The Most Convenient Number Came First

The latest National Drug Strategy Household Survey should have been an opportunity to give Australians a clear and honest account of what has happened to smoking, vaping and nicotine use since the federal government overturned the legal vaping market and forced lawful supply into participating pharmacies. Instead, the Australian Institute of Health and Welfare has released a remarkably convenient headline, surrounded it with the language of progress, and postponed much of the information needed to test it.

According to the initial report, daily smoking among Australians aged 14 and over fell from 8.3% in 2022–23 to just 5.6% in 2025. That is not a small continuation of the established downward trend. It is an extraordinary 2.7-percentage-point fall, equivalent to a relative reduction of almost one-third in little more than two years. In population terms, AIHW estimates that the number of daily smokers fell from approximately 1.8 million to 1.3 million. In other words, around half a million daily smokers apparently disappeared during a period in which Australia’s legal tobacco market contracted, the illicit market exploded, vaping was driven almost entirely underground, and the Australian Bureau of Statistics estimated that the total quantity of nicotine consumed was increasing.

If the 5.6% estimate is accurate, it is unquestionably good news. Nobody committed to tobacco harm reduction should be disappointed when fewer people smoke. But an unusually favourable result deserves unusual scrutiny, particularly when it is politically useful to the same government that funds the survey and defends controversial tobacco and vaping policies. Instead of releasing the result alongside the complete technical information required to examine it, AIHW has given the public the answer without showing enough of the working.

The basic calculation is straightforward. AIHW does not simply divide the number of respondents who smoke daily by the total sample. Survey participants are assigned weights so that their responses are intended to represent Australians of similar ages, sexes, locations and demographic characteristics. The estimated smoking prevalence is calculated by adding the survey weights of respondents who reported smoking daily and dividing that total by the combined weights of all respondents. The resulting 5.6% is then multiplied by the estimated Australian population aged 14 and over, yielding an estimated 1.3 million daily smokers.

That is standard survey practice. Weighting itself is not evidence of manipulation. The problem is that the initial report does not provide enough information to determine how much the weighting altered the result.

More than 17,500 people completed the survey. A crude calculation would suggest that approximately 980 respondents correspond to 5.6% of the sample, but that is not necessarily the actual number who reported smoking daily because the published figure is weighted. AIHW has not provided the raw number of daily smokers, the unweighted prevalence, the distribution of the weights or a comparison showing how the estimate changed before and after weighting. It has not published the 2025 response rate, a complete breakdown of completion modes, the design effect or detailed confidence intervals in this initial release.

This means that an independent reader cannot determine whether the raw smoking rate was close to 5.6% or whether weighting moved it substantially. We cannot see whether a relatively small number of respondents received unusually large weights. We cannot assess whether smoking was concentrated among groups with low response rates, or whether the final sample differed from previous waves in ways that ordinary demographic weighting could not fully correct.

These questions matter because this was the first survey in the series conducted by Ipsos after previous waves were conducted by Roy Morgan. Changing the contractor does not automatically make the survey unreliable, and AIHW may have done considerable work to maintain comparability. However, comparability must be demonstrated, not presumed merely because the report places the latest number at the end of an historical graph.

The recruitment process also appears to have changed. In 2022–23, Roy Morgan used a multi-mode process in which selected respondents could complete the survey on paper, online or by telephone. Approximately 72% completed it on paper and 28% online. The 2025 questionnaire describes online completion as the “best method”, while Ipsos explained that postal invitations were used to encourage people to complete the survey online before interviewer visits and paper follow-up.

That could produce a different mixture of respondents. People who promptly answer an online government survey may differ from those who complete a paper questionnaire after personal contact with a fieldworker. They can differ not only in age and sex, which weighting may address, but also in trust, lifestyle, socioeconomic circumstances, willingness to disclose illegal behaviour and the likelihood of smoking or purchasing illicit products. Weighting a respondent to represent more people of the same age and sex does not necessarily make that respondent representative of smokers who refused to participate.

AIHW has previously acknowledged that survey mode can affect responses and that caution is required when comparing results over time. It has also acknowledged that some non-response bias is expected in surveys dealing with tobacco, alcohol and illegal drug use, but that this bias has not been measured.

The previous survey already had a serious participation problem. In 2022–23, 21,663 usable questionnaires were received from 49,389 contacted eligible households, producing a reported response rate of 43.9%. When eligible households that could not be contacted were included in the denominator, the effective response rate fell to 30.6%. More than two-thirds of the eligible sample, therefore, did not produce a usable response under the broader calculation.

For 2025, AIHW said it aimed to obtain responses from at least 20,000 people. The final sample was only a little over 17,500. Missing the target does not prove the results are wrong, but it makes the missing response-rate and weighting information more important, not less. If the intended sample was not reached, the public should be shown where the shortfall occurred, which groups were underrepresented and how heavily the final data had to be adjusted.

The scope of the survey introduces another limitation. It covers people living in private dwellings but excludes people experiencing homelessness, prisoners, residents of rehabilitation facilities, nursing homes and other institutional settings, military establishments and university residences. Some of these excluded or difficult-to-reach populations may have higher rates of smoking, dependence and illicit-product use than the general population. AIHW acknowledges this limitation elsewhere, yet it is barely visible in the celebratory public presentation of the 5.6% headline.

Self-reporting introduces another problem. Respondents were being asked not only whether they smoked or vaped, but whether they obtained products through channels that may be illegal. Although the survey is confidential, people can still minimise, forget or conceal behaviour they perceive as embarrassing, stigmatised or unlawful. Previous AIHW data-quality statements have explicitly acknowledged that tobacco and illicit-drug consumption tends to be underestimated. In a market that has become more illegal since the previous survey, the potential for underreporting has not become less important.

None of this proves that the 5.6% estimate is false. It shows that the initial report does not provide the evidence needed to judge how robust it is. That distinction matters. The responsible criticism is not that AIHW has been caught fabricating a number, but that it has released an exceptional number without releasing the exceptional level of transparency required to support it.

The report becomes even more questionable when it moves from individual product estimates to its constructed measure of “past-year nicotine use”. AIHW says the proportion using any nicotine product in the previous year fell from 17.4% in 2022–23 to 15.2% in 2025. On its face, this appears to support a broad story that nicotine use itself is declining.

But AIHW does not measure every product over the same period. Smoking is included when a person currently smokes daily, weekly or less than weekly. Vaping is included when a person currently vapes daily, weekly, monthly or less than monthly. Nicotine pouches, traditional snus and other nicotine products are included if they were used at any time during the previous 12 months.

That is not a consistent past-year measure. Someone who smoked throughout the first half of 2025 but quit six months before completing the survey may not be counted as a past-year nicotine user under the published definition. Someone who tried one nicotine pouch during that same period is counted. Current smoking and vaping are being combined with any past-year use of other products and presented as though the reference periods were equivalent.

This construction can make the overall nicotine figure appear lower than a genuine measure of everyone who used nicotine at any point during the year. It also treats vastly different behaviours as interchangeable. Trying one pouch once and smoking cigarettes every day are both reduced to a yes-or-no indicator of nicotine use, despite representing completely different levels of exposure and risk.

The report then expresses concern that the proportion using three or more forms of nicotine “almost doubled”, rising from 0.5% to 0.9%. Mathematically, that is an 80% relative increase, but in absolute terms, it is only 0.4 percentage points. More importantly, the questionnaire changed.

For the first time in 2025, oral nicotine pouches and traditional tobacco-containing snus were presented as separate response categories. AIHW concedes that respondents in earlier surveys may have recorded nicotine pouches under snus or other nicotine products. By creating more individual categories in 2025, the survey made it easier for a respondent to be classified as having used multiple forms of nicotine.

A person who had previously ticked a single combined or approximate category could now tick both traditional snus and nicotine pouches. That alone can move people into a higher “poly-use” category without any genuine increase in the number of products they use. Presenting the result as an emerging behavioural concern without quantifying the effect of the questionnaire change risks turning a measurement artefact into a public-health warning.

The report’s concern about nicotine use is also strangely detached from relative risk. Daily smoking represents repeated exposure to the products of combustion and is responsible for an enormous burden of disease. Using a nicotine pouch or vaping does not carry the same risk as inhaling cigarette smoke. Yet the report repeatedly collapses these behaviours into a single category of “nicotine use”, as though the public-health objective should be reducing every form of nicotine rather than moving people away from its deadliest delivery system.

The young-adult results expose this problem particularly clearly. Among people aged 18–24, current vaping reportedly fell from 20.6% to 14.0%. That will undoubtedly be presented as evidence that the vaping reforms worked. But most of the decline occurred among people who vaped less frequently than daily. Daily vaping moved from 9.3% to 8.3%, a decline of only one percentage point, while less-frequent vaping fell from 11.3% to 5.8%, a decline of 5.5 percentage points.

In other words, the apparent policy effect was much larger among occasional users than among daily users, who are more likely to be nicotine-dependent or to use vaping as a stable substitute for smoking. The headline can say that youth and young-adult vaping declined, but the underlying pattern suggests that the people most committed to using the product were far less affected.

At the same time, 8.4% of 18–24-year-olds reported using nicotine pouches during the previous year and 3.8% reported using traditional snus. These percentages cannot simply be added because some respondents may have used both, just as smoking, vaping and pouch percentages cannot be totalled without accounting for overlap. Nevertheless, the figures show substantial interest in smoke-free oral nicotine products despite Australia providing almost no ordinary regulated consumer market for them.

This is the predictable outcome of a policy that tries to suppress nicotine demand by restricting individual products. Consumers adapt. Some move from vapes to pouches, some obtain vapes illegally, some use several products, and some may return to cigarettes. The government can prohibit a preferred route of access, but it cannot repeal demand.

Nowhere is the failure of the pharmacy model clearer than in the source-of-supply findings. Only 2.5% reportedly obtained vapes from a physical pharmacy, while another 4.8% used an online pharmacy. Together, those sources account for 7.3%. Subtracting that from 100% leaves 92.7%, which AIHW rounds to 93%.

The arithmetic is simple and devastating. After imposing a national pharmacy-only supply model, approximately 93% of people using vapes obtained them outside pharmacies, most commonly from tobacconists. The physical-pharmacy estimate is itself marked as having a relative standard error between 25% and 50%, meaning that even this tiny estimate must be treated cautiously.

There is, however, an unresolved question about the denominator. The survey questionnaire directs current users, former users and people who tried vaping only once or twice through questions about where they usually obtained their products. The initial report refers variously to “people who used e-cigarettes” and “people using e-cigarettes” without clearly stating whether the source percentages cover current users, past users or everyone who had ever tried the product. The detailed tables may answer this, but they should have been released with the headline.

Even with that qualification, the broad conclusion is unavoidable: the legal pathway has been rejected by almost the entire market. A regulatory system used by around 7% of its intended consumers cannot reasonably be described as functioning. It is a marginal legal annex attached to a vast illicit market.

The tobacco figures are equally grim. AIHW reports that 34% of current smokers had recently used illicit tobacco, up from 16.7% in 2022–23. That is an increase of 17.3 percentage points and a relative increase of more than 100%. The prevalence of recent illicit use among smokers has more than doubled.

The 34% consists of 22.6% who purchased branded tobacco without Australian plain packaging and graphic health warnings and 16.4% who currently smoked unbranded tobacco, commonly known as chop-chop. These figures overlap because some people did both. The overlap can be reconstructed mathematically:

22.6 plus 16.4 minus 34.0 equals 5.0 percentage points.

That means approximately 17.6% used branded illicit tobacco only, 11.4% used unbranded tobacco only, 5% used both and 66% reported neither. The arithmetic is internally consistent.

But the figure is not a direct estimate of illicit tobacco’s market share. Someone who purchased one illegal packet during the previous three months is placed in the same “recent illicit use” category as someone who buys every cigarette illegally. The measure tells us how widespread contact with the illicit market has become, not what proportion of all cigarettes consumed came from that market.

Even this limited measure produces an alarming result. More than half of those purchasing branded illicit tobacco said they usually bought it from a tobacconist. Australia’s illegal tobacco trade is no longer confined to obscure sellers operating secretly at the edges of society. It has become embedded in an ordinary-looking retail network that operates openly enough for consumers to identify tobacconists as their usual suppliers.

This is where the latest AIHW story collides with the Australian Bureau of Statistics. In June 2026, the ABS released an experimental analysis estimating that the quantity of nicotine consumed in Australia increased by almost 40% between 2017 and 2025, far exceeding population growth. It also estimated that consumption from illicit sources rose dramatically and that illicit tobacco accounted for an extraordinarily large share of total tobacco consumption in 2025.

The NDSHS and ABS measure different things. The household survey estimates the proportion of people reporting particular behaviours, while the ABS attempts to estimate quantities consumed. It is therefore technically possible for smoking prevalence to fall while total consumption remains stable or increases if the remaining smokers consume more, if underreporting changes, or if the two methods capture different parts of the illicit market.

But that is precisely why the discrepancy deserves serious examination. One government statistical agency describes a substantial rise in the quantity of nicotine consumed and an illicit market overwhelming legal supply. Another report states that daily smoking has fallen by almost one-third, and overall nicotine use has reached a historic low. These results should not be placed in separate institutional compartments and selectively quoted according to which one produces the preferred political message.

The report also cannot establish that the 2024 vaping reforms caused the reported decline. It is a repeated cross-sectional survey, not a controlled policy evaluation. It did not follow the same individuals before and after the reforms, identify which smokers switched products, measure what caused people to quit or construct a credible comparison group unaffected by the policy.

The timing of the survey does not solve that problem. A measurement taken after a policy change cannot automatically be credited to that policy. Many other influences could have contributed, including long-term smoking trends, cost-of-living pressure, illicit-market substitution, changes in the survey contractor, recruitment and completion modes, demographic shifts, underreporting and ordinary differences between samples.

The findings may eventually support a careful policy evaluation when combined with other evidence. On their own, they cannot prove that the pharmacy-only system reduced smoking or that Australia’s vaping restrictions produced a net public-health benefit. Any politician or organisation claiming otherwise is asking the survey to answer a causal question it was not designed to answer.

That is the central problem with this initial release. Its limitations are not merely technical footnotes. They affect the meaning of almost every major conclusion. The smoking estimate is weighted, but the unweighted result and weighting details have not been provided. The contractor changed, but the effect on comparability has not yet been demonstrated. The sample missed its stated target, but the response rate has not been released. “Past-year nicotine use” mixes current and past-year measures. Poly-use is compared across a change in product categories. Youth vaping reductions are driven mainly by occasional use. Pharmacy sourcing is overwhelmingly marginal, but its denominator is not clearly explained. Illicit-tobacco use has doubled, but the measure cannot quantify market share.

AIHW says that more detailed findings will be released later. Perhaps those findings will resolve many of these concerns. Perhaps they will show that the 5.6% smoking estimate remains stable across survey modes, unweighted and weighted results, demographic adjustments and alternative assumptions. If so, that would strengthen confidence in the result.

But transparency should accompany the headline, not arrive after it.

By the time the technical qualifications appear, ministers, government departments, health organisations and media outlets may already have repeated the preferred narrative: smoking is collapsing, vaping has stabilised, nicotine use is falling, and Australia’s “world-leading” reforms are working. Corrections and methodological details rarely receive the attention given to an attractive headline on release day.

The most politically useful number came first. The means to scrutinise it were postponed.

That does not prove deliberate manipulation, but it is a form of information control. It allows the public narrative to be established before independent observers can examine how much the change of contractor, recruitment process, response pattern, completion mode and weighting affected the estimate. It foregrounds the apparent success while relegating the collapse of legal supply to the later paragraphs.

A genuinely transparent release would have published the response rate, raw sample counts, unweighted estimates, weighted estimates, confidence intervals, design effects, completion-mode distribution, weighting methodology, questionnaire comparability assessment and sensitivity analysis at the same time as the 5.6% headline. It would have explained precisely how Ipsos maintained comparability with the Roy Morgan series. It would have reconciled the household findings with the ABS consumption estimates. It would have described the limits of causal interpretation prominently rather than leaving readers to infer causation from the report’s policy chronology.

Instead, Australians have been asked to admire the headline and wait for the evidence needed to interrogate it.

The latest report may eventually prove that daily smoking really did fall to 5.6%. That would be welcome. But it already proves something else: the pharmacy-only vaping model has attracted only a tiny fraction of consumers, illicit vaping dominates supply, illicit-tobacco use among smokers has doubled, tobacconists have become leading black-market outlets, and young adults are experimenting with smoke-free nicotine products that Australian policy refuses to regulate proportionately.

This is not a simple story of public-health success. It is the story of a government potentially reducing one visible statistic while losing control of the market beneath it. It is a story in which legal sales disappear more quickly than consumer demand, safer products are pushed underground, survey categories struggle to capture the resulting substitution, and the black market becomes an ordinary part of daily life.

If the government wants to celebrate 5.6%, it should first show exactly how that number was produced. If AIHW wants the public to trust an extraordinary result, it should release extraordinary evidence. If public-health organisations want to claim the reforms worked, they should explain why 93% of people who vape bypassed the legal pathway and why illicit-tobacco use among smokers doubled.

Until then, the report should be treated for what it is: a preliminary and incomplete statistical release containing one highly convenient success claim and several devastating admissions of policy failure.

The number may be correct. The narrative built around it is far less convincing.