Impact of bundle implementation on the incidence of peri/intraventricular hemorrhage among preterm infants: a pre-post interventional study

ABSTRACT BACKGROUND: Peri/intraventricular hemorrhage (PIVH) is a frequent cause of death and morbidity among preterm infants. Few studies have addressed the use of bundles for preventing PIVH. OBJECTIVE: To evaluate the efficacy of a bundle of interventions designed to decrease the incidence of intraventricular hemorrhage at hospital discharge among preterm infants. DESIGN AND SETTING: Pre-post interventional study with retrospective and prospective data collection performed before and after bundle implementation in the neonatal intensive care unit of a university hospital. METHODS: Infants with gestational age < 32 weeks without malformations, who survived > 6 days were included. The bundle consisted of the following actions during the first 72 hours of life: maintenance of head in neutral position with the body in supine position, minimal handling, including delay of lumbar puncture until after 72 hours and absence of respiratory therapy maneuvers. Cranial ultrasound was performed on days 3, 7 and 28, or later if needed. The effect of the bundle was analyzed through logistic regression and results were adjusted for confounding variables. RESULTS: 167 infants met the inclusion criteria; 146 (87%) were analyzed. Bundle implementation was associated with decreased chances of PIVH at hospital discharge (odds ratio 0.29; 95% confidence interval 0.13-0.65). Cerebrospinal fluid collection within the first 72 hours increased the odds of PIVH of any grade during the hospital stay fourfold, after adjustment for all variables included in the model. CONCLUSION: Implementation of a bundle of interventions to avoid intraventricular hemorrhage was effective for decreasing the incidence of all grades of PIVH in preterm infants.


INTRODUCTION
The survival of preterm infants reflects the quality of prenatal and labor and delivery care, as well as the infrastructure available for neonatal care. According to a report published by the World Health Organization in 2012, titled "Born Too Soon: The Global Action Report on Preterm Birth", 1 Brazil ranks 10 th highest in the number of premature live births and 16 th highest in the number of deaths associated with complications of prematurity. Adequate and effective care in the neonatal intensive care unit (NICU) might modify the short, medium and long-term prognoses 2 for infants who require long hospital stays, especially for those who are extremely preterm.
Peri/intraventricular hemorrhage (PIVH) is a frequent cause of death and morbidity among preterm infants and contributes to an adverse neurological prognosis. [3][4][5] Several studies have sought to identify risk factors associated with PIVH in general, and in grade III and IV lesions in particular, 6 in order to establish preventive strategies for this condition. 7 However, because of the complex and multifactorial etiopathogenesis of PIVH, adoption of preventive measures alone is not expected to have much impact on the incidence of this severe neonatal complication. 8 The outcomes from interventions aimed at improving the quality of hospital care indicate that the use of bundles, i.e. sets of simultaneously applied measures, reduces the incidence of infections, [9][10][11] and central catheter-related late neonatal sepsis in particular, along with reducing other conditions particular to preterm infants. 12,13 However, few studies have addressed the use of bundles for prevention of PIVH. 8,14,15 Carteaux et al. 14 implemented a set of practices to reduce the incidence of intracranial hemorrhage and periventricular leukomalacia among very low birth weight infants, although the impact of those measures was not reported. Schmid et al. 8 implemented a bundle of measures targeting delivery care and the initial care of neonates in the delivery room and neonatal intensive care unit (NICU) that emphasized minimal handling, and they reported that these measures reduced the incidence of PIVH by 50% among preterm infants with birth weight under 1,500 g. A nursing intervention bundle applied in two Dutch centers reduced the risk of severe PIVH, and it was also associated with a lower risk of any degree of IVH, cystic periventricular leukomalacia and/or mortality. 15

OBJECTIVE
Thus, the aim of the present study was to establish whether a bundle of clinical measures implemented during the first 72 hours of life among preterm infants with gestational ages less than 32 weeks could reduce the incidence of PIVH of any grade during the hospital stay.

METHODS
This was a pre-post interventional study with retrospective (pre) and prospective (post) data collection performed before and after implementation of a bundle of measures at the NICU of a university hospital. The study was approved by the institutional research ethics committee (protocol number: 226.656; approved on March 22, 2013). The study periods were as follows: 1. Data were retrospectively collected before bundle implementation, over a period covering from March 2009 to April 2011.
Thus, informed consent for this period was waived.
2. Data were prospectively collected after bundle implementation, from May 2011 to April 2013. For these data, informed consent was required.
All newborn infants with gestational ages less than 32 weeks, as established according to the best obstetrical estimate, without lethal congenital abnormalities or malformations of the central nervous system, were included. Infants who died and/or who underwent surgical procedures within the first 168 hours of life were excluded.
The presence of PIVH was investigated using head ultrasound (US) on days 3, 7 and 28 of life, and later as indicated by the medical staff. The scans were performed using the Acusson X300 US device (Siemens, Erlangen, Germany), with a multifrequency micro-convex 8-5 MHz transducer. The presence and severity of PIVH was assessed in accordance with the methods of Papile. 6 All US examinations were performed by specialized pediatric radiology staff, and the reports were reviewed by the supervising radiologist.
The frequency of PIVH of any grade and the frequency of grade III/IV lesions among newborn infants with gestational ages less than 32 weeks and without malformations, who survived for more than 12 hours, were 64% and 45%, respectively in the NICU of the present study in 2010. Assuming that the bundle of measures would reduce the incidence rates of PIVH of any grade and grade III/IV lesions to the average seen among 20 hospitals included in the Brazilian Network of Neonatal Research 16 (35% and 18%, respectively), with power of 90% and two-tailed alpha error of 5%, the required sample size for the periods before and after bundle implementation was 60 individuals.
The and environmental policies. 14,15 Demographic data on the mothers and newborn infants were Neonatal characteristics and variables associated with adherence to the bundle, stratified in relation to the presence of PIVH, were analyzed by means of logistic regression for the outcome of PIVH of any grade during the hospital stay. As a sensitivity analysis, PIVH of any grade at three days of age was also analyzed. Two models were fitted for each outcome, in which all variables that had a P-value < 0.20 on univariate analysis were included as independent variables.
In addition, model one also included a "period" variable indicating whether the infants received the bundle. In model two, the "period" variable was removed, and the following bundle-related variables were included: head position (adequate or inadequate); body position (adequate or inadequate); tracheal aspiration (present or absent); respiratory physical therapy maneuvers (present or absent); and CSF sample collection before 72 hours of age (present or absent). Non-significant variables were removed from the model one at a time, and the model goodness-of-fit was tested using the Hosmer-Lemeshow test in SPSS 21.0 0 (IBM SPSS Statistics for Windows, version 21.0; IBM Corporation, Armonk, NY, United States).

RESULTS
A total of 274 infants with gestational ages less than 32 weeks were born during the study period, and 221 (81%) met the inclusion criteria (Figure 1). Of these, 46 (21%) were excluded because they died within the first week of life, and eight (4.5%) were excluded because they underwent surgical procedures.
Among the remaining 167 eligible infants, data could not be obtained for 21 (12.6%). Consequently, 146 infants were included in the analysis: 61 (42%) during the pre-intervention period and 85 (58%) during the post-intervention period.
The mothers' and newborn infants' characteristics stratified according to intervention group are described in Table 1; no significant differences were found between the groups.
The incidence of PIVH of any grade at three days of age was 41% among the infants in the pre-intervention group and 29% in the post-intervention group (P = 0.146). These values were 5% and 3% (P = 0.281), respectively, for grade III/IV lesions on the third day of life. Comparing the incidence of PIVH during the infants' hospital stay, post-intervention reductions were found both for PIVH of any grade (66% versus 49%; P = 0.05) and for grade III/ IV lesions (16% versus 6%; P = 0.03).
The results from the logistic regression analysis performed to investigate risk factors associated with occurrence of PIVH of any grade during the hospital stay are described in Table 2 GA = gestational age. * The infants that did not meet the inclusion criteria in the pre-intervention period presented the following: central nervous system malformation (n = 7); or congenital malformations incompatible with life, consisting of multiple malformations (n = 4) or renal dysplasia (n = 1). ** The infants that did not meet the inclusion criteria in the post-intervention period presented the following: central nervous system malformation (n = 15); or congenital malformations incompatible with life, consisting of multiple malformations (n = 9), renal dysplasia (n = 7), pulmonary hypoplasia (n = 3), skeletal dysplasia (n = 2) or anencephaly (n = 5).

DISCUSSION
This observational study showed that after adjustment for con- This study, as well as the study conducted by Schmid et al., 8 focused predominantly on measures aimed at stabilizing infants during the period of cardiopulmonary transition, when infants are vulnerable to hemodynamic fluctuations that impact the germinal matrix. 23 Despite the positive result, it should be highlighted that the incidence of PIVH at the NICU studied here was much higher than the incidence of 20-25% reported in the United States for infants with birth weight < 1,500 g. 24 Measures aimed at improving the quality of hospital care may have greater impact on conditions of higher prevalence.
In designing this study, it was decided to exclude infants who died within the first 168 hours of life because death and PIVH are competing risks. The best strategy for analyzing competing risks is to use composite outcomes, but these are difficult to use and lead to errors of interpretation and sample size calculation. 25 Moreover, composite outcomes are generally inadequate, thus implying that the results apply to the individual components of the composite outcome rather than only to the overall composite. 26 Since nearly all occurrences of PIVH develop within the first week after birth, it was therefore decided, in this study, to exclude infants who died before the occurrence of PIVH. We also excluded infants who underwent surgical procedure during this same period, because the procedure itself and the anesthetic procedure might have biased our outcome.
With regard to the mechanisms underlying PIVH, turning the infant's head to one side might occlude or obstruct drainage of the ipsilateral jugular vein, which might consequently increase local venous congestion. 17 The results from the present study stress the relevance of this pathophysiological mechanism, given that to assess whether head midline position would be more effective than any other position for preventing or extending PIVH and they did not find any significant differences in the outcomes. 27 They included a total of 110 infants in their review, from two randomized controlled trials, and found that the difference in the risk of PIVH, comparing the supine midline head position with the supine rotated head position, was 0.03 (95% CI: -0.13 to 0.18). 27 The results from this systematic review did not provide a definitive answer to the review question. 27  increases. 20,29 These factors, together with the pain caused by the procedure, 30

CONCLUSIONS
These limitations notwithstanding, the present study demonstrated that a set of measures to prevent PIVH in preterm infants with gestational age less than 32 weeks was effective in decreasing the incidence of all grades of intraventricular hemorrhage.
Implementation of these measures mainly requires education for the NICU staff regarding the need for care practices that minimize fluctuations in cerebral blood flow. These measures are lowcost and easy to implement, although they are difficult to maintain in the long run because consistent implementation requires staff to be continuously reminded of their value.