Article Text

Original research
Development and validation of a mechanical power-oriented prediction model of weaning failure in mechanically ventilated patients: a retrospective cohort study
  1. Yao Yan1,2,
  2. Jiye Luo3,
  3. Yanli Wang3,
  4. Xiaobing Chen3,
  5. Zhiqiang Du2,
  6. Yongpeng Xie3,
  7. Xiaomin Li1,3
  1. 1Department of Emergency Medicine, Lianyungang Clinical College of Nanjing Medical University, Lianyungang, Jiangsu, China
  2. 2Department of Critical Care Medicine, The Second People’s Hospital of Lianyungang, Lianyungang, Jiangsu, China
  3. 3Department of Emergency Medicine, The First People’s Hospital of Lianyungang, Lianyungang, Jiangsu, China
  1. Correspondence to Dr Xiaomin Li; lyglxm1{at}163.com; Dr Yongpeng Xie; xyp8285{at}163.com

Abstract

Objective To develop and validate a mechanical power (MP)-oriented prediction model of weaning failure in mechanically ventilated patients.

Design A retrospective cohort study.

Setting Data were collected from the large US Medical Information Mart for Intensive Care-IV (MIMIC-IV) V.1.0, which integrates comprehensive clinical data from 76 540 intensive care unit (ICU) admissions from 2008 to 2019.

Participants A total of 3695 patients with invasive mechanical ventilation for more than 24 hours and weaned with T-tube ventilation strategies were enrolled from the MIMIC-IV database.

Primary and secondary outcome Weaning failure.

Results All eligible patients were randomised into development cohorts (n=2586, 70%) and validation cohorts (n=1109, 30%). Multivariate logistic regression analysis of the development cohort showed that positive end-expiratory pressure, dynamic lung compliance, MP, inspired oxygen concentration, length of ICU stay and invasive mechanical ventilation duration were independent predictors of weaning failure. Calibration curves showed good correlation between predicted and observed outcomes. The prediction model showed accurate discrimination in the development and validation cohorts, with area under the receiver operating characteristic curve values of 0.828 (95% CI: 0.812 to 0.844) and 0.833 (95% CI: 0.809 to 0.857), respectively. Decision curve analysis indicated that the predictive model was clinically beneficial.

Conclusion The MP-oriented model of weaning failure accurately predicts the risk of weaning failure in mechanical ventilation patients and provides valuable information for clinicians making decisions on weaning.

  • adult intensive & critical care
  • respiratory medicine (see thoracic medicine)
  • intensive & critical care

Data availability statement

Data are available upon reasonable request. The data set analysed to generate the findings for this study are available from the corresponding author on reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

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.

Strengths and limitations of this study

  • Multiple imputation was used to impute variables with <15% missing data to minimise the bias caused by missing values.

  • Continuous predictors with non-linear trends were transformed into categorical variables based on their distribution and clinical significance, increasing the utility of this prediction model.

  • The nomogram was constructed using the multivariable logistic regression analysis with the R package ‘rms’.

  • The area under the receiver operating characteristic curve, calibration curves and decision curve analysis were enrolled to evaluate the performance of the prediction model in the development and validation cohort.

  • We could not compare the performance of the mechanical power-oriented model with the existing model (eg, the modified Burns Wean Assessment Program scores).

Introduction

Mechanical ventilation is an advanced respiratory support technique widely used in the intensive care unit (ICU).1 Both prolonged ventilation and premature weaning are associated with poor patient outcomes, resulting in an increased risk of ventilator-associatedpneumonia, longer hospital stays and higher mortality.2 Therefore, it is important to accurately predict the risk of weaning failure in mechanically ventilated patients and optimise the weaning time.3 The reasons for weaning failure are complicated, with airway and pulmonary dysfunction, and the imbalance of respiratory load and respiratory muscle function as main influencing factors.4–6 Traditional weaning evaluation methods include Rapid Shallow Breathing Index (RSBI) and spontaneous breathing test (SBT). However, the specificity of RSBI is affected by various factors such as ventilator settings, health state and body position.7 In addition, between 3% and 19% of the patients who passed the SBT were reintubated due to weaning failure,7 8 which may be related to the inaccuracy of short-term SBTs in reflecting airway and lung function, and the lack of objectivity in assessing the endurance of respiratory muscles to spontaneous breathing load.

Mechanical power (MP) is the energy delivered by the ventilator to the entire respiratory system per unit time.9 MP can be used as a dynamic and objective measure of the energy load on the respiratory muscles before weaning, and accurately reflects the airway and lung function status. Based on multiple studies, Ghiani et al10 11 concluded that MP can be used to assess the workload of the respiratory muscles before SBT and to guide the weaning of patients with long-term mechanical ventilation. In this study, we aimed to further develop and validate an MP-oriented weaning failure prediction model through a retrospective analysis of the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database and use nomograms to visualise the model for evaluation of weaning failure to assist clinicians in making decisions about weaning.

Methods

Data source

We performed a retrospective analysis of data from the large US MIMIC-IV V.1.0, which integrates comprehensive clinical data from 76 540 intensive care unit (ICU) admissions at Beth Israel Deaconess Medical Center in Boston, Massachusetts, USA, from 2008 to 2019. Since all patient identification information was de-identified, the requirement for informed consent was waived.12 The researcher (YY) completed the National Institutes of Health (NIH) ‘Protecting Human Research Participants’ online course and obtained access to the database (Certification Number: 41699414).

Study cohort

After screening the MIMIC-IV database, a total of 3695 patients with invasive mechanical ventilation (IMV) for more than 24 hours and weaned with T-tube ventilation strategies were included in this study. The research cohort was randomly divided into development and validation cohorts at a ratio of 7:3. The development cohort was used to build the predictive model, and the validation cohort was used for validation. Each cohort was further divided into weaning success and weaning failure groups according to the weaning outcome (figure 1).

Figure 1

Flow chart of the study. ICU, intensive care unit; MIMIC-IV, Medical Information Mart for Intensive Care-IV; MP, mechanical power; SBT, spontaneous breathing test.

Data extraction

Data extraction was performed using structured query language with the following analysis variables: (1) basic demographic data (age, sex, body mass index (BMI), smoking history and Sequential Organ Failure Score (SOFA)); (2) time-related data, time to first intubation, the start and end time of mechanical ventilation, the start time of the first SBT, the successful and aborted time of SBT, the time of the first extubation, the time of the second intubation, the time of the first non-invasive ventilation after extubation, the length of ICU stay and the duration of IMV before SBT; (3) combined symptoms, extracting comorbidities (hypertension, diabetes, chronic obstructive pulmonary disease, congestive heart failure, chronic kidney disease, stroke) according to the International Classification Of Diseases (ICD)-9 codes recorded in the MIMIC-IV database; (4) the average value of respiratory mechanics parameters (tidal volume (VT), respiratory rate (RR), peak inspiratory pressure (Ppeak), plateau pressure (Pplat), end-expiration positive pressure (PEEP), minute ventilation, inspired oxygen concentration (FiO2)) 4 hours before the first SBT; (5) laboratory indicators (white blood cell count (WBC), serum creatinine (SCr)) before SBT, and hourly urine output before SBT (urine output rate (Uorate)); and (6) vital signs (heart rate (HR), respiration (breathing frequency (BF)), mean arterial pressure, blood oxygen saturation (SPO2); arterial blood gas analysis during SBT, including PH, arterial oxygen partial pressure (PO2), arterial partial pressure of carbon dioxide (PCO2), oxygenation index (PO2/FiO2, PF)) during SBT.

Calculation of MP

After excluding patients with missing variables required to calculate MP, including patients with missing Pplat (ie, all patients in the study had Pplat measurements in volume control mode before SBT), we extracted data according to the simplified MP equation in the volume-controlled model proposed by Gattinoni et al9 as follows:

MP (J/min)=0.098×VT×RR×(Ppeak–0.5×ΔP).

Where VT represents tidal volume, RR represents respiratory rate, Ppeak represents peak inspiratory pressure and ΔP represents driving pressure.

Driving pressure (∆P) in the ventilation mode was calculated using Pplat and PEEP:

∆P (cmH2O)=Pplat−PEEP.

Where Pplat represents plateau pressure, and PEEP represents end-expiration positive pressure.

Dynamic lung compliance (Cdyn) refers to the change in lung volume caused by a unit pressure change, reflecting the compliance of the overall respiratory system11 and is calculated as follows:

Cdyn (mL/cmH2O)=VT/(Ppeak–PEEP).

Where VT represents tidal volume, Ppeak represents peak inspiratory pressure and PEEP represents end-expiration positive pressure.

Definition of weaning failure

Weaning failure was defined as failure of SBT (ie, premature termination of SBT), or the need for reintubation or non-invasive ventilation within 48 hours of cessation of mechanical ventilation, or death within 48 hours of extubation.13 Early termination of SBT in the MIMIC-IV database was assigned as follows: RR >35 beats/min >5 min; heart rate >140 beats/min; blood pressure >180 or <90 mm Hg; new-onset arrhythmia; pulse oximetry (SpO2) <90% >2 min; with use of accessory respiratory muscles. SBT was discontinued when the clinicians at the bedside observed that the patient’s vital signs exceeded the above indicators. Only patients on T-tube ventilation during weaning were included in this study to reduce the influence and bias of different SBT modalities on weaning outcomes.14

Statistical analysis

Variables with >15% missing data in the study were excluded, and multiple imputation was used to impute variables with <15% missing data to minimise the bias caused by missing values.15 A linear trend test was performed on continuous predictors.16 Variables with non-linear trends in predictors and weaning outcomes were transformed into categorical variables based on the distribution of the independent variables and their clinical significance. Normally distributed measurement data were expressed as the mean±SD, and t-test was used for comparisons between groups. Non-normally distributed measurement data were expressed as the median and IQR and compared using the Mann-Whitney U test or the Kruskal-Wallis H test. Enumeration data were expressed as numbers (percentages), and the χ2 test was used for comparison between groups.

A logistic risk regression model was used to screen important predictors of weaning outcome, and the results were expressed as OR with 95% CI. To limit the variables and increase the practicability of the final model, variables with p<0.05 in the univariate analysis were included in the multivariate regression model for variable screening using the backward method. A nomogram was constructed based on the results of the multivariate analysis, and the discrimination and accuracy of the model were evaluated by receiver operating characteristic curve (ROC) and calibration curve.17 The accuracy of the nomogram, MP and Cdyn in predicting the outcome of weaning failure was further compared by area under the ROC (AUC). Decision curve analysis (DCA) was used to evaluate the clinical validity of the predictive model.

All tests were two-tailed, and p<0.05 was set as the threshold for statistical significance. Data analysis was performed using Stata V.16.0 (StataCorp, Texas, USA) software and R software V.4.1.2 (2021-11-01).18 Graphs were drawn with the R package ‘ggplot 2’ V.3.3.5.19

Patient and public involvement

Patients and/or the public were not directly involved in this study.

Results

Baseline characteristics of the development cohort and validation cohort

By screening data in the MIMIC-IV from 2008 to 2019, we identified 3695 patients with IMV for more than 24 hours who were weaned by T-tube ventilation strategy. This cohort comprised 2274 patients (61.5%) who were successfully weaned and 1421 patients who failed weaning (38.5%) (figure 1). Weaning failure patients included 1138 patients (80.1%) who failed SBT, and 283 patients (19.9%) who were reintubated, received non-invasive ventilation or died 48 hours after weaning. Eligible patients were randomised into a development cohort (n=2586, 70%) and a validation cohort (n=1109, 30%). Table 1 summarises the demographic and clinical baseline characteristics of the different weaning outcome groups in the development and validation cohorts. See detailed comparison of continuous variables between groups in supplementary materials (online supplemental tablea S1 and S2). The baseline characteristics of the development and validation cohorts were balanced.

Table 1

Baseline characteristics of the development and validation cohorts

Prognostic factors in the development cohort

Variables such as basic demographics, and respiratory mechanics, laboratory and clinical parameters in the development cohort were further tested by univariate regression analysis (table 2). BMI, SOFA, RR, PEEP, Pplat, Ppeak, MP, Cdyn, FiO2, WBC, SCr, Uorate, HR, BF, MBP, SPO2, PH, PF, the length of ICU stay and duration of IMV at the first SBT were identified as potential predictors of weaning failure (p<0.05). Incorporating these predictors into a multivariate logistic regression equation showed that PEEP, MP, Cdyn, FiO2, the length of ICU stay and duration of IMV before the first SBT were independent predictors of weaning failure (table 2). Analyses showed that higher PEEP is associated with an increased risk of weaning failure (<5 vs 5–8, ≥8, OR=1.34, 3.52, both p<0.05), and patients with high MP had the highest risk of weaning failure (≤5 vs 5–10, 10–15, ≥15, OR=2.52, 3.90, 4.55, all p<0.001), followed by patients with low Cdyn (≥50 vs 40–50, 30–40, ≤30, OR=3.02, 3.42, 4.44, all p<0.001). The risk of weaning failure in patients with high FiO2 was higher than that in patients with low FiO2 (OR=1.37, p=0.002). Additionally, longer ICU days (<7 vs ≥7, OR=2.43, p<0.001) and IMV duration (<3 vs ≥3, OR=2.33, p<0.001) were associated with a higher risk of higher weaning failure.

Table 2

Univariate and multivariable analyses for the relationship between weaning success and weaning failure in the development cohort

A prognostic nomogram of weaning failure

A predictive model of weaning failure in patients with IMV was constructed based on the six independent predictors identified in the multivariate logistic regression model, and presented as a nomogram (figure 2). As shown in the nomogram, corresponding scores were assigned on the scale according to the OR value of each factor in these variables, with higher OR values corresponding to higher risk scores. The probability of weaning failure is predicted by summing the scores calculated for each prognostic factor in the nomogram. For instance, one patient with IMV with a PEEP of 8 cmH2O (83 points), an MP of 12 J/min (89 points), a Cdyn of 35 mL/cmH2O (81 points), an FiO2 of 45% (24 points), an ICU length of 7 days (59 points) and an IMV duration of 3 days (56 points) had a total score of 392 points, which corresponded to a weaning failure probability of approximately 86% in the nomogram.

Figure 2

Nomogram predicting the probability of weaning failure. Cdyn, dynamic lung compliance; FiO2, inspired oxygen concentration; ICU, intensive care unit; IMV, invasive mechanical ventilation; MP, mechanical power; PEEP, positive end expiratory pressure.

Evaluation of the prognostic nomogram performance

Internal cross-validation of nomograms using the bootstrap method (bootstrap=1000 resampling) in the development cohort. As shown in figure 3A, the calibration plot yielded a straight line with a slope close to 1, indicating that the nomogram was well calibrated for predicting weaning failure. Using ROC curves, we evaluated the effectiveness of the nomogram in predicting weaning failure in both the development and validation cohorts, with an AUC of 0.828 (95% CI: 0.812 to 0.844) for the development cohort and 0.833 (95% CI: 0.809 to 0.857) for the validation cohort (figure 3C and D). In addition, by comparison, the accuracy of the nomogram in the development cohort and the validation cohort in the prediction of weaning failure was significantly higher than that of the single indexes MP and Cdyn (development cohort AUC, 0.828 vs 0.746, 0.692, both p<0.001; validation cohort AUC, 0.833 vs 0.743, 0.682, both p<0.001) (figure 3C and D). Based on DCA, we concluded that the nomogram was clinically valid in the validation cohort (figure 3B).

Figure 3

Evaluation of the prognostic nomogram performance in the development and validation cohort. Calibration plot of the nomogram for the probalility of weaning failure within development (A). Decision curve for treatment failure within validation cohort (B). Area under the receiver operating characteristic curve (95% CI) for nomogram, MP and Cdyn within development (C) and validation cohort (D). Cdyn, dynamic lung compliance; MP, mechanical power.

Discussion

This study is the first to establish and validate a mechanical power-oriented prediction model for weaning outcomes based on a large database. The model visualises six simple and easily obtained variables through a nomogram, and can be used to evaluate the risk of weaning failure before the SBT, thereby assisting clinicians in making decisions related to weaning in critically ill mechanically ventilated patients.

Increased respiratory load and respiratory muscle work resulting from increased airway resistance combined with decreased respiratory system compliance are major causes of weaning failure.5 6 MP integrates multiple factors of mechanical ventilation, and the total energy delivered by the ventilator to the lung parenchyma can be calculated by combining parameters such as VT, PEEP, Pplat, Ppeak and RR.9 20 The measurement of MP is simple and non-invasive, and the workload required to maintain optimal alveolar ventilation acting on the respiratory muscles per unit time can be obtained without disconnecting the ventilator at the bedside; consequently, MP has recently become a new guideline for clinical weaning.10 11 21 The MP-oriented weaning outcome prediction model has certain advantages in the assessment of respiratory load before weaning and provides a comprehensive judgement of weaning decisions combined with clinical feasibility.

Among the 3695 mechanically ventilated patients in this study, 38.5% (1421/3695) failed weaning after the first SBT. Furthermore, 11.07% (283/2557) of patients required reintubation, non-invasive ventilation or died 48 hours after successful SBT weaning, which was consistent with the multicentre observational study by Jaber et al.22 Among a total of 32 variables were assessed in the study, the following 20 key variables related to weaning outcomes were identified through screening: BMI, SOFA score, respiratory mechanics indicators (RR, PEEP, Pplat, Ppeak, MP, Cdyn, FiO2), inflammatory markers (WBC), organ function status (SCr), fluid management (Uorate), physiological status at weaning (HR, BF, MBP, SPO2, PH, PF), the length of ICU stay and duration of IMV (table 2). Our study and previous research shows that higher BMI23 and SOFA score,24 abnormal vital signs,25 acid-base balance,26 degree of infection control,27 organ function and fluid levels and management23 are important predictors of weaning failure risk. However, after incorporating these potential predictors into a multivariate logistic regression model, only six predictors (PEEP, MP, Cdyn, FiO2, the length of ICU stay and duration of IMV) were found to be independently associated with weaning failure. Four of these were respiratory mechanics-related indicators (table 2). These findings suggest that respiratory factors have greater weight in the prediction of weaning outcomes, which is consistent with the results reported by Heunks and van der Hoeven.4 Although reversible factors leading to weaning failure are treated aggressively, objective assessment of airway and lung function is still an important aspect of avoiding weaning failure.

PEEP can prevent lung collapse and reduce intrapulmonary shunting, thereby maintaining alveolar recruitment and increasing arterial oxygenation.28 The lower the level of PEEP required to achieve the therapeutic goal before weaning reflects a lower number of collapsed alveoli and better uniformity of lung ventilation.28 Zhao et al8 also used PEEP as an independent risk indicator for predicting weaning failure. FiO2 levels before weaning reflect the severity of hypoxia, as well as the state of circulatory function and oxygen transport capacity.29 Our results are consistent with those of Yan Jia et al,30 but differ from the findings of Savi et al,31 showing that FiO2 is a better predictor of the risk of weaning failure than PO2/FiO2. This discrepancy may be related to significant influence of FiO2 and PEEP levels on PO2/FiO2.32 PEEP and FiO2 are also important indicators for weaning screening tests.26 In accordance with the findings of Baptistella et al,3 our research supported the conclusion that dynamic lung compliance is a respiratory mechanics parameter that can be used as a predictor of weaning outcome. Cdyn represents the pressure required to generate an appropriate volume to meet physiological needs, reflecting the ease with which the lung undergoes volume change under the action of external force.33 Cdyn is affected by both lung tissue elasticity and airway resistance, with greater lung compliance during weaning associated with lower risk of weaning failure.3 As a comprehensive respiratory mechanics index, MP is a quantitative measure of the energy required to overcome pulmonary resistance and maintain alveolar opening and optimal oxygenation during mechanical ventilation, and can reflect the severity of lung lesions.34 In this study, we found that larger MP values before weaning were associated with a greater energy load that must be overcome by the respiratory muscles during spontaneous breathing, and a higher the risk of weaning failure, which is consistent with the findings of Ghiani et al.11

In accordance with previous studies,3 35 the length of ICU stay and duration of IMV were also independent predictors of weaning failure. With a length of stay in ICU >7 days and duration of IMV >3 days, the OR values of the risk of weaning failure increased to 2.43 (95% CI: 1.96 to 3.02) and 2.33 (95% CI: 1.87 to 2.90), respectively, (both p<0.001) (table 2). This may be related to the increased risk of weaning failure due to prolonged mechanical ventilation and prolonged ICU stays leading to increased risk of diaphragmatic dysfunction, ventilator-related morbidity and mortality.2 36 Although a single index such as MP and Cdyn can predict the weaning outcome to a certain extent, our ROC analysis provided evidence that the nomogram (AUC=0.828) constructed using a combination of parameters is more accurate in predicting weaning failure than a single index, which is consistent with the conclusions reported by Torrini et al.37 In clinical practice, the MP-oriented prediction model constructed by combining the respiratory system parameters and the overall condition of the patient can be used to improve the prediction of weaning failure. Given that there are no identified risk factors with the need for laboratory parameters and all variables in the final model are available at the bedside, the prediction model has better generalisability and simplicity than previous predictive scoring tools (eg, Extubation Predictive Score).3

Several limitations of this study should be pointed out. First, we mainly extracted the data for patients with complete Pplat measurements and MP calculated in volume control mode before SBT. Since this study is a secondary analysis of the data set in the MIMIC-IV for clinical purposes, there is no guarantee that the parameters analysed were collected under standard conditions without spontaneous breathing and adequate levels of sedation. Second, due to database limitations and missing data for some variables, we cannot rule out the possibility that other variables that were not included in our study, such as serological markers B-type natriuretic peptide38 and central veins pressure,39 may also have predictive value for weaning outcomes. In addition, we could not compare the performance of the MP-oriented model with the existing model (eg, the modified Burns Wean Assessment Program scores).40 Finally, although we randomly assigned a validation cohort of 30% of the total sample size to verify the superiority of our model, analysis of a large external cohort will further enhance the credibility and validity of our model.

In conclusion, this study is the first to establish and validate an MP-oriented prediction model for weaning failure based on a database and provides an intuitive and visualisation of the model with a nomogram that predicts weaning failure with good accuracy and clinical validity. The model is simple to use and can be used with ease to provide information with clinical practicability. Moreover, this model can be used by clinicians as a decision support tool in the weaning process.

Data availability statement

Data are available upon reasonable request. The data set analysed to generate the findings for this study are available from the corresponding author on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

The use of the database for research purposes was approved by the Institutional Review Boards of the Massachusetts Institute of Technology and Beth Israel Deaconess Medical Center.

Acknowledgments

We would like to thank all the individuals at the MIT Computational Physiology Laboratory and the Beth Israel Deaconess Medical Centre who designed, built and maintained the Medical Information Mart for Intensive Care (MIMIC)-IV database for open access.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • YX and XL contributed equally.

  • Contributors YY: Study design, data extraction, data analysis and drafting the manuscript. JL and YW: Conceptualisation, writing review and editing. XC: Data analysis and curation. ZD: Literature search and data interpretation. YX: Study design, data analysis, writing review and editing. XL: Conceptualisation, writing review and editing and data curation. All authors read and approved the final manuscript. XL is the guarantor.

  • Funding This study was supported by grants from the Plan Funding Project of Jiangsu Commission of Health (H2019109) and Jiangsu Science and Technology Department Social Development Project (BE2020670).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.