The Influence of Food Texture and Liquid Consistency Modification on Swallowing Physiology and Function: A Systematic Review

Collectively, the selected studies clearly show a reduction in the risk of penetration–aspiration with liquids, as they progress from the thin to the very thick end of the viscosity continuum. This finding is limited, by definition, to studies in which objective measures of penetration–aspiration were available, which for the purposes of the present review meant studies involving barium swallowed under videofluoroscopy. Evidence regarding penetration–aspiration was also limited to studies involving adult participants with dysphagia.

However, an important cautionary note arises from this review given the convergence of evidence across several studies, showing a heightened risk of post-swallow residue in the pharynx for liquids with higher viscosities. This points to an important clinical challenge in terms of identifying suitable and safe consistencies for patients with dysphagia; namely, that of identifying liquids that are thick enough to be swallowed safely (without penetration–aspiration) while avoiding the pitfall of post-swallow residue.

As a post-script on this particular question, an additional source of data was brought to the attention of the authors after completion of the qualitative synthesis of the selected articles. This as-yet unpublished doctoral dissertation [72] involved rigorous videofluoroscopic exploration of swallowing with thin and nectar-thick Varibar™ barium by infants aged 3 weeks to 3 months, referred for evaluation of swallowing secondary to respiratory compromise. This dissertation is particularly noteworthy given the relative dearth of information regarding pediatric feeding and swallowing uncovered in our search process [17‐19]. The Gosa study [72] reports several findings that concur with the observations gleaned from our qualitative synthesis, including prolongations of oral transit time and a reduction in penetration–aspiration with the nectar-thick stimulus compared to the thin liquid barium. Additionally, residue was reported to be present for 80 % of the nectar-thick swallows compared to only 44 % of thin liquid swallows. An additional gap to highlight with respect to the lack of identified studies in either the healthy or dysphagic infant population is the challenge of matching assessment stimuli to the rheological properties of breast milk or infant formula. This is a question of emerging interest in the dysphagia literature [9, 73] and definitely an area where additional research is needed.

Although this systematic review finds a convergence of evidence showing that thicker liquids are less likely to be aspirated, and more likely to cause post-swallow residue, the available data are insufficient to suggest particular viscosity values or other quantitative measures of material properties along the continuum from thin to extremely thick liquids that represent boundaries of clinical importance [74]. Historically, clinical guidelines regarding the use of thickened liquids have proposed quantitative boundaries that arise either from clinical consensus, or represent an educated guess. For example, the National Dysphagia Diet in the United States proposes 4 levels of liquid viscosity, labeled “thin”, “nectar-thick”, “honey-thick” and “spoon-thick” and corresponding to apparent viscosity ranges of 1–50, 51–350, 351–1,750, and ≥1,751 mPa s, measured at a shear rate of 50/s [10]. The Japanese guideline provides a larger number of categories with viscosity ranges of 1–50, 51–150, 151–300, 301–500, and >500 mPa s, again measured at a shear rate of 50/s [14, 15]. In the current review, we found no specific evidence to support or refute the specific numeric categorical viscosity boundaries suggested in these or other guidelines. We found no evidence to suggest that there are transitions of clinical relevance occurring at the boundaries between categories in these guidelines. Furthermore, the available evidence to date does not help us to ascertain how large a viscosity difference needs to be, in order to have a beneficial and measurable effect with respect to reducing penetration–aspiration, nor at what point the risk of residue accumulation becomes a real concern. Similarly, the available evidence does not provide clear evidence to suggest how many incremental levels of increasing viscosity might be meaningful in the clinical context. The available data also lack evidence regarding the important possibility that properties of a liquid bolus other than apparent viscosity, such as density, yield stress, cohesiveness, or slipperiness (to list a few) might influence swallowing physiology and function. On this latter point, we are aware of a recent publication describing differences in the rates of occurrence of penetration–aspiration for liquids, depending on the type of thickener used (starch vs xanthan-gum), albeit thickened to different degrees [75]. A recent conference abstract also reports differences in residue accumulation for liquids thickened with corn-starch versus xanthan-gum thickeners, and attributes these differences to subjectively judged cohesive properties of the bolus [76]. Similarly, several articles reviewed in this study revealed that different thickening agents produced products that had different rheological or material property profiles, as suggested in prior studies [60], and were shown to require different degrees of oral processing and suggested to have different rates of flow [43, 44]. Thus, it is naïve and not appropriate to assume that liquids thickened to similar viscosities using different agents will behave similarly in the oropharynx. The possibility that properties other than viscosity may have clinical relevance is both intriguing and important, and poses a challenge to the scientific community to develop rigorous studies that characterize such properties according to validated methods, in order to explore such phenomena.

Given the recognition that particular numeric viscosity boundaries for levels of liquid thickening are neither empirically supported nor empirically refuted, we conclude that the most appropriate clinical course of action with respect to identifying the optimal consistency of liquids for a patient who aspirates thin liquids is to increase viscosity in relatively small increments until swallowing safety is demonstrated. How large these increments might need to be can perhaps be informed by evidence from the sensory literature, based on the assumption that changes in behavior result from perceived differences in bolus consistency. The literature shows that the scaling of oral viscosity perception does not grow in a linear manner, but rather in an exponential fashion [77, 78]. Human ability to discriminate viscosity is proportional to the viscosity of the sample itself, as described by Weber’s law [79]. Recently, Withers and colleagues [80] manipulated the cream/fat content, and viscosity of skimmed milk using 0.1 % w/v increments of a starch-based thickener to explored the thresholds of just-noticeable differences (JND) in perceived thickness by healthy adults. They found no age differences in viscosity discrimination for liquids with apparent viscosities between 45 and 135 mPa s at 44/s, and reported that the average JND was between 0.26 and 0.32 % w/v in terms of thickener concentration, which equated to approximately a 1.8-fold increase in apparent viscosity (i.e., from 45 to 83 mPa s). Another recently published study explored just-noticeable differences using 0.1 % w/v increments in the concentration of a xanthan-gum thickener for sweetened cordial stimuli between 190 and 380 mPa s at 50/s [56]. In this case, the JNDs were found to be narrower, namely 0.38-fold for thickener concentration, equating to a 0.67-fold increase in viscosity [56]. Differences in the nature of the liquids studied (i.e., dairy products vs cordials), the viscosity ranges probed (i.e., 45–135 vs. 190–390 mPa s) and the testing methodologies used (i.e., two-stimulus forced choice comparisons versus three-stimulus triangle test paradigms) may have contributed to the observed differences in the resolution of perceivably different viscosities across these two studies. The authors of the second study [56] extrapolated from their findings to suggest that an array of liquids with apparent viscosities of 5, 8, 13, 22, 36, 60, 100, 170, 280, 470, 790, 1320, and 2200 mPa s at 50/s might provide a starting point for evaluating the influence of perceivably different viscosities on swallowing. However, it should also be noted that all of the viscosity discrimination studies cited [56, 77, 78, 80] were conducted using healthy volunteers with intact sensory systems. Alterations to oral or pharyngeal sensation, such as may be seen in individuals with dysphagia secondary to stroke, may alter perceptual viscosity discrimination abilities compared to healthy individuals. It should also be noted that the perceptual thresholds for noticeable viscosity differences arising from both of these studies are smaller than the categorical boundaries suggested by current clinical guidelines. Certainly, the results of our qualitative synthesis point to a significant gap both in literature and knowledge regarding the impact of small increments of viscosity on swallowing, and illustrate the need for new studies, which explore both the physiological and functional consequences of thickening in both narrow and larger increments.

As a final comment in this section on thickened liquids, some attention is required on the issue of viscosity measurement and its dependence on shear rate, which is a quantified measure of the speed of flow. The addition of either starch or xanthan-gum thickeners to liquids will result in non-Newtonian characteristics [56‐62], meaning that the apparent (measured) viscosity is strongly dependent on shear rate. This is also true of most, but not all contrast media used for swallowing evaluation [54, 55]. Thus, if a measure of viscosity is reported, the shear rate used in the measurement is critical to understanding the measurement and to enabling comparison across studies. Although it has been common to report apparent viscosity at a shear rate of 50 reciprocal seconds (i.e., 50/s) [10, 14, 56], none of the studies identified for inclusion in this systematic review followed this convention. Notably, available information regarding the viscosity of Varibar™ (a line of commercially available barium products for swallowing evaluation used frequently in the United States) is quoted at a shear rate of 30/s. The actual shear rates involved in oral processing and swallowing depend on the rate and degree of pressures applied as well as the material properties of the fluids, and as such, can vary widely. The oral preparatory phase probably involves low shear rates (particularly for thicker fluids) and the perception of thickness has been shown to be best-related to objective measurements of viscosity taken at 10/s [64]. The pharyngeal and esophageal stages of swallowing are thought to involve much more rapid flow, with computer simulations suggesting shear rates in the order of 400/s for water [81].

Until such time as new research is available to describe the shear rates that are actually operating during swallowing [82], both in healthy and impaired contexts, it is paramount that apparent viscosities of thickened liquids intended both for assessment and therapeutic clinical purposes be reported across a range of shear rates. As a starting point, we recommend that shear rates of 1, 10 [64], 30, 50, and 100 reciprocal seconds would provide a reasonable basis for comparison. We particularly encourage consideration of liquid flow behaviors at low shear rates due to the likelihood that motoric deficits in dysphagia may impact a person’s ability to generate the shear forces and physiological behaviors that are typical of healthy swallowing.

If the literature on thickened liquids is sparse, this is even more apparent when reviewing the literature regarding texture-modified foods and swallowing. As illustrated in Table 5, the identified literature discussed only a small number and variety of solid foods. With the exception of longer duration and higher amplitude masseter surface electromyography signals when ingesting solid foods with increasing thickness or hardness [19, 30, 36‐38, 43, 44, 46], the findings of the identified studies do not clearly point to measurable differences in either oral processing or swallowing parameters across the particular solid foods tested. We did not, for example, find literature that specifically explored the particle size of solid foods after a specific timeframe of chewing by people with partial or missing dentition or with reduced chewing strength. Penman and Thomson suggest that particles of 1.5 cm² constitute a choking hazard for people with dysphagia [83], but the studies that we found describing the characteristics of solid foods after oral processing focused more on textural profiling than on particle size. Although this information may exist in the dental or food oral processing literature, it was not found given the specified search strategy, and, as previously acknowledged, the term “chok*” was not included in our search. Data regarding solid food particle size after oral processing under both normal and abnormal dental conditions would be interesting to consider alongside autopsy results suggesting that individuals with partial or missing dentition are more prone to choking on food [84]. A recent report by the Japanese Food Safety Commission [85], concludes that food texture (surface smoothness, elasticity, hardness), size, and shape are all relevant with respect to choking risk. In their investigations, sticky rice cakes were found to be the leading cause of choking accidents, but jelly cups were also mentioned as a not infrequent cause of choking. The report highlights that the risk of choking on a particular food item needs to be understood both in terms of the textural properties of the bolus and of the physiological behaviors commonly used during ingestion of that item. Thus, the jelly cups, for which they describe a common behavior of tilting the head backwards to suck the jelly out of the cup, are not without risk.

The review revealed common use in the food oral processing literature of accepted terminology to describe the textural attributes of solid foods as laid out in ISO guidelines [68, 69]. The construct of cohesiveness, mentioned earlier, is one example of such terminology. One term, which was encountered in the food oral processing literature, but which remains poorly understood, is “ease of swallowing” [86]. This appears to be an attribute that is commonly captured in sensory profiling of food textures; however, whether and how this attribute maps to objective, quantifiable measures of bolus flow or physiology remains unclear.

An interesting question arising from this review is whether adhesive paste consistency stimuli such as cheese spread or peanut butter should be considered to be semi-solid foods or extremely thick liquids? These items can be compressed and spread across the palate with the tongue, and do not fracture; as such, in a physiological sense, they behave quite differently and involve different oral processing behaviors from foods that require mastication [87]. On the other hand, they do not flow either under gravity, or under the typical pressures applied by the tongue, and require handling by the tongue for transport through the oral cavity. In this respect, they are quite different from liquids. It is recommended that future investigations with respect to differences in oral processing and swallowing of solid foods and thickened liquids make a clear distinction based on the physiological processes that are required for transport (i.e., mastication, oral containment, tongue-sweeping, and propulsion or simple tongue compression), rather than using texture descriptors derived based on physical properties alone [88]. Furthermore, it may be important to note that a given stimulus may behave more like a liquid for one person and a semi-solid for another person, based on the person’s ability to generate forces or movement with their tongue. As such, physiological definitions of texture may have different boundaries for different consumer groups. Considerations of temperature inside the mouth and the slipperiness of the oral surfaces given differences in the levels of saliva across the duration of oral processing are undoubtedly also relevant to developing a texture classification system for the dysphagia population, which is founded on a physiological framework.

The paucity of studies captured in our search describing oral processing or swallowing of texture-modified foods comes as both a disappointment and a surprise. On reflection, we believe that the rules of our search strategy may have overly limited the search results, given mandatory inclusion of the MeSH terms “Swallowing” or “Deglutition” or “Dysphagia”, even when the supplementary search for articles was performed with the additional MeSH heading term of “food texture”. We are aware that there is an entire field of scholarship known as “food oral processing”, with its own journals and conferences. Although our search strategy employed search engines intended to tap the engineering and non-medical domains, it may well be that key words related to swallowing and dysphagia are not commonly used in publications within this subspecialty, leading our search to capture only a limited number of articles from this domain. Certainly, a direction for future research will be to explore this literature in greater detail for relevant evidence regarding differences in oral processing behaviors for foods with different textural characteristics.

From a clinical perspective, the lack of guidance regarding the classification, labeling, and preparation of texture-modified foods for people with dysphagia is a concern. It is not uncommon for coroner’s inquiries into fatal choking episodes in people at risk for dysphagia to conclude that food of an inappropriate consistency was ingested [84, 89‐91]. On the basis of the current review, we are obliged to point out that the best available evidence regarding the selection of an optimal food consistency for a person with dysphagia comes from the careful exploration of tolerance for different foods in a comprehensive clinical swallowing assessment. This systematic review found a lack of research evidence providing support for the selection or avoidance of specific consistencies. Our review points to an urgent need to generate empirical evidence to describe different classes of chewable food, so that the corresponding expected differences in oral processing and swallowing behavior can be defined. Additionally, the development of valid methods for observing, describing, and measuring oral stage behaviors during assessment tasks that probe a variety of different solid foods would be a valuable addition to current subjective clinical methods. Collaboration with the research field food oral processing is strongly advised as a first step in developing such methods.