Clinician-patient Interaction During Addiction Consultation (CLiPID)

Background

Drug misusers present to addiction services a wide range of problems with the most serious consequence of drug overdose being the threat to life. Accurately predicting the risk of suffering a drug overdose therefore plays a crucial role in effective prevention of drug overdose. So far, most effect have been focused on investigating predictive variables that are related to drug misusers' subjective perceptions of coping abilities and measures of life experience or treatment history. However, very little attention has been directed to the factors that are associated with the aspects of the addiction consultation process, where clinicians have a greater degree of control over patient interaction.

Some initial empirical evidences suggest that clinicians' and patients' facial expressions were predictive of suicidal reattempt. It was found that doctors showed a higher frequency and intensity in frowning working with suicide repeaters than with non-repeaters. In addition, repeater patients were found to have a higher activity of the mouth and to look downwards significantly more often. This line of study has laid some foundation for a new way of prediction through observation of clinician-patient interactions. We are therefore interested in exploring some non-verbal behaviours of both clinicians and patients during the consultation process that are potentially predictive of a patient's risk of suffering a drug overdose.

Regarding how verbal communicative behaviours during the consultation process might influence patient care outcomes, findings from other healthcare communication areas have consistently demonstrated that patient-centred communication is positively correlated with improved patient health outcomes. One key feature of a patient-centred consulting style is clinicians' ability to verbally recognize and address individual patients' emotional distress. To our knowledge, it is the first time to investigate how clinicians manage patients' emotional distress affects patient care outcomes in a drug service setting.

No validated coding schemes so far have been developed to explore the relationship between aspects of interaction and drug service outcomes. In other areas of healthcare communication, the Verona Coding Definitions of Emotional Sequences (VR-CoDES) has been frequently used to code patient expressions of emotional distress and provider responses. Although it has been widely used in a variety of contexts from psychiatric contexts to dental settings, it has not yet been used in consultations involving drug users. We are therefore interested to test the applicability of the system in a drug service setting after a suitable modification. Drug addiction consultation is usually an emotion-provoking environment and accurate perception of patient emotions has become a precondition for empathic responses. The modification of a validated coding scheme to be suitable for the drug service context will be the first useful tool to help clinicians to recognize patients' emotional concerns and thereby improving the quality of patient care.

With the modified version of the VR-CoDES as an analysis tool, we aim to identify factors that are associated with aspects of the consultation process, particularly the verbal and non-verbal behaviours of both clinicians and patients, to improve our understanding of drug misusers' risk of suffering a drug overdose. The main research questions are summarized below.

Research questions

• What communicative behaviours observed in the interaction between clinicians and drug user patients are predictive of patent's risk of suffering a drug overdose? In other words, we are interested in exploring what verbal and non-verbal behaviours of both clinicians and patients during the consultation process predict a patient's risk of suffering a drug overdose.
• We are also interested in exploring how clinicians respond to patients' expressions of emotional distress during the consultation process and how successful management of emotional worries might be related to improved patient satisfaction. Furthermore, we are keen to investigate whether the Verona Coding Definitions of Emotional Sequence (VR-CoDES) can be reliably applied to the drug consultation context after suitably modified. In other words, whether patients' expressions of emotional cues and concerns and clinicians' responses to them can be identified and reliably coded using the modified VR-CoDES.

Research design

It will be a one year, pilot observational study with video recording of clinician-patient consultation within a single site of NHS Fife Addiction Services, based in the Whyteman's Brae Hospital in Kirkcaldy. In the future, more sites may be considered to achieve a larger sample size when this pilot study is complete on NHS Fife single site, largely due to the time constraint.

The Overdose Risk Assessment Questionnaire (ORA) and the Treatment Outcomes Profile Questionnaire (TOP) are two questionnaires that have already been routinely used by the Addiction Nurses within the NHS Fife Addition Services. We hope to exert a minimum extra workload to the staff participants and will respect their normal implementation procedure of the questionnaires. Two pieces of information obtained from the ORA questionnaire will be included in the analysis. First, the overall risk assessment score on the first assessment with a key worker before the video observation will be used to compare with the behavioural factors emerged from the interactions for predicting patient drug overdose risk. For example, high risk assessment scores might be correlated with a high frequency of patient 'looking down' non-verbal behaviour. Second, some relevant information, such as depression and emotional abuse, will be used as predictors for patient expressions of emotional distress during consultation. Therefore some regression and correlational analysis will be conducted.

Similarly, only two pieces of information selected from the TOP questionnaire will be included for analysis for this study. First, the number of days patients injected non-prescribed drugs in the past four weeks obtained one months after the video recorded session. This information will be used as an indicator of a patient risk of suffering a drug overdose, as the primary outcome of the study. Frequency and duration of some behavioural data gathered from the consultation interaction (e.g., clinician's 'frowning', patient's 'looking down') will be correlated with this outcome variable (i.e., the number of days patients inject drugs). Second, patient's rating of overall quality of life obtained from the TOP questionnaire will be used to measure patient satisfaction as the secondary outcome of the study. Patient satisfaction will be also measured by a specifically developed questionnaire immediately after consultation. Again, some multiple regression analysis will be employed to explore relationship between predictor and outcome variables.

Methods

• Main ethical issues The main ethical issues involved in this study relate to the recording and storage of the images of clinicians and drug user patients and anonymity and confidentiality of participants' information. We have taken every possible steps to address these ethical issues and included some additional safeguard actions to deal with unexpected incidents. First, initial raw data recorded on DVDs will be locked in a metal case for temporary storage in the field and they will be secured in a locked cabinet in the University of St Andrews once transportation from the field to the University is complete. Second, each of the raw DVD files (i.e., each patient consultation) will be assigned a unique code and then be digitalized and stored on a university computer internal hard disc for coding and analysis. Third, these digitalized files will then be permanently removed from the computer once the coding and analysis is complete. They will be exported, after face images being digitally blurred, onto an external hard drive that will have all data encrypted and access to data files password protected. These encrypted data will be stored for two years following the end of the study and the original raw data on DVDs will be permanently destroyed at the end of the project. An audit record will be compiled to allow inspection of the progress of video data files from video DVD to computer hard disc for coding and then exporting to external media for safe long-term storage. Audit will be monitored by the School Ethics Committee on a 6-month basis.

It is important to note that the faces of participants on the video data will be permanently digitally obscured after the coding has been completed. We will only display video files to a wider audience (e.g., at conferences) after face images have been blurred and informed consent from the relevant participants have been obtained.

In order to protect anonymity of personal details and confidentiality of information provided by participants, we will assign a unique code to each participant so that participants will not be identified on the data files. This coding book will be locked in a secure place and only the research team members will have access.

• Participants Sampling framework: In order to answer the research question of how aspects of clinician-patient interaction during addiction consultation predict drug misusers' risk of suffering a drug overdose, it seems that the population of interest is all clinicians and patients providing and using the addiction services in the UK. In practice, the ideal random sample or a proportionate stratified sample drawn from the population is often not feasible due to limited resource or time constraints. The sampling procedures adopted for the current study is a mix of opportunity and purposive sampling. Opportunities presented themselves in the form of patients being referred to the NHS Fife Addiction Services and all eligible staff members and patients based in the Whyteman's Brae service site will be invited to take part. Sampling is also purposive in that only the first session with a CDAS Addiction Nurse, where the patient is naive to the Addiction Nurse with regard to interaction experience, will be used to study the interaction. It is also considered important that sample patient participants should include approximately equal proportions of men and women; and the age profile should have representation from young and middle aged groups. Additionally, it is also important to consider some demographic information such as age, sex and working experience when sampling staff participants.

Inclusion and exclusion criteria: All CDAS Addiction Nurses, who provide a one-to-one consultation service to patients, based in the Whyteman's Brae Hospital as part of the NHS Fife Addiction Services will be eligible. All patients referred to the NHS Fife Addiction Services by a Fife wide triage contact point or their GPs, who are to have a first session with their CDAS Addiction Nurses, will be eligible. Patient with mental health problems as well as those whose first language are not English will be excluded.

Recruitment procedures: Staff participants are recruited through a combination of top-down and bottom-up approach. Meetings with head of the service and team leaders (top-down approach) and discussions with staff members (bottom-up approach) will take place simultaneously, to introduce the project and encourage participation. At the staff team meeting, information about the project and consent form with an opt-in reply slip will be left with them after the meeting so that they have time (about 3-4 days) to decide whether they are willing to participate. The consent forms will be returned to the research assistant in a prepaid and addressed envelope. We need a minimum number of 16 staff participants preferably with four male and four female between 25-65 years old. Newly qualified stff members are encouraged to participate as well as those more experiences ones.

Obtaining patient consent will be achieved mainly through the help of the addiction services. An invitation package is delivered to all potential patient participants from the Addiction Services on behalf of the research team. As we need a minimum number of 16 consultation sessions (eight clinicians with two patients per clinician), the patient participants will preferably be eight women (four between 18-40 years old and four between 41-65 years old) and eight man (four between 18-40 years old and four between 41-65 years old). The invitation package will include an invitation letter, a project information sheet and a consent form. Patients will be also given 3-4 days before they decide to participate. The consent forms will be returned to the research assistant in a prepaid and addressed envelope.

• Procedures of data collection Observations of clinician-patient interactions during the addiction consultation will take place at the consultation room at the Whyteman's Brae Hospital in Kirkcaldy in the form of audio-visual recording. The whole process of recording will take about 30 minutes depending on the complexity of individual cases. It is not expected that the presence of the camera and researchers will have any detrimental effect on the clinicians' delivery of the service, nor will this likely to upset the patients. We will use a small sized digital camera and place it in a good distance (approximately 2m from the clinician and the patient). The operation of the recording system will be well practised and the setup will be completed in good time to minimize any possible disturbance to the participants. The recording and storage of video files will strictly follow the ethical procedures described in the main ethical issues in this application form, such as digitally obscure the face images of relevant participants and secure data files in a locked cabinet.

According to the research design (to collect a minimum of 16 consultation sessions with about eight staff participants with two patients per staff member) and the availability of the Addiction Nurses at the Whyteman's Brae Hospital, which is about 10 up to the date of this proposal, we expect that each Addiction Nurse will participate in an average number of approximately two video recording sessions, which will last about 60 minutes in total. This average 60-minute total observation time for each Addiction Nurse could spread out in 5 or 6 days depending on the appointment with his/her patients. The total participation time for each patient participant in the video recording session is about 30 minutes depending on the complexity of the case. If either the clinician or the patient feels unhappy with the recording at any point of the observation, the recording will be stopped immediately and any recorded data will be destroyed.

The Overdose Risk Assessment Questionnaire and the Treatment Outcomes Profile Questionnaire will be implemented as part of the routine practice by the Addiction Nurses. The Patient Satisfaction Survey, specially developed for this study, is handed to the patient by the research assistant before the video observation. It can be completed within 5 minutes at the end of the consultation. Patients are advised to complete the survey with the absence of their clinicians and return it to the letter box placed at the Reception. This is to avoid possible pressure to provide positive comments in front their clinicians and protect confidentiality of the information. There will be no follow-up questions after this initial survey has been completed.

• Data analysis All video data will be analysed using the Observer XT system. First, all verbal and non-verbal behaviours of both clinicians and patients will be coded using the modified VR-CoDES coding scheme. Both inter- and intra-coder reliabilities will be checked to a satisfactory level (i.e., Kappa ≥ .70). Second, frequency or duration of certain behaviours will be calculated to identify possible relationships between certain behaviours and outcome variables (i.e., risk of suffering a drug overdose and/or patient satisfaction). Third, sequential analyses will be employed to detect patterns of behaviours and identify more complex behavioural relationships. In addition, some coded behaviours will be exported to other statistical packages for further analyses such as SPSS. Data gathered from the survey method, such as patient satisfaction with the consultation and patent previous depression will be used as either outcome or predictor variables to be correlated with behavioural variables from the interactions.