Background

Telephone triage is increasingly used to navigate patients to the appropriate level of care and manage the patient flow [1, 2]. Telephone triage involves assessing the patient’s symptoms, determining the level of urgency and type of healthcare needed and providing self-care advice, if appropriate. A review shows that approximately 50% of the calls handled by nurses or doctors in medical call centres in primary care can be handled with self-care advice alone [3], consequently reducing the pressure on the services [4].

The same review also concluded that there is no greater safety concern with telephone triage than with traditional face-to-face care [3]. However, studies show that the accuracy of decisions is positively associated with high-quality communication [5, 6], and conversely, inadequate communication can reduce patient safety [7,8,9,10]. In addition, the caller’s satisfaction with the conversation is associated with an increased likelihood of following the medical advice given [11]. Thus, ensuring good communication is a way of ensuring patient safety and a good quality of the healthcare delivered.

Good communication in telephone triage has been described in various communication assessment tools developed via Delphi processes and with professional grounding [12,13,14]. While studies that have evaluated actual calls show noticeable variations in quality [5, 6, 15], limited attention has been directed to exploring the factors that contribute to the observed variation. Therefore, this review aims to identify factors affecting communication during telephone triage in emergency primary healthcare and to describe how these factors affect communication.

Methods

Design

A mixed-method systematic review was chosen to encompass findings from all types of primary studies. The review was performed using the 24-step guide for systematic review published by Muka et al. [16] as a guideline and reported according to the Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) checklist [17]. The checklist is available in Additional file 1. The study protocol was registered in PROSPERO January 2021 (CRD42022298022).

Eligibility criteria

To identify and describe the research question, we used the SPIDER methodology (Sample, Phenomenon of Interest, Design, Evaluation, Research study) [18] (Table 1).

Table 1 SPIDER specifications
Full size table

Original studies with qualitative, quantitative, and mixed-method design were included if (a) they described factors affecting communication between callers and operators in medical call centres, (b) the setting was medical call centres in primary care that managed all types of medical conditions from an unselected population, (c) they were published in English and (d) the full-text version was available.

Search strategy and identification of studies

The search strategy was designed by three of the review group members (SLSF, VM, JV) and a research librarian (HW). The search was conducted in the databases MEDLINE (Ovid SP), Embase (Ovid SP), Cumulative Index to Nursing and Allied Health Literature (CINAHL) and Web of Science by the librarian on 15 April 2021. A search for new literature was conducted by another librarian (ISKS) on 26 June 2023, with the same search strategy (Additional file 2). In addition to the database search, the reference list of included studies was manually screened for supplementary literature.

Selection of studies

Search results from the different databases were combined in an EndNote library file and uploaded to the systematic review management tool Covidence [19]. Duplicates were removed before two authors independently reviewed the titles and abstracts of the studies included. The authors voted for either inclusion or exclusion and were blinded to the other author’s vote during the screening process. When there was disagreement on inclusion/exclusion, an additional author reviewed the study and gave a third vote. The same strategy was used for the full-text screening. Each step of the screening process was conducted by the first author, SLSF, in collaboration with one or several of the co-authors, VM, JV and IHJ.

Data collection, synthesis and quality assessment

Information about the first author, title, year of publication, country, and characteristics of the study (design, objective(s), sample size, study population characteristics, setting) was retrieved in Covidence by SLSF. A results-based convergent synthesis design was used to extract and synthesise the data [20]. The qualitative findings from the studies included were analysed using thematic synthesis, following three steps: free line-by-line coding, generation of descriptive themes and generation of interpretative/analytical themes [21]. The web-based programme EPPI reviewer [22] was used for free line-by-line coding. To create consensus on the extraction, the entire author group read a selection of studies and independently identified factors that were later discussed for agreement at joint meetings. SLSF then coded identified factors in all the studies and extracted all coded text from the EPPI reviewer into Excel files, where each theme was summarised and described.

The factors (analytical themes) identified in the qualitative studies constituted a framework for analysing the quantitative studies. The qualitative and quantitative results were presented individually and subsequently consolidated as an integration of findings. Throughout the process, the group of authors collaborated in discussions to overcome challenges and determine the next course of action.

Methodological limitations for each study were independently assessed and then discussed by SLSF and IHJ using the Mixed Methods Appraisal Tool (MMAT) [23].

Results

The search yielded 6620 studies. After removing duplicates, 5087 titles and abstracts were screened. Only 173 studies were assessed in full text, and 62 studies were finally included in the review (Fig. 1). The 62 studies included comprised 40 qualitative studies, 16 quantitative studies, and six studies using mixed methods. Characteristics of each included study are available in Table 2. In the mixed methods studies, factors affecting communication were found in the quantitative part of one study [24], the qualitative part of three studies [25,26,27] and in both parts of two studies [28, 29]. The year of publication ranged from 1990 to 2023 (Fig. 2). The studies were performed in eight countries, with Sweden as the main contributor, accounting for 47% of the studies (Fig. 3). The primary data sources for the qualitative studies comprised interviews with either the operator or the caller. Additionally, open-ended survey questions and audio recordings of interactions between operators and callers were employed. In the case of quantitative studies, the predominant data sources were surveys and audio recordings. Out of all the studies, the operator’s view was explored in 28 studies, and the caller’s view in 18, while the remaining 16 studies explored both viewpoints. Altogether, 13 factors were identified and categorised into four main groups: organisational factors, factors related to the operator, factors related to the caller and factors in the interaction. Additional core elements describing each factor were also identified. An overview of the main themes, factors and core elements is presented in Table 3.

Fig. 1
figure 1

Prisma flowchart

Full size image
Table 2 Key information about the included studies (N = 62)
Full size table
Fig. 2
figure 2

Year of publication of the studies (N = 62)

Full size image
Fig. 3
figure 3

Country in which the studies (N = 62) were performed

Full size image
Table 3 Overview of factors and core elements belonging to each main theme
Full size table

Qualitative findings

Organisational factors

The organisational theme contained three factors: availability, working conditions, and decision support systems.

Availability

The availability of the service affected the communication. Queuing and waiting to get through to the operator were described by both callers and operators as negatively affecting communication [27, 30,31,32,33]. Callers found the uncertainty of the waiting time frustrating, which could make them irritated and angry, requiring operators to spend extra time to calm them down [27, 30,31,32].

When operator resources did not correspond to needs, the queues of callers increased and the operators reported a high level of stress [9, 27, 31, 32, 34,35,36]. The operators communicated faster and more mechanically, which could lead to quick decisions being made based on little information [27, 31, 32, 34,35,36,37,38]. Time to express their needs was highlighted as important for the callers [39, 40]. Sufficient operator resources also allowed for collaboration with other colleagues to discuss difficult issues [8, 9, 27, 35], which increased operators’ self-confidence [41] and callers’ faith in the advice given [39, 42, 43].

The operators’ ability to make decisions during the conversation was affected by shortages of other resources, such as doctors on call, ambulances and mental health services [9, 27, 31, 32, 34, 37, 44, 45]. When a shortage of resources, the operators had to spend time on explaining the lack of available resources [31, 32].

Working conditions

The construction of the workspace was described as influencing communication [31, 41]. The opportunity to move and stretch helped the operators to concentrate better [31], and proximity between the operators’ workspaces allowed for collaboration [41].

Organisational attitudes affected how rigidly the operators used guidelines [31, 35, 46,47,48]. If the employer had high-efficiency expectations, the operators felt they were monitored and described the same effect as we found under lack of operator resources: less time for good communication [9, 32, 34, 45]. The organisation of work shifts could reduce the quality of communication, due to physical and psychological limitations during long shifts or night shifts [35, 37, 47].

Technology, such as the use of video, could clarify or prevent misunderstandings [28]. However, multitasking between different technical aids while gathering and interpreting information from the callers was described as cognitively demanding [35, 48, 49]. Beginners spent most of their cognitive capacity on technology, which suppressed the use of their own medical knowledge and communication skills [35]. Technical failure caused problems with focusing, stress and a lack of control, which led to disruptions in the conversation [27, 31, 34, 45, 50, 51].

Decision support systems

Decision support systems (DSS) were described by the operators as ensuring quality by giving structure to the conversations, providing evidence-based knowledge, and being a checklist for important questions when the operators’ own clinical knowledge and experience were limited [8, 41, 48, 50, 52]. Also, callers were more receptive to advice when informed that it came from the DSS [36, 48, 50].

However, DSS structured the conversation in a way that deviated from normal conversations, particularly when operators followed a DSS with a fixed checklist structure [49, 53, 54] that used standardised sentences and closed yes/no questions [54, 55]. Consequently, this restricted the caller’s provision of a detailed description of the situation and made the operators convert all reasons for calling into a problem that suited the system [54]. The operators described feeling controlled, directed, passive and less attentive to the callers when using a fixed checklist structure [50, 56]. The callers described the operators’ use of checklists as frustrating and impersonal, due to an increased number of questions that sometimes seemed irrelevant [38, 57, 58]. If the DSS lacked information, was non-intuitive or used medical terminology, the operators spent time searching the tool or translating words into everyday language, which led to pauses in the conversation [49,50,51]. In contrast, the operators’ attention was notably more oriented to the ongoing conversation when they did not use DSS [53].

Factors related to the operator

Factors related to the operator comprised three categories: knowledge and experience, personal qualities, and communication strategies.

Knowledge and experience

Sufficient medical knowledge was described by the operators as essential for asking the right questions and being confident when gathering information and making decisions [32, 47]. Similarly, the operators’ organisational knowledge was described as a basis for decisions and thereby also the information conveyed to the callers [8, 9]. Callers described greater trust when they spoke to operators who were more competent than themselves [7, 33, 39, 42].

During calls, operators used personal and professional experience to assess symptoms and problems. This influenced the questions asked, the operators’ perception of the situation and the advice given [34]. As they became more experienced, operators developed tacit knowledge and the ability to visualise the patient’s situation. Tacit knowledge was explained as intuition or a gut feeling, which made the operators able to read between the lines and catch information that was not verbalised [8, 27, 31, 37, 58,59,60]. Visualisation was described as crafting a mental image of the patient’s circumstances and served as a means of promoting understanding [31, 44].

Training and education affected communication by increasing operators’ competence in, e.g. communication strategies and medical knowledge, which enhanced the operators’ overall performance and sense of security [8, 9, 27, 31, 35, 41, 46, 48].

Personal qualities

The callers reported that the operators’ appearance in the conversation influenced their experience of the call. The operators’ positive attitude was important for the callers, as they felt vulnerable when calling the service [30, 40, 42, 58]. A positive attitude was also emphasised as important for achieving good communication [26, 31, 44,45,46]. If the callers experienced the operators as dismissive, unfriendly, arrogant or disrespectful, this could lead to communication characterised by anger, irritation and mistrust [7, 33, 39, 42]. Both operators and callers described a correlation between confident operators and feeling reassured [30, 36, 40].

In response to callers’ emotional state (anger, indignation, sadness) or situation (death, abuse), the operators could become emotionally affected [9, 31, 32, 47], which made the conversation quite demanding, especially if the caller was aggressive. The operators described that having control over their reactions during the calls was crucial to achieve effective and good communication with the caller [7, 27, 31, 40, 41, 47].

Communication strategies

Different articulation (tone and rhythm) was used actively to create a calm atmosphere (speaking calmly), emphasise important information (articulating clearly), calm an aggressive caller (speaking calmly and in a deeper voice) and show that one has understood the seriousness of the problem (speaking faster) [8, 31, 40, 41, 59].

Taking control and structuring the communication helped the operator maintain the direction of the conversation and choose what to investigate further [8, 31, 56]. The necessary control and structure had to be balanced against giving the caller enough time to explain the situation [31, 42, 59]. Listening actively and communicating in a calm, empathetic and attentive way helped the operator grasp the situation and gain the caller’s trust [8, 9, 38, 40, 41, 54, 59]. It was a pitfall if the operator had interpreted the patient’s symptoms in one direction, and worked to confirm that direction, thereby overlooking important information from the caller [61]. Pausing the conversation was a helpful strategy if the conversation got out of control emotionally [31, 37, 41].

Providing information, instructions, reassurance, and confirmation were described by operators and callers as beneficial communication strategies [7, 31, 37,38,39, 41,42,43, 59] that could calm the callers [30, 37, 41]. Callers emphasised the need to understand the reasoning behind the operators’ questions and decisions, and they also wanted to be informed about the causes of symptoms and how to deal with them [38,39,40, 57]. By summarising the conversation and allowing the callers to participate in joint decision-making, operators facilitated a common understanding of the situation and a feeling of security [7, 8, 38, 39, 43, 61]. The use of common language and tailored advice created a shared understanding of the content of the conversation and enabled the caller to apply the advice given [26, 28, 31, 37, 39, 41, 42]. Operators who did not expect a highly urgent situation allowed callers to speak freely to a greater extent than if an urgent situation was expected [62].

The design of the questions affected the quality of the information collected. Asking questions in the present tense (“How can I help you now?”) helped the callers focus on the here and now [41]. Operators described that open-ended questions gave a broader overview of the patients’ situation, compared to closed questions [9, 35]. Yet using closed questions seemed to be a good strategy to clarify specific elements [53, 54, 63]. However, simple questions based on yes/no and either/or could be challenging to answer if they did not suit the situation [54, 63]. Callers tended to only answer the last question if the operators asked two or more questions immediately after each other [55].

Exploring the caller’s concern was described as a good strategy to obtain additional information [25, 55]. Furthermore, assessment techniques such as listening for respiratory sounds and instructing a third person to perform physical examinations were described as useful [42, 59].

Factors related to the caller

The factors related to the caller were categorised into two groups: individual differences and the presented medical problems.

Individual differences

Sociodemographic factors, such as gender, age, culture, level of education, and place of residence, affected communication [27, 31, 57, 64, 65]. Different ways of describing illness and expressing needs between cultures made it difficult for callers and operators to understand each other [27, 64, 65]. Age, gender, and place of residence could affect how symptoms were presented and how actively the caller participated in finding solutions. Older people, people living in rural areas, and men tended to neglect or underreport serious symptoms [31, 64], and the operators had to take this into account.

If the callers were under the influence of drugs or alcohol, the conversations could become more aggressive and challenging due to how the drugs affected the callers’ ability to talk, describe their problems and listen [9, 26, 27, 46].

The callers’ attitude toward the system and the operators was described as affecting the tone of the conversation [8, 50, 59]. Callers described how experience from previous encounters affected their trust in the operators and their decisions [39, 40]. Callers’ expectations of the service did not always match what the healthcare system should or could offer, which could make it challenging to reach an agreement [27, 32, 35, 36, 50]. This was particularly challenging if the caller had decided on the outcome in advance and was not receptive to the operator’s assessment [27, 32, 36]. Callers described how they exaggerated their symptoms or the situation if they had to legitimise the contact [43].

Emotional stress and callers’ ability to control their emotions affected how the conversation developed. Diminished control hindered rational thinking, augmented problems with listening and taking in what was said and made it challenging to describe the symptoms and adhere to the operator’s instructions [9, 26, 27, 32, 34, 37, 38].

The callers’ level of knowledge and experience influenced communication. Operators described how the conversation became more challenging when callers lacked knowledge about illness, normal bodily functions, and the organisation of the health services [27, 47, 56]. The knowledge gap acted as a barrier to shared understanding, as callers then struggled to describe their situation, held back information due to uncertainty or responded without comprehending the operators’ questions [49, 55, 56]. A disparity in situational awareness between operators and callers could arise when the callers had conducted online searches prior to their call and misinterpreted the information [34, 43, 45].

Factors in the interaction

The following factors were related to the interaction: faceless communication, connection between operator and caller, third-person caller and communication barriers.

Faceless communication

Operators described themselves as completely dependent on the caller’s description, which made it difficult to create a correct picture of the symptoms and situations, especially when describing skin symptoms or characteristics of children’s symptoms [27, 31, 32, 34, 35, 44, 59, 65]. Not being able to read or use body language made it more difficult to clarify words, read expressions and interpret and monitor responses to the information given, compared to face-to-face conversations [27, 31, 32, 39, 59, 65]. Simultaneously, a lack of visual clues was described as preventing premature judgments based on visual impressions alone [59].

The anonymity of the faceless conversation had advantages and disadvantages. It could make it easier to discuss embarrassing topics [47, 58, 59]. The operators also described it as positive that they could express emotions through body language with no effect on the caller [31]. However, operators described how uncertainty regarding caller identity limited information sharing [59, 65].

Connection between operator and caller

A positive relation between the caller and operator facilitated seamless communication, which made it more likely that the operator would gain a comprehensive understanding of the caller’s situation [8]. Similarity between caller and operator (e.g. the same gender) was described as a factor that eased the connection [64]. The operators described how sympathy came more naturally when they could recognise themselves in the situation [32]. Feeling sympathy for the caller’s situation made the operators more engaged in the conversation, in contrast to calls without a sympathetic approach, where the communication became more direct and technical [56].

A pre-established relationship between the caller and the operator promoted continuity, insight, and trust, and increased the possibility of finding personalised solutions [9, 39, 46]. Previous knowledge could also be a disadvantage, as in the case of frequent callers, who call many times about the same problems. For this group, the operators described the conversations as stressful, frustrating, time-consuming, less empathetic and with a different structure compared to other calls, with less listening and less use of decision-making support [27, 45, 46].

For the callers, having to repeat information created frustration [43, 58]. Information from previous contacts in the patient’s medical records could illuminate the situation, help the operator build on previous conclusions and prevent the caller from repeating information [8, 31, 32, 36, 45].

A power asymmetry in the interaction was described, where the operators, as the professionals and the “door openers’” to medical help, had the most power [26, 38, 55, 56, 66]. Yet, callers described themselves as personal experts and emphasised the importance of not being devalued and considered solely as a source of information [7, 38, 39, 43]. A power struggle could affect the communication [26, 33, 38, 43]. To equalise the balance of power, callers used methods such as expressing concern to elicit empathy, arguing until they achieved the desired action, and presenting the problem as an order or as a recommendation from someone with greater authority [66].

Third-person caller

Second-hand information from a third-person caller (not the patient) was described as difficult to trust and interpret and required more questions to be asked [27, 28, 31, 32, 64]. Parents expressed how it was challenging to interpret and describe children’s symptoms objectively and that feelings and fear influenced how the information was provided [42]. Nevertheless, talking to a third person provided information of great importance when the patient was prevented from speaking for themselves [31, 39]. The third person’s proximity to the patient, in physical or relational distance, affected the quality of the information given [26, 64].

Confidentiality requirements hindered the flow of information between third-person callers and the operators, as the operators had to be careful not to share specific details with a third person [31]. Talking about a patient without the patient’s consent was considered an ethical dilemma, where the operator had to deal with safeguarding both the patient’s and the caller’s autonomy [65].

Communication barriers

Distractions in the caller’s or operator’s environment made it difficult to hear the other party and took focus away from the conversation. As a result, it became uncertain whether the caller or operator was listening properly or understanding the content of the conversation [31, 35, 39].

Language barriers were described as creating uncertainty and misunderstandings, delaying and complicating the progression of the conversation and making it more problematic to use decision support systems [27, 28, 32, 34, 36, 42, 48, 67]. Similar effects were also observed in calls where the caller had impaired hearing or speech [26, 31, 39].

Quantitative findings

Organisational factors

Quantitative studies had results that shed further light on all three organisational factors: availability, working conditions, and decision support system. The results from a survey of callers’ satisfaction showed a negative correlation between perceived unreasonable waiting time before the call was answered and overall satisfaction with the service [68]. An observational study exploring the cognitive impact of stress in operators found that higher stress levels increased the number of cognitive failures and affected the operator’s decisions, but also made the operators process information more quickly [69].

A study that assessed the quality of communication using a list of assessment items (RICE communication list) found a positive correlation between time spent per call and quality of communication [15]. However, a study comparing calls ending in malpractice claims with other calls found no significant difference in time spent per call [70].

A survey investigating how the use of video was experienced by both operators and callers showed that the integration of video was considered to enhance safety and elevate the overall quality of communication [28].

Two observational studies using audio records to compare calls handled by nurses utilising DSS and doctors not utilising them observed that nurses tended to ask more questions, often adopting a checklist-style approach [5, 71]. When communication quality was compared between the two groups using a validated quality assessment tool (AQTT—Assessment of Quality in Telephone Triage), there was no significant difference in the overall quality [5].

Factors related to the operator

In relation to the operator, the factors of knowledge and experience and communication strategies were supported by quantitative studies. Personal qualities were only described in qualitative studies.

A study examined predictors of callers’ satisfaction by matching the content of audio-recorded calls with caller questionnaire data. They found that expectations fulfilled by the operator in terms of listening, clarity, cooperation and perceived competence were strong predictors of caller satisfaction [72]. An intervention study examining the effect of educational activities on the quality of the information provided during telephone triage found a positive impact on quality in the short term, but the effect did not persist over time [73].

A study comparing calls that resulted in reported medical errors with other calls found that active listening and checking a shared understanding were less present in calls that resulted in reported medical errors [70]. This same study also showed that when operators used more open-ended questions, callers provided significantly more medical information than in calls with fewer open-ended questions. Another study examining callers’ satisfaction with receiving self-care advice found that callers’ satisfaction with the help received was dependent on whether the caller felt reassured after the call [68].

Factors related to the caller

Evidence from quantitative studies elaborated on both factors related to the caller: individual differences and presented medical problem. A user satisfaction survey showed that people with a lower level of education were less satisfied with the information and advice given, due to problems with understanding the content [74]. Another survey study found that mismatch of expectations was associated with low satisfaction with the call [75]. Callers with a high degree of concern needed more reassurance due to worries about negative consequences of their illness [29].

An observational study using a communication quality measurement tool showed a negative association between urgency level and the quality of communication [6]. Another study found that very urgent calls were characterised by less rude and aggressive behaviour than less urgent calls [76].

Observational studies found that calls regarding mental illness lasted longer than the other calls [77] and had a greater risk of misunderstandings [78] and that callers with mental health problems were at greater risk of being rude and aggressive when talking to healthcare professionals than other callers [76].

Factors in the interaction

The only interactional factor supported by quantitative results was communication barriers.

Two observational studies comparing non-fluent speakers to fluent speakers found that trust and satisfaction were lower in non-fluent than in fluent speakers [67, 77]. One of the studies found that communication time was on average longer for the non-fluent speakers [67].

Integration of quantitative and qualitative findings

An overview of the factors and where the quantitative findings either expand upon or corroborate the established qualitative framework is presented in Fig. 4. In addition, a more comprehensive overview of the factors distributed over each individual study can be found in Additional file 3. The assessed quality of the studies was generally high, and 48 of 62 studies were found to have the maximum score. The results of the assessment are further described in Additional file 4.

Fig. 4
figure 4

Number of studies (N = 62) supporting each factor

Full size image

Organisational factors

The number of studies and the diversity in terms of population, country and perspectives studied suggested that availability [8, 9, 15, 27, 30,31,32,33,34,35,36,37,38,39,40,41,42,43, 50, 68,69,70] strongly affected communication. Resource shortages affected accessibility of the service, operators’ stress and callers’ satisfaction negatively. Time pressure and stressful working environment made the operators rush through the communication to the extent that it could become a patient safety problem. Having enough time to gain a comprehensive overview of the patient’s situation seemed essential for quality. However, divergent findings when examining the link between the duration of the conversation and the quality of the communication imply that time alone may not serve as a reliable indicator of conversation quality [15, 70].

Decision support systems were also well-documented in both qualitative and quantitative studies [5, 8, 36, 38, 41, 48,49,50,51,52,53,54,55, 57, 58, 71]. The DSS’s role in structuring the conversation was described particularly thoroughly [5, 38, 48,49,50,51,52,53,54,55,56,57,58, 71]. If the operators adhered strictly to the tool’s algorithms during the information collection, the conversation was more strained and did not develop naturally. Also, the use of DSS increased the number of questions asked and the frequency of checklist-style questions used. Nevertheless, quantitative data did not reveal any difference in the overall quality of the conversation when using DSS compared to not using DSS [5].

The factor working conditions was mainly supported by qualitative studies [9, 27, 31, 32, 34, 35, 37, 41, 45,46,47,48,49,50,51]. One mixed methods study from Denmark examined the effect of video as a technical aid [28]. This study examined both callers’ and operators’ perspectives and found that video reassured both parties by contributing to an expanded understanding of the situation. However, the quality of this study was compromised due to the substandard description of some of the methods and data sources mentioned in the paper.

Factors related to the operator

Knowledge and experience [7,8,9, 27, 31,32,33,34,35, 37, 39, 41, 42, 44, 46,47,48, 58,59,60, 72, 73] were supported by both qualitative and quantitative findings. The integration indicated that the operators’ medical and communication skills affected the information obtained. These skills also emerged as strong predictors for establishing trust and confidence between the two parties during communication. Training and education were described by the operators as improving quality, which was also quantitatively supported by one study [8, 9, 27, 31, 35, 41, 46, 48, 73]. However, the effect seemed to decrease over time, suggesting that training needs to be repeated [73].

The number and diversity of studies [7,8,9, 24,25,26,27,28, 30, 31, 35, 37, 38, 40,41,42,43, 53,54,55,56,57, 59, 61,62,63, 68, 70, 72] supported that operators’ use of communication strategies affected communication. A multitude of strategies was described, with varying degrees of support. Listening [8, 9, 40, 41, 70], being clear and informative [7, 31, 37,38,39, 41,42,43, 59, 68] and facilitating two-way communication [7, 8, 38, 39, 43, 61, 70] were the most supported strategies. Furthermore, different question designs seem to give different perspectives on the patient’s situation and should be used actively.

There appears to be a connection between the use of communication strategies and the other factors linked to the operator: inherent personal characteristics, knowledge and experience. Some strategies could be a natural part of the operator’s personality and therefore part of the operator’s natural communication response, while others might need to be learned through training and education. Personal qualities, such as attitude and emotional control, did impact communication. However, it was a less substantiated factor mentioned only in qualitative studies [9, 26, 27, 31, 32, 36, 41, 45,46,47, 65].

Factors related to the caller

A multitude of methodologically diverse studies described that individual differences between callers affected the conversation [8, 9, 26, 27, 29, 31, 32, 34,35,36,37,38,39,40, 42, 43, 45,46,47,48,49,50, 55,56,57, 59, 64, 74, 75, 79]. Mismatch between the caller’s expectations and what the service could provide hindered reaching an agreement on measures, which made the communication more difficult [27, 32, 35, 36, 50, 75]. The studies also suggested that callers with lower levels of education and less knowledge were struggling when describing their situation and understanding the information given, which could impede the shared understanding of the situation [27, 47, 56, 74]. Furthermore, the callers’ knowledge and experience affected how the medical problems were presented [6, 8, 9, 26, 27, 31, 32, 34,35,36, 47, 54, 56, 62, 63, 66, 76,77,78].

Urgency level and calls regarding mental illness were described in both qualitative and quantitative studies as affecting communication. High-urgency calls were described as streamlined and easy to handle in a qualitative study [8] and of lower quality than less urgent calls in a quantitative study [6]. However, the quality assessment tool used in the latter study did not account for how urgency levels might impact conversation dynamics. Both quantitative and qualitative studies reported that calls regarding mental health symptoms differed from calls regarding somatic symptoms, in being more time-consuming and emotionally demanding [8, 9, 27, 47, 76,77,78].

Factors in the interaction

Factors in the interaction were almost exclusively described in qualitative studies. Faceless communication [27, 31, 32, 34, 35, 39, 44, 47, 52, 58, 59, 65], third-person callers [26,27,28, 31, 32, 39, 42, 64, 65] and communication barriers [26,27,28, 31, 32, 34,35,36, 39, 42, 48, 67, 77] were all supported by a smaller number of studies than many of the other identified factors. The connection between operators and callers was mentioned in many and diverse qualitative studies [7, 9, 26, 27, 31,32,33, 36, 38, 39, 43, 45,46,47, 55, 56, 58, 64, 66]. However, both qualitative and quantitative studies described the core element language barriers as hindering good communication [27, 28, 32, 34, 36, 42, 48, 67].

Discussion

The objective of this study was to identify factors affecting communication during telephone triage and to describe how these factors affected communication. A total of 12 factors were identified to affect the structure, content and flow of communication. The factors were organised into four main themes: organisational factors, factors related to the operator, factors related to the caller and factors in the interaction. All factors were supported by a range of studies (n = 9–31). The findings showed that the organisational factors mainly influenced communication by facilitating or complicating the operator’s communication. This suggests that the organisation of the medical call centre has the strongest impact on communication.

Strengths and limitations

The search strategy was piloted and adjusted ahead of the main search, until known literature was identified in the search result. We chose to search in the main medical databases: the two major medical databases (Medline and Embase) and a database for nursing/allied health professionals (CINAHL). In addition, we supplemented with a search in Web of Science, which is a large interdisciplinary index database. Additional searches in databases such as Scopus might have identified even more studies. However, during the process of analysing the data, it became clear that this material, like other qualitative material, had a saturation point after which no new factors were identified.

The methodological heterogeneity of the studies made the choice of inclusion challenging. Some studies described the factors more subtly than others, and these were often discussed before inclusion. All studies were screened by two or more of the authors, while the synthesis was mainly carried out by one author. Although this approach made the synthesis more consistent, it might also have increased the risk of omitting details of factors or references.

In this review, the data from the quantitative studies was transformed into textual descriptions, to allow for integration with the qualitative data. The transformation might increase the risk of interpretation bias. The methodological heterogeneity in the studies included made the interpretation of the extracted data complex. However, it can be argued this heterogeneity, in which elements are illuminated from different perspectives, strengthens the probability that the phenomenon described is real. Some of the core elements had few references and could have been omitted from the results. We chose to include them, create an overall picture of factors and thereby also describe elements that should be examined more closely. Forty-seven percent of the studies were conducted in Sweden, which could challenge the generalisability of the findings. However, none of the factors identified in this review were only described in studies from one country.

Implications for further research

The majority of studies included were qualitative studies based on interviews with operators and callers. Interviews shed light on the participants’ experiences and attitudes. While this approach can be used to form hypotheses and models, further research is needed to establish the actual relevance of the identified factors.

A subset of the included studies used conversation analysis or quality measurement tools to assess audio recordings from real conversations. Such methods can capture more of the actual dynamics of the communication process. Employing conversation analysis on audio recordings not only provides a comprehensive understanding of operator-caller interactions, but also allows for a more detailed examination of individual elements in the conversation. However, when using audio recordings, information about the context surrounding the call might be lacking. The results from our review suggest that external factors influence the conversation and need to be considered when analysing audio recordings.

There is an overall lack of studies that explore and substantiate the effects of individual factors. Such studies are needed to balance factors against each other and to gain a deeper understanding of how the factors affect communication and their specific impact on the communication process.

Implication for practice

Figure 5 shows the influence the organisation, the operator, and the caller has on each other during communication. As parties in the interaction, operators and callers have a direct influence on the communication and thus each other. Nevertheless, the organisation emerged as the most important facilitator of good communication, since availability, working environment and decision support tools had a great influence on how the operator communicates. The operator must rely on the organisation to provide a conducive working environment that promotes communication and learning opportunities, thereby enhancing their competence.

Fig. 5
figure 5

The influence between the organisation, the operator and the caller during communication

Full size image

Organisational policies and guidelines are examples of how the organisation influences operators’ communication. Currently, work is being done to standardise and simplify descriptions of symptoms into decision support systems, to ease the assessment of urgency. The organisations argue that standardisation ensures patient safety. Yet this might oversimplify, overlooking the individual differences that need to be considered. Individual differences and differences in how the caller presents the medical problem are well-known challenges in the interaction between healthcare personnel and patients. For decades, researchers have described how patients seek to explain their symptoms by using their own experience, knowledge and surrounding context, and how this affects their health-seeking behaviour, e.g. Leventhal [80]. The use of strict guidelines might challenge the ability to see the overall picture. Additionally, it could discriminate against patients who do not fit the standardisation or fail to describe their symptoms in a way that suit the system. Those who have mental health symptoms, for example, can often initially describe physical symptoms [9]. They need operators with good communication and assessment skills who understand the complexity of the narratives.

Through training and education, the organisation can increase the operator’s communication and assessment skills. Active use of communication strategies seems to be beneficial to communication, but many of the strategies need to be learned. Increased awareness of communication strategies and how they can be utilised can improve the operator’s ability to influence communication.

The organisation can also directly influence the interaction and the caller. One example is the introduction of a video that allows the operator to see the patient, removing some of the barriers to faceless communication. Callers’ expectations and use of the service are influenced by information campaigns [8]. Such campaigns can therefore be used to reduce the mismatch between the callers’ expectations and what the service can provide, and help provide a common starting point for the conversation. In essence, the organisation is the primary driver of enhancing the quality of communication by defining the framework dictating service capabilities and operational procedures.

The dialogue process during telephone triage from an operator’s perspective is well documented and described as containing five phases: opening, listening, analysing, motivating and closing [81]. Still, as found in this review, there are many factors that can complicate the communication, which implies that the operator alone cannot control the communication. There is a need to examine communication from a wider perspective to find what is effective and good communication in medical call centres. The framework developed in this review can be utilised to enhance awareness of the complexity of communication, serving as a foundation for internal quality improvement within the organisation. It can also serve as a starting point for further research.

Conclusions

Many factors affect the structure, content and flow of communication during telephone triage. The operator influences the quality of communication directly but relies on the organisation to provide a working environment that facilitates good communication. The framework of factors identified in this review can serve as a tool for raising awareness of measures that can improve communication, thereby increasing patient safety. However, the results are mainly based on qualitative studies with a focus on operators’ and callers’ experiences and attitudes. There is limited research into the actual relevance and effect of the identified factors. This needs further investigation.