The headline finding doesn’t mean drugs are harmless or that there are no risks associated with taking them. But the general perception of how harmful they are doesn’t line up with the experience of many users.
“MDMA is a classic example. It has a very high profile in terms of the media and popular consciousness, but based on our initial findings, it’s one of the lower-risk drugs,” Wilkins told the Herald.
He warned that those risks can be easily amplified if it’s a strong dose, adulterated with other drugs (which can be checked at a drug-checking service), or combined with intoxication or over-exertion, such as dancing for long periods.
Of particular interest are the various components of harm in the data that could have practical applications for reducing the negative impacts of particular substances across drug policy, treatment for problem users and harm-prevention strategies.
“It’s not just whether this drug is harmful. We want to know how it’s harmful,” he said.
“The people using meth with really serious problems: who are they and why are they [having serious problems]? We want to look at breaking it down by not only drug type, but also frequency of use, amount used and also the people having the problems.
“Is it mental illness? Neurodivergence? It’s then about trying to help them with those issues. This could be a really important insight in terms of allocating health services.”
The data is summarised in the Substance Outcome Harm Index, which has heroin/morphine (9/10) at the top, followed by methamphetamine (8.21), and GHB/GBL (8), also known as fantasy.
These were significantly higher in cumulative harm than the rest and roughly twice as harmful as alcohol (4.2), which was ahead of cannabis, tobacco, nitrous oxide (nangs) and vaping. The least harmful was LSD, followed by cocaine and then MDMA (ecstasy).
An important caveat is the methodology is still being developed, with the data and findings – shared exclusively with the Herald – in only a preliminary stage.
It’s also a measure of acute harm rather than something that might emerge years or even decades later.
“So with tobacco, ‘I smoke now and I get cancer 20 years from now’. This index is not going to pick that up very well,” Wilkins said.
Almost one in five users (19%) reported a mental health problem related to their substance use. This was the most commonly cited issue, ahead of money problems (18.2%), physical health impacts (14.6%) and relationship difficulties (14.3%).
Unsafe driving (6.7%) and personal safety concerns (6.7%) were comparatively rarely experienced, as was having legal issues (3.1%) linked to substance use.
Of those who reported a negative experience, most said it was due to a single substance, while a quarter said it was two substances, and one in 10 said it was three. To avoid confusion over which drugs might be responsible for which type of harm, the data from those who ticked more than one substance was excluded from the index.
The results show – and advocates for drug law reform have repeatedly said – the Misuse of Drugs Act does not align well with the level of harm associated with particular substances.
LSD and cocaine, for example, are Class A substances, meaning their misuse can lead to the highest penalties, even though they are at the lower end of the harm scale.
Alcohol and tobacco are legal, while MDMA and opium – each with significantly different harm profiles – are both Class B substances.
Wilkins’ latest research follows two major reports into drug law reform: one by the Drug Foundation, which recommended decriminalising personal drug use while boosting resources for harm reduction, based largely on what has happened in Portugal; and the other by the Helen Clark Foundation, which, contrary to popular political opinion, found majority support for shifting from a punitive law to a health-focused one.
The Government has already ruled this out, saying it would be counter to its stated goal of toughening law and order.
Perception v reality
A key attribute of the Substance Outcome Harm Index is that it’s about individual risk rather than total societal harm.
The latter would factor in prevalence of use, which was a major contributor to alcohol topping a New Zealand-specific drug harm index in a 2023 study; “harm to others” formed a bigger share of overall alcohol harm in the study than “individual harm”.
The stark difference in prevalence across substances is evidenced in the Health Ministry’s annual health survey: 3.25 million adults (15 and older) had consumed alcohol in the previous year, almost five times as many as those who had smoked cannabis, more than 15 times as many as those who had used MDMA, and almost 100 times as many as those who had used opioids.
For his survey, Wilkins said the respondents were broadly representative of the demographics of the population, but with one key difference: they all said they had used drugs in the previous six months.
The survey asked about 12 dimensions of harm: mental health, physical health, relationships, money, driving, personal safety, violence, legal, property crime to pay for drug use, drug dealing to pay for drug use, parenting and work/study.
Respondents were then asked whether they experienced any of these “related to your alcohol and drug use in the past six months”, and if so, how severe: from one (mild) to four (very serious).
The results are a snapshot of the period from December 2024 to May 2025.
Wilkins said his team was applying for more funding so they could dive more deeply into the data.
He cannot yet say, for example, what proportion of methamphetamine users reported no harm from their use in the previous six months.
Preliminary findings may surprise those who associate the drug with “violence”. This scored very low on its harm profile compared with “money” or “relationship” problems.
“This is not saying that methamphetamine use does not contribute to violence, but that most users, on average, do not often experience these issues at a high level of seriousness,” Wilkins said.
He said the stigma of violence might make it less likely to be self-reported.
Driving problems also rated low, and Wilkins noted reports the drug is sometimes used to help tired drivers stay alert.
In contrast, the data aligns with the perception of meth addicts being “forced to sell drugs to pay for drug use”, which had a high harm rating.
“It’s very early days, but there are some of the really important insights that are surprising,” Wilkins said.
“You have this idea with heroin, for example, that it’s just about dependency, but there are lots of others: relationship problems, mental health, money problems. It’s a broad cross-section.
“With nitrous oxide [nangs], the popular impression is that it’s kind of harmless. But if you dig a little deeper, that’s actually not correct. The use of it in a very repetitive way can be responsible for some serious brain injuries.”
It’s also a substance that has lately become available in such high concentrations that the Government announced a crackdown.
The addictiveness of some substances – opioids, methamphetamine, tobacco – is probably a contributing factor to their high scores when it comes to “money” problems. For the latter, increasingly higher prices – due to excise tax on tobacco – are also often cited as a leading reason for wanting to quit.
A higher likelihood of drug dependency could also be a factor in GHB/GBL (fantasy), methamphetamine and opioids scoring highest for “relationship” problems, while MDMA – sometimes used by couples to enhance their relationships – scored the lowest.
Wilkins’ team is now developing a method for measuring harm components and their respective severity levels. Some people may consider harms related to parenting more serious than those related to property crime, for example, or poor work performance.
No perfect data
This work seeks to iron out inherent issues with self-reported data: how can people’s subjective judgments of harm severity be accounted for? Another issue is how reliable the data is, as opposed to a clinician making more objective observations about harm.
Some of this can be mitigated with the large sample size (4838 respondents), but questions remain on the cocaine data due to its relatively low availability, and hence low frequency of use (and low problematic use). This is despite cocaine consumption skyrocketing in 2024, along with methamphetamine, which has triggered a multi-pronged government response.
Wilkins believed people assumed, incorrectly, that those who used drugs were unreliable.
“We generally find that most of the people who take the effort to do the survey, they do it in a pretty honest way. There’s no real reason for them to lie about harm.”
NZ Drug Foundation executive director Sarah Helm acknowledged the research’s limitations – the lack of insight into chronic harm, for example – but said it was nevertheless important.
“In part, it helps to address the silencing impact of both the criminalisation and stigma surrounding drug use by surfacing people’s real experiences,” she said.
“Some substances have therapeutic benefits increasingly recognised by science. Some have pleasant effects with low harms, such as psilocybin or MDMA, while others carry higher harms.
“Even those with higher harms, it is important to better understand what the real experiences are of people who take it, and why they take it, and why some run into problems.”
Wilkins added that existing data about drug harm failed to capture the total harm spectrum.
Emergency department admissions or accidental fatal overdose data, for example, represent only those at the end of the total harm spectrum. Questions around causality are also rarely definitively answered, “whether [the harm] was the alcohol or the meth or the synthetic cannabinoids”.
“Most people don’t die of an overdose. They can still have quite a bit of harm in their lives from drug use, like a car crash or a relationship break-up,” he said.
“So that’s what we’re trying to do. We can take driving criteria and then rank all the drug types by users reporting how it affects their driving.”
The Government in November outlined the rollout of a new drug-driving regime that will test for THC (cannabis), MDMA, methamphetamine and cocaine.
Critics have pointed to how a positive result doesn’t necessarily translate to an impaired driver, while those using medicinal cannabis might also be captured.
Transport Minister Chris Bishop has defended the regime by saying about 30% of all road deaths involve a driver who has tested positive for a drug.
Wilkins can see the rationale. “Drug-driving is a good thing to respond to because some people are drugged up and that makes them very dangerous on the road.
“But there’s not a whole lot of evidence behind it. The whole policy process tends to be, ‘Well, we’re against drugs, so if people are using drugs when they’re driving, let’s design a law to detect that’. It’s not particularly sophisticated in terms of finding out the risk.”
The index, regardless of whether policymakers will notice, showed the risk of driving harm related to using MDMA was near zero (0.04).
“We could show how taking MDMA is actually related to crashes. That’s an illustration of where this could be going.
“There’s quite a way for us to go in terms of this journey. But I think it is potentially groundbreaking work.”
Derek Cheng is a senior journalist who started at the Herald in 2004. He has worked several stints in the press gallery team and is a former deputy political editor.