INTRODUCTION

Diaphragm elevation caused by diaphragm dysfunction due to phrenic nerve injury is a well-recognized complication after cardiac surgery, with a reported incidence ranging from 1.2% to 60%^[1]. Several technical risk factors associated with this phenomenon include internal mammary artery harvesting and cold injury of the phrenic nerve due to intrapericardial application of topical ice slush for myocardial protection^{[1, 2, 3, 4, 5]}. Additionally, a higher incidence of diaphragm elevation is found in patients with chronic obstructive pulmonary disease (COPD) and/or diabetes mellitus^{[6, 7]}. Diaphragm dysfunction can lead to adverse postoperative outcomes such as the need for prolonged mechanical ventilation^[8] atelectasis, and recurrent pneumonia^[9], as well as increased intensive care unit and hospital stay, morbidity, and mortality^[10].

However, reports on the incidence of diaphragm dysfunction and its consequences during recovery after cardiac surgery remain historical^[8]. The aim of this study is to provide contemporary insights on the incidence of diaphragm dysfunction in patients undergoing cardiac surgery, its effect on postoperative outcomes, and the potential recovery of phrenic nerve injury during follow-up.

METHODS

A retrospective cohort study was performed considering all patients who underwent cardiac surgery through sternotomy at the Radboud University Medical Centre (or Radboudumc) in Nijmegen, the Netherlands, between January 2015 and December 2016. Patients were excluded when death occurred during surgery, preoperative imaging was missing, postoperative imaging was missing, or pre and postoperative imaging could not be judged adequately (due to pleural effusion or atelectasis). This retrospective study was approved by the institutional review board (file number 2020-6728); no individual patient consent was required.

Data were obtained from digital patient charts and hospital registries and included detailed patient-, surgery-, and postoperative outcome-related information. The principal data used for the current analysis was based on the standardized Dutch National database of cardiac surgery (Begeleidingscommissie Hartinterventies Nederland [or BHN], a supervisory committee for heart interventions in the Netherlands) in which postoperative outcome parameters are prospectively being collected by the Department of Cardiothoracic Surgery.

In addition, diaphragm position and potential diaphragm elevation were evaluated on chest radiography (CR) at certain timepoints. Namely, the latest CR prior to surgery and the latest eligible CR prior to discharge but within a month after surgery. Right-sided diaphragm elevation was defined as the right diaphragm being > 3.0 cm above the left diaphragm^[11]. Left-sided diaphragm elevation was defined as < 0.5 cm below or above the level of the right diaphragm. When available, follow-up imaging was evaluated as well to determine the occurrence of recovery in case of diaphragm elevation. In case of multiple follow-up images, the latest one was used for review. Possible follow-up outcomes were recovered elevation, persistent postoperative elevation, new elevation, or still no elevation present.

If no CR was available, other types of imaging were used when possible (e. g, computed tomography- or magnetic resonance imaging-scan). All CR were evaluated by the two main authors (SI, TS) independently. Disagreement was resolved by consensus, or after consultation with the final author (WWLL).

Patients were divided into three groups for statistical analyses: group A — pre-existing (hemi)diaphragm elevation, group B — new (hemi)diaphragm elevation, and group C — no (hemi) diaphragm elevation.

Primary Endpoints

The primary endpoints were pulmonary complications (defined as pneumothorax with or without treatment, pleural effusion requiring drainage, pulmonary embolism, exacerbation of COPD, and/or special ventilatory requirements [ventilation in prone position]), in-hospital mortality, and recovery of new diaphragm elevation. Pulmonary complications combined with the need for reintubation formed our primary composite endpoint (composite 1).

Secondary and Additional Endpoints

Secondary outcomes included the composite endpoint of pulmonary complications, reintubation, and in-hospital mortality (composite 2) and the composite endpoint of any form of complication (cardiac, pulmonary, renal, infectious, neurological complication, or reintubation) and/or in-hospital mortality after surgery (composite 3). Furthermore, when follow-up data was available, recovery of postoperative diaphragm elevations was evaluated.

Statistical Analyses

All data was entered into an electronic database, Castor Electronic Data Capture (Castor EDC, Ciwit B.V., Amsterdam, the Netherlands), according to institutional regulations. Statistical analyses were performed using the software IBM SPSS Statistics for Windows, version 25.0 (Released 2017), Armonk, NY: IBM Corp. Continuous data are presented using means (± standard deviation). Categorical variables are presented with counts and percentages. Continuous data analysis was performed using the independent samples t-test, and categorical variables were compared using the chisquared test.

Multivariate logistic regression analysis was performed to determine whether change in diaphragm height difference is predictive of complications, in particular pulmonary complications. This change in diaphragm height was separated in groups of 2 cm, ranging from 0 to > 4.0 cm. Composite endpoints were formed to analyze the relationship between diaphragm elevation, respiratory complications, and mortality as previously defined. Multivariate logistic regression analysis was also performed to determine whether the presence of diaphragm elevation is predictive of complications. Interrater variability was compared using Pearson’s correlation coefficient. A P-value of < 0.05 was considered statistically significant.

RESULTS

A total of 1510 patients have undergone cardiac surgery through sternotomy during the study period (Figure 1). Of these, 12.8% (n = 194) were excluded, either due to missing preoperative imaging (n = 40), indistinct preoperative imaging (n = 18), or indistinct postoperative imaging (due to pleural effusion or atelectasis) (n = 136). This resulted in a total of 1316 patients included in the final analysis (Figure 1).

Fig. 1
Research setting on diaphragm elevation after cardiac surgery. A flow diagram of the study depicting the number of excluded patients with their respected reasons and the three groups with pre-existent, new, and no (hemi) diaphragm elevation. Radboudumc=Radboud University Medical Centre.

Of these 1316 patients, 13% (n = 179) had pre-existing diaphragm elevation (group A), 27% (n = 351) had a new diaphragm elevation postoperatively (group B), and 60% (n = 786) had no diaphragm elevation (group C). Of the patients with new postoperative diaphragm elevation, 64% (n = 223) had a left-sided elevation and 36% (n = 128) had right-sided elevation (P < 0.001). None of the patients had bilateral diaphragm elevation.

Preoperative Demographic and Clinical Data

Baseline characteristics of patients for all three groups are presented in Table 1. For the total group, 74% were male, with a mean age of 65.87 ± 10.43 years. Patients in group A were significantly older than patients from groups B and C (A vs. B P = 0.008 and A vs. C P = 0.021). Additionally, patients from group A had significantly higher body mass index (BMI) when compared with those from group B (27.64 ± 3.94 vs. 26.83 ± 3.85, P = 0.023). Concerning preoperative comorbidities, more patients with diabetes were found in group C compared to group B (18% vs. 23%, P = 0.049). No other statistical differences were found in the baseline characteristics and clinical history.

Table 1

Patients’ characteristics.

Baseline characteristics of all study patients and separated for the three groups with pre-existing (hemi)diaphragm elevation (group A), new (hemi)diaphragm elevation (group B), and no (hemi)diaphragm elevation (group C). Results are depicted as mean ± standard deviation or as absolute numbers with percentages. P-values < 0.05 are deemed statistically significant BMI=body mass index; CABG=coronary artery bypass grafting; OPCAB=off-pump coronary artery bypass grafting

Type of Surgery

Regarding the effect of the type of cardiac surgery on the incidence of new diaphragm elevation postoperatively (Table 2), we found that patients were most often affected after aortic surgery (37%), however this difference was not statistically significant. No statistically significant differences were observed between the other different types of surgery, nor between cases using left internal mammary artery (LIMA) and/or right internal mammary artery (RIMA) in the coronary artery bypass grafting (CABG) or off-pump coronary artery bypass grafting.

Table 2

The incidence of diaphragm elevation within each type of surgery.

Absolute number of pre-existing, new, or no (hemi)diaphragm elevation with percentages separated for each type of surgery performed. In case of CABG, the use of LIMA or RIMA is shown for all three groups CABG=coronary artery bypass grafting; LIMA=left internal mammary artery; OPCAB=off-pump coronary artery bypass grafting; RIMA=right internal mammary artery

Postoperative Outcomes and Follow-up

As seen in Table 3, infectious and cardiac complications occurred most frequently (7-8%), whereas renal complications and hospital mortality occurred the least (1-2%). No statistically significant differences were found between the groups. Analysis on hospital mortality found no relationship between the groups and outcome. There were no statistically significant differences regarding the composite endpoints between the three groups (Table 3).

Table 3

Postoperative outcomes.

The postoperative outcomes and complications for all study patients and separated for the three groups with pre-existing (hemi) diaphragm elevation (group A), new (hemi)diaphragm elevation (group B), and no (hemi)diaphragm elevation (group C). Results are depicted as mean ± standard deviation or as absolute numbers with percentages. Composite endpoint 1 = pulmonary complication + reintubation; composite endpoint 2 = pulmonary complication + reintubation + hospital mortality; composite endpoint 3 = all complications together. P-values < 0.05 are deemed statistically significant *one patient who also had a pneumothorax requiring drain; ^#one patient who also had pleural drainage; ^!one patient who also had urinary tract infection; ^$one patient who also had pneumonia AKI=acute kidney injury; COPD=chronic obstructive pulmonary disease; CVA=cerebrovascular accident; CVVH=continuous veno-venous hemofiltration; ECMO=extracorporeal life support; HAP=hospital-acquired pneumonia; ICU=intensive care unit; TIA=transient ischemic attack; UTI=urinary tract infection

In multivariate analysis (Table 4), neither newly developed diaphragm elevation (odds ratio [OR] 1.087, 95% confidence interval [CI] 0.622-1.901, P = 0.769 and OR 1.046, 95% CI 0.607-1.803, P = 0.871, for composites 1 and 2, respectively) nor postoperative diaphragm elevation in centimeters (OR 1.539, 95% CI 0.841-2.817, P = 0.162 and OR 1.062, 95% CI 0.356-3.169, P = 0.914; OR 1.344, 95% CI 0.739-2.445, P = 0.332 and 0.854 95% CI 0.287-2.534, P = 0.775, respectively 2-4 cm and > 4 cm for composites 1 and 2) were significant predictors for composite endpoint 1, or for composite endpoint 2.

Table 4

Logistic regression analyses.

Logistic regression analyses to determine the association between change in diaphragm height (top panel) and the presence of diaphragm elevation (bottom panel) for pulmonary complication and reintubation (composite 1) and pulmonary complication with reintubation and mortality (composite 2). P-values < 0.05 are deemed statistically significant BMI=body mass index; CABG=coronary artery bypass grafting; CI=confidence interval; OPCAB=off-pump coronary artery bypass grafting

Almost a third of all patients had follow-up imaging available (n = 404), of which the results are shown in Table 5. The median follow-up time is 17.5 months (range 0 - 58 months). Interestingly, as seen in Table 6, 65% of patients who developed new diaphragm elevation postoperatively recovered in the follow-up period. Of all surgical interventions, patients who underwent aortic surgery most often had imaging available in the follow-up period (93% vs. < 50% for all other interventions).

Table 5

Follow-up (patient) characteristics.

Characteristics of all study patients separated for presence or absence of follow-up data. Results are depicted as mean ± standard deviation, as absolute numbers with percentages or as median with range. P-values < 0.05 are deemed statistically significant BMI=body mass index; CABG=coronary artery bypass grafting; ICU=intensive care unit; OPCAB=off-pump coronary artery bypass grafting

Table 6

Follow-up for group B (new elevation).

Characteristics of patients that were assigned to the group with new postoperative (hemi)diaphragm elevation (group B) of whom follow-up data was present. Results are depicted as absolute numbers with percentages or a median with a range

Interrater Reliability

For 240 randomly chosen patients, pre, postoperative, and, when available, follow-up diaphragm distances have been measured by two observers. This has resulted in 542 pairs of data. The mean difference between these measurements was 0.0304 cm (limits of agreement = 0.030 +/- 1.96 × 0.123). Using Pearson’s r test, a correlation coefficient of 0.939 (P < 0.001) was found. This indicates a very strong level of agreement between both observers.

DISCUSSION

Elevation of the diaphragm is a known complication after cardiothoracic surgery. This study presents that around a quarter of all patients (27%) suffers from new postoperative diaphragm elevation, regardless the type of surgery, although the incidence is seen after aortic surgery. No significant differences were found in postoperative outcomes. Neither was there a positive correlation between the level of diaphragm shift relative to each other and any of the composite endpoints. Strikingly, almost three quarters of all patients who develop diaphragm elevation post cardiac surgery recovered from this elevation in the months after discharge. However, when comparing both the group with follow-up data and the group without it, it is found that the difference in demographic data is statistically significant, meaning that the group with follow-up data might not be representative for the entire population. This could be clarified by the fact that patients undergoing aortic surgery have a much more stringent follow-up protocol including imaging compared to the patients undergoing standard CABG or valve surgery. The aortic surgery patients were also most affected, although not significantly, by new diaphragm elevation.

As mentioned before, the reported incidence of diaphragm elevation ranges between 2% and 60%^{[2, 10, 12, 13, 14]}. Most of these studies present historical data, with the most recent articles on diaphragmatic dysfunction after cardiac surgery with significant numbers published between 1994 and 2001^{[8, 15, 16]}. Our study, using contemporary data, reports an incidence of new postoperative diaphragm elevation of 27%, meaning that one in four patients undergoing cardiac surgery will suffer from (often temporary) diaphragm elevation. This high number is a consequence of the way diaphragm dysfunction was evaluated. As stated by Chetta et al.^[17], CR is not suitable for predicting diaphragm function, but it is a great tool for determining the height of the diaphragm and the corresponding diaphragmatic height index^[18]. As such, it is still the most commonly used first step in diagnosing possible diaphragm dysfunction^[19].

The number of patients with (new) diaphragm elevation is of course dependent on the cutoff value used to define diaphragm elevation. Many agree upon the left diaphragm elevation if it is at the level of the right or above, however for the right hemidiaphragm this is less distinct. This study used 3 cm as cutoff point, as this is commonly described in the literature^{[11, 20, 21]}. However, other studies suggest lower values, which would further increase the incidence numbers of diaphragm elevation and could be an explanation for the wide range in the previously described incidence numbers^[19]. Previous studies mainly focus on the presence of diaphragm elevation using different definitions^{[1, 2, 17, 18, 19]}. However, most of the new diaphragm elevations were minor changes (< 2 cm), which could explain why no significant difference was seen between the pre-existent and no elevation group.

Most new diaphragm elevations were seen in patients undergoing aortic surgery. This could be related to the use of topical ice slush in our centre. Multiple studies have already shown that the use of ice slush for topical hypothermia in cardiac surgery is associated with diaphragm paralysis^{[5, 22]}. Since it was shown to have no additional benefit, topical ice should be discouraged^[23]. In most cases, this is a transient paresis^[24], as described in our study with a 65% recovery rate of patients with new postoperative diaphragm elevation. Therefore, in case of new diaphragm dysfunction after cardiac surgery, either “wait and see” or intensive physiotherapy can be initiated in the early postoperative phase. More definite injury to the phrenic nerve is seen in CABG when using the cautery in close proximity to the nerve during harvesting of LIMA and/or RIMA^{[2, 25]}. Diaphragm plication is the proposed therapy in this setting of persistent diaphragm elevation combined with significant complaints of pulmonary deterioration^[19]. Although early spontaneous recovery is rare, previous studies confirm spontaneous recovery in over half of patients on the long term^{[2, 26, 27]}.

Limitations

A limitation of this study was the use of radiographic technique. Atelectasis or pleural effusion complicated the analysis of exact diaphragm heights resulting in certain patients being excluded, even though a suitable number of patients remained to be included for analyses. Also, in case of diaphragm elevation, this was not confirmed by functional imaging (ultrasound or fluoroscopy). However, the retrospective character of this study and the limited availability of functional imaging hampered more detailed evaluation. Another issue is the loss to follow-up. As this study was performed in a tertiary referral centre, many patients had follow-up in another hospital.

CONCLUSION

Diaphragm elevation is a complication that occurs frequently after cardiac surgery. However no significant correlation was found between diaphragm elevation, the distance in diaphragm height, and the outcomes after surgery. In most cases, the elevation recovers spontaneously. Future directions should focus on a larger number of patients with longer follow-up and functional testing as well as the consideration of slushed ice and use of cautery in CABG as a risk factor to explore amendments required for clinical practice.

REFERENCES

Aguirre VJ, Sinha P, Zimmet A, Lee GA, Kwa L, Rosenfeldt F. Phrenic nerve injury during cardiac surgery: mechanisms, management and prevention. Heart Lung Circ. 2013;22(11):895-902. doi:10.1016/j.hlc.2013.06.010.

Deng Y, Byth K, Paterson HS. Phrenic nerve injury associated with high free right internal mammary artery harvesting. Ann Thorac Surg. 2003;76(2):459-63. doi:10.1016/s0003-4975(03)00511-3.

Curtis JJ, Nawarawong W, Walls JT, Schmaltz RA, Boley T, Madsen R, et al. Elevated hemidiaphragm after cardiac operations: incidence, prognosis, and relationship to the use of topical ice slush. Ann Thorac Surg. 1989;48(6):764-8. doi:10.1016/0003-4975(89)90667-x.

Canbaz S, Turgut N, Halici U, Balci K, Ege T, Duran E. Electrophysiological evaluation of phrenic nerve injury during cardiac surgery--a prospective, controlled, clinical study. BMC Surg. 2004;4:2. doi:10.1186/1471-2482-4-2.

Efthimiou J, Butler J, Woodham C, Benson MK, Westaby S. Diaphragm paralysis following cardiac surgery: role ofphrenic nerve cold injury. Ann Thorac Surg. 1991;52(4):1005-8. doi:10.1016/0003-4975(91)91268-z.

Tripp HF, Bolton JW. Phrenic nerve injury following cardiac surgery: a review. J Card Surg. 1998;13(3):218-23. doi:10.1111/j.1540-8191.1998.tb01265.x.

Yamazaki K, Kato H, Tsujimoto S, Kitamura R. Diabetes mellitus, internal thoracic artery grafting, and risk of an elevated hemidiaphragm after coronary artery bypass surgery. J Cardiothorac Vasc Anesth. 1994;8(4):437-40. doi:10.1016/1053-0770(94)90284-4.

Chang CZ, Chen YF, Lin HR, Lin PY, Chiu CC, Lin YT. Postoperative diaphragmatic dysfunction in patients undergoing open-heart surgery. Gaoxiong Yi Xue Ke Xue Za Zhi. 1995;11(11):626-32.

Diehl JL, Lofaso F, Deleuze P, Similowski T, Lemaire F, Brochard L. Clinically relevant diaphragmatic dysfunction after cardiac operations. J Thorac Cardiovasc Surg. 1994;107(2):487-98.

Mehta Y, Vats M, Singh A, Trehan N. Incidence and management of diaphragmatic palsy in patients after cardiac surgery. Indian J Crit Care Med. 2008;12(3):91-5. doi:10.4103/0972-5229.43676.

Gaillard F, Bell D, Luijkx T, et al. Normal position of hemidiaphragms on chest radiography. Reference article, https://Radiopaedia.org [updated 02-02-2021; cited 2021. Available from: https://doi.org/10.53347/rID-9138.

Deng Y, Sun Z, Ma J, Paterson HS. Semi-skeletonized internal mammary grafts and phrenic nerve injury: cause-and-effect analysis. J Huazhong Univ Sci Technolog Med Sci. 2006;26(4):455-9. doi:10.1007/s11596-006-0420-z.

Dimopoulou I, Daganou M, Dafni U, Karakatsani A, Khoury M, Geroulanos S, et al. Phrenic nerve dysfunction after cardiac operations: electrophysiologic evaluation of risk factors. Chest. 1998;113(1):8-14. doi:10.1378/chest.113.1.8.

Wheeler WE, Rubis LJ, Jones CW, Harrah JD. Etiology and prevention of topical cardiac hypothermia-induced phrenic nerve injury and left lower lobe atelectasis during cardiac surgery. Chest. 1985;88(5):680-3. doi:10.1378/chest.88.5.680.

Ando M, Endo M, Nishida H, Tomioka H, Sodenaga Y, Koyanagi H. [Phrenic nerve injury following coronary artery bypass surgery: its clinical implication and management]. Rinsho Kyobu Geka. 1994;14(6):501-5. Japanese.

Sarnowski W, Kulesza J, Ponizyήski A, Dyszkiewicz W. Uniesienie przepony po operacjach kardiochirurgicznych. Pol Merkur Lekarski. 2001;10(55):24-6. Polish.

Chetta A, Rehman AK, Moxham J, Carr DH, Polkey MI. Chest radiography cannot predict diaphragm function. Respir Med. 2005;99(1):39-44. doi:10.1016/j.rmed.2004.04.016.

Pornrattanamaneewong C, Limthongthang R, Vathana T, Kaewpornsawan K, Songcharoen P, Wongtrakul S. Diaphragmatic height index: new diagnostic test for phrenic nerve dysfunction. J Neurosurg. 2012;117(5):890-6. doi:10.3171/2012.8.JNS111734.

Dubé BP, Dres M. Diaphragm dysfunction: diagnostic approaches and management strategies. J Clin Med. 2016;5(12):113. doi:10.3390/jcm5120113.

Debnath J, Sreedhar CM, Mukherjee R, Maurya VK, Patrikar S, Reddy YRN. Revisiting anatomical variants on screening chest radiographs in Indian adolescents: a cross sectional observational pilot study. Med J Armed Forces India. 2018;74(4):337-45. doi:10.1016/j.mjafi.2017.07.010.

Puddy E, Hill C. Interpretation of the chest radiograph. Continuing Education in Anaesthesia Critical Care & Pain. 2007;7(3):71-5. doi:10.1093/bjaceaccp/mkm014.

Oliemy A, Singh S, Butler J. Role of topical application of iced slush in the development of phrenic nerve palsy after cardiac surgery. J Cardiothorac Surg. 2015;10(1):A8. doi:10.1186/1749-8090-10-S1-A8.

Alassar A, Bazerbashi S, Moawad N, Marchbank A. What is the value of topical cooling as an adjunct to myocardial protection? Interact Cardiovasc Thorac Surg. 2014;19(5):856-60. doi:10.1093/icvts/ivu259.

Rousou JA, Parker T, Engelman RM, Breyer RH. Phrenic nerve paresis associated with the use of iced slush and the cooling jacket for topical hypothermia. J Thorac Cardiovasc Surg. 1985;89(6):921-5.

O’Brien JW, Johnson SH, VanSteyn SJ, Craig DM, Sharpe RE, Mauney MC, et al. Effects of internal mammary artery dissection on phrenic nerve perfusion and function. Ann Thorac Surg. 1991;52(2):182-8. doi:10.1016/0003-4975(91)91334-r.

Smith BM, Ezeokoli NJ, Kipps AK, Azakie A, Meadows JJ. Course, predictors of diaphragm recovery after phrenic nerve injury during pediatric cardiac surgery. Ann Thorac Surg. 2013;96(3):938-42. doi:10.1016/j.athoracsur.2013.05.057.

de Leeuw M, Williams JM, Freedom RM, Williams WG, Shemie SD, McCrindle BW. Impact of diaphragmatic paralysis after cardiothoracic surgery in children. J Thorac Cardiovasc Surg. 1999;118(3):510-7. doi:10.1016/S0022-5223(99)70190-X.

Glossary

AKI: Acute kidney injury

BMI: Body mass index

CABG: Coronary artery bypass grafting

CI: Confidence interval

COPD: Chronic obstructive pulmonary disease

CR: Chest radiography

CVA: Cerebrovascular accident

CVVH: Continuous veno-venous hemofiltration

ECMO: Extracorporeal life support

HAP: Hospital-acquired pneumonia

ICU: Intensive care unit

LIMA: Left internal mammary artery

OPCAB: Off-pump coronary artery bypass grafting

OR: Odds ratio

RIMA: Right internal mammary artery

TIA: Transient ischemic attack

UTI: Urinary tract infection

Author notes

Correspondence Address:Tim Somers Department of Cardiothoracic Surgery, Radboud University Medical Centre Geert Grooteplein Zuid 10 (Route 615), Nijmegen, Netherlands, Zip Code: 6525 GA, E-mail: tim.somers@radboudumc.nl