See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/279448149
A Novel Risk Score to Predict Dysphagia After Cardiac Surgery Procedures ARTICLE in THE ANNALS OF THORACIC SURGERY · JUNE 2015 Impact Factor: 3.85 · DOI: 10.1016/j.athoracsur.2015.03.077
CITATION
READS
1
17
11 AUTHORS, INCLUDING: Joshua Collin Grimm
J. Trent Magruder
Johns Hopkins Medicine
Johns Hopkins Medicine
35 PUBLICATIONS 38 CITATIONS
33 PUBLICATIONS 114 CITATIONS
SEE PROFILE
SEE PROFILE
Alicia K. Vose University of Florida 4 PUBLICATIONS 3 CITATIONS SEE PROFILE
Available from: Joshua Collin Grimm Retrieved on: 09 February 2016
A Novel Risk Score to Predict Dysphagia After Cardiac Surgery Procedures Joshua C. Grimm, MD, J. Trent Magruder, MD, Rika Ohkuma, MD, Samuel P. Dungan, BA, Andrea Hayes, MS, CCC-SLP, Alicia K. Vose, MA, CCC-SLP, Megan Orlando, BS, Marc S. Sussman, MD, Duke E. Cameron, MD, and Glenn J. R. Whitman, MD Division of Cardiac Surgery and Speech-Language Pathology, Physical Medicine and Rehabilitation, The Johns Hopkins Hospital, Baltimore, Maryland
Background. Although the exact cause of dysphagia after cardiac operations is unknown, timely diagnosis is critical to avoid a devastating aspiration event. Accordingly, we sought to generate the risk of dysphagia in cardiac surgery (RODICS) score to identify patients at risk for its development after heart surgery. Methods. All adult heart surgery patients at our institution between January 2011 and March 2012 were analyzed. A videofluoroscopic swallow study stratified patients into two groups based on the presence or absence of dysphagia. Covariates (p < 0.20) were included in a multivariable model to determine the strongest independent predictors of postoperative dysphagia. Based on the relative odds ratios of significant variables, the RODICS score was generated. Risk cohorts were then created based on easily applicable, whole-integer score cutoffs. Results. During the study period, 115 of 1,314 patients (8.8%) undergoing heart surgery were diagnosed with clinically significant dysphagia. The 38-point RODICS
score comprises seven patient-specific characteristics and perioperative factors. The low risk (less than 4), intermediate risk (5 to 9), and high risk (more than 9) cohorts had postoperative dysphagia rates of 3.0%, 6.8%, and 21.6%, respectively (p < 0.001). The intermediate-risk cohort (odds ratio 2.3, 95% confidence interval: 1.33 to 4.27, p [ 0.01) and high-risk cohort (odds ratio 8.9, 95% confidence interval: 5.22 to 15.32, p < 0.001) were at significantly higher risk of dysphagia developing. The RODICS score demonstrated excellent discriminatory ability (area under the curve 0.75). Conclusions. The incidence and impact of dysphagia after open cardiac operations is significant. This novel scoring system could lead to prompt identification of patients at high risk for postoperative dysphagia and potentially minimize the complications of aspiration.
D
normal physiologic process of swallowing, few studies have categorized the type of postoperative dysphagia based on radiographic imaging. Accordingly, we sought to generate a risk of dysphagia in cardiac surgery (RODICS) score to identify patients at high risk for postoperative dysphagia, and also characterize the specific pathophysiologic processes responsible for its development.
ysphagia after cardiac surgery can result in clinically important aspiration and, if unrecognized, increase the risk of postoperative pneumonia [1–3]. Although previous studies have identified risk factors for postoperative dysphagia, most were conducted before the routine use of transesophageal echocardiogram (TEE) and, therefore, focused mainly on the impact of this intraoperative adjunct on postoperative dysphagia rates [2–5]. Currently, TEE is used in most heart surgery operations. Therefore, a better understanding of the influence of other patient-specific characteristics and perioperative factors could aid in the identification of high-risk patients, and might prevent aspiration events through early consultation with a speech-language pathologist (SLP) to manage dysphagia symptoms and expedite initiation of enteral nutrition. Furthermore, although the mechanisms of dysphagia are known to involve a derangement in the
Accepted for publication March 23, 2015. Address correspondence to Dr Whitman, Division of Cardiac Surgery, The Johns Hopkins Hospital, 1800 Orleans St, Zayed Tower 7107, Baltimore, MD 21287; e-mail:
[email protected].
Ó 2015 by The Society of Thoracic Surgeons Published by Elsevier
(Ann Thorac Surg 2015;-:-–-) Ó 2015 by The Society of Thoracic Surgeons
Material and Methods Approval for this study was obtained from The Johns Hopkins Institutional Review Board (00001068).
Patient Population All adult patients (aged 18 years or more) undergoing heart surgery at our institution between January 2011 and March 2012 were included in the analysis. Patients who required more than one operation during the same hospital admission were characterized by their index procedure. The Society of Thoracic Surgery database was used to abstract patient demographics, preoperative comorbidities, intraoperative variables, and postoperative outcomes. At our institution, enteral nutrition is considered 0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2015.03.077
2
GRIMM ET AL DYSPHAGIA IN CARDIAC SURGERY
Abbreviations and Acronyms BMI CI IQR OR RODICS SLP TEE VFSS
= = = = = = = =
body mass index confidence interval interquartile range odds ratio risk of dysphagia in cardiac surgery speech-language pathologist transesophageal echocardiogram videofluoroscopy
within 2 hours of extubation after successful completion of a bedside, nurse-administered swallow test. By protocol, patients with certain baseline characteristics (history of cerebrovascular accident, intubation more than 72 hours, stridor on extubation, or signs and symptoms of dysphagia at nurse-administered bedside screen) are formally evaluated by an SLP before permitting any oral intake. Formal SLP consultation involves an initial bedside evaluation for signs and symptoms of aspiration. The presence of dysphonia, dysarthria, cough with swallowing, or an abnormal gag reflex or voluntary cough have been established as signs of increased risk for aspiration [6, 7]. Patients demonstrating any of these signs are then referred for an instrumental assessment using videofluoroscopy (VFSS). In addition to confirming the presence of aspiration, the VFSS can also identify the type of swallowing impairment, facilitating a targeted intervention, such as chin tuck or selection of certain food consistencies.
Diagnosis of Dysphagia, Videographic Classification, and Benefit of Dietary Modifications For the purposes of this study, a diagnosis of dysphagia required documentation by a formal VFSS performed by an attending radiologist in concert with a consulting member SLP. If present, dysphagia was classified as being due to (1) retention (contrast material remaining in the pharynx after the first swallow); (2) laryngeal
Ann Thorac Surg 2015;-:-–-
penetration (contrast entering the airway above the vocal folds); (3) tracheal aspiration (contrast entering the airway below the vocal folds); or (4) a combination of those processes (Figs 1 and 2). The SLP therapeutic recommendations consisted of dietary modifications (eg, nothing by mouth status, thickened liquids, or texture modifications), volume modifications, or compensatory swallowing strategies. Furthermore, the SLP also engaged the patient in practicing certain maneuvers to rehabilitate aspects of the swallow and to promote long-term beneficial mechanical changes. To verify that most episodes of dysphagia were captured in our analysis, the medical records of all patients undergoing heart surgery during the study period were reviewed for the occurrence of postoperative aspiration pneumonia (documented infection with a computed tomography scan exhibiting the typical findings consistent with an aspiration-driven pulmonary infection) or the coding of dysphagia in a discharge problem list.
Statistical Analysis Stratification of the study population based on dysphagia was performed. Patient-specific characteristics, intraoperative variables, and postoperative outcomes were compared between patients with and patients without dysphagia. Patient-specific variables included age, race, sex, and body mass index (BMI) as well as a history of diabetes mellitus, tobacco use, congestive heart failure, chronic lung disease, cerebrovascular disease, dialysis dependence, dyslipidemia, and hypertension. Intraoperative variables included the index operation and utilization of circulatory arrest as well as operative and bypass times. Postoperative outcomes included duration of mechanical ventilation, need for reintubation, length of stay, inhospital mortality, stroke, pneumonia, and renal failure. Continuous variables, reported as mean SD or median (interquartile range), were compared using Student’s t test (parametric) or the Wilcoxon rank sum test (nonparametric) when appropriate. Categoric variables, reported as number and percentage, were analyzed using
Fig 1. Videofluoroscopic swallow study demonstrating (A) no penetration or aspiration, (B) penetration (black arrow indicates contrast above the level of the vocal cords), and (C) aspiration (black arrow indicates contrast below the level of the vocal cords).
Ann Thorac Surg 2015;-:-–-
GRIMM ET AL DYSPHAGIA IN CARDIAC SURGERY
3
Fig 2. Videofluoroscopic swallow study demonstrating (A, B, increasing retention in the pyriform sinuses (black solid arrows), resulting in (C) overflow aspiration (outlined arrow).
Pearson’s c2 analysis. Findings were considered significant at a p value less than 0.05. Lowess smoothing plots were used to transform continuous variables into categoric variables based on graphically interpreted inflection points. Univariate logistic regression modeling was performed to determine the impact of each patient-specific characteristic and perioperative variable on the development of postoperative dysphagia. Covariates (p < 0.20) were then included in a multivariable model in a systematic, stepwise fashion. The Akaike information criteria, likelihood ratio test, and Hosmer-Lemeshow goodness of fit test were used to compare model strengths. Scores were assigned based on the relative odds ratio of each independent predictor (p < 0.05) in the multivariable model. A RODICS score was then calculated for each patient in the analysis. Risk cohorts were generated based on clinically relevant score cutoffs. Stata 12.1 software (StataCorp LP, College Station, Texas) was utilized for the statistical analysis.
Results Classification of Dysphagia and Dietary Modifications Of the 1,314 patients who underwent cardiac surgery during the study period, 307 (23.4%) were referred to an SLP for formal evaluation. After consultation, a VFSS was recommended for 44.6% of those referred (137 of 307). We ultimately diagnosed dysphagia in 83.9% (115 of 137) of patients undergoing VFSS. Therefore, the incidence of radiographically proven dysphagia within the entire cohort was 8.8% (115 of 1,314). Based on the findings from the VFSS, the type of dysphagia was further characterized as retention in 14.8% of patients (17 of 115), laryngeal penetration in 12.2% (14 of 115), isolated tracheal aspiration in 4.4% (5 of 115), and aspiration due to combined pathologies in 68.7% (79 of 115). In the cohort of patients with dysphagia, dietary modifications were implemented in 64.3% (74 of 115), with the most common being no oral intake (82.4%, 61 of 74). Conversely, of patients who did not require formal radiographic examination (170), only 18.2% (31) were given nothing by mouth until the cause of their transient
swallowing dysfunction (narcosis, and so forth) resolved. For patients with dysphagia, enteral nutrition was initiated with a nasogastric or nasoduodenal tube. A percutaneous gastrostomy or gastrojejunostomy was performed in 31.3% of patients (36 of 115) before discharge.
Differences in Key Clinical Variables in Patients With and Patients Without Dysphagia After stratification based on the presence or absence of dysphagia, differences in several patient-specific characteristics, intraoperative variables, and postoperative outcomes were appreciated (Tables 1 and 2). Dysphagia patients had a lower BMI (27.0 7.1 versus 29.4 7.3 kg/ m2, p ¼ 0.001), and a higher incidence of congestive heart failure (35.6% versus 22.7%, p ¼ 0.002), more chronic lung disease (33.0% versus 16.2%, p < 0.001), and more cerebrovascular disease (19.1% versus 10.1%, p ¼ 0.003). Patients with dysphagia had more complex procedures, such as ventricular assist device implantation, heart transplantation or combined cardiac procedures (53.9% versus 41.9%), and fewer isolated coronary artery bypasses (20.0% versus 32.7%, p < 0.001). Additionally, patients with dysphagia underwent more procedures requiring hypothermic circulatory arrest (6.1% versus 1.3%, p ¼ 0.002) as well as operations with longer cardiopulmonary bypass (137.5 75.9 versus 110.8 53.1 minutes, p < 0.001) and operative times (4.7 1.8 versus 4.0 1.7 hours, p ¼ 0.001). Furthermore, they had longer durations of mechanical ventilation (median 26.4 hours [IQR: 14.4 to 63.6] versus 13.4 hours [IQR: 9.9 to 19.6], p < 0.001) and a greater incidence of postoperative pneumonia (17.0% versus 2.2%, p < 0.001) and reintubation (25.2% versus 3.0%, p < 0.001). As expected, inhospital mortality was higher (7.9% versus 3.8%, p ¼ 0.03) and total length of stay was longer (median 19 days [IQR: 12 to 33] versus 7 days [IQR: 5 to 10], p < 0.001) for patients with dysphagia.
RODICS Score After univariate analysis, 11 covariates met criteria for inclusion in the multivariable model. The final model consisted of 9 variables (hypertension and perfusion
4
GRIMM ET AL DYSPHAGIA IN CARDIAC SURGERY
Ann Thorac Surg 2015;-:-–-
Table 1. Patient-Specific Risk Factors Stratified by Dysphagia
Variable
No Radiologic Evidence of Dysphagia, Not Tested (n ¼ 1,199)
Radiologic Evidence of Dysphagia (n ¼ 115)
61 (52, 71)
65 (53, 74)
73.3 (878) 17.1 (205) 9.6 (115) 64.1 (769) 29.4 7.3 29.9 (358) 16.7 (200) 22.7 (272) 16.2 (195) 10.1 (121) 2.8 (34) 64.1 (769) 69.6 (834)
79.0 (90) 14.0 (16) 7.0 (8) 72.2 (83) 27.0 7.1 27.0 (31) 13.9 (16) 35.6 (41) 33.0 (38) 19.1 (22) 2.6 (3) 58.3 (67) 60.9 (70)
Age, years, median (IQR) Race Caucasian Black Other Male BMI, kg/m2, mean SD Diabetes mellitus Tobacco use Congestive heart failure Chronic lung disease Cerebrovascular disease Dialysis dependence Dyslipidemia Hypertension
p Value 0.11 0.41
0.09 0.001 0.52 0.44 0.002
