On this page
Rotavirus causes severe, dehydrating diarrhoea in babies, and for years vaccines that worked well elsewhere seemed to work less well in India. Gagandeep Kang has spent much of her career at Christian Medical College (CMC), Vellore, finding out why and helping test a vaccine made in India. This post covers two of the studies she worked on and shows you how to turn the raw counts from a vaccine trial into the efficacy figure that makes the news.
Who is Gagandeep Kang?
The Royal Society, which elected her a Fellow in 2019, describes her as known for interdisciplinary research on the transmission, development and prevention of gut (enteric) infections in children in India. Its citation says she built national rotavirus and typhoid surveillance networks, set up laboratories to support vaccine trials and ran phase 1 to 3 clinical trials, work that supported two WHO-prequalified vaccines made by two Indian companies. It lists her position as Professor of Microbiology in the Division of Gastrointestinal Sciences at CMC Vellore, and notes the training programmes she built there for students and young doctors in clinical research.
What did the Vellore birth cohort find?
To understand how children build protection, Kang's team followed babies from birth. The study, Protective Effect of Natural Rotavirus Infection in an Indian Birth Cohort (NEJM, 2011), recruited 452 newborns in urban slums in Vellore, and 373 completed three years of follow-up. Field workers visited homes twice a week, collected stool samples every two weeks and on alternate days during diarrhoea, and took blood every six months.
The results explained a puzzle. In a Mexican birth cohort, two natural infections had given complete protection against moderate or severe disease. In Vellore, infection came early, with 56% of children infected by six months of age. Reinfection was common: only about 30% of the infections found were first infections. Protection against moderate or severe disease rose with each infection but was still only 79% after three. The authors concluded that early infection and frequent reinfection with many strains could explain why rotavirus vaccines had worked less well than expected in Asia and Africa.
How well did the Indian rotavirus vaccine work?
The next question was whether a vaccine developed in India could protect Indian babies. A multicentre trial in Delhi, Pune and Vellore, with Kang among the authors, tested the 116E vaccine, later sold as ROTAVAC. The Lancet 2014 trial report describes a randomised, double-blind, placebo-controlled design: 4,532 infants received three doses of vaccine and 2,267 received placebo, at about 6, 10 and 14 weeks of age.
Worked example: from raw counts to vaccine efficacy
The trial reports 71 cases of severe rotavirus gastroenteritis in 4,752 person-years of follow-up in the vaccine group, and 76 cases in 2,360 person-years in the placebo group. Here is how those become an efficacy figure.
Step 1: why person-years? Children were followed for different lengths of time. One child followed for two years counts as two person-years. Dividing cases by person-years gives a fair rate.
Step 2: the rates. Vaccine: 71 ÷ 4,752 = 0.01494, or about 1.5 cases per 100 person-years. Placebo: 76 ÷ 2,360 = 0.03220, or about 3.2 per 100 person-years. These match the paper.
Step 3: the rate ratio. 0.01494 ÷ 0.03220 = 0.464. The paper reports 0.46.
Step 4: efficacy. Efficacy = 1 − rate ratio = 1 − 0.464 = 0.536, or 53.6%. The paper reports exactly 53.6%, with a 95% confidence interval of 35.0% to 66.9%, and 56.4% in the first year of life.
Notice that 71 and 76 look almost the same. The difference is hidden in the denominators, because twice as many babies got the vaccine as the placebo.
What mistakes do students make with trial numbers?
Comparing raw counts. With a 2:1 split, equal counts would mean the vaccine did nothing. Always compare rates.
Misreading efficacy. 53.6% does not mean 53.6% of vaccinated babies were protected. It means the rate of severe disease in the vaccine group was 53.6% lower than in the placebo group.
Dropping the interval. The confidence interval of 35.0% to 66.9% tells you how precise the estimate is.
Skipping safety. The same report recorded six cases of intussusception (a bowel blockage) in the vaccine group and two in the placebo group, all after the third dose. Honest reading includes the safety table, not just the headline.
For the immunology behind vaccines, see our post on human health and immunity, and for the history of the first vaccine, Edward Jenner and smallpox vaccination.
What can a student take from Kang's way of working?
Kang's work moves in a clear order: measure the disease carefully where it actually happens, explain why existing solutions fall short, then test a new one rigorously. The cohort study came before the trial, and it told researchers what to expect.
When you meet a number in biology, do what the worked example did: ask what was counted, over how long, and compared with what. If rates and ratios trip you up, the Fermi Problem Library has a step-by-step solver that explains the reasoning, so you can check each stage of a calculation like this one.
Open the 2014 trial abstract today and recompute the first-year efficacy yourself from its numbers.
