Introduction
Weaning patients with respiratory failure from ventilatory support is one of the most challenging problems in intensive care. Unnecessary mechanical ventilation poses increased risk of complications to patients; however, premature liberation from mechanical ventilation may also be harmful[
1]. In the past few decades, a variety of predictors have been developed to identify patients ready to breathe independently[
2]. Although to date, spontaneous breathing trial (SBT) is considered the most accurate index for predicting weaning success, 15% to 20% of patients succeeding in SBT require reintubation[
3,
4]. The pathophysiology of weaning failure is complex and involves interaction between cardiopulmonary reserve, autonomic function, and musculoskeletal capacity[
5,
6]. Thus, it may be hard to assess the interplay between those factors based on a single or a few predictors.
Heart-rate variability (HRV) has been related to the balance between parasympathetic and sympathetic regulation of cardiac activity, respiration, baroreflex, and thermal regulation[
7‐
9]. It is a noninvasive and valuable tool to characterize autonomic function and cardiorespiratory interaction[
10]. The impact of mechanical ventilation on HRV has been studied in newborn babies, children with brain death, and healthy young adults placed on sedation and paralysis[
11‐
13]. Change of HRV between different ventilator settings has also been described in a canine model[
14]. Accordingly, measurement of HRV may help evaluate physiological responses to the weaning process. In a case series by Shen and associates,[
15], decrease in HRV is the main finding in patients with weaning failure, and the authors suggested that change of HRV components may be a potential tool of automatically gathered parameters during ventilator weaning. However, the result has not yet been replicated in larger-scale studies. Further, change of HRV between SBT and extubation was not explored in that study.
Thus, the aim of the present study was to investigate change of HRV during the entire weaning process in patients recovering from respiratory failure. The potential predictive value of change of HRV on SBT and extubation outcomes is also evaluated.
Discussion
This is the largest study to date demonstrating the relation between change of HRV and outcomes of SBT and extubation in critically ill patients undergoing weaning from mechanical ventilation. The two major findings are that first, reduced HRV was significantly associated with SBT failure among patients with their first SBT; second, inability to increase HRV after extubation correlated with extubation failure in patients who passed SBT.
During the weaning process, the use of SBT could impose cardiopulmonary stress on certain patients. Specifically, patients for whom SBT fails are subjected to higher cardiovascular and pulmonary stress than those who succeed in the trial[
21‐
23]. When the stress system is activated, it responds with consequent release of catecholamines due to sympathetic stimulation[
24]. Decrease in HRV may reflect elevated sympathetic activity and an impairment in the physiological regulatory and adaptive mechanisms[
25,
26]. In line with the concepts and the findings of Shen[
15], the present study showed reduced HRV in patients with failed SBT. Further, more-reduced HRV was an independent predictor of SBT failure in multivariate analysis. Continuous monitoring of electrocardiography is the routine practice in modern ICU care; thus, HRV measurements should be easily accessible and helpful in evaluating patients receiving SBT. However, real-time analysis of HRV remains challenging, because greater effective lengths of observation are required to provide better spectral resolution. Yet, in this era of rapidly advancing information technology in healthcare, we believe that this drawback will be solved in the near future, and point-of-care monitoring of HRV could be clinically feasible.
Similar to HRV, the rhythmic activity of respiration is not monotonous but is characterized by breath-to-breath variability in the tidal volume, respiratory rate, and inspiratory time of sequential breaths[
27]. An inverse relation between breathing variability and respiratory loading has been observed in healthy subjects[
28,
29]. As expected, in patients undergoing SBT, breathing variability is greater in those successfully weaned from mechanical ventilation[
30]. Breathing variability is inversely correlated with HRV in subjects free of cardiovascular and respiratory diseases[
31]; however, it seems not to be the case in patients during the weaning process. Further study of simultaneously measuring breathing variability and HRV during ventilator weaning should be carried out to clarify this issue and the role it plays in predicting weaning outcomes.
One major strength in this study is that we, for the first time, assessed change of HRV after extubation. This study found significant association between increased HRV and successful extubation in patients who passed SBT. In other words, incapability to increase HRV suggested a higher probability of extubation failure. Increased HRV indicates health, fitness, and a well-functioning autonomic control mechanism[
32]. Therefore, it is anticipated that patients would do better regarding the extubation outcome if they had increase in HRV after extubation. Patients requiring reintubation have a high mortality rate[
1], and a direct and specific effect of extubation failure and reintubation on patient outcomes was demonstrated in a recent study[
33]. Prophylactic use of noninvasive ventilation may reduce reintubation and mortality rates in patients at high risk for extubation failure[
34].
However, high risk for extubation failure is difficult to define. In this study, HRV was measured shortly after extubation, and failure to increase HRV independently predicted the need for reintubation among patients being extubated after successful SBT. Early identification of at-risk patients will allow clinicians to use preventive measures such as noninvasive ventilation support, adjustment of medications, and so on, before the adverse event takes place.
In the present study, dynamic change of TP was the only recognized HRV component associated with both SBT and extubation outcomes. The TP is a net effect of all possible physiological mechanisms contributing to HRV and thus represents the overall variability of the cardiac autonomic nervous system. Of the two major components of power-spectral analysis, HF reflects modulation of parasympathetic activity; and LF indicates sympathetic modulation on the heart, although it may be also associated with parasympathetic contribution[
7]. Usually, change of TP is accompanied by corresponding change in either or both of HF and LF, but we did not find such a result. This could be explained by the complex pathophysiologic process of ventilator weaning. This study enrolled a heterogeneous group of critically ill patients, and they should respond differently to the weaning process and have a variety of causes of weaning failure. Thus, the change of certain HRV components may be obscured. Further studies aiming to explore the association between HRV change and mechanisms of weaning failure will make clear the puzzle.
Other than TP, which is a reflection of total HRV, we found that change of VLF was useful in risk stratification for predicting extubation outcome. The VLF contains rhythms from several physiological variables like hormones, temperature, and vasomotion, and depends on parasympathetic outflow[
10,
35]. Evidence suggests that the power in this range is substantially influenced by physical activity, reflecting the activity of both parasympathetic and renin-angiotensin systems[
36]. Recently, the value of VLF measurements has also been proven in risk stratification in patients with multiple organ dysfunction syndrome, because it may represent a viable interorgan communication[
37]. In that study, lower VLF was significantly associated with a worse outcome. Accordingly, it is not surprising that increased VLF served as a useful index of patient well-being after extubation in the present study.
Weaning parameters have been held as a holy grail for ICU clinicians to predict weaning outcomes. However, recent studies argued against their utility in general ICU patients, neurocritical care patients, and patients with prolonged mechanical ventilation[
38‐
40]. Similarly, weaning parameters were not associated with SBT or extubation outcomes in our work. Weaning failure is usually multifactorial; thus, weaning parameters that assess a single physiological function may have limited predictive accuracy[
16,
20]. Several studies have demonstrated poor specificity of these parameters for predicting weaning success[
41,
42]. Instead, HRV probably represents the complex interplay of multiple physiological inputs to the heart, and it would intuitively be better to predict weaning outcomes, as shown in this study.
HRV can be significantly influenced by various categories of medications; thus, the impact should be considered while interpreting HRV. The use of β
2 agonists and anticholinergics were known to affect HRV and were more commonly seen in our patients with failed SBT[
43,
44]. Because all the measures in our study were completed within hours, and we compared change of HRV between study periods by using a paired
t test, the influence of drug effects on HRV could be minimized or even eliminated. Moreover, to reduce this source of error, we tried not to administer possible confounding drugs during HRV measurements. In addition, multivariate analysis in the present study did not show an independent role of these medications on the prediction of SBT outcomes.
Certain limitations of the study must be considered. First, the results obtained must be interpreted with caution, as external factors other than SBT or extubation could have influenced the HRV values. This might hinder appropriate interpretation of the data, even though we tried hard to eliminate these intervening factors. Second, our study was conducted in a single center, and the findings may not apply similarly to other institutions, although incidence rates of SBT and extubation failure were within previously reported ranges[
16]. Future multicenter studies would be helpful in this regard. Third, for the accurate measurement of HRV, we excluded patients with arrhythmia or taking antiarrhythmics, and these patients may have different HRV responses to the weaning process. More-sophisticated methods for dealing with arrhythmia during HRV analysis are needed to allow investigation of clinical implications of HRV in this specific patient population. Fourth, the HRV thresholds derived have not been prospectively validated, although the findings were pathophysiologically reasonable and predictable. Undoubtedly, validation with independent data sets will consolidate our results.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
CTH, YJT, and CJY conceived the study and participated in its design and coordination. CTH, JWL, SYR, and HDW participated in the data analysis and interpretation. CTH drafted the manuscript. CTH, YJT, JWL, SYR, HDW, and CJY participated in the revision of the manuscript. All authors read and approved the final manuscript.