A kidney disease cluster in paddy farmers: what the survey can and cannot tell you
A rural dry-zone district reports a cluster of young men presenting with advanced kidney failure. A rapid survey of 1,200 residents aged 30–60 in two MOH areas found reduced eGFR in 8.4% of men working in paddy cultivation (n = 700) compared with 2.1% of non-farming residents of the same age (n = 500). Cases are often diagnosed late, at eGFR below 15, when they present with vomiting and breathlessness. The district hospital has four dialysis slots. Community leaders are demanding screening of all farmers and an immediate ban on agrochemical spraying. Water sources are shallow wells with high hardness and fluoride. Some households use a reverse-osmosis plant installed by a charity, which has never been tested.
1. Critique the inference that farming causes the kidney disease. What alternative explanations would you test first?
Read the worked answer
The survey measures prevalence, not risk. Prevalence = incidence × duration. A high prevalence among current farmers can arise because cases survive longer in that group, because non-farmers with early disease leave farming (and leave the exposed group), or because affected farmers migrate out of the area and are not counted. Cross-sectional data cannot establish temporality: reduced eGFR may be the consequence of an exposure, or the reason the person stopped or continued a particular job.
Alternative explanations to test.
- Survivor/healthy-worker bias - the exposed group contains only those well enough to keep farming; the unexposed group may include people who left farming because of illness.
- Confounding by age, socio-economic position, water source, heat exposure, alcohol, NSAID and herbal medicine use, snake bite, leptospirosis, and hepatitis or other nephritides.
- Exposure misclassification - 'paddy farmer' is not a dose. Pesticide class, duration, protective behaviour, well depth and water chemistry all matter, and none was measured.
- Outcome misclassification - a single eGFR value, uncorrected for hydration and muscle mass, will misclassify; chronic kidney disease requires abnormality persisting beyond three months.
- Selection - how the 1,200 were chosen is unspecified; clinic-based recruitment would inflate prevalence.
The 8.4% versus 2.1% difference is a signal demanding investigation, not a causal conclusion. A pesticide ban on this evidence would impose large economic costs and might not address the cause.
2. What study design would you propose, with what exposure measurement, and what would make it credible?
Read the worked answer
Design. A prospective cohort of adults aged 30–60 in the affected MOH areas, recruited from the community rather than clinics, with baseline measurement of renal function and exposure, and follow-up for three to five years. A parallel nested case-control study of incident cases would give more efficient exposure measurement. A cross-sectional survey with better exposure data would be cheaper but would reproduce the temporality problem.
Outcome. Standardised serum creatinine with a validated equation, ideally paired with cystatin C to reduce muscle-mass confounding; repeat at 12 months to confirm chronicity; protocol for referral and care of anyone found to have reduced eGFR.
Exposure. Individual-level, not job-title-level: water source and measured water chemistry (hardness, fluoride, and a panel for nephrotoxic metals) at each household; a heat-strain and hydration index; detailed lifetime pesticide and fertiliser history by crop and year; NSAID, alcohol and herbal remedy use; episodes of snake bite and leptospirosis; and socio-economic and dietary data.
Credibility. Pre-specified hypotheses and analysis plan; documented loss to follow-up, with comparison of leavers; adjustment for the confounders above; measurement of exposure blinded to outcome where possible; and community-level environmental sampling so that individual water intake can be linked to measured water quality. I would also test the untested reverse-osmosis plant and compare renal outcomes in its users and non-users - an accidental natural experiment worth exploiting.
3. Community leaders demand mass screening of all farmers. How would you respond, ethically and operationally, given four dialysis slots?
Read the worked answer
Acknowledge the legitimate demand. People are dying undiagnosed at eGFR below 15. Refusing to act is not neutral - it is a decision to let late presentation continue.
Distinguish screening from case finding. Population screening means offering a test to asymptomatic people in a programme designed to benefit them. That requires, at minimum, a test that predicts something, an effective intervention for those detected, agreed criteria for whom to treat, and capacity to investigate and follow up. Here there is no disease-specific treatment, no proven way to halt progression, and dialysis capacity is four slots for a district. Under those conditions, mass screening will generate diagnoses that the system cannot act on, cause anxiety, and expose people to out-of-pocket costs.
What I would offer instead.
- Targeted case finding with a clear pathway - blood pressure and urinalysis for proteinuria, plus creatinine where available, in primary care for adults in the affected MOH areas, with an explicit protocol for what happens to a positive result. Hypertension and proteinuria are treatable today and reduce progression.
- Honest communication - explain that the cause is not yet established and that a ban on spraying is a policy decision to be taken on evidence, not a substitute for it; commit to publishing the study results in the community.
- Water - test the reverse-osmosis plant now and publish the result; advise on safe water without overstating what is known.
- Capacity - quantify the need for dialysis and transplant against the four slots and take that to the provincial level as a resource case.
Screening without a treatment pathway converts an epidemiological problem into an ethical one.