Diaphragm Paralysis After Cardiac Surgery: A Frequent Cause of Post-Operative Respiratory Failure

Diaphragmatic dysfunction (DD) is found in 1.2-60% of patients after cardiac surgery. The aim of this study was to reinvestigate the incidence, risk factors and outcomes of DD with actual cardiac surgery procedures. This is an observational study based on a prospectively collected database in one cardiac surgery centre. The DD group included patients with clinically perceptible diaphragmatic paralysis, which was conrmed by chest ultrasound (amplitude of the diaphragm movement in time-motion mode [TM] at rest, after a sniff test). The primary endpoint was the incidence of DD. The data collected included demographic and medical data; type of cardiac disease; EuroSCORE II; ECG ndings; type of surgery; operating data, including duration of cardiopulmonary bypass (CPB) and aortic cross-clamping (ACC); pre- and post-operative echocardiography; and main complications after surgery, including use of inotropes, sepsis, pneumonia, ischaemia, stroke, reoperation, mediastinitis, length of stay and hospital mortality.


Conclusion
The incidence of symptomatic DD after cardiac surgery was 7.6%, leading to respiratory complications and increased ICU stay. Coronary bypass grafting was the principal factor associated with DD.
DD can be transient after cardiac surgery and can be related to mechanical factors, such as pain, pleural effusion, presence of pleural and mediastinal tubes, decubitus position, and sternotomy, with recovery after a few days [22,23]. It may be more prolonged in connection with partial or complete damage to the phrenic nerve and/or diaphragmatic devascularization [9,24], with more prolonged or de nitive effects [25].
The aim of our study was to investigate the actual incidence, risk factors and outcomes of patients who present with DD after cardiac surgery.

Material & Methods
This observational study was based on the analysis of a prospectively collected database. According to French national regulation, systematic written informed consent was requested for all patients on admission for the secured storage of clinical and para-clinical data obtained during hospitalization and for potential anonymous use in clinical research studies. All adult patients who underwent a cardiac surgery between January 2016 and September 2019 in one teaching hospital were included (Clinique Ambroise Paré, Neuilly-Sur-Seine, France). Operative indications for surgery agreed with the current European guidelines during the study. Two groups were de ned. The DD group included all patients with clinically perceptible post-operative diaphragmatic paralysis. The diagnosis of DD was suspected clinically and/or radiologically and con rmed by chest ultrasound. The diagnosis of DD was considered con rmed when a decrease in the TM amplitude of the movement of the diaphragm was found below the lower limit of normal for at least one of the following measurements: 1) at rest, < 9 mm for women and < 10 mm for men; 2) after a sniff test, < 16 mm for women and < 18 mm for men; and 3) after a deep inspiration, < 37 mm for women and < 47 mm for men [21]. The control group included all other patients.
The data collected included demographic and medical data; type of cardiac disease; EuroSCORE II; ECG ndings; type of surgery; operating data, including duration of cardiopulmonary bypass (CPB) and aortic cross-clamping (ACC); pre-and post-operative echocardiography; and main complications after surgery, including use of inotropes, sepsis, pneumonia, ischaemia, stroke, reoperation, mediastinitis, length of stay and hospital mortality.
When DD was con rmed, patients were treated according to our local protocol, including half-sitting position, reinforced muscular and respiratory kinesitherapy, intermittent non-invasive ventilation, nocturnal continuous positive airway pressure therapy, and bowel prokinetic drugs, if necessary, using trimebutine 100 mg/24 h + erythromycin 375 mg/24 h + macrogol 26 g/ 24 h. If no transit was obtained after 48 h, neostigmine 1.5 mg/24 h subcutaneously was added.
The primary endpoint was the incidence of DD. Secondary outcomes were as follows: 1) the search of DD predictive factors among the collected data and 2) the main complication associated with DD as compared to the control group.

Statistical analysis
Shapiro-Wilk tests were used to test the normality of distribution of the studied variables. Continuous variables were expressed as mean ± standard deviation (SD) when distributed normally or median [interquartile range] when not. Categorical variables were expressed as a number (percentage).
Categorical variables in two groups were compared using Chi-squared and Pearson's test. Comparison of variables normally distributed used Student's t test. Other continuous variables were compared by the Mann-Whitney U nonparametric test.
A multivariate analysis was conducted by a logistic regression, including comparisons with p ≤ 0.1 in univariate analyses. A constant was added. All predictors were considered clinically meaningful, and none was rejected. The signi cance threshold adopted was p < 0.05. The software used was SPSS 25.0®, IBM®, Armonk United States.

Results
During the studied period, 3577 patients underwent cardiac surgery. Among them, 272 had DD. Baseline pre-operative characteristics of the whole cohort are presented in Table 1. Patients in the DD group had more arterial hypertension and higher body mass index (BMI) compared to the control group; other baseline characteristics were similar.   Predictors of DD in univariate analysis are shown in Table 4. In multivariate analysis (Table 5)

Discussion
Despite recent advances in cardiac surgery procedures, we found a 7.6% postoperative incidence of DD. To our knowledge, this is the largest study of DD after cardiac surgery. Not surprisingly, we found that DD was associated with respiratory complications and prolonged ICU and hospital stays. Half of the patients with DD required non-invasive ventilation and one-quarter developed pneumonia. DD was also the cause of 20% of all reintubations and of 50% of all tracheotomies. However, the early diagnosis and systematic therapeutic protocol allowed a low in-hospital mortality rate (0.7%) that was not different from other patients.
Indeed, DD must be treated speci cally. Although spontaneous recovery can be observed within 96 h after extubation [1,3,4], complications may have started during this time delay. Non-invasive ventilation may help to improve respiratory capacity after extubation. Some authors have suggested to keep the patient under mechanic ventilation until recuperation of diaphragm function, even if a tracheotomy needs to be used [1,3,4]. Our local protocol, applied to the patients in this study, gave priority to early extubation; bringing the patient on the armchair as soon as possible with intense respiratory physiotherapy to reeducate the diaphragm; and systematic use of non-invasive ventilation with positive pressure, including continuous positive airway pressure during sleep, to maintain residual functional capacity as close as possible to normal. Tracheostomy was restricted to patients with persistent DD with respiratory failure needing reintubation. Although our mortality was low, this was not an objective of our protocol and controlled studies are needed to evaluate this strategy.
Respiratory complications with or without pneumonia are frequent (2-6%) after cardiac surgery, leading to an increase of hospital stay and mortality [1,3,4]. DD is probably an underestimated cause of respiratory failure through other complications, such as atelectasis and pneumonia.
Our study con rmed the link between DD, hypertension, BMI, and coronary bypass grafting but not with diabetes, as previously observed. The role of diabetes in this pathology is still unclear [14,24]. DD can be transiently due to many factors, including pain, hyperglycaemia, sepsis, neuromyopathy, electrolyte disorders (such as hypophosphoraemia), mechanical harm by tubes, and pleural or pericardial effusion. Prolonged DD is most often the consequence of a phrenic nerve injury but may be caused by multiple factors. The use of ice slush for myocardial protection during CPB was a well-known cause of phrenic nerve injury in the past, but it is no longer in use [8,26]. It is uncertain whether phrenic nerve injury is primarily related to direct surgical injury or to ligation of its blood supply during internal mammary artery (IMA) dissection, especially when using electric cautery. Stretching of the phrenic nerves during chest opening may also be a possible mechanism and may explain why BMI, which supposes higher strength to open the chest, was associated with DD in our study and in others [10,27]. The anatomic relationship between the phrenic nerve and the IMA is inconstant, and thus, caution must be taken when dissecting this artery. Furthermore, the blood supply to the phrenic nerve comes from the pericardiophrenic artery, which is a branch usually originating from the upper 1 to 3 cm of the IMA. In the 1990s, when the IMA was harvested, the reported incidence of DD was very high, ranging from 42-69% [8,28,29], but it decreased with the knowledge of this pathology, improvement in surgical techniques and increase in surgeon's skills. Almost three-quarters (72.3%) of patients with DD were found after coronary bypass grafting, with quite a perfect parallelism between the side of the IMA harvesting and the side of the DD (100% for left DD and 94% for right DD). In the same way, the higher post-operative creatine kinase peak in the DD group, without differences in the troponin peak, may cause ischaemia of phrenic muscle after IMA harvesting. Nevertheless, the last quarter of our patients with DD (25.7%) were found after valvular surgery, con rming that DD does not have a unique mechanism.
Our study has several limitations. First, being a monocentric study, the results correspond to a speci c experience. The retrospective nature of the analysis brings only a low risk of bias since the database was prospectively collected on a registry. However, this prevents the assertion of the causal link of the observed associations. The DD incidence we studied was the incidence of clinically perceptible DD. A systematic analysis of DD will probably nd a higher incidence. Furthermore, we did not report the link between the side of DD paralysis and the side of internal jugular vein catheterization, systematically placed before surgery, which could be another risk factor for phrenic nerve injury. Last, we only studied intra-hospital events without a long-term follow-up. Data from long-term outcomes would be interesting, especially to analyse the potential recovery.

Conclusion
Symptomatic DD was found in 7.6% of patients after cardiac surgery. It leads to an increase of respiratory complications, such as pneumonia and prolonged ventilation and ICU stay. Coronary bypass grafting with IMA harvesting was the principal factor associated with DD, along with obesity and systemic hypertension in multivariate analysis. Early diagnosis and appropriate support can probably minimize its consequences. -Consent for publication: Not applicable.
-Availability of supporting data: The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.
-Competing interests: The authors declare that they have no con ict of interest.
-Funding: The authors received no speci c funding for this work -Authors' contributions: D.L, P.E, and P.S contributed to the design and implementation of the research. P.M.L, R.M, A.S and A.B analyzed and interpreted data. DL, MPL, PS were major contributors in writing the manuscript. All authors read and approved the nal manuscript.
-Acknowledgements: Thank you to Emma Taylor from Proof Reading Services for her proofreading of this paper.