Clinical reasoning is highly dependent on the practicing clinician (Jessee, 2018; Langridge, Roberts, & Pope, 2016). Cognition, training and experience are unique to each clinician, but are all influential within the clinical reasoning process (Norman et al., 2017). The role of these factors have informed various clinical reasoning models within the literature (Norman et al., 2017), that are broadly classified within two categories. The subconscious and conscious (Gummesson et al., 2018) models of clinical reasoning are discussed below.

The subconscious or pattern-recognition model assumes clinical reasoning is influenced by knowledge acquired from experience and intuition (Johansen & O’Brien, 2016). This model is informed by the premise that HCPs associate clinical problems presenting in current patients with problems seen previously and thus manages patients in a similar manner (Benner, Hughes, & Sutphen, 2008; Johansen & O’Brien, 2016). According to this model, novice clinicians initially rely on training and established guidelines to make clinical decisions. Progressively, they move towards intuitive reasoning as they become more experienced (Benner & Tanner, 1987; Jessee, 2018). In contrast, the conscious model, the hypothetico-deductive model, states that HCPs develop a hypothesis, based clinical information which is informed by training and experience (Abdallah, 2020; Trimble & Hamilton, 2016). This hypothesis is then exhaustively tested through several stages such as cue recognition, hypothesis generation, cue interpretation and hypothesis evaluation, to develop a clear management strategy (Abdallah, 2020; Trimble & Hamilton, 2016).

Clinical reasoning is determined by various factors related to context and political consciousness (Kathard & Pillay, 2013; Young et al., 2020). This is particularly true for LMICs all over the world, who are haunted by political and economic injustices (Browne et al., 2016; Lavoie et al., 2018; Umeh & Feeley, 2017; Ziltener & Künzler, 2013). These injustices have resulted in several consequences such as severe inequity, high unemployment rates and insufficient health systems (Baatiema et al., 2020; Browne et al., 2016; Ruelas, Gómez-Dantés, Leatherman, Fortune, & Gay-Molina, 2012). This in turn has caused unsurmountable personnel and resource constraints, high patient numbers, as well as a complex burden of disease (Ludwick, Morgan, Kane, Kelaher, & McPake, 2020; Mayosi et al., 2012; Meng et al., 2020; Puchalski Ritchie et al., 2016). These constraints have forced HCPs to practice in a creative and flexible manner to combat the effect of these inequities on their patients (Baatiema et al., 2020; Saito et al., 2016; Van Graan, Williams, & Koen, 2016). For perceived non-essential services such as speech therapy, these difficulties are often compounded because of their lack of priority when disseminating resources (Andrews & Pillay, 2017; Bright, Wallace, & Kuper, 2018; Rech, Hugo, Schmidt, Goulart, & Hilgert, 2019). As a result, according to Andrews and Pillay (2017), Pierpoint and Pillay (2020) and Modi and Ross (2000), many Speech-Language Therapists (SLTs) chose to alter their practice patterns to cope with the demands of the contexts in which they work, to allow for best possible practice within challenging conditions (Doeltgen et al., 2019; Saito et al., 2016).

According to Kathard and Pillay (2013), another facet of contextual influence is determined by political climates and historical biases. This accounts for how political, social, cultural and systemic structures impact clinical reasoning (Kathard & Pillay, 2013). For example, HCP’s clinical reasoning is informed by knowledge acquired from professional training (Doeltgen et al., 2019). Historically, academic instruction has subscribed to a privileged approach to knowledge which has been taught under the guise of universality (Dawson, 2020; Grosfoguel, 2013; Kathard & Pillay, 2013). This is problematic as a rigid approach to knowledge which gives deference to colonial ideals cannot account for the complex social, economic and historical realities of the rest of the world (Dawson, 2020; Ndlovu-Gatsheni, 2018). Therefore, it is important for us to understand the influence of these biases on clinical reasoning to assist in managing patients in a manner that is more relevant and socio-politically aware (Findyartini et al., 2016; Kathard & Pillay, 2013; Zahir, Miles, Hand, & Ward, 2020).

In this article, we have chosen to foreground the CSE given its subjective nature which relies on clinical reasoning (Garand, McCullough, Crary, Arvedson, & Dodrill, 2020). The CSE is an assessment which evaluates patients to make inferences about the presence, nature and cause of dysphagia within the oral and pharyngeal phase of the swallow (Carnaby-Mann & Lenius, 2008; Garand, McCullough, Crary, Arvedson, & Dodrill, 2020). This is achieved through chart reviews, patient observations, oral motor examinations and multiple swallow trials (Doeltgen et al., 2019; Garand et al., 2020). The CSE is a detailed assessment which informs hypotheses regarding the anatomy and physiology of the swallow (Rangarathnam & McCullough, 2016). This assessment is necessary as it allows SLTs to make decisions regarding oral intake and consistency, additional compensations and ultimately the need for instrumental assessments (Namasivayam-MacDonald & Riquelme, 2019; Virvidaki, Nasios, Kosmidou, Giannopoulos, & Milionis, 2018). There is significant controversy regarding the utility of the CSE as a stand-alone assessment to guide dysphagia management (Doeltgen, McAllister, Murray, Ward, & Pretz, 2018; Riquelme, 2015), given its reliance on perceptual information and varied practice patterns (Riquelme, 2015). However, the CSE is still the most commonly used form of dysphagia assessment (Virvidaki et al., 2018). It is regarded as a time efficient, cost-effective and non-invasive assessment, which allows therapists to understand a patient’s dysphagia in relation to their medical history and displayed symptoms (Carnaby-Mann & Lenius, 2008; Riquelme, 2015; Virvidaki et al., 2018).

The studies included were those which:

Information for this review was sourced from both electronic and physical journals. The electronic databases searched were: South African Bibliographical Network (SABINET), EBSCOhost, Medline, PubMed, Clinical Key, Google Scholar and Cochrane Library. These databases were selected as they are highly reputable and extensive, which ensured that all the relevant literature was identified. Furthermore, published conference papers of American Speech and Hearing Association (ASHA), and South African Association of Speech, Language and Hearing association (SASLHA), as well as postgraduate dissertations available on ResearchSpace were all searched to identify grey literature. The final electronic search was conducted on 03 June 2019.

Handsearching was performed at three South African university libraries. The Biomedical Journal, Academic Medicine, Medical Education, South African Journal of Communication Disorders, International Journal of Speech and Language, New England Journal of Medicine, and Journal of Medical Research were searched from 1970 to 2019 for keywords via table of contents and subject indexes. The final manual search was conducted on 07 June 2019.

The JBI’s (Peters et al., 2020), three-step search strategy was utilised to guide a comprehensive search of the literature. This included searching two databases namely, Google Scholar and MEDLINE, for articles which related clinical reasoning to the CSE to identify keywords contained within the title, abstract, and index (Moola et al. 2015). The keywords identified through this strategy were: clinical reasoning, clinical decision-making, CSE, dysphagia and speech therapy. Following this, a second search was carried out to determine which combination of these keywords would yield the most relevant results across databases. The final combination of keywords used were: [clinical reasoning] OR [decision-making] AND [dysphagia assessment] OR [swallow assessment]. These terms were placed into the databases using advanced searching options to filter articles by date and search terms. Finally, reference lists of relevant articles were searched post data collection to ensure no relevant information was missed.

Two reviewers blindly screened the titles, abstracts and full articles of the search results independently. Reviewer 1 performed the manual search and screening, and Reviewer 2 searched and reviewed the six electronic databases. Upon completion of the screening processes, the reviewers met to discuss the eligibility of each article for the dataset. The suitability of each of the articles was evaluated based on the inclusion criteria. An agreement rate of 85.7% was achieved between the reviewers. In the two instances where the reviewers could not reach a consensus, a third reviewer was consulted to make the final decision.

The data extraction and charting process was iterative, allowing for the dataset to evolve as useful elements were identified (Moola et al., 2015; Schultz et al., 2018). The data charted included: author, year of publication, country, aims, study population and size, methodology, intervention type, outcomes and key findings. The extracted data served as a basic summary to help orient readers to the nature of the research collected.

Thematic analysis as described by Braun and Clarke (2006), was utilised to provide a rich and complex account of the data. Braun and Clarke (2006), described a six-step recursive method for thematic analysis which included: (1) familiarisation, (2) generation of initial codes, (3) searching for and (4) reviewing themes, (5) production of final themes, and (6) a report. Furthermore, the computer analysis software NVivo 12 was used to support data coding processes by streamlining complex functions like syntactical word recognition, word frequency counting, coding and the creation of visual representations of the data in the form of word clouds, tree diagrams and word matrices.

Finally, the researcher chose to include an optional consultation as described by Colquhoun et al. (2017), within the third phase of this review. Although largely unknown, the consultation phase is a historic (Arksey & O’Malley, 2005; Levac, Colquhoun, & O’Brien, 2010) and current (Faulkner, Taylor, Ferrence, Munro, & Selby, 2006; Faulkner et al., 2011; Khan et al., 2015; Moore et al., 2019; Myers, Schaefer, & Coudron, 2017; Smartt et al., 2019; Youssef, Chaudhary, Wiljer, Mylopoulos, & Sockalingam, 2019; Zwiep et al., 2018) feature of scoping reviews which engages stakeholders to understand the implications of a review’s findings within practice. The aim of this consultation phase was to contextualise the data to South Africa. The reliability, value and applicability of the results were addressed within the context of healthcare in South Africa. Four dysphagia practitioners were consulted through a focus group to allow for a multidimensional discussion around the collected data (Nyumba, Wilson, Derrick, & Mukherjee, 2018). The focus group was conducted using open-ended questions surrounding the applicability of the results to the South African context. Upon completion, the focus group was transcribed and thematically analysed in relation to the themes of the literature search. This allowed the researchers to highlight the applicability of the results in the context of South Africa.

This study was approved by the Biomedical Research Ethics Committee (BREC) of the University of KwaZulu-Natal (BREC Reference number: BE275/19).

Figure 1 depicts the data collected through both the electronic and manual searches. There were 49,000 articles identified in the initial searches across the six databases, however there were extensive numbers of duplicates between the searches. After the duplicate studies were removed, 8,800 studies were screened. Of the total number of articles screened by title, only approximately 0.14% of the articles were included within the final review. This could demonstrate a poor focus within the literature on clinical reasoning within the CSE.

Appendix 2 represents the characteristics of the 12 studies included in the study. Each of the selected articles made explicit references to the use of clinical reasoning within the CSE as per the eligibility criteria. The studies were published between the period 1999 and 2019. The methodologies of the study included: two literature reviews, seven descriptive survey-based studies, two intervention studies and one case study. All the included studies were conducted in largely urbanised, high-income countries. Four studies were carried out in Australia, two in Ireland, four in the United States of America, one in Canada, and one in England. This influenced the narratives which could be explored given the homogenous economic context of the dataset (see Appendix 2).

The results of the review revealed three main themes related to the value and use of clinical reasoning within the CSE. The results of both the review and the consultation highlighted several key issues including: the importance of clinical reasoning, the impact of training and experience, the importance of individualised service delivery and the value of flexible and critical reasoning within dysphagia assessment. Each of these themes will be elaborated on further in the discussion below.

The review highlighted the importance of knowledge acquired from training and experience in the clinical reasoning process. As expected, university training was described as influential in the development of clinical reasoning skills. As mentioned previously, training is the foundation on which all clinical knowledge is built and therefore impacts how all decisions are made (Doeltgen et al., 2019). According to seven of the articles within the dataset, a solid theoretical understanding of assessment and management of dysphagia is needed to develop appropriate knowledge schemas to guide reasoning (Brodsky, Mayfield, & Gross, 2019; Doeltgen et al., 2018; Jones et al., 2018; Mathers-Schmidt & Kurlinski, 2003; McAllister, Kruger, Doeltgen, & Tyler-Boltrek, 2016; McCurtin & Carter, 2015; McCurtin & Healy, 2017).

However, an interesting finding was the focus on experience rather than training as the guiding factor within clinical reasoning. Several articles within the dataset referred to the level and quality of experience amongst SLTs as being the preferred factor to guide clinical reasoning (Brodsky et al., 2019; Doeltgen et al., 2018; Jones et al., 2018; Mathers-Schmidt & Kurlinski, 2003; McAllister et al., 2016; McCurtin & Carter, 2015; McCurtin & Healy, 2017; Pettigrew & O’Toole, 2007; Rumbach et al., 2018; Walshe, Ryan, & Regan, 2017). Table 1 demonstrates this with several extracts from the dataset which support the use and preference of experience in clinical reasoning. Experience was described in terms of the Recognition-Primed Decision Model (RPDM), which theorises that HCPs make decisions through the recognition of patterns (Klein, 1999). All 12 articles described pattern recognition as useful, as it allows HCPs to quickly process information to develop a small number of plausible diagnoses. This preference for pattern-recognition was attributed to it being a resource- and time-efficient process whilst avoiding haphazard or inaccurate decisions. The RPDM relies heavily on intuition, which is described as ‘the immediate realisation of risk, usually learnt […] without conscious awareness’ (Farr-Wharton, Brunetto, & Shacklock, 2012, p. 1392). According to articles 1 (Doeltgen et al., 2018), 7 (McCurtin & Carter, 2015) and 11 (McCurtin & Healy, 2017), intuition is developed through diverse and high-quality clinical experiences. Article 1 (Doeltgen et al., 2018) explains that these experiences help SLTs to develop complex networks of illness scripts and patterns of symptoms that are relied on to guide decision-making. Furthermore, articles 1 (Doeltgen et al., 2018) and 7 (McCurtin & Carter, 2015) highlighted that the amount of experience which HCPs have, impacts on both the speed and accuracy of their pattern-recognition approach. Novice HCPs have less information within their knowledge schemas to develop diagnoses at an intuitive level. Therefore, they have to engage in an exhaustive conscious approach to make diagnoses (Klein, 1999).

The sentiment that experience is arguably more influential than training was echoed in the results of the consultation phase. This was attributed to inconsistency and irrelevance of training of SLTs at tertiary institutions. The results showed that practitioners preferred to rely on their own experiences to guide their approach to dysphagia management. As seen in Figure 2, the focus group felt that universities focused more on a theoretical understanding of dysphagia and trained clinicians in a linear outcome-based model which is inappropriate to the LMIC contexts that they find themselves in. They reported that this inappropriateness of training was exacerbated by insufficiency in opportunities to develop these clinical skills in real-life scenarios. These sentiments are echoed in recent literature which describes the doubt and under-preparedness voiced by many novice SLTs in response to their training (Caesar & Kitila, 2020; Coutts, 2019; Singh et al., 2015).

Globally, SLTs often work in a variety of settings as seen in Figure 3. As a result of this diversity in setting, every article within the dataset highlighted contextual factors as a major contributing factor to clinical reasoning which should not be discounted. Article 7 describes this by stating: ‘the clinical decisions [which] health professionals [make] are strongly influenced by the context in which they are made’ ( McCurtin & Carter, 2015, p. 72). This makes sense given that context often determines resource availability, prioritised health goals, institutional policies and popularised assessment practices (Schwarz,Coccetti, & Cardell, 2020). As a result of this increased variability, HCPs are required to engage in assessment in a highly nuanced and complex manner. Therefore, we can deduce that the decision which clinicians make cannot be removed from the context in which it occurs. For example, the data showed a discrepancy between acute/subacute and community-based settings as demonstrated in Figure 4. SLTs favoured instrumental assessments within the acute/subacute settings. This higher frequency can be attributed to equipment usually being more accessible as well as the complex nature of the patient population. Conversely, SLTs in community settings were more likely to rely on the CSE to guide their decisions as instrumental assessments are considered costly and difficult to access. This is in line with the literature explored earlier by Virvidaki et al. (2018) and Namasivayam-MacDonald and Riquelme (2019).

The results of the consultation frequently described how influential context was in making clinical decisions. The clinicians felt that clinical reasoning was imperative to ensure that the management of patients reflected both the needs of the population as well as best practice even in challenging conditions. There was consensus that because of the complex quadruple burden of disease, severe resource constraints and high patient loads which are mirrored in most LMICs, SLTs have to alter the way in which they practice to cope with demands of their settings (Andrews & Pillay, 2017). This was described to be unique to LMICs as often there are fewer confounding variables to consider within high income contexts. Therefore, they felt that SLTs often relied on clinical reasoning when developing context-specific assessments which are crucial in effective and equitable service delivery.

The dataset (as seen in Table 2) showed that SLTs have moved to a purposeful and process-driven approach to assessment which allows them to easily adapt to conditions. Articles 1 (Doeltgen et al., 2018), 4 (Jones et al., 2018), 7 ( McCurtin & Carter, 2015) and 12 (Martino , Pron, & Diamant, 2004) suggested that this has occurred because SLTs have a network of core assessment components which are influenced by a hierarchical model. According to this hierarchy, SLTs begin with the most common CSE components and then move on to less utilised components as required. This process continues until the clinician has developed a satisfactory diagnosis. González-Fernández, Ottenstein, Atanelov and Christian (2013), corroborated this idea by stating that component omission by SLTs was purposeful and non-random. This suggests that component selection in the CSE is reflective of the practice philosophy amongst SLTs who ‘[do] what works best for their patients, in their setting’ (Jones et al., 2018, p. 73). In addition, it is clear to see the positive impact which clinical reasoning has on the utilisation of the CSE. Within complex cases better outcomes are reported when SLTs employ high-quality clinical reasoning processes to guide the selection of assessment components. This conclusion is supported by Odderson, Keaton and McKenna (1995), who state that:

The positive impact of clinical reasoning includes assisting SLTs in guiding management decisions, considering multiple factors, preventing unnecessary assessment and synthesising information from multiple sources.

This positive impact was extrapolated within the results of the consultation phase. The focus group agreed that there was substantive value in clinical reasoning especially within the complex socio-economic context of South Africa as an example of an LMIC. The participants reasoned that clinical reasoning was a key tool which clinicians used to adapt dysphagia assessment protocols to make them relevant to their context. They felt that this study was invaluable in creating a link between research, service delivery and training to meet the needs of the complex and diverse communities within LMICs in an effective manner.