Usefulness of barium sulfate and iohexol as contrast agents for VFSS in visualizing components of swallowing predictable of poor outcomes

Barium sulfate and iohexol are commonly used as contrast agents for videofluoroscopic swallowing study (VFSS). This study compared their usefulness as contrast agents in visualizing components of swallowing predictable of subsequent pneumonia and unintentional weight loss after VFSS. This was a randomized, controlled, crossover trial. The two contrast agents were alternately used in the same participants, and the order in which the contrast agent was tested first was randomly assigned. After VFSS, we followed the participants for 3 months and the association between VFSS findings of each contrast agent and the subsequent pneumonia and unintentional weight loss were analyzed. A total of 30 participants were included in the analysis. We recorded 11 cases of subsequent pneumonia and 13 of unintentional weight loss. Regarding the risk of subsequent pneumonia after VFSS, only the oral transit time and number of swallows tested with barium sulfate indicated significant differences between participants with and without subsequent pneumonia. For unintentional weight loss, oral transit time and pharyngeal wall coating after swallowing tested with barium sulfate, as well as oral transit time, nasal penetration, residue in the valleculae, PAS scores, and number of swallows when testing with iohexol demonstrated significant differences between those with and without unintentional weight loss.


Participants
The inclusion criteria were as follows: (1) patients who were admitted in a tertiary university hospital and referred to the Department of Physical Medicine and Rehabilitation for VFSS to evaluate the safety of swallowing, (2)  patients who could maintain a sitting position during VFSS, (3) patients who could stay alert for at least 20 min, and (4) patients who have sufficient cognitive function to follow simple directions during VFSS.The exclusion criteria were as follows: (1) patients aged < 20 years, (2) patients with laryngeal pathology including laryngeal cancer, stenosis, paralysis, presence or history of tracheostomy, and postoperative head and neck surgery, (3) patients who could not complete VFSS due to aspiration episodes more than once, and (4) patients who were newly diagnosed with other diseases that could aggravate dysphagia during the 3-month follow-up period.
All the inpatients who were referred to our department for VFSS were assessed for eligibility based on medical records.All participants were provided study-related information directly from the researchers and gave written informed consent 1 or 2 days prior to VFSS. Figure 1

Intervention
VFSS was performed using an X-ray flat-panel detector system (FPD, Zexira®, Toshiba, Tokyo, Japan).Swallowing images were digitally recorded at a rate of 30 frames/s from a lateral viewing plane.The VFSS protocol consisted of (1) 3-ml thin liquid (barium sulfate solution or iohexol) provided with a spoon, (2) 5-ml yogurt mixed with each contrast agent provided with a spoon, (3) 5-ml banana mixed with each contrast agent provided with a spoon, and (4) 10-ml thin liquid (barium sulfate solution or iohexol) provided with a cup.Each of the consistencies was given twice to the participants, and the two contrast agents were given alternately for each time (a total of 8 boluses; 4 boluses mixed with barium sulfate and 4 boluses mixed with iohexol) (Fig. 1).Which contrast agent was used first was randomly assigned.To maintain a constant concentration of contrast agent to be mixed with yogurt and banana, 2 ml of each contrast agent was mixed with 10 g of yogurt and banana.According to IDDSI level, liquid barium sulfate solution was level 1, iohexol was level 0, yogurt mixed with each contrast agent were both level 4 and banana mixed with each contrast agent were both level 6.For patient safety, www.nature.com/scientificreports/ the VFSS was terminated if a second aspiration was observed at any stage of the VFSS.Imaging acquisition is maintained until the clearing of residue is completed.In addition, to minimize radiation exposure during VFSS, we performed VFSS at lateral viewing plane and total time for radiation exposure was controlled within 5 min by frequent ceasing of image acquisition 25 .If a radiation exposure time exceeded 5 min, VFSS was terminated and the participants were excluded from the analysis.VFSS findings were categorized according to the penetration-aspiration scale (PAS) 26 and functional dysphagia scale (FDS) 27 .The FDS is a 100-point scale originally developed for quantifying functional aspect of swallowing by supplementing PAS which measures only the anatomical depth of food material penetration 27 .Although it was initially developed for stroke patients, previous studies have revealed that it is also useful to evaluate dysphagia in patients with Parkinson's disease, head and neck cancer, and swallowing disorders due to frailty or general weakness 6 .The highest score (higher scores indicate worse swallowing function) for any type of food tested on the VFSS was used in the analysis.Additionally, the efficiency of swallowing was evaluated using the number of swallows required to eliminate food from the pharyngeal space 28 .If the participants could not eliminate the contrast after repetitive swallows of over ten, the count was considered at ten.Since mild coating of pharyngeal space with contrast agent was considered as normal, only prominent residue in valleculae and pyriform sinuses were considered as remained bolus.We followed up the participants for 3 months after the VFSS and analyzed the association between the VFSS findings of each contrast agent and the risk of subsequent pneumonia and unintentional weight loss.

Outcome measures
Development of subsequent pneumonia after VFSS during the follow-up period was the primary outcome of this study and secondary outcome was unintentional weight loss.A case of pneumonia was identified based on chest imaging and antibiotic prescription.Aspiration pneumonia is accepted as a part of the continuum of community-acquired pneumonia and hospital-acquired pneumonia rather than a distinct entity, and there have been no robust diagnostic criteria for the diagnosis of aspiration pneumonia 29 .In clinical settings, the diagnosis of aspiration pneumonia is generally made based on the combination of known risk factors for aspiration and involvement of characteristic location of the lungs in chest imaging 29 .In this study, all participants had known risk factors for aspiration (because they were referred for evaluating dysphagia), therefore, if the involvement of characteristic anatomical location of the lungs in the chest imaging with antibiotics prescription is confirmed, we regarded it as subsequent aspiration pneumonia.If the participant was discharged to another hospital and the chest imaging and antibiotic prescription could not be directly confirmed by researchers, it was checked by asking the participant or guardian if they had heard that from the medical staffs at the hospital they visited.Unintentional weight loss was defined as a loss of at least 5% of the usual body weight in < 3 months 30 .

Statistical analysis
This study is a preliminary study and required sample size was calculated using G*Power 3.1 31 .Assuming that the value of alpha to be 0.05, statistical power to be 95%, effect size to be 0.55, 39 participants were calculated to be needed.Due to the lack of preliminary studies on this topic, the effect size had to be chosen arbitrarily 32 .Assuming a dropout rate of 15%, a total of 45 participants were calculated to be needed.
Statistical analyses were performed using the Statistical Package for Social Sciences (version 22.0; IBM Corp., Armonk, NY, USA).The Wilcoxon signed-rank test was used to compare the VFSS findings (PAS, FDS, and number of swallows) for barium sulfate and iohexol.The Mann-Whitney test was used to compare the VFSS findings of each contrast agent in participants with or without subsequent pneumonia and unintentional weight loss.Receiver operating characteristic (ROC) curve analysis was performed to evaluate the predictive accuracy of VFSS findings for subsequent pneumonia and unintentional weight loss during the 3-month follow-up period.An area under the ROC curve (AUC) of 0.5, 0.7 to 0.8, 0.8 to 0.9, and more than 0.9 suggests no discrimination, acceptable diagnostic value, excellent diagnostic value, and an outstanding diagnostic value, respectively 33 .Statistical significance was set at p < 0.05.

Results
A total of 45 participants were enrolled to this study, and 23 participants were enrolled to the group where barium sulfate was used as a contrast agent prior to iohexol, and 22 participants were enrolled to the group where iohexol was used prior to barium sulfate during the VFSS procedures (Fig. 1).Among them, 15 participants could not complete the VFSS due to aspiration episodes more than once, according to the VFSS protocol.Finally, 30 participants were included in the analysis (15 in barium sulfate-first group and 15 in iohexol-first group), and 11 cases of subsequent pneumonia (36.7%), 13 cases of unintentional weight loss (48.1%), and three cases of death (10%) were recorded during the follow-up period.All deaths were caused by sepsis due to pneumonia, and the cases of subsequent pneumonia included all the participants who died.The incidence of unintentional weight loss was analyzed including only 27 participants who survived.Among the 13 participants with unintentional weight loss, seven (54%) also experienced subsequent pneumonia.At 3 months after the VFSS, 13 participants were on a normal diet (48.2%), 10 participants were on a modified diet (37%), and 4 participants (14.8%) were on artificial feeding using a nasogastric tube.The demographic data of the participants and 3-month outcomes are presented in Table 1; no significant differences in the demographic characteristics and 3-month outcomes were observed between the groups tested for barium sulfate first and iohexol first (Table 1).
When comparing the VFSS findings using barium sulfate and iohexol, no significant differences in PAS and FDS total scores were identified (Table 2).In the FDS subcategories, only "residue in pyriform sinuses" yielded higher scores when barium sulfate was used than when iohexol was used (p = 0.039).In addition, the number of www.nature.com/scientificreports/swallows required to maximally eliminate food from the pharyngeal space was higher when barium sulfate was used than when iohexol was used (p = 0.005).Table 3 compares the VFSS findings between participants with and without subsequent pneumonia, tested using barium sulfate and iohexol as contrast agents.When analyzing PAS scores, FDS scores, and number of swallows required to maximally eliminate food from pharyngeal space, only the FDS subcategory of "oral transit www.nature.com/scientificreports/time" (p = 0.048) and number of swallows (p = 0.005) when testing with barium sulfate had significantly higher values in participants with subsequent pneumonia than that in participants without subsequent pneumonia.When iohexol was used as a contrast agent, no significant differences in VFSS findings were observed between the participants with and without subsequent pneumonia.Table 4 presents comparison of VFSS findings between participants with and without unintentional weight loss during the 3-month follow-up period.When barium sulfate was used as a contrast agent, the FDS subcategories of "oral transit time" (p = 0.013) and "coating of pharyngeal wall after swallowing" (p = 0.027) indicated significantly higher values in participants with unintentional weight loss, as well as the number of swallows (p = 0.017).Meanwhile, when iohexol was used as a contrast agent, the FDS subcategories of "oral transit time", "nasal penetration", and "residue in the valleculae" demonstrated significantly higher values in participants with unintentional weight loss than in participants without (p = 0.033, p = 0.027, and p = 0.036, respectively).Moreover, the PAS scores for cup drinking and number of swallows also demonstrated significantly higher values in participants with unintentional weight loss (p = 0.016 and p = 0.006, respectively).
To predict subsequent pneumonia, the AUC of the number of swallows when testing with barium sulfate was 0.806 (p = 0.006) (Table 5).When predicting unintentional weight loss, the AUC for "oral transit time' and number of swallows when testing with barium sulfate were 0.736 and 0.766, respectively (p = 0.037 and p = 0.019) (Table 6).By contrast, when iohexol was used as a contrast agent in VFSS, the AUC of the PAS scores for cup drinking was 0.761 (p = 0.021), while that of the number of swallows was 0.805 (p = 0.007) (Table 6).

Discussion
No significant differences were observed in the PAS and FDS total scores according to the contrast agent used in the test.This might suggest that the difference in physical properties of the two contrast agents is not critical enough to make a difference in the anatomical depth to which food bolus penetrates when ingested.However, significant differences between the two contrast agents were seen in the FDS subcategories, scores for "residue in pyriform sinuses".The reason for this difference between PAS and FDS scores could be because the FDS was developed to supplement PAS and it can report more qualitative aspects of VFSS findings than PAS.Additionally, a greater number of swallows were required with barium sulfate than with iohexol.According to a previous study, iohexol demonstrated a faster transit time in gastrointestinal studies than barium sulfate because iohexol has higher water solubility 34 .Furthermore, barium sulfate is reportedly more viscous than iohexol 35 .Therefore, higher scores of "residue in pyriform sinuses" in FDS and a greater number of swallows when using barium sulfate than when using iohexol might be due to iohexol having higher water solubility and lower viscosity than barium sulfate.However, considering the small effect size of the difference between scores for "residue in pyriform sinuses", the result should be carefully interpreted.
Our study reveals that a longer oral transit time and greater number of swallows when testing with barium sulfate were associated with the development of subsequent pneumonia after VFSS.Of note, when these two www.nature.com/scientificreports/factors were analyzed using ROC analysis, only the number of swallows demonstrated an excellent diagnostic value for predicting subsequent pneumonia.According to previous studies, prolonged swallowing time, including oral transit time, is associated with an increased risk of aspiration pneumonia 6,36 .Moreover, the amount of post-swallowing residue in the pharyngeal space after swallowing is known to be associated with a greater risk of airway penetration or aspiration 37 .Our findings are consistent with those of previous studies; however, in the ROC analysis, a significant differences was observed only in the number of swallows when barium sulfate was used.The reason why the number of swallows emerged as a stronger predictor than PAS scores or residue after swallowing, which are well-known predictors of pneumonia, may be the VFSS protocol implemented in this study.There is still no clear consensus on when to measure pharyngeal residue after swallowing, for example, after first swallowing response or after multiple swallowing response.We measured the residue after the participants tried multiple swallows in order to maximally eliminate residues from the pharyngeal space through multiple swallowing.Consequently, the amount of residue might be underestimated during the VFSS compared to a typical meal time in a hospital room or at home because it is difficult to visually check the amount of residue in the pharyngeal space and this might hinder the caregivers from encouraging multiple swallows.Instead, the number of swallows may better indicate the impaired ability to eliminate food bolus from the pharyngeal space.
In addition, this significant result was found only when barium sulfate was used as a contrast agents; this might be due to the abovementioned different characteristics of the two contrast agents.
Our study also found significant differences in VFSS findings including oral transit time, coating of the pharyngeal wall after swallowing, and number of swallows in participants with and without unintentional weight loss when using barium sulfate.By contrast, when using iohexol, oral transit time, nasal penetration, residue in the valleculae, PAS scores, and number of swallows demonstrated significant differences between those with and without unintentional weight loss.Among these findings, oral transit time and the number of swallows when using barium sulfate had acceptable diagnostic values for predicting unintentional weight loss; however, the number of swallows when using iohexol may have an excellent diagnostic value for predicting unintentional weight loss, while the PAS scores when using iohexol had an acceptable diagnostic value.Similarly to subsequent pneumonia discussed earlier, due to the VFSS protocol of our study, oral transit time and number of swallows are thought to be able to better indicate the efficiency of swallowing compared to other metrics.Additionally, it is well known that unintentional weight loss is significantly associated with dysphagia 38 .Our findings are in accordance with those of a previous study in that seven out of 13 participants (54%) with unintentional weight loss also had subsequent pneumonia during the follow-up period.Moreover, the VFSS findings that revealed significant differences between participants with and without unintentional weight loss were consistent with those of previous reports 28,38,39 .Interestingly, the number of swallows was identified as a significant predictor of weight loss regardless of the use of either barium sulfate or iohexol as a contrast agent for VFSS, which has an acceptable diagnostic value for barium sulfate and an excellent diagnostic value for iohexol.This might be because the number of swallows reflects the global swallowing capacity and efficiency 28 .Another notable finding was that PAS scores were a significant predictor of unintentional weight loss only for iohexol.Considering that high-viscosity boluses have a reduced risk of aspiration during swallowing 40,41 , it might well be because barium sulfate is far more viscous than iohexol 35 .
To the best of our knowledge, this is the first study designed as a randomized, controlled, crossover trial to directly compare barium sulfate and iohexol as contrast agents for VFSS in predicting subsequent pneumonia and unintentional weight loss in patients with dysphagia.Because the two contrast agents were compared in the same participants, minimizing the influence of different patient characteristics on the results of the analysis was possible.Previous studies that compared barium sulfate and iohexol as contrast agents for VFSS focused on safety when the participants aspirated during the test, and since each contrast agent was administered to two different groups, the possibility of selection bias could not be completely excluded even if factors including age and sex were controlled.
However, this study has several limitations.First, the sample size was relatively small; future studies with larger sample sizes would provide more information for selecting the superior contrast agent between barium sulfate and iohexol.Second, underlying diseases of the participants were varied without restrictions.For example, although the characteristics and/or clinical course of dysphagia in patients with acute stroke and those with neurodegenerative diseases can be very different, our study was conducted without excluding specific diseases or targeting only specific diseases.To compare the ability of contrast agent to predict long-term prognosis of dysphagia, including various underlying diseases may have advantages in terms of generalizability, but the limitations due to this are also clear.Also, it should be noted that that although the statistical significance was not reached, more participants in barium sulfate-first group had pulmonary disease as medical comorbidity, which has close association with pneumonia and unintentional weight loss.Third, inability for blinding to the contrast agents used in the VFSS was another limitation of this study.Forth, the IDDSI levels of the two contrast agents were different, and this might well affect the study outcomes.Future study using equal IDDSI level of the liquid barium sulfate and iohexol would be necessary to better understand how the different physical properties of the two contrasts other than viscosity affect findings of VFSS.Fifth, subsequent pneumonia does not necessarily mean aspiration pneumonia.Despite efforts to include only aspiration pneumonia in the analysis, the possibility that other types of pneumonia may also be included cannot be completely ruled out.Sixth, although efforts were made to reduce the difference and make the conditions for testing the two contrast agents as similar as possible, muscle fatigue, which can be induced by performing multiple swallowing motions during the VFSS, may have had an effect for the results.To reduce the effect of fatigue, further study testing the two contrast agents in a differentiated participants would be needed.Last, we assumed that the previously administered contrast agent did not affect the subsequent VFSS findings; however, the lack of a washout period might have affected the study results.Although the previously used contrast agent was not detected using fluoroscopy, it might have affected the subsequent VFSS findings.

Conclusion
In conclusion, barium sulfate could be superior to iohexol in predicting subsequent pneumonia, whereas iohexol could be superior to barium sulfate in predicting unintentional weight loss after VFSS.However, considering that iohexol could be safer than barium sulfate when aspirated, it might be a better option as a contrast agent for VFSS.In addition, understanding the influence of the physical properties of the contrast agent on VFSS findings can help clinicians in making better clinical decisions.
presents the Consolidated Standards of the Reporting Trials (CONSORT) flow diagram.

Table 1 .
Demographic data of the study participants.

Table 2 .
Comparison of VFSS findings using barium sulfate and iohexol contrast agent.IQR interquartile range, FDS functional dysphagia scale, PAS penetration aspiration scale.a p-value was calculated by the Wilcoxon singed rank test.*p < 0.05.

Table 3 .
Comparison of VFSS findings between patients with and without subsequent pneumonia during 3-month follow up period.IQR interquartile range, FDS functional dysphagia scale, PAS penetration aspiration scale.a p-value was calculated by the Mann-Whitney test.*p < 0.05.

Table 4 .
Comparison of VFSS findings between patients with and without unintentional weight loss during 3-month follow up period.IQR interquartile range, FDS functional dysphagia scale, PAS penetration aspiration scale.
a p-value was calculated by the Mann-Whitney test.*p < 0.05.