bs_bs_banner
Pediatrics International (2014) 56, 726–730
doi: 10.1111/ped.12371
Original Article
Correlation between hyperglycemia and retinopathy of prematurity Mousa Ahmadpour-Kacho,1 Alireza Jashni Motlagh,2 Seyed Ahmad Rasoulinejad,3 Tahereh Jahangir,4 Ali Bijani5 and Yadollah Zahed Pasha1 1 Non-Communicable Pediatric Diseases Research Center, Department of Pediatrics, 2Amirkola Children’s Hospital, Department of Pediatrics, 3Department of Ophthalmology, 4Amirkola Children’s Hospital, and 5Non-Communicable Pediatric Diseases Research Center, Babol University of Medical Sciences, Babol, IR Iran Abstract
Background: Several risk factors are attributed to retinopathy of prematurity (ROP). This study was done to determine any association between hyperglycemia and ROP in premature infants. Methods: In a retrospective case–control analysis, all infants with a gestational age (GA) < 34 weeks and a birthweight (BW) < 2000 g admitted and treated in the Neonatal Intensive Care Unit at Amirkola Children’s Hospital, Iran, during March 2007–September 2010 were included. Hyperglycemia was defined as a plasma glucose level of >150 mg/dL during the hospital stay. The duration of being hyperglycemic was also recorded. All of these neonates were examined for ROP by a retinologist unaware of group assignment. The difference in the ROP incidence and also the severity of ROP was compared between the hyperglycemic and non-hyperglycemic infants. Matching was done for GA, BW, and also Clinical Risk Index for Babies score. The data were analyzed by t-test, χ2-test and logistic regression test and a P < 0.05 was considered significant. Results: In total, 155 neonates were examined. Seventy (45.2%) of them developed ROP but 85 (54.8%) did not show any evidence of ROP. The frequency of hyperglycemia in patients with ROP was 33 (47.2%), but in those without ROP, hyperglycemia occurred in five (5.9%) (P = 0.0001). The severity of ROP showed no significant differences between the two groups (P = 0.35). The logistic regression for GA and BW showed a significant correlation between hyperglycemia and ROP (P = 0.0001). Conclusions: Hyperglycemia is an important risk factor for ROP that can be prevented along with other risk factors by accurate supervision.
Key words hyperglycemia, infant, premature, retinopathy of prematurity.
Premature infants under 34 weeks’ gestational age and 2000 g birthweight (BW) have become a major clinical population in the neonatal intensive care unit, as a result of improved neonatal care and enhanced survival. Many factors are involved in the development and progression of retinopathy of prematurity (ROP). Many of these infants, especially those under 1000 g, have glucose intolerance resulting in hyperglycemia.1 In fact, a neonate with BW less than 1000 g is 18 times more likely to develop hyperglycemia than one who weighs more than 2000 g.2 An increased risk of reaching threshold retinopathy of prematurity was found to be associated with apnea, longer use of aminophylline, hyperglycemia, and higher score for neonatology acute physiology (SNAP) during the first 24 h from admission to the neonatal intensive care unit (NICU).3 Some researchers have evaluated the long-term adverse effects of hyperglycemia in the neonates, such as intraventricular Correspondence: Mousa Ahmadpour-Kacho, MD, Neonatal Intensive Care Unit, No. 19, Amirkola Children’s Hospital, Amirkola, Babol, Mazandaran 47317-41151, Iran. Email: mousa_ahmadpour@hotmail .com Received 2 March 2011; revised 30 August 2013; accepted 19 February 2014.
© 2014 Japan Pediatric Society
hemorrhage (IVH), ROP and death.4,5 Animal and human studies support the relation between hyperglycemia and circulatory retinal changes, which, in conjunction with hypoxemia, can accelerate the onset and progression of diabetic retinopathy.6–8 Understanding the metabolic complications associated with extremely low birthweight (ELBW) infants is important due to the increased survival of these infants and their extensive morbidities. Of particular interest is the growing concern of cell programming during critical periods of development. Specifically, we do not understand the programming of glucose metabolism regulation and how the pancreas functions during this critical stage of extrauterine development for extremely premature infants and the subsequent development of disease later in life.9–11 Hence, it is important to elucidate glucose metabolism if optimal glucose management may have an impact on the outcomes of these vulnerable infants.12 Furthermore, it is well established that the prevention of proliferative retinopathy in diabetic adults requires the tightest possible glucose control.13 In order to assess the effects on clinical outcomes of interventions for treating neonatal hyperglycemia in the very low birthweight (VLBW) neonate receiving total or partial parenteral nutrition, Bottino et al. in a review concluded that “it is not clear whether
Hyperglycemia and ROP hyperglycemia per se is a cause of adverse clinical outcomes or how hyperglycemia should be treated.”14 Although the relation between proliferative retinopathy and poor glucose control in adult diabetic patients is well known, the relation between ROP and hyperglycemia in VLBW infants is questionable. So, this study was carried out to determine any association between hyperglycemia and ROP.
Methods In a retrospective case–control study, we enrolled all live-born infants with a GA less than 34 weeks and BW less than 2000 g who were admitted to an NICU and the Newborn Services at Amirkola Children’s Hospital, a referral hospital that is affiliated with Babol University of Medical Sciences, Babol, northern Iran, from March 2007 to September 2010. GA was determined by the attending neonatologist based on clinical data along with maternal dates and/or prenatal ultrasound when available. All data were obtained by reviewing the existing patients’ medical records. A patient with ROP was considered as a case and a neonate without ROP as a control group. A total of 184 charts were reviewed. Twenty-nine infants were excluded; they were defined as infants who received no ophthalmologic examination due to early death (27 infants) or congenital eye anomalies (two patients). According to the local protocol for ROP examination (Table 1), all of these neonates were examined for ROP by a retinologist. Follow-up examinations were planned, based on the findings of the previous first examination. The patients with highgrade ROP were referred for laser, cryotherapy, surgery or antivascular endothelial growth factor (VEGF) therapy. All infants were followed up for ROP after hospital discharge and the patients with ROP needing treatment were referred for further treatment. The stages and the zone of ROP were expressed based on the international classification of ROP (IC-ROP).15All babies with severe ROP defined by the presence of prethreshold and threshold ROP were referred for the treatment. Matching was done between ROP and control groups for GA, BW and Clinical Risk Index for Babies (CRIB) score. Clinical and demographic data
Table 1 Schedule for first indirect ophthalmoscopy and follow-up exam in premature infants for ROP in Amirkola Children’s Hospital, Babol, Iran Who • All infants with a gestational age < 34 weeks and a birthweight < 2000 g admitted and treated in the Newborn Services and/or NICU When • By the end of 31 weeks postmenstrual age† or 4 weeks after birth • Recommend first examination before discharge from the hospital Follow-up • Prethreshold and threshold ROP: refer for treatment • Stage 2, zone 2: Repeat examination after 1 week • Other stages: Repeat examination after 2 weeks † Postmenstrual age in weeks is equal to the gestational age at birth plus the chronological age in weeks after birth. NICU, neonatal intensive care unit; ROP, retinopathy of prematurity.
727
extracted from medical charts included: GA, BW, sex, ABG records, transfusion, underlying diseases, CRIB score and blood glucose level. Prenatal data collected included: antenatal corticosteroid use, as reported by an obstetrician. Also the following data were recorded during the hospital stay: first serum glucose level, the duration of hyperglycemia, glucose infusion rate and insulin use with regard to the day of initiation and the number of days used. The measuring of glucose levels was performed using the glucose oxidation technique (AU 640 Immune Analyzer, Olympus, Center Valley, PA, USA), first at the time of admission then at least twice a day up to 5 days and then once a day until the time the baby goes to oral full fed. Glucose measuring was repeated two to three times a week or at a frequency determined by the clinical status of each infant until the baby was discharged. If the glucose value was >150 mg/dL, the number of days with hyperglycemia was recorded. During the study period, the usual practice for advancing oral milk and decreasing parenteral dextrose 10% in water (D10W) was the clinical status and blood sugar of patients. The goal of the management was to keep the serum glucose level between 80 and 150 mg/dL. The glucose infusion rate (GIR) was determined by using glucose rate calculator introduced by Klaus and Fanaroff.16 Hyperglycemia was defined as a plasma glucose concentration of greater than 150 mg/dL (8.3 mmol/L) on at least two different occasions during the hospital stay; this level has been previously identified as being clinically relevant. The criterion for insulin therapy was determined by blood glucose level more than 300 mg/dL on at least two occasions. Outcomes variables included: ROP and its stages and grades, also the need for any treatment and final visual outcome. All statistical analyses were performed using spss 19 (spss, Chicago, IL, USA). χ2-tests and t-tests were used for the distributions and means of demographic and clinical variables to compare across the study subgroups. Standard deviations are shown for continuous variable. Logistic regression was employed to assess the correlation of hyperglycemia, and the association of hyperglycemia to ROP in relation to the other risk factors, after adjusting for GA and BW. As numerous risk factors might have some relation to ROP, we only included in the final multivariate analysis those risk factors that had a close association to ROP in our study, which was indicated by a P-value less than 0.05.
Results One hundred and fifty-five neonates were included in this study in which 70 (45.2%) of them developed ROP, but 85 (54.8%) did not show any evidence of ROP. Thirty-eight (24.5%) of ROP patients had hyperglycemia during the hospital stay, but only five (5.9%) patients in the control group experienced hyperglycemia in their critical hospital period (P = 0.0001). No differences in sex, BW, GA and CRIB score were found between ROP and control infants as shown in Table 2. In our study, variables like transfusion, PaO2 ≥ 75 and blood sugar (BS) ≥ 150 continued to reach statistical significance after adjusting the other study covariates (Table 3). © 2014 Japan Pediatric Society
728
M Ahmadpour-Kacho et al.
Table 2 Demographic characteristics of ROP and non-ROP premature neonates at the time of enrolment ROP infants (n = 70) 29.91 ± 2.46 1238.57 ± 344.77 5.63 ± 2.45 34/36 (49%/51%) 48 (43%)
Study characteristics Gestational age (weeks) (mean ± SD) Birthweight (g) (mean ± SD) CRIB score (mean ± SD) Sex (male/female) Antenatal steroid received (%)
Control infants (n = 85) 30.59 ± 1.97 1327.53 ± 293.03 5.42 ± 2.24 46/39 (54%/46%) 64 (57%)
P 0.06 0.09 0.59 0.49 0.35
CRIB, Clinical Risk Index for Babies; ROP, retinopathy of prematurity.
Both PaO2 ≥ 75 and hyperglycemia had correlation with ROP. Sixty-nine patients (98.6%) in the retinopathy group had PaO2 ≥ 75 but 62 neonates (72.9%) in the control group had PaO2 ≥ 75 (P < 0.001). Although mothers of 112 patients had received antenatal steroid (betamethasone), no differences were detected between the two groups. The mean plasma glucose level and duration of being hyperglycemic were higher in the ROP group than in the control group (Table 4). The glucose infusion rate (GIR) in ROP infants was significantly higher than that in the non-ROP infants (5.8 vs 5.1 mg/kg/ min, respectively, P = 0.01). The median number of days to reach maximum serum glucose between case and control groups was 2.3 and 1.3 days, respectively (P = 0.04). Hyperglycemia was associated with an increase in ROP even after matching in terms of GA, BW and omitting the transfusion
Table 3 Bivariate analysis for factors associated with retinopathy of prematurity Study characteristics PaO2 ≥ 75 FiO2 ≥ 60% Blood sugar ≥150 Transfusion Mechanical ventilation
OR 25.59 1.27 14.27 2.60 1.10
95%CI 3.35–195.14 0.586–2.78 5.16–39.50 1.20–3.53 0.38–1.87
P 0.002 0.537 0.000 0.008 0.687
Table 4 Glycemic characteristics of ROP and non-ROP premature neonates (mean ± SD) Study characteristics Mean total serum glucose (mg/dL) Mean first serum glucose (mg/dL) Min & max of serum glucose (mg/dL) Glucose infusion rate (mg/kg/min) Duration of hyperglycemia (days) Insulin use
ROP infants (n = 70) 147.53 ± 71.14
Control infants (n = 85) 93.80 ± 37.37
0.0001
141.14 ± 114.60
88.16 ± 72.31
0.001
15–351
0.001
20–560
P
5.80 ± 2.15
5.10 ± 1.08
0.01
2.32 ± 2.37
1.33 ± 0.87
0.04
2 (100%)
0 (0%)
0.2
ROP, retinopathy of prematurity.
© 2014 Japan Pediatric Society
effect (Table 5). Therefore, only covariates of primary conceptual interest were used in the multivariate analysis. Severity of ROP did not reach statistical significance between the hyperglycemic and normoglycemic infants (P = 0.1). Among babies with ROP, 18 (54.5%) patients in the hyperglycemic group had prethreshold and threshold ROP, but 25 (67.6%) patients in the normoglycemic group developed prethreshold and threshold ROP (P = 0.264).
Discussion This study has demonstrated that hyperglycemia is a significant risk factor for retinopathy of prematurity. The elicited cut-off point for hyperglycemia in studies by Ertl et al.,17 Garg et al.5 and Blanco et al.1 was 150 mg/dL, which is in line with this study. Those authors also concluded that hyperglycemia might play an important role as a risk factor for ROP. In our study population, the overall incidence of hyperglycemia was 24.5%, which was close to the 19.4% in the study by Ertl et al.17 The investigation by Garg et al. showed that glucose infusion rate and duration of hyperglycemia like our study were risk factors eligible for ROP.5 Mean serum glucose level and duration of hyperglycemia play effective roles in the occurrence of ROP, as reported by Vendal et al. and our study supports this.18 In another study by Mohamed et al., also evaluating hyperglycemia as a risk factor for ROP in premature infants less than 32 weeks gestation, it was shown that neonatal hyperglycemia was significantly associated with an increased risk of the development of ROP in premature infants.19 The majority of previous studies were performed on several risk factors of ROP other than hyperglycemia. Postnatal serum insulin like growth factor-1 (IGF-1) level in addition to weight development may correctly identify neonates at risk for severe ROP.20
Table 5 Multivariate analysis for factors associated with retinopathy of prematurity Study characteristics Gestational age (weeks) Birthweight (g) Transfusion (times) Mean blood sugar (mg/dL) Glucose infusion rate (mg/kg/min) Duration of hyperglycemia (days) PO2 (torr)
OR 0.95 0.99 1.99 1.03 1.36 1.07 1.01
95%CI 0.64–1.4 0.99–1.0 0.70–5.60 1.01–1.05 0.92–2.02 0.54–2.12 0.99–1.02
P 0.801 0.566 0.191 0.001 0.331 0.844 0.275
Hyperglycemia and ROP Beardsall et al. suggested a significant correlation between ROP and hyperglycemia after adjusting for birthweight, gestational age and CRIB score, which was comparable with our results.21 The adjustment of birthweight, gestational age and steroid use was done by Blanco et al.1 and hyperglycemia was associated with a statistically significant increase in ROP, which was compatible with the findings in our study. Intraventricular hemorrhage had no relation to hyperglycemia in either of these studies.1 The duration of oxygen and ventilator therapies and IVH was not associated with ROP according to the reports of Kim et al., although hyperglycemia was detected as a significant risk factor.3 This result has been confirmed by our study. As the numbers of the bronchopulmonary dysplasia patients in our NICU graduates are very low (about 1% of VLBW infants) we did not decide to include the days on oxygen use as a variable. The reason for low incidence of bronchopulmonary dysplasia patients in our NICU graduates may be explained either by the higher gestational age of the NICU population, ethnicity or geographical parameters.22 Instead, we tested for FiO2 and PaO2. PaO2 ≥ 70 played an important role but a FiO2 ≥ 60% did not get a significant correlation with ROP. In our study, hyperglycemia was not more important than PaO2 ≥ 75 based on the odds ratio for PaO2 ≥ 75 and BS ≥150 (Table 3) but it is the second most important factor for the occurrence of ROP. Hyperoxemia during resuscitation after birth and during the mechanical ventilation in premature infants born at ≤32 weeks of gestation is the most important risk factor for ROP. For prevention, it is recommended to keep the oxygen saturation between the 85% and 93%. In this range of SpO2 in about 90% of patients, the PaO2 would be in the desirable range (40– 80 mmHg).23 In three recent international randomized, controlled trials, a group of researchers from the UK, Australia, and New Zealand evaluated the effects of targeting an oxygen saturation of 85–89%, as compared with a range of 91–95%, on disability-free survival at 2 years in infants born before 28 weeks’ gestation. These studies showed that those infants in the lower-target group for oxygen saturation had a reduced rate of ROP and an increased rate of necrotizing enterocolitis and death. Therefore, targeting an oxygen saturation of less than 90% among such infants should be avoided.24 GA and BW are the major determinant risk factors for ROP, but contrary to the report by Ertl et al., gestational age and CRIB score did not affect ROP in our study.17 Sutija et al. suggested that ROP developed more frequently in lower GA and BW than in the control and they also had more IVH than control groups. However, after adjusting BW and GA, we, like others, have not found any association between ROP and IVH.25 Although GA and BW were lower in ROP infants in the study by Vendal et al., longer duration of ventilation and higher level of blood sugar were reported by those authors, which was not detected in our study.18 The patients in the study by Garg et al. needed more insulin for normalizing blood sugar than the control group, which was not detected in our cases. A study by Yoo et al. on 7-day-old
729
rat pups concluded that “insulin treatment substantially increased VEGF levels, extraretinal vessel formation, matrix metalloproteinase activity, and the extent of retinal hemorrhage in rat pups with mild oxygen-induced retinopathy compared with saline controls.”26 Also, in a review by Kaempf et al., the authors concluded that “hyperglycemia and especially insulin use in premature infants may increase the risk of ROP.” They also found that “slower NICU growth velocity, but not rate of head or length growth, was predictive of ROP.”27 However, insulin infusion for hyperglycemia in premature infants was not associated with an increased risk on ROP in a study by Heald et al.28 Investigations by Blanco et al. showed a higher incidence (88%) of hyperglycemia in their patients than in ours (24.5%). It was partially because of their lower birthweight (
Pediatrics International (2014) 56, 726–730
doi: 10.1111/ped.12371
Original Article
Correlation between hyperglycemia and retinopathy of prematurity Mousa Ahmadpour-Kacho,1 Alireza Jashni Motlagh,2 Seyed Ahmad Rasoulinejad,3 Tahereh Jahangir,4 Ali Bijani5 and Yadollah Zahed Pasha1 1 Non-Communicable Pediatric Diseases Research Center, Department of Pediatrics, 2Amirkola Children’s Hospital, Department of Pediatrics, 3Department of Ophthalmology, 4Amirkola Children’s Hospital, and 5Non-Communicable Pediatric Diseases Research Center, Babol University of Medical Sciences, Babol, IR Iran Abstract
Background: Several risk factors are attributed to retinopathy of prematurity (ROP). This study was done to determine any association between hyperglycemia and ROP in premature infants. Methods: In a retrospective case–control analysis, all infants with a gestational age (GA) < 34 weeks and a birthweight (BW) < 2000 g admitted and treated in the Neonatal Intensive Care Unit at Amirkola Children’s Hospital, Iran, during March 2007–September 2010 were included. Hyperglycemia was defined as a plasma glucose level of >150 mg/dL during the hospital stay. The duration of being hyperglycemic was also recorded. All of these neonates were examined for ROP by a retinologist unaware of group assignment. The difference in the ROP incidence and also the severity of ROP was compared between the hyperglycemic and non-hyperglycemic infants. Matching was done for GA, BW, and also Clinical Risk Index for Babies score. The data were analyzed by t-test, χ2-test and logistic regression test and a P < 0.05 was considered significant. Results: In total, 155 neonates were examined. Seventy (45.2%) of them developed ROP but 85 (54.8%) did not show any evidence of ROP. The frequency of hyperglycemia in patients with ROP was 33 (47.2%), but in those without ROP, hyperglycemia occurred in five (5.9%) (P = 0.0001). The severity of ROP showed no significant differences between the two groups (P = 0.35). The logistic regression for GA and BW showed a significant correlation between hyperglycemia and ROP (P = 0.0001). Conclusions: Hyperglycemia is an important risk factor for ROP that can be prevented along with other risk factors by accurate supervision.
Key words hyperglycemia, infant, premature, retinopathy of prematurity.
Premature infants under 34 weeks’ gestational age and 2000 g birthweight (BW) have become a major clinical population in the neonatal intensive care unit, as a result of improved neonatal care and enhanced survival. Many factors are involved in the development and progression of retinopathy of prematurity (ROP). Many of these infants, especially those under 1000 g, have glucose intolerance resulting in hyperglycemia.1 In fact, a neonate with BW less than 1000 g is 18 times more likely to develop hyperglycemia than one who weighs more than 2000 g.2 An increased risk of reaching threshold retinopathy of prematurity was found to be associated with apnea, longer use of aminophylline, hyperglycemia, and higher score for neonatology acute physiology (SNAP) during the first 24 h from admission to the neonatal intensive care unit (NICU).3 Some researchers have evaluated the long-term adverse effects of hyperglycemia in the neonates, such as intraventricular Correspondence: Mousa Ahmadpour-Kacho, MD, Neonatal Intensive Care Unit, No. 19, Amirkola Children’s Hospital, Amirkola, Babol, Mazandaran 47317-41151, Iran. Email: mousa_ahmadpour@hotmail .com Received 2 March 2011; revised 30 August 2013; accepted 19 February 2014.
© 2014 Japan Pediatric Society
hemorrhage (IVH), ROP and death.4,5 Animal and human studies support the relation between hyperglycemia and circulatory retinal changes, which, in conjunction with hypoxemia, can accelerate the onset and progression of diabetic retinopathy.6–8 Understanding the metabolic complications associated with extremely low birthweight (ELBW) infants is important due to the increased survival of these infants and their extensive morbidities. Of particular interest is the growing concern of cell programming during critical periods of development. Specifically, we do not understand the programming of glucose metabolism regulation and how the pancreas functions during this critical stage of extrauterine development for extremely premature infants and the subsequent development of disease later in life.9–11 Hence, it is important to elucidate glucose metabolism if optimal glucose management may have an impact on the outcomes of these vulnerable infants.12 Furthermore, it is well established that the prevention of proliferative retinopathy in diabetic adults requires the tightest possible glucose control.13 In order to assess the effects on clinical outcomes of interventions for treating neonatal hyperglycemia in the very low birthweight (VLBW) neonate receiving total or partial parenteral nutrition, Bottino et al. in a review concluded that “it is not clear whether
Hyperglycemia and ROP hyperglycemia per se is a cause of adverse clinical outcomes or how hyperglycemia should be treated.”14 Although the relation between proliferative retinopathy and poor glucose control in adult diabetic patients is well known, the relation between ROP and hyperglycemia in VLBW infants is questionable. So, this study was carried out to determine any association between hyperglycemia and ROP.
Methods In a retrospective case–control study, we enrolled all live-born infants with a GA less than 34 weeks and BW less than 2000 g who were admitted to an NICU and the Newborn Services at Amirkola Children’s Hospital, a referral hospital that is affiliated with Babol University of Medical Sciences, Babol, northern Iran, from March 2007 to September 2010. GA was determined by the attending neonatologist based on clinical data along with maternal dates and/or prenatal ultrasound when available. All data were obtained by reviewing the existing patients’ medical records. A patient with ROP was considered as a case and a neonate without ROP as a control group. A total of 184 charts were reviewed. Twenty-nine infants were excluded; they were defined as infants who received no ophthalmologic examination due to early death (27 infants) or congenital eye anomalies (two patients). According to the local protocol for ROP examination (Table 1), all of these neonates were examined for ROP by a retinologist. Follow-up examinations were planned, based on the findings of the previous first examination. The patients with highgrade ROP were referred for laser, cryotherapy, surgery or antivascular endothelial growth factor (VEGF) therapy. All infants were followed up for ROP after hospital discharge and the patients with ROP needing treatment were referred for further treatment. The stages and the zone of ROP were expressed based on the international classification of ROP (IC-ROP).15All babies with severe ROP defined by the presence of prethreshold and threshold ROP were referred for the treatment. Matching was done between ROP and control groups for GA, BW and Clinical Risk Index for Babies (CRIB) score. Clinical and demographic data
Table 1 Schedule for first indirect ophthalmoscopy and follow-up exam in premature infants for ROP in Amirkola Children’s Hospital, Babol, Iran Who • All infants with a gestational age < 34 weeks and a birthweight < 2000 g admitted and treated in the Newborn Services and/or NICU When • By the end of 31 weeks postmenstrual age† or 4 weeks after birth • Recommend first examination before discharge from the hospital Follow-up • Prethreshold and threshold ROP: refer for treatment • Stage 2, zone 2: Repeat examination after 1 week • Other stages: Repeat examination after 2 weeks † Postmenstrual age in weeks is equal to the gestational age at birth plus the chronological age in weeks after birth. NICU, neonatal intensive care unit; ROP, retinopathy of prematurity.
727
extracted from medical charts included: GA, BW, sex, ABG records, transfusion, underlying diseases, CRIB score and blood glucose level. Prenatal data collected included: antenatal corticosteroid use, as reported by an obstetrician. Also the following data were recorded during the hospital stay: first serum glucose level, the duration of hyperglycemia, glucose infusion rate and insulin use with regard to the day of initiation and the number of days used. The measuring of glucose levels was performed using the glucose oxidation technique (AU 640 Immune Analyzer, Olympus, Center Valley, PA, USA), first at the time of admission then at least twice a day up to 5 days and then once a day until the time the baby goes to oral full fed. Glucose measuring was repeated two to three times a week or at a frequency determined by the clinical status of each infant until the baby was discharged. If the glucose value was >150 mg/dL, the number of days with hyperglycemia was recorded. During the study period, the usual practice for advancing oral milk and decreasing parenteral dextrose 10% in water (D10W) was the clinical status and blood sugar of patients. The goal of the management was to keep the serum glucose level between 80 and 150 mg/dL. The glucose infusion rate (GIR) was determined by using glucose rate calculator introduced by Klaus and Fanaroff.16 Hyperglycemia was defined as a plasma glucose concentration of greater than 150 mg/dL (8.3 mmol/L) on at least two different occasions during the hospital stay; this level has been previously identified as being clinically relevant. The criterion for insulin therapy was determined by blood glucose level more than 300 mg/dL on at least two occasions. Outcomes variables included: ROP and its stages and grades, also the need for any treatment and final visual outcome. All statistical analyses were performed using spss 19 (spss, Chicago, IL, USA). χ2-tests and t-tests were used for the distributions and means of demographic and clinical variables to compare across the study subgroups. Standard deviations are shown for continuous variable. Logistic regression was employed to assess the correlation of hyperglycemia, and the association of hyperglycemia to ROP in relation to the other risk factors, after adjusting for GA and BW. As numerous risk factors might have some relation to ROP, we only included in the final multivariate analysis those risk factors that had a close association to ROP in our study, which was indicated by a P-value less than 0.05.
Results One hundred and fifty-five neonates were included in this study in which 70 (45.2%) of them developed ROP, but 85 (54.8%) did not show any evidence of ROP. Thirty-eight (24.5%) of ROP patients had hyperglycemia during the hospital stay, but only five (5.9%) patients in the control group experienced hyperglycemia in their critical hospital period (P = 0.0001). No differences in sex, BW, GA and CRIB score were found between ROP and control infants as shown in Table 2. In our study, variables like transfusion, PaO2 ≥ 75 and blood sugar (BS) ≥ 150 continued to reach statistical significance after adjusting the other study covariates (Table 3). © 2014 Japan Pediatric Society
728
M Ahmadpour-Kacho et al.
Table 2 Demographic characteristics of ROP and non-ROP premature neonates at the time of enrolment ROP infants (n = 70) 29.91 ± 2.46 1238.57 ± 344.77 5.63 ± 2.45 34/36 (49%/51%) 48 (43%)
Study characteristics Gestational age (weeks) (mean ± SD) Birthweight (g) (mean ± SD) CRIB score (mean ± SD) Sex (male/female) Antenatal steroid received (%)
Control infants (n = 85) 30.59 ± 1.97 1327.53 ± 293.03 5.42 ± 2.24 46/39 (54%/46%) 64 (57%)
P 0.06 0.09 0.59 0.49 0.35
CRIB, Clinical Risk Index for Babies; ROP, retinopathy of prematurity.
Both PaO2 ≥ 75 and hyperglycemia had correlation with ROP. Sixty-nine patients (98.6%) in the retinopathy group had PaO2 ≥ 75 but 62 neonates (72.9%) in the control group had PaO2 ≥ 75 (P < 0.001). Although mothers of 112 patients had received antenatal steroid (betamethasone), no differences were detected between the two groups. The mean plasma glucose level and duration of being hyperglycemic were higher in the ROP group than in the control group (Table 4). The glucose infusion rate (GIR) in ROP infants was significantly higher than that in the non-ROP infants (5.8 vs 5.1 mg/kg/ min, respectively, P = 0.01). The median number of days to reach maximum serum glucose between case and control groups was 2.3 and 1.3 days, respectively (P = 0.04). Hyperglycemia was associated with an increase in ROP even after matching in terms of GA, BW and omitting the transfusion
Table 3 Bivariate analysis for factors associated with retinopathy of prematurity Study characteristics PaO2 ≥ 75 FiO2 ≥ 60% Blood sugar ≥150 Transfusion Mechanical ventilation
OR 25.59 1.27 14.27 2.60 1.10
95%CI 3.35–195.14 0.586–2.78 5.16–39.50 1.20–3.53 0.38–1.87
P 0.002 0.537 0.000 0.008 0.687
Table 4 Glycemic characteristics of ROP and non-ROP premature neonates (mean ± SD) Study characteristics Mean total serum glucose (mg/dL) Mean first serum glucose (mg/dL) Min & max of serum glucose (mg/dL) Glucose infusion rate (mg/kg/min) Duration of hyperglycemia (days) Insulin use
ROP infants (n = 70) 147.53 ± 71.14
Control infants (n = 85) 93.80 ± 37.37
0.0001
141.14 ± 114.60
88.16 ± 72.31
0.001
15–351
0.001
20–560
P
5.80 ± 2.15
5.10 ± 1.08
0.01
2.32 ± 2.37
1.33 ± 0.87
0.04
2 (100%)
0 (0%)
0.2
ROP, retinopathy of prematurity.
© 2014 Japan Pediatric Society
effect (Table 5). Therefore, only covariates of primary conceptual interest were used in the multivariate analysis. Severity of ROP did not reach statistical significance between the hyperglycemic and normoglycemic infants (P = 0.1). Among babies with ROP, 18 (54.5%) patients in the hyperglycemic group had prethreshold and threshold ROP, but 25 (67.6%) patients in the normoglycemic group developed prethreshold and threshold ROP (P = 0.264).
Discussion This study has demonstrated that hyperglycemia is a significant risk factor for retinopathy of prematurity. The elicited cut-off point for hyperglycemia in studies by Ertl et al.,17 Garg et al.5 and Blanco et al.1 was 150 mg/dL, which is in line with this study. Those authors also concluded that hyperglycemia might play an important role as a risk factor for ROP. In our study population, the overall incidence of hyperglycemia was 24.5%, which was close to the 19.4% in the study by Ertl et al.17 The investigation by Garg et al. showed that glucose infusion rate and duration of hyperglycemia like our study were risk factors eligible for ROP.5 Mean serum glucose level and duration of hyperglycemia play effective roles in the occurrence of ROP, as reported by Vendal et al. and our study supports this.18 In another study by Mohamed et al., also evaluating hyperglycemia as a risk factor for ROP in premature infants less than 32 weeks gestation, it was shown that neonatal hyperglycemia was significantly associated with an increased risk of the development of ROP in premature infants.19 The majority of previous studies were performed on several risk factors of ROP other than hyperglycemia. Postnatal serum insulin like growth factor-1 (IGF-1) level in addition to weight development may correctly identify neonates at risk for severe ROP.20
Table 5 Multivariate analysis for factors associated with retinopathy of prematurity Study characteristics Gestational age (weeks) Birthweight (g) Transfusion (times) Mean blood sugar (mg/dL) Glucose infusion rate (mg/kg/min) Duration of hyperglycemia (days) PO2 (torr)
OR 0.95 0.99 1.99 1.03 1.36 1.07 1.01
95%CI 0.64–1.4 0.99–1.0 0.70–5.60 1.01–1.05 0.92–2.02 0.54–2.12 0.99–1.02
P 0.801 0.566 0.191 0.001 0.331 0.844 0.275
Hyperglycemia and ROP Beardsall et al. suggested a significant correlation between ROP and hyperglycemia after adjusting for birthweight, gestational age and CRIB score, which was comparable with our results.21 The adjustment of birthweight, gestational age and steroid use was done by Blanco et al.1 and hyperglycemia was associated with a statistically significant increase in ROP, which was compatible with the findings in our study. Intraventricular hemorrhage had no relation to hyperglycemia in either of these studies.1 The duration of oxygen and ventilator therapies and IVH was not associated with ROP according to the reports of Kim et al., although hyperglycemia was detected as a significant risk factor.3 This result has been confirmed by our study. As the numbers of the bronchopulmonary dysplasia patients in our NICU graduates are very low (about 1% of VLBW infants) we did not decide to include the days on oxygen use as a variable. The reason for low incidence of bronchopulmonary dysplasia patients in our NICU graduates may be explained either by the higher gestational age of the NICU population, ethnicity or geographical parameters.22 Instead, we tested for FiO2 and PaO2. PaO2 ≥ 70 played an important role but a FiO2 ≥ 60% did not get a significant correlation with ROP. In our study, hyperglycemia was not more important than PaO2 ≥ 75 based on the odds ratio for PaO2 ≥ 75 and BS ≥150 (Table 3) but it is the second most important factor for the occurrence of ROP. Hyperoxemia during resuscitation after birth and during the mechanical ventilation in premature infants born at ≤32 weeks of gestation is the most important risk factor for ROP. For prevention, it is recommended to keep the oxygen saturation between the 85% and 93%. In this range of SpO2 in about 90% of patients, the PaO2 would be in the desirable range (40– 80 mmHg).23 In three recent international randomized, controlled trials, a group of researchers from the UK, Australia, and New Zealand evaluated the effects of targeting an oxygen saturation of 85–89%, as compared with a range of 91–95%, on disability-free survival at 2 years in infants born before 28 weeks’ gestation. These studies showed that those infants in the lower-target group for oxygen saturation had a reduced rate of ROP and an increased rate of necrotizing enterocolitis and death. Therefore, targeting an oxygen saturation of less than 90% among such infants should be avoided.24 GA and BW are the major determinant risk factors for ROP, but contrary to the report by Ertl et al., gestational age and CRIB score did not affect ROP in our study.17 Sutija et al. suggested that ROP developed more frequently in lower GA and BW than in the control and they also had more IVH than control groups. However, after adjusting BW and GA, we, like others, have not found any association between ROP and IVH.25 Although GA and BW were lower in ROP infants in the study by Vendal et al., longer duration of ventilation and higher level of blood sugar were reported by those authors, which was not detected in our study.18 The patients in the study by Garg et al. needed more insulin for normalizing blood sugar than the control group, which was not detected in our cases. A study by Yoo et al. on 7-day-old
729
rat pups concluded that “insulin treatment substantially increased VEGF levels, extraretinal vessel formation, matrix metalloproteinase activity, and the extent of retinal hemorrhage in rat pups with mild oxygen-induced retinopathy compared with saline controls.”26 Also, in a review by Kaempf et al., the authors concluded that “hyperglycemia and especially insulin use in premature infants may increase the risk of ROP.” They also found that “slower NICU growth velocity, but not rate of head or length growth, was predictive of ROP.”27 However, insulin infusion for hyperglycemia in premature infants was not associated with an increased risk on ROP in a study by Heald et al.28 Investigations by Blanco et al. showed a higher incidence (88%) of hyperglycemia in their patients than in ours (24.5%). It was partially because of their lower birthweight (
