Chills in 'early sepsis': good for you?

We evaluated the predictive value of chills, bacteraemia and endotoxaemia for in-hospital mortality and survival at 5-10 years long-term follow-up in a prospective cohort of 'early sepsis' patients presenting with fever resulting from community-acquired pneumonia or pyelonephritis. Febrile patients with chills had bacteraemia more often (RR 3.1, 95% CI 1.8-5.4) than those without chills. Neither chills nor bacteraemia were significantly related to in-hospital mortality, but patients with endotoxaemia had a higher in-hospital mortality rate than those without endotoxaemia. Patients with chills had a significantly higher survival rate at long-term follow-up than those without chills on admission: the estimated risk of dying was 0.644 (95% CI 0.43-0.95, P = 0.029) for an individual with chills, compared to a person without chills, adjusting for the other factors [age cohort, underlying disease and the pro-inflammatory response in the blood, i.e. tumour necrosis factor-alpha (TNF-alpha) and blood leucocyte number, as scored on hospital admission] in the Cox proportional hazards model. Chills may characterize a patient subpopulation that upon pulmonary and urinary tract infection is able to raise a more rapid and/or efficient host response.