Abstract
Background:
Dysphagia is a frequent complication that may increase morbidity and mortality in Parkinson’s disease (PD). Nevertheless, there is limited data on its objective impact on healthcare outcomes.
Objective:
To investigate the outcomes associated with dysphagia in hospitalized patients with PD and associated healthcare costs and utilization.
Methods:
We performed a retrospective cohort study using the National Inpatient Sample (NIS) data from 2004 to 2014. A multivariable regression analysis was adjusted for demographic, and comorbidity variables to examine the association between dysphagia and associated outcomes. Logistic and negative binomial regressions were used to estimate odds or incidence rate ratios for binary and continuous outcomes, respectively.
Results:
We identified 334,395 non-elective hospitalizations of individuals with PD, being 21,288 (6.36%) associated with dysphagia. Patients with dysphagia had significantly higher odds of negative outcomes, including aspiration pneumonia (AOR 7.55, 95%CI 7.29–7.82), sepsis (AOR 1.91, 95%CI 1.82–2.01), and mechanical ventilation (AOR 2.00, 95%CI 1.86–2.15). For hospitalizations with a dysphagia code, the length of stay was 44%(95%CI 1.43–1.45) longer and inpatient costs 46%higher (95%CI 1.44–1.47) compared to those without dysphagia. Mortality was also substantially increased in individuals with PD and dysphagia (AOR 1.37, 95%CI 1.29–1.46).
Conclusion:
In hospitalized patients with PD, dysphagia was a strong predictor of adverse clinical outcomes, and associated with substantially prolonged length of stay, higher mortality, and care costs. These results highlight the need for interventions focused on early recognition and prevention of dysphagia to avoid complications and lower costs in PD patients.
INTRODUCTION
Parkinson’s disease (PD) is a prevalent neurodegenerative condition that presents with a broad range of motor and non-motor manifestations. Among them, oropharyngeal dysphagia has been recognized as a common symptom, affecting the majority of patients with PD during the disease [1–3]. Although dysphagia can present at any time in the disease course, its presence has been considered a marker for late-stage PD and may result in significant morbidity and quality of life impairment [4–7]. Among the most well-recognized complications of impaired swallowing is aspiration pneumonia, which is associated with increased mortality in PD patients [8, 9].
Inpatient and economic evaluations of dysphagia have been studied in hospitalized patients with other conditions, demonstrating an increased risk of adverse outcomes, prolonged admission and greater costs [10, 11]. For instance, a large study demonstrated that individuals with Alzheimer’s disease and dysphagia were found to have a 38%increase in length of stay and US$10,703 higher admission costs when compared to patients without dysphagia [10, 12]. Moreover, among the general population, the economic impact of dysphagia in the US between 2009 to 2013 revealed that this complication may increase admissions costs by 42–44%, which was estimated to account for an additional cost of 4.3–7.1 billion dollars annually [13]. However, similar data for PD are not available. Therefore, this study aims to investigate and quantify the association of dysphagia among hospitalized PD patients with length of stay, hospitalization cost, and serious adverse outcomes.
METHODS
Ethics statement
The database used in this study is publicly available and limited to de-identified information; therefore, research ethics board review was not required.
Data source
Our study utilized data from the National Inpatient Sample (NIS) database from 2004 to 2014. The NIS database is part of the Healthcare Cost and Utilization Project (HCUP) in the United States (US), which contains the largest collection of all-payer, inpatient data starting in 1988. It is estimated to contain data on 7 to 8 million inpatient stays annually. The NIS database aims to provide representative nationwide data. Prior to 2012, a stratified sample of US hospitals was collected and weighted to ensure that the data were nationally representative. Starting in 2012, the NIS database sampling method was changed to include 20%of discharges from all hospitals in this database. The new redesigned strategy aimed to reduce sampling error and to provide more precise estimates [14]. The documented variables include discharge diagnoses and procedures for each inpatient stay using the International Classification of Diseases, Ninth Revision, Clinical Modification codes (ICD-9-CM), plus demographic variables, expected payor, and costs of delivered care.
Study samples and variables
We performed a retrospective cohort analysis of all hospitalizations of individuals ages 18 and older. The International Classification of Diseases, 9th version (ICD-9-CM) codes were used to identify patients with Parkinson’s Disease (332.0) and dysphagia (787.2). We excluded elective admissions to avoid the possibility of analyzing patients who could have been admitted for elective dysphagia-related procedures, such as percutaneous endoscopic gastrostomy (PEG) placement. The primary outcomes were length of hospital stay (measured in days), overall cost per admission, and death. Secondary outcomes included direct potential complications of dysphagia, including PEG placement, mechanical ventilation, aspiration pneumonia, respiratory failure, electrolyte disorders, and sepsis. Other outcomes that were not considered a direct consequence of dysphagia were also explored, such as encephalopathy, palliative care consultations, and diagnosis of pressure ulcer.
We describe the following baseline patient characteristics by dysphagia status: age, gender, race/ethnicity, insurance type, median neighborhood income, admission source and discharge status. Insurance types were characterized as private, Medicare, Medicaid and others (unknown and uninsured). Medicare is a federal program that provides insurance and health coverage for patients over 65 years of age, while Medicaid primarily serves the low-income population of the United States [15, 16]. Total charges were used to calculate the average inpatient cost per hospitalization, termed the “inpatient cost per case”. In the NIS database, total charge values are rounded to the nearest dollar and numbers are considered inconsistent if excessively low or high, therefore excluded from the analysis. Between 1998 to 2006, allowable values were between $25 and $1.0 million; from 2007 to 2010 between $100 and $1.5; and starting in 2011 between $100 and $5.0 million. Generally, this variable does not include professional fees and non-covered charges.
The burden of comorbid conditions was defined using the Elixhauser Comorbidity Index [17]. This index contains at least 30 comorbid conditions with differential weights and scores and was included to control for other comorbidities which may correlate with the presence of dysphagia. Several studies have demonstrated the utility of this index in hospital settings, particularly when examining mortality [18, 19].
Results containing cell sizes equal to or less than 10 patients were removed from our tables due to privacy concerns and to comply with the publishing requirements by HCUP.
Statistical analyses
Baseline characteristics were compared using standardized differences. To compare our PD groups with and without dysphagia, continuous outcomes were presented as mean (standard deviation), and categorical variables were presented as frequency (proportion). Binary outcomes were evaluated using odds ratios obtained from logistic regression models with adjustment for potential confounders of age, sex, race, Elixhauser Comorbidity Index, dementia, and insurance type. Age and comorbidity index were treated as continuous variables in our models. Due to the high prevalence of dysphagia in patients with stroke and diseases of esophagus, we also adjusted our analysis for these conditions [11].
Continuous outcomes were compared using incidence rate ratios obtained from negative binomial regression models, adjusted for covariates as in the logistic regression models. Given the large number of outcomes assessed, statistical significance was adjusted for multiple testing using the Bonferroni correction. Statistical significance was defined as a p-value less than 0.002 (0.05/25) [20]. All tests were two-tailed. The R software (version 3.6.3) was used for all statistical analyses [21].
RESULTS
From 2004 to 2014, 616,843 hospital admissions in the NIS database recorded a diagnosis of PD. After excluding elective admissions or missing data regarding admission sources, a total of 334,395 hospitalizations were analyzed. Of those, 21,288 (6.36%) were also coded as having dysphagia. Baseline characteristics in PD patients with and without dysphagia are displayed in Table 1. PD patients with dysphagia were older (average age 79.34 years (SD 9.16) vs. 77.62 years (SD 9.85) and more frequently male (60%vs. 52.6%). The mean Elixhauser Comorbidity Index score was similar in both groups (3.88, SD 1.72 without dysphagia vs. 3.80, SD 1.67 with dysphagia). As expected, the presence of comorbid ischemic or hemorrhagic stroke and disease of the esophagus, commonly associated with dysphagia, were more prevalent in patients with dysphagia (18%vs. 10.8%, and 20.7%vs. 16.3%, respectively). Patients with dysphagia more frequently lived in a neighborhood where the median income was in the 76th–100th percentile for the US, more frequently had health insurance through the Medicare program (reflecting age greater than 65). More patients with PD and dysphagia were discharged to a skilled nursing facility, or home health care. Similarly, patients in the dysphagia subgroup more often had documentation of a do-not-resuscitate (DNR) code status by (3.4%vs. 1.2%).
Baseline demographics and characteristics of hospitalized patients with PD with and without dysphagia between 2004–2014
SD, standard deviation; DNR, do not resuscitate. p-values refer to two-tailed test.
Regarding the primary outcome measures of our analysis, PD patients with dysphagia had higher inpatient costs (mean costs per case 46,575 vs. 31,937 US dollars), a longer length of stay (mean number of days 8.01 vs. 5.56), and more procedures per case (mean number of procedures of 1.39 vs. 1.13). After adjusting for confounders, PD patients with dysphagia still had more inpatients costs (IRR 1.46, CI 1.44–1.47), prolonged stay (IRR 1.44, CI 1.43–1.45), and higher number of procedures (IRR 1.28, CI 1.26–1.30) (Table 2).
Clinical outcomes of hospitalized patients with PD with and without dysphagia between 2004–2014
SD, standard deviation; USD, US dollars; CI, confidence interval. p-values refer to two-tailed test.
Related outcomes and procedures in hospitalized PD patients with and without dysphagia between 2004–2014
SD, standard deviation; OR, odds ratio; CI, confidence interval. p-values refer to two-tailed test.
After adjusting for the potential confounders, PD patients with dysphagia had significantly higher adjusted odds of adverse clinical outcomes (Table 2). Potential direct and known dysphagia consequences, such as aspiration pneumonia, were almost eight times higher in the dysphagia group (AOR 7.55, CI 7.29–7.82). PD patients with dysphagia also had higher adjusted odds of respiratory failure (AOR 2.15, 2.06–2.24), pressure ulcer (OR 1.99, CI 1.90–2.08) and malnutrition (AOR 4.03, CI 33.83–4.24). Furthermore, patients with PD and dysphagia had an increased likelihood of urinary tract infection (OR 1.42, CI 1.38–1.47), and sepsis (AOR 1.91, CI 1.82–2.01). This subgroup also had higher odds of undergoing invasive procedures. As expected, the most striking association of a dysphagia diagnosis was with PEG tube placement (AOR 17.46, CI 16.58–18.40), but the adjusted odds of central venous catheterization (AOR 2.00, CI 1.90–2.11), intubation (AOR 2.12, CI 1.96–2.29), and mechanical ventilation (AOR 2.00, CI 1.86–2.15) were also greater. Finally, the dysphagia group was more likely to receive palliative care consultation (OR 2.73, CI 2.51–2.97) or die (AOR 1.37, CI 1.29–1.46) during the hospitalization.
DISCUSSION
This retrospective cohort study identified that hospitalized PD patients with dysphagia had significantly higher mortality, length of stay, and overall costs when compared to PD patients without dysphagia between 2004 and 2014. Dysphagia symptoms were also associated with higher rates of other serious clinical outcomes, such as mechanical ventilation, PEG tube placement, ICU admission, and fluid and electrolyte disorders. To our knowledge, this is the first study to quantitatively estimate the increased risk for these poor outcomes, including costs, associated with dysphagia in a large cohort of hospitalized patients with PD. Our results highlight the significant clinical and economic burden of a common manifestation in PD patients admitted to a hospital.
Impaired swallowing in PD is a common symptom and has been estimated to affect at least a third of all patients with PD during the disease course [2]. Despite the complex interaction of comorbidities, dysphagia is often associated with a higher incidence of pneumonia and other adverse health outcomes, leading to increased mortality [22]. A retrospective study from 2005 demonstrated that dysphagia was an overall marker of worse prognosis and mortality for many conditions, including PD patients [11]. Large studies evaluating patients with stroke and Alzheimer’s disease have also demonstrated a longer length of stay and morbidity in patients with impaired swallowing in the inpatient setting. Our findings show that the adverse implications of dysphagia also extend to PD in even larger proportions, with additional 2.52 admission days in the presence of dysphagia (44%increase).
Having a diagnosis of dysphagia carries the risk of several other comorbidities and diagnoses. In addition to the known association with worse quality of life and impairments in psychological domains, the presence of impaired swallowing may have additional consequences to the PD population, including insufficient medication intake, malnutrition, and dehydration [4, 23]. In keeping with this, PD patients in our cohort with dysphagia had higher odds of dehydration, altered mental status, and pressure ulcers. These results are similar to other studies examining dysphagia in the general population [11].
Given these findings, it is unsurprising that dysphagia in PD was associated with significantly higher hospital-related costs than hospital costs in the absence of dysphagia (additional $14,638). This issue is fundamental in PD, where hospitalizations are frequent, and several studies have considered inpatient hospitalization to be the main driver of total direct costs in PD [24]. For instance, the proportional contribution of inpatient admissions to the total costs of PD has been estimated to range from 25–39%in Europe and up to 72%in the US [25]. The estimates of the economic impact of PD in the US have varied, although one study estimated the expected medical expenses for patients with PD to be $12,800 per capita higher when compared to patients without PD (total of $8.1 billion of extra costs for one year). The nonmedical cost of PD-related lack of productivity and other associated factors was similarly high, estimated to total $6.3 billion in 2010 [26].
Several studies have suggested that PD patients have a longer length of stay than non-PD patients [27]. Prolonged inpatient admissions for PD patients may lead to increased costs, higher comorbidities, and the potential for adverse outcomes [28, 29]. Previous analyses identified that dysphagia may prolong the length of stay up to 1.64 days in the general population and up to 2.16 days in patients with dementia [10, 11]. In our cohort, the presence of impaired swallowing in patients with PD was associated with an increase in mean total length of stay by 2.5 days. Therefore, it is expected that the early diagnosis and treatment of dysphagia in hospitalized PD patients may reduce the total length of stay and associated costs of care.
There are several limitations to our study. First, we were unable to stratify PD cases by severity or clinical stage, as the NIS database does not offer detailed information regarding PD diagnosis, including specific motor symptoms, duration of disease, or medication use. Consequently, we cannot objectively characterize if the associated dysphagia is a marker of late-stage PD and not causally associated with the adverse outcomes found in this dataset. We were also unable to determine the underlying reason for dysphagia, which may include PD but also other medical conditions. To partially correct this issue, we adjusted our analysis for the most frequent causes associated with dysphagia, including stroke and disease of the esophagus. We have also adjusted for the Elixhauser Comorbidity Index analysis, which measures patient’s comorbidities based on more than 30 ICD-9 diagnosis codes and is designed to predict hospital mortality and resource use. Another significant limitation to any health service database includes the possibility of inaccurate coding, which may not directly reflect a patient’s diagnosis, performed interventions or procedures. For instance, it is possible that some of the included patients might have been mistakenly categorized as PD, when they in fact were affected by another form of parkinsonism. Although the ICD-9 code 787.2 for individuals with dysphagia has been previously validated with high specificity (89.5%), its sensitivity was considered low [30]. This is reflected in the low proportion of PD patients with the dysphagia code in this database. Likely, mild dysphagia is commonly missed, and the total frequency and morbidity of this symptom may be higher than we report here. Consequently, the relative morbidity of those with and without dysphagia that we report here likely represents the more severe instances of dysphagia. Finally, given that this large longitudinal database does not include individual identifiers, we cannot identify readmissions. Therefore, more than one encounter might have been associated with the same patient.
CONCLUSION
In summary, in our large cohort evaluating more than 330,000 patients, dysphagia was associated with worse clinical outcomes, higher admission costs, and prolonged stay among hospitalized patients with PD in the US. After controlling for patient’s comorbidities and most frequent diagnoses associated with dysphagia, patients with PD and impaired swallowing still had higher odds of serious complications, including death.
Our study highlights the objective burden and estimated costs of dysphagia in hospitalized patients with PD. Further research focusing on preventing dysphagia complications in PD patients is necessary to alleviate negative clinical outcomes and healthcare associated costs and utilization. Additional epidemiological studies should also focus on identifying individuals at risk and the influence of potential modifiable factors, such as medication use. With a better understanding of the clinical course of dysphagia, standardized and validated screening methods should be developed. Given the high prevalence of dysphagia in PD, the implementation of objective interventions to mitigate the morbidity associated with dysphagia should be prioritized and pursued in future clinical research.
CONFLICT OF INTEREST
Funding Sources: Safra Foundation and Porridge for Parkinson’s Foundation. The authors declare that there are no conflicts of interest relevant to this work.