CASE NOTES


The North Coast LHPA and Emerging Disease Detection—Risk analysis by a worked example

Matt Ball - Senior District Veterinarian North Coast LHPA

Posted Flock & Herd April 2015

Table of Contents

KEY FINDINGS

BACKGROUND
Risk analysis and emerging disease surveillance
The North Coast of NSW and emerging disease risk
The North Coast general surveillance system
This risk analysis

METHODS
Steps in the risk analysis
Expert Opinion and the Pert Distribution

RESULTS

DISCUSSION

BIBLIOGRAPHY

KEY FINDINGS

BACKGROUND

Risk analysis and emerging disease surveillance

The detection of new and emerging disease is typically listed as one of the main functions of a National or State livestock health surveillance system (van Wuijckhuise et al, 2011; Arthur, 2011). But how can one assess whether a surveillance system can adequately fulfil its role in detecting new and emerging disease?

One technique could be to retrospectively analyse the successes or failures of the system. As an example, the Australian Veterinary Journal (AVJ) recently published a detailed supplement detailing the Equine Influenza event in Australia (AVJ, 2011). Many lessons can be learnt with the benefit of hindsight. Risk analysis has been identified as a prospective tool that can also be used to identify the effectiveness of components in a surveillance system (Ball et al, 2011).

The North Coast of NSW and emerging disease risk

The North coast of NSW is a popular subtropical region that could be particularly prone to new and emerging livestock disease. High densities of livestock, wildlife and humans ensure that pathogens have ample opportunities for species transfer.

Flying foxes, a genus commonly associated with emerging disease, are numerous in the region and they have close contact with livestock because many farms have both horticultural and livestock pursuits. Abundant tick and fly vectors increase the range of pathogens that could successfully establish in the region. Legislative constraints in peri-urban areas, like much of the North coast of NSW, limit vertebrate pest control, ensuring that high numbers of feral pests are available to act as disease reservoirs. When these factors are combined with the increased disease risks from global warming disease emergence seems a legitimate concern.

The North Coast general surveillance system

The North Coast Livestock Health and Pest Authority (NCLHPA) employs three district veterinarians and three para-veterinary officers (Rangers), to carry out a range of livestock health functions as specified in the NSW Rural Lands Protection Act 1998. Amongst other activities, the Act outlines that the Authorities are required to provide resources for conducting animal disease surveillance programs and are to collect, collate, interpret and report animal disease surveillance information.

The NCLHPA provides a disease investigation service. Phone advice is available to the public. NCLHPA ratepayers are entitled to property visits to investigate any herd health problem. If the reported syndrome could be consistent with a notifiable disease, then non-ratepayers are also entitled to a property visit. Routine individual animal problems are referred to a private veterinarian. The NSW Department of Primary Industries (NSW DPI) maintains a State Diagnostic Laboratory (SDL) and covers the cost of laboratory testing to exclude notifiable diseases. Producers are generally responsible for all other laboratory tests. However, together NSW DPI and the NCLHPA contribute a total of $24,000 per annum to pay laboratory fees for selected cases.

Private veterinarians carry out the majority (90%) of livestock disease investigations on the North Coast of NSW. These veterinarians report any suspected notifiable diseases to the NCLHPA. The NCLHPA only undertakes 10% of the disease investigations in the region but carries out 60-70% of the necropsies (Ball et al, 2011).

There are other differences between private veterinarians' activities and the legislated activities of the NCLHPA. The latter has a 100% focus on the diagnosis of herd livestock disease with no resources allocated for treatments. NCLHPA veterinarians have easier access to laboratory test subsidies and the flexibility to spend longer time on a single investigation. The NCLHPA takes active steps to spread surveillance information beyond the individual producer. Each month a surveillance newsletter is distributed to NSW DPI, private veterinarians and producers. In the event of emergency disease the NCLHPA takes a leading role in both investigations and management.

This risk analysis

This quantitative risk analysis explores the capability of the disease investigation service of the NCLHPA to diagnose the progressive spread and thus significance of an emerging disease. NSW has had ongoing depressed State Government budgets for livestock health. It is likely that after the initial detection of some pathogens the Government's ability to fund targeted surveillance would be limited. Political decisions may be made that there is no advantage to fund extensive surveillance. The NCLHPA as an industry funded livestock health service may be able to assist but would also have limited ability to fund a costly surveillance operation. This risk analysis considers this issue by estimating the 'cost neutral' option of the NCLHPA normal disease investigation service to 'passively' diagnose ongoing disease from an emerging disease agent.

The risk analysis explores this concept through a worked example that estimates the probability of the NCLHPA investigation service diagnosing the mortality events associated with the emergence of a new pathogenic blood parasite. No particular blood parasite is singled out as the hazard but the analysis has considered biological features of disease agents such as Trypanosoma evansi (Surra), Theileria sp. (Theileriosis), Babesia sp. (Babesiosis) and Anaplasma sp. (Anaplasmosis).

Blood parasites were considered a suitable example because the region can support adequate vectors for their maintenance and they can have some particular challenges in diagnosis.

The scope of this risk analysis is limited to analysing data relevant to the activities of the NCLHPA.

METHODS

Steps in the risk analysis

Pfeiffer (1997) and OIE (2004) outline the steps of carrying out a quantitative risk analysis. The following steps were utilized in the development of this risk analysis:

Step 1: Define the question to be answered

Step 2: Identify the populations of interest

Step 3: Decide on appropriate initiating and outcome events

Step 4: Draw, discuss and progressively modify a scenario tree that includes relevant variables

Step 5: Identify the pathways that lead to a successful outcome

Step 6: For each pathway summarise relevant information from peers, scientific literature and technical experts that will help assign probabilities in the scenario tree

Step 7: Use deterministic principles to create single parameter values for probabilities that need to be calculated within each pathway

Step 8: Replace the single parameter values with probability distributions using stochastic principles. The POPTOOLS add in to Excel was used for this step. The POPTOOLS excel sheets are attached as an appendix

Step 9: Calculate a total probability for each pathway by multiplying the median probabilities within each pathway

Step 10: Calculate a total probability for the scenario by summing the median probabilities of each pathway

Step 11: Interpretation and communication of the results

Expert Opinion and the Pert Distribution

To calculate the probabilities in this risk analysis it was necessary to consider a range of expert opinions. The Pert distribution is considered one of the most useful for modelling expert opinion (Murray, 2002). It is a modified beta distribution that allows minimum, most likely and maximum estimates to be modelled.

RESULTS

Step 1: Question to answer

What is the probability that the disease investigation service of the NCLHPA will be able to diagnose mortality events associated with the progressive spread of a new pathogenic bovine blood parasite?

Step 2: Population of interest

The region has approximately 370,000 beef cattle and 26,000 dairy cattle principally spread out as small groups on many properties. 723 farms have dairy cattle, with only 87 of those farms having more than 100 cows. There are 6295 farms with beef cattle and these farms have an average herd size of only 60 animals. 15 of the beef herds have more than 1000 animals.

Step 3: Initiating event and outcomes

Initiating event: At least two bovines are found dead by a farmer within a 2 week period.

Successful outcome: Diagnosis

Unsuccessful outcome: No diagnosis

Step 4 and 5: Scenario trees and pathways to success

Four pathways were identified that would lead to a diagnosis:

  1. Event Pathway One- dead animal with blood and/or tissue smears submitted (Figure One);
  2. Event Pathway Two - live animals in a 'suspect' herd and blood smears submitted (Figure Two);
  3. Event Pathway Three- live animals not in a 'suspect' herd and haematology requested (Figure Three);
  4. Event Pathway Four- dead animal histopathology requested (Figure Four).
Figure 1: Initial Scenario Tree. Event Pathway One is shown in green (click to enlarge)
Figure 2: Secondary Scenario Tree showing Pathway Two in green. (click to enlarge)
Figure 3: Secondary Scenario Tree showing Event Pathway Three in green. (click to enlarge)
Figure 4: Secondary scenario tree showing Event Pathway Four in green. (click to enlarge)

Steps 6, 7, 8 and 9: Gathering information relevant to each probability, assigning probabilities and calculating the total probability for a pathway.

Each of the four pathways in the scenario tree is discussed in detail.

Event Pathway One: Dead animal with blood and/or tissue smears submitted (Figure 1).

Initiating event: At least two bovines have died on a farm within a 2 week period.

probability farmer calls LHPA vet (P1) probability LHPA vet attends farm (P2) probability herd on farm is infected (P3) probability dead animal suitable for necropsy (P4) probability dead animal was paristaemic (P5) probability appropriate samples taken (P6) probability pathologist identifies blood parasite (P7).

P1: What is the probability of a North Coast farmer contacting the NCLHPA after two bovines die within 2 weeks?

Palmer (2006) discuses that farmers decide whether or not to contact a veterinarian for a mortality investigation by considering factors such as the value of their stock, whether other animals have died, whether the symptoms were unusual, what the distance is from the nearest veterinary office, whether other farmers in the district were experiencing similar problems and the state of their finances.

The NCLHPA has widely advertised its free diagnostic service for herd disease. Distance is not an issue as the NCLHPA will drive anywhere within its Authority region (Map 1).

Map 1: Region of the North Coast LHPA with a green mark placed on approximate location of all farm investigations between July and December 2010 (NCLHPA, 2011).

There is significant variation in the disease reporting between farmers. North Coast cattle farmers can be divided into three roughly equal groups and a fourth smaller group (NCLHPA, 2011):

Of these farmer groups, underreporting to the NCLHPA is common in Company and Traditional farmers. Company farms typically find solutions to problems within their own range of consultants. Diagnoses made by these consultants are not typically shared with the NCLHPA. A large proportion of traditional farmers dispose of dead stock on farm without consulting veterinary opinion. However, the probability of any farmer type contacting the NCLHPA is increased when there are multiple mortalities.

Each month the NCLHPA livestock health manager summarises all disease investigations according to suspected diagnosis and location of outbreak. In addition, the three North Coast district veterinarians provide a report on any trends in their district. This reporting process is generally effective at detecting a suspicious temporal or spatial disease cluster. The monthly surveillance activities are then reported on a syndromic basis to farmers and private veterinarians through an electronic surveillance newsletter. NCLHPA experience with diseases such as Yersiniosis, Theileria and Hendra virus demonstrates that this surveillance reporting process increases farmer reporting and the consequent sampling for a syndrome.

During the recent media attention on Hendra virus reports of horse deaths were so common that the livestock health manager was able to estimate that a horse dies almost every day in the region. Of course only a minute proportion of these will have had Hendra virus. Prior to the public attention on Hendra virus it would be unusual if more than 10 horse mortalities were reported annually. However horses are companion animals and there is less likelihood of such increased reporting after a bovine disease alert.

In fact, some cattle farmers may be less likely to report a disease syndrome being seen commonly in the district. They may interpret it as a 'normal' event. A staff member of the NCLHPA was informed by a beef farmer in 2009 that:

he had over ten stock die and he couldn't work out what the problem was - we didn't use the vet because it was obvious that the district had the problem' - it was a bad season.

In 2010 a local dairy farmer casually informed a NCLHPA district veterinarian that:

3 months ago I introduced 50 dairy cows from Queensland. Over 2 weeks 15 of them died. I didn't call you as they were introduced cattle and I put it down to them not being any good for here - other farmers have had the same issue.

Even in the face of an obvious epidemic there will still be underreporting. In the Equine Influenza outbreak in NSW the rate of underreporting in some regional areas was as high as 10% (Sergeant, 2011).

The probability of contacting the NCLHPA is further decreased by the growing number of cattle farmers (almost 1000) that are non-ratepayers because their farm is less than 10 hectares in size. These farmers either know they cannot use NCLHPA services or have no knowledge of the NCLHPA or, if they do contact the NCLHPA, a staff member may decide not to service them on the assumption that a notifiable disease is unlikely and they are not entitled to service.

There are also a proportion of farmers that are reluctant to involve the NCLHPA in a disease investigation because of the farmers' focus on avoiding the risk of regulatory action if a notifiable disease is found. Probability of contacting the NCLHPA is also decreased by the increasing percentage of absentee farmers who may not even notice the mortalities.

If the second mortality occurred on a Saturday experienced farmers will not contact the NCLHPA because they are aware that NCLHPA staff does not work on weekends except for emergency disease work. Farmers less familiar with the NCLHPA may still contact the district veterinarians on the weekend including for a single mortality.

A variety of individuals were asked, after the issues were discussed, for their opinion on this probability of a farmer contacting the NCLHPA after a second mortality. For example, the NSW DPI Regional Veterinary Officer, who was also a North Coast district veterinarian for 2O years, estimated that 1 in 3 North Coast farmers would report at least two mortalities. A number of LHPA staff felt the number would be slightly higher. One North Coast cattle farmer who had also worked in the Cattle tick program for many years thought only 1 in 5. A number of other farmers were uncertain. The minimum, most likely and maximum of the expert opinion were entered into a Pert distribution in Poptools and generated the data in the attached Excel sheet PROB 1. Table 1 summarises relevant statistics for P1.

Table 1: P1-What is the probability of a North Coast cattle farmer contacting the NCLHPA after 2 bovines die within 2 weeks?
Probability Median Minimum Maximum 95% Confidence Interval
P1 0.30 0.11 0.48 (0.16, 0.44)

P2: What is the probability that a NCLHPA veterinarian will be able to attend the farm after a request to investigate bovine mortality?

The district veterinarians of the NCLHPA attempt to service all requests to investigate bovine disease as long as the request is consistent with the NCLHPA surveillance guidelines. A single sick animal that is likely to require treatment is an example of an initiating event that would not be consistent with the surveillance guidelines. This would be referred to a private veterinarian. The initiating event in this scenario, multiple bovine mortalities, is considered a syndrome that the NCLHPA district veterinarians would prioritise and schedule a farm visit.

One NCLHPA district veterinarian is based in Casino, one in Lismore and another in Grafton. Each district veterinarian principally undertakes disease investigations in the area surrounding his office. However, they are all part of one workplace and any of the district veterinarians can be rostered to attend calls in any part of the NCLHPA operational region. This means that staff leave/absence or an inability of an individual veterinarian to attend a farm due to other work commitments should not prevent the capacity of the NCLHPA to attend a disease investigation.

The following reasons may prevent the NCLHPA district veterinarians attending a bovine mortality event:

NCLHPA records and the expert opinion of the three district veterinarians were collected to develop a point estimate for this probability. On average the NCLHPA will be unable to attend 16 of 300 bovine mortalities reported to it. Therefore a deterministic estimate is that there are 284 successful events, creating a probability of 95%. This figure was used as the 'most likely' parameter in a Pert distribution to generate the statistics in Table 2.

Table 2: P2- What is the probability that a NCLHPA veterinarian will be able to attend the farm after a request to investigate bovine mortality?
Probability Median Minimum Maximum 95% Confidence Interval
P2 0.94 0.87 0.98 (0.89, 0.97)

P3: What is the probability that the herd, with the mortalities and being visited by the LHPA, is infected with the new blood parasite?

If a new pathogenic blood parasite was to enter and become established in a proportion of the North Coast cattle population there would be an overall increase in the population mortality rate and consequently the level of reporting to the NCLHPA. The NCLHPA would still receive all of its current mortality investigation reports as well as some additional calls because of the effect of the new pathogen.

The NCLHPA at a 'base' level investigates approximately 270 'more than one' mortality incidents annually. Using P1 (median 0.30) it can be estimated that the true number of 'more than one' mortality incidents is close to 900.

It is conservatively estimated that with the progressive spread of a new pathogenic blood parasite through a naive population at least one extra 'more than one' mortality incident would occur every day. An annual total of 1200 'more than one' mortality investigations in the first year is estimated. In this 'epidemic' state it is also estimated that the rate of reporting and subsequent NCLHPA investigation would increase for those farmers experiencing a 'new' problem. While the P1 median would continue to apply for the reporting of the 900 'normal' incidents a different probability would apply for the reporting to and the number of investigations by the NCLHPA of the 'epidemic' incidents. As the epidemic continued the proportion of incidents investigated by the NCLHPA that were due to the new pathogen would likely increase more and more.

Following discussion on this issue with other district veterinarians a range of expected probabilities was generated. The 'most likely' situation agreed upon was that, given 12 months, on average 70% of the 'new' mortality cases would be reported and investigated. From this assumption it was calculated that during the year in which the new pathogen was progressively spreading through the population the NCLHPA would investigate a total of 480 'more than one' mortality investigations of which 210 (43%) would be from the blood parasite. This figure was used as the 'most likely' parameter in a Pert distribution to generate the statistics in the attached Excel sheet and Table 3.

Table 3: P3- What is the probability that the herd with the mortalities and being visited by the LHPA is infected with the new blood parasite?
Probability Median Minimum Maximum 95% Confidence Interval
P3 0.42 0.26 0.55 (0.30, 0.52)

P4: What is the probability that the dead animal is suitable for a necropsy?

The North Coast of NSW has a subtropical climate. Necropsies performed on the day after deaths are seldom useful. Approximately 10% of the time a suitable carcass is not available for an investigation to continue. This information was used in a Pert distribution to generate the statistics the attached Excel sheet and in Table 4.

Table 4: P4- What is the probability that the dead animal is suitable for a necropsy?
Probability Median Minimum Maximum 95% Confidence Interval
P4 0.89 0.81 0.95 (0.84, 0.94)

P5: What is the probability that the post-mortem animal from an infected herd has a detectable parasitaemia?

The estimation of this probability is challenging because of the variation in the epidemiology of a hypothetical blood parasite. Based on the biology of known pathogenic protozoa; such as T.evansi, Theileria sp, Babesia bovis, Babesia bigemina and Anaplasma marginale; a number of possibilities were considered for a dead animal in an infected herd:

A 'success' as a diagnosis in the scenario of this risk analysis may occur with the first, second or third possibilities. Possibility three is included because it does not matter if the parasite was not the cause of disease because its detection alone is highly significant. Success is most likely with the first possibility.

The rate of infection within a herd will have direct impact on the probability of an animal being parasitaemic. It is not possible to accurately estimate prevalence for the scenario of this risk analysis because it is a hypothetical situation. However it can be assumed that with the progressive spread of a new blood parasite through a na-ve herd the prevalence of infection is likely to be high. Dargantes et al (2010) describe a within herd seroprevalence for T.evansi in an endemic area of 40-50% while the Tick Fever Centre (2010) describe a within herd seroprevalence of 48% in areas of endemic stability for Babeisa bovis. It was not possible to find data for seroprevalence during an epidemic.

It is also assumed that acute illness will be common in a naive population. This would commonly result in a high parasitaemia at the time of death. However there are a number of situations where a new pathogenic blood parasite causing acute disease may still not be detectably parasitaemic at death:

There is a high degree of uncertainty with this probability which is reflected in the summary statistics from a Pert distribution (Table 5).

Table 5: P5- What is the probability that the post-mortem animal from an infected herd has a detectable parasitaemia?
Probability Median Minimum Maximum 95% Confidence Interval
P5 0.54 0.31 0.92 (0.35, 0.79)

P6: What is the probability that appropriate samples will be taken and submitted for testing?

The NSW DPI Lab Manual (2010) indicates that if a blood parasite is suspected it would be ideal to collect the following samples from a dead animal:

In addition to these suggestions the OIE (2010) also lists bone marrow and lung as useful fixed tissues to help exclude T.evansi in a dead animal.

The collection of fixed tissues for histopathology is considered useful to exclude other diagnoses, to suggest possible aetiology of blood parasites if none are identified and occasionally to directly detect Trypanosomes. However histopathology is considered of low value in comparison to the value of blood films, impression smears and fresh brain squash and will not be considered as 'appropriate' in this probability pathway. The probabilities in pathway four specifically address fixed tissues.

Whether or not a NCLHPA district veterinarian collects all, some or none of the above suggestions will depend on the following factors:

Blood, blood smears and organ impression smears are not routinely collected by the NCLHPA in bovine necropsies compared to 'no sampling' and the collection of a range of fixed and fresh tissues. Relevant gross pathology that may alert the district veterinarian to a blood parasite and to include blood smears in their sampling could include some or all of: enlarged spleen, pale lungs, jaundice, red urine in bladder, emaciation, petechial haemorrhages, enlarged lymph nodes and ascites. All the NCLHA district veterinarians are experienced in gross pathology and two of the three district veterinarians have undertaken post graduate certification in gross pathology. Unfortunately some blood parasites, such as T.evansi can have non-specific gross pathology (OIE, 2010).

In comparison to private veterinarians the NCLHPA district veterinarians would normally have enough time to undertake a thorough necropsy and collect a range of samples. On occasions, however, competing work priorities may necessitate a less than thorough necropsy.

The NSW SDL operates on a cost recovery basis so in general fees are charged to the submitting veterinarian. Fees are not charged for notifiable disease exclusion. Unless a subsidy is allocated, the NCLHPA passes this laboratory cost onto the farmer. Since the development of a cost recovery laboratory the rate of 'no sampling' has increased by district veterinarians (Kemsley, 2010). No samples are submitted when either the district veterinarian is satisfied with a diagnosis on the basis of gross pathology or when the farmer elects not to pay for laboratory charges.

However the NCLHPA manages a budget to subsidise investigations that it considers 'significant'. The 2011-2012 NCLHPA Surveillance Guidelines suggest the following situations where a subsidy is relevant:

Initial outbreaks according to the scenario in this risk analysis are highly likely to qualify for a full cost laboratory subsidy. This will increase the rate of sampling and submission by the NCLHPA veterinarian. If the 'epidemic' progressively spreads in the district and a diagnosis has not been made any testing to exclude exotic disease will also be free. In addition, NSW DPI has a specific Theileria project that allows free blood smear examination.

Table 6 [and the attached Excel sheet] summarises statistics calculated from the opinion of the three district veterinarians of the NCLHPA.

Table 6: P6- What is the probability that appropriate samples will be taken and submitted for testing?
Probability Median Minimum Maximum 95% Confidence Interval
P6 0.79 0.63 0.90 (0.67, 0.88)

P7: What is the probability that the pathologist will identify the parasite?

Once the appropriate samples have been collected from a parasitaemic animal the following factors will determine whether a pathologist can diagnose the infection:

Two laboratories, including four pathologists, were consulted for expert opinion. The laboratories included the SDL of NSW and the Tick Fever Centre in Wacol, QLD.

Smears from clinically ill animals with an infection like B.bovis will be very likely to contain parasites at a detectable level. However this is not the case for many other blood parasites, including ones like T.evansi.

There was a difference in opinion between the two laboratories when information was gathered about the rate of inadequate smears. A much higher rate of inadequate smears was reported by the SDL (10%) than by the Tick Fever Centre (2%).

The experience of the pathologist at the tasks becomes an important factor. Smears from the NCLHPA would either be examined at the SDL or at the Tick Fever Centre at Wacol, QLD. In both of these laboratories an experienced pathologist will always look at smears (EMAI, 2011 and Tick Fever Centre, 2011). In the event of an inexperienced pathologist examining the smears the results would be validated by another pathologist. These pathologists are highly experienced at identifying organisms such as Theileria, Babesia and Anaplasma. They do not routinely identify Trypanosomes, but these protozoa should be obvious to identify as long as they are present. Truly 'unknown' parasites would be a particular challenge as most pathologists would recognise parasites by means of 'pattern recognition'.

Sensitivity calculations are much more subjective for a task such as smear examination compared to a serological test. On the basis of comments from laboratory staff the sensitivity for detecting blood parasites in a blood smear with an appearance like Babesia bovis would be close to 99% as long as there are at least 0.02% red blood cells infected (1 in ten fields). If there were only parasites in 0.004% (1 in 50 fields) sensitivity would drop to 50%. The examination of brain squash material also has a sensitivity of close to 99% for this type of parasite. Impression smears from other organs have a much lower sensitivity.

A parasite like T.evansi can be challenging to detect and sensitivities are difficult to predict in the Australian context because it is an exotic disease. In a thick blood smear the sensitivity may approximate 40-50% but in a thin smear the sensitivity would be considerably lower. Other testing techniques, such as specific molecular tests, would improve the chance of detection but they may not be available for a new or emerging parasite and the submitting veterinarian and laboratory staff would be unlikely to suspect the need for them.

The sensitivity to detect a truly new pathogen would be highly variable depending on its appearance in a blood film. If it does not look like a Theileria, Babesia, Anaplasma or Trypanosome it may initially be considered to be artefactual.

There are clearly limitations in a laboratory being able to detect a new or emerging blood parasite on the basis of a single submission. However, it is likely that as an unusual 'epidemic' continues and other diagnoses are excluded the chance of establishing a diagnosis increases rapidly because more submissions are received and a greater variety of testing is initiated. For example, in 2009-2010 an unusual cluster of mortalities associated with bovine anaemia was identified on the south, mid and north coasts of NSW (Izzo et al, 2010). It is now known that these, and ongoing disease outbreaks in a range of locations, were associated with a change in the strain dynamics of the pre-existing Theileria organism. The new syndrome was identified rapidly but it took multiple laboratory submissions and almost a year to establish the molecular basis of the problem (Kamau et al, 2011).

The overall probability for a pathologist to detect a new pathogen is increased when there are multiple chances for success. The discussion for P3 suggested that the NCLHPA may undertake 210 investigations on infected herds. By considering this information in combination with the information related to a single laboratory submission the chances for success is considered to be high. This is reflected in the summary statistics from a Pert distribution (Table 7).

Table 7: P7- What is the probability that a pathologist will identify the parasite?
Probability Median Minimum Maximum 95% Confidence Interval
P7 0.90 0.81 0.99 (0.83, 0.97)

Total probability for Event Pathway One

The median values of the probabilities in this pathway can be multiplied to calculate a total probability.

P1 X P2 X P3 X P4 X P5 X P6 X P7 = 0.04 (4%)

The concept of 'further work' (Figure 1)

Figure One summarises, in yellow, the points of the scenario tree where neither a successful (diagnosis) or unsuccessful (no diagnosis) outcome has occurred and referral to Pathways Two, Three and Four is indicated.

The concept of 'further work' is introduced. This further work involves a range of activities that could increase or decrease the suspicion of a blood parasite being involved in the recent mortalities. For example, if a dead animal is not suitable for necropsy the district veterinarian may elect to collect a more detailed history, examine the environment and undertake a clinical examination of live animals. When a dead animal was suitable for necropsy further work would include that listed above but also involve results of a detailed gross pathological exam and any other sampling the district veterinarian elects to do.

The probability of further work is described as Pfw (Figure 1). If further work does not occur the probability is 1-Pfw. A variety of opinions were collected on what Pfw would be equal to. Most opinions emphasised that LHPA district veterinarians typically try to undertake further work if a diagnosis is not immediately forthcoming. A range of factors including farmer motivation and the availability of further animals can reduce the probability. The attached Excel sheet Pfw and Table 8 summarise statistics calculated for this probability.

Table 8: Probability of further work (Pfw)
Probability Median Minimum Maximum 95% Confidence Interval
Pfw 0.80 0.67 0.90 (0.71, 0.87)

Once the further work is completed the district veterinarian may either suspect a blood parasite or have no suspicion of a blood parasite. The probability of whether or not further work would lead to a suspicion of a blood parasite is described as Psbp (Figure 1). The attached Excel sheet Psbp and Table 9 summarise statistics calculated for this probability. The probability is relatively low because a more detailed history, environmental examination, clinical examinations and gross pathology do not reliably suggest blood parasites especially if indicative live animals are not available on the day.

Table 9: Probability of suspecting a blood parasite (Psbp)
Probability Median Minimum Maximum 95% Confidence Interval
Psbp 0.32 0.21 0.55 (0.22, 0.49)

If a blood parasite is suspected district veterinarians would often progress to collect peripheral blood smears from the tail tips of live animals (Event Pathway Two). If a blood parasite is not suspected one option may be to locate any sick animals and submit blood for routine haematology and biochemistry (Event Pathway Three). If necropsy findings were not obviously suggestive of a blood parasite the veterinarian may submit a range of fixed tissues to seek a diagnosis by histopathology (Event Pathway Four).

Less detail is provided for the calculation of probabilities (P8-P20) in Pathways Two, Three and Four because many of the assumptions have been made in discussions relating to P1-P7.

Probabilities leading to Event Pathway Two (Figure 1)

There are a number of events and options that need to occur before Event Pathway Two will happen. These are summarised as:

(P1) X (P2) X (P4) X (1-P6) X Pfw X Psbp = 0.0135

OR

(P1) X (P2) X (P4) X (P6) X (1-P7) X Pfw X Psbp = 0.005

OR

(P1) X (P2) X (1 -P4) X Pfw X Psbp = 0.007

The total probability leading to Event Pathway Two is considered to be 0.0135 + 0.005 + 0.007 = 0.025 (2.5%)

NB: P3 and P5 were not included in these calculations because these probabilities that relate to herd status and parasitaemia are covered by new probabilities within Pathway 2.

Event Pathway Two: Examination of live animals in a 'suspect' herd and smears initially made (Figure 2).

Initiating event: Veterinarian has decided to examine and sample live animal/s in a herd with a 'suspicion' but not a diagnosis of a blood parasite problem based on necropsy, histopathology or other findings. The veterinarian elects to request smear examination from live animals at the laboratory.

Probability herd on farm is infected (P8) probability selected animal is parasitaemic (P9) probability appropriate samples are collected and appropriate test is requested (P10) probability pathologist identifies parasite (P11).

P8: What is the probability that a 'suspect' herd on a farm is infected?

This event has a similarity to P3 but involves significant sampling bias because the herd is already considered to be 'suspect'. As the 'epidemic' continues, a clinical case definition may be determined for a 'suspect case'. This would further increase the probability of the investigated herd being truly infected. A statistical summary for this probability is provided in the attached Excel sheet for P8 and in Table 10.

Table 10: P8- What is the probability that a 'suspect' herd on the farm is infected?
Probability Median Minimum Maximum 95% Confidence Interval
P8 0.72 0.47 0.85 (0.54, 0.83)

P9: What is the probability that a selected 'suspect' animal is parasitaemic?

This event has a similarity to P5 but a number of issues are considered to increase the chance of the selected animal being parasitaemic in this pathway:

A statistical summary for this probability is provided in the attached Excel sheet for P9 and in Table 11.

Table 11: P9- What is the probability that a selected 'suspect' animal is parasitaemic?
Probability Median Minimum Maximum 95% Confidence Interval
P9 0.74 0.54 0.90 (0.58, 0.86)

P10: What is the probability that appropriate samples will be selected from a live 'suspect' animal and an appropriate test requested?

Appropriate samples and test requesting would preferentially involve both thick and thin field blood smears with a request for parasitological examination. Collection of EDTA blood with a request to the laboratory to make smears and examine them for parasites is also acceptable but not as ideal.

It is highly probable that in this event a district veterinarian will intend to collect at least some appropriate samples and request appropriate tests. The creation of both thick and thin blood smears may not always be done. Some blood parasites, such as T.evansi are very difficult to detect in a thin blood smear (OIE, 2010).

A statistical summary for this probability is provided the attached Excel sheet for P10 and in Table 12.

Table 12: P10- What is the probability that appropriate samples will be selected from a live 'suspect' animal and an appropriate test requested?
Probability Median Minimum Maximum 95% Confidence Interval
P10 0.98 0.92 1.00 (0.94,1.00 )

P11: What is the probability that a pathologist will identify the parasite in appropriate samples from a suspect animal in a suspect herd?

This probability is similar to P7 although the sensitivity of smear examination from a live versus dead animal is considered higher, especially if peripheral tail tip and thick and thin smears are made (Tick Fever Centre, 2011). This is one of the major factors for the difference between P7 and P11.

A statistical summary for this probability is provided the attached Excel sheet for P11 and in Table 13.

Table 13: P11- What is the probability that appropriate samples will be selected from a live 'suspect' animal and an appropriate test requested?
Probability Median Minimum Maximum 95% Confidence Interval
P11 0.98 0.92 1.00 (0.94, 1.00)

Total probability for Event Pathway Two

The median values of the probabilities in this pathway can be multiplied to calculate a total probability:

0.025 X P8 X P9 X P10 X P11 = 0.01 (1%)

On average at least two live animals would be sampled by the district veterinarian so this probability is doubled.

= 0.02 (2%)

Probabilities leading to Event Pathway Three

There are a number of events and options that need to occur before Event Pathway Three will happen. These are summarised as:

(P1) X (P2) X (P4) X (1-P6) X Pfw X (1-Psbp) = 0.015

OR

(P1) X (P2) X (P4) X (P6) X (1-P7) X Pfw X (1-Psbp) = 0.01

OR

(P1) X (P2) X (1-P4) X Pfw X (1-Psbp) = 0.017

Total probability = 0.015 + 0.01 + 0.017 = 0.04 (4%)

Event Pathway Three: Examination of live animals and haematology initially requested

Initiating event: A district veterinarian has decided to examine and sample live animal/s in a herd as a diagnosis has not been made on gross pathology or the dead animal was not suitable for necropsy. Because a blood parasite is not specifically suspected the veterinarian elects to request routine haematology and biochemistry.

Probability herd on farm is infected (P12) probability haematology results will be suggestive of infection and further blood smear examination occurs(P13) probability pathologist identifies parasite (P14).

P12: What is the probability that a herd with live animal testing on a 'less suspect' farm is infected?

An assumption is made that P12 will approximate P3. A separate distribution was not generated in Excel.

Table 14: P12- What is the probability that a herd with live animal testing on a 'less suspect' farm is infected?
Probability Median Minimum Maximum 95% Confidence Interval
P12 0.42 0.26 0.55 (0.30, 0.52)

P13: What is the probability that routine haematology results will be suggestive of infection and that blood smear examination occurs?

Anaemia is a common abnormality with blood parasites. Anaemia is also a common event in many other bovine disease syndromes on the North Coast of NSW. Internal parasites, bracken fern and chronic disease are some examples. Theileria is an endemic blood parasite that causes anaemia.

As a new pathogenic blood parasite progressively spreads through a naive population the rate of anaemia will increase in disease investigations. The number of herds involved during this process has already been estimated for P3 and were used also for estimating P13. The ratio of anaemia as a finding in haematology reports during a 'baseline' period was compared to the likely ratio of anaemia during an 'epidemic'. Early in the 'epidemic' a diagnosis of anaemia may not prompt the requesting or collection of blood smears because other diagnoses would be considered more likely. Later on in the 'epidemic' haematology results suggestive of anaemia may lead to routine blood smear submission. The attached Excel sheet for P13 and Table 15 summarise the statistics for P13.

Table 15: P13- What is the probability that routine haematology results will be suggestive of infection and that blood smear examination occurs?
Probability Median Minimum Maximum 95% Confidence Interval
P13 0.48 0.30 0.76 (0.34, 0.68)

P14: What is the probability that a pathologist will identify the parasite?

P14 can be assumed to equal P11.

Table 16: P14- What is the probability that appropriate samples will be selected from a live 'suspect' animal and an appropriate test requested?
Probability Median Minimum Maximum 95% Confidence Interval
P14 0.98 0.92 1.00 (0.94, 1.00)

Total probability for Event Pathway Three

The median values of the probabilities in this pathway can be multiplied to calculate a total probability:

0.04 X P12 X P13 X P14 = 0.0076 (0.76%)

Probabilities leading to Event Pathway Four

There are a number of events and options that need to occur before Event Pathway Four will happen. These are summarised as:

(P1) X (P2) X (P4) X (1-P6) X Pfw X (1-Psbp) = 0.015

OR

(P1) X (P2) X (P4) X (P6) X (1-P7) X Pfw X (1-Psbp) =0.01

Total probability = 0.015 + 0.01 = 0.025 (2.5%)

Event Pathway Four: Dead animal. Tissues submitted for histopathology.

Initiating event: A necropsy was undertaken and gross pathology did not suspect a blood parasite. A range of fixed tissues were submitted to the laboratory.

Probability herd on farm is infected (P15) probability results are suggestive of infection and blood smear examinations occur (P16) probability pathologist identifies parasite (P17).

P15: What is the probability that the herd on farm is infected?

An assumption is made that P15 will approximate P3 and P12. A separate distribution was not generated in Excel.

Table 17: P15- What is the probability that on farm is infected?
Probability Median Minimum Maximum 95% Confidence Interval
P15 0.42 0.26 0.55 (0.30, 0.52)

P16: What is the probability that histopathology results will be suggestive of infection and subsequent smear examinations occur?

Histopathology is not a sensitive test for blood parasites. Occasionally trypanosomes or other blood parasites can be identified in the tissues of an infected animal. Evidence of hypoxia and anaemia may be seen in some tissues. It is relatively unlikely that a histopathology report would trigger the district veterinarian to re-attend the farm and collect blood smears. Summary statistics for P16 are provided in the attached Excel sheet for P16 and Table 18.

Table 18: P16- What is the probability that histopathology results will be suggestive of infection and subsequent smear examinations occur?
Probability Median Minimum Maximum 95% Confidence Interval
P16 0.14 0.06 0.20 (0.09, 0.19)

P17: What is the probability that a pathologist will identify the parasite?

An assumption is made that P17 will approximate P11 and P14. A separate distribution was not generated in Excel.

Table 19: P17- What is the probability that a pathologist will identify the parasite?
Probability Median Minimum Maximum 95% Confidence Interval
P17 0.98 0.92 1.00 (0.94, 1.00)

Total probability for Event Pathway Four

The median values of the probabilities in this pathway can be multiplied to calculate a total probability:

0.025 X P15 X P16 X P17 = 0.0015 (0.15%)

Step 10: Assigning a total probability for the scenario

The probability of each pathway can be added to calculate a total probability for the scenario:

Scenario Probability = 0.04 + 0.02 + 0.0076 + 0.0015

= 0.07 (7%)

Step 11: Interpretation and communication of the results

It is estimated that as a new pathogenic blood parasite spreads through the North coast cattle population the disease investigation service of the NCLHPA has a 7% chance of establishing a diagnosis in each multiple mortality event.

DISCUSSION

Research undertaken for this risk analysis suggests that the disease investigation service of the NCLHPA is a busy and capable component of the regions livestock disease surveillance system. This NCLHPA component is 'passive' in nature because it simply involves the investigation of clinical cases reported by farmers or private veterinarians (Salman, 2003).

The figure of 7% estimated by this risk analysis may confirm that the passive collection of data will not ensure the early detection of disease or allow a full understanding of the spread of a new disease. Despite this it is suspected that the disease investigation activities of the NCLHPA give enough information to the NCLHPA livestock health manager to monitor disease over time and detect changes in disease patterns. It is hoped that a more active surveillance system can then be implemented once 'epidemic events' are identified.

The reliability and interpretation of the calculated probability needs to be questioned. The risk analysis was undertaken in a limited time period which reduced the opportunity to collect significant expert opinion. The scenario tree created was complex and the opportunities for errors great because so many probability calculations were performed. The author was uncertain if different weightings should have been applied to each pathway before calculating the total probability. Detailed peer review is necessary to improve reliability. Once the reliability of the final calculation is improved expert opinion is also needed to clarify the most appropriate interpretation of the figure. Is this figure actually of any use for evaluating the system?

The scope of this risk analysis has been limited to the activities of the NCLHPA. Private veterinarians are a major contributor to the passive surveillance system and would also detect many more of the mortality events associated with the new blood parasite.

Undertaking a risk analysis can identify areas of low probability where improvements can be made. NCLHPA staff should continue to focus their efforts on the encouragement of disease reporting by farmers. They should also ensure that they collect adequate samples to allow effective screening for a range of blood parasites. The collection of whole blood and the making of thick and thin blood smears should be a more routine task in bovine mortalities. Live animals should also be selected in any case where a blood parasite is suspected.

The model developed by this risk analysis could equally allow other components of the surveillance system, such as private veterinarian activities, to be assessed. Perhaps the figure calculated by this risk analysis is only really useful in assessing the performance of the NCLHPA if a figure could be calculated for other agencies. Making comparisons between the performances of surveillance systems that include an industry funded Authority versus those that do not have an industry funded Authority could also be undertaken using this model and may be relevant for general surveillance discussions occurring at a Commonwealth level.

BIBLIOGRAPHY

  1. Arthur R, 2011, 'Surveillance Program' In: NSW Animal Health Plan 2011-12, NSW DPI, Orange
  2. AVJ 2011, 'Equine Influenza in Australia in 2007', AVJ, Vol. 89, Supplement 1
  3. Ball M, Finlayson K and Freeman P 2011, 'A review of a regional animal disease investigation surveillance system', Epidemiol.et sante anim. Vol59-60, pp305-307
  4. Dargantes AP, Dobson RJ., Robertson I.D., Piedrafita, D. and Reid, S.A. 2010 Natural infections with surra and fasciolosis in southern Philippines: seroprevalence, risks and impact on buffalo health. In: XIIth International Conference for Parasitology (ICOPA), 15 - 20 August , Melbourne, Australia
  5. EMAI 2011, Email and phone correspondence with pathologists at State Diagnostic Laboratory, Elizabeth Macarthur Agricultural Institute
  6. Izzo MM, Poe I, Horadagoda NJdVA, House JK 2010, 'Haemolytic anaemia in cattle in NSW associated with Theileria infections', Aust Vet J, Vol. 88, pp45-51
  7. Kemsley, P 2010, District Veterinarian Casino. Personal communication
  8. Kamau J, de Vos A, Playford M, Salim B, Kinyanjul, P and Sugimoto C, 2011, 'Emergence of new types of Theileria orientalis in Australian cattle and possible cause of theileriosis outbreaks', Parasites and Vectors, Vol. 4, No. 22
  9. Murray N 2002, 'Useful Probability Distributions'. In: Import Risk Analysis- Animals and Animal Products. NZ MAF, pp.83-100
  10. NCLHPA 2011, Data kept on file at the Lismore office of the North Coast Livestock Health and Pest Authority
  11. NSW DPI Lab Manual, 2010. Retrieved Online on 27th October from:
  12. http://www.dpi.nsw.gov.au/agriculture/vetmanual
  13. OIE 2004, 'Guidelines for developing a quantitative risk assessment model', In: Handbook on Import Risk analysis for animals and animal products- Volume 2 Quantitative risk Assessment, OIE
  14. OIE 2010, 'Trypanasoma evansi infection (surra)', In: OIE Terrestrial Manual, Chapter 2.1.17, pp1-14
  15. Palmer 2009, 'Factors affecting livestock disease reporting and biosecurity practices: a study of West Australian sheep and cattle producers'. Retrieved online on 23rd October from:
  16. http://researchrepository.murdoch.edu.au/2972/1/01Front.pdf
  17. Pfeiffer, DU 1997 'Quantitative risk assessment'. In: Risk analysis and animal health- a course manual, Dubendorf, July 13-18, pp. 701-721
  18. Salman MD 2003, 'Animal Disease Surveillance and Survey Systems- Methods and Applications', Blackwell Publishing, Iowa
  19. Sergeant ESG and Dhand NK 2011, 'Assessment of the Proportion of under-reporting during the 2007 equine influenza outbreak in NSW', Australia, AVJ, Vol.89 Supplement 1, pp.73-75
  20. Tick Fever Centre 2010, 'Tick Fever- specialist training course', Wacol, QLD
  21. Tick Fever Centre 2011, Email and phone correspondence with Dr Peter Rolls.
  22. van Wuijckhuise LV, Kick P, van Schaik G 2011, 'Cattle health surveillance in the Netherlands GD- the re-active surveillance component' Epidemiol.et sante anim. Vol59-60, pp126-127

 


Site contents and design Copyright 2006-16©