Predictors of Percutaneous Endoscopic Gastrostomy Placement in Patients with Progressive Neuromuscular Dysphagia

Introduction: In the progression of dysphagia secondary to neurologic disorders (ND), a percutaneous gastrostomy tube (PEG) is commonly employed to achieve long-term nutritional goals safely. We aimed to characterize predictors for PEG tube placement in patients evaluated for dysphagia secondary to progressive ND. Methods: Patients with dysphagia related to ND were identified retrospectively across 4 tertiary academic centers from December 2010 – December 2021. The primary endpoint was placement of gastrostomy. We performed multivariable logistic regression for the primary outcome and Cox proportional hazards regression for the primary endpoint to identify predictors. The severity of the patient’s underlying clinical condition was gauged by the Charlson Comorbidity Index, the Clinical Fragility Scale (CFS), American Association of Anesthesiology (ASA) classification, and clinical features. Results: A total of 273 patients were included (median age 63.9, male 42.6%). The most common disorders were multiple sclerosis (27.1%), myasthenia gravis (22.3%) and amyotrophic lateral sclerosis (ALS, 15.4%). A gastrostomy was recommended in 38 patients (13.9%) at the initial consultation. At follow-up, a total of 51 patients had a gastrostomy (median time: 48 days, interquartile range: 7-181). Characteristics that predicted recommendation of PEG were: diagnosis of ALS (HR 10.39, 95% CI: 5.34-20.19, P < 0.0001), CFS per point (HR 1.49, 95% CI: 1.23-1.81, P < 0.0001), ASA Classification per point (HR 2.23, 95% CI: 1.28-3.89, P = 0.0046), under evaluation for ND including electromyography (HR 2.06, CI 95%: 1.10-3.82), and dysphagia to both liquids and solids (HR 2.57, CI 95%: 1.44-4.57, P = 0.0014). Increased BMI was protective (HR 0.92, CI 95%: 0.87-0.98, P = 0.0212). Conclusion: In this cohort of patients with neurologic disease, the diagnosis of ALS was the greatest identified risk factor for PEG placement. Other than this diagnosis, the presence of solid and liquid dysphagia was the main clinical finding which predicted PEG placement. We found that both the ASA score and CFS were highly predictive of the need for PEG placement. We suggest that when considered in conjunction with the predictive clinical feature of dysphagia to both solids and liquids, these scores may help clinicians anticipate PEG when applied to neurologic disorders as a whole (Figure 1, Table 1).Figure 1.: A: Kaplan-Meier Curve, Incidence of Gastrostomy Placement by Diagnosis of ALS. P < 0.001. B: Kaplan-Meier Curve, Cumulative Incidence of Gastrostomy Placement by Type of Dysphagia, P < 0.001. Table 1. - A Total of 51 out of 273 Patients had a Gastrostomy Placed at Follow up Univariable Multivariable* HR (95% CI) P Value HR (95% CI) P Value Age at Evaluation, per 1 year 1.04 (1.01-1.06) 0.0049 1.02 (0.99-1.04) 0.2755* Male 2.60 (1.46-4.62) 0.0011 1.30 (0.70-2.42) 0.4078 White Race 0.71 (0.28-1.78) 0.4604 1.32 (0.51-3.42) 0.5726 Body Mass Index, per 1 kg/m2 0.92 (0.87-0.96) 0.0060 0.92 (0.87-0.98) 0.0092 Mayo Arizona 3.62 (2.08-6.29) < 0.0001 1.16 (0.60-2.23) 0.6619 Amyotrophic Lateral Sclerosis 17.7 (9.60-32.60) < 0.0001 10.39 (5.34-20.19) < 0.0001* Myasthenia Gravis 0.20 (0.06-0.65) 0.0074 0.63 (0.18-2.16) 0.4624 Hereditary Neuropathy 0.11 (0.02-0.80) 0.0296 0.27 (0.04-2.01) 0.2002 Multiple Sclerosis 0.16 (0.05-0.48) 0.0015 0.45 (0.13-1.56) 0.2089 Clinical Frailty Scale, per 1 point 1.92 (1.61-2.30) < 0.0001 1.49 (1.23-1.81) < 0.0001* ASA Classification, per 1 point 3.76 (2.32-6.10) < 0.0001 2.23 (1.28-3.89) 0.0046 Charlson Comorbidity Index, per 1 point 1.03 (0.92-1.17) 0.5903 0.99 (0.80-1.22) 0.9088 Arm Weakness Present 4.22 (2.31-7.70) < 0.0001 0.94 (0.43-2.04) 0.8782 Leg Weakness Present 2.62 (1.50-4.57) 0.0007 1.23 (0.67-2.27) 0.5076 EMG Performed 3.01 (1.68-5.37) 0.0002 2.06 (1.10-3.82) 0.0229 Aspiration Pneumonia 2.05 (1.12-3.76) 0.0198 0.95 (0.51-1.78) 0.8673 Speech Swallow Evaluation 2.13 (1.21-3.75) 0.0085 0.91 (0.49-1.68) 0.7519 Portable Ventilator 12.75 (6.46-25.13) < 0.0001 1.66 (0.79-3.47) 0.1775 Pulmonary Function Tests Performed 2.32 (1.32-4.08) 0.0036 0.74 (0.39-1.40) 0.3525 Diabetes Mellitus 0.36 (0.13-1.00) 0.0502 0.50 (0.17-1.42) 0.1926 Time with Neuromuscular Disorder, per 365 days 0.95 (0.91-0.99) 0.0190 0.99 (0.94-1.04) 0.7021 Type of Dysphagia Solids Only 1.00 (reference) NA 1.00 (reference) NA Liquids Only 2.82 (0.66-12.15) 0.1634 2.73 (0.64-11.63) 0.1735 Both 3.97 (2.23-7.07) < 0.0001 2.57 (1.44-4.57) 0.0014* Referred by Neurology 3.19 (1.71-5.97) 0.0003 0.83 (0.40-1.71) 0.6172 Evaluated by esophagologist 0.58 (0.28-1.19) 0.1361 1.13 (0.52-2.48) 0.7608 Given the outcome with the lowest event rate (n = 51), the maximum number of variables allowed in the multivariable models was 5. Multivariable Cox Proportional Hazards Regression models were adjusted for the strongest variables associated with gastrostomy placement at < 0.05 on univariable regression analysis. * The 3 strongest variables adjusted in the baseline model were 1) amyotrophic lateral sclerosis, 2) clinical frailty scale, per 1 point, and 3) dysphagia type. Age at evaluation was added to the baseline 4-variable model as a covariable.