It is 1991. You are a senior officer at the New York City Health Department.
"Tuberculosis control is really very simple. Just one rule: No cheating. Every patient in your area, you are responsible for their outcome."
This exercise has two connected parts. In Part I — The Inbox, you work through a sequence of messages arriving at the health department as the MDR-TB crisis unfolds. Each message demands a decision. Your choices reveal how surveillance, law, and programme design interact under pressure.
In Part II — The Cohort Dashboard, the programme is underway. You receive a cohort of patients mid-treatment and must allocate limited outreach resources. The dashboard then shows you how your decisions shaped outcomes — and whether you could answer Styblo's question.
Work through each message in order. Some decisions have a clearly better answer. Others involve genuine trade-offs. The debrief at the end draws on the same analytical framework that guided New York City's actual response.
Health Department Inbox
I am writing to alert you to a pattern I cannot explain and cannot ignore. Over the past several months, patients presenting at Harlem Hospital with tuberculosis are not responding to standard first-line therapy. Several have returned weeks or months after discharge, sicker than before. Laboratory results — when they arrive — suggest resistance to isoniazid and rifampicin in multiple unconnected patients.
I do not yet know whether this is a true outbreak or a cluster of unrelated treatment failures. But the number of cases, and the severity of resistance, is unlike anything I have seen. If this is spreading, we may have been losing ground for years without knowing it.
I am bringing this to your attention because the city's current surveillance system has no reliable mechanism for detecting drug resistance at the population level. We are reporting cases, but not susceptibility results. We cannot tell whether this is one strain or many. We cannot tell whether hospitals are amplifying spread.
I believe this warrants urgent investigation. I am available to discuss at any time.
January 1993. Karel Styblo visits New York City.
Styblo asks one question: "Of the 3,811 patients with tuberculosis diagnosed in New York City last year, how many did you cure?"
You cannot answer. The programme has case counts, treatment reports, staff hours, and budget figures. It does not have the answer to this question. The next morning, you begin quarterly cohort reviews. Below is your first cohort — eight patients, all diagnosed in the same quarter. Each has a status flag. You have limited outreach capacity. What do you do?
What happened — and what it means
New York City controlled the MDR-TB epidemic. TB cases fell from 3,811 in 1992 to 1,307 in 2001 — a 66% reduction. MDR-TB cases fell even more dramatically, from 441 in 1992 to 40 in 2001 — a 91% decline. By 2023, the city reported just 381 total TB cases and 3 MDR-TB cases. What many believed unstoppable was brought under control using tools that already existed — once they were organised, funded, and sustained.
Discussion: See, Believe, Create
-
Dr. Brudney's alert worked because she had a connection with the health department. What structural conditions allowed a single clinician's observation to catalyse a system-wide response — and what conditions would have prevented it?
Sentinel surveillance depends on both clinical acuity and communication pathways. Consider: what proportion of important outbreak signals are lost because no such pathway exists, or because the person who noticed was not in a position to be heard?
-
Much of the MDR-TB transmission was occurring in hospitals, not in the community. Why did molecular epidemiology reveal this when traditional surveillance had not — and what does this imply for how surveillance systems should be designed?
DNA fingerprinting established transmission chains that case reports could not. The methodological question is: what class of public health failures are only detectable by methods that most programmes do not routinely use? What is the cost of that gap?
-
Fewer than 2% of TB patients in the first two years of the programme were detained. Yet legal authority under §11.21 is described as indispensable. How can a power used very rarely be essential to a programme's success?
The deterrent function of legal authority operates independently of its application. The existence of credible legal backstops changes the behaviour of patients, providers, and institutions — even when the backstop is never invoked. This is the structural logic of a 'last resort' that also shapes how frequently it becomes necessary.
-
Budget approval took a median of 117 days. The health department did not eliminate the budget review — it moved it to after the hire. Why was this redesign acceptable when simply eliminating the step would not have been? What does this reveal about navigating bureaucratic constraints?
Incremental proposals that preserve the form of existing processes while changing their sequence or timing are harder to reject than proposals that eliminate steps entirely. The logic generalises: changing when accountability occurs is less politically costly than eliminating it. Commissioner Hamburg's instruction — "don't let the perfect be the enemy of the good" — reflects a pragmatism that is both strategically necessary and ethically defensible.
-
Styblo's question — "how many did you cure?" — changed the direction of the programme. What was the programme measuring before, and why did measuring the wrong thing persist for so long without being challenged?
Process indicators are easier to collect, almost always higher, and allow programmes to report progress while patients are lost. Structural incentives — funding tied to activity counts, reporting systems that aggregate inputs — make process measurement self-reinforcing. The shift to outcome accountability requires not just a different metric but a different set of organisational incentives and a willingness to confront uncomfortable results.
The Illusion of Inevitability
A senior CDC tuberculosis official told city leaders that controlling MDR-TB would take at least 10 to 15 years, and likely would not be possible without new diagnostics and new drugs. This was not an unreasonable view. It was also wrong. The illusion of inevitability is not always grounded in bad evidence — it is grounded in the failure to assess what existing tools, properly organised, can accomplish. New York City proved that with surveillance to see, evidence to believe, and systems to create change, public health can succeed.