Background Emergency department (ED) communication has been demonstrated as requiring improvement and ED patients have repeatedly demonstrated poor comprehension of the care they receive. Through patient focus groups, the authors developed a novel tool designed to improve communication and patient comprehension.
Study design This is a prospective, randomised controlled clinical trial to test the efficacy of a novel, patient-centred communication tool. Patients in a small community hospital ED were randomised to receive the instrument, which was utilised by the entire ED care team and served as a checklist or guide to the patients' ED stay. At the end of the ED stay, patients completed a survey of their comprehension of the care and a communication assessment tool-team survey (a validated instrument to assess satisfaction with communication). Three blinded chart reviewers scored patients' comprehension of their ED care as concordant, partially concordant or discordant with charted care. The authors tested whether there was a difference in satisfaction using a two-sample t test and a difference in comprehension using ordinal logistic regression analysis.
Results 146 patients were enrolled in the study with 72 randomised to receive the communication instrument. There was no significant difference between groups in comprehension (OR=0.65, 95% CI 0.34 to 1.23, p=0.18) or communication assessment tool-team scores (difference=0.2, 95% CI: −3.4 to 3.8, p=0.91).
Conclusions Using their novel communication tool, the authors were not able to show a statistically significant improvement in either comprehension or satisfaction, though a tendency towards improved comprehension was seen.
- Care systems
- risk management
- quality assurance
- patient education
- emergency department
Statistics from Altmetric.com
- Care systems
- risk management
- quality assurance
- patient education
- emergency department
Communication between physicians and patients is a critical component of healthcare delivery. Effective communication has been shown to correlate to improved rates of emergency department (ED) follow-up, patient medication compliance and health outcomes, as well as increased patient satisfaction.1–3 The Institute of Medicine, the Society for Academic Emergency Medicine and the recent Kalamazoo Consensus Statement have targeted patient–physician communication as a point of emphasis in recent years.3
The factors that contribute to deficient communication in the ED include a fast-paced, high acuity and chaotic environment; the need for rapid decision making; and lack of established patient–provider relationships. In recent studies, gaps in ED communication have been identified. When audiotapes of patient–physician interactions were reviewed, physicians spent the majority of time on collecting rather than dispensing information.4 Vashi and Rhodes5 reviewed audiotapes to specifically evaluate the quality and content of discharge instructions. Only 55% of patients received any verbal discharge instructions; and only 22% of providers confirmed that patients understood the instructions. This represents a missed opportunity to clear up any misunderstandings and to reinforce compliance.4 Patients repeatedly demonstrate poor understanding of the care that they receive in the ED. In a study by Engel et al in 2009, 78% of patients showed a comprehension deficiency in at least one domain of their ED visit (diagnosis and cause, ED care, post-ED care, return instructions), with 61% demonstrating a deficiency with comprehension of their ED care.6
Despite the recognition of poor physician–patient communication in the ED and the resultant deficiency in patient comprehension, there is a lack of proven methods for improvement. Previous efforts, including workshops for ED staff focusing on communication skills, and videos or handouts explaining ED processes, have been shown to decrease patient complaints and increase patient satisfaction.7–10 These efforts are limited by their general nature. They are not personalised to the specific patient's clinical scenario, ED course, or questions and have not been tested for their effect on patient comprehension.
The goal of this study was to assess the effect of a standardised, yet customisable communication instrument on patient comprehension of emergency care and patient satisfaction. One of the hospitals in our system had begun using a structured erasable white board in the patient's room that contained preprinted areas for the staff to fill in which studies were planned and how long until results were expected. Anecdotally, providers felt that these white boards helped patients understand their care and improved satisfaction. We wanted to evaluate whether a tool that providers and patients could both use would improve comprehension and satisfaction.
We developed a writing tool to actively engage the patient and their family while in the ED. Note taking can improve memory and understanding through enhanced recall and reflection on significant events.11 ,12 Patient care journals are promoted as a method to help families track the care their loved one is receiving. One prominent example of this is the written and phone application journal provided by the Josie King foundation, a patient safety organisation with community and care provider outreach programs.13 While a writing tool presents a barrier to some patients in the ED with low literacy, the check box portion of the tool can serve as a guide for the staff in their conversation with the patient and provide rudimentary ability for confirmation that specific areas are addressed. The communication instrument acts as a checklist of areas of care and as a starting point for further discussions about care. This written tool is inexpensive and easy to implement in the ED.
The communication tool that was used in this study was developed through patient focus groups at the St Joseph Mercy Hospital (SJMH) in Ann Arbor, Michigan, USA, an 85 000 visit Level II trauma centre (figure 1). Patients were recruited after completing an ED visit. They were invited to participate in a group discussion scheduled at a later time. Participants were asked to refer to their recent ED visit and recall what information they would have liked to have known or understood better, as evaluation and treatment were proceeding in the ED. They were also asked to report what other information related to their diagnosis, treatment and follow-up would have been helpful for them to have at hand after their discharge from the ED. Information that the participants felt was important while in the ED included what imaging tests or labs were going to be completed, their anticipated length of stay, and whether or not they should eat. These and other suggestions were added to the tool in the final iteration.
This study was conducted at the Saline Emergency Department, in Saline, Michigan, USA. This is a 10-bed, single covered ED with approximately 15 000 visits/year. Subjects were approached, screened and consented by a research assistant after they were placed in a treatment room but before the physician entered the room.
It was emphasised during consent that participation in the study would not affect the care received. Patients who met inclusion criteria and agreed to participate were assigned to the intervention group according to whether the randomly sequenced envelope they were handed contained the communication instrument or a blank sheet of paper. Each envelope also included a pencil. The instrument, shown in figure 1, consists of predesignated places to mark any planned labs, imaging studies or other interventions as well as how long the patient could expect to wait for these to occur and for results to return. In addition, there was a space to write whether and when the patient could eat and spaces for questions the patient or family might want answered as care proceeded.
Caretakers or family members were allowed to write on the tool if they agreed to be present throughout the entire ED stay. A family member could be the main user of the tool if, for instance, the patient was obtunded and not the primary contact for communication in the ED.
The entire medical team including the attending physician, nurses, respiratory therapists and radiology technicians was encouraged to use the instrument to help communicate the plan of care and to fill in information where appropriate. The study team educated staff about how to use this instrument and the conduct of the study during staff meetings held before the study began. It was emphasised that the conduct of the study and use of the instrument should not alter patient treatment plans in any way.
At the end of the patients' ED stay (prior to discharge, admission or transfer), the research assistant administered the assessment surveys including the communication assessment tool-team (CAT-T) and patient comprehension surveys. The CAT-T is a validated instrument used to assess communication and satisfaction with the medical team10 ,14 (figure 2). The comprehension survey is a tool created by the authors to assess patient comprehension of the treatment that they received in the ED. On this form, patients are asked in an open-ended manner to report the diagnoses they were given in the ED, the types of treatment they received in the ED, and their discharge instructions, if they were discharged, or admission plans, if they were admitted (figure 3). They were allowed to refer to any discharge instructions provided.
The comprehension surveys were compared with the patients' ED chart by two emergency nurse reviewers. Chart reviewers were blinded to the patients' intervention group. Each answer given by the patient was coded as concordant, partially concordant or discordant as compared with the information contained in their chart. Absent information was scored as discordant by the nurse reviewers. When the nurses disagreed (33/720 items), an emergency medicine resident was the tiebreaker. There was 95.4% agreement between the two initial reviewers. An example of a partially concordant answer for ‘Diagnosis’ would be if a patient reported her diagnosis as ‘potassium level slightly lower than normal’ but did not mention the chest pain that had brought her into the ED and for which she was admitted. Examples of concordant answers would be where the patient listed both chest pain and low potassium as her ED diagnoses or if, under ‘Treatments’ the patient was able to name all of the medications she received in the ED.
Many of the comprehension instruments that were scored as concordant contained detailed information, such as a patient presenting with a migraine headache who was able to state that she received, ‘IV fluids, compazine, toradol, and dilaudid’. There were several cases in which the question ‘What are 3 reasons you should return to the ER’ was misunderstood, prompting answers such as ‘reliable, kindness, convenient location’, but in general if this question was scored as discordant it was because patient left it blank or reported ‘none’. Concordant answers in this domain included reasons specifically addressed in the chart such as, ‘worsening abdominal pain, fever, or vomiting’. The most common reason for ‘Time to follow-up’ being scored incorrectly was an answer of ‘no follow-up needed’, followed by cases where the incorrect time interval for follow-up was reported by the patient.
The written comprehension assessment was designed to assess whether patients understood the parts of their visit that we felt most important. Our patient population during this pilot did not seem to have any difficulty completing the assessment due to literacy or legibility of responses. No assessments were left entirely blank, though a few had minimal information given. Blank questions or answers of ‘none’, when not accurate, were scored as discordant.
All analyses, unless otherwise specified, were done using SAS V.9.2 (SAS Institute). Demographic information was summarised using frequencies and percentages. Missing data from the satisfaction questionnaire were assumed to be missing at random and multiple imputation (five singly imputed data sets) to minimise bias and to calculate appropriate standard errors was accomplished with IVEware, 2002 (Regents of the University of Michigan).
The primary outcome was the number of discordances as judged by comparing the patient comprehension tool with the ED record.
Secondary outcomes were the number of misunderstandings (defined as discordances or partial concordances), also judged by comparing the patient comprehension tool with the ED record, and the total point score on the CAT-T.
Post hoc analysis was performed to evaluate the domains of comprehension.
IVEware was used to analyse the data from multiple imputation. We evaluated whether there was a difference in satisfaction with care between groups using a two-sample t test on the total scores from the CAT-T. We investigated whether the communication domain (questions 7, 9 and 13) was different between groups by comparing the combined scores from these three questions, using the two-sample t test. A post hoc analysis was done to see whether the proportion of excellent responses to each individual question from the CAT-T was different between groups using logistic regression analysis with the dependent variable being ‘excellent’ or ‘not excellent’. A Bonferroni correction for multiple comparisons was made, reducing the level of statistical significance to p<0.0031 for this analysis.
We evaluated whether there was a difference in comprehension between groups in two ways. The total number of discordances was turned into an ordinal measure with 0, 1, or 2 or more discordances out of a possible total of 5 in the discharged patients and 3 in the admitted patients. The same thing was done for the total number of discordances plus partial discordances (also referred to as ‘misunderstandings’). These outcomes were compared between groups using ordinal logistic regression analysis with age, sex and whether the patient was admitted to the hospital as covariates in the model. The models were tested for compliance with the proportional odds assumption using the score test.
For discharged patients, we tested which domains showed more discordance using a generalised linear mixed model with a multinomial distribution and a cumulative logit link function. In this model, the dependent variable was a 3-level ordinal measure of concordance with 0 being discordant, 1 being partially concordant and 2 being concordant. The independent variables were group, age, sex, domain, and an interaction between group and domain. The interaction term was removed if statistical significance for that term did not reach p<0.20. A random intercept for subject was used in this analysis to account for correlation within subjects. Statistical significance was set at p<0.05, unless otherwise stated.
One hundred and forty-six patients were enrolled in the study and 60 (41%) were male subjects. Seventy-two were randomised to the comprehension tool and 74 were randomised to usual care. All but 13 patients (9%) filled out the questionnaires themselves. The distributions of age and race are shown in table 1. Twenty-four patients were admitted to the hospital from the ED, 13 in the intervention group and 11 in the control group.
Seventeen patients skipped at least one question on the satisfaction survey and these answers were multiply imputed. The mean satisfaction score for the intervention group was 72.0 and for the control group, 71.8 (difference=0.2 (95% CI: −3.4 to 3.8), p=0.91). The means of the three question communication domain were also not different between groups, 13.2 in the intervention group versus 13.5 in the control group (p=0.42).
Responses to individual questions on the communication satisfaction survey are shown in table 2. There was no statistical difference between groups for the proportion of ‘excellent’ responses to any question, even those specifically geared towards assessing communication.
The five domains of comprehension and the frequency of responses for each domain are shown in table 3. Ordinal logistic regression analysis satisfied the cumulative odds assumption. It showed the OR for increasing discordances from 0 to 1 and from 1 to 2 or more was not statistically significant but favoured the intervention group (OR=0.65, 95% CI 0.34 to 1.23, p=0.18). A similar result was found for the OR for partial concordance or discordance (OR=0.63, 95% CI 0.34 to 1.17, p=0.15). Thus, while the direction of effect favours the instrument decreasing partial or full discordances in comprehension, it does not reach statistical significance.
In patients who were not admitted, the individual comprehension domains were significantly related to overall degree of comprehension (p=0.04). Comprehension of follow-up time and reasons to return to the ED were more likely to show higher levels of discordance than comprehension of diagnosis and treatment (see table 4). There was no effect of the intervention on this relationship (p=0.63). Patients were more likely to understand diagnosis and treatment in the ED (which were explicitly covered by the communication instrument) than follow-up time and reasons to return to the ED (which were not explicitly covered).
Communicating the events, purpose and outcome of emergency care to the patient can be complicated and difficult. Standard communication has persistently been shown to be faulty, and our study re-demonstrates that there are many deficits in comprehension among our patients There is an increasing interest in medical communication as a means of increasing efficiency and decreasing the cost of medical care. Proposed Hospital Outpatient Quality Reporting measures for 2013 include the use of a transitional record to be received by discharged patients that specifies their care while in the hospital and would include ED care.15
Our objective was to test a potentially feasible and low cost communication tool that we developed using focus groups of discharged ED patients. We also developed a comprehension test based on the information we felt was most important for our patients to understand. It is clear from our comprehension tests that patients have a varied degree of understanding about their care. Our nurse reviewers were able to reliably score this when compared with the chart, indicating that this instrument can be used as an accurate representation of patients' comprehension. The legibility of the patient's writing did not prevent scoring of any response and even in cases where medication names or diagnoses were misspelled, in the context of chart review, it was possible to determine accuracy.
Our comprehension test did not show a statistically significant difference in scores between groups. We do not know why the tool did not result in statistically significant improvement in the measured outcomes, but we have some ideas.
Because the comprehension test had not been used before, we had no prior information to power the study for a presumed difference in comprehension survey scores. We did not know whether the number of patients evaluated would be adequate to find a difference in comprehension and suspect that the sample size may have been too small.
Another explanation may be found in the characteristics of the site that we used for our pilot. Saline hospital is a rural ED with globally high satisfaction scores and fast-throughput times for patient care. This is different from the site where the CAT-T was validated and may have made the addition of the communication tool a minor improvement there. This limits the generalisability of results to the majority of EDs, where it might have greater impact.
In addition, the Hawthorne effect may have blunted the differences between the treatment and control groups. While the communication instrument was in use, physician and nurse performance of communications tasks may have improved, even when not using the instrument, because of a teaching effect of using the tool. This may be one of the instances where a before and after design is preferable to a randomised controlled design.
We were unable to show that our comprehension tool improved patient comprehension of ED care or satisfaction with communication in this small study performed in a rural ED. We did show a decrease in the odds of discordances and misunderstandings, although these differences were not statistically significant. Multiple studies have shown difficulties in communication in the ED. Further studies evaluating possible solutions should be attempted in an effort to improve care for our patients.
Funding This work was supported by St Joseph Mercy Hospital grant number RC-10-133.
Competing interests None.
Ethics approval This study was reviewed and approved by the St Joseph institutional review board, study number HSR-11-1281.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Our database of comprehension survey and CAT-T results are available on request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.