Inside the chaos before the M. bovis ‘surge’
What the Minister was told about a massive backlog in tracing Mycoplasma bovis – and how much turned out to be wrong.
In the lead-up to last Christmas, more farms were being confirmed with the highly infectious cattle disease Mycoplasma bovis.
Ministry for Primary Industries field staff tasked with tracing infected animals from farm-to-farm, to try and stop the spread – part of an ambitious world-first eradication programme – were anticipating a busy New Year.
But the extra work, on top of its existing backlog, didn’t come. For months, those anticipated referrals for follow-up action weren’t being made. (Which was lucky, in a way, considering many field staff contracts had expired and replacements weren’t sought for months.)
Properties identified for tracing – some dating back to early 2018 – remained un-investigated and farmers were free to sell and move their animals. That potentially presented a huge problem, if infected animals were shifted within the country’s semi-nomadic farming system.
Quick tracing of animals, particularly from known infected properties, is acknowledged as the best way to stop the spread of M. bovis. Not acting on that information is almost unforgivable for a programme whose sole purpose is the disease’s eradication.
The situation would bubble into a crisis that, last month, prompted a public apology from MPI director-general Ray Smith, after two investigations identified multiple failures – and recommendations to fix them.
An urgent briefing paper sent to Biosecurity Minister Damien O’Connor in April detailed the embarrassing backlog and the planned surge in tracing, and recruiting. It’s telling, however, to compare that paper – released to Newsroom as part of a sheaf of ministerial updates about M. bovis – with those investigation reports.
Steady as she goes
Post-Christmas, the briefings to O’Connor, submitted in the name of Biosecurity New Zealand head Roger Smith, raise no serious alarm bells.
Sure, slow progress is noted on follow-up testing from a bulk milk survey in spring, as is a backlog of animals awaiting slaughter. But the impression given is that MPI is on top of things.
January 25’s briefing, for example, says “only a few farms are still being followed up” from the spring milk surveillance, and the survey is “largely complete”. Yet it takes until May 6 for MPI to confirm that all farms have been done.
The programme “remained fully operational” over the Christmas-New Year period, the January briefing says. However there were “very few calls” and “farmers indicated that they did not want contact” over this period. It comes out later that the holiday lull enlarges the backlog.
On March 1, Smith provides an aide-memoire ahead of O’Connor’s oral update to Cabinet. The message is nuanced. It notes poor record-keeping by farmers and “varying rates of compliance” with NAIT, the national animal tracing system. That makes tracing the movements of cattle to and from infected farms a “painstaking task”.
But it proudly states the percentage of farms confirmed as infected is decreasing steadily over time. “This is consistent with the expectation that the number of undetected cases in the national herd is decreasing.” (The paper also notes the over-budget programme will need to access an $88 million contingency.)
The surge arrives
On April 15, in Smith’s 12th regular M. bovis update for the year to O’Connor, cracks appear. It notes the difficulty in contacting the last few farmers about the spring bulk milk survey – “mainly due to their details not being kept up to date in databases (such as Agribase, NAIT, FarmsOnline)”. The briefing notes the frustrations of farmers, including “slow follow-up of stock movements from their farm to other farms”.
Two days later, in a separate out-of-cycle briefing entitled “Surge in farm casing workload”, news of the backlog lands, with a thud, on the Minister’s desk. (MPI puts out a press statement about the surge the following day.)
The briefing’s key messages begin: “There has been a substantial surge in the number of farms which need to ‘cased’ – i.e. called and checked for animal movements which may increase the risk of finding M. bovis on the property.”
It’s the first time this year that O’Connor’s written briefings have mentioned the word “cased”. The increase in the numbers of farms needing to be cased has “developed over the last six weeks”, the briefing says.
Somewhat confusingly, the Minister is told the issue began “with a boost in surveillance resources in December”, which boosted the number of infected farms found and generated “a surge of further work that has now reached the labour-intensive casing stage”. Another listed problem is “the compounding effects of several disease management initiatives”.
The backlog is 1100 farms – 300 tagged “urgent and high priority” and 800 considered “medium and low priority”. Based on previous experience, officials predict 30 more farms will be confirmed as infected. However, the identity of those farms won’t be known by June 1, Moving Day.
O’Connor is assured the surge in work will be matched by a surge in hiring, with all calls and checks expected to be completed by May 17. Importantly, Biosecurity NZ is still confident the eradication of M. bovis will be a success.
Important questions arise. Why did it take months to discover the backlog? And why wasn’t the alarm raised earlier? Helpfully, answers can be found in two investigation reports released last month.
(O’Connor’s Official Information Act response contained no comment. Asked by Stuff about the surge in May, he said eradication was a 10-year programme and activity was expected to peak and trough. “We need to be flexible enough to scale up when needed,” he said. “I continue to have confidence that eradication is achievable and is on track.”)
“Managers were to an extent operating in the dark, with their vision of the response limited largely to their own silos of work.” – Roger Paskin’s report
“The ‘backlog’ did not develop over the preceding six weeks,” a report by Dr John Roche, MPI’s chief science adviser, finds. In fact, field staff, farmers and veterinarians called MPI’s national control centre for months, warning that farms of interest weren’t being followed up.
The numbers were wrong, too. Roche’s team determines the correct size of the backlog was 1481 properties, of which 644 were urgent or high priority. (Later, in a section headed “True size of the backlog”, Roche clarified 882 “risk events” – some properties could have multiple risk events – could be attributed to the backlog, of which 644 were urgent or high priority, while 599 events were considered “routine” because they were within an acceptable timeframe.)
By April, seven supposedly high priority properties had been in the queue for 14 months – when, according to best practice to stop the spread of the disease, they should have been followed up within a month.
That’s a problem because most identified infected properties generate up to 100 to 150 traces each. It’s estimated that a “super spreader” dairy farm could infect another eight farms within six months.
Sure, staff did additional surveillance work. But Roche’s report says planning for that was done in the middle of last year and should have been “anticipated and resourced”.
The other investigation report, commissioned by Dairy New Zealand, was written by disease management expert Dr Roger Paskin. It concludes: “Managers were to an extent operating in the dark, with their vision of the response limited largely to their own silos of work.”
Units within the M. bovis response were “inflexible, solidified and difficult to cut across” with a “turf protection mentality”, Paskin says. There was no single, shared database, for managing properties through the various stages of MPI’s process. (A new database management system is on the way.)
There was a “spreadsheet culture”, Paskin finds – possibly encouraged by software packages with limited functionality and reporting capability. “Due to an overwhelming workload, some of the spreadsheets were often temporarily ‘overlooked’, updating was haphazard, there was no version control, and crucially no sharing of these data between functions.”
No individual manager had a grasp of what all the different parts were doing, the report says.
After the alarm was raised, it took MPI three weeks to piece together data from their different spreadsheets to build a comprehensive picture of the backlog. Even when the true extent emerged, in mid-April, “it seems that this was not necessarily fully understood by senior managers”, Paskin writes.
Key reason for failure
Roche, the MPI chief science adviser, points to a crucial moment, when management of the response shifted from Wallaceville, where MPI’s animal health laboratory is based, to MPI’s Wellington headquarters, Pastoral House. Before then the intelligence manager was a senior epidemiologist who regularly produced reports tracking properties through the disease management queue – from identification, through to casing, testing, confirmation of infection, and depopulation. But after the shift, in May last year, this reporting appears to have been dropped.
“The failure to continue these reports is a key reason for the failure to detect this backlog earlier,” the Roche report says. The teams held meetings but there was no forum to discuss workload or if extra staff was needed.
Paskin says collaboration and communication between teams was “discouraged”. He describes decision-making as slow, badly-informed and centralised. Staff were recruited hastily, and lacked “skills, qualifications, and experience”.
“While the backlog may not affect programme timelines in the long run, there are concerns that a failure to lock down potentially infected properties may have contributed to disease spread.”
It’s not all doom and gloom, however. Because 65 percent of the backlog is beef farms, the situation is “less concerning”, Roche writes, because animal movements off dairy farms are more risky. However, he notes dairy farms comprise the majority of urgent properties in the backlog.
Also, MPI thought up to 250 notices of direction restricting cattle movements might be issued before Moving Day, but the actual number was 171, it said on July 4.
According to the latest figures, there are 22 “active” infected properties. A further 236 farms have movement-restricting notices of direction, and 605 are under what’s called active surveillance. The big question is how will that change when testing is finished on the “backlog” farms.
Why is this important?
M. bovis is highly infectious and most often spreads by infected cattle mixing with uninfected ones. The highly mobile dairy industry involves sharemilkers, the contracted rearing of replacement stock, and wintering of South Island herds in different locations. That can lead to animals from different properties grazing together for more than a year.
Tracing infected animals is crucial, therefore, to stopping the disease’s spread. Infected properties are identified through regular surveillance, and the movement of animals from that farm are traced to known or suspected properties. But to be effective, tracing has to done as soon as possible.
MPI has rightly been criticised for letting the tracing backlog build up. But disease management expert Paskin gives the ministry credit for taking immediate action when the flaws became apparent.
The elephant in the room for the eradication programme, however, remains NAIT. The beleaguered tracing scheme is crucial to New Zealand’s ability to manage any animal disease.
Roche writes in his report: “Current [NAIT] compliance remains inadequate and anything that can be done to encourage farmers to comply with requirements will greatly accelerate the efforts in tracing, property prioritisation, and, ultimately, disease eradication.”
* We have updated this story to add Dr John Roche’s clarification about the “True size of the backlog”.