Visit Content Analysis: Doctor-Patient Communication in Patients with Type 2 Diabetes

Abstract

Introduction:

The primary care visit is an important opportunity to discuss and modify diabetes management.

Objective:

To gain insight into doctor-patient communication during primary care visits among English and Spanish speaking patients with type 2 diabetes and suboptimal glycemic control (HbA1c > 7%).

Methods:

We conducted a quantitative content analysis of audiotaped primary care visits in 2 patient cohorts. In Study 1 (31 English-speaking patients), we examined factors associated with management changes, and in Study 2 (20 Spanish-speaking patients and their Spanish-speaking providers), we examined the association of question asking with HbA1c control. This study was conducted between November 2017 and January 2020 across 8 primary care practices within Kaiser Permanente Northern California.

Results:

In Study 1, the only factor significantly associated with a diabetes management change was patient identification of diabetes as a priority prior to the visit (91.7% had a management change vs 52.6% of patients who did not identify diabetes as a priority; p = 0.02). In Study 2, patients with poorer glycemic control (HbA1c ≥ 10.0) asked significantly fewer questions (3.4 ± 1.8 vs 10.7 ± 6.9 questions per 15 minutes; p = 0.004). Overall, despite receiving primary care from language-concordant providers, Spanish-speaking Study 2 patients asked fewer questions than English-speaking Study 1 patients (4.5 ± 2.9 vs 7.5 ± 3.7 questions per 15 minutes, respectively; p = 0.004).

Conclusion:

Our results highlight 2 potential strategies (preparing patients for their visits through identifying priorities and learning how to ask more questions during visits) for improving diabetes primary care.

INTRODUCTION

The majority of the 34 million people with type 2 diabetes (T2D) in the US have suboptimal glycemic control (defined as an HbA1C of 7.0% or higher).¹ Effective diabetes control is essential for reducing the micro- and macrovascular complications that contribute to the higher morbidity and mortality seen in T2D.² Because the management of diabetes has become increasingly complex for both clinicians and patients, it requires well-informed strategies for treatment.³

The primary care visit represents a critical opportunity for addressing suboptimal diabetes control because it allows for coordination of care⁴ and contextualization of diabetes with other medical and nonmedical concerns.⁵ Unfortunately, providers often have limited time with patients and commonly need to address multiple, competing health concerns. Recent research has shown that the number of clinical items addressed during visits has outpaced changes in visit duration, increasing the strain on the already resource-limited primary care visit.⁶ Such time limits may leave important topics, such as diabetes care plans, deferred to the end of the visit or entirely unaddressed.^7,8 A better understanding of patient-provider communication during the primary care visit offers the chance to identify potential opportunities to promote effective diabetes management.

We conducted 2 studies to quantitatively analyze in-person visit interactions between primary care providers and their patients with suboptimal glycemic control. In the first study, we examined visit characteristics associated with diabetes management changes among patients who were prompted to identify their visit priorities prior to their visit. In the second study, we built on prior work among Spanish-speaking patients^9,10 to examine the relationship between question-asking and glycemic control. Together, these 2 studies provide novel insights into doctor-patient communication during the primary care visit.

METHODS

Setting

Visit audiotaping took place between November 2017 and January 2020 with 14 primary care providers across 8 primary care practices in Kaiser Permanente Northern California (KPNC). KPNC is a nonprofit integrated care delivery system providing care for more than 4.4 million adult members throughout Northern California,¹¹ including over 360,000 members with diabetes.¹² The distribution of member demographic and socioeconomic characteristics is diverse and similar to that of the area population.¹¹

Study Design

We used quantitative analysis of visit communication patterns obtained via professionally transcribed visit audio recordings to conduct a detailed examination of real-time primary care visit communication. We applied this methodology to 2 separate patient-provider cohorts.

Eligibility and Recruitment

Patient eligibility criteria for both studies included at least 1 year of membership at KPNC, T2D with a last measured suboptimal HbA1c (HbA1c > 7.0%), and an upcoming encounter with their primary care provider. Visits were restricted to language concordant (Spanish or English) patient-provider dyads. Informed consent was obtained from patients and providers to audio-record eligible visits. Study 1 was a visit audio recording substudy of a larger clinical trial (Pre-Visit Prioritization or Complex Patients With Diabetes, ClinicalTrials.gov NCT02375932)¹³ in which eligible English-speaking patients were sent a secure electronic message prior to their primary care visit asking them to select their top 1 or 2 visit priorities from a list of 5 options (important changes in your life, new/important health issues, medication concerns, diabetes-related concerns, and mood/motivation).¹⁴ Study 2 was conducted independently using the same analytic methods but focusing instead on Spanish-speaking patients with Spanish-speaking providers. Because they were not part of the larger clinical trial, these patients did not receive a previsit survey. Audio recordings were professionally transcribed for all visits. The Kaiser Permanente Institutional Review Board approved this study.

Statistical Analysis

The visits transcripts were reviewed by members of the study team (DA and RG for English language visits, and LR [a native Spanish speaker] for the Spanish language transcripts). For each transcript reviewed, the reviewers abstracted the following visit characteristics: visit length, number of patient-provider exchanges, number of questions asked by providers and by patients, and the timing of diabetes-related discussions after standardization of visit length. “Exchanges” were determined by the number of times each individual spoke, and “timing of diabetes-related discussion” was defined by whether initial mention of diabetes occurred in the first or second half of the visit. “Question-asking” was defined as a care-related inquiry that required a direct response either from patient or provider. In Study 1, the 2 reviewers also identified intravisit diabetes management changes, defined as either diabetes-related lifestyle counseling or medication intensification. We then examined the associations between visit characteristics and patient-selected visit priorities with these changes in diabetes management. In Study 2, we examined the association of question-asking with poor glycemic control (defined as HbA1c ≥ 10.0%). Visit communication patterns were also compared between the 2 study populations. In both analyses, differences in the outcomes by levels of the exposure variables were compared using χ² or Fisher’s exact test. For Study 1, independent predictors of diabetes management change were additionally investigated using a multivariate logistic regression model that adjusted for visit length, HbA1c, and selection of T2D as a visit priority. Analyses were conducted using SAS version 9.4.

RESULTS

We Analyzed 51 Unique Patient-Provider Visits (Table 1).

Table 1.

Summary of patient and visit characteristics, studies 1 and 2

Variable	Study 1 (n = 31)	Study 2 (n = 20)	p value
Patient characteristics
Age, y (SD)	62.9 (9.7)	55.1 (9.3)	0.01
Women, n (%)	18 (58.1)	10 (50.0)	0.57
A1C, % (SD)	9.0 (1.2)	9.4 (1.2)	0.24
Race/ethnicitya, n (%)
White	11 (42.3)	-	-
Black	4 (15.4)	-	-
Asian	9 (34.6)	-	-
Other	2 (7.7)	-	-
Visit characteristics
Visit length, min (SD)	23.4 (7.7)	28.5 (9.4)	0.04
Word count per 15 min, n (SD)
Total	2001.4 (344)	1501.1 (437)	0.64
Patient	841.8 (255)	524.1 (204)	0.63
Physician	1159.6 (309)	977.0 (349)	0.06
Questions per 15 min, n (SD)
Physician	34.9 (15.4)	27.7 (10.6)	0.07
Patient	7.5 (3.7)	4.5 (2.9)	0.004
Exchanges per 15 min, n (SD)	137.0 (42.8)	94.3 (49.5)	0.002

^a.Race/ethnicity information was missing for 5 participants.

SD = standard deviation.

Study 1: Factors Associated with Change in Management

In Study 1 (n = 31), the average patient age was 62.9 ± 9.7 years, 58.1% (18/31) were women, 57.7% were non-White (15/26, including 4 Black, 9 Asian, and 2 other; 5 were missing race/ethnicity information), and the mean HbA1c was 9.0% ± 1.2%. Visits averaged 23.4 ± 7.7 minutes and included 5.6 discussion topics (range, 2-9), 137 ± 42.7 exchanges between patients and providers, and 42.4 ± 17.4 questions asked per 15 minutes (34.9 ± 15.4 by provider and 7.5 ± 3.7 by patient).

Two-thirds of patients (21/31, 67%) selected diabetes as either a first or second visit priority. Identification of diabetes as a visit priority by the patient prior to the visit was the only factor significantly associated with changes in diabetes management during the visit (Table 2). The results remained unchanged when limited to either first or second priority. Nearly all patients (11/12, 91.7%) who had changes in diabetes management during the visit had selected diabetes as a visit priority. In contrast, among patients with no changes in care arising from the visit, only 52.6% (10/19) had selected diabetes as a priority (p = 0.02). Selection of diabetes as a visit priority by the patient remained significantly associated with diabetes management change after adjusting for visit length and HbA1c (adjusted odds ratio = 10.1; 95% confidence interval = 1.06-95.5; p = 0.045). Among the patients who did not select diabetes as a visit priority, the topic of diabetes was more likely to be discussed toward the latter half of the visit (31.6% of visits vs 8.3% of visits for which patients prioritized diabetes), although this difference did not reach statistical significance (p = 0.2).

Table 2.

Patient and visit characteristics by change in diabetes management (study 1)

Variable	No change (n = 19)	Lifestyle and/or medication change (n = 12)	p value
Patient characteristics
Age, y (SD)	62.2 (10.2)	64.0 (9.1)	0.61
Women, n (%)	11.0 (57.9)	7.0 (58.3)	0.98
A1C, % (SD)	9.1 (1.4)	9.0 (0.9)	0.83
Visit characteristics
Visit length, min (SD)	22.8 (8.0)	24.2 (7.5)	0.63
Distinct discussion topics, n (SD)
Total	5.5 (2.0)	5.7 (2.0)	0.79
Non-diabetes related	3.8 (1.9)	3.5 (1.5)	0.66
Diabetes related	1.7 (0.9)	2.2 (0.9)	0.16
Word count per 15 min, n (SD)
Total	1929.9 (379)	2114.6 (254)	0.15
Patient	827.4 (263)	864.7 (250)	0.70
Physician	1102.5 (301)	1250.0 (312)	0.20
Questions asked per 15 min, n (SD)
Physician	31.1 (11.7)	41.0 (18.8)	0.08
Patient	7.1 (3.6)	8.1 (4.0)	0.47
Verbal exchanges per 15 min, n (SD)	92.3 (26.7)	90.9 (22.3)	0.89
Patient-defined visit priorities, n (%)
Diabetes is any priority	10.0 (52.6)	11.0 (91.7)	0.02
Diabetes priority			0.02
First	5.0 (26.3)	9.0 (75.0)
Second	5.0 (26.3)	2.0 (16.7)
All priorities are addressed	16.0 (84.2)	11.0 (91.7)	1.00
Diabetes discussion
Diabetes topic not mentioned until second half of visit, n (%)	6.0 (31.6)	1.0 (8.3)	0.20

SD = standard deviation.

Study 2: Question-Asking by Latinos

In Study 2 (n = 20), all patients were Latino. Mean age was 55.0 ± 9.3 years, 50% (10/20) were women, and the mean HbA1c was 9.4% ± 1.2%. Asking fewer questions was significantly associated with higher HbA1c. Patients with poor glycemic control (HbA1c ≥ 10.0%) asked fewer questions compared with patients with HbA1c < 10% (3.4 ± 1.8 vs 10.7 ± 6.9 questions per 15 minutes; p = 0.004).

Compared with the English-speaking patients in Study 1, visit lengths in Study 2 were longer (28.5 ± 9.4 minutes) with fewer exchanges between patients and providers (94.3 ± 49.5 exchanges). After standardizing visit length per 15 minutes, Study 2 patients also asked significantly fewer questions than Study 1 patients (4.5 ± 2.9 vs 7.5 ± 3.7 per 15 minutes; p = 0.004) (Table 1).

DISCUSSION

Quantitative content analyses of primary care visit discussions provide a unique opportunity to gain insight into doctor-patient interactions and their association with diabetes-related health outcomes. In our analyses of patients with suboptimally controlled T2D, we found that patients who prioritized diabetes as a topic of discussion prior to the visit were more likely to have a change made to their diabetes management during the visit. Among Spanish-speaking patients and their Spanish-speaking providers, we found that even in the absence of language barriers patients who asked fewer questions had poorer glycemic control.

Previously described barriers to optimizing the primary care visit include short visit lengths and the need for more collaborative communication.¹⁵ Because all patients in our study had elevated HbA1c measured prior to the analyzed primary care visit, evidence-based guidelines would have recommended that these patients would have a change in their diabetes management (eg, lifestyle modification advice, referral to health education or other services, medication initiation or dose adjustment) as a consequence of seeing their primary care provider. In Study 1 we found that factors such as visit length, number of topics discussed, questions asked, and current HbA1c level were not associated with a management change. Rather, patients who selected diabetes as a priority for their visit were more likely to have a management change. In contrast, patients with suboptimal glycemic control who did not prioritize diabetes were more likely to discuss the topic during the latter half of the visit and less likely to have a regimen change. These findings suggest that structured visit preparation may help facilitate more productive doctor-patient interactions related to diabetes management. This strategy of planned patient preparedness may help patients bring their diabetes-related needs to the forefront of the visit so that the physician may be able to better address them. It may also help mitigate the “hand on the doorknob” phenomenon, where important issues are raised at the end of the visit⁷ and thereby allow for more time to discuss and implement changes in diabetes management. Structured visit preparation may also act as a reminder for physicians to check in on diabetes management, given its importance, regardless of whether patients perceive it as a priority.

The Study 1 results confirm prior qualitative work showing that having patients identify priorities can help organize visits and enable more productive doctor-patient interactions.¹⁵ Previous research has also shown that patients who received previsit prioritization surveys were more likely to prepare a list of questions for their physicians and were given more choices about their treatment.¹³ The elicitation of the patient perspective allows clinicians and patients to engage in meaningful conversations and contributes to patient-centered care.¹⁶ Moreover, research has demonstrated that patient participation in care has made patients more effective in garnering information from physicians and has improved diabetes management.¹⁷ Our results support these findings by suggesting that having diabetes as a previsit priority may promote interactions that facilitate a change in diabetes management.

Prior studies of Spanish-speaking patients with T2D have shown improvements in glycemic control with language-concordant providers and suggest that language discordance can be a barrier to effective care.^10,18 Removing the language barrier in our study allowed us to take a closer look at the relationship between question-asking and HbA1c, previously identified as an important issue among Latino patients.⁹ We found a significant association between less question-asking and poor glycemic control in our study. This finding suggests that improving patients’ confidence with and skills for question-asking may represent a potentially powerful strategy in diabetes management. Prior research has underscored that individuals with diabetes have one of the highest desires for information among individuals with various chronic conditions^19,20 and that having more information improved self-management and disease-related patient engagement in the decision-making process.^19,21 Because questions are a key indicator of patient engagement, patients who ask more questions are likely to receive more detailed explanations and have a greater role in decision-making.²² The frequency with which patients ask questions is also related to the degree to which doctors provide diagnostic and treatment information.²³ Prior findings show that an intervention helping patients develop a list of questions for their visit with a physician resulted in increased patient activation levels.²⁴ This further emphasizes the significance of question-asking as it relates to doctor-patient communication.

The comparison of Study 1 and Study 2 demonstrated that, even with language barriers removed, Spanish speakers asked fewer questions than their English-speaking counterparts. A common construct in Latino culture is the deference to physician authority,^9,25 which may lead to a lack of visit preparation and question-asking given the cultural expectation that it is the responsibility of the provider to set the agenda and know what to ask.²⁶ Our study further builds on prior research that emphasizes the development of question-asking skills to promote patient engagement specifically for Latinos.⁹ Furthermore, patient-facing communication interventions, such as modeling of question-asking behaviors, have shown success in helping patients obtain more information from physicians and exhibit greater involvement during the visit.²⁷ Future research should investigate question-asking behaviors of other cultural groups and comparing men vs women in larger sample sizes.

Some limitations of our study should be considered. The relatively small number of transcribed visits limits our power to show smaller differences between groups. However, this is balanced by a greater depth of detail in the visits included in our study and the relatively large effect sizes of our significant predictors. Further research with a larger study sample might reveal more differences between groups. In addition, though we were able to show several meaningful associations, our observational study cannot be used to determine causality. Future interventional work is needed to confirm the value of our findings. Finally, although there may be some concern that recording primary care visits may have caused participants to shift their behaviors due to the Hawthorne effect, other investigators have previously shown that visit recordings do not change visit behaviors.²⁸

CONCLUSION

Quantitative visit content analysis can help shed light on the content of patient-provider discussions during primary care visits. Our analyses identified patient previsit topic prioritization and increased question-asking as 2 potential domains for further intervention. These results suggest that new interventions to improve patient visit preparedness and confidence with asking questions represent a promising strategy for improving diabetes primary care.