A simultaneous investigation into certain epidemiological aspects of scarlet fever was carried out over a 2-year period at Oxford, Cambridge and Cardiff. The results obtained at the three centres are compared and contrasted.

Nose and throat swabs were taken from 648 patients on admission to hospital; group A haemolytic streptococci were demonstrated in the swabs of 75·3 %. Eight days after admission 81·9 % had given positive swabs.

Serological typing showed that 56·9 % of strains were types 1–4. The only other common types were types 11 and 8/25.

The average length of stay in hospital was estimated for all cases, divided into groups according to whether or not they had suffered from complications or been cross-infected. The average for the whole series was 30·6 days; this was increased to 33·4 days for all complicated cases and to 37·6 days for all those cross-infected. It was longest (42·3 days) when there were both complications and cross-infections.

The incidence of cross-infection with fresh serological types of haemolytic streptococci was noted both in hospital and for the first 3 weeks after discharge. In the whole series 20·1 % of cases were cross-infected in hospital. The cross-infection rate was highest at Cardiff (27·9 %) and lowest at Oxford (13·5 %). Of the 123 cross-infected cases, complications occurred in 23·6%.

A total of ninety-eight patients (16·4 %) suffered from complications; in twenty-three of these (23·4%) complications were attributed to cross-infecting strains.

Nose and throat swabs were taken from all patients shortly before discharge; 60·3% were still carrying haemolytic streptococci.

Data were obtained regarding the rate at which cases became free of haemolytic streptococci. By the 10th week of the disease only 3·2 % of those who had not been cross-infected were carrying streptococci. Cross-infected patients were slower in becoming negative; at the 10th week 25·8% were still carriers.

There were fourteen return cases (return case rate 2·1%) following the return home of eleven primary cases (infecting case rate 1·7 %).

Swabbing of home contacts at the time of the patient's admission to hospital showed that one or more contacts of 25·5% of cases carried in their naso-pharynx the same type of haemolytic streptococci as that infecting the patient; 12·8% of the contacts swabbed were positive. During the first 3 weeks after discharge 24·6 % of contacts, that is one or more contacts of 55·5% of cases, carried a type that had been harboured by the patient while in hospital.

It was noted that in a particular area the common types that gave rise to scarlet fever also commonly caused other streptococcal infections of the upper respiratory tract. Type 12, however, though frequently found in other streptococcal infections and in healthy carriers, appeared only rarely to cause scarlet fever.

It is suggested that the low incidence of complications, less intimate relationship between complications and cross-infections and low return case rate found in this investigation, as compared to earlier observations made in England, are associated with lowered virulence and invasive powers of the strains of Str. pyogenes responsible for the much milder scarlet fever now prevalent.

The bacteriological findings confirm the wisdom of modern fever hospital practice of early discharge from hospital. It is suggested that discharge within 3 weeks should be the universally adopted rule.