Those coefficients of variation have been will be:
determined by researchers. Gullberg reviews the pub-
lished literature and suggests suitable values, that is Co bt ¼ 188 14:8 5 ¼ 188 74 ¼ 114 mg%
er ¼ 0.092, ez ¼ 0.03 and eb ¼ 0.22. These will be
adopted here, because this paper re-works Hence from Formula 1 the coefficient of variation
Gullberg’s example and the comparison is made of the calculated BAC will be:
easier by adopting the same input. All the variables
are assumed uncorrelated except r and b, where the vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
covariance is –0.135. u 188 742
u 0:052 þ 0:032 þ 02 þ 0:0922 þ
Once the coefficients of variation for the input par- u 114 1882
ameters have been established, the uncertainty ec of ec ¼ u
u
t 74
the calculated BAC can be obtained from Formula 1. 0:222 þ 02 0:27 0:092 0:22
Appendix 2 presents a similar exercise for the cal- 188
culation in the other direction, that is the Reverse
¼ 0:21:
Widmark calculation of the volume of drink con-
sumed from a measurement of the BAC at a later
time. The formula for the volume drunk is: The coefficient of variation of the calculated BAC
is therefore 0.21, which is the same result as Gullberg
ðB þ btÞ obtained (21%). On the basis of that example,
v¼ zad where B is the measured level of BAC:
100 rM Gullberg concluded that:
All the variables are assumed uncorrelated except r ‘‘When reporting an estimated BAC, a 2CV [i.e. twice
and b, where the covariance is – 0.135. The coefficient the coefficient of variation] uncertainty interval
of variation of the calculated value of v, the volume of should be approximately 42%’’.
drink ingested, will be:
That is simply not so. The coefficient of variation is
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi not a constant 21% for all circumstances, but must be
u
u ½B=Bo 2 e2 þ ½bt=Bo 2 e2 þ e2 þ e2 þ e2 þ e2 calculated on a case by case basis. That may be seen
ev ¼ t B b t r a z
from Gullberg’s own example, by noting that after a
0:27½bt=Bo er eb
further 7 hours the BAC will have fallen to a calcu-
ð2Þ lated 10 mg%. It hardly needs saying that the coeffi-
cient of variation of that figure is far greater than the
where Bo ¼ B þ bt 21% of 10 mg%, which would be only 2.1 mg%. A
Once again the coefficients of variation for the constant percentage as suggested by Gullberg will
input parameters must be estimated for each of not do.
them. The value of eB, that is the accuracy of blood Furthermore, Gullberg has chosen an example
alcohol analysis, is about 0.0375 in the UK. where uncertainty in absorption (ea) can be ignored,
Gullberg stated only the general principle of error as can uncertainty in the duration of the drinking ses-
propagation, without deriving any formula by which sion (et).
the uncertainty could be calculated. He did however In that same paper, Gullberg gives an example of
give an example and, without showing any working, the uncertainty of a Reverse Widmark calculation,
stated the result he had calculated for it. Gullberg’s again without giving any formula. It is based on the
example is presented here in metric units, but this time same data, except now it is the measured blood alco-
giving the formula and calculating through to the hol B which is given, as 120 mg%, and the volume
result. of drink is to be calculated. That calculation is
In Gullberg’s example, a man of mass 81.6 kg straightforward and the result is 3662 millilitres. A
having an estimated Widmark Factor of 0.73 back calculation of Bo, the BAC at time zero, gives:
(er ¼ 0.092) drinks 3.55 0.178 litres of beer 120 þ 14.8 5 ¼ 194 mg%.
Searle 7
Appendix 2
where C is the BAC at the relevant time Estimation of the uncertainty of a reverse
v is the volume of drink consumed in millilitres
BAC calculation
a is the proportion of the alcohol absorbed With the reverse calculation, the volume of drink that
z is the strength of the drink as percentage has been consumed is to be calculated from a mea-
ABV 7 100 sured level of blood alcohol, denoted by B, obtained
d is the density of alcohol (¼ 0.789 grams per milli- from a sample taken at the relevant time. The uncer-
litre, constant) tainty of the blood alcohol measurement is eB.
r is the subject’s proportion of body water in litres/ Using the nomenclature of the main paper,
kilogram, divided by the proportion of water in also presented in Appendix 1, the reverse BAC calcu-
blood in litres/litre (Widmark Factor) lation is:
M is the mass of the subject, in kilograms
b is the subject’s elimination rate, in mg% per zad
v ¼ ðB þ btÞ
hour 100 rM
t is the duration from the start of the session to the The standard deviation of the calculated value of v,
relevant time, in hours the volume of drink ingested, will be:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Uncertainty in any of the input parameters will add u
u ½@v=@C2 S2 þ ½@v=@b2 S2 þ ½@v=@t2 S2
to the uncertainty in C, that is the calculated BAC. u c b t
u
Writing S with a suffix to denote the standard devia- u 2
S v ¼ u þ ½@v=@z Sz þ ½@v=@a Sa 2 2 2
tion of each parameter, the uncertainty (standard t
deviation) of the BAC is: þ ½@v=@r2 S2r þ 2½@v=@r ½@v=@b Covðr bÞ
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u All variables are assumed uncorrelated except r
u ½@C=@v2 S2v þ ½@C=@a2 S2a þ ½@C=@z2 S2z
u and b, where the covariance is –0.135 SrSb
u
Sc ¼ u 2 2
u þ ½@C=@r Sr þ ½@C=@b Sb þ ½@C=@t St
2 2 2 2 Performing the partial differentiations we obtain
t
þ 2½@C=@r ½@C=@b Covðr, bÞ @v v @v vt @v vb
¼ ¼ ¼
@C ðB þ btÞ vb ðC þ btÞ @t ðB þ btÞ
All the variables are assumed uncorrelated except r @v v @v v @v v
and b, where the covariance is –0.135 SrSb. ¼ ¼ ¼
@z z @a a @r r
Performing the partial differentiations, we obtain:
Putting those partial derivatives into the formula
@C Co @C Co @C Co @C Co for the uncertainty of v and writing ev ¼ Sv/v and so
¼ ¼ ¼ ¼
@v v @a a @z z @r r on for the other variables:
@C @C vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼t ¼b u
@b @t u B2 e2c B2 t2 eb 2 B2 t2 et 2
u 2 2 2
u ðB þ btÞ2 þ ðB þ btÞ2 þ ðB þ btÞ2 þ ez þ ea þ er
ev ¼ u
u
Putting those partial derivatives into the formula for t 0:135bt er eb
2
the uncertainty of the BAC, and writing Sv ¼ vev and ðB þ btÞ
so on for the other variables:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Denoting B þ bt by Bo we have:
u 2 h iffi
u e þ e2 þ e2 þ e2 þ ðbt=Co Þ2 e2 þ e2 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
Co u v a z r b t u
ec ¼ t u ½B=Bo 2 e2 þ ½bt=Bo 2 e2 þ e2 þ e2 þ e2 þ e2
C
0:27ðbt=Co Þer eb ev ¼ t B b t r a z
0:27½bt=Bo er eb
ð1Þ
ð2Þ
DOI 10.1177/0025802414524385
ðB þ tÞ rM
v¼
100 zad
ðB þ tÞ rM
v¼
100 zad
In that same column the symbol C appears three times, and it should be B. Minus signs have been missed in the
expressions -zv and -av
@v @v
The expression @ appears incorrectly as V