Introduction

Mechanical ventilation is frequently administered in pediatric intensive care units (PICUs) to support children with respiratory failure as well as cardiac dysfunction or neurologic diseases. However, prolongation of mechanical ventilation increases the risk of serious complications, mainly barotrauma and ventilator-associated pneumonia. Therefore, it is important to shorten the duration of mechanical ventilation and to start weaning patients from mechanical ventilatory support as soon as possible. Farias et al. reported that up to 46% of the total time on mechanical ventilation was devoted to weaning [1]. In this regard, reducing the duration of weaning could be important to reduce the length of mechanical ventilation and thus the associated complications. The process of weaning is dependent on several factors: (1) the clinical status of the patient, (2) the availability of caregiver(s) who prescribe and apply the mechanical ventilatory settings, (3) the institutional policies that allow weaning within certain periods of the day and (4) the clinical criteria used by the attending intensivist to conduct weaning.

Given the limited human resources, protocols of mechanical ventilation have been developed to accelerate weaning. For instance, protocols on spontaneous breathing trials have been implemented daily or more frequently [2, 3, 4, 5]. Also, expert-systems integrated into a mechanical ventilator facilitate medical decision making and thus can react rapidly and potentially avoid delays in prescribing and applying mechanical ventilatory settings [6]. In both cases, the criteria that guide weaning depend on the actual medical knowledge provided by literature and physicians' clinical experience.

Indicators or surrogate markers commonly used in clinical practice to assess patient's progression and ability to breathe spontaneously are: adequacy of gas exchange [oxygen saturation (SaO2), arterial pH and PCO2], ventilator parameters [including fraction of inspired oxygen (FiO2), respiratory rate (RR), positive end-expiratory pressure (PEEP), VT], signs of increased work of breathing (W), hemodynamic stability, level of consciousness, discomfort and diaphoresis [7]. The quantifiable parameters that reflect ability of CO2 removal during weaning are RR, VT and the resulting PCO2. These parameters were recently used in an algorithm incorporated in an expert weaning system for adults and children [8, 9, 10, 11]. However, no data are currently available on the acceptable physiological limits of these three parameters for weaning during mechanical ventilation in children. Pertinent input from pediatric intensivists is essential to reach a consensus that would then allow the establishment of guidelines or algorithms on weaning during mechanical ventilation in pediatric patients.

The aim of this study was to determine the limits of respiratory parameters that are tolerated by pediatric intensivists while weaning children from mechanical ventilation.

Methods

Study design

A cross-sectional self-administered survey designed to determine the acceptable physiological limits of respiratory parameters on weaning during mechanical ventilation was developed for pediatric intensivists. Participants were asked to detail the lower and upper limits of specific respiratory parameters in three different clinical scenarios. Demographic information describing the physicians and their ICUs were also collected.

Study population

The survey was sent via conventional postal services to 222 pediatric intensivists working in 63 different PICUs located in four countries: Canada, France, Switzerland and Belgium. Their addresses were obtained by telephone through communication with the departmental secretaries in each of the 16 participating Canadian pediatric intensive care units and by mail from the “Groupe Francophone en Urgences et Réanimation Pédiatrique” (GFRUP) for the European countries [12]. We decided a priori to exclude physicians who had ceased to work in a PICU, who cared only for children less than 1 month old or adults, and intensivists who had outdated addresses.

Development of written questionnaire

Three case-scenarios were developed by two of the investigators (F.G., P.J.). We designed relevant clinical scenarios addressing three “virtual” patients (Appendix). The underlying pathology was different in these three cases, but reflected the most common diseases treated by pediatric intensivists.

For each of the “virtual” patients, the intensivists were asked to report what they would tolerate as an acceptable physiological limit, while weaning mechanical ventilation, for each of the following variables: minimal and maximal RR, minimal and maximal VT, minimal and maximal PetCO2.

Demographic data on the respondents were also recorded.

Questionnaire format

Predetermined ranges of values were provided in the questionnaire. For minimal RR, the available options ranged from 10 breaths per min (bpm) to 35 bpm in steps of 5 bpm. Regarding maximal RR, options ranged from 20 bpm to 70 bpm in steps of 10 bpm. Options regarding VT ranged from 3 ml/kg to 14 ml/kg in steps of 1 ml/kg. Finally, PetCO2 ranged from 20 mmHg to 70 mmHg in steps of 5 mmHg.

The intensivists were then asked at what time of the day—in the event that the weaning process were well tolerated—they would proceed with extubation of the patient. The available options were 08:00–12:00, 12:00–16:00, 16:00–20:00, 20:00–24:00 and 24:00–08:00. They could select all possible options.

The questionnaire was developed and validated in English. The questionnaire was anonymous and the responders could not be identified.

Questionnaire pretesting

A first version of the questionnaire was developed. Two pediatric intensivists and five PICU fellows from Sainte-Justine Hospital rated the questionnaire for discrimination, clarity, utility, face validity, content validity, construct validity and redundancy [13]. Each qualitative scale was rated according to a seven-point Likert scale ranging from 1 (complete disagreement) to 7 (complete agreement). A response of 5, 6 or 7 was considered satisfactory. The proportion of satisfactory response ranged from 71% to 86% for all items. The questionnaire was then revised, taking into account the comments received, and the same two pediatric intensivists and five PICU fellows rated the questionnaire once again. The proportion of satisfactory responses for this revised version of the questionnaire was 85% for discrimination and content validity and 100% for clarity, utility, face validity, construct validity and redundancy.

Administration of questionnaire

A first mailing was carried out on 25 October 2004, followed by a second mailing on 29 November 2004. Following collection of fully completed questionnaires and transcription of data to an Access database, the database was closed on 4 January 2005.

Statistical analysis

Descriptive data were expressed as frequencies (%), mean ± standard deviation and median (range). We used Student's t-test for group comparisons and chi-square with Yates correction for comparison of proportions. The level of statistical significance was set at p < 0.05. We also employed Pearson's r correlation coefficient to study the relationship between the respiratory parameters considered acceptable for weaning and the following parameters: gender of intensivist, country where he/she was working, years of experience and number of beds in the ICU.

Results

Demographic results

Ninety-seven (43%) pediatric intensivists answered the survey. Seven were ineligible either because they cared only for children less than 1 month old or because they were not working in an ICU. Therefore, input from 90 pediatric intensivists is reported in this survey. The response rate was 58% in Canada, 56% in Belgium and Switzerland and 26% in France. Demographic data of the responders and the ICUs they were working in are presented in Table 1.

Table 1 Respondents' characteristics (n = 90)
Full size table

Most respondents worked in Canada (47%) or France (36%); 80% of them were pediatricians and 86% worked full time in an ICU. Furthermore, 57% had practiced critical care medicine for at least 10 years. The median number of beds in the units they worked in was 12 (range 4–44), and most of the units (96%) treated medical as well as surgical patients. Forty-nine percent of the units were combined PICU and neonatal ICU (NICU). These ICUs used three different types of ventilators but some units had up to eight different ventilators. Finally, only 12% of the units were using a written protocol on weaning during mechanical ventilation.

Mechanical ventilation weaning

The minimal and maximal accepted RR, VT and PetCO2 for the three scenarios are reported in Fig. 1. There was a wide range of acceptable RR in children being weaned during mechanical ventilation according to pediatric intensivists, regardless of the patient's age. This range was greater for the younger patients. The median (25th;75th percentile) minimal acceptable RR was 20 bpm (15;25) for the 3-month-old patient, 15 bpm (10;15) for the 2-year-old patient and 10 bpm (10;15) for the 10-year-old patient. For the maximal RR, the median values were 50 bpm (40;60) for the 3-month-old patient, 40 bpm (30;40) for the 2-year-old patient and 30 bpm (30;40) for the 10-year-old patient.

Fig. 1
figure 1

Acceptable minimal and maximal respiratory rate (RR), tidal volume (V T ) and end-tidal CO 2 (PetCO 2) for the three scenarios. The y-axis corresponds to the number of respondents

Full size image

The median (25th;75th percentile) minimal acceptable VT was 5 ml/kg (4;6) for the 3-month-old and 2-year-old patients and 5 ml/kg (5;6) for the 10-year-old child. The median maximal acceptable VT was 10 ml/kg (10;10) for the 3-month-old patient and 10 ml/kg (10;12) for the 2- and 10-year-old patients.

The median (25th;75th percentile) for the minimal acceptable PetCO2 was 35 mmHg (30;35) for all three patients. For the maximal PetCO2 the median was 55 mmHg (50;60) for the 3-month-old patient, 50 mmHg (45;50) for the 2-year-old patient and 50 mmHg (50;55) for the 10-year-old patient.

For all three parameters, there was no statistically significant relationship between the responses of the intensivists and their gender (Pearson r 2 < 0.04), country (r 2< 0.08), years of practice in critical care medicine (r 2 < 0.14) or size of ICU they were working in (r 2 < 0.08).

The intensivists were mostly ready to extubate in the morning (> 98%); fewer considered extubation on a24-h basis (24% for the 3-month-old patient, 32% for the 2-year-old patient and 28% for the 10-year-old patient).

Regarding the mode of ventilation used for weaning, 86% of the intensivists used pressure support ventilation, 79% used gradual reduction of mechanical ventilatoryparameters, 48% employed continuous positive airway pressure and 25% used a T-piece.

Discussion

In this survey, the acceptable physiological limits on weaning during mechanical ventilation according to pediatric intensivists were broad, reflecting the absence of clear guidelines and supportive literature.

In the pediatric population, the range of normal values for respiratory rate is large even in the healthy child. This is mainly due to the decrease in respiratory rate with age and the difference in respiratory rate between sleeping and awake children [14, 15]. This could partially explain the great variation seen among the answers of the pediatric intensivists we surveyed. The wide range of these results, at a given age, probably also reflects different levels of tolerance regarding respiratory efforts during weaning. Interestingly, these results also suggest that for younger patients, the upper limit of the minimal acceptable respiratory rate overlaps with the lower limit of the maximal acceptable rate. Thus, the same respiratory rate was considered the highest acceptable rate by some intensivists and the lowest acceptable rate by others.

In our survey, there was a good consensus for minimal acceptable tidal volume: over 85% of the intensivists agreed on a minimal tidal volume between 4 and 6 ml/kg. However, for the maximal acceptable tidal volume, there was a higher variability that could not readily be explained. Ideal limits of tidal volume on ventilators have been studied and published only for ARDS [16]. Historically, technical problems have complicated measurements of tidal volumes on ventilators in pediatric patients. Nowadays, the values are more accurate due to technological advance including new spirometers, smaller connecting tubes and cuffed tubes with balloon pressure monitoring. As the measures become more accurate, we should be able to establish acceptable limits of tidal volume measures in mechanically ventilated pediatric patients. In this survey, tidal volume was normalized per body weight since it is usually prescribed in ml/kg. However, weight is not directly correlated to tidal volume. In our chosen scenarios the children had a weight close to the ideal body weight at the age concerned. Consequently, we believe that intensivists' responses were not affected by the patients' weight.

The opinions concerning maximal and minimal PetCO2 were also diverse. The minimal acceptable PetCO2 ranges from 20 to 40 mmHg and the maximum from 45 to 70 mmHg. In several studies in PICUs and NICUs, the accuracy of the PetCO2 to estimate the PaCO2 was quite good, but there was a significant systematic bias. The mean bias (PaCO2–PetCO2) ranged from 7 to 4 mmHg [17, 18, 19]. For example, McDonald et al. [17] observed that a PaCO2 of 40 mmHg corresponded to a PetCO2 of 37.3 mmHg, with a 95% probability of being between 35.5 and 39.2 mmHg. This limitation of PetCO2 with respect to accuracy and precision may have affected, at least in part, the responses to the three scenarios.

Pressure support ventilation was the ventilator-weaning mode chosen a priori in the scenarios, as it is the mode most frequently used in weaning protocols as well as in expert weaning systems constructed to support decision making about weaning [2, 3, 10]. However, in the literature, synchronized intermittent mandatory ventilation is the mode of weaning most frequently used by pediatric intensivists [1]. Nevertheless, irrespective of the weaning mode used, this should not affect the acceptable limits of RR, VT and PetCO2.

This study was conducted in 63 different PICUs. However, the vast majority of respondents worked in Canada or France. Therefore, these results reflect weaning strategies practiced by pediatric intensivists mainly in these countries. We did not find any significant difference in opinion regarding all parameters according to the different countries. The wide variety of opinions seems to differ from intensivist to intensivist and is not limited to a specific country.

The goal of developing new weaning strategies would be to enable us to reduce the duration of mechanical ventilation. However, as shown in this survey, few pediatric intensivists were ready to extubate a patient on a 24-h basis. Therefore, even if protocols or expert weaning systems could reduce the weaning time, the intensivists would have to extubate patients with limited time restrictions—otherwise, reduction in total ventilation time would not be possible.

Different possible biases could have affected the answers to this questionnaire. First, the fact that the three children presented in the case scenarios suffered from different underlying pathologies could have influenced the intensivists' answers. However, these patients represent the real PICU population and we wanted to point out limits that would be applicable to a large proportion of PICU patients in order to use them to develop algorithms for weaning of mechanical ventilation. Secondly, we realized later on that intensivists from Europe filled out only one questionnaire in each center. The intensivist filling out the questionnaire was trying to reflect the opinion of his entire group and not only his personal opinion.

This is the first study describing stated practice of weaning during mechanical ventilation in infants and children treated in PICUs. However, the study design reflects the way the intensivists would treat a “virtual” patient, which does not capture the true daily clinical practice in PICU. This is another limitation of surveys on stated practice pattern. The answers of the respondents cannot entirely reflect what is actually done for the real patients under their care. Therefore these results report the clinicians' perception of what they do and may not reflect reality.

Conclusion

According to this survey, physiologically acceptable limits of RR, VT and PetCO2 on weaning during mechanical ventilation are broad. No statistically significant demographic or geographic parameter could explain the wide range of responses. This survey demonstrates the diverse opinions amongst pediatric intensivists and the absence of evidence-based guidelines on weaning during mechanical ventilation. In order to develop weaning algorithms, we need to organize and consolidate our knowledge and thinking on weaning during mechanical ventilation. A study evaluating the bedside clinical practice of weaning from mechanical ventilation in children would be necessary before any guidelines could be suggested. We believe that this survey is an important step to further develop protocols on weaning children during mechanical ventilation using computerized expert systems.