A cough that will not go — what it is, and what it is not

The two-week rule, and what it misses

India's National TB Elimination Programme defines a person as presumptive for pulmonary tuberculosis if they have "cough for 2 weeks or more, fever for 2 weeks or more, significant weight loss, haemoptysis, any abnormality in chest radiograph", and directs that such patients have sputum examined.^[1] That is the rule, it is correct, and it should be applied.

It should also be understood for what it is. When cough of two weeks or more is used as a screening test in its own right, its pooled sensitivity across 40 studies and 6,737 patients with TB is 0.42 (95% CI 0.36–0.48), with specificity 0.94.^[2] In plain terms: fewer than half of people who have tuberculosis will report a cough of two weeks or more. The rule is a floor, not a filter. Someone with three weeks of fever and weight loss and no cough at all is still presumptive TB, and someone whose cough has lasted ten days but who is losing weight should not be told to come back in four days.

The scale of the background matters. India's national prevalence survey found microbiologically confirmed pulmonary TB in 316 per lakh of the population aged 15 and over,^[3] and the national notification rate in 2023 was 178.8 per lakh, with treatment success for drug-sensitive TB at 87.6%.^[4] This is a high-prevalence setting, and the pretest probability of TB in a person with a long cough here is not the pretest probability in a textbook written elsewhere.

Which sputum test — this is the part that changes outcomes

Sputum smear microscopy has been the traditional first test and it is still what many people are given. The national prevalence survey tested the same specimens both ways and found MTB in 825 people (2.22%) by CBNAAT against 445 (1.27%) by AFB smear.^[3] Smear found a little over half of what the molecular test found.

The pooled diagnostic accuracy figures say the same thing. Xpert MTB/RIF has a sensitivity of 85% (82–88%) and specificity 98% (97–98%) for pulmonary TB across 70 studies and 37,237 participants — and critically, a sensitivity of 67% (62–72%) in smear-negative, culture-positive disease, which is exactly the group smear microscopy has already failed.^[5] Truenat, the platform actually deployed at peripheral Indian laboratories, performs comparably: Truenat MTB sensitivity 87.6% (81.6–91.8%), Truenat MTB Plus 90.6% (83.7–94.8%) with specificity 95.7%.^[6]

Both also report rifampicin resistance directly — Xpert with sensitivity 96% and specificity 98%,^[5] which matters because it identifies drug-resistant disease at diagnosis rather than after months of failing treatment. India diagnosed 63,929 people with MDR or rifampicin-resistant TB in 2023.^[4]

Despite that, in 2023 a rapid molecular test was offered in only 21% of presumptive TB examinations in India, though the number of NAAT facilities rose from 5,090 to 6,496 in a year and 31.13 lakh Truenat tests were performed.^[4] The practical point for a patient is simple: ask which test was done on your sputum. If it was smear only and it was negative and you are still coughing, that is not a negative result for tuberculosis.

When the cough is not tuberculosis

Most long coughs in this district are not TB, and the alternatives are specific enough to be worth listing.

The blood pressure tablet nobody connects to the cough

This one is worth its own paragraph because it is common, entirely reversible, and routinely missed. ACE inhibitors — ramipril, enalapril, lisinopril, perindopril — cause a dry cough in 5–35% of people who take them. The timing is what defeats the diagnosis: in one prospective cohort, 28.5% of cases began within the first week, but 25.7% began between two and six months after starting, 11.4% between seven and eleven months, and 8.5% only after one to five years.^[8] A cough that starts eight months after a tablet was begun is never attributed to that tablet by the patient, and often not by the doctor either. Bring your medicine strips to the consultation; this is one of the reasons why.

The cooking fire, again

In West Bengal only 20.5% of rural households use clean cooking fuel.^[9] Chronic bronchitis affects 8.7% of never-smoking women cooking on solid biomass against 4.3% on LPG in the nearest comparable eastern Indian population studied.^[10] A productive morning cough in a woman who has never smoked and cooks on a wood stove is a diagnosis, not a mystery. Chronic bronchitis affects 3.49% of Indian adults aged 35 and over overall.^[11]

Tobacco still matters, and it is under-reported because of how the question is usually asked. In West Bengal 48.1% of men and 10.8% of women aged 15 and over use tobacco in some form.^[9] Someone who has never smoked a cigarette may have chewed khaini for thirty years, and will answer "no" to "do you smoke?" quite truthfully.

What is done here

A cough of two weeks or more gets a chest X-ray and sputum sent for molecular testing, arranged locally, alongside a blood count. The medicine list is reviewed for an ACE inhibitor. The kitchen and the tobacco history are asked about specifically rather than in passing. Where the picture is obstructive, spirometry is arranged. Reports are reviewed with you at follow-up, which is not charged within 10 days — deliberately, so that the cost of coming back to have a result explained is not a reason to skip it.

What is not done is a course of antibiotics and a cough syrup for a cough that has already lasted a fortnight, without testing. That is the sequence that turns two weeks of tuberculosis into six months of it.