Cessation model: exclusion of former occasional smokers

[DougManuel]

Background

The cessation APC model (Stage 7) is restricted to former daily smokers (SMKDSTY categories 3 and 4), consistent with the original Manuel et al. (2020) SAS implementation. Former occasional smokers are excluded because their cessation timing variable (SMK_06A_cont) is only available in cycles 2001–2014 and is absent from 2015 onward.

Current decision

Exclude former occasional smokers from the cessation model for the initial PUMF implementation. Cessation age is derived as:

age_cessation = DHHGAGE_cont - time_quit_smoking   # time_quit_smoking = SMK_09A_cont

This covers SMKDSTY 3 (former daily) and 4 (occasional, formerly daily). SMKDSTY 5 (former occasional, never daily) is excluded.

Why this matters

• SMKDSTY 5 (former occasional, never daily) accounts for ~14% of the draft sample
• Their cessation timing is partially available via SMK_06A_cont for 2001–2014 cycles but absent thereafter
• Excluding them means the cessation model reflects only daily smoking trajectories, potentially understating overall cessation rates for cohorts where occasional smoking is more prevalent (younger cohorts)

Proposed future work

1. Assess the proportion of SMKDSTY 5 respondents by cohort and cycle to quantify the gap
2. Evaluate whether SMK_06A_cont can be used for 2001–2014 cycles without introducing cycle-specific bias
3. Consider whether occasional smokers should be handled in a separate sub-model
4. Flag in the study protocol (§3.4.2) as a known limitation

References

• docs/development/apc-plan.md Issue 4
• Manuel et al. (2020) SAS implementation