Fewer cigarettes, fewer smokers? Read the study's actual outcome

01

A report with three sets of people

Everything in this exercise is invented, not a real intervention or efficacy result. All 100 adults averaged 20 cigarettes per day during the seven days before baseline. At six months, their averages for the preceding seven days are: 20 people at zero, 30 at ten, and 50 still at twenty. Assume complete, accurate, unrounded daily histories, with no missing observations.

For this exercise, the reduction outcome is at least a 50% fall in each person's daily average, including zero consumption. The abstinence outcome is no cigarette smoking, not even a puff, in that final seven-day window. The 20 zero-history people also explicitly report no puffs; the other 80 report some smoking. A rounded daily average alone is not used to establish abstinence.

The table asks two questions of the same 100 people. Those at ten meet the reduction rule but not abstinence. Those at zero meet both. These labels belong only to the invented definitions, not to a reader’s personal status.

02

35%, 50% and 20% can all describe this report

The follow-up group mean is (20×0 + 30×10 + 50×20)/100 = 13 cigarettes per day. From 20 to 13 is a 35% fall in the mean: (20−13)/20. It does not mean that 35% of people quit. Here the baseline is identical for everyone; in other data, a percentage change in group means need not equal the mean of individual percentage changes.

The reduction threshold is met by 20+30=50 people, giving 50/100=50%. The stated abstinence rule is met by 20/100=20%. Adding 50%+20%=70% would count the same 20 people twice, not identify 70 distinct ‘successes’.

If a separate analysis includes only the 80 who still smoke, 30 of those 80 meet the preset reduction threshold: 30/80=37.5% of people. That answers a different population question from 50/100. An accurate summary is: ‘In this invented complete cohort, the mean fell 35%; 50% met the inclusive reduction rule and 20% the seven-day cigarette-abstinence rule.’ With no comparator, these counts do not establish an intervention effect or predict later quitting.

03

Find the rule behind the reported percentage

CONSORT asks reports to specify the outcome variable, metric, aggregation and assessment time. For reduction, locate the baseline period, unit, follow-up period and exact threshold: ‘at least 50%’ differs from ‘more than 50%’. A group mean, average individual change and proportion reaching a threshold are different summaries. Identify whether zero-consumption participants are included, excluded or reported separately.

For abstinence, keep the product definition, covered interval, lapse rule and any verification with the result. Read the analysis denominator and missing-data handling for each outcome; they may be the same or different. Do not choose a favourable threshold or change the primary outcome after seeing the figures. An unknown cigarette count or abstinence response is not a known zero.

[1][2]

04

A cigarette result is not automatically an all-product result

SRNT distinguishes product scopes and recommends recording other tobacco or alternative-product use so the intended outcome can be interpreted. A report about cigarette quantity or cigarette abstinence does not silently answer whether all tobacco or nicotine products were used. Retain the actual scope instead of changing it to ‘nicotine-free’.

Less consumption at one visit also cannot replace a later, separately defined abstinence assessment. Reading endpoint names does not select a gradual or abrupt strategy, establish a health-risk change or judge a person’s progress. Those need their own questions and evidence.

[1]

05

Support does not require earning a research label

NHS Better Health describes stop-smoking support and professional advice. For personal options, contact a qualified service relevant to your location and check current arrangements. This page needs no daily counts, smoking history or health details; a study category is not an admission test for help.

[3]

06

What to keep in mind

• A 35% fall in a mean is not 35% of people quitting.
• A reduction outcome may include abstinent people; check overlap.
• Preserve each result’s definition and analysis population.