A treatment-completion percentage can change without a single outcome changing if the denominator changes. A tuberculosis outbreak investigation involving methamphetamine-using social networks illustrates this issue, while also separating evidence of transmission from a proposed drug-related immune mechanism.
What outbreak was investigated?
Pevzner and colleagues investigated a 2005–2006 Snohomish County, Washington, cluster linked to an earlier outbreak. Ten disease cases had epidemiological links; seven also had genotyping links. Eight of the nine adult patients reported methamphetamine use.
A linked cluster is not a random sample of people who use methamphetamine. Its case proportion cannot estimate the probability that a user will develop tuberculosis. Social contact, locations and organism typing provide evidence about a particular network; they do not supply a universal drug-attributable risk.
What did the contact counts mean?
Investigators identified 372 contacts, screened 319, and found 80 with positive tuberculin skin tests. Seventy-one began treatment for latent infection and 57 completed it. Thus, completion was 57/71, or 80.3%, among starters; it was 57/80, or approximately 71.3%, among all test-positive contacts.
Both completion percentages are valid descriptions when their denominators remain visible. They answer different service-delivery questions: retention after treatment begins versus completion across the broader eligible group. The second includes people who did not start; the first does not.
The 80 positive tests should also not be relabeled 80 additional active disease cases. The report distinguished latent infection from disease. A positive test in a contact does not, by itself, identify the date of infection or prove which person transmitted it.
How did investigators find contacts?
Initial patient interviews yielded incomplete contact lists. The team used location-based screening and interviews with shared contacts, alongside outreach, incentives and practical assistance. The database did not separately quantify the number of extra contacts identified by each strategy.
This makes the report useful operational evidence without making it a randomized comparison of outreach components. High completion after a combined program does not identify the independent contribution of transport assistance, incentives, housing or interview technique. Nor does it show what completion would have been in the same network without those measures.
Did methamphetamine itself transmit tuberculosis?
The investigation documented exposure opportunities in a social network. It did not experimentally test a pharmacological effect of methamphetamine on infection susceptibility. A behavior that brings people together and a drug effect on immune cells are different possible pathways.
Related organism genotypes strengthen a transmission investigation when interpreted with contact and timing evidence. They do not establish that the drug, rather than the social exposure setting, caused an individual infection. Assigning that mechanism would exceed what this outbreak design measured.
How should the historical findings be reported?
Keep the stages visible: contacts identified, contacts screened, positive tests, treatment initiation and completion. Use the numerator and denominator corresponding to the question being asked. Separately describe active disease cases and the evidence linking them.
The report demonstrates successful case finding and treatment completion within a difficult-to-reach network. It does not establish a current tuberculosis treatment regimen, a methamphetamine-specific biological risk estimate or the comparative effectiveness of each outreach measure. Its value lies in the documented investigation and the clearly defined outcomes.
Source
Pevzner ES et al. Tuberculosis transmission and use of methamphetamines in Snohomish County, WA, 1991–2006. American Journal of Public Health. 2010;100(12):2481–2486. doi:10.2105/AJPH.2009.162388. Full text. Historical outbreak investigation with program outcomes.