CoordenaÃÃo de AperfeiÃoamento de NÃvel Superior / A sepse neonatal à atualmente a infecÃÃo mais freqÃente e importante causa de Ãbito de RN internados nas UTIN de paÃses desenvolvidos. Para conhecer a dimensÃo desse problema em um Hospital UniversitÃrio de atendimento terciÃrio localizado no Nordeste do Brasil, foi realizado um estudo de coorte retrospectivo de todos as crianÃas que nasceram e foram admitidas na UTIN no perÃodo de outubro de 1997 a abril de 1998. Uma coorte de 422 pares de recÃm-nascidos e suas respectivas mÃes foi formada; os RN foram seguidos do nascimento à alta ou Ãbito na UTIN ou atà a idade de 28 dias enquanto internados na UTIN. Ao todo, cerca de 34 variÃveis maternas, do neonato e procedimentos hospitalares foram pesquisadas em cada membro da coorte. Os testes estatÃsticos utilizados foram: Teste do Qui-quadrado e o Teste exato de FISCHER, cÃlculo do risco relativo com os respectivos intervalos de confianÃa. Em seguida procedeu-se a anÃlise multivariada com transformaÃÃo para logÃstica dos fatores mais significativos (p<0,05). Ao final, cinco fatores foram selecionados como preditores independentes da sepse neonatal: cateterizaÃÃo venosa central (OR=8,7, IC95%=2,3 a 32,6), faixa ponderal 1000 a 1499g (OR=4,8, IC95%=2,3 a 9,9), transfusÃo de hemoderivados (OR=3,6, IC95%=1,8 a 7,4), gravidez Ãnica (OR=2,3, IC95%=1,0 a 5,4) e faixa ponderal 1500 a 2499g (OR=2,3, IC95%=1,3 a 4,0). A incidÃncia de sepse na coorte foi de 40,4 para cada 100 RN admitidos (167/413). As bactÃrias mais prevalentes dos casos confirmados foram os bacilos gram-negativos; 67% dos episÃdios surgiram nos seis primeiros dias de vida. A internaÃÃo dos RN com sepse foi 4,3 vezes superior a internaÃÃo dos RN nÃo acometidos. A mortalidade global na UTIN foi de 25,59 para cada 100 RN admitidos, enquanto a letalidade pelo desfecho foi de 41,31%, com risco relativo de morte por sepse de 2,8. Este estudo poderà ser Ãtil para futuras estratÃgias com vistas a diminuir a morbimortalidade por sepse neonatal. / OBJECTIVE: Neonatal sepsis is currently the most frequent infection and an important cause of death among the newborns admitted at NICU. In order to evaluate the extension of this problem in a tertiary care University Hospital of Northeastern Brazil, a retrospective cohort survey was carried out on all inborn and admitted infants at the Assis Chateaubriand NICU from October 1997 to April 1998. METHOD: the survey design was a retrospective cohort carried out on all inborn infants admitted at the Neonatal Intensive Care Unit during seven consecutive months; 422 newborns were enrolled in the study and each one was followed up from birth to discharge from NCIU or death at the NICU. To compare the levels of the risk factors, two groups were formed: one by the all subjects who developed the outcome and another by all those who did not to. Each member of the cohort was investigated for 34 potential predictors variables concerning mothers factors, neonates factors and hospital procedures. In case of presence of sepsis, the variables were measured just up to the outcome. Standard National Nosocomial Infection Surveillance (NNIS-CDC) definitions of sepsis were used. Chi Square and Fischerâs exact tests were applied for comparison of frequencies; relative risk (RR) with their respective confidence interval of 95% (CI95%) was calculated. Subsequently, a multivariate analysis was done using logistic regression of most significant factors (OR). The level of statistical significance considered was p=0,05. RESULTS: The cohort sepsis incidence was 40,4 for each hundred of newborn admitted at NICU. The bacterias more prevalent of the confirmed cases were the gram-negative bacilli. Most sepsis episodes appeared in the first six days of life (67%). The time of NICU hospitalization of the sick newborn was 4,3 times longer compared to that non-sick newborn. Five factors were selected as independent predictors for neonatal sepsis: central venous catheter (OR=8,7, CI95%=2,31 to 32,69, p=0,001), birth weight of 1000-1499g (OR=4,8, CI95%=2,39 to 9,97, p=0,000), blood transfusions (OR=3,6, CI95%=1,81 to 7,45, p=0,003), singular gestation (OR=2,3, CI95%=1,04 to 5,44, p=0,04) and birth weight of 1500<2500g (OR=2,3, CI95%=1,34 to 4,04, p=0,002). Global mortality reached 25,59% of the cohort. Mortality associated to sepsis was 41,31% with Relative Risk for death = 2,8. CONCLUSION: neonatal sepsis incidence and mortality rates found are higher than in developed countries rates. Birth weight under 2500g, singular gestation, central venous catheter and blood transfusions proved be independent predictors related to neonatal sepsis. This study may contribute for the future strategies for reduction of neonatal sepsis rates and its sequels in our hospital.