In the Chocolate

On 16 July, Cadbury was fined £1 million, having pleaded guilty to charges that they had put unsafe chocolate on sale, had failed to alert the authorities that salmonella was in the chocolate, had breached hygiene controls, and had committed six other food safety offences at their Marlbrook manufacturing plant in Herefordshire. Essentially, Cadbury came unstuck because of bad risk-management policies. The worst was a decision they made in 2003 to introduce an ‘allowable tolerance’ level for salmonella in their chocolate. Instead of destroying contaminated product, they would let it onto the market if the number of bacteria in it was very small. This was bad science. It is very difficult to measure numbers of bacteria in foods like chocolate; underestimates are very likely. Also, the bacteria will probably not be evenly distributed, so some portions will contain more than the estimated number. Worst of all, Cadbury failed to draw rational conclusions from the evidence revealed by the study of previous chocolate outbreaks. A possible and plausible explanation is that the scientists advising Cadbury were still being influenced in their thinking by experiments done long ago. The feeding of ‘volunteers’ in US prisons in the 1950s – not with chocolate but with bacterial cultures – showed that the lowest dose that caused infection was about 125,000 live organisms. In the case of one strain, fifty million were needed. The textbooks still say that large numbers are needed to cause infection.

But chocolate and salmonella are good for each other. The abundant fat in chocolate is thought to protect the bacteria as they pass through the stomach, which is why the dose required for infection is very low. Low moisture levels and high sugar content help preserve the bacteria; so, like the chocolate, they have a very long shelf life. That outbreaks are uncommon suggests that manufacturers don’t do too badly. But two or three big chocolate-based outbreaks happen every decade: in Sweden in 1970 (110 cases), North America in 1973-74 (200 cases from Canadian chocolate balls), England and Wales in 1982 (272 cases from Italian chocolate bars), Canada in 1985-86 (from Belgian chocolate coins), Norway and Finland in 1987 (349 cases) and Germany, Denmark and other European countries in 2001-02 (439 cases from German chocolate). Nearly all these incidents came to light because they were caused by rare salmonella serotypes that suddenly became common: S. durham in 1970, S. eastbourne in 1973-74, S. napoli in 1982, S. nima in 1985-86 and S. oranienburg in 2001-02.

There are many salmonella serotypes; about 2000 have been described. In the 1930s, a convention was established that each new one would be named after the place at which it was first isolated. The name means nothing more than that, though the fact that one is named after my home town and another after its most desirable suburb – S. aberdeen and S. rubislaw – means that our bacteriologists must have been testing diarrhoea with enthusiasm over the years.

The Cadbury strain was S. montevideo, another rare serotype. Chocolate outbreak history repeated itself. A strain increases in incidence and so draws attention to itself; when sufferers are questioned, it turns out that the same chocolate was eaten by nearly all of them; the bacterium is then found in the chocolate. That it is identical to the bacteria in the stools of sufferers is proved beyond reasonable doubt by DNA fingerprinting, and the outbreak stops when the chocolate is recalled. S. montevideo was Cadbury’s bad luck. If the strain that dripped into the chocolate crumb from a leaking waste pipe had been a common one, like S. enteritidis, which infects thousands every year in Britain, the forty or so extra cases would almost certainly not have drawn any special attention.

S. enteritidis caused the resignation of Edwina Currie from the Tory government in 1988 when she said ‘most of the egg production in this country, sadly, is now infected with salmonella.’ She was right. A pandemic was in progress. In 1981 there were 395 human infections in England and Wales. In the year she spoke there were 12,522. S. enteritidis does well in the reproductive tract of the chicken. That is why it can occur inside eggs. But like all other salmonellas, it is happiest in the intestine. Any food that can be contaminated with faeces can be a vector. The most notorious British outbreak in recent times was caused by another common serotype, S. typhimurium: 451 patients and staff fell ill at Stanley Royd Hospital in 1984; 19 elderly patients died.

So, salmonella can kill. The young and the old are more at risk: they eat more chocolate. The number of bacteria found in chocolate implicated in outbreaks has been tiny, usually about two per gram. Its presence indicates contamination with bowel contents. Cadbury’s decision to let them go forward in their products was wrong. Even stranger was its failure to have a Hazard Analysis and Critical Control Point system to deliver safer chocolate. HACCP systems analyse risks in ingredients, processes and final foods, identify risk controls and corrective actions for when things go wrong and establish record-keeping and verification procedures. HACCP was developed by Nasa, the Pillsbury food company and the US Army Natick Laboratory in the 1960s (the mere idea of diarrhoea and vomiting in zero gravity was unacceptable). It has been the best spin-off from the space programme, by far. It works. Every food safety authority in the world promulgates it; the United Nations Codex Alimentarius Commission has codified it; it is enshrined as a central component of European Food Law. To develop one would have cost Cadbury thousands. But not having one has cost the company a £1 million fine, the £30 million it lost by recalling its product, and untold damage to its brand.