The headline figures from a police intelligence report caught Steve Dawson's attention. Cannabis was causing 2000 hospital admissions a year costing more than $30 million, the report claimed, and was the "cornerstone" of drug harm in this country.
"I'd never met any of the 2000 a year who were clogging up the hospitals – you know, the stoned wandering the streets and A&Es saying 'help me!' – so I thought I'd look into it."
A "practical sort of person" with a degree in sociology from Canterbury University, Dawson would spend the better part of the next five years battling red tape and Government stalling tactics in his efforts to find the truth.
The 111-page report in question was called New Cannabis: The Cornerstone of Illicit Drug Harm in New Zealand, and was produced in 2007 by the National Drug Intelligence Bureau (NDIB), a police-led operation also involving Customs and Health.
Its author, NDIB strategic drug analyst Les Maxwell, painted a grim picture of New Zealand's most popular drug: cannabis was an "increasing threat" because it was getting stronger and was a "gateway" drug for young people.
But it was the number of people who were ending up in hospital, according to the report, that caught the media's eye.
"Perhaps surprisingly," Maxwell wrote, "cannabis related hospital admissions between 2001 and 2005 exceeded admissions for opiates, amphetamines and cocaine combined", with about 2000 people a year ending up in hospital because of the drug.
The report put cannabis admissions into two categories: primary and secondary diagnosis. Primary diagnosis cases had tracked between 210 and 250 a year, while secondary cases ranged between 1799 and 2012 a year.
Most of the primary admissions were because of psychotic disorder, the report said, while "harmful use" accounted for most secondary cases.
Maxwell put the cost of these admissions at $31m in 2005, up from $19m the year before, based on a whopping 58,000 hospital bed nights on average.
Senior police seized on Maxwell's findings. Detective Inspector Stuart Mills, the NDIB's co-ordinator, said the report provided the first big picture of cannabis' harmful effects.
"We talk about methamphetamine, but here we can see the harm it causes with the number of hospital admissions caused solely by cannabis," he told reporters.
Tony Ryall, Health Minister at the time, suspected the cost to the health system would be "significantly higher ... when you consider its contribution to accidents and family breakdown".
To law reform campaigners, it signalled a "war on cannabis", culminating in a massive operation targeting the Switched on Gardener chain in 2010.
Dawson, formerly of Christchurch but now living in New Lynn, West Auckland, didn't buy the claims. A thin, pony-tailed man who looks younger than his 53 years, he is in some ways a caricature of a cannabis user: t-shirt, shorts and bare feet his dress code.
He is not a member of any pressure group or political party, but feels the cannabis debate has become emotional and politicised. Since he'd fallen on hard times after some failed business ventures, he had the time to pick apart the methods used in preparing the Maxwell report.
If someone was going to make "sole causation" medical claims, they needed to be able to produce the evidence, he figured. So began his magnificent obsession.
In 2010, using the Official Information Act, Dawson set about trying to get his hands on the hospital data used in the report. The Ministry of Health said it was unavailable and referred him to the NDIB.
The NDIB provided a spreadsheet, but it was clearly not the raw data Dawson had requested, so he asked for the original file. It was lost, the bureau said, perhaps incorrectly saved.
Dawson went back to Health, but it too claimed the data was lost. "I had to get pretty cross with Health to finally get the file in July, 2013."
Having studied the literature and spoken to experts, it took him a few minutes to see what the bureau had done.
Hospitalisations are recorded by a worldwide system known as ICD-10, which contains thousands of codes for diseases, symptoms, complaints, social circumstances and suchlike.
After a patient is discharged, a clinical coder goes through the doctor's notes and attributes a principal reason for the admission, as well as codes for other matters on the file.
What the NDIB had done for the Maxwell report, Dawson found, was look for all cannabis-related codes, removed all other codes, leaving just a "primary" and "secondary" reason for the admission.
Dawson flicks through the data and finds an example; a case where a woman has been admitted to hospital because of birthing complications. The original data shows several different codes associated with obstetrics and a live birth, as well as F122, which is "cannabis dependence".
But the Maxwell data ignores the other factors, says the "secondary" reason for the woman's admission was cannabis and attributes the cost of the hospital stay to the drug. "It's ludicrous," Dawson says. "It was a birthing matter, it's got nothing to do with cannabis".
Elsewhere, he finds a case where someone has spent 240 days in hospital due to prostate cancer – again counted as a "cannabis-related" admission in the Maxwell data. "There's no study on the planet that makes that link."
Michael Baker, a professor in public health at the University of Otago in Wellington, confirms Dawson's thesis.
The ICD-10 system records the principal reason for the hospital admission, Baker explains, and all other codes are "additional". They can record trivial things such as the person having a cold or to record social circumstances.
The additional codes are not ranked, he says, so it's incorrect to refer to "primary" and "secondary" diagnoses.
"Those [additional] things have not put the person in hospital, they are just things that were noted on their chart.
"It gives you totally the wrong answer if you go on a fishing expedition for the codes you're interested in."
Analysts should focus on the principal code when drawing conclusions about hospital admissions, Baker says.
Using this logic, just over 200 people a year are admitted to hospital because of cannabis. Police had exaggerated the numbers ten-fold.
Dawson thought he had proved his point and demanded that the Maxwell report be retracted and the Justice Ministry advised. But it seemed police hoped it would quietly go away.
In a letter dated October, 2013, national manager of intelligence Detective Superintendent Stephen Vaughan said the report had been removed from the police website and the internal police intranet.
He said the report was no longer referenced, "due to the fact that it is over six years old", there had been "significant changes" to intelligence practices and processes within the NDIB, and none of the current staff were in their roles when the report was released.
"The issues that you raise ... would not occur in the current intelligence system," Vaughan wrote, saying there was no need for further action.
Dawson was not satisfied and continued to fire off letters. Finally, in August 2015, he received what he considered to be the "smoking gun" document, from associate Health Minister Peter Dunne.
It included an email from Simon Ross, the Ministry of Health's manager of analysis and reporting, who had met with police NDIB members to discuss Dawson's concerns about their use of health data.
Ross had urged caution in using the ICD-10 data and said the NDIB should focus only on cases where a drug-related diagnosis was the primary reason for a hospital admission.
"Presentation of numbers of primary and secondary diagnoses in the same graph are problematic because they imply to the reader that these have the same significance. Since this is demonstrably not true ... this practice should be avoided," he wrote.
Ross said the cost estimates for the cannabis-related hospital admissions were also incorrect as they were based on the "secondary" diagnosis. The estimated yearly cost of $25m to $30m was incorrect, he wrote: $2.5m was "much more realistic".
It was important that in future his unit provide "robust peer review" of the way the NDIB presented health data. "Where we have not offered or provided this in the past, we should have."
Maxwell, who is still with the police, says he doesn't want to discuss his report.
NDIB co-ordinator John O'Keeffe says the bureau accepts the section of the report relating to the assessment of cannabis harm "was not as robust or as clear as it could have been, and could therefore be open to misinterpretation".
There was never any intention to mislead or misinform, and the NDIB stopped referring to the data some time ago, he says.
O'Keeffe says the Maxwell report was a high-level strategic document to assist policy and decision-makers. "It was not, nor was it ever intended to be, an operational document to inform drug enforcement operations."
But Dawson remains suspicious. He believes the Maxwell report did inform police operations, the Switched on Gardener raids the prime example, and that people went to jail because of it.
It's not good enough for police to say the report is "outdated", he says, a public retraction is warranted.
A drug harm sheet from a 2013 intelligence report, which he obtained under the OIA, shows police continuing to combine "primary" and "secondary" data on the same graph, albeit with a caveat in the small print.
Dawson believes it's in police interests to overstate the harm that cannabis causes, to maintain their funding and operational powers.
He accepts cannabis causes harm, as does alcohol and tobacco, but thinks the authorities want to scare us, just like the Reefer Madness propaganda films of the 1930s.
"Now we have Reefer Madness 2.0, which is the New Zealand police fiddling health data."
THE MAXWELL REPORT
Claim: There are more than 2000 cannabis-related hospital admissions each year
Fact: Between 200 and 250 people a year are admitted to hospital primarily because of cannabis
Claim: The cost of these admissions is $31m.
Fact: The cost is more like $2.5m.
- Sunday Star Times