Page 79 of 201
from experienced AAVs and exporters. It is pleasing to see the results of statistical analyses supporting well accepted hypotheses and anecdotal opinions since it adds confidence in the
interpretation of the findings and in using these results to inform preventive strategies.
It is also interesting that the apparent prevalence for musculoskeletal injury conditions is much closer to the true prevalence estimate than was observed for respiratory disease. The
apparent prevalence is a little higher than the true prevalence, reflecting the same issue as discussed in the previous paragraph.
When the diagnostic information for musculoskeletal injury conditions is considered, it suggests that musculoskeletal injury conditions can be effectively detected as a cause of
death without requiring necropsy examination for diagnostic classification – in contrast to
respiratory disease. This is not to say that lame animals that die should not be subjected to necropsy. The point
is that necropsy of lameinjured animals that die is of value not only to confirm a necropsy diagnosis of lamenessinjury as a cause of death, but to investigate other causes of death
such as respiratory disease that can only be diagnosed effectively at necropsy and not by clinical category information alone. This information can be used potentially to fine tune the
approach to necropsy to make it more efficient without reducing its diagnostic effectiveness. This will be revisited in more detail in later sections of this report.
7.10.5 Inappetence detection using gross necropsy
Inappetence at gross necropsy was mapped most closely to ketosis as a final cause of death.
Table 22: Summary of classification of 215 cases of mortality by gold standard test final cause of death and the gross necropsy diagnosis, using inappetenceketosis as the outcome of
interest.
Ketosis vs inappetence Gold standard test
Final cause of death Disease +
Disease - Gross necropsy
diagnosis
Test + 10
5 15
Test - 2
198 200
12 203
215
Page 80 of 201 Table 23: Statistical measures of diagnostic performance for gross necropsy diagnosis as a
test for detecting inappetenceketosis as a cause of death in cattle. Based on data in Table 22. Se= sensitivity; Sp=specificity; PPV = positive predictive value; NPV = negative predictive
value; App Prev = apparent prevalence; True Prev = true prevalence; CI = Confidence Interval.
Parameter Estimate
95 CI Lower
Upper Se
0.83 0.55
0.95
Sp
0.98 0.94
0.99
PPV
0.67 0.42
0.85
NPV
0.99 0.96
1
App Prev
0.07 0.04
0.11
True Prev
0.06 0.03
0.1 The Se and Sp measures are moderate to high.
The PPV is only moderate and reflects in part the fact that the prevalence of the condition is generally low 6-7 as well as the fact that some animals classified with inappetence at
gross necropsy were ultimately determined to have died from some other cause as the final cause of death. It is recognised that inappetence ketosis can be an important cause of
mortality risk under certain conditions such as shipments of pregnant cattle or older cows.
As an example of the impact of prevalence on estimates of PPV, the PPV was re-calculated using a standard formula involving Se, Sp and prevalence
19
, while holding the Se and Sp constant at the values shown in Table 23 and allowing prevalence to vary from 0.1 to 0.4.
��� =
�∗ �∗
+ −� ∗ −
P=prevalence As prevalence rises from 0.1, 0.2, 0.3, 0.4 the PPV rises to 0.82, 0.91, 0.95, and 0.97,
respectively.
7.10.6 Enteric disease detection using gross necropsy
Table 24: Summary of classification of 215 cases of mortality by gold standard test final cause of death and the gross necropsy diagnosis, using enteric disease as the outcome of interest.
Enteric disease Gold standard test
Final cause of death Disease +
Disease - Gross necropsy
diagnosis
Test + 5
5 10
Test - 5
200 205
10 205
215
19
Gardner and Greiner 1999
Page 81 of 201 Table 25: Statistical measures of diagnostic performance for gross necropsy diagnosis as a
test for detecting enteric disease as a cause of death in cattle. Based on data in Table 24. Se= sensitivity; Sp=specificity; PPV = positive predictive value; NPV = negative predictive value;
App Prev = apparent prevalence; True Prev = true prevalence; CI = Confidence Interval.
Parameter Estimate
95 CI Lower
Upper Se
0.5 0.24
0.7
Sp
0.98 0.94
0.99
PPV
0.5 0.24
0.76
NPV
0.98 0.94
0.99
App Prev
0.05 0.03
0.08
True Prev
0.05 0.03
0.08 While the NPV is very high, the PPV for a gross necropsy diagnosis of enteric disease was
relatively poor. This is likely to be the result of a relatively poor Se as well as the low prevalence of the condition. Enteric conditions include diseases such as salmonellosis and
under favourable conditions shedding animals, rough voyage with associated stress and presence of concurrent disease it is possible that multiple cases of enteric disease could
occur.
If Se and Sp are held at the values above and prevalence is increased to 0.1 and 0.2, the PPV does rise to 0.74 and 0.86, respectively.
7.10.7 Other diagnostic classifications