Pregnancy Disorders Appear to Modify the Risk for Retinopathy of Prematurity Associated With Neonatal Hyperoxemia and Bacteremia

Abstract

Objective

To explore (1) whether extremely low gestational age newborns exposed to inflammation-associated pregnancy disorders differ in retinopathy of prematurity (ROP) risk from infants exposed to placenta dysfunction-associated disorders, and (2) whether ROP risk associated with postnatal hyperoxemia and bacteremia differs among infants exposed to these disorders.

Methods

Pregnancy disorders resulting in preterm birth include inflammation-associated: preterm labor, prelabor premature rupture of membranes (pPROM), cervical insufficiency, and abruption and placenta dysfunction-associated: preeclampsia and fetal indication. The risk of severe ROP associated with pregnancy disorders was evaluated by multivariable analyses in strata defined by potential effect modifiers, postnatal hyperoxemia and bacteremia.

Results

Compared to preterm labor, infants delivered after pPROM were at reduced risk of plus disease (Odds ratio = 0.4, 95% confidence interval: 0.2–0.8) and prethreshold/threshold ROP (0.5, 0.3–0.8). Infants delivered after abruption had reduced risk of zone I ROP (0.2, 0.1–0.8) and prethreshold/threshold ROP (0.3, 0.1–0.7). In stratified analyses, infants born after placenta dysfunction had higher risks of severe ROP associated with subsequent postnatal hyperoxemia and bacteremia than infants born after inflammation-associated pregnancy disorders.

Conclusion

Infants exposed to placenta dysfunction have an increased risk of severe ROP following postnatal hyperoxemia and bacteremia compared to infants exposed to inflammation-associated pregnancy disorders.

INTRODUCTION

Retinopathy of prematurity (ROP) is a leading cause of impaired vision and blindness in children born prematurely (1). The risk of severe ROP increases in premature infants, especially in those exposed to high levels of oxygen(2, 3), as well as inflammatory stimuli(4, 5), including those that occur prenatally(6, 7). Antenatal exposures to the developing fetus depend on the intrauterine environment and can therefore be modified by associated maternal pregnancy disorders. Pregnancy disorders that lead to preterm birth can be classified into two broad groups based on the presence of intrauterine inflammation or evidence of aberrant placenta implantation(8). Given this, we question if processes associated with individual pregnancy disorders differ in their ability to influence the development of severe ROP and if these exposures modify their influences on the risk of severe ROP (9, 10)?

Given the complex relationships between pregnancy disorders and ROP, we hypothesized that pregnancy disorders involving inflammation are associated with (1) an increased risk of ROP, and/or (2) modify the ROP risk associated with exposure to postnatal inflammatory events, such as hyperoxemia and bacteremia.

METHODS

The ELGAN (Extremely Low Gestational Age Newborns) study was designed to identify characteristics and exposures that increase the risk of structural and functional neurologic disorders in ELGANs(11, 12). During the years 2002–2004, women who were delivered before 28 weeks gestation at one of 14 participating institutions in 11 U.S. cities were asked to enroll in the study.

Mothers were approached for consent either upon antenatal admission or shortly after delivery. A total of 1249 mothers consented. Infants with birth defects and/or aneuploidy were excluded, as well as infants who did not survive until completion of all eye exams.

Demographic and pregnancy variables

After delivery, a trained research nurse interviewed each mother in her native language following standardized procedures to limit possible bias. The clinical circumstances that led to each maternal admission and initiated each preterm delivery were defined using both data from the maternal interview and data abstracted from the medical record (8). When they conflicted, the mother’s report was accepted.

Pregnancy disorders that lead to preterm delivery have previously been classified into two broad groups(8). The first group, labeled inflammation-associated, included preterm labor, prelabor premature rupture of membranes (pPROM), placental abruption, and cervical insufficiency. The placentas from these pregnancy disorders showed evidence of inflammation and often harbored microbes. Clinically, preterm labor was defined as progressive cervical dilation with regular contractions and intact membranes. pPROM was defined as the presence of vaginal pooling with either documented nitrazine positive testing or ferning prior to regular uterine activity. For a diagnosis of cervical insufficiency, a woman had to present with cervical dilation of greater than two centimeters in the absence of membrane rupture and detected or perceived uterine activity. Placental abruption was defined as presentation with significant amount of vaginal bleeding and a clinical diagnosis of placental abruption.

The second group of pregnancy disorders, placenta dysfunction-associated disorders, included preeclampsia and fetal indication/intrauterine growth restriction. These placentas revealed a lack of inflammation but did have infarcts and increased syncitial knots(8). Clinically, preeclampsia was defined as new onset hypertension of sufficient severity to warrant delivery. Presentations under the category of fetal indication/IUGR included severe intrauterine growth restriction based on antepartum ultrasound examination, non-reassuring fetal testing, oligohydramnios, and Doppler abnormalities of umbilical cord blood flow.

Infant variables

The gestational age estimates were based on a hierarchy of the quality of available information. Most desirable were estimates based on the dates of embryo retrieval or intrauterine insemination or fetal ultrasound before the 14^th week of gestation. When these measures were not available, reliance was placed sequentially on a fetal ultrasound at 14 or more weeks, last menstrual period without fetal ultrasound confirmation, and gestational age recorded in the log of the neonatal intensive care unit.

The birth weight Z-score is the number of standard deviations the infant’s birth weight is above or below the mean weight of infants at the same gestational age in a standard data set. We used the Oxford England standard because it excluded fetal-growth-restricting pregnancy disorders(13). This was the only available standard that excluded “electively delivered preterm infants … from the calculation of new birth weight and head circumference centiles.”

Maternal aspirin consumption was defined as use of aspirin, prescribed or not, during pregnancy while maternal leukocytosis was defined as a white blood cell count >20,000 per mm³ within 48 hours before or after delivery. Hyperoxemia of the infant was defined as a PaO₂ in the highest quartile for gestational age and postnatal day on two of the first three postnatal days for the entire study sample. Bacteremia was defined as a positive blood culture at any time between postnatal days 8 and 28.

Eye examinations

Participating ophthalmologists participated in sessions to minimize observer variability. Subsequently, they completed a standardized data collection form for each examination.

In keeping with ROP screening guidelines,(14) the first ophthalmologic examination was within the 31^st to 33^rd post-menstrual week. Follow-up exams were as clinically indicated until normal vascularization occurred in zone III.

Definitions of terms were those accepted by the International Committee for Classification of ROP(15). Severe ROP was defined using four definitions: stage 3–5, plus disease (retinal vascular dilation and tortuosity), prethreshold/threshold, and any ROP in zone I. We elected to use multiple categories of severe ROP rather than just ROP stage 3 or greater because each of these categories of ROP is considered severe and has implications for follow up or immediate treatment. For example, treatment is typically indicated for plus disease in zones I or II, or stage 3 in zone I(16). Without including ROP outcomes for zone I disease or plus disease, one could miss important associations for these examples of severe ROP requiring treatment.

Data analysis

Our primary outcome was ROP deemed worthy of treatment, defined as type 1 ROP according to ET-ROP Study recommendations(3). We also examined four categories of severe ROP, including stages 3 to 5, plus disease (retinal vascular dilation and tortuosity), prethreshold/threshold ROP, and zone 1 ROP.

Within the entire ELGAN study sample, the frequency of each form of severe ROP varied with gestational age at delivery. In early sets of analyses, we adjusted for gestational age in two ways, by both week of gestation (23, 24, 25, 26, 27), and by groups of weeks (23–24, 25–26, 27). Similar qualitative results were obtained and we present the data adjusted for gestational age in groups of weeks.

To identify potential confounders, we compared the frequency of each category of ROP among infants in strata defined by the characteristics and exposures of their mothers, pregnancies, and deliveries (Tables 1 and 2). We also compared the frequency of each pregnancy disorder among infants in strata defined by the characteristics and exposures of their mothers, pregnancies, and deliveries (data not shown).

Table 1.

The percent of each group of newborns who had the pregnancy characteristic on the left who had each category of ROP^*. These are row percents.

Pregnancy Characteristics		Stage 3–5	Plus disease	Zone I	PreThresh/ Thresh	ET-ROP treatable§	Row N
Smoking during pregnancy	Yes	29	11	8	17	16	166
	No	29	11	7	14	13	1002
Vaginal bleedings after 12 weeks	Yes	26	9	5	11	10	341
	No	29	11	8	16	14	821
Fever	Yes	37	17	10	18	18	78
	No	29	11	7	14	13	1078
Vaginal/cervical infection	Yes	31	15	12	21	19	155
	No	28	10	7	13	12	1008
Urinary tract infection	Yes	31	12	6	14	14	187
	No	28	11	8	14	13	976
Highest WBC	>20K	25	11	10	16	14	241
	≤20K	28	11	7	15	13	936
Any medication	Yes	29	11	8	15	14	1013
	No	23	9	5	11	9	148
Aspirin	Yes	43	21	8	24	24	62
	No	28	10	7	14	12	1095
Antenatal steroid course	Complete	27	10	7	13	12	759
	Partial	34	12	9	18	15	313
	None	30	15	7	17	16	123

^*ROP categories are not mutually exclusive

^§Met ET-ROP criteria for retinal ablative therapy for type I ROP

Table 2.

The percent of each group of newborns who had the delivery characteristic on the left who had each category of ROP^*. These are row percents.

Delivery Characteristics		Stage 3–5	Plus disease	Zone I	PreThresh/ Thresh	ET-ROP treatable§	Row N
Pregnancy Disorder
Inflammation-Associated	PPROM	28	7	5	10	9	261
	Abruption	25	10	3	9	11	126
	Cerv Insuf’ncy	27	11	8	17	15	70
	PTL	28	13	9	18	15	537
Placental Dysfunction	Preeclampsia	37	12	8	15	15	154
	Fetal Indicat'n	34	16	8	22	20	50
Cesarean Delivery	Yes	29	10	7	13	12	795
	No	29	14	9	18	16	404

^*ROP categories are not mutually exclusive

^§Met ET-ROP criteria for retinal ablative therapy for type I ROP

We then created multivariable models to quantify the contribution of each pregnancy disorder to each category of ROP (Table 4). In these logistic regression models, we adjusted for potential confounders such as gestational age and birth weight z-score, maternal WBC > 20,000, and maternal aspirin use. Maternal leukocytosis is often linked to infection and was associated with zone 1 disease in our sample. Maternal consumption of aspirin was included because in our sample it was prescribed preferentially for women with preeclampsia (18) and was associated with stage 3–5, plus, and prethreshold/threshold disease.

Table 4.

Odds ratio (and 99% confidence interval) of each category of ROP listed at the top of each column associated with the risk factor identified on the left. In this logistic regression model adjustment is made for gestational age (GA), fetal growth restriction (BWZ < −2 and BWZ < −1), maternal leukocytosis, maternal aspirin use, hyperoxemia and bacteremia, and birth hospital. Of these adjustment variables, the odds ratios of only hyperoxemia, bacteremia, and their interaction term are shown here. Infants born after preterm labor are the referent group. Bold italic type indicates odds ratios significantly <1 (p<0.05) and bold type indicates odds ratios significantly >1 (p<0.05). N=1199.

	Odds Ratios (95% confidence intervals)
	Stage 3–5	Plus disease	Zone I	PreThresh/ Threshold	ET-ROP treatable§
Preterm Labor (referent)	1.0	1.0	1.0	1.0	1.0
pPROM	1.1 (0.7, 1.6)	0.4 (0.2, 0.8)	0.5 (0.2, 1.03)	0.5 (0.3, 0.8)	0.5 (0.3, 0.9)
Abruption	0.6 (0.3, 1.04)	0.5 (0.2, 1.1)	0.2 (0.1, 0.8)	0.3 (0.1, 0.7)	0.6 (0.3, 1.2)
Cervical Insufficiency	0.8 (0.4, 1.6)	0.5 (0.2, 1.3)	0.6 (0.2, 2.9)	0.6 (0.3, 1.4)	0.7 (0.3, 1.6)
Preeclampsia	1.2 (0.7, 2.01)	0.6 (0.3, 1.3)	0.8 (0.3, 2.0)	0.7 (0.3, 1.3)	0.8 (0.4, 1.5)
Fetal Indication	1.2 (0.5, 2.7)	0.9 (0.3, 2.6)	1.8 (0.5, 6.2)	1.1 (0.4, 2.9)	1.3 (0.5, 3.5)
Hyperoxemia and Bacteremia	1.7 (0.9, 3.3)	5.1 (2.4, 11)	3.4 (1.4, 8.1)	5.3 (2.6, 11)	6.2 (3.1, 12)
Hyperoxemia Only	1.4 (0.9, 2.2)	1.2 (0.6, 2.3)	1.8 (0.9, 3.6)	1.8 (1.00, 3.1)	1.5 (0.9, 2.8)
Bacteremia Only	1.1 (0.8, 1.7)	1.3 (0.8, 2.3)	10.8 (0.4, 1.6)	1.3 (0.9, 2.2)	1.4 (0.8, 2.3)

^§Met ET-ROP criteria for retinal ablative therapy for type I ROP

Because ROP has been associated with both blood gas abnormalities (19) and bacteremia (20), we also adjusted for postnatal hyperoxemia and bacteremia. Because differences in obstetric management and population characteristics between the individual institutions might influence our findings, we also included a “cluster” term for birth hospital in our regression analysis to adjust for possible non-independence. We included indicator variables for missing data. The referent group consisted of infants delivered because of preterm labor, the largest subgroup of pregnancy disorders. The contribution of each variable to each category of ROP is presented as a risk ratio with 95% confidence intervals.

In a final set of logistic regression models, we further evaluated the effects of postnatal hyperoxemia and bacteremia as effect modifiers on the risk of ROP separately in the subsample of infants delivered for maternal or fetal indications (identified as the placenta dysfunction group), and separately in the subsample of infants delivered for spontaneous indications (identified as the infection-associated disorders group) (Table 5).These models adjusted for gestational age, birth weight z-score, maternal WBC >20,000, maternal aspirin consumption, and birth hospital.

Table 5.

Odds ratio (and 99% confidence interval) of each category of ROP listed at the top of each column associated with the risk factor identified on the left evaluated separately in two strata defined by pregnancy disorder subtype. The logistic regression models adjust for gestational age, fetal growth restriction, maternal leukocytosis, maternal aspirin use, and birth hospital. Bold italic type indicates odds ratios significantly >1 (p<0.05). N=1199.

	Odds Ratios (95% confidence intervals)
	Stage 3–5	Plus disease	Zone I	PT/ Threshold	ET-ROP treatable§
Intrauterine Inflammation Associated Disorder
Hyperoxemia and bacteremia	1.8 (0.9, 3.8)	4.0 (1.7, 8.5)	3.3 (1.3, 8.3)	4.2 (1.9, 9.3)	4.6 (2.0, 10)
Hyperoxemia only	1.3 (0.8, 2.2)	0.9 (0.4, 1.9)	1.1 (0.5, 2.4)	1.2 (0.6, 2.3)	1.0 (0.5, 2.0)
Bacteremia only	1.2 (0.8, 1.8)	1.3 (0.7, 2.3)	0.9 (0.4, 1.8)	1.3 (0.7, 2.2)	1.4 (0.8, 2.4)
No hyperoxemia or bacteremia	1.0	1.0	1.0	1.0	1.0
Placental Dysfunction Disorders
Hyperoxemia and bacteremia	1.4 (0.4, 4.8)	17 (2.3, 130)	11 (1.6, 68)	37 (4.5, 299)	78 (7.4, 830)
Hyperoxemia only	1.7 (0.5, 5.1)	7.1 (1.1, 47)	24 (3.9, 154)	35 (5.2, 234)	43 (5.7, 339)
Bacteremia only	1.2 (0.5, 3.4)	4.1 (0.7, 23)	3.0 (0.5, 18)	3.7 (0.7, 20)	3.9 (0.6, 22)
No hyperoxemia or bacteremia	1.0	1.0	1.0	1.0	1.0

^§Met ET-ROP criteria for retinal ablative therapy for type I ROP

Model fits were evaluated using standard techniques(17).

RESULTS

The final sample for this study consisted of 1199 infants who were enrolled in the ELGAN study after parental consent, and who underwent standard, routine evaluation for ROP (Figure 1).

Figure 1.

Description of the sample of infants studied who survived to discharge and completed ROP screening per protocol

Pregnancy Characteristics (Table 1)

The majority of mothers did not report smoking, vaginal bleeding, or having fever during the pregnancy. A slightly higher than expected percentage of infants born to mothers with vaginal/cervical infection developed ROP in zone I (12%) and prethreshold/threshold disease (21%). The incidence of ROP was not increased in infants whose mothers had fever, UTI, antibiotic use, or a WBC ≥ 20,000. Infants born to women who took aspirin during pregnancy had a slightly higher percentage of stage 3–5 ROP (43%), PLUS disease (21%), and prethreshold/threshold (24%) ROP than those whose mother did not take aspirin. Other medications were not associated with increased incidence of ROP. A complete course of antenatal corticosteroids was associated with a very modestly lower incidence of each characteristic of ROP

Delivery Characteristics (Table 2)

The majority of infants were delivered by cesarean delivery with a short latency following membrane rupture. Infants delivered for a fetal indication had the highest percentage of prethreshold/threshold disease (22%) and those delivered following abruption or prelabor rupture of membranes had the lowest percentage of ROP in zone I (3 and 5% respectively).

Characteristics of the newborn (Table 3)

Table 3.

The percent of each group of newborns who had the characteristic on the left who had each category of ROP^*. These are row percents.

Infant Characteristics		Stage 3–5	Plus disease	Zone I	PreThresh/ Thresh	ET-ROP treatable§	Row N
Sex	Male	30	11	9	16	14	630
	Female	29	12	6	14	13	569
Gestational age (weeks)	23–24	54	25	16	31	28	248
	25–26	31	11	9	17	15	556
	27	11	3	1	3	3	395
Birth weight (g)	≤ 750	47	19	13	25	24	443
	751–1000	24	8	6	12	10	519
	> 1000	7	2	0.4	2	2	237
BW Z-score**	< −2	49	18	11	23	23	65
	≥ −2, < −1	39	17	8	21	20	155
	≥ −1	26	10	7	13	12	979
Lowest quartile PO2on 2 of first 3 postnatal days	Yes	28	15	7	18	17	216
	No	33	12	9	16	15	787
Bacteremia between days 8 to 28	Yes	37	17	10	22	20	304
	No	27	9	7	13	11	892

^*ROP categories are not mutually exclusive

^**Birth weight z-score based on Oxford England standard

^§Met ET-ROP criteria for retinal ablative therapy for type I ROP

Six-hundred-thirty male and 569 female infants were included in the final sample. Most infants were born at gestational age 25–26 weeks, and with a birth weight between 751–1000g. A total of 155 infants had a birth weight between one and two standard deviations below the expected mean (i.e, a Z-score ≥ −2, < −1), while 65 had a birth weight more than two standard deviations below the expected mean (i.e, a Z-score < −2). The incidence of severe ROP, regardless of how defined, increased as gestational age and birth weight decreased. Girls had an ROP incidence that was similar to that of boys.

Multivariable Analyses (Table 4)

We compared the ROP risk associated with each pregnancy disorder to that of preterm labor in multivariable analyses, adjusting for maternal aspirin consumption, gestational age, fetal growth restriction, leukocytosis, birth hospital, postnatal hyperoxemia and bacteremia. Abruption was associated with a decreased risk of any ROP in zone I (odds ratio = 0.2; 95% confidence interval; 0.1, 0.7) and prethreshold/threshold disease (OR: 0.3; 95% CI: 0.1, 0.8), while pPROM was associated with reduced risk of plus disease (OR: 0.4; 95% CI: 0.2, 0.8) and prethreshold/threshold disease (OR: 0.5; 95% CI: 0.3, 0.8).

Multivariable Analyses in Pregnancy disorder strata (Table 5)

Among infants exposed to in utero inflammation-associated pregnancy disorders, the ROP risks associated with either neonatal hyperoxemia or bacteremia alone tended towards the null, while the ROP risks associated with exposure to both were increased. Among infants antenatally exposed to placenta dysfunction-associated disorders, however, relative risk estimates for plus disease, zone 1 disease, and prethreshold/threshold ROP associated with hyperoxemia alone were increased (ORs >7), and those associated with both postnatal hyperoxemia and bacteremia together were prominently increased (ORs >11). Indeed, these risks associated with both postnatal hyperoxemia and bacteremia were much higher among ELGANS whose mother had a placental dysfunction-associated pregnancy disorder than among ELGANs whose mother had an intrauterine inflammation-associated pregnancy disorder.

DISCUSSION

Three of our findings deserve emphasis and comment. First, pPROM and abruption appear to be associated with a lower risk of severe ROP than preterm labor. Second, although hyperoxemia and bacteremia individually are not associated with increased risks of severe ROP, the combination is associated with severe ROP in the total sample, and especially in strata defined by whether the pregnancy disorder leading to delivery was characterized by inflammation or abnormal placenta implantation. Third, while maternal and fetal indications for preterm delivery were not associated with severe ROP in univariable analyses, they were prominently associated with increased risk when the ELGAN also had recurrent hyperoxemia.

ROP Pathogenesis and Disorders Associated with Intrauterine Inflammation

The current pathogenesis of ROP is thought to involve two sequential phases.(18, 19) The first (vaso-obliterative) phase can be triggered by hyperoxia in the immediate postnatal period. High oxygen suppresses vascular endothelial growth factor (VEGF), and results in the cessation of normal vessel growth and regression of existing vessels. The second phase occurs when areas of the developing retina become relatively hypoxic, which stimulates VEGF release and neovascularization, one of the hallmarks of severe ROP.(20) Given our analyses, it appears that the pregnancy disorders leading to preterm birth modify the risk of developing severe ROP when exposed to postnatal events.

Severe ROP has been associated with intrauterine inflammation (9, 21), and with increased cytokine levels in postnatal blood attributed to intrauterine inflammation(5). What we found appears to contradict this. Placental abruption and pPROM, pregnancy disorders that have been associated with intrauterine inflammation(8, 22–27) were associated with reduced risk of severe ROP in our sample. We are not sure why our findings differ from those of others. One explanation invokes our analytic strategy. We compared all pregnancy disorders that lead to preterm delivery to preterm labor. Thus, one interpretation of our findings is that preterm labor, clearly an inflammation-related disorder(28), is associated with an increased risk of severe ROP when compared to other inflammation-related pregnancy disorders.

Given our findings that pPROM and abruption are associated with a reduced risk of severe ROP, we speculate that the risk reduction seen in inflammation-associated pregnancy disorders might be due to preconditioning/tolerance effect, the reduced probability of damage that can follow an insult if a less-than-damaging level of a potentially-adverse exposure occurred earlier(32). In utero preconditioning by a sub-clinical infection that contributes to abruption or pPROM might decrease the postnatal hyperoxemia and intrauterine inflammation-associated risk for ROP.

Placental Dysfunction and Sensitization

We observed a significant interaction between pregnancy disorders and postnatal events in the placental dysfunction group. The ROP risks were most prominently increased in ELGANs born after exposure to a placenta dysfunction pregnancy disorder who postnatally were exposured to hyperoxemia.

During preeclampsia, abnormal placenta implantation is presumed to result in inadequate vascular remodeling of the uterine spiral arteries, which in turn results in restricted fetal blood flow and growth retardation.(29) Placentas of pregnancies complicated by preeclampsia and IUGR have infarcts and syncytial knots, which are viewed as indicators of placental insufficiency.(8) The resulting presumed placenta ischemia induces the release of many bioactive factors including vascular endothelial growth factor (VEGF), cytokines, reactive oxygen species (ROS), hypoxia-inducible factors, and matrix matalloproteases (MMPs).(29) Perhaps these circulating proteins then alter vascular development, thereby contributing to the onset and/or progression of ROP.

The increase in ROP risk seen with exposure to postnatal hyperoxemia, as well as in the “triple-hit” scenario of in-utero placental dysfunction, postnatal hyperoxemia, and postnatal bacteremia, might be explained in part by a sensitization effect that primes the system for a more intense response to subsequent infections or insults, which then leads to worsening ROP(30).

Prenatal Phase of ROP

The present study suggests a complex relationship between the intrauterine environment and the pathogenesis of severe ROP. In particular, prenatal phenomena associated with inflammation or placental dysfunction might influence retinal maturation and development of ROP. We raise the possibility that ROP pathogenesis can begin in utero when the developing retina is exposed to varying amounts of inflammation or processes associated with restricted blood flow. Subsequent (mainly postnatal) exposure to hyperoxemia and bacteremia, therefore might have different effects on ROP development and/or progression depending on the extent of preconditioning or sensitization initiated in utero.

ROP has to be sufficiently severe to warrant ablative therapy. The columns in Tables 1–5 for individual indicators of ROP severity provide some of that information, but not adequately. This prompted us to include a column for treatable disease. Enrollment for the ELGAN Study began in 2002 when the indications for treatment were based on the findings of the CRYO-ROP study(31). A year and one half later, and before enrollment was complete, the results of the ET-ROP Study(32) indicated that the criteria for ablative therapy should be expanded. Consequently, some ELGAN Study subjects were treated based on CRYO-ROP criteria, while others were treated based on ET-ROP criteria. The column in each of the five tables for disease worthy of treatment according to ET-ROP criteria provides the best information about clinically-significant ROP. This is most evident in Tables 4 and 5 for the combination of recurrent early hyperoxemia and bacteremia during the second through fourth weeks.

Limitations and Strengths

The limitations of our study are those common to observational studies. In addition, the sickest infants were more likely to be treated more aggressively than those who were not quite so sick, making our study prone to confounding by indication.(33, 34) We are also well aware of the relatively small sample size of the placental dysfunction group compared to those in the placental infectious group. Still, the significant relative decrease in severe ROP risk observed with inflammatory pregnancy disorders deserves further exploration.

Despite the limitations, our study has several strengths. First, we included a large number of infants, making it unlikely that we have missed important associations due to lack of statistical power or claimed associations that might have reflected the instability of small numbers. Second, we selected infants based on gestational age, not birth weight, in order to minimize confounding due to factors related to fetal growth restriction.(35) Third, we collected all of our data prospectively. Finally, examiners were not aware of the medical histories of the children they examined, thereby minimizing “diagnostic suspicion bias.”(36)

CONCLUSION

In summary, pregnancy disorders appear to convey information about processes that can modify the risk of severe ROP associated with neonatal hyperoxemia and bacteremia among extremely preterm infants. This might be evidence of a prenatal phase of ROP development. We encourage others to explore the relationships among pregnancy disorders, postnatal exposures, and the risks of neonatal conditions, including ROP.