SciELO - Scientific Electronic Library Online

 
vol.22 número1Prognóstico do paciente cirrótico admitido na terapia intensivaRetirada do leito após a descontinuação da ventilação mecânica: há repercussão na mortalidade e no tempo de permanência na unidade de terapia intensiva? índice de autoresíndice de assuntospesquisa de artigos
Home Pagelista alfabética de periódicos  

Serviços Personalizados

Journal

Artigo

Indicadores

Links relacionados

Compartilhar


Revista Brasileira de Terapia Intensiva

versão impressa ISSN 0103-507X

Rev. bras. ter. intensiva vol.22 no.1 São Paulo mar. 2010

http://dx.doi.org/10.1590/S0103-507X2010000100005 

ORIGINAL ARTICLE

 

Risk factors for neonatal death in neonatal intensive care unit according to survival analysis

 

 

Susana de Paula RissoI; Luiz Fernando C. NascimentoII

IMedical Graduation Student of Universidade de Taubaté – UNITAU – Taubaté (SP), Brazil
IIPhD, Assistant Professor of the Internal Medicine Department of the Universidade de Taubaté – UNITAU –Taubaté (SP), Brazil

Author for correspondence

 

 


ABSTRACT

OBJECTIVE: To identify risk factors associated with death in infants admitted to neonatal intensive care unit of Taubaté University Hospital.
METHODS: This is a longitudinal study based on information from medical records regarding newborns admitted to the neonatal intensive care unit of Taubaté University Hospital. The outcome, either discharge or death, was the dependent variable. The independent variables were maternal and gestational variables: maternal age, hypertension, diabetes, corticosteroid therapy and delivery; the newborn variables: birth weight, gestation length, Apgar score in the first and fifth minutes of life, multiple birth, congenital malformations and gender; hospital variables: reports of mechanical ventilation, positive pressure ventilation, prolonged parenteral nutrition, sepsis, intubation, cardiac massage, phototherapy, hyaline membrane disease, oxygen and fraction of inspired oxygen. A model was built with three hierarchical levels for the survival analysis by the Cox model; the software Stata v9 was used, and the final model had variables with p value <0.05. The risks were estimated by the effect measure known as hazard ratio (HR) with 95% confidence intervals. Newborns transferred to other services during the hospitalization were excluded from the study.
RESULTS: Were admitted during the study period 495 newborns, with 129 deaths (26.1%). In the final model, only the variables of corticosteroid use (HR 1.64, 95% CI 1.02-2.70), malformation (HR 1.93, 95% CI 1.05-2.88), very low birth weight (HR 4.28, 95% CI 2.79-6.57) and Apgar scores lower than seven by 1 min (HR 1.87, 95% CI 1.19-2.93) and 5 min (HR 1.74, 95% CI 1.05-2.88) and the variables phototherapy (HR 0.34; 95% CI 0.22-0.53) and intubation (HR 2.28, 95% CI 1 .41-3.70).
CONCLUSION: Factors related primarily to the newborn and the hospitalization (except corticosteroids therapy) were identified to be associated with mortality, highlighting a possible protective factor of phototherapy and the risk for very low birth weight infants.

Keywords: Infant mortality; Intensive care units, neonatal; Infant, low birth weight; Intensive care; Survival analysis; Infant, newborn; Congenital abnormalities


 

 

INTRODUCTION

Survival analysis consists in determining study subjects' survival when exposed to the variables considered risk factor. It is currently known that the study of risk factors for infant mortality is very important, as, particularly in the newborn, it can be considered one of the best quality indicators for health care, as well as an indicator of population social and economic welfare.(1,2)

It is determined by multiple and complex factors related to biological, assistantial and social-economic variables, mainly in its neonatal component.(2-4)

Mosley & Chen(5) studies showed that this close and complex variables interaction may be better studied using hierarchical models, which allow the study of infant mortality determinants and their inter-relationships. Thus, distal factors (antecedents) influence intermediate factors, which in turn influence proximal factors (those acting more directly on the outcome). Thus, by means of a hierarchical structure, it is possible to consider and model distinct factors according to their time precedence and relevance for the outcome determination.(6)

With identification of risk factors, it is possible their prevention, particularly aiming the improvement of newborn children care.(1)

Here, neonatal intensive care units (NICUs) surge as hope, and one of the most effective tools to face this country's newborn mortality issue.(7) In this context, it can be said that the emergence of NICUs brought a widened universe for newborn assistance, offering improved survival for those who, few years ago, wouldn't be likely to survive.(8) It is important to highlight that it useless having a very complex assistance available if appropriate attention is not given to the prenatal and perinatal periods, and to the immediate newborn assistance in the delivery room.

NICU is an area set for treating newborns with a group of birth problems.

Thus, this study aims to identify the risk factors associated with death in newborns staying in the Taubaté University Hospital neonatal intensive care unit between in the years 2005 to 2007.

 

METHODS

This is a longitudinal study, using the survival analysis technique. It was based on information extracted from the medical records of newborns staying in the neonatal intensive care unit of the Taubaté University Hospital (TUH).

The population consisted of born alive infants admitted between January 1st 2005 and December 31, 2007. Were considered cases the newborns with the outcome death, which in survival analysis is called "failure".

The dependent variable was the outcome, either discharge or death. The independent variables were maternal and gestational: maternal age, hypertension, diabetes, corticosteroid therapy and type of delivery; newborn variables: birth weight, pregnancy length, first and fifth minute Apgar score, multiple birth, congenital malformations and gender; hospitalization-related variables: reports of mechanic ventilation, positive pressure ventilation, long term parenteral nutrition, sepsis, intubation, cardiac massage, phototherapy, hyaline membrane disease, stay length (days) and inspired oxygen fraction.

A hierarchical model was built with three survival analysis levels in a Cox model; the Stata v9 software was used, and p value <0.05 variables remained in the final model. The risks were estimated by the hazard ratio (HR) with respective 95% confidence intervals. Were excluded from the study newborns transferred to other services during the hospital stay.

The maternal age variable was categorized as appropriate age (20 to 34 years old) or non-appropriate (others). Hypertension was categorized as present/absent, as well as were diabetes and prenatal corticosteroids use (administration time, dose and number of doses were not considered), and the delivery was categorized as either vaginal or surgical.

The newborns were categorized in three birth weight categories: very low birth weight (< 1,500 g), low birth weight (1,500 g to 2,499 g) and normal birth weight (> 2,500 g). Later, two categories were considered: very low birth weight (below 1,500 g) and others. Regarding the pregnancy length, the newborns were categorized as very preterm (up to 32 weeks of pregnancy), preterm (33 to 36 weeks + 6 days) and full term (37 weeks or more), later categorized in preterm (up to 36 weeks + 6 days) and full term. The fifth minute of life Apgar score was categorized in up to 6 (poor vitality) and 7 to 10 (good vitality). The first minute of life Apgar score was equally categorized. Multiple delivery was categorized as either present or absent.

Hospitalization-related variables: mechanic ventilation, long term parenteral nutrition and sepsis (clinical and/or laboratory diagnosis) categorized as yes/no (present/absent). Inspired oxygen fraction was categorized as room air when no oxygen inspiration was installed other than room air (21%) and high for other values. Intubation, cardiac massage, phototherapy, hyaline membrane disease (clinical-radiological diagnosis) and oxygen therapy were also categorized as either present/absent (yes/no), and the hospital stay length was considered in days.

The data were bivariately analyzed to identify factors with p values <0.20. Next, these variables were evaluated together – multivariate analysis – by Cox regression. This approach provides the hazard ratio (HR), which dependent variable is failure as a function of hospital stay, death in this study, and the independent multivariate analysis variables are those with p<0.20 in the bivariate analysis, having a proportional behavior. Remained in the final model the p < 0.05 variables.

In this analysis strategy a three hierarchical levels model was set, introducing stepwise variables, starting with the more distal variables and simultaneously introducing only variables in the same level. Thus, these variables can function e.g. as confounding factors for the proximal, and mediator for the distal factors.

Each variable effect on the outcome is interpreted as adjusted for the variables belonging to the hierarchically previous levels (more distal) and for the same level variable effects.

The first model level comprised the maternal biological features, as well as gestation-associated conditions, birth assistance and newborn gender. The second level consisted of newborn health and birth status. Proximally to the outcome, third level, neonatal assistance-related aspects were included (Figure 1).

 

 

The Stata version 9 software was used, and the 5% significance level adopted . This study was approved by the UNITAU Ethics Committee (EC), process 325/07.

 

RESULTS

Between 2005 and 2007, 523 newborns were admitted to the TUH NICU, being 28 (5.4%) excluded due to transference. From the 495 newborns study group, 129 died (26.1%), 54% male and 46% female. The mean newborns stay length among those who were discharged was 19 days, while the mean stay length for those who died was 8 days, p<0.001.

These variables underwent survival analysis by the bivariate model and Kaplan-Meier curve. The survival curve is shown in figure 2; 50% of the deaths were seen up to the 10th hospital day. Later, the final result was obtained from the Cox multivariate model. No censored data were considered, as this possibility was prevented by excluding the children transferred to other services.

 

 

In the first level, the variables showing risk significance in the bivariate analysis were non-appropriate maternal age and prenatal steroids use. Meanwhile, cesarean delivery and multiple delivery were within the boundaries of risk factors significance. When hierarchized, the only factor remaining in the model was maternal steroid therapy (Table 1).

 

 

In the second level, all variables were considered significant in the bivariate analysis, remaining in the final model congenital malformation, very low birth weight and 1st and 5th minutes Apgar score below seven (Table 2).

 

 

During immediate post-delivery assistance, third level, in the first analysis all variables were considered significant, remaining in the model as a risk factor intubation, and phototherapy as protective factor (Table 3).

 

 

Finally, were found maternal corticosteroid therapy, congenital malformation, very low birth weight, below seven/poor vitality 1st and 5th minute Apgar score, phototherapy (protective) and intubation (Table 4).

 

 

DISCUSSION

The final result was based on hierarchical analysis, which allowed identifying the factors leading to increased first 28 days of life mortality, by approaching the outcome. With this method it was possible the exclusion of confounding variables and maintaining only pertinent variables.

Thus, it is known that at the distal level there is a close relationship between significant variables in the bivariate analysis – cesarean delivery, non-appropriate maternal age, multiple gestation and prenatal corticosteroid therapy – however remaining as the Cox model adjusted variable, only prenatal corticosteroid use.

This drug therapy aims lung function maturation, acting as a protective factor for mechanic ventilation and oxygen therapy.

Prenatal corticosteroids use is generally considered for women in risk of premature delivery or in other relevant clinical conditions such as preeclampsia, premature amniorhexis, placenta praevia, intrauterine growth restriction, among others, both maternal and fetal. However it should be emphasized that the maternal corticosteroid therapy benefits are observed in the 24-34 weeks of pregnancy range.(9)

The Rede Brasileira de Pesquisas Neonatais [Brazilian Neonatal Research Network] registered in 2004 that in 78% of the study treated pregnant women, this therapy was inappropriately given, leading the intended protective effects to eventually turn into harmful effects for the newborn.(10) Thus, inappropriate use may have been a reason for prenatal corticosteroid therapy appear in this study as a risk factor. The missing information on dosing time and doses may have contributed to this variable appearance as risk factor in our study. This is a possible study limitation.

In the second level, the group requiring more careful neonatal intensive care unit assistance is premature babies – particularly the very premature, their degree of immaturity and Apgar score and birth weight.(11)

The data regarding malformations in this study coincide with the literature. It is currently known that congenital malformation is an important concern for fetal death, first month of life death, and can itself increase the death odds.(12)

According to Amorim(13) in some world regions, particularly in developed countries, this cause overcomes prematurity. In the USA it was found that 45% of neonatal intensive care unit deaths were related to malformations, compared with 40% in this study.

Low birth weight and prematurity are universally acknowledged as the major neonatal mortality risk factors.(4) In this study, a significant association for very low birth weight – below 1,500 grams, was identified, 57%, in agreement with other studies showing that this group mortality is still very high in Brazil, specially in those below 750 grams(14) and far away from the rate in the best services, with about 10% mortality for very low weight infants.(15)

For the Araujo et al.(16) study, low birth weight is associated with unfavorable social-economic situation, making the newborn more vulnerable to conditions such as prematurity. Although the variable gestational age was out the hierarchical model, it is important to take into consideration its influence in the newborn initial clinical status, as it is known that the organs and systems maturation physiological response changes each single gestation week.(14)

According to Shrimpton(17), a mean 100 grams birth weight increase in a low birth weight newborn is associated with 30% to 50% neonatal mortality rate change, showing the importance of programs aimed to improve maternal nutritional status.

It is still worthy emphasizing that neonatal mortality studies have found that the lower the fifth minute of life Apgar score, the lesser is the survival chance.(15) However, it is known that it is possible to see an Apgar score close to six in newborns from high risk pregnancies, cesarean sections or complicated births, being healthy premature babies, with a lower than in a full term baby muscle tonus.(18)

For this study it was found that poor vitality newborns – 5 minutes Apgar below 7 – had a 25% lower survival than those with good vitality – 5 minutes Apgar above 5. The same was seen for the 1 minute Apgar. It is generally found that an asphyxia full term baby has a 10% risk of death, and a neurological dysfunction risk of 10-45%. But, in a pre-term baby, the risks increased by 80% to 95%.(2)

It is exactly the initial newborn clinical severity that will influence the association of the third level variables identified, a reflex of previous levels as obstetric events, delivery and the birth assistance.

Variables as oxygen fraction above 21%, mechanic ventilation and DMH lost their significance in the hierarchical model. The explanation for this may be related to the NICUs ability to provide effective assistance for these infants.

Additionally, other variables in this level which weren't included in the final model could be confounded with increased mortality risk, because these children are clinically more severely ill, needing more specific interventional therapies, such as venous access, cardiac massage, VPP and NPP.

Phototherapy appeared as a protective factor; it was not possible to find an explanation for this; possibly, the longer hospital stay lead these infants to develop jaundice, in turn leading to phototherapy.

Additionally, the need of intubation in a newborn suggests clinical severity, being considered a mortality risk factor during NICU stay in this study.

Mortality risk evaluation by within variables correlation has been used in several countries, allowing to identify new and important factors such as low birth weight and prematurity, which were considered as important isolate variables.(14)

It was thus possible to identify risk factor associated with death in newborns staying in the Taubaté University Hospital neonatal ICU in the years 2005, 2006 and 2007, being these results comparable to others in literature reported populations. In addition, this model allowed hierarchically understanding the relationship between these factors in the causality of infant mortality, using a still seldom used study frame.

Acknowledgements: FAPESP, for the scientific initiation fellowship process 2008/05663-0 and Source of Financing - Process: 08/51485-6.

 

REFERENCES

1. Weirich CF, Andrade ALSS, Turchi MD, Silva SA, Morais-Neto OL, Minamisava R, Marques SM. Neonatal mortality in intensive care unit of Central Brazil. Rev Saúde Pública = J Public Health. 2005;39(5):775-81.         [ Links ]

2. Maran E. Mortalidade neonatal: fatores de risco no município de Maringá-PR em 2003 e 2004 [dissertação]. Paraná: Universidade Estadual de Maringá; 2006.         [ Links ]

3. Schoeps D, Almeida MF, Alencar GP, França Júnior I, Novaes HMD, Siqueira AAF, et al. Risk factors for early neonatal mortality. Rev Saúde Pública = J Public Health. 2007;41(6):1013-22.         [ Links ]

4. Almeida MF, Novaes HMD, Alencar GP, Rodrigues LC. Mortalidade neonatal no Município de São Paulo: influência do peso ao nascer e de fatores sócio-demográficos e assistenciais. Rev Bras Epidemiol. 2002;5(1):93-107.         [ Links ]

5. Mosley WH, Chen LC. An analytical framework for the study of child survival in developing countries. Popul Dev Rev. 1984;10 Suppl:25-45.         [ Links ]

6. Lima S, Carvalho ML,Vasconcelos AGG. Proposta de modelo hierarquizado aplicado à investigação de fatores de risco de óbito infantil neonatal. Cad Saúde Pública = Rep Public Health. 2008;24(8):1910-6.         [ Links ]

7. Nascimento LFC. Fatores de risco para óbito em Unidade de Terapia Intensiva Neonatal. Rev Paul Pediatr. 2009;27(2):186-92.         [ Links ]

8. Reichert APS, Lins RNP, Coullet N. Humanização do cuidado da UTI neonatal. Rev Eletronica Enferm. [Internet] 2007;9(1):200-13. Avaible from: http://fen.ufg.br/revista/v9/n1/v9n1a16.html        [ Links ]

9. Albuquerque ICC, Amorim MMR, Meneses J, Katz L, Santos LC. Avaliação do impacto da corticoterapia antenatal para aceleração da maturidade pulmonar fetal nos recém-nascidos em maternidade-escola brasileira. Rev Bras Ginecol Obstet. 2002;24(10):655-61.         [ Links ]

10. Rede Brasileira de Pesquisas Neonatais. Uso antenatal de corticosteróide e condições de nascimento de pré-termos nos hospitais da Rede Brasileira de Pesquisas Neonatais. Rev Bras Ginecol Obstet. 2004;26(3):177-84.         [ Links ]

11. Sarquis ALF, Mitsuru M, Miyaki M, Cat MNL. Aplicação do escore CRIB para avaliar o risco de mortalidade neonatal. J Pediatr (Rio J). 2002;78(3):225-9.         [ Links ]

12. Almeida MFB, Guinsburg R, Martinez FE, Procianoy RS, Leone CR, Marba STM, et al. Perinatal factors associated with early deaths of preterm infants born in Brazilian Network on Neonatal Research centers. J Pediatr (Rio J). 2008;84(4):300-7.         [ Links ]

13. Amorim MMR, Vilela PC, Santos ARVD, Lima ALMV, Melo EFP, Bernardes HF, et al. Impacto das malformações congênitas na mortalidade perinatal e neonatal em uma maternidade-escola do Recife. Rev Bras Saúde Matern Infant. 2006;6(Supl 1):S19-25.         [ Links ]

14. Brito ASJ, Matsuo T, Gonzalez MRC, Carvalho ABR, Ferrari LSL. Escore CRIB, peso ao nascer e idade gestacional na avaliação do risco de mortalidade neonatal. Rev Saúde Pública = J Public Health. 2003;37(5):597-603.         [ Links ]

15. Araújo BF, Tanaka ACA, Madi JM, Zatti H. Estudo da mortalidade de recém-nascidos internados na UTI neonatal do Hospital Geral de Caxias do Sul, Rio Grande do Sul. Rev Bras Saúde Matern Infant. 2005;5(4):463-9.         [ Links ]

16. Araújo BF, Bozzetti MC. Mortalidade neonatal precoce no município de Caxias do Sul: um estudo de coorte. J Pediatr (Rio J). 2000;76(3):200-6.         [ Links ]

17. Shrimpton R. Preventing low birthweight and reduction of child mortality. Trans R Soc Trop Med Hyg. 2003;97(1):39-42. Review.         [ Links ]

18. Ural SH. What is the Apgar Score? [Internet] Kids Healts, october 2004. [cited 2007 Nov 3]. Available from: http://kidshealth.org/parent/newborn/first_days/apgar.html#        [ Links ]

 

 

Author for correspondence:
Luiz Fernando Costa Nascimento
Rua Durval Rocha, 500 - Vila Paraíba
CEP: 12515-710 - Guaratinguetá (SP), Brazil.
Phone: (12) 3625-4271
E-mail: lfcn@unitau.br

Submitted on September 16, 2009
Accepted on February 9, 2010.

 

 

Received from the Universidade de Taubaté – UNITAU – Taubaté (SP), Brazil.
Conflict of Interest: none to declare.

Creative Commons License Todo o conteúdo deste periódico, exceto onde está identificado, está licenciado sob uma Licença Creative Commons