Association of Nutritional Support with Survival Time in Hospitalized Older Patients with Dysphagia: A Retrospective Study

Dysphagia results from age-related changes in swallowing physiology and contributes to various health status changes, including higher risks for malnutrition, pneumonia, and mortality [1, 2]. It also occurs in age-related diseases, including stroke [3] and cancer [4], and progressive neurological diseases, such as dementia [5] and Parkinson’s disease [6]. These associations, and global aging, have increased the importance of dysphagia management.

Artificial nutrition can support patients with dysphagia caused by various clinical conditions. Percutaneous endoscopic gastrostomy (PEG) is an artificial nutrition method that secures an enteral nutrition (EN) route in dysphagia patients [7]. PEG tubes have been widely used in Japan as a convenient and effective EN route since approximately 1990; however, the use of PEG feeding in dementia patients is controversial [8-10]. Total parenteral nutrition (TPN) is an alternative nutritional management method [11] and has been the most frequently used nutritional support method in Japan since 2010. A comparison of long-term outcomes associated with PEG feeding and TPN in older patients found that older patients, and those with poorer nutritional status and severe dementia, were more likely to receive TPN, whereas patients with cerebrovascular disease were more likely to receive PEG [12]. A 2012 Japan Geriatric Society position statement on end-of-life care for older adults suggested that patients’ families or caregivers should decide whether feeding tubes should be withheld or withdrawn [13].

Although advanced directives are now widely used, the decision of whether to institute enteral tube feeding is an important issue for older patients and their families. However, global clinical evidence for the effect of artificial nutritional support on survival time, and for differences between administration routes, is lacking. The rate of aging in Japan is more rapid than in any other country, and more research is needed on clinical decision-making for the older population. The study aim was to clarify the effects of artificial nutritional support route on survival time of older terminal patients with dysphagia.

We conducted a retrospective observational study in Fukushimura Hospital, Toyohashi, Japan. The hospital admits many older patients in the terminal stages of various diseases. We reviewed patient charts and medical reports. Inclusion criteria were hospitalized patients who died September 2010 to August 2016 and who underwent artificial nutritional support during hospitalization. Exclusion criteria were mortality within 30 days of hospital admission, mortality within 5 days of nutritional support initiation, artificial nutritional support before admission, and central venous nutrition support without high-calorie infusion. Data on patients’ general and clinical background, clinical courses, and routes and types of artificial nutritional support, including EN, parenteral nutrition (PN), central venous, and PEG, were extracted. Cause of death was obtained from death certificates.

Eligible patients were allocated to three groups according to type of nutritional support administered at admission: EN support only (EN group), EN support and subsequent switch to PN (EN/PN group), and PN support only (PN group). Between-group comparisons were performed of patient clinical and general backgrounds. The EN/PN group comprised patients introduced to EN support and then switched to PN support for any reason (reasons for and time of the switch were not recorded). In this study, no patients were switched from PN to EN. Patients were also allocated to three groups according to time from artificial nutritional support initiation to death (survival time): <180 days, 180–359 days, and ≥360 days, and their backgrounds and causes of death were compared. Risk factors for 180- and 360-day mortality were analyzed in survivors and non-survivors 180 and 360 days after artificial nutritional support initiation. Cutoff points for body mass index (BMI) and serum albumin concentration level (Alb) were set as median values.

The Kruskal-Wallis test was used for continuous variables and χ² or Fisher’s exact tests for categorical variables. Survival curves of the number of days from initiation of artificial nutritional support to death, according to type of artificial nutrition, were analyzed using the Kaplan-Meier method, and comparisons were made using the log-rank test. To evaluate independent risk factors for 180- and 360-day mortality, a stepwise selection method was used to select significant variables for a logistic regression model. SPSS Statistics 25.0 (IBM, Armonk, NY, USA) was used for analysis. For all analyses, significance levels were two-tailed, and p values <0.05 considered significant.

During the study period, 1,397 patients died. Of these, 1,226 patients received artificial nutrition during hospitalization, and 1,061 met the inclusion criteria and were included in the analysis (online suppl. Fig. s1; see www.karger.com/doi/10.1159/000524085 for all online suppl. material). Most eligible patients received PN support (n = 873, 82.3%), with fewer in the EN/PN group (n = 159, 15.1%) and very few in the EN group (n = 29, 2.7%). Table 1 shows the general characteristics of patients by group. Most patients in all three groups were >75 years; patients in the EN/PN group had the lowest median age. The PN group contained the highest proportion of women. Of patients, 81.7% underwent artificial nutritional support because of dysphagia. Alb, but not BMI, differed significantly among the three groups. Thirty-five percent of patients had neurological diseases, including dementia, but there were no significant between-group differences. Most patients admitted to hospital were bedridden.

The number of days from artificial nutrition initiation to death was significantly lower in the PN group than in the EN and EN/PN groups (p < 0.001) (Table 1). The effect of artificial nutritional support type on survival time was compared among patients receiving different types of artificial nutritional support (Fig. 1). Patients in the PN group had significantly shorter survival time than those in the EN and EN/PN groups (log-rank test, p < 0.001). In the PN group, 70% of patients died within 180 days of artificial nutritional support initiation. The 360-day mortality was lowest in the EN group (55.2%) and highest in the PN group (85%; p < 0.001) (Table 1). Many patients died from heart failure, particularly in the PN group (p = 0.001) (Table 1). By contrast, more patients in the EN group died of respiratory failure than heart failure (p < 0.001). Approximately 30% of patients in each group died of pneumonia, but there were no significant between-group differences (p = 0.927).

Patient background and cause of death were categorized according to survival time from artificial nutritional support initiation: <180 days, 180–359 days, or ≥360 days (online Suppl. Table s1). Of patients who died within 180 days, 94.8% received PN. The proportion of patients with low BMI was significantly higher in patients with longer survival time (p < 0.001). Approximately 50% of patients with neurological dysfunction survived for over a year. The proportions of patients who died from respiratory failure or malignant neoplasm were highest in the <180-day mortality group (p = 0.034 and p = 0.001, respectively). However, no significant differences were found for other causes of death.

Factors affecting patient survival time were evaluated for 180- and 360-day mortality using logistic regression. Factors associated with 180-day mortality were being in the PN group, BMI, cerebral infarction, neurological dysfunction, and male sex (Table 2). Factors associated with 360-day mortality were being in the PN or EN/PN groups, BMI, Alb, neurological dysfunction, and male sex (Table 2). PN was associated with high-mortality risk at either 180 days (odds ratio [OR], 7.401) or 360 days (OR, 1.823). Conversely, the EN/PN combination reduced the risk for 360-day mortality (OR, 0.132). Patients who displayed neurological dysfunction also had lower risks for 180- and 360-day mortality (OR, 0.471 and 0.300, respectively).

The findings showed that older patients supported with PN had 7.3- and 1.8-fold higher risks for 180- and 360-day mortality, respectively, compared with patients supported with EN or a combination of EN and PN. Therefore, the use of EN support for older patients with dysphagia may help maximize their remaining lifespan.

In over 80% of patients, artificial nutritional support was initiated because of swallowing difficulty, and 63.6% of patients had diseases that cause dysphagia, including cerebral infarction, dementia, and other types of neurological dysfunction, which indicates that diseases in hospitalized older adults frequently cause difficulty swallowing. The introduction of artificial nutritional support is crucial to maintain these patients’ lives. However, making decisions about the type of artificial nutrition is a burden for patients and their families. One systematic review indicated that emotional stress induced by the treatment decision can persist for over 6 months [14]. Therefore, the availability of scientific evidence to aid decision-making may reduce stress. Many patients in the present study received PN. This is because EN patients are often cared for in nursing homes rather than in hospitals, and the use of nasogastric feeding/PEG tends to be avoided in Japan. In addition, most patients in the present study were bedridden older patients in the terminal stage. Therefore, these data may not be generalizable to other patient populations, particularly as current guidelines recommend EN as the standard treatment.

In Japan, most older patients with dysphagia receive either tube feeding or total PN [15]. Additionally, nearly 100% of our patients were bedridden. The survival curves showed a significantly shorter survival time for patients in the PN group than for those in the EN and EN/PN groups. Furthermore, PN support was a mortality risk factor at both 180 and 360 days. A previous study showed that bacterial translocation occurred with long-term PN use, whereas the use of EN prevents this phenomenon [16]. A systematic review that compared the use of EN with PN and of EN with a combination of EN and PN in adults in intensive care identified fewer deaths in the first 30 days when EN and PN were used and a lower incidence of sepsis when EN was used rather than PN [17]. Although the review authors stated that the evidence was not robust, this suggests that EN may be associated with control of disease progression, and the findings are comparable with the present findings.

In the present study, the most frequent cause of death in the PN group was heart failure, whereas it was pneumonia in the EN and EN/PN groups, which is the most common cause of death in older adults with dysphagia. According to guidelines for the nutritional support of critically ill adult and pediatric patients, early EN initiation reduces the incidences of infection and mortality [18]. The intestinal tract plays an important role in immunity, as well as digestion and absorption, and it may be that EN prolongs survival, even in the terminal stage of an illness. The present findings differ from existing data from Western countries [19-22] but are consistent with those of previous Japanese studies [12, 23]. However, because PEG is thought to prolong life, guardians frequently choose this unphysiological and more expensive method of nutritional management. Therefore, from an ethical perspective, it is essential to inform the relevant parties about evidence for the success of various types of artificial nutrition before implementing an advanced directive.

There were several study limitations. First, this was a retrospective cohort study conducted in a single center, and groups containing unequal numbers of patients were compared. Second, the duration of administration of the EN and PN combination was not evaluated. Third, although we studied patients with clinically diagnosed dysphagia, swallowing function was not directly assessed. Fourth, we did not study the effects of patient medications, which may interfere with swallowing functions. Fifth, more than 90% of the data were from people aged ≥75 years, and more than 80% were from people aged ≥80 years, most of whom were elderly, so the data cannot be compared with those from other age-groups. Finally, the very high proportion of PN among this group of patients is not representative of most patients with dysphagia, so the results cannot be extrapolated to other patient groups. Despite these limitations, and given the lack of evidence on the use of artificial nutrition for dysphagia in older patients, these findings could help to facilitate decision-making by clinicians who provide medical care for older adults.

In conclusion, the use of PN use was associated with higher risks of mortality, compared with EN or EN/PN, but residual confounding cannot be excluded. The findings should be explored further in prospective studies.

We thank the staff of Fukushimura Hospital for their assistance with the present study. In particular, we thank the members of the nutrition support team at Fukushimura Hospital who recorded the key pathological and nutritional events in the clinical records during hospitalization.

The present study was approved by the Ethics Committee of Fukushimura Hospital (approval number 187). The requirement for written informed consent was waived by the Ethics Committee of Fukushimura Hospital for this retrospective study, with public notification of the study made by public posting.

The authors have no conflicts of interest to declare.

We have no funding resources for conducting the present study.

Conceived and designed the experiments: Hiroyasu Akatsu, Yoshiyuki Masaki, and Toshie Manabe. Performed the experiments: Yoshiyuki Masaki, Yoshiko Yamamoto, Kayoko Yamamoto, Osamu Kohashi, Takayuki Yamamoto, and Hiroyasu Akatsu. Analyzed the data: Toshie Manabe and Yoshiyuki Masaki. Interpreted the study results: Toshie Manabe, Yoshiyuki Masaki, Yoshiko Yamamoto, Kayoko Yamamoto, Osamu Kohashi, Takayuki Yamamoto, Hirotaka Ohara, and Hiroyasu Akatsu. Wrote the first draft of the manuscript: Toshie Manabe and Yoshiyuki Masaki. All the authors read and approved the final manuscript.

All data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author.

Toshie Manabe and Yoshiyuki Masaki contributed equally to this work.