Watery diarrhoea after a funeral feast in a rural Kano community
A rural community of about 1,400 people in Kano State reports 52 cases of acute watery diarrhoea over four days, and two elderly women have died. Forty-seven of the 52 patients attended a funeral feast three days before falling ill; 210 people attended that feast. The community draws water from two hand-dug wells. Well A is in the compound where the feast was held; a pit latrine stands about 8 metres from it and the concrete apron around the well head is cracked. Well B serves the rest of the village. Among the 210 feasters, 120 drank from well A and 90 drank only from well B; 38 of those who drank from well A became ill, compared with 9 of those who drank only from well B. Stool microscopy from four patients shows no red cells and few pus cells. The local government area has one overstretched primary health centre and no laboratory.
1. Calculate the attack rate among attendees overall and by water source, estimate the relative risk, and interpret what these figures do and do not tell you.
Read the worked answer
Calculations
- Overall attack rate among the 210 attendees = 47/210 = 22.4%.
- Well A drinkers: 38/120 = 31.7%.
- Well B drinkers: 9/90 = 10.0%.
- Relative risk = 31.7/10.0 = 3.17, so well A drinkers were about three times as likely to fall ill.
- Risk difference (attributable risk) = 31.7 - 10.0 = 21.7 percentage points.
- Attributable fraction in the exposed = (3.17 - 1)/3.17 = about 68%, meaning roughly two-thirds of illness among well A drinkers could be attributed to that water if the association is causal.
Interpretation The association is strong, biologically plausible - a cracked apron 8 metres from a pit latrine offers a direct faecal contamination route into a shallow hand-dug well - and the timing fits a point source, with illness clustering three days after a single shared exposure. That triad supports well A as the vehicle, but attack rates alone do not prove causation.
What the numbers do not tell you
- The 10.0% attack rate among well B drinkers is well above what one would expect as background, and needs explaining: secondary person-to-person spread in households, a food-borne exposure at the feast, people misremembering or mixing sources, or an unrelated concurrent outbreak.
- Confounding has not been examined: age, prior exposure and immunity, and whether ill people simply drank more fluid.
- The sample is the feasting cohort. This is a retrospective cohort study, and the correct measure is a risk ratio; a case-control design with an odds ratio would be used if the population at risk could not be enumerated.
- Attack rates depend entirely on the case definition used, which must be stated and applied consistently.
2. Outline what you would do in the first 48 hours, in order of priority.
Read the worked answer
Treat and count at the same time
- Clinical care first: establish oral rehydration points at the health centre and in the community, stock oral rehydration salts and zinc, define referral criteria for severe dehydration, and investigate the two deaths by verbal autopsy.
- Verify the outbreak with a written case definition (for example, three or more watery stools in 24 hours in a resident, with onset after a stated date) and an active case search in households and at the health centre, so that the denominator is not just people who came to the clinic.
- Build a line list and a spot map, then do descriptive analysis by time (an epidemic curve), place (compound) and person (age, sex, attack rates). A single sharp peak within one incubation period would support a point source.
- Test the hypothesis with the cohort already defined: the feast attendees. Calculate attack rates by water source and food item, and report risk ratios with confidence intervals.
- Environmental investigation: sanitary inspection of both wells - apron, lining, rope and bucket, latrine distance and gradient, chlorination - and take water samples for faecal coliform counts.
- Laboratory: stool culture or polymerase chain reaction on fresh specimens for Vibrio cholerae, Shigella, Salmonella and rotavirus, dispatched to the state reference laboratory since the LGA has none.
- Notification: report immediately to the LGA disease surveillance and notification officer and the state epidemiology unit. Cholera and shigellosis are notifiable and may trigger a rapid response team.
- Begin control measures at once; do not wait for laboratory confirmation. Keep a written record so that a full outbreak report can be produced.
3. What immediate and longer-term control measures would you recommend, and how would you know they had worked?
Read the worked answer
Immediate
- Take well A out of use and seal or cordon it; provide alternative safe water by tanker, and distribute household treatment (chlorine tablets or sachets) with instructions.
- Clean and superchlorinate well A before it is returned to service, and test free residual chlorine (aim for 0.2-0.5 mg/L at point of use).
- Free oral rehydration salts and zinc at the health centre and at community distribution points; treat severe dehydration with intravenous fluids on site or refer.
- Hygiene promotion with soap at the four critical times, safe reheating and storage of food, and an immediate stop to open defecation near the wells.
- Disinfect latrines and household surfaces; ensure safe disposal of the faeces of patients.
- If cholera is confirmed, a reactive oral cholera vaccine campaign may be used as an adjunct - never as a substitute for water and sanitation.
- Maintain intensified surveillance for at least two incubation periods (about 10 days) after the last case before declaring the outbreak over.
Longer term
- Replace the hand-dug wells with a protected borehole or a piped supply; if wells are kept, line them, build raised aprons with drainage, and fit a pump rather than an open bucket.
- Relocate or rebuild the pit latrine at least 30 metres from, and downhill of, any water source.
- Community-led total sanitation, water quality testing at intervals, and a stockpile of chlorine, oral rehydration salts and sample kits at LGA level.
How you would know it worked The epidemic curve should fall sharply within one incubation period of the intervention, attack rates in newly exposed households should fall, and repeat water samples should grow no coliforms. A final report with a line list, attack rates and recommendations should go to the LGA and state.