Comparing light and noise levels before and after a NICU change of design

To compare light and sound levels before and after a change of design and evaluate these levels considering recommended NICU standards. A pre-test/post-test design. Light and sound levels were compared between the former open ward (OW) NICU of 34 beds and the current 40-bed unit composed of both pods and single-family rooms (SFR). Light levels were significantly higher in the pod/SFR unit for all levels of care, days of the week and time of the day. These findings could be attributed to the number and configuration of windows in the new pod/SFR unit allowing for more daylight entry compared to the OW. Sound levels were significantly lower in the current NICU (pod/SFR) compared to the former OW. Following the change of design, the pod/SFR unit are less noisy than the OW, although light levels are higher indicating the necessity to measure light levels.


Introduction
Appropriate light and sound levels in the Neonatal Intensive Care Unit [NICU] are essential to promote preterm infants' growth and development [1][2][3][4] and a fundamental consideration in any plan to change the design of a unit [5].
A small change in the NICU light environment creates physiological instability in preterm infants [6] and disrupts their sleep if light protection is insufficient [7]. Similar adverse effects, such as physiological and motor instability [8,9] and sleep disruption [10] have been reported in preterm infants exposed to NICU sound. These environmental factors could also prevent parents from staying close to their infants [11], while sound restrict their presence [12], and interferes with their sleep in the NICU [13]. An optimal NICU environment includes developmentally appropriate lighting for preterm infants, is exempt of excessive sound [14] and encourages family involvement [15]. In an evidence-based NICU design, light levels should respect diurnal variations and should be adjusted to perform care and procedures, while a combination of environmental control for sound is essential [16]. NICU standards recommend that ambient lighting in the infant care area should be adjustable from ten to 600 lux [17]. White et al. [17] also advise that sound levels should not exceed an hourly average of 45 decibels (dBA) and should never exceed 65 dBA.
In the past few years, many NICUs world-wide have modified their unit configuration from an open ward [OW] design to single-family rooms [SFRs] as the latter has been the recommended NICU design [18][19][20]. SFRs enable familycentered care [21] and allow for better control of light and sound levels and promote families' involvement [22] Still, a pod design which may consist of a cluster of four to six infants cared for in one space has been recently proposed to optimize infants' development and interaction with caregivers and avoid the isolation of infants' that may occur in SFRs if families are not sufficiently present [23]. In the context of a change in design from an OW to 6-bed pods and SFRs at one hospital, the aim of our study was to compare the light and sound levels before and after this NICU change of design and evaluate if these levels respect the recommended NICU standards. We hypothesized that: (a) light levels (in lux) would be more appropriate in the pod/ SFR unit compared to the former OW; (b) sound levels (in decibels) would be lower in the pod/SFR unit compared to OW; and (c) the percentage of time light and sound levels would meet the recommended levels would be higher in the pod/SFR unit.

Methods
The study was a pre-test/post-test design and was conducted in a NICU in Montreal, Canada. Ethics approval was obtained from the institutional review board at the study site (Federal Assurance number 0796). In January 2016, the 34-bed OW design NICU moved to a newly constructed unit with a 40-bed combination design consisting of three pods of six beds for intensive level III care, two pods of six beds for intermediate level II care, and 10 SFRs for level I care prior to discharge of the infant (including two isolation rooms) (see Fig. 1 for the pod/SFR unit design). The former OW had all three care levels in one large open space of 400m 2 : critical (14 beds), semi-critical (12 beds), and stepdown (8 beds). In the pod/SFR unit of 1145-m 2 , infants are admitted to an intensive or intermediate pod and then moved to a SFR when their health status is stable. The nurse-patient ratio remained unchanged between the former OW and the pod/SFR; 1:2 in acute care, 1:3 in intermediate care, and 1:4 in step-down/SFR. In the former unit, nurses sat at a central station at the entry of the NICU and in the new unit nurses sit at stations designed for this use in the pods and at a central station for the SFR area (see Fig. 1). Creating optimal environmental conditions for newborn infants was an important goal in the design of the new NICU, so the Recommended Standards for Newborn ICU Design [17] in addition to the guidelines for design and construction of health care facilities [24] were consulted and recommendations followed. In the pod/SFR unit, indirect ceiling neon and procedural lights were installed beside each infant for care or emergency purposes. In addition, outside windows were triple glazed and equipped with light filtering blinds with double roller shades (one being a blackout shade). For sound, floors of the pods and SFRs were covered in sound absorbent tiles.

Light and sound measurements
Light levels were measured in lux with the Omega ® HB3336-03 light meter in a horizontal plane and sound levels were captured with a dBA-weighted scale sound meter (Sound Examiner SE-402), which measures environmental sound as heard by the human ear [25]. The sound measurements were obtained with one decimal in equivalent level (Leq), which is the appropriate measure to obtain an average sound level. Light and sound levels were recorded over 24 h and for an entire week (7 days). Light levels were recorded every minute and sound levels every second, but for sound a mean was computed every minute for analysis. After measurements were obtained, recordings were downloaded into an Excel spreadsheet for analysis. Measurements in the former and current NICUs were collected over the same months of the year and same time of the day to control for daylight associated with seasonal variations as well as sound associated with nursing workload.

Former NICU
Light and sound levels in the OW were assessed during the summer in 2014, 6 months prior to the NICU's planned move to the pod/SFR unit configuration. In the OW, light and sound meters were placed in four locations representing the three levels of care: critical, semi-critical, and step-down. These locations were selected as they replicated locations used for a previous evaluation of sound levels conducted by the biomedical department of the hospital.

Current NICU
In the pod/SFR unit, light and sound levels were measured during the summer of 2016, 6 months after the move, which had been delayed from January 2015 to January 2016. Measurements were taken at five different locations representing the same levels of care as in the former unit: critical, semi-critical, and step-down. For critical care, measurements were taken in two different pods composed of six beds, one pod with windows adjacent to an atrium and the other pod with windows exposed to the exterior of the building. For semi-critical care, measurements were obtained in one 6-bed pod, while for the step-down area; measurements were taken in two SFRs-one with a window and the other without.

Analysis
Analysis were conducted with Stata (version 16). Independent Student t-test were conducted to compare light and sound levels between the former and current NICUs. A test of variance was conducted to compare variability of measures between the former and current NICUs. The percentage of time light and sound levels exceeded the recommended levels was calculated using the two-tailed proportion z-test. These analyses were also performed using nonparametric statistics, and since all findings remained significant, results of parametric statistics are reported. Significance was set at p < 0.05 two-sided for all tests.

Findings
A total of 93,620 light readings were collected over both phases of this study. Of these, two light readings from the step-down unit were above 20,000 lux and thus considered outliers and removed from the analysis dataset. In the OW, 43,412 (46.37%) readings of light were captured, and in the pod/SFR unit 50,208 (53.63%). Overall, light readings ranged from 0 to 4030 lux, with a mean of 189.90 and standard deviation of 359.79 lux. For sound, a total of 5,549,651 measurements were captured over both phases of this study. Of these, 114,896 readings were removed from the dataset as there was missing data (no data recorded), leaving 5,434,755 measurements. For the purpose of analysis 4,792,810 measurements were used: 1,953,612 measurements (40.76%) taken in the OW unit, and 2,839,198 in the pod/SFR unit (59.24%). Overall, sound levels ranged from 37.4 to 97.3 dBA with a mean of 53.10 dBA, and standard deviation of 6.80 dBA. Means and standard deviations of light and sound levels over 24 h for every day of the week in the OW and current pod/SFR unit are shown in Table 1. Means and standard deviations of light and sound per 24 h periods for the three time periods corresponding to the timing of shifts for nursing staff: day (7h30 to 15h30), evening (15h30 to 23h30), and night shifts (23h30 to 7h30) in the OW and current pod/SFR are shown in Table 2

Light variability
To better understand the variability that appeared to be notable in our study, an F-test for equality of variances was performed to test the hypothesis that the variance of the two units differed. In all three care levels of the new pod/SFR unit, that is in critical care, F(20,157, 12,822) = 138.12, p ≤ 0.001, semi-critical care F(10,006, 9737) = 106.35, p ≤ 0.001 and step-down care, F(20,037, 10,526) = 36.86, p ≤ 0.001, there was significantly more variability in light levels compared to the former OW (see Fig. 2).

Light levels versus recommendations
In the respective critical care areas, the current pod/SFR unit had significantly more (11.3%) light readings that surpassed the ambient light recommendation of 600 lux compared to the OW where it never exceeded 600 lux (z = 39.36, p ≤ 0.001). The pod/SFR unit's semi-critical care pod also had significantly more readings over 600 lux (27.7%) than the OW's semi-critical area that had none (z = 56.01, p ≤ 0.001), a trend that continued in the SFRs (step down area), where the proportion of light readings over 600 lux was significantly higher (9.5%) than those of the OW which were 0% (z = 32.57, p ≤ 0.001) (See Supplementary file, Table 1).

Pods versus SFRs
Comparing light data recorded in the pod/SFR in the current NICU, we noticed differences in the estimated means. However, as the two mean distributions overlapped, we could conclude that there was no significant difference between the mean levels observed in the pods and those from the SFR.

OW versus Pod/SFRs
Irrespective of the day of the week, sound levels were always lower in the pod/SFR unit compared to the OW (see Fig. 3). Independent samples t-tests confirmed that this difference was significant overall, with lower levels in the

Pods versus SFRs
For the comparisons of the mean levels obtained in the pods and SFRs in the current unit, even if there are differences in the estimated means, we could conclude that there was no significant difference between the means as the two distributions overlapped.

Sound levels versus recommendations
Maximum recommended sound levels of 45 dBA were met 100% of the time in the OW for all levels of care (critical, semi-critical, and step-down areas); however, in the pod/ SFR unit, minimum levels were only met 76.7%, 79.5% and 75.8% of the time for critical and semi-critical pods as well as the SFRs, respectively (Supplementary files, Table 2). With respect to sound surpassing the recommended upper level of 65 dBA, the former OW had significantly more readings above 65 dBA than the critical care pods (7.1% OW versus 0.7% pods), the semi-critical care pods (11.1% OW versus 0.5% pods) and the SFRs (6.6% OW versus 1.5% SFRs). All differences were statistically significant (p ≤ 0.001).

Light levels
Our study findings do not support the hypothesis that light levels would be more appropriate in the pod/SFR unit compared to the former OW nor that the percentage of time light levels would meet the recommended levels would be higher in the current NICU. Light levels were significantly   higher in the pod/SFR unit for all levels of care, days of the week and time of the day. However, these findings should be interpreted with caution as they more likely reflect architectural design decisions and not that light levels are necessarily higher in pods/SFR. These findings could be attributed to the number and configuration of windows in the new pod/SFR unit allowing for more daylight entry compared to the OW. In fact, had the SFR unit rooms with north-facing windows or a suitable window overhang, excessive light entering the rooms could have been avoided.
In the former OW, all the windows were located on only one side of the large open space on the only exterior wall therefore limiting daylight entry into many areas of the OW, especially the intermediate and step-down areas that were further from the windows. In contrast, all pods and five of ten SFRs in the new unit have windows either exposed onto an outdoor atrium or a fully exposed exterior wall allowing for more daylight entry (see Fig. 1). The effect of windows on light levels in the current unit is particularly noticeable when comparing the readings taken in a SFR with and without windows. In the SFR without windows (SFR B), light levels never go beyond 10 lux (Tables 3 and 4, Supplementary files). Our findings are consistent with one study where the mean light levels were found to be higher in a SFR unit compared to an OW due to the higher number of windows in the new SFR unit [25]. Windows guarantee the entry of daylight, which is the optimal lighting for care procedures and to observe infants' skin color [16,17]. Windows may increase light levels in the NICU but are also recommended so that both professionals and parents have access to daylight for psychological benefits [17]. Nonetheless, control of light levels in important for the wellbeing of the infants. Seasonal and diurnal variability is also a factor influencing light intensity in the NICU and infants' exposure to light [26]. In our study light readings from the former and current NICU were taken during the summer when sunny days tend to be more frequent and the days are longer. Our readings reflect the influence of natural light entering the pod/SFR unit as levels of light increased and decreased in a typical diurnal cycle in all three care areas, peaking near 3 pm daily (see Fig. 2). Independent of the day of the week, and without reaching statistically significant differences, light levels were always higher in the semi-critical pod (overall mean of 407 lux) compared to critical care (overall mean of 246 lux), a trend which was also observed for daytime and across weekdays with the highest mean level reaching 741 lux in the semi-critical pod during the day, a mean above the upper recommended ambient level of 600 lux (see Tables 1 and 2). The orientation of the windows to the sun may explain these findings as the entry of light is more direct in this critical care pod compared to the semicritical care pod. Direct sunlight entering the critical care pod may be more noticeable and prompt NICU nurses and other professionals to close windows blinds to lower both light and heat; whereas in the semi-critical care pod where the light is more indirect, windows blinds may not be closed as often or as much and thus contributing to the higher levels observed. Thus, NICU staff should be aware of the light exposure provided by windows in various areas of their unit in addition to abrupt changes in lighting [27], as small light variations in the NICU have been found to disrupt preterm infants' physiological stability and sleep [6,7]. It might be difficult for NICU staff to estimate when light levels are too high, thus regular light meter readings should be readily available for NICU professionals to facilitate appropriate control of the environment. For instance, sound-level system providing direct and visual color feedback to staff when a 50 decibels threshold is exceeded in the NICU was found to successfully reduce sound levels in patient care areas [28]. Also, a noise-sensor light alarm activating in the NICU when noise levels reached more than 65 dBA was reported to decrease noise levels inside incubators [29]. Similar technology could therefore be designed for a continuous reading of NICU light levels and automatic feedback provided to staff.
Due to weather conditions (cloudy, partly cloudy, or sunny), ambient light was occasionally above 600 lux in the current pod/SFR unit. In contrast, light in the OW never surpassed this recommended upper limit (Table 1, Supplementary file). Generally, in the current pod/SFR unit for the critical and semi-critical levels of care, readings were more frequently above 600 lux on sunny days. Still, in the SFR with a window, light levels were on average, frequently higher on partly cloudy days compared to sunny or cloudy days suggesting again that NICU staff or parents, may tend to close blinds more often on sunny days compared to partly cloudy days where daylight enters the unit but it may be less apparent that the level is high. For the SFR without a window, the recommended maximum level of 600 lux was always respected (Table 1, Supplementary file). As suggested, all NICU windows should be equipped with adequate shading [17] to avoid directly exposing infants to direct sunlight and allow for the control of light levels at any time of day.

Sound levels
For sound levels, our findings support the hypothesis that levels would be more appropriate in the pod/SFR unit compared to the former OW and that the percentage of time sound levels would meet the recommended levels would be higher in the current NICU. For every day of the week, every time of the day and for all levels of care, sound levels were significantly lower in the new pod/SFR unit compared to the OW. We also compared sound levels between pods and SFRs. A recent systematic review concluded that sound levels in SFRs or enclosed unit design are usually lower than in OW [19]. More precisely, in one study, the sound level was lower in the SFRs compared to an unoccupied OW (i.e., no patients, no staff) [20] and lower in pods of 6 beds compared to OW of 11 beds [30]. In contrast to our study where no significant difference was found between the 6-bed pods compared to our SFRs for sound levels, two studies reported significantly lower levels in SFRs compared to NICU design similar to pods. Sound levels were reported to be lower in SFR units compared to an 8-bed OW [25] and significantly more time at lower sound levels was observed in the SFRs compared to an OW of six to ten infants [31]. In our study, the critical and intermediate care pods had comparable mean levels of sound to the former NICU (i.e., critical care pods: 50 dBA; semi-critical care pods: 49 dBA; SFRs: 49 dBA). Others reported a significantly higher mean sound level in their critical versus semi-critical care areas of their NICU [32]. In the current NICU the similar levels recorded in the different room configurations (i.e., pods vs. SFRs) and every day of the week may be due to the use of construction materials (i.e., floor, wall materials) meeting the current NICU recommendations, which may contribute to lower sound levels even if there is more than one infant in a pod. The NICU acoustic environment also depends on factors such as sound containment with walls, sound absorption by flooring materials and doors with an acoustic seal that prevent intrusive sounds from entering [17]. It might be that both the unit design as well as the construction materials used in the new unit may be effective in addressing these issues leading to comparable sound levels in the various care areas of our pod/SFR unit.
Our findings are important since lower NICU sound levels are essential to promote preterm infants' growth and development. Studies report that preterm infants exposed to lower sound levels by wearing earmuffs have longer quiet sleep [33,34] in addition to improved physiological stability [34], which in turn could limits their energy expenditure and favor growth [35][36][37]. In addition to appropriate sound levels, it is equally important that preterm infants are exposed to developmentally appropriate auditory experiences during their NICU hospitalization for optimal brain development [38]. Preterm infants are more susceptible to poorer language development which may be attributed to being exposed to high sound levels, as well as low language exposure during their NICU hospitalization. Positive, appropriate auditory stimulation is essential as infants exposed to a higher number of adult words in the NICU were found to have better language and cognitive development at 7-and 18-months corrected age [39].
In the former OW NICU mean sound levels ranged from 54 to 61 dBA and were consistently (100%) above the recommended hourly level of 45 dBA [17], whereas mean levels in our current NICU ranged from 47 to 51 dBA and exceeded 45 dBA no more than 75% of the time. In addition, in the pod/SFR NICU sound levels almost never exceeded 65 dBA (less than 1.1%). Although our mean sound levels in the pod/SFR unit were not always lower than the recommended 45 dBA, we had a significant mean decrease of 7-10 decibels compared to the former OW. Nonetheless, in order to sustain appropriate sound levels in NICUs, there is a need to continue to educate NICU staff and provide reminders to maintain a quiet sound level [40]. Every NICU should have a program of sound control and be aware of the effects of sound on preterm infants.
A significant contribution of our study is both the comparison of light and sound levels between an OW and a pods/SFR NICU, in addition to pods versus SFRs. We found that light levels were higher in the current pod/SFR unit compared to the former OW, which may be explained by not only updated lighting but also a higher number of windows and their orientation to the sun. Accordingly, the higher mean light levels measured in the new pod/SFR unit may reflect architectural design decisions and should not be interpreted to indicate that OWs are more conducive to appropriate light levels. Measurement of the light levels in a vertical plane could have provided, from another angle, more precise information about the sun's contribution to the intensity of lighting in the pods. Sound levels were significantly lower in the current NICU compared to the former NICU and those levels were similar between pods and the SFRs. Although we collected data on confounding factors which may influence light and sound levels (i.e., numbers of: infants, staff, windows with blinds, ventilators, and phototherapy lamps in function), it was not possible to interpret meaningfully these comparisons between the OW and pods/SFR units. For example, we noted the total of infants hospitalized as well as staff nurses in both the former and current NICU, but those units were not occupied the same way in the pod/SFR unit (i.e., a maximum of six infants and two nurses in the pods; one infant and one nurse in SFR) precluding comparisons with OW.
Recommended light and sound levels should be respected in NICUs to promote infant growth and development as well as to encourage family involvement and optimize work conditions for staff. Future studies should evaluate pod compared to SFRs on preterm infants' health outcomes as well as parents' perceptions.
Senior Research Scholar Award from the Fonds de recherche du Québec-Santé (FRQS). The funding bodies had no role to play in the design of the study and collection, analysis, and interpretation of data nor in the writing of this manuscript.
Author contributions MA: supervised data collection and analysis and drafted the manuscript with judicious input from all authors (NF, SR, LC, PDD). NF: obtained funding, supervised data collection and analysis. SR: supervised data collection. NF, MA, SR, LC: conceived the study and interpreted the data. PDD: conducted the statistical analysis. All authors read and approved the final manuscript.