The Use Of Supportive Communication When Responding To Older Peoples Emotional Distress In Home Care An Observational Study

Design and setting
The study had a descriptive observational design and was part of a large international research project on person-centred communication in home care (COMHOME) [27]. This study reports the Norwegian part of the COMHOME project.

Data were collected from four home care units: three units located in a city of approximately residents and one unit in a rural municipality of approximately 5000 residents. The encounters took place in the private homes of the older people. Real-time communication unfolding between the older person and the registered nurse (RN) or nursing assistant (NA) during the visits was audio-recorded and analysed. The units of analysis were the older person’s expressions of emotional distress; concerns, and cues or non-verbal vocal cues, identified by the Verona Coding Definition of Emotional Sequences (VR-CoDES), and the subsequent responses of the RN or NA [28].

The study population comprised older people (≥65 years) receiving home care in Norway, and the RNs and NAs providing home care services. In Norway, RNs must have at least a bachelor degree, and NAs must have completed formal education to at least upper secondary school level plus vocational training in nursing.

The nursing staff sample
The managers of the home care units recruited in total 33 nursing staff, 16 RNs and 17 NAs. All had received oral and written information about the study from the research group before recruitment. The inclusion criteria were being an RN or an NA providing home care in private homes (not apartments located in nursing homes) with a permanent position. The target sample of nursing staff was 10 participants for each unit, aiming for an equal distribution of RNs and NAs. There are more female nursing staff in the population; therefore, no measures were taken to ensure equal distribution of genders in the sample, besides ensuring that both genders were represented among both RNs and NAs. Regions 1 and 4 organized home care in two units with allocated personnel. Therefore, only seven and six staff members were included, respectively, from regions 1 and 4. The work experience of participating nursing staff ranged from 5 months to 45 years, with an average experience of 17 years (SD, 11; missing data for one NA). Information about age and gender is shown in Table1.

Table 1 Sample of older people, nursing staff and visits

The sample of older people
In total, 48 older people participated in the study; see Table1. Average activities of daily living (ADL) scores, the level of assistance needed to perform a range of daily tasks (0 = no assistance needed, 5 = full assistance needed) [29], varied between 0 and 3.7 (SD, 2.1). Hours of home care included in the care plan ranged from 0.3 to 21.5 h per week (average, 5.2; SD, 5.3). The inclusion criteria were being an older person receiving home care and being able to provide informed consent. Older people considered too frail, based on medical records and clinical experience (e.g., severe dementia, cognitive challenges, final-stage cancers), were excluded by the unit manager.

Data collection procedure
All participating nursing staff took part in the recruitment of older people, after having received proper instructions by the research group. Data were collected during the period of December 2013 to April 2014. The recruitment started a week before the planned start of the data collection. The data collection was organized for the four units consecutively and aimed to be completed within a week.

The audio recordings of entire home care visits (n= 196) were collected using a digital audio recorder (H1 Zoom), worn on the upper arm of the participating nursing staff. One hundred and twenty-two visits included 1–2 identified tasks to be completed, 60 visits had 3–4 tasks and 14 visits had five or more tasks. The duration of the visits ranged from 1 to 72 min, with an average duration of 17 min (SD: 14). The participating RN or NA could encounter the same older person in multiple visits, and the older person could encounter different nursing staff in multiple visits. All care providers encountered at least three different patients.

Coding methodology: Verona Coding Definitions of Emotional Sequences (VR-CoDES)
To identify emotional talk sequences, all visits were coded with the VR-CoDES – Cue and Concern [30, 31], identifying the older person’s emotional distress as cues and concerns; and the VR-CoDES – Provider Response [32, 33], identifying the nursing staff’s responses to emotional distress.

The coders were the first author (LH) and a research assistant. The first author was trained in the use of VR-CoDES by the last author (HE), an experienced coder and one of the founders of the coding system. When there was a need to resolve issues concerning the implementation of the VR-CoDES in a home care setting or disagreements on the interpretation of specific expressions between the two coders, and to establish consensus about the appropriate use of the VR-CoDES, the last author was consulted.

Older people’s expressions of emotional distress
VR-CoDES defines concerns as clear and unambiguous expressions, where the emotion is current or recent and explicitly verbalized. A cue is defined as a verbal or non-verbal hint of an underlying unpleasant emotion, but the expression is lacking clarity. Cues are categorized into seven mutually exclusive categories: vague or unspecific words for the emotion (cue a), verbal hints of implicit emotions or unpleasant states/circumstances (cue b), phrases emphasizing unpleasant cognitive or physical states (cue c), expressions of potential importance (cue d), repetition of neutral words or phrases (cue e), non-verbal expressions or hints of negative emotions (cue f) and verbalized references to an emotion that occurred more than a month ago (cue g). Further, all concerns and cues were coded either as expressed by the older people themselves on their own initiative (patient expressed, PE) or as elicited (concerns/cues being solicited, explored or facilitated) by the nursing staff (health-care provider elicited, HPE).

The two coders analysed 32% of the recordings independently to reach acceptable inter-rater reliability. The steps taken in this process followed given recommendations [28]. An inter-rater reliability of a Cohen’s kappa above 0.6 (substantial agreement) was considered sufficient for further analysis [34]. The inter-rater reliability was calculated for the following outcome variables: expressions of emotional distress (n= 63, Cohen’s kappa = 0.74); concerns (n= 15, Cohen’s kappa = 0.68); and whether the older person took the initiative to express the concern/cue, or the expression was elicited by the health-care provider (n= 119, Cohen’s kappa = 0.68). Cohen’s kappa was not calculated for specific cue categories. The inter-rater reliability was comparable to that found in other studies using VR-CoDES [35,36,37]. The coding system is considered to have high ecological validity, capturing concerns that are experienced as real by the patient [38].

Categorization of cues and concerns
Three sum categories of expressed emotional distress were computed: expressions of an emotion by referring to it in words, references to unpleasant states/circumstances and contextual hints of emotion. These categories were respectively labelled as emotional references, emotional states/circumstances and contextual hints of emotion. Emotional references (4 VR codes) were expressions with either clear (concerns and cue g) or vague words (cue a) or non-verbal vocal cues (cue f) related to a negative emotion. Cue f was coded if the patient sighed, whined, moaned, cried or sobbed, as these are considered self-evident vocal expressions (no words used) of a negative emotion, e.g., crying is understood as a reference to sadness. Emotional states/circumstances (2 VR codes) were verbal hints of implicit emotions or unpleasant states/circumstances (cue b), or phrases emphasizing unpleasant cognitive or physical states (cue c). Contextual hints of emotion (2 VR codes) were neutral expressions coded as utterances of emotion because of contextual factors or hints, such as how the expression emerged from the narrative background (cue d) or was repeated by the patient (cue e); see Table2.

Table 2 The distribution of concerns and cues codes and elicitation by VR-CoDES sum categories and nursing staff

Nursing staff responses
VR-CoDES – Provider Response codes the nursing staff response immediately following a concern or cue [33]. The coding has two dimensions. First, the coder identifies whether the response refers to the concern/cue explicitly or not explicitly, e.g., maintaining wording or key elements of the concern/cue or not. Second, the coder determines whether the response performs the function of providing space or reducing space for further disclosure of the concern/cue, e.g., allowing the patient to talk more about their expressed emotional distress or not. The coding system includes 17 mutually exclusive response codes. Application of the system to audio recordings does not permit use of the code “silence – non-explicit providing space”, and by default the code “ignore – non-explicitly” is used. The code “postponing – reducing space” was not identified in our data. Therefore, these two codes are not included in the sum categories for provider responses (Table3).

Table 3 Frequency of response categories defined by communicative function as described in VR-CoDES- Provider Response

Cohen’s kappa was calculated for coder agreement for the response codes that differentiated between providing space and reducing space (and also those differentiating between explicit and non-explicit references to the concern/cue, as these codes occurred within the providing/reducing space dimension) (n= 31, Cohen’s kappa = 0.75). Cohen’s kappa was not calculated for the individual response codes.

Categorization of provider responses
The categorization of the response codes (Table3) was based on the function assigned to each response as described in the VR-CoDES manual; that is, whether the response showed that the health-care provider had noticed the patient’s expressed emotion and focused on the content of the patient’s expression, or ignored or blocked the expression [33].

Emotion-focused responses (7 VR response codes)
All seven codes in this category indicate that the response functioned to provide space for emotional expression, either explicitly or non-explicitly. The nursing staff notice the emotion and provide a response allowing further disclosure of the emotion. Non-explicit responses (no explicit reference to the previous concern/cue), include back channels, acknowledgement, an active invitation to speak further and implicit empathy and, explicit responses (specifically mentions either the content/topic or the emotion in the previous concern/cue) include acknowledging or exploring the emotion or providing explicit empathy.

Content-focused responses (5 VR response codes)
Two response codes in this category focus explicitly on the content of the cue or concern and provide space for, acknowledge or explore the content of the cue or concern. Two response codes explicitly refer to the concern/cue but reduce space for further disclosure by switching (e.g., refers the patient to a third party/agency to talk to them about the concern/cue) or giving advice and specific information. One response code reduces space for further disclosure by providing non-specific information or advice.

Ignoring or blocking responses (3 VR response codes)
These codes include responses that seem to ignore the concern/cue completely, non-explicitly diverge from the concern/cue or actively block the concern/cue by refusing to talk more about the topic and devaluing what is said by the patient.

Statistical analysis
Statistical analysis was performed with IBM SPSS Statistics, version 24.0 (IBM Corp, New York, USA). The dataset was checked for missing data. Variables describing the older people had no missing data.

The outcome variable for the study was whether nursing staff responses opened space for further disclosure of emotion (emotion-focused responses). A binary variable was computed, differentiating between “emotion-focused responses” and “responses focusing on content or responses ignoring or blocking the concern/cue”. The data were fitted to a logistic model to identify predictors of emotion-focused responses.

Possible explanatory variables of the nursing staff responses, such as characteristics of the care provider and patient, the individual visit, and the type and expression of concerns/cues, were explored using frequency and summation tables. Percentages were rounded to whole numbers. Group differences were analysed using Fisher’s exact tests and the significance level was set at 39]. In crosstab analysis, available data were analysed, whereas in univariate logistic regression, the default solution in SPSS of listwise deletion was used.

In order to explore patient expressions of emotion (verbal or non-verbal) versus less direct suggestions of emotion, a binary variable capturing this distinction was computed. This variable differentiated concerns and cues that referred to an emotion (i.e., emotional references) from cues referring to states/circumstances or contextual hints of emotions (i.e., emotional states/circumstances and contextual hints of emotions). This nominal binary variable allowed exploration of whether linguistic properties of the patient’s communication influenced the nursing staff’s responses, and thus was important for answering our research questions. This also strengthened statistical power by combining two categories with limited numbers of observations: emotional states and contextual hints of emotion (Table2). The binary variable was used in the logistic regression analysis.

Ethical considerations
This study has been conducted in compliance with the World Medical Association Declaration of Helsinki: ethical principles for medical research involving human subjects [40].

Local newspapers informed the public about the study in advance of data collection, to inform the community about why care providers would be wearing audio recorders. Nursing staff who were well known to the older people receiving home care provided verbal and written information about the study, before asking for written consent. All participating older people received a telephone call from the unit manager after the audio-recorded visits, and were given the opportunity to withdraw from the study. If the patient had problems answering the phone, a member of staff who had not been part of the visits addressed this in person. None of the patients used the opportunity to withdraw.