With a shift to non-invasive ventilation of larger preterm infants, clinicians most often now face the challenge of ventilating extremely low birthweight (ELBW) infants with limited information about the optimal settings that would best serve this vulnerable population. Despite increasing acceptance of volume-targeted ventilation, there are limited data to guide the clinician in selecting optimal settings in this unique group of infants. In addition, the impact of the proportionally larger fixed dead space of the flow sensor in these extremely small infants may require some degree of tidal volume (V_T) compensation. Selection of an optimal target V_T is critical in ensuring the success of all volume-targeted modes of mechanical ventilation.1 An excessive V_T setting will lead to inadvertent hyperventilation, increase dependence on mechanical ventilation, and could cause volutrauma. A target V_T that is too low may increase the work of breathing, energy expenditure and oxygen consumption, increase the risk of atelectasis, and potentially lead to more lung injury, rather than less.2

The Volume Guarantee (VG) option available on the Dräger Babylog 8000+ ventilator (Dräger, Lübeck, Germany) is the most widely used and most extensively studied type of volume-targeted ventilation. In VG, inspiratory pressure is regulated in response to changing lung compliance and patient effort using exhaled V_T measurements.3 Earlier studies of VG generally included relatively large infants, often in the recovery phase of the acute respiratory disease4^–7; normative data are lacking for the smallest preterm infants who now constitute an increasing proportion of ventilated infants and often remain ventilator-dependent for extended periods of time. There is no consensus as to the optimal means of weaning from mechanical ventilation with VG. Many clinicians believe that the V_T settings should be progressively lowered during weaning, but our clinical experience suggests otherwise.

On the basis of empirical observations of progressively increasing V_T needed to maintain adequate partial pressure of carbon dioxide (Pco₂) with advancing postnatal age, we hypothesised that changes in anatomical and alveolar dead space caused by evolution of the underlying disease lead to alterations in the appropriate V_T settings in ELBW infants with prolonged ventilator dependence. The objective of this study was to establish the evolution over time of V_T associated with normocapnia/mild permissive hypercapnia in ELBW infants ventilated with VG during the first 3 weeks of life.

METHODS

Study design

This was a retrospective analysis of data collected in the course of normal patient care. The study was approved by the Georgetown University Institutional Review Board, which granted a waiver of informed consent requirements.

Patients

All ELBW infants with a birth weight <800 g admitted to Georgetown University Hospital between January 2003 and August 2005 were eligible for the study if they received mechanical ventilation with VG during the first 3 weeks of life. Infants with chromosomal anomalies, pulmonary hypoplasia or other anomalies of the lungs/upper airway and those with documented sepsis were excluded.

Ventilators and ventilation strategies

During the study, the Dräger Babylog 8000+ was used exclusively in patients receiving conventional mechanical ventilation. VG was the standard mode of ventilation, combined with either assist control (AC) or pressure support ventilation (PSV), with the latter used far more often. The initial target V_T during the first few days of life was 4–6 ml/kg, the exact value being left to the judgement of the clinical team. Subsequent adjustments to target V_T settings were based on blood gas values and on clinical observation of the work of breathing, tachypnoea or other signs of distress. Standard practice was to set the peak inspiratory pressure 3–5 cm above the prevailing working pressure. The target Paco₂ was not dictated by a specific unit policy. The general consensus in our neonatal intensive care unit (NICU) is that, during the first days of life, we aim to keep Paco₂ in the range 35–50 mm Hg, as long as the pH remains >7.20. Subsequently, we allow increasing degrees of permissive hypercapnia with Paco₂ 45–65 mm Hg, as long as the pH remains >7.20.

Data acquisition and statistical analysis

Infants meeting inclusion criteria were identified from our clinical database. The patient’s medical record was then retrieved, and basic demographic data were extracted and entered into the study database. At the time of each blood gas measurement, the respiratory rate, fraction of inspired oxygen, target and measured V_T, ventilator settings, respiratory rate, arterial or capillary pH and Pco₂ were documented during the first 48 h of life, days 5–7 and 14–21. Minute ventilation was calculated by multiplying the recorded V_T and respiratory rate.

Descriptive statistics were used to define the patient population and baseline respiratory variables. The relationship between postnatal age and Pco₂, minute ventilation and V_T at normocapnia was evaluated by analysis of variance for repeated measures. Because the number of paired V_T and blood gas observations for individual patients varied, the mean V_T/kg and Pco₂ for each infant for each of the study periods was calculated, and these mean values were then used for the analysis of variance. For the purposes of this study, normocapnia was defined as Pco₂ of 35–55 mm Hg during the fist week and 40–65 mm Hg during the third week. Because our intent was to determine the V_T that is associated with normocapnia, V_T associated with Pco₂ values outside of these ranges were excluded from the final analysis. Birth weight was used to calculate V_T/kg on days 1–7, and actual weight was used for days 14–21. During the latter two periods, many infants no longer had arterial blood gas values available. Mean capillary and arterial Pco₂ values did not differ from each other and were thus combined for analysis. Statistical significance was accepted at p<0.05.

RESULTS

A total of 79 infants with birth weight <800 g were admitted to our NICU during the study period. Forty-nine did not meet the eligibility criteria: 21 were extubated before the third week of life, 12 were switched to other modes of ventilation (usually high-frequency ventilation), nine were transported to us at >2 days of age, and seven died during the first days of life. Four records were unavailable, leaving 26 eligible infants. Twenty-one percent (160/757) of CO₂ values were outside the defined “normocapnic” range during days 1–7 and 12% (67/541) during days 14–21, and were therefore excluded. A total of 828 paired normocapnic blood gas and V_T sets were analysed: days 1–2 = 251; days 5–7 = 185; days 14–21 = 392. Because the last period was longer than the others and contained the largest number of paired measurements, we subsequently further subdivided this period into days 14–17 (n = 216) and days 18–21 (n = 176). All infants had been intubated with a size 2.5 endotracheal tube at birth, making it very unlikely that, given the babies’ size, any endotracheal tube leak could be present at least during the first week. Fifteen of the 26 infants (58%) were male. Their mean (SD) birth weight was 615 (104) g (range 400–790) and mean (SD) gestational age was 25.5 (1.4) weeks (range 24–29). Twelve infants were white, 10 were African–American, and four were of other racial origin. Table 1 shows the baseline ventilator variables during the first 48 h.

Consistent with the generally accepted practice of mild permissive hypercapnia, Pco₂ progressively increased from 44.0 (5.40) mm Hg on days 1–2 to 53.90 (6.2) mm Hg by the end of the third week (p<0.01). Measured V_T rose from 5.15 (0.62) ml/kg on days 1–2 to 6.07 (1.4) ml/kg on days 18–21 (p<0.01) (table 2). The minute ventilation rose in parallel with the V_T from 287 (46) ml/kg/min on days 1–2 to 337 (67) ml/kg/min by days 18–21 (p<0.01) (table 2).

The standard deviation of both Pco₂ and V_T progressively increased as well, indicating that there is substantial individual variability in the rate and degree of increase; in other words, the group of infants was becoming more heterogeneous. Furthermore, it can be seen in table 2 that there was an increasing difference between the set and measured V_T with advancing postnatal age and a modest increase in the difference between the pressure limit and actual working pressures, indicating that some infants were spontaneously generating larger V_T than those set by the NICU staff. When the data were reanalysed without excluding the “out-of-range” values, the mean Pco₂ and V_T differed only slightly from the data presented in table 2 and the differences remained highly significant.

DISCUSSION

Volume-targeted ventilation is a promising modality of mechanical ventilation because of its documented ability to achieve more stable V_T, reduce the proportion of excessively large V_T values, and reduce the incidence of hypocapnia.5^–7 The potential to reduce the risk of chronic lung disease has been clearly demonstrated by Lista et al,8 who showed that proinflammatory cytokine concentrations in bronchoalveolar lavage could be reduced and duration of mechanical ventilation shortened with the use of VG ventilation. Although other types of volume-targeted ventilation have not been studied as extensively as VG, most modern ventilators have some form of volume-targeted mode available. The need for normative data to guide V_T settings thus potentially applies to all neonatal ventilators.

Limited information is available to guide the clinician’s choice of V_T setting in volume-targeted modes, especially in ELBW infants, who often remain ventilator-dependent for many weeks. Although there is general consensus that a V_T of 4–5 ml/kg (measured at the airway opening) is an appropriate starting point for VG, there has not been a systematic attempt to determine whether these values remain optimal as the underlying pulmonary condition evolves. Clinicians have relied on empirical adjustments in response to blood gas measurements, but these adjustments are not based on published data. Weaning infants from mechanical ventilation has always been an inexact science with no consensus regarding optimal strategies even when the standard pressure-limited modes are used. In general, weaning is accomplished by lowering inspiratory pressure and allowing the infant’s own effort to gradually take over the work of breathing. Implicit in the strategy is the understanding that, with improved lung compliance and spontaneous respiratory effort, the infant is able to maintain adequate V_T and minute ventilation as the inspiratory pressure is reduced. With volume-targeted ventilation, V_T becomes the primary control variable. Some clinicians mistakenly believe that, similar to lowering pressure with pressure-limited ventilation, target V_T should be progressively lowered in volume-targeted modes. However, that reasoning is not physiologically sound, because the normal physiological V_T does not decrease; rather, the ventilator pressure needed to achieve that V_T decreases as lung compliance improves and the infant generates more spontaneous inspiratory pressure.

Our data confirm this premise and indicate that the physiological V_T needed to maintain mild permissive hypercapnia gradually increases over the first 3 weeks of life in ELBW infants. There are probably at least two mechanisms that contribute to the increasing V_T requirement. With positive pressure ventilation, the immature trachea and bronchi are exposed to cyclic stretch as many as 86 000 times/day, resulting in progressive dilatation of the trachea and bronchi (acquired tracheomegaly), which leads to increased anatomical dead space by as much as 90%.9 With evolution of chronic lung disease, the lung inflation becomes more heterogeneous and the lungs often become somewhat hyperinflated, resulting in increased alveolar (or physiological) dead space. There may also be a gradual increase in CO₂ production as the infant matures, generates more spontaneous activity, and receives increased enteral nutrition. The relative contribution of these factors is unknown, but, in combination, they appear to result in increasing V_T and minute ventilation requirement with advancing postnatal age. Because we only examined data for the first 3 weeks, we cannot comment on whether there is further increase in these values beyond the period of the study.

We noted increasing heterogeneity of the Paco₂ and V_T values with increasing postnatal age, along with some divergence between set and observed V_T and working pressure and peak inspiratory pressure limit. This observation indicates that many infants were spontaneously generating larger V_T than those set by the NICU staff. This progressive discrepancy between set and observed V_T may reflect attempts at reducing V_T by the clinical team in an effort to wean the infant from mechanical ventilation, and suggests that, in some instances, the set V_T was below the infant’s physiological need. This phenomenon is a common occurrence that we have noted in clinical practice in our NICU and elsewhere.

The effectiveness of volume-targeted ventilation largely depends on the selection of an appropriate target V_T. Empirically, a value of 4–5 ml/kg has become the standard target V_T, based on early clinical studies of VG.4^–7 However, these studies included much larger infants, and some used synchronised intermittent mandatory ventilation (SIMV) rather than AC or PSV. It has been shown that the V_T needed to achieve an equivalent minute ventilation is larger with low-rate SIMV than with AC, a mode in which every spontaneous breath is supported.10 Therefore, target V_T values need to be specific to the basic underlying mode. We have shown previously that VG was more effective when combined with AC than with SIMV at a rate of 30 breaths/min.11 The present data were generated in infants on VG combined with AC or PSV modes, and therefore these results may not be directly applicable to low-rate SIMV + VG. Eighty-nine percent of the paired values were obtained while the infant was on PSV and 11% on AC. Both modes support every spontaneous breath, with the only difference being that PSV is flow-cycled rather than time-cycled, and therefore we believe it appropriate to combine the findings. Because the number of values obtained on AC was quite small, meaningful statistical comparison between PSV and AC was not feasible.

Retrospective data from Dawson and Davies12 support the validity of initiating ventilation with a target V_T of ∼4 ml/kg (range 2.9–5.1) in larger preterm infants (mean weight 955 (306) g) ventilated with SIMV +VG with a relatively rapid set rate of 57.3 (9.8) breaths/min. They reported acceptable Pco₂ levels during the first 48 h of life (defined as 25 mm Hg>Paco₂<65 mm Hg) 96% of the time. We defined a narrower target range of 35–55 mm Hg, which explains the larger proportion of out-of-range values in our patients. There was no correlation between V_T and Paco₂, nor between calculated minute ventilation and Paco₂, possibly because they did not take into account the effect of instrumental dead space. In addition, because with SIMV there may be spontaneous breathing between mechanical breaths, the total minute ventilation may have been higher than the values calculated from the set respiratory rate and V_T.

The importance of choosing an appropriate V_T is illustrated by a pair of studies by Lista et al.8 These investigators initially showed the benefits of VG combined with AC when a V_T of 5 ml/kg was used. They then compared AC + VG for a V_T of 3 ml/kg with AC + VG for a V_T of 5 ml/kg in a similar group of preterm infants with respiratory distress syndrome. They found an increase in proinflammatory cytokines in the bronchoalveolar lavage fluid with the smaller V_T,13 probably as a result of atelectasis resulting from insufficient distending airway pressure.2 In addition, an inappropriately low V_T target has been shown to shift the work of breathing almost entirely to the infant.6 Although this was tolerated in the short-term study, prolonged periods of breathing through narrow endotracheal tubes with an inadequate target V_T are likely to increase the work of breathing, energy expenditure and oxygen consumption, thus interfering with adequate growth.

There are several limitations of this study. The V_T data were collected retrospectively from nursing and respiratory flow sheets and not downloaded in real time from the ventilator output port. Therefore, the recorded V_T is only a point estimate of the measured V_T at the time of the blood gas measurement. This is because, in actively breathing infants, the actual V_T fluctuates somewhat from breath to breath. However, because the notation obviously took place without knowledge that the data would be used for a study like this, there is no reason to believe that a systematic bias was introduced. The same is true for the respiratory rate and therefore the calculated minute ventilation. Secondly, it is not possible to be sure that, especially during the later periods when tracheal/laryngeal dilatation may have occurred, some leak around the endotracheal tube did not develop. However, because VG uses the exhaled V_T, the leak would cause the measured V_T to be underestimated, rather than overestimated, and therefore we do not believe that this would have been a major problem. The development of a significant leak would be expected to result in some degree of dead space washout as well, which in turn would probably reduce the Pco₂ levels and allow V_T settings to be reduced.14 Neither of these phenomena was in evidence, strongly suggesting that a leak around the endotracheal tube was not a material issue. Furthermore, because we are well aware that VG does not work reliably in the presence of a large leak, it is our practice to actively look for a leak around the endotracheal tube and reintubate with a larger size tube when such a leak is present. Finally, VG displays and uses exhaled V_T to regulate the working pressure. Because any leak around the tube is always less during expiration than inspiration, the exhaled V_T more accurately reflects the true V_T. Therefore, if some degree of leak was present in a few infants, the net effect, if any, would have been to underestimate the increase in V_T requirement over time that is demonstrated by our observations.

SUMMARY

We believe that this study provides useful information to guide clinicians tasked with managing volume-targeted ventilation of ELBW infants, who often require mechanical ventilation for prolonged periods. Although the data were generated using VG, the findings should be applicable to other forms of volume-targeted ventilation, with caveats regarding differences in specific ventilator performance and way of measuring V_T. Further studies are needed to prospectively evaluate different approaches to weaning from volume-targeted ventilation modes. Such studies should prospectively compare clearly defined alternative weaning strategies, assessing such variables as the work of breathing and weight gain, in addition to time to extubation.