Purpose To assess the association between respiratory muscle weakness (RMW) at intensive care unit (ICU) discharge and 5-year mortality and morbidity, independent from confounders including peripheral muscle strength.
Methods Secondary analysis of the prospective 5-year follow-up of the EPaNIC cohort (ClinicalTrials.gov: NCT00512122), limited to 366 patients screened for respiratory and peripheral muscle strength in the ICU with maximal inspiratory pressure (MIP) after removal of the artificial airway, and the Medical Research Council sum score. RMW was defined as an absolute value of MIP <30 cmH2O. Associations between RMW at (or closest to) ICU discharge and all-cause 5-year mortality, and key measures of 5-year physical function, comprising respiratory muscle strength (MIP), hand-grip strength (HGF), 6 min walk distance (6MWD) and physical function of the SF-36 quality-of-life questionnaire (PF-SF-36), were assessed with Cox proportional hazards and linear regression models, adjusted for confounders including peripheral muscle strength.
Results RMW was present in 136/366 (37.2%) patients at ICU discharge. RMW was not independently associated with 5-year mortality (HR with 95% CI 1.273 (0.751 to 1.943), p=0.352). Among 156five-year survivors, those with, as compared with those without RMW demonstrated worse physical function (MIP (absolute value, cmH2O): 62(42–77) vs 94(78–109), p<0.001; HGF (%pred): 67(44–87) vs 96(68–110), p<0.001; 6MWD (%pred): 87(74–102) vs 99 (80–111), p=0.009; PF-SF-36 (score): 55 (30–80) vs 80 (55–95), p<0.001). Associations between RMW and morbidity endpoints remained significant after adjustment for confounders (effect size with 95% CI: MIP: −23.858 (−32.097 to −15.027), p=0.001; HGF: −18.591 (−30.941 to −5.744), p=0.001; 6MWD (transformed): −1587.007 (−3073.763 to −179.253), p=0.034; PF-SF-36 (transformed): 1.176 (0.144–2.270), p=0.036).
Conclusions RMW at ICU discharge is independently associated with 5-year morbidity but not 5-year mortality.
- respiratory muscles
- critical care
- pulmonary rehabilitation
Data availability statement
Data are available on reasonable request. Data sharing is considered under the format of collaborative projects. Proposals can be directed to the senior authors (GH and GVdB).
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
What is the key question?
Does intensive care unit (ICU)-acquired respiratory muscle weakness during critical illness impact long-term outcome, independently from confounders including peripheral muscle weakness?
What is the bottom line?
Respiratory muscle weakness at ICU discharge independently associates with reduced peripheral and respiratory muscle strength, physical function, and quality-of-life at 5 years post-ICU, but does not independently increase the 5-year mortality risk.
Why read on?
In a large population of adult critically ill patients (mortality N=366, morbidity N=156) prospectively assessed for peripheral and respiratory muscle strength on the ICU, we illustrate that respiratory muscle weakness acquired in the ICU contributes to the legacy of critical illness, and could therefore hold potential as a target to improve long-term prognosis in ICU survivors.
Respiratory muscle weakness (RMW), in particular diaphragm dysfunction, is a frequent complication of critical illness, occurring in over 60% of patients requiring mechanical ventilation.1–3 Risk factors include higher age, infection, systemic inflammation, illness severity, mechanical ventilation—both overassistance and underassistance, and certain drugs such as sedatives.1–5 RMW often co-occurs and shares common features with peripheral muscle weakness acquired in the intensive care unit (ICU), labelled as ICU-acquired weakness (ICUAW). Both are associated with increased ICU and hospital mortality.6 However, recent research revealed that RMW is more frequent than ICUAW, risk factors do not completely overlap, and short-term impact on morbidity differs. Whereas RMW contributes to weaning failure, increased ICU mortality3 7 8 and early readmission,1 6 ICUAW predominantly affects the duration of mechanical ventilation, ICU and hospital stay.6 9 As such, RMW and ICUAW may be considered separate, though overlapping, entities.1 10 In contrast to ICUAW, of which we and others recently described the impact on long-term outcomes,11 12 studies of RMW and its associated outcomes beyond the index hospitalisation, when patients have overcome the major short-term risks of RMW, are scarce. Current evidence is confined to two case series with follow-up periods of respectively 1 and 2 years that revealed conflicting results.10 13 Identifying a possible relationship between RMW in the ICU and the legacy of critical illness is highly relevant as preventive strategies for the long term, multidimensional limitations reported for survivors of critical illness are lacking. Respiratory muscle strength training in ICU survivors may hold promise in improving the long-term outcomes of critically ill patients.
We hypothesised that RMW in critically ill patients, as assessed by the maximal inspiratory pressure (MIP) after removal of (or without) artificial airway at (or closest to) ICU discharge, associates with 5-year mortality and morbidity, independent of confounders, including peripheral muscle strength. To this aim, we investigated the 5-year outcomes of a subgroup of the post-EPaNIC prospective follow-up cohort,14 who received systematic in-ICU respiratory and peripheral muscle strength testing.
Study design and patient population
This is a secondary analysis of the prospective 5-year follow-up study of EPaNIC patients. Study design and methods of the EPaNIC trial and the EPaNIC follow-up study (ClinicalTrials.gov: NCT00512122) have been published previously14 15 and are summarised in the online supplemental file. We here focus on those patients who were systematically screened in the ICU for respiratory and peripheral muscle strength. This screening was performed for long-stay patients (ICU stay ≥8 days). Additionally, a random set of short stayers (ICU stay <8 days) were tested on the ward on day 8±1. The last evaluations performed in the ICU closest to ICU discharge, or on day 8±1 on the ward are further referred to as values ‘at ICU discharge’.
Measurement of respiratory muscle strength and diagnosis of RMW
In patients without an artificial airway, MIP was measured according to the ATS guidelines16 as described earlier.9 Contraindications included flail chest, pneumothorax, haemodynamic instability, intracranial hypertension, respiratory distress or high flow oxygen therapy. A mouthpiece, incorporating a small leak to prevent glottic closure during the forceful inspiration, was used. Measurements were performed with the micro medical respiratory pressure metre, CareFusion with Puma PC software. The patients were instructed to perform a maximal inspiratory manoeuvre starting from functional residual capacity. Maximal static inspiratory pressure was determined as the pressure maintained for 1 s. The best of three consecutive measurements was recorded. RMW was defined as an absolute value of MIP <30 cmH2O.13 17
The primary endpoint was all-cause 5-year mortality. Secondary endpoints included respiratory muscle strength (MIP, cmH2O) at 5 years follow-up as well as three key measures of physical function frequently compromised in ICU survivors, amounting to the so-called legacy of critical illness. These comprise hand-grip strength (HGF, %predicted), 6 min walk distance (6MWD, %predicted), and the physical function domain of the 36-item Short Form Health Survey (PF-SF-36, range 0–100, higher values indicating better scores). Additionally, the associations between RMW and other 5-year morbidity outcomes were explored, including peripheral muscle strength (Medical Research Council (MRC) sum score and hand-held dynamometry), Barthel index (indicating the degree of physical independence) and the Physical and Mental Component Score of the SF-36.
Descriptive statistics included median and IQR for continuous variables and numbers and percentages for categorical variables. Continuous data were compared with the Mann-Whitney U test, categorical variables with the χ2 or Fisher’s exact test, as appropriate. As this was a secondary analysis, sample size was not modifiable, however, it was estimated that the sample size was sufficient for the primary endpoint (see online supplemental file). Bootstrap resampling (N=1000) was performed to obtain more robust estimates of effect sizes (denoted as bias-corrected and accelerated CI, Bca CI). Analyses were performed with SPSS V.26 (IBM) and R V.3.6.1. Two-sided p values ≤0.05 were considered statistically significant.
Primary endpoint: explanatory modelling of the association between RMW at ICU discharge and all-cause 5-year mortality
Crude 5-year mortality of patients with and without RMW was compared as a time-to-event analysis with log-rank test and visualised with Kaplan-Meier plots. The effect size was estimated with Cox proportional hazard analyses, adjusting for a priori defined confounders, identified through a systematic literature search.18 Potential confounders comprised demographic variables, comorbidities and ICU treatments and events, particularly including peripheral muscle strength. To identify the latter, the MRC sum score at ICU discharge was dichotomised at 55, according to recent findings concerning its association with long-term outcomes.11 Further details on the search strategy, confounders identified and modelling are provided in online supplemental file.
Secondary endpoints: explanatory modelling of the association between RMW at ICU discharge and 5-year morbidity
The association between 5-year morbidity outcomes and RMW at ICU discharge was explored with the Mann-Whitney U test. The association between RMW at ICU discharge and respiratory muscle strength at 5 years, as well as the three key measures of physical function, comprising HGF, 6MWD and the PF-SF-36, was further assessed by multivariable linear regression analyses, adjusted for confounders identified through a systematic literature search (see online supplemental file for further details).
As the cut-off for MIP to define RMW has not been validated for long-term outcomes, we assessed the functional form of the relationship between MIP at ICU discharge and the primary and secondary outcomes through an LOESS-smoother on the Martingale residuals plot19 for survival analyses and on the scatter plots for linear regression analyses. If appropriate, analyses were repeated with alternative modelling of MIP.
Multiple sensitivity analyses were performed. This involved restricting 5-year mortality analyses to hospital survivors. Also, the proportional hazard assumption was checked for each variable in each of the Cox regression models with the Schoenfeld residuals test. Where appropriate, sensitivity analyses were performed by adding factors for which the assumption was violated as time-dependent covariates. For both 5-year mortality and morbidity, sensitivity analyses also involved adjusting for MRC at ICU discharge as a continuous variable as well as dichotomised for the previously defined cut-off of 48.
Patient cohorts and characteristics
The patient cohort with both peripheral and respiratory muscle strength assessment at ICU discharge consisted of 368 patients (figure 1). Five-year mortality data were available for 366 patients. Among 246 survivors, 156 (63.4%) were evaluated for 5-year morbidity. For the 5-year mortality cohort, the median age was 60 (50–72) years, duration of mechanical ventilation 4 (2–10) days and ICU stay 10 (3–17) days. For the 5-year morbidity cohort, the median age was 60 (51–70) years, duration of mechanical ventilation 3 (1–9) days and ICU-stay 8 (2–15) days. Further details are provided in table 1 and online supplemental table 1. RMW was present in 136/366 (37.2%) patients at ICU discharge. Patients with RMW were older, more likely to be female, to have a high nutritional risk score and to suffer from diabetes. Patients with RMW were sicker on admission and were longer exposed to corticosteroids and benzodiazepines, more frequently acquired a new infection in the ICU, and had a longer duration of mechanical ventilation and ICU stay. Interestingly, the incidence of sepsis on admission was not significantly different for patients with and those without RMW.
Five-year mortality analyses
Total 5-year mortality in the studied cohort was 104/366 (28.4%), involving 51/136 (37.5%) of patients with RMW and 53/230 (23.0%) patients without RMW. As mortality data were obtained from the national registry, no information on cause of death was available. Time to death was significantly different for patients with vs those without RMW (log-rank p=0.002), the former experiencing a higher event rate (online supplemental figure 1). However, when adjusted for confounders, RMW was not associated with 5-year mortality (HR 1.273 (95% Bca CI 0.751 to 1.943), p=0.352) (table 2).
Five-year morbidity analyses
Five-year survivors with RMW at ICU discharge had worse 5-year morbidity outcomes than those without RMW (MIP (absolute value): 62 (42–77) cmH2O vs 94 (78–109) cmH2O, p<0.001; HGF (%pred): 67% (44%–87%) vs 96% (68%–110%), p<0.001; 6MWD (%pred): 87% (74%–102%) vs 99% (80%–111%), p=0.009; PF-SF-36: 55 (30–80) vs 80 (55–95), p<0.001). Further exploratory analyses showed that patients with RMW also scored worse for all other 5-year morbidity endpoints, including MRC sum score, hand-held dynamometry, Barthel index and quality-of-life questionnaires (online supplemental figure 2). Adjusted for confounders, RMW remained independently associated with all secondary endpoints (MIP (absolute value, cmH2O): B:−23.858 (95% Bca CI −32.097 to −15.027), p=0.001; HGF (%pred): B: −18.591 (95% Bca CI −30.941 to −5.744, p=0.001; 6MWD (%pred, transformed) B: −1587.007 (95% Bca CI −3073.763 to −179.253), p=0.034; PF-SF-36 (transformed with higher transformed values indicating lower scores): B: 1.176 (95% Bca CI0.144 to 2.270), p=0.036) (table 2). Additional information on rehabilitation trajectory, working and living conditions is provided in online supplemental table 3.
Martingale residual plots and scatter plots did not support a clear cut-off at 30 cmH2O for MIP at ICU discharge, for prediction of the studied long-term endpoints (online supplemental figure 3). Reanalyses with both MIP as a continuous variable and with MIP at a cut-off of 45 cmH2O, possibly suggested by the plots as a better cut-off, did not alter the conclusions (online supplemental table 2).
RMW at ICU discharge was not independently associated with 5-year mortality, neither when the analysis was restricted to hospital survivors, nor when MRC was modelled as a continuous variable or when dichotomised at 48. Also, the addition of time-dependent covariates to the survival model for those factors violating the proportional hazard assumption did not alter the conclusion (online supplemental table 2). For the morbidity endpoints, alternative modelling strategies for the MRC sum score did not affect the conclusions (online supplemental table 2).
In this patient cohort of mixed adult critically ill patients, systematically screened for respiratory and peripheral muscle strength at ICU discharge and prospectively followed, we demonstrated that RMW at ICU discharge was independently associated with lower respiratory muscle strength at 5 years follow-up. Furthermore, RMW at ICU discharge independently associated with key morbidity outcomes that characterise the postintensive care syndrome,11 12 14 20–23 comprising HGF, 6MWD and the PF-SF-36 quality-of-life questionnaire. Importantly, these analyses were adjusted for confounders, including peripheral muscle strength at ICU discharge. In contrast, RMW at ICU discharge did not independently associate with 5-year mortality.
The incidence of RMW in this mixed cohort of critically ill patients, as assessed by a volitional MIP manoeuvre in cooperative patients without artificial airway at ICU discharge, was 37.2%. Previous research reported diaphragm dysfunction in 64% of patients on ICU admission, strongly associated with the presence of sepsis and with illness severity.3 At the start of the weaning process, the incidence of RMW ranged between 54% and 63%,1 17 with severe diaphragm dysfunction present in 31%.10 In patients with peripheral weakness, up to 80% also suffer from diaphragm dysfunction.7 The lower incidence of RMW in our study may have several explanations. First, factors contributing to RMW may improve over time. Indeed, in contrast to the findings on ICU admission3 or up to a week prior to successful extubation,24 we and others who assessed RMW at a later stage in the ICU1 13 found no association between sepsis and RMW. Second, by measuring RMW after removal of the artificial airway at ICU discharge, we evidently eliminated the sickest patients from this analysis, including those who were never weaned or were not cooperative after weaning. As such, we likely excluded patients with severe RMW and poor short-term outcomes. This allowed us to focus on the long-term impact of RMW, studying only patients who overcame the major short-term risks.
The detrimental short-term impact of RMW has been well documented.6 This includes increased ICU and hospital mortality rates,1 3 6 possibly linked with longer duration of mechanical ventilation,25 later weaning24 and increased risk of weaning failure.1 2 13 The combination of respiratory and peripheral muscle weakness in particular associates with poor outcomes.17 Data on the impact of RMW beyond the acute timeframe are scarce. We demonstrated increased 5-year mortality in patients with RMW (37.5%) at ICU discharge as compared with patients without RMW (23.0%). However, when adjusted for baseline characteristics, illness severity, comorbidities, ICU exposures and peripheral muscle strength at ICU discharge, the risk of death within 5 years was not increased in patients with RMW. These data were confirmed in all sensitivity analyses, including the scrutinised analysis limited to hospital survivors. Previous data obtained from 124 patients suggested that 1-year mortality was significantly increased in patients with RMW (31% vs 7%) at the time of successful extubation.13 However, 21% of patients with RMW experienced extubation failure, and 47.6% of deaths occurred on the ICU. The difference in mortality after hospital discharge for those with, and those without RMW, was not statistically significant. Saccheri et al recently showed that 2-year mortality was not different between patients with and those without diaphragmatic dysfunction at the start of the weaning process (36% vs 29%).10 These data were confirmed when analyses were limited to hospital survivors. Survival at 2 years was worse for patients suffering from both ICUAW and RMW (36%), though the patient sample was too small to allow adjustment for confounders. Our study extended the follow-up time window and suggests that, if patients with RMW overcome the acute challenges, long-term survival is not independently affected by RMW.
We subsequently evaluated the association between RMW at ICU discharge and 5-year morbidity. We demonstrated that, among 5-year survivors, RMW was independently associated with reduced inspiratory muscle strength, but also with peripheral muscle weakness as assessed with HGF, poor physical function as assessed with 6MWD and low quality-of-life as assessed with PF-SF-36. These findings are novel, as data on the association between RMW at ICU discharge and morbidity beyond the index hospitalisation are virtually absent. Saccheri et al recently found no association between RMW in critically ill patients and the quality of life assessed in 40 survivors at 2 years follow-up.10 In light of the data we here present, the lack of effect in the aforementioned paper may be due to a lack of power. The finding of an independent relationship between RMW at ICU discharge and typical features of the long-term legacy of critical illness may have important consequences as, currently, strategies to prevent or cure this long-term burden of critical illness are lacking.26 27 Respiratory muscle training has demonstrated multiple positive effects when performed in various patient populations, including patients with COPD, chronic heart or kidney disease, as well as in healthy volunteers and athletes. These benefits surpass mere improvement of respiratory muscle strength and include improved submaximal and maximal exercise capacity, functional capacity and quality of life.28–32 Several mechanisms have been proposed to explain such beneficial effects,28 33 34 which are relevant at both strenuous and submaximal exercise effort, and reversible by inspiratory muscle training, in the aforementioned patient populations.35 The role for inspiratory muscle training as an intervention to specifically reduce the long-term burden of critical illness therefore deserves consideration and further evaluation through randomised clinical trials.
Our study has several strengths. To the best of our knowledge, this is the largest dataset on the association between RMW in critically ill patients and long-term outcomes. By focusing on RMW at ICU discharge, in contrast to previous studies, we dissociated the outcomes of interest from the acute risks of RMW. Furthermore, the dataset is unique, as we have concomitant information on peripheral muscle strength, well known to contribute to the long-term sequelae of critical illness, which allowed us to assess the independent effect of RMW. Finally, this is the first study to assess the association between RMW at ICU-discharge and morbidity at 5 years post-ICU, a time window of follow-up far exceeding currently available perspectives. This study also has several limitations. First, we measured respiratory muscle strength with the MIP, rather than with transdiaphragmatic (or transtracheal) twitch pressure generated in response to bilateral magnetic phrenic nerve stimulation (BAMPS), which is considered to be the golden standard in critically ill patients.3 25 36 MIP is a simple screening measure of global inspiratory muscle strength, not limited to evaluation of the diaphragm. MIP correlates with twitch tracheal pressure7 37 and several researchers have demonstrated its clinical relevance in the ICU setting.6 13 17 24 37 MIP requires consciousness and cooperation, while BAMPS is not dependent on patient effort. As we performed measurements at ICU discharge and after removal of the endotracheal tube aiming to study postacute phase impact of RMW, conscious cooperation was a reasonable inclusion criterion, and MIP was considered a suitable measure for our research purpose. Additionally, while MIP can be easily performed at the bedside, BAMPS is a time-consuming and technically challenging procedure, requiring expensive equipment that is only available in a few ICUs worldwide.16 36 38 Second, although we adjusted our analyses for confounders identified through a systematic literature search, we cannot exclude unmeasured confounding. Consequently, as this is an observational study, we cannot draw causal conclusions. Third, morbidity endpoints are subject to selection bias, as not all survivors were evaluated. Fourth, as no systematic evaluations at intermediate time points were made, no conclusions can be drawn with regard to differential recovery trajectories of respiratory muscle strength. Fifth, as maximal expiratory pressure (MEP) was not a prespecified outcome in our study, we cannot draw any conclusions on a potential association of expiratory muscle weakness with long-term outcomes. However, recent data could not define an independent effect of MEP when MIP was taken into account with regard to short-term outcomes.39 Finally, all patients were included in the EPaNIC trial, possibly limiting generalisability.
We conclude that RMW at discharge from the ICU did not independently associate with 5-year mortality. However, RMW at ICU discharge was associated with peripheral and respiratory muscle strength, physical function and quality of life in 5-year survivors, independent of baseline characteristics, comorbidities, illness severity, ICU exposures and peripheral muscle strength. As inspiratory muscle training has the potential to improve respiratory muscle strength as well as other outcomes, our findings suggest that inspiratory muscle training is an attractive and promising intervention that should be investigated in ICU survivors in randomised studies with the aim to improve long-term outcomes of critical illness.
Data availability statement
Data are available on reasonable request. Data sharing is considered under the format of collaborative projects. Proposals can be directed to the senior authors (GH and GVdB).
The study protocol and informed consents of EPaNIC and its long-term follow-up were approved by the Ethical Committee Research UZ/KU Leuven (ML4190).
We are indebted to all patients and controls for their participation in this study; to Helena Van Mechelen, Tine Vanhullebusch, Sanne Verweyen, Tim Van Assche for acquisition of data, to Helena Van Mechelen, Tine Vanhullebusch, Sanne Verweyen, Tim Van Assche, Alexandra Hendrickx, Heidi Utens, Sylvia Van Hulle for their technical and administrative support.
GVdB and GH contributed equally.
Contributors GH and GVdB designed the study and planned the statistical analysis; data were acquired by GH, PM, YD, AW, JG, MC, NVA, GVdB and GH analysed and interpreted the data, and drafted the manuscript; critical revision of manuscript content was done by all authors, funding was obtained by MC, JG, GH, NVA and GVdB; administrative and technical support was provided by PJW.
Funding This work was supported by the Research Foundation – Flanders, Belgium (grant G.0399.12, Fundamental Clinical Research fellowship to GH: 1805116N, MPC: 1700111N; aspirant PhD fellowship to NVA: 1131618N), the clinical research fund (KOF) of the University Hospitals Leuven, Belgium (postdoctoral research fellowship to JG); the Methusalem programme of the Flemish Government (METH/08/07, renewed as METH/14/06 via KU Leuven) to GVdB; the European Research Council ERC Advanced grants (AdvG-2012-321670 from the Ideas Program of the EU FP7 and AdvG-2017-785809 from the Horizon 2020 Programme of the EU) to GVdB. An unrestricted and non-conditional research grant was given to KULeuven by Baxter between 2007 and 2010.
Disclaimer The authors have no conflict of interest with the sponsors of the study.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.