The introduction of BCG vaccination to neonates in Northern Sweden, 1927-31: Re-analysis of historical data to understand the lower mortality among BCG-vaccinated children.

Background: Following the introduction of oral Bacille Calmette-Guérin (BCG) a century ago, Albert Calmette suggested that BCG both provided protection against death from tuberculosis (TB) and other causes. The findings were not pursued. Today, there is considerable evidence that intradermal BCG have beneficial non-specific effects (NSEs). We re-analyzed data from BCG's introduction 1927-1931 in Sweden hypothesizing that BCG reduced infectious deaths.

Methods: In three papers published by Dr Carl Näslund, the progress of oral neonatal BCG rollout provided free-of-charge and the effects on child mortality in the highly TB-prevalent region Norrbotten was sequentially updated. We analyzed cause-specific post-neonatal mortality by vaccination status excluding deaths from congenital conditions. Due to apparent differences in effects during study years, effects were assessed overall and separately in two periods (1927-1929, 1930-1931).

Results: According to Näslund, TB households were slightly more likely to accept vaccination; fewer newborns that were sick or had congenital problems were vaccinated. BCG coverage was 28.3% (5659/20,012); 8.7% (1746/20,012) died. The BCG/unvaccinated Risk Ratio (RR) of post-neonatal childhood death was 0.53 (0.45-0.62). BCG was associated with 80% (49-92%) reduced mortality from TB. From 1927 to 29, BCG appeared to protect strongly against deaths from all diseases, including the non-infectious, RR = 0.09 (0.02-0.36), presumably reflecting selection bias. From 1930 to 1931, there was no protection against non-infectious deaths, RR = 0.92 (0.49-1.70) indicating less bias (p = 0.004 for same effect). During 1930-1931, BCG was associated with reductions in non-TB infectious deaths (RR = 0.75 (0.58-0.97)); 2/3 were caused by respiratory infections, against which the BCG/unvaccinated RR was 0.61 (0.43-0.84). Other causes of death were less frequent and provided no clear pattern, except that BCG was associated with more meningitis deaths, RR = 6.85 (2.20-21.4).

Conclusion: Healthy vaccinee bias, particularly in 1927-1929, resulted in strongly beneficial overall BCG effects. However, the 1930-1931 data provided some support that BCG both protected against TB deaths and deaths from respiratory infections.