Prediction of hyperglycemia in preterm newborn infants

Predição de hiperglicemia em recém-nascido pré-termo

ORIGINAL ARTICLES

PREDICTION OF HYPERGLYCEMIA IN PRETERM NEWBORN INFANTS

Mário Cícero Falcão and José Lauro Araújo Ramos

RHCFAP/2952

FALCÃO, M. C. et al. - Prediction of hyperglycemia in preterm newborn infants. Rev. Hosp. Clín. Fac. Med. S. Paulo 54 (1): 3 - 8, 1999.

SUMMARY: Many conditions are associated with hyperglycemia in preterm neonates because they are very susceptible to changes in carbohydrate homeostasis. The purpose of this study was to evaluate the occurrence of hyperglycemia in preterm infants undergoing glucose infusion during the first week of life, and to enumerate the main variables predictive of hyperglycemia. This prospective study (during 1994) included 40 preterm neonates (gestational age <37 weeks); 511 determinations of glycemic status were made in these infants (average 12.8/infant), classified by gestational age, birth weight, glucose infusion rate and clinical status at the time of determination (based on clinical and laboratory parameters). The clinical status was classified as stable or unstable, as an indication of the stability or instability of the mechanisms governing glucose homeostasis at the time of determination of blood glucose; 59 episodes of hyperglycemia (11.5%) were identified. A case-control study was used (case = hyperglycemia; control = normoglycemia) to derive a model for predicting glycemia. The risk factors considered were gestational age (£31 vs. >31 weeks), birth weight (£1500 vs. >1500 g), glucose infusion rate (£6 vs. >6 mg/kg/min) and clinical status (stable vs. unstable). Multivariate analysis by logistic regression gave the following mathematical model for predicting the probability of hyperglycemia:

1/exp{-3.1437 + 0.5819(GA) + 0.9234(GIR) + 1.0978(Clinical status)}

The main predictive variables in our study, in increasing order of importance, were gestational age, glucose infusion rate and, the clinical status (stable or unstable) of the preterm newborn infant. The probability of hyperglycemia ranged from 4.1% to 36.9%.

DESCRIPTORS: Glycemia. Hyperglycemia. Prediction. Parenteral infusion. Preterm newborn infant.

The clinical significance of hyperglycemia during the neonatal period differs from center to center and from study to study ^16,17. The upper limit at which a decrease in glucose infusion rate or the installation of insulin therapy is indicated is a blood glucose level between 125 and 240 mg/dl²⁵. In practice, however, intervention in hyperglycemia depends on individual clinical circumstances and on the risk of complications, which occur most commonly in newborn preterm infants submitted to glucose infusion or parenteral nutrition^4,5,15.

In general, the risk of hyperglycemia is greatest in neonates who are immature and have some degree of perinatal asphyxia and respiratory distress^18,23. The most common exogenous factor is parenteral glucose administration³. Hyperglycemia can also be triggered by intravenous lipid infusion⁷ or by surgical intervention1^11,12.

In addition to prematurity, other contributing factors include inappropriate hormone levels and conditions conducive to hormonal imbalance. Thus, intrauterine stress¹³, treatment with aminophylline¹ or corticoids¹⁴, and hypoxemia are associated with elevated incidence of hyperglycemia¹⁶.

It should also be remembered that during the neonatal period, pain causes increases in the liberation of epinephrine, norepinephrine, cortisol, and glucagon and a decrease in insulin^3,11. These pain-induced hormonal changes could result in exacerbating hyperglycemia in preterm neonates, causing a catabolic state undesirable in a newborn infant with already limited nutritional reserves³.

The question of assuring an adequate supply of glucose to the tissues of preterm neonates without provoking hyperglycemia and consequent complications constitutes a real concern, especially when there are factors which prevent oral feeding.

The purpose of the present study was to determine the principal predictive variables of hyperglycemia in preterm neonates undergoing parenteral glucose infusion by correlating the incidence of hyperglycemia with gestational age, birth weight, the rate of glucose infusion, and clinical stability or instability at the time of the blood sample.

PATIENTS AND METHODS

A prospective clinical study was carried out on a series of cases treated in the Nursery of the Santa Catarina Hospital (São Paulo, Brazil), during 1994. The study included 40 neonates classified as preterm (gestational age <37 weeks) that received parenteral solutions containing glucose during the first week of life, and were monitored for glycemic status at least three times daily with the objective of optimizing the glucose infusion rate. Subjects were included in the study only after consent of their parents or guardians.

The clinical status at the time of blood glucose determination was classified as unstable (UCS) or stable (SCS) by clinical and laboratory criteria. The objective of this classification was to evaluate the importance of the presence or absence of clinical stability with regard to the maintenance of glucose homeostasis.

The neonates classified as SCS were normothermic (temperature 36-37°C); showed satisfactory hemodynamics and respiration (O₂ saturation ³92%); had adequate urine output (³1 ml/kg/hr); demonstrated appropriate neurological development for gestational age (examined using a standardized systematic semiological technique incorporating tonus, reflexes, and motor evaluation)²⁰; had normal abdominal palpation and bowel movement; and had adapted to the nutritional regime (complete retention the diet on offer, with the possibility of daily incrementation by 20 ml/day).

When respiration was assisted, the following clinical and laboratory parameters were taken into account in considering the status to be stable, in addition to the factors cited above: respiration rate <60 movements per minute, heart beat 100-160/min, average arterial pressure 45-60 mmHg, O₂ saturation ³92% measured by pulse oximeter, pH 7.25-7.45, PaO₂ >60 mmHg, PaCO₂ <45 mmHg, and urine output ³1 ml/kg/hr.

When the clinical status at any time did not fulfill the conditions above, it was considered to be unstable.

The site chosen for sampling of capillary blood was the inside plantar surface of the heel¹⁰. Blood was collected in three heparinized microtubes (240 ml)⁹. There was a delay of at least one hour after altering the rate of glucose infusion before taking a new blood sample. Blood glucose was determined by the glucose oxidase enzymic/colorimetric method⁹. Determinations were completed within 30 minutes of blood sampling⁹. Hyperglycemia during the neonatal period was defined as a blood glucose concentration in excess of 125 mg/dl^21,22. This figure corresponds to a plasma glucose concentration of approximately 150 mg/dl²⁴.

The glucose solution was administered by peripheral or central venous infusion, avoiding the umbilical blood vessels, using a continuous infusion pump at a known glucose infusion rate that varied depending on the age, nutritional needs, and results of glycemia monitoring of the infant.

For the purposes of statistical analysis, blood glucose was stratified by birth weight, gestational age, glucose infusion rate, and stable or unstable clinical status. A case-control study was chosen in order to construct a multi-variate model for predicting hyperglycemia. The risk factors analyzed were: gestational age (wk) £31 vs. >31, birth weight (g) £1500 vs. >1500, glucose infusion rate (mg/kg/min) £6 vs. >6, and clinical status unstable vs. stable. The influence of these factors was analysed by logistic regression⁸. In all cases, statistical significance was estimated at a p value < 0.05.

RESULTS

The population for the study comprised 40 preterm neonates. Background factors are listed in table 1.

There were 511 determinations of blood glucose, with an average of 12.8 determinations per infant. Of the determinations, 438 (85.8%) were normal, 59 (11.5%) revealed hyperglycemia, and 14 (2.7%) revealed hypoglycemia (Table 2).

Blood samples for 283 (55.4%) determinations of blood glucose were taken when neonates were clinically stable, and 228 (44.6%) were taken when neonates were clinically unstable (p >0.05).

In order to study variables that are predictive of hyperglycemia, a case-control study was constructed. In the study, the cases comprised the times at which blood samples revealed hyperglycemia, and the controls comprised the times at which the blood samples revealed normoglycemia. The cut-off point was blood glucose >125 mg/dl for cases and £125 mg/dl for controls. There were thus 59 cases and 454 controls, giving a case/control ratio of 1:8. These were subjected to univariate and multivariate analysis by logistic regression (Tables 3 and 4).

DISCUSSION

The large number of factors that influence glucose metabolism during the perinatal period make it especially difficult to interpret values for blood glucose and variations from normality, and to identify the causes of hyperglycemia and its complications^13,18.

The first fact that needs to be stressed is that in neonatal practice, it is very difficult to establish limits for stability or instability of carbohydrate homeostasis, especially in preterm neonates²³.

A more precise and reproducible indicator would have been desirable in order to establish an index of clinical instability. However, we were unable to use the various indices validated in the literature (SNAP¹⁹, NTISS⁶, CRIB²), since in our cases, glycemic status was determined repeatedly during the first week of life, with the corresponding clinical status being determined at each sampling. The SNAP index, for example, which is a valid measure of the seriousness of the clinical condition and is applied on admission to neonatal intensive care units, incorporates laboratory data that were apparently not considered to be necessary to the routine care of many of the patients enrolled in the study.

However, some of the parameters employed in SNAP are used in our model to characterize stability: respiration rate <60 movements/minute, heart rate 100-160 beats/minute, mean arterial pressure 45-60 mmHg, PaCO₂ <45 mmHg; urine output ³1 ml/kg/hr (all with SNAP zero), pH 7.25-7.45 (closer to zero), and PaO₂ >60 mmHg (within SNAP zero).

On the basis of the criteria used to characterize stability in our model, it is reasonable to consider that the corresponding SNAP scores would be very close to zero. It is thus probable that stability as defined in our study is equivalent to the most favorable end of the SNAP score range, considered to be normal. Although it is possible that variables that were not analysed or were not included (serum sodium, calcium and urea, etc.) might have increased the respective scores, this does not seem to be very probable given the general conditions of stability of the neonates.

The classification of the clinical status of the infant at the time of sampling as unstable or stable was based on clinical presuppositions. Our results confirm that we were correct in our classifications, since unstable clinical status correlated positively with the development of hyperglycemia.

A case-control study was used in order to reveal the relationships among the most important variables for predicting hyperglycemia, with the cases being episodes of hyperglycemia and the controls being episodes of normoglycemia. The univariate analysis of the four variables studied (Table 3) showed that each presented an odds ratios greater than 1, and p <0.05, and were therefore significant.

In order to determine the weight of each of the variables, multivariate analysis was carried out using logistic regression so as to determine interdependence. The variable "birth weight £1500 g" had an odds ratio less than 1 and a non-significant p value, and was therefore eliminated (Table 4).

Using the formula for determining the probability of the occurrence of an event, that incorporates a constant (a) and coefficients (b) determined by logistic regression, a mathematical model was derived that showed the relationships among the remaining variables predicting hyperglycemia: gestational age £31 wk, glucose infusion rate >6 mg/kg/min, and unstable clinical status⁸ showed that the most important predictors, in increasing order, are gestational age £31 wk, glucose infusion rate>6 mg/kg/min, and unstable clinical status, with the final probability of hyperglycemia being 4.1% in the most favorable cases, and as much as 36.9% in the least favorable cases.

In this study, 11.5% of the determinations showed hyperglycemia, with the value being closer to the lower limit for prediction, showing that this metabolic disturbance can be avoided by frequent monitoring of glycemic status and optimization of the rate of glucose infusion⁴.

Despite theoretical considerations associating the development of high and undesirable levels of blood glucose with immaturity, very low birth weight, and total load of infused glucose, the present study has demonstrated that the fundamental factor for the occurrence of hyperglycemia, and thus of a failure in the mechanisms which regulate glucose homeostasis is, in the final analysis, an imbalance in the overall status of the neonate. It is also very important to emphasize that clinical instability, when combined with prematurity and parenteral feeding at a high glucose infusion rate, leads to a considerably increased risk of this sort of metabolic disturbance^3,16.

Finally, we can also conclude that in the case of healthy neonates, the neonatologist in the day-to-day practice of the Neonatal Intensive Care Unit has more leeway when it comes to levels of glycemia, and can accept higher limits with the aim of improving nutritional status and decreasing morbidity and mortality, and more especially to decrease the time that preterm neonates need to spend in intensive care.

On the other hand, more rigor is needed in relation to levels of glycemia in sick neonates, since it is in this group that hyperglycemia can promote its undesirable effects. It seems clear that the most important factor, especially in extremely premature neonates, is to prevent hyperglycemia through cautious infusion of solutions containing glucose and effective serial monitoring of levels of glycemia¹⁷.

In conclusion, the more unstable the clinical status, the higher the glucose infusion rate, and the lower the gestational age, the higher the incidence of hyperglycemia in preterm neonates given parenteral glucose infusion. Birth weight is not significantly related to the incidence of hyperglycemia and therefore should not be considered as a predictive variable. The principal predictive variables in this study, in decreasing order of importance, were an unstable clinical status, the glucose infusion rate >6 mg/kg/min, and the gestational age £31 wk.

ACKNOWLEDGMENTS: The authors would like to express their thanks to Dr. Crésio Romeu Pereira for his collaboration in the statistical analyses in this study.

RESUMO

RHCFAP/2952

FALCÃO, M. C. e col. - Predição de hiperglicemia em recém-nascido pré-termo. Rev. Hosp. Clín. Fac. Med. S. Paulo 54 (1): 3 - 8, 1999.

A multiplicidade de condições que provocam ou se associam à hiperglicemia em recém-nascidos pré-termo justificam a susceptibilidade destes a desequilíbrios na homeostase dos carboidratos. O objetivo deste é rastrear hiperglicemia em recém-nascidos pré-termo submetidos à infusão de glicose e relacionar as principais variáveis preditoras de hiperglicemia. Foi realizado um estudo prospectivo, durante 1994, com recém-nascidos pré-termo recebendo glicose parenteral na primeira semana de vida. No momento da determinação da glicemia eram conhecidos: idade gestacional, peso de nascimento, velocidade de infusão de glicose e condição clínica do recém-nascido (segundo critérios clínicos - hemodinâmicos e respiratórios - e laboratoriais). Estes "momentos clínicos" foram divididos em "instáveis" ou "estáveis", pela presença ou não de estabilidade na homeostase da glicose no momento da determinação da glicemia. Realizou-se um estudo caso-controle (caso-hiperglicemia; controle-normoglicemia) para o modelo multivariável (regressão logística), preditor de hiperglicemia. Os fatores de risco foram dicotomizados em: idade gestacional £31 e >31sem, peso de nascimento £1500 e >1500g, velocidade de infusão de glicose £6 e >6mg/kg/min e "momento clínico" (estável e instável). Realizaram-se 511 determinações de glicemia, obtendo-se 59 (11,5%) episódios de hiperglicemia em 40 recém-nascidos pré-termo. Pela análise multivariada e regressão logística, o modelo matemático da probabilidade da ocorrência de hiperglicemia foi:

1/expon{-3,1437+0,5819(IG)+0,9234(VIG)+1,0978 (Momento clínico)}

e a probabilidade de hiperglicemia variou de 4,1 a 36,9%. Neste estudo, as principais variáveis preditoras de hiperglicemia, em ordem crescente de importância foram: idade gestacional, velocidade de infusão de glicose e a condição clínica instável do recém-nascido.

DESCRITORES: Glicemia. Hiperglicemia. Predição. Infusão parenteral. Recém-nascido pré-termo.

Received for publication on the 31/08/98

From Santa Catarina Hospital (São Paulo) - Nurscry. Department of Pediatries, Children's Institute, Clinical Hospital, University of São Paulo, Medical School, São Paulo, Brazil.

• 01. BHATT-MEHTA, V.; JOHNSON, C. E.; DONN, S. M. et al. - Accuracy and reliability of dosing equations to individualize theophylline treatment of apnea of prematurity. Pharmacotherapy 1995; 5:246-50.
• 02. COCKBURN, F.; COOKE, R. W. & GAMSU, H. R. - The CRIB (clinical risk index for babies) score: a tool for assessing initial neonatal risk and comparing performance of neonatal intensive care units. Lancet 1993; 342:193-8.
• 03. COLLINS, J. W. et al. - A controlled trial of insulin infusion and parenteral nutrition in extremely low birthweight infants with glucose intolerance. J Pediatr 1991; 118:921-7.
• 04. FALCÃO, M. C. & RAMOS, J. L.. - A. Complicações da hiperglicemia em recém-nascidos pré-termo submetidos à infusão parenteral de glicose. Pediatria (São Paulo) 1997; 19:128-33.
• 05. FINBERG, L. - Danger to infants caused by changes in osmolal concentration. Pediatrics 1967; 40: 1031-4.
• 06. GRAY, J. E.; RICHARDSON, D. K.; MCCORMICK, M. C. et al. - Neonatal therapeutic intervention scoring system: a therapy-based severity-of-illness index. Pediatrics 1992; 90:561-67.
• 07. HALE, D. E. & BENNETT, M. J. - Fatty acid oxidation disorders: a new class of metabolic diseases. J Pediatr 1992; 121:1-11.08.
• 08. HENNEKENS, C. H. & BURING, J. E. - Evaluating the role of confounding. In: MAYRENT, S. L., eds. Epidemiology in medicine. Boston, Little, 1987. p 287-323.
• 09. HOWANITZ, P.; HOWANITZ, J. A. & HENRY, J. B. - Carbohydrates. In: HENRY, J. B., eds. - Clinical diagnosis and management by laboratory methods. Philadelphia, Saunders, 1991. p. 172-87.
• ¹⁰
  HUGHES, W. T. & BUESCHER, E. S. - Procedimentos técnicos em pediatria. Rio de Janeiro, Interamericana, 1980.
• ¹¹
  JONES, M.O.; PIERRO, A.; HAMMOND, P. et al. - The metabolic response to operative stress in infants. J Pediatr Surg 1993; 28:1258-63.
• ¹²
  KLEIN, J. O. & MARCY, S. M. - Bacterial sepsis and meningitis. In: REMINGTON, J. S. & KLEIN, J. O., eds. - Infections disease of the fetus and newborn. Philadelphia, Saunders, 1990. p. 601-89.
• ¹³
  OGATA, E. S. - Carbohydrate homeostasis. In: AVERY, G. B.; FLETCHER, M. A. & MACDONALD, M. G., eds. - Neonatology: pathophysiology and management of the newborn. Philadelphia, Lippincott, 1994. p 568-84.
• ¹⁴
  PAPILE, L. A.; TYSON, J. E.; STOLL, B. J. et al. - A multicenter trial of two dexamethasone regimens in ventilator-dependent premature infants. N Engl J Med 1998; 338:1112-8.
• ¹⁵
  PEREIRA, G. R. - Nutritional care of the extremely premature infant. Clin Perinatol 1995; 22:61-75.
• ¹⁶
  PILDES, R. S. - Neonatal hyperglycemia. J Pediatr 1986; 5:905-7.
• ¹⁷
  PILDES, R. S. & PYATI, S. P. - Hypoglycemia and hyperglycemia in tiny infants. Clin Perinatol 1986; 2:351-75.
• ¹⁸
  RAMOS, J. L. A. - Metabolismo dos hidratos de carbono. In: CARRAZZA, F. R. & MARCONDES, E., eds. - Nutrição clínica em pediatria. São Paulo, Sarvier, 1991. p 45-60.
• ¹⁹
  RICHARDSON, D.K. ; GRAY, J. E.; MCCORMICK, M. C. et al. - Score for neonatal acute physiology: a physiologic severity index for neonatal intensive care. Pediatrics 1993; 91:617-23.
• ²⁰
  SAINT-ANNE DARGASSIES, S. - Normality and normalization as seen in a long term neurological follow-up of 286 truly premature infants. Neuropediatrie 1979; 10:226-44.
• ²¹
  SRINIVASAN, G.; PILDES, R. S.; CATTAMANCHI, G. et al. - Plasma glucose values in normal neonate: a new look. J Pediatr 1986; 109:114-7.
• ²²
  STANLEY, C. & LEVITT-KATZ, L. E. - Disorders of glucose and others sugars. In: SPITZER AR, ed. - Intensive care of the fetus and neonate. St Louis, Mosby, 1996. p 982-92.
• ²³
  TYRALA, E. E.; CHEN, X. & BODEN, G. - Glucose metabolism in the infants weighing less than 1100 grams. J Pediatr 1994; 125:283-7.
• ²⁴
  WILKER, R. E. - Hypoglycemia and hyperglycemia. In: CLOHERTY, J. P.; & STARK, A. N., - Manual of neonatal care. Philadelphia, Lippincott, 1998. p 545-53.
• ²⁵
  ZARIF, M; PILDES, R. S. & VIDYASAGAR, D. - Insulin and growthhormone responses in neonatal hyperglycemia. Diabetes 1976; 25: 428-33.

Publication Dates

• Publication in this collection
  12 Sept 2000
• Date of issue
  Feb 1999

History

• Received
  31 Aug 1998