The Gestation and Birth of Somatic Cell Score
SCS "Born" at 1983 NMC Annual Meeting
The question "What
is the proper reference for Somatic Cell Score (SCS)?" is often raised,
both by students and veteran mastitis researchers. Regrettably, the
references are in rather obscure places and few people know of them.
For members of NMC, the gestation of SCS is detailed in: Shook, G. E.,
1982, Approaches to summarizing somatic cell count which improve interpretability.
Proc. Nat'l. Mastitis Council, p. 150-166. In the early 1980s, seven
record processing centers were operating in the Dairy Herd Improvement
system. Three centers reported somatic cell counts and four used one
of three different scoring approaches. To facilitate educational programs
across the country, it was important to establish a common scoring method
across all processing centers. The 1982 NMC article outlined six alternative
scores, discussed their properties, and compared their advantages. The
article concluded with a recommendation to use a natural log score or
base 2 log score. The natural log and base 2 log were preferred because
the scores occur over a 10-point range; in contrast, base 10 logs for
SCC have a range of only 3 points.
If the 1982 NMC paper represented the gestation of SCS, the birth of
SCS was announced in a poster, "The New DHIA Linear Score for Somatic
Cell Count" during the Technology Transfer session at the 1983 NMC annual
meeting. That poster was presented by Anne Saeman, then a research specialist
at the University of Wisconsin and since 1987 executive director of
NMC.
Also in 1982, the NMC paper was presented to the Research Committee
of the National Cooperative Dairy Herd Improvement Program (NCDHIP)
with a recommendation to adopt the base 2 log score. During committee
deliberations, the formula (shown further in the article) was changed
in a small but important way - as proposed, the addition of a constant
4 was included; however the committee voted to change that constant
to 3. The effect is shown in this table for selected values of SCS:
|
Somatic
Cell Score
|
SCC
as proposed (constant = 4) |
SCC
as adopted (constant = 3) |
|
0
|
6,250
|
12,500
|
|
1
|
12,500
|
25,000
|
|
3
|
50,000
|
100,000
|
|
4
|
100,000
|
200,000
|
|
9
|
3,200,000
|
6,400,000
|
|
10
|
6,400,000
|
12,800,000
|
In the early 1980s, SCCs above 6 million were not uncommon so, as proposed,
some SCSs would be >10. One of the processing centers argued that to
save column space on printed reports and memory space in the computers,
the highest scores should be <10. In retrospect, it was a mistake:
a) computer memory has become incredibly cheap; b) for
diagnostic purposes, whether SCC is 6 million or 12 million is of no
consequence; c) SCCs have been managed to lower levels, so the
upper range seldom goes into those extreme highs; d) we now have
SCCs as low as 10,000 - with 3 as the constant, SCC = 12,500 translates
to SCS = 0, so SCC <12,500 calculates to a negative score. In retrospect,
it would have been better to truncate the upper end of the range to
get better resolution at the low end of the range.
The recommendation of the NCDHIP Research Committee was adopted by the
NCDHIP Policy Board, creating the national standard that is still in
use. As originally described, the score was called Linear Score for
SCC owing to the linear (straight-line) relationship between the score
and milk loss associated with increasing SCC. Later, a shorter name,
Somatic Cell Score, was adopted by many users. Both terms are now in
use, and they both refer to the same scoring formula.
To answer the original question - "What is the proper reference for
somatic cell score (SCS)" - the preferred citation is: Shook, G. E.,
1993, Genetic improvement of mastitis through selection on somatic cell
count, Veterinary Clinics of North America - Food Animal Practice 9:563-581.
Details are in the Appendix of the article. This issue of volume 9 (pages
421-630) was titled Update on Bovine Mastitis and was edited by Kevin
Anderson at North Carolina State University. Briefly, the formula is
SCS = log2 (SCC/100) + 3 where somatic cell count (SCC) is in units
of 1,000 cells/mL. This formula is used in several other countries including
Canada, New Zealand, France, and Germany.
Reasons Why Somatic Cell Scores Are More Informative than Somatic
Cell Counts: A Review
Why it's important at the dairy:
-
There is a straight-line relationship of SCS with milk yield. A one-point increase in SCS is associated with a 400 lb (200 lb in first lactation) decrease in yield at all levels of SCS. For SCC, the decrease in yield per 100,000 cells is greater at low counts than high counts. (Raubertas and Shook, 1982, J. Dairy Sci 65:419-425)
-
Ability to detect presence vs. absence of intramammary infections is greater by a factor of two for SCS than SCC. (Berning and Shook, 1992, J. Dairy Sci. 75:1840-1848)
-
Average SCS closely approximates the value that separates the lowest 50% from the highest 50% of cows and this relationship is fairly consistent at all levels of herd average SCS. For SCC, the average corresponds to percentiles considerably above 50%, and the percentile associated with average SCC varies widely among herds. A few extremely high SCCs inflate average SCC, and they distort the average as a measure of the middle value of a herd. (Ali and Shook, 1980, J. Dairy Sci. 63:487-490)
-
Heritability of SCS is 12%, but only 6% for SCC. More of the differences among cows are attributable to genetics and less to environmental factors when SCS is used. As a result, reliabilities of bulls' genetic evaluations are higher for SCS. (Monardes et al., 1983, J. Dairy Sci 66:1707-1713)
Why it matters in the research lab:
-
The variance of SCS among cows within herds or among daughters within sires is reasonably consistent from herd to herd or sire to sire. Because variances are similar among groups, a common value characterizes variation in all groups. Only the group average is needed to characterize differences between groups when using SCS. For SCC, variances are quite variable from group to group. (Ali and Shook, 1980)
-
The SCS conforms to the familiar symmetrical, bell-shaped Normal frequency distribution; the frequency distribution of SCC is strongly skewed and non-symmetrical. The Normal distribution is required for many of the common methods for statistical hypothesis testing. (Ali and Shook, 1980)