Which medical interview skills are associated with patients’ verbal indications of undisclosed feelings of anxiety and depressive feelings?

Background In medical practice, obtaining information regarding patients’ undisclosed “feelings of anxiety” or “depressive feelings” is important. The purpose of this study was to determine which interview skills are best suited for eliciting verbal indications of undisclosed feelings, for example anxiety or depressive feelings in patients. Methods Our group videotaped 159 medical interviews at an outpatient department of the Department of Family Medicine, Mie University Hospital (Mie, Japan). Physicians’ medical interview skills were evaluated using a Medical Interview Evaluation System and Emotional Information Check Sheet for assessing indications of “feelings of anxiety” or “depressive feelings”. We analyzed the relationship between the interview skills and patients’ consequent emotional disclosure using generalized linear model (GLIM). Results The usage of interview skills such as “open-ended questions” “asking the patient’s ideas about the meaning of illness” “reflection” and “legitimization” were positively associated with the number of anxiety disclosure, whereas “close-ended questions” and “focused question” were negatively associated. On the other hand, only “respect” was positively associated with the number of depressive disclosures, whereas “surveying question” was negatively associated. Conclusions The results revealed that there are several interview skills that are effective in eliciting verbal indication of undisclosed “feelings of anxiety” or “depressive feelings”.


Background
One of the important functions of the medical interview is to understand the mental status of patients. It is important that the physician checks the mental status of the patient during the medical interview, because patients often do not come to see physicians suspecting that they might have mental disorder [1,2]. Paying attention to the patients' emotions could not only contribute to early detection of psychiatric disease but is said to affect the patient's satisfaction [3][4][5]. This approach will also shorten the consultation time and reduce the extra stress of the physicians [6]. However, in many cases, patients are said to only send indirect signals with no explicit emotional expressions [7] and show only an ambiguous response [8]. The cue expressed by the patient may be overlooked due to the physician's low awareness [9,10] or insufficient clinical skills due to the lack of training [11,12]. Furthermore, even when they are encouraged by the physicians, many patients would feel embarrassed, hesitated or simply feel that they do not deserve to disclose emotions [13].
In recent years, many studies have been made on the relationship between the patients' emotional response and the physicians' interview skills to address these needs. Many of them have discussed that physicians' communication skills should be employed to help patients' express their emotions [6,8,14,15]. However, most of the reports on the patients' emotional response are made in the context of cancer treatment. Sufficient studies have not been made on the physicians' skills and the patients' emotional response during the common medical interviews like ones at primary care facilities.
The aim of the present study is to explore the association between physicians' interview skills and patients' disclosure about anxiety or depression in the primary care setting.

Participants and procedures
One hundred and fifty nine different new patients (74 males and 85 females) participated in this observational, cross-sectional study. They are patients at outpatient department of the Department of Family Medicine of large-scale hospital, Japan. The patients came from both rural and urban areas and were from a wide range of socioeconomic classes. All of the patients were over 15 years of age, and most patients generally had common medical problems such as hypertension, hypercholesterolemia, and diabetes. First, we checked whether this was a first visit and asked if the prospective participant were willing to participate, and if he or she declined, no further effort was made. Second, we gave them instructions about the purpose and meaning of the study and told them that they could decline from this study anytime when they want to. Third, we obtained each participant's signature. Parental permission was obtained for patients whose age is less than 20 years old.
Each patient was interviewed by one of 26 family physicians, 98 family medicine residents, or 35 medical students having family medicine clinical clerkship. These interviews were videotaped for evaluation.
Informed consents were obtained from patients and physicians prior to any observation of the medical interviews. The research protocol of this study was approved by the Research Ethical Committee of the Mie University School of Medicine.

Outcome measurements
The videotaped medical interview skills were reviewed by trained researchers using the Takemura Medical Interview Rating Scale (TMIRS) ( Table 1) [16]. Some systems of the Roter interaction analysis system were used to build this scale [17]. "Feelings of anxiety" and "depressive feelings" were also operationally defined ( Table 2). These items were all combined into a single sheet to allow us to count the skills that doctor's used and patients' emotional responses.
This scale was assessed for reliability. First, for each videotaped interview, an evaluator used the scale to evaluate the number of skills used by the physician. Second, the same evaluator did a second review a random selection of 25 interviews 2 months after the first review. The kappa value was calculated and found to be 0.93.
Interview length was considered as a confounder, because longer interview duration was presumed to positively influence our findings [18]. In addition, patients' sex, sex matching between patient and physician, and affiliation (student, resident and family physician) were also obtained as other possible confounders.

Statistical analyses
Demographic data were obtained first. Then, we used a generalized linear model (GLM) to assess the relationship between a physician's use of these interview skills and verbal indications of feelings of anxiety or depression. Because our observation of physicians' interview skills and patients' emotional disclosures were recorded as count data, we modeled the relationship using the Poisson distributions and log link function. After the generalized linear model was estimated, we verified the variance inflation factors of each independent variable and the covariates (VIF). Finally, we evaluated goodness of fit of the estimated model using Akaike Information Criteria (AIC).
Statistical significance was set as p < 0.05 for all measures. We used R version 3.1.1 [19] for statistical analyses. Generalized linear model analyses were conducted using the "glm" function. There was no missing data.

Results
Characteristics of patients and physicians are shown in Table 3. Additionally, the table shows basic statistics including mean value, standard deviations, and quartile values of each interview skill and emotional disclosure.
The estimated model for disclosures of anxiety is shown in Table 4, and the estimated model about that of depression is shown in Table 5. For the purposes of our analyses, only modified models will be considered.
Patients' disclosures of anxiety were positively associated with "open-ended questions" "legitimization" "reflection" and "asking the patient's idea about the meaning of the illness (meaning of the illness)". Conversely, they were negatively associated with "close-ended questions" and "focused questions". As none of the interview-skills and covariates showed VIF above 10 (data not shown), multicollinearity which was found to be within acceptable norms [20][21][22][23].
Patients' disclosures of depression were positively associated with "respect for patients" and negatively associated with "survey questions". As with the result for disclosures of anxiety, multicollinearity was found to be within acceptable limits, with no VIF above than 10 (data not shown).
We compared the respective association strengths of the various independent variables. "Open-ended questions" and "legitimization" were found to have the strongest association with respect to indications of undisclosed "feelings of anxiety". "Close-ended questions" and "focused question" were also significant; however, their respective association strengths were close to zero. As for depression, "survey questions" and "respect for patients" were the only notable associations.

Discussion
There were differences in which skills were effective for eliciting feelings of anxiety and depression clearly. "Open-ended questions" "meaning of the illness" "reflection" and "legitimization" had a positive association with indications of "feelings of anxiety". "Open-ended questions" were effective for eliciting patients' feelings of anxiety, but "close-ended questions" or "focused questions" were not. A lot of previous studies recommended using the open-ended questions to soliciting patients' information [24][25][26] and the result of this study is in agreement with these investigations.
Previous study showed that "reflection" and "legitimization" were associated with patient satisfaction [16]. Another study also identified a significant positive association between the amount of information elicited from patients and patient satisfaction [27]. Furthermore, an investigation also indicated that patient satisfaction is associated with discussion of      [3,4,28]. This study might provide an explanation as to why "reflection" and "legitimization" have been shown to increase patient satisfaction, as enhanced disclosure has been shown to increase patients' satisfaction. "Meaning of the illness" was effective for eliciting patient's feelings of anxiety, but "close-ended questions" and "focused questions" might conceal it. That results are in accordance with the previous investigation that patients are more likely talk about psychosocial issues in the atmosphere created when a physician uses less close-ended questions and less dominating [3,4]. These results may show it is difficult for patients to express their anxieties when the patients were asked to verbalize the feeling directly by "close-ended questions" or "focused questions".
In this study, "respect for patients" had positive association with indications of feelings of depression. Previous research showing that patients were more willing to divulge personal information; such as therapeutic, lifestyle, and psychosocial information to physicians who show respect supports the result of this study [29]. Cape and McCulloch also explored patients' reasons for not presenting emotional problems in medical interviews and they found that 45 percent of reasons related to psychological embarrassment or hesitation to trouble the general practitioner or family physician [13]. Their result shares similarity in our result.

Strength and limitations
Strength of this study is the use of an objective method to assess medical interview skills and patients' emotion. Conversely, there are several limitations. First, there may be other confounding factors not measured in this study; for example, other relevant patient background information, non-verbal communications, and so on. Second, it is difficult to evaluate subjective information such as mental feelings objectively.

Conclusions
This study reveals that several interview skills including "open-ended questions" "legitimization" would be effective in eliciting undisclosed "feelings of anxiety". On the other hand, "respect for patients" was associated with undisclosed "depressive feelings".
Authors' contributions MG conceived of the study, analyzed and interpreted the videotaped data and drafted the manuscript. YT participated in the design of the study, revised it critically for important intellectual content, performed the statistical analysis and drafted the manuscript. Both authors contributed to defining the research question, participated in the coordination of the study. Both authors read and approved the final manuscript.