Understanding the influences and impact of patient‐clinician communication in cancer care

Abstract Background Patient‐clinician communication is thought to be central to care outcomes, but when and how communication affects patient outcomes is not well understood. Objective We propose a conceptual model and classification framework upon which the empirical evidence base for the impact of patient‐clinician communication can be summarized and further built. Design We use the proposed model and framework to summarize findings from two recent systematic reviews, one evaluating the use of shared decision making (SDM) on cancer care outcomes and the other evaluating the role of physician recommendation in cancer screening use. Key results Using this approach, we identified clusters of studies with positive findings, including those relying on the measurement of SDM from the patients’ perspective and affective‐cognitive outcomes, particularly in the context of surgical treatment decision making. We also identify important gaps in the literature, including the role of SDM in post‐surgical treatment and end‐of‐life care decisions, and those specifying particular physician communication strategies when recommending cancer screening. Conclusions Transparent linkages between key conceptual domains and the influence of methodological approaches on observed patient outcomes are needed to advance our understanding of how and when patient‐clinician communication influences patient outcomes. The proposed conceptual model and classification framework can be used to facilitate the translation of empirical evidence into practice and to identify critical gaps in knowledge regarding how and when patient‐clinician communication impacts care outcomes in the context of cancer and health care more broadly.

endorsements and incentives, how and when different communication exchanges impact specific patient outcomes is only beginning to be understood.
As a body of empirical evidence emerges, a framework is needed to facilitate the understanding of what constitutes effective patientclinician communication in given contexts. Early ecological models of patient-clinician communication often did not consider the diversity of patient outcomes that communication may impact (see for example, Street;7 Feldman-Stewart et al. 8 ), or were void of environmental and contextual considerations (see for example Street et al.; 5 Kreps et al. 9 ).
Alternatively, models that have more broadly focused on care quality and which consider not only a diversity of patient outcomes, but the complexity of environmental or contextual factors that can impact care processes, including communication, fail to address the complexities and nuances specific to patient-clinician communication (see for example, Zapka et al.; 10 Wagner's chronic care model 11 ). Thus, while a number of thoughtful models and frameworks are available, none, on its own, can be used to guide a systematic summary of existing empirical evidence on patient-clinician communication and its impact on patient outcomes or to identify critical gaps in existing knowledge.
In this study, we propose a conceptual model and classification framework upon which the empirical evidence base for the impact of patient-clinician communication on patient outcomes can be summarized and further built. To illustrate the usefulness of the proposed classification system, we use the approach to summarize findings from two recent systematic reviews of the impact of patient-clinician communication and patient outcomes 12,13 focusing on cancer-related studies.

| COMMUNICATION-OUTCOMES MODEL
As originally proposed by the Transformation Model of Communication and Health Outcomes (Transformation Model), if communication is an essential process in promoting effective health care, one should be able to demonstrate how communication impacts health and other outcomes. 9 Our proposed model, therefore like the Transformation Model, has its origins in the systems theory model of input-process-output.
As depicted in Figure 1, at the centre of the model is the patientclinician communication exchange. Under the umbrella of high-quality patient-clinician communication, there are many types of communication exchanges patients and clinicians may use, ranging from specific communication strategies such as action planning, informed decision making or shared decision making (SDM) to specific functions such as information exchange, question asking or rapport building. These communication exchanges can (but may not) be adapted to fit the needs or preferences of the patient (eg, health literacy levels or language preference). Additionally, these exchanges may occur through a variety of channels (eg, face-to-face, telephone or email).
To the left of the communication exchange are the antecedent contextual conditions and attributes patients and clinicians bring with them to the exchange. These include characteristics that can alter the exchange itself, individuals' interpretation of the exchange or the impact of the exchange on subsequent patient outcomes. 14 Patient socio-demographic (eg, age, race, income, and education level) and other characteristics (eg, self-efficacy, health insurance status) as well as characteristics of the clinician (eg, age, race, medical specialty), are known to be associated with differences in the occurrence of different types of communication exchanges and may also moderate the effect of communication on subsequent outcomes. [15][16][17][18] In particular, race, gender and age concordance between patients and clinicians has been shown to impact both the communication exchange itself and associated patient outcomes. [19][20][21] In most cases, communication between patients and clinicians is not limited to a single encounter. Rather, patients form relationships with their clinician(s) over time and often discuss the same health or other topics repeatedly over time. Thus, the length and quality of the emotional and other relationship between the patient and clinician represents one of the first layers of contextual factors that may impact the association between a specific communication exchange and a subsequent patient outcome. [22][23][24] Following other ecologically based models, 10 the model highlights not only this relationship context, but also the clinical and delivery system contexts as well as the wider community context. Such an acknowledgement is important as patientclinician communication exchanges are likely quite different depending on the disease or treatment being discussed as well as the specific objective of the interaction (eg, initial diagnostic discussion between an oncologist and a woman with early stage breast cancer vs a discussion regarding surgical treatment options between the same woman and a surgeon). Examples of delivery system contextual factors that could influence the communication exchange and its impact on outcomes include the professional and leadership culture of the organization, available resources and procedures within the specific practice setting, and care delivery/management policies. At the community/policy level, the communication exchange and outcomes may be affected by public policy or regulation, professional standards and/or market pressures. 10 For example, CMS recently issued a requirement for the use of SDM as a prerequisite for lung cancer screening reimbursement. 25 To the right of the patient-clinician communication exchange is a range of patient outcomes. Similar to the Transformation Model, 9 the model conceptualizes outcomes in three broad domains: affectivecognitive outcomes; behavioural outcomes; and health outcomes. 12 Affective-cognitive outcomes include a patient's understanding, knowledge, satisfaction, self-efficacy, anxieties and the like.

| COMMUNICATION-OUTCOME CLASSIFICATION FRAMEWORK
While the model depicted in Figure 1   London, England), where algorithms can be used to rotate the blocks to achieve the desired solution, the purposeful selection and organization of key elements can result in the identification of important patterns. As such, the flexibility of a classification table approach affords a practical and transparent visualization of results across diverse studies.
As with a Rubik's Cube ® , any one side (or classification table)  When the patient-clinician communication is explicitly measured within the context of a study, it is typically measured in one of three ways: patient self-report, clinician self-report or observer rating. Even within each of these different measurement perspectives, there are multiple measurement instruments available, often with no agreed upon gold standard. 26 For example, there are a variety of measures used to ascertain patient-reported assessments of SDM. [27][28][29][30][31][32][33] Likewise, there are multiple coding systems available to obtain observer-rated measures of SDM. 34-37 Further complicating matters, patient selfreports of SDM are often not associated with observer ratings of SDM. [38][39][40] Thus, as illustrated in Figure 2 42 video-based telemedicine 43 and even social media such as Twitter and Facebook. 44 As of yet, limited research has explored how these channels alter patient-clinician communication exchanges or alter the outcomes associated with those exchanges.
Specificity is also important in the type of patient outcome(s) considered. While some of the cognitive-affective outcomes (eg, satisfaction and trust) are by definition patient-reported measures, behavioural and health-related outcomes may be measured either by patient or clinician self-report, or by more objective measurement methods. For example, a patient could be asked to rate their current stress level or their stress could be measured via salivary cortisol-neither of which has been found to correlate highly with one another 45 and each of which may be worthy of consideration within a specific study context. As the effect of patient-clinician communication on patient outcomes may vary across outcomes and how those outcomes are measured, transparency in outcomes and measurement methods is also needed. impact of communication exchanges on diverse patient outcomes, but also the ability to identify knowledge gaps where subsequent research is needed. For example, we recently applied this approach to summarize the evidence for the effect of SDM on patient outcomes. 12 In that application, we held the communication exchange type constant (ie, SDM), but considered the different types of outcomes (ie, affective-cognitive, behavioural, or health outcome) that had been studied in relationship with SDM as well as the different perspectives from which the measurement of SDM had occurred. 12 In so doing, we were able to highlight the importance of the communication measurement perspective used, finding that SDM, as reported by patients as occurring, was associated with improvements in affective, and in some cases, behavioural outcomes. Table 2 reports findings from the 48 cancer specific studies included in that review. As illustrated in Table 2, similar clusters of studies with positive findings can be seen among the cancer-specific studies, including those relying on the measurement of SDM from the patients' perspective and affective-cognitive outcomes. Also illustrated is the void in studies that have considered clinician perceptions of SDM as well as those evaluating the impact of SDM patient on health outcomes.

| APPLICATION OF MODEL AND FRAMEWORK TO SHARED DECISION-MAKING COMMUNICATION
We further illustrate the usefulness of the model and classification framework by using it to consider findings from the same systematic review albeit from a different perspective. 13 In this second example, (Table 3)

| DISCUSSION
We propose the use of a conceptual model and framework to consider the impact of patient-clinician communication on cancer care outcomes. The model highlights the complexity of the relationship between patient-clinician communication and patient outcomes, the diversity of communication exchanges that transpire between T A B L E 2 Summary of results by measurement perspective and patient outcome category: impact of shared decision making on patient outcomes

Patient outcome category
Affective-cognitive Behavioural Health Total n % n % n % n % Positive  8  50  3  30  1  9  12  32   NS  6  38  7  70  10  91  23  62   Negative  2  13  0  0  0  0  2  5   Total measured  16  10  11  37 Clinician reported Positive  across studies-whether systematically or otherwise. 26 By using the model and framework proposed here, we have illustrated the importance of the measurement perspective used to assess the communication exchange as well as the type of outcome considered when evaluating the impact of SDM. 12 In the application here, we further demonstrate how this approach can be used to explore multiple dimensions within an existing evidence base to highlight important gaps and methodological variability. Despite this ability, it nonetheless is important to note that the model and accompanying framework approach remain a simplification. As such they may omit from consideration other important factors that impact either the communication exchange itself or its impact on outcomes. Furthermore, it is important to acknowledge challenges have and will continue to exist in finding or constructing databases that enable any individual study to encompass all the components of the model presented here.

Patient reported
It is now accepted that patient-clinician communication can and does impact patient outcomes. Using studies from two systematic reviews and the conceptual model and classification framework proposed, we are able to begin to understand how and when patient-