Feeding Immaturity in Preterm Neonates

Risk Factors for Oropharyngeal Aspiration and Timing of Maturation

Abstract

Objective: 

Oropharyngeal aspiration (OPA) is a common cause of morbidity in premature neonates. We sought to identify significant risk factors and determine the timing of feeding maturation in preterm infants with OPA.

Methods: 

We performed a retrospective medical record review of 148 former preterm neonates referred for modified barium swallow study (MBSS) during a 3-year period. We looked at the MBSS results and potential clinical and demographic risk factors and performed bivariate and multivariable logistic regression analyses to evaluate predictors of failure. We evaluated the timing of resolution of OPA based on MBSS results.

Results: 

Of the 148 infants comprising our study cohort, 47 (32%) passed their initial MBSS and 101 (68%) were found to aspirate thin liquids. Infants who aspirate had younger corrected gestational age (CGA) and postnatal age at the time of testing. Increasing CGA led to lower odds of failing. Infants passed a MBSS at a median CGA of 53 weeks and those who failed an initial MBSS eventually passed after a median of ∼3.4 months from the first study. Infants with bronchopulmonary dysplasia had lower odds of failing their MBSS (odds ratio 0.16), but those taking inhaled corticosteroids and diuretics had higher odds of failing, and infants who were multiples also had higher odds of failing.

Conclusions: 

We determined the median CGA for passing MBSS and identified risk factors for OPA in this population. This information can help guide providers when counseling families on timing of feeding maturity and anticipation of appropriate timing for follow-up MBSS testing.

Oropharyngeal aspiration (OPA), the presence of liquids/foods below the level of the vocal folds, is highly prevalent in former premature infants ^(1,2). Up to 70% of very-low-birth-weight (VLBW; <1500 g) infants demonstrate swallowing abnormalities and up to 30% have impaired airway protection and are at risk for OPA ⁽¹⁾. In premature infants, increased risk of OPA has been attributed to neuromuscular immaturity and the delayed development and coordination of the suck-swallow-breath pattern ^(3,4). Most infants begin to coordinate their sucking-swallowing-breathing enough to begin oral feeding by approximately 34 weeks’ corrected gestational age (CGA), and maturation to successful feeding without OPA has been shown to drastically improve between 34 and 36 weeks and is frequently accomplished by 40 weeks’ CGA ^(4–6). For the subset of infants in whom maturation is delayed and OPA occurs, the timing of maturation has not been well described.

Aspiration into the lungs can lead to significant problems in this population, ranging from pneumonia and chronic cough to damage of the developing lungs and worsening of chronic lung disease ^(3,7–9); however, a majority of premature infants who aspirate do so silently, leading to underdiagnosis of OPA ⁽¹⁰⁾. Accurate assessment of aspiration is therefore critical to allow interventions to protect respiratory function. Presently, modified barium swallow studies (MBSSs) are the criterion standard for OPA diagnosis, but they carry the risk of radiation exposure. Factors commonly seen in former preterm infants, such as younger age and smaller size/height, have been significantly associated with increasing effective dose of radiation during MBSS ⁽¹¹⁾. It is therefore of the utmost importance that we attempt to minimize the number of studies performed. An understanding of the natural history of feeding immaturity in this population would better guide providers in the timing of MBSS to minimize this risk.

Previous work describing the epidemiology of OPA in infants has focused on term neonates or evaluated subjects with static genetic, anatomical, or neurologic comorbidities that affect feeding ^(7,12–14). Little information exists describing the timing of resolution of feeding immaturity in otherwise healthy former premature infants with known OPA.

We therefore performed this study to determine the prevalence of aspiration among premature infants referred for MBSS, describe the natural course of OPA and feeding maturation in premature infants, and identify clinical/demographic variables that may be predictive of aspiration in premature infants. We hypothesized that infants born more prematurely would be more likely to have evidence of aspiration.

METHODS

Study Population

This study was approved by the Boston Children's Hospital (BCH) institutional review board. We conducted a retrospective medical record review of all premature neonates born at <37 weeks’ gestational age (GA) referred for initial MBSS during a 3-year period between May 1, 2009 and April 30, 2012 at BCH in Boston, MA. Data were collected through December 31, 2012 for subsequent follow-up MBSS performance. Inclusion criteria included birth GA <37 weeks and referral for MBSS by inpatient/outpatient provider. Exclusion criteria included the following: intraventricular hemorrhage (IVH) greater than grade II, cyanotic heart disease, genetic syndromes, airway malformations, and hypoxic-ischemic encephalopathy. These criteria were chosen to isolate the role of immaturity in our population by excluding confounding risk factors for aspiration in an otherwise healthy former preterm population.

MBSS

MBSSs were performed by the BCH Radiology Department in conjunction with the BCH Feeding Team as per standardized clinical protocol. A “failed” MBSS was defined as demonstrating evidence of OPA, or the presence of liquids/foods below the level of the vocal cords with feeding.

Statistical Methods

We compared baseline demographic and clinical information between subjects who aspirated thin liquids (failed MBSS) and those who did not aspirate thin liquids (passed MBSS). We used Fisher exact testing, χ² testing, and nonparametric Wilcoxon rank sum testing as appropriate for binary, categorical, and continuous variables. We reported median and interquartile ranges (IQRs), given the number of outliers and non-normality of the data. Logistic regression models were used to model predictors of MBSS failure. Effects are reported as odds ratios (ORs) and 95% confidence intervals (CIs). We performed bivariate logistic regression analysis of the crude relation between the outcome of MBSS failure and each of the demographic predictors (birth weight, birth GA, sex, race, gestation number); clinical predictors (respiratory support requirements, patent ductus arteriosus, necrotizing enterocolitis [NEC], pneumothorax, nephrocalcinosis, IVH grade I or II, requiring tracheostomy, retinopathy of prematurity, requiring gastrostomy-tube [G-tube] placement, and diagnosis of bronchopulmonary dysplasia [BPD] defined as an oxygen requirement once corrected to 36 weeks’ CGA); medications (surfactant, postnatal steroids, inhaled corticosteroids [ICS] at time of testing, bronchodilator treatment at time of testing, diuretics at time of testing, antireflux medications at time of testing); and infant characteristics at the time of testing (CGA, postnatal age). The initial model was built upon those covariables with P ≤ 0.2 on bivariate analysis. Predictors were retained in the model if they were found to be significant at P < 0.05. We added back each covariable to assess for confounding. We assessed for collinearity in our model by evaluating the effect of each covariable on standard error and P values of the others. All of the statistical analyses were done using SAS 9.2 (SAS Institute, Cary, NC).

RESULTS

We identified 964 infants referred for MBSS between May 1, 2009 and April 30, 2012. Of these, 622 were born full-term and excluded, whereas 342 were born preterm. Of the preterm infants, 194 met exclusion criteria because of genetic, metabolic, or anatomic comorbidities; the remaining 148 infants comprised our study cohort. Of these, 47 (32%) passed their initial MBSS and 101 infants (68%) were found to aspirate thin liquids and failed their initial MBSS. The range of birth GA was 23^+3/7 to 36^+6/7 weeks, median of 31 weeks. Late preterm infants (birth GA 34–36^+6/7 weeks) made up 28% of the cohort (n = 42). Birth weight data were available on 127 subjects, and of these, 57% (n = 73) were VLBW and 35% (n = 45) were extremely-low-birth-weight (ELBW, <1000 g).

Those who passed their initial MBSS had older CGA and older postnatal age at the time of study, and were more often singletons than those who failed (Table 1). There were no differences in birth GA, birth weight, sex, race, respiratory support required, medications, and/or other comorbidities between those who passed and failed (Table 2). In this cohort, there was a trend toward increased BPD in those that passed, although this was not statistically significant (P = 0.12). Of the 42 late preterm infants referred, 57% (n = 24) failed initial MBSS.

TABLE 1: Unadjusted age and weight predictors based on initial MBSS result

TABLE 2: Unadjusted demographic and clinical predictors based on initial MBSS result

Of those who failed their initial MBSS (n = 101), 1 patient died before retesting, and 63 went on to pass a subsequent MBSS by the completion of data collection for this study. One death was unrelated to OPA. The remaining 37 had not passed a subsequent MBSS by the end of data collection. We therefore had 3 groups of infants: those who passed their initial MBSS (n = 47), those who failed initial MBSS but passed a subsequent test (n = 63), and those who had no documented MBSS pass (n = 37) by the end of the study's follow-up period.

For those who passed initial study, the median CGA at passing was 46^+3/7 weeks and the median postnatal age was 114 days (Table 1). For those who failed their initial MBSS but went on to eventually pass, the median CGA at passing was 56^+6/7 weeks (IQR 31.6 weeks) and the median postnatal age was 181 days (IQR 202 days). Median time from failure to first pass was 13.4 weeks, or 3.4 months (range 1–124 weeks) (Fig. 1). The mean and median number of MBSS performed from initial failed test to passage was 3, with a range of 2 to 8 tests.

FIGURE 1: Time-to-pass after initial failed modified barium swallow study (MBSS), weeks (n = 63).

A total of 110 infants passed an MBSS during the study period (47 passed initially and 63 failed initially and then passed a subsequent test). Figure 2 demonstrates the CGA at which these infants first passed their MBSS. The median CGA of first pass was 53 weeks (IQR 25 weeks) with the first quartile (25th percentile) passing at 44 weeks’ CGA and the third quartile (75th percentile) passing at 68.5 weeks’ CGA.

FIGURE 2: Corrected gestational age at the time of first modified barium swallow study (MBSS) pass (n = 110).

Of those who failed their initial MBSS, 37 had not passed by the end of data collection. These subjects had a median of 2 MBSS performed, with a range of 1 to 7. The mean number of weeks from their last MBSS to the end of data collection on December 31, 2012 was 65, or almost 1.3 years, with a minimum of 53 days and a maximum of 1153 days. Thirty-one of these subjects (84% of those who had not yet passed) had gone >6 months since their last MBSS, and 21 (57% of those who had not passed) had gone >1 year.

Predictors of OPA

We performed logistic regression to identify risk factors for OPA on initial MBSS (Table 3). The final multivariable model demonstrated that the significant risk factors associated with increased odds of OPA in this population included multiple gestation (twin/triplet), diuretic use at the time of testing, and ICS use at the time of testing. Increasing CGA at the time of testing decreased the odds of OPA. Interestingly, diagnosis with BPD decreased the odds of OPA. Adjusting for sex, race, birth GA, respiratory support, NEC, G-tube placement, postnatal steroid use, pneumothorax, and IVH grade I or II did not significantly affect the model, so all were excluded from the final model.

TABLE 3: Multivariable model of predictors of OPA

DISCUSSION

This study evaluated a large cohort of former premature infants without significant neurologic or anatomic comorbidities to assess the role of maturation in the resolution of OPA in this population. Our data demonstrated that being a multiple gestation infant and having younger CGA at the time of MBSS were significantly associated with OPA, but surprisingly, a diagnosis of BPD was not. We also showed that resolution of feeding immaturity in this population most often occurs within the first few months after neonatal intensive care unit discharge, but may take >1 year.

Recent studies have demonstrated a correlation between respiratory- and oral-based feeding immaturity and lower GA, unrelated to neurologic disorders and developmental delay ^(2,13). Maturation of sucking-and-swallowing and the ability to coordinate these functions with breathing have been shown to more accurately relate to birth GA rather than postnatal age ⁽¹⁵⁾. In our cohort, we found no difference in birth GA or birth weight between those who failed and those who passed their initial MBSS, nor were these significant predictors in our multivariable model. We believe the discrepancy between our findings and other published reports is that previous studies looked at a range of GAs including term infants, whereas our study looked exclusively at former preterm infants. When looking at overall risk of aspiration, it is highly likely that lower birth GA and preterm birth are important risk factors, but within our cohort of exclusively preterm infants, lower birth GA did not influence the odds of having OPA. Of note, late preterm infants were 28% of those referred for MBSS, and 57% of these were found to aspirate. This highlights the fact that late preterm infants are at risk for aspiration despite having fewer comorbidities than early preterm infants. Providers should be aware of this risk when caring for late preterm infants, whose level of respiratory and feeding immaturity is often underestimated.

In contrast to birth GA or birth weight, CGA and postnatal age at time of MBSS were significantly different in infants who passed initial MBSS compared with those who failed. Infants who passed an initial MBSS were more mature, with older CGA (median 46^+3/7 vs 41^+6/7 weeks, respectively) and older postnatal ages than those who failed (114 vs 92 days, respectively). Other studies have noted that the maturation and coordination of the sucking-swallowing-breathing pattern are more strongly correlated with increasing CGA than with postnatal age and feeding experience ⁽⁴⁾. Our findings seem to support the idea that feeding difficulties improve with older CGA. Specifically, our study found that for each additional day of CGA, the odds of aspiration on MBSS decreased by 1% (OR 0.992, 95% CI 0.986–0.997). In other words, for every additional week of CGA, the odds of failing the MBSS decrease by 7%.

We found no differences in race, sex, respiratory support requirement, medications, or other comorbid conditions such as patent ductus arteriosus, NEC, or IVH grades I/II between those who passed and those who failed an initial MBSS. Unexpectedly, we found that infants with a diagnosis of BPD had lower odds of failing an initial MBSS than those without (OR 0.16, 95% CI 0.046–0.569). This was an interesting finding given that infants with BPD are at risk for lower saturations during and after feeds ⁽¹⁶⁾. We did find increased odds of failure of MBSS in those infants who were taking ICS at the time of testing (OR 39.7, 95% CI 3.2–494) and who were taking diuretics at the time of testing (OR 5.34, 95% CI 1.1–26). These medications are commonly used to treat infants with BPD. All of the infants who were taking diuretics carried the diagnosis of BPD in our cohort. It is possible that there was selection bias in our population and that those infants with BPD who were poorly controlled with worsened distress may not have been tested or may have been tested at older CGA. Upon subgroup analysis, infants with BPD in this cohort were more likely to have a G-tube (31% vs 8% in those without BPD; P = 0.0014) and were chronologically older at the time of testing than those without BPD (median 105 vs 91 days; P = 0.0073); however, there was no difference in the CGA at the time of initial testing for infants with and without BPD. Lee at al ⁽¹⁾ previously evaluated swallowing dysfunction in VLBW infants and found that BPD was not a risk factor for impaired airway protection and hypothesized that aspiration during swallowing may not be a major cause of increased desaturations associated with oral feeding in infants with BPD. Given the increased odds of failing for infants on ICS and diuretics, we remain concerned about the increased risk of aspiration in infants with chronic lung disease. Our study highlights the fact that BPD diagnosis alone may not be adequate to reflect lung disease after neonatal intensive care unit discharge; requirements for respiratory medications may be an additional method to identify patients with respiratory fragility.

We noted that those who passed an initial MBSS were more likely to be singleton gestations, and in our final multivariable model, being a multiple increased the odds of OPA (OR 4.9, 95% CI 1.9–12.9). The etiology of this association is unclear, but it did not change when adjusting for the other covariates in the model or for numerous clinical factors. This is the first association we have noted in the literature between multiple gestation and increased risk of OPA. Multiple gestation infants are more likely to be born premature, be conceived by IVF, and be small for GA. There may be confounders that we were unable to identify based on our retrospective study, but this is an important association for future investigation.

For infants who failed their first MBSS, the median CGA at which they subsequently passed was 53 weeks, with the top 25th percentile passing at CGA of 44 weeks and the 75th percentile passing at 68.5 weeks’ CGA. More than half (52%) of infants passed by the time they were corrected to 54 weeks’ CGA. These data provide information for clinicians to use when giving anticipatory guidance to families regarding potential duration of feeding immaturity and OPA risk. These data may also guide timing for scheduling follow-up MBSS to limit the number of tests and subsequent radiation exposure. This provides evidence that close monitoring for feeding maturity is important not only in the first few days to weeks after discharge but also often for many months.

The strengths of our study include the relatively large sample size of premature infants without additional comorbidities and the length of follow-up. There were several potential limitations to this study. First, it is a retrospective review of infants referred for MBSS by their clinical care providers. There were likely many infants who were at risk for aspiration but were not referred for MBSS. It is also important to note that aspiration is not the only definition of feeding immaturity in neonates, nor is resolution of aspiration the only marker of feeding maturity. Passing an MBSS indicates safety for attempting thin liquid feeds, but does not necessarily indicate that an infant has attained complete feeding maturity; however, our study describes a large cohort of premature infants with OPA and therefore provides generalizable data for this population. The use of MBSS to detect duration of feeding immaturity is confounded by variables that affect the timing of the study, including availability of MBSS scheduling and care provider referral pattern. It is therefore difficult to pinpoint the exact timing of feeding maturity based on our data; however, our center is a large tertiary care facility with frequent availability for MBSS appointments. Timing of MBSS follow-up is guided by experienced feeding specialists, and is likely in a range that allows use for clinical guidance.

A subset of infants in our cohort did not have a documented MBSS pass by the end of our assessment period, so we do not have complete data on all of the infants referred. Of those who failed their initial MBSS, 37 had not passed by the end of data collection (25% of the study population). Thirty-one of these subjects had gone >6 months since their last MBSS by the end of our data collection and 21 had gone >1 year. This may be because of loss to follow-up as well as that the feeding team often avoided additional MBSS testing by working on feeds as clinically tolerated. A small number of infants, for example, did not “pass” an MBSS but continued to be followed by the feeding team, which recommended empiric adjustment to feeds (either thickening or weaning back to thin liquids) based on their clinical observation of a feeding session, presumably to avoid the risk associated with MBSS. Because MBSS is the criterion standard to assess OPA, we chose to limit analysis of time-to-resolution to those infants with a documented pass.

Finally, the sensitivity and specificity of a MBSS are not 100%. It is possible that because of limitations on duration of the test, related to the concern about radiation exposure, all of the infants who aspirate were not identified; however, the MBSS remains the accepted standard to use for identification of OPA.

CONCLUSIONS

We identified a large cohort of former preterm infants without significant anatomic or neurologic comorbidities who were referred for MBSS because of concern about aspiration. Infants who aspirate had younger CGA and postnatal age at the time of testing, and increasing CGA led to lower odds of failing. Infants passed MBSS at a median CGA of 53 weeks and those who failed an initial MBSS eventually passed after a median of ∼3.4 months from the first study. Infants with BPD had lower odds of failing their MBSS, but those taking ICS and diuretics had higher odds of failing, and infants who were multiples also had higher odds of failing. This information can help guide providers in counseling families on timing of feeding maturity and anticipation of appropriate timing for follow-up MBSS.

Acknowledgments

The authors thank the feeding team at BCH for their ongoing assistance. The authors also thank the Division of Newborn Medicine at BCH and the Scholars in Clinical Science Program for their support of Dr Davis.