Health Coaching by Medical Assistants to Improve Control of Diabetes, Hypertension, and Hyperlipidemia in Low-Income Patients: A Randomized Controlled Trial

As a class, we felt the opening sentence was a strong way to start the introduction. The introduction conveyed both the significant mortality and economic cost that cardiovascular disease (CVD) inflicts on our society. We were surprised to read about the low adherence rate to medications and lifestyle changes by patients. We discussed other barriers associated with proper health care. One major barrier we discussed was the lack of education and knowledge about CVD. If patients are not knowledgeable regarding their condition, it will be harder for them to properly manage their condition.

Regarding the methodology of the study, we discussed a preference for a double blinded study. Journal club members were concerned with patients being aware of the condition they were placed in (usual care condition or intervention condition) prior to the study resulting in a possible Hawthorne effect. Regarding the procedure, we discussed the follow-up time period (initial visit and 12 months after). Three and six-month follow up to examine patterns and changes in A1c, blood pressure, and fasting lipids readings would have added to the study. The inclusion criteria presented by the researchers was clear and feasible, however, the exclusion criteria was vague. We reviewed the prior published article on the methodology for details but still felt there was limited information. The paper listed uncontrolled schizophrenia as the only exclusion criteria, but it alluded to some other possible exclusion criteria. We wondered if other criteria, for example, patients diagnosed with sleep apnea and Cushing's disease were considered.

The study group conducted the health coaching training themselves rather than going through an accredited health coach certification. The journal club members explored how training offered by the study group may have been similar or different from training offered by an accredited one-two year health coaching program. The results section was interesting, particularly the fact that while patients met their goals, no significant difference was found in blood pressure measurements. After discussion, we decided that blood pressure measurement may not be stable over time due to multiple factors that could impact results. For example, blood pressure is very reliant on user competency, anxiety state (white coat hypertension), and any other co-morbidity present. We feel these type of factors may have contributed to the nonsignificant results. Lastly, the journal club members pondered if the length of time since diagnoses of three medical conditions could impact the results. For instance, someone who has had their disease for a longer period of time could be more treatment resistant than those recently diagnosed.

Overall, we believe that the implementation of health coaching can have a positive impact on patient care, reduce physician burnout, and address physician time constraints. More studies need to be conducted to evaluate the cost and clinic efficiency of health coaching in addition to its impact on the patient-doctor relationship.

Competing interests: None declared

Dear Editors,

We thank John Mihelcic, Timothy Feldheim, Robert Horsley, and Alesia Jones for their thoughtful comments and questions regarding our medical assistant health coaching study published in the March/April issue of Annals of Family Medicine. We wanted to share more information about the study in response to their questions.

While Mihelcic and colleagues are correct that patients may differentially elect to take part in a research study and that participating patients could differ from others in their motivation for change, we would observe that a relatively high proportion (67%) of patients who were identified as eligible elected to take part in the study. Moreover, patients were randomized by arm, so their degree of motivation would in theory be randomly distributed across arms. We did assess several factors at baseline which support the success of randomization. In addition to those reported in the paper (e.g., no difference in demographics or clinical values), there was no significant difference between arms in reported self-efficacy for chronic disease management (mean of 7.6 versus 7.8 for coaching versus usual care group; p=.29); medication adherence (mean of 10.2 v. 10.4; p=.33); or knowledge of clinical values at baseline.

Regarding exclusion criteria for serious health conditions, we did not systematically exclude patients based on diagnosis; rather, we asked PCPs to exclude patients if they believed that severe or terminal illnesses or cognitive dysfunction would not allow them to meaningfully participate with a health coach. There were, for example, people with well -managed schizophrenia or who were under treatment for cancer who took part in the study. Health coaches for this study were not part of existing staff; they were hired for the health coaching program, and therefore they did not have existing relationships with patients. In settings where such prior relationships exist, these may serve as a catalyst to a successful coaching relationship.

Both arms of the study had access to all resources available within the clinical sites, such as chronic disease management classes, health educators, or behavioral health resources. There was no difference at baseline between arms in their engagement with educational classes, health educators, or behavioral health. However, at 12 months, patients with a health coach were more likely than those in usual care to report having engaged with health educators (31.0% v. 18.5%; p<.01) and social workers (23.8% v. 6.9%; p<.001). This suggests that health coaches successfully encouraged patients to take advantage of existing clinical resources. Connecting patients with social workers and community resources was one important way that health coaches addressed patients' social needs.

In regard to the definition of usual care, the two sites shared many characteristics. Both sites involved their medical assistants in proactively identifying preventive and chronic care needs for their patients through panel management. Both sites had integrated behavioral health programs and health educators. Neither site had an EHR at the beginning of the study; both acquired one by the end of the study. One prominent difference in usual care was that at the time of this study medical assistants in Clinic A worked with different clinicians every day, while Clinic B paired medical assistants and clinicians in stable teamlets that worked together every day.

Mihelcic and colleagues offer some insightful ideas on how the limitations of the study might be addressed in practice. We agree that considering cultural concordance in the hiring process, providing a clear protocol and supervision, and possibly changing the coach would be appropriate ways to ensure that a high quality of coaching is delivered. We also agree that some clinical measures may require more time to impact than others; to that end, we are completing analysis on an observational study of this cohort that was conducted 12 months after the end of the intervention.

Competing interests: None declared

The overall purpose of this article was to assess whether or not in-house health coaching would be beneficial to patients with cardiovascular and metabolic risk factors. Health coaches are individuals in the health care field whose focus is on changes in lifestyle and health wellness, usually for chronic disease management, but can be any health challenges. Health coaches also give education on mental health and perform motivational interviewing for behavior change. They employ a psychosocial approach to medication use and lifestyle change. For proper health coaching to occur, long term coordination and interaction with the patient is desired. Health coaching programs vary from 40 hour training sessions to 1-2 year long programs.

The introduction of this article focused on the importance of gaining a control on health care costs of chronic diseases, especially with cardiovascular disease. The authors made a strong case for health coaching to control cost and decrease morbidity and mortality. Also described were the particular problems associated with chronic disease management and patient adherence to treatment plans. The authors noted that disparities in the care of minorities seem to be more significant than other demographics. Several social barriers to care are mentioned in the article, but the authors fail to investigate this aspect when describing the role of health coaching.

The authors reported while previous studies have examined the utilization of health coaches in chronic disease management, studies were limited due to methodological issues such as small sample size or lack of power, and other study gaps. The authors were clear on the intent of their study and then addressed gaps that existed prior to the study. However, our discussion group discussed variation in the term "usual care" and wanted more details on how that term was defined in their study especially due to the fact that two different clinics were used and may have different clinical practices.

Our group also discussed the possibility of participant bias as patients who are compliant to treatment may be more likely take part in a research study than those who are not compliant. The group suggested future studies examine frequency of visits by the patients in both the health coaching and control group. Serious health conditions were excluded; group was interested in the list of medical conditions that were excluded. Patients were self-randomized into either an intervention group (those who would receive health coaching) and control group (those who would receive normal interventions and resources) by selecting envelope with random binary sequence. The group felt this was a strength of the study. The group thought it would be interesting to assess motivation to change and knowledge of diseases at baseline. Health coaches were utilized at both sites and received similar training of 40 hours. It was unclear if the health coaches were a part of the existing staff and may have already developed a relationship with some patients prior to the study.

The group wondered if patients in both conditions received resources and if the resources were monitored for usage. It was also unclear if different lab panels were used at each site. This could be problematic when comparing lipid levels as a primary outcome since it may not have been standardized between both clinics. The primary goal of achieving 1 of the 3 goals was considered reasonable by the group, however, the authors could have described how patients were already meeting one or more of the goals.

Authors noted limitations related to study issues at Clinic B. First, the health coach at clinic B was absent for 8 weeks. The health coach had fewer patient interactions and appeared to not implement coaching principles properly. These problems could have been overcome by having an alternative coach ready in such situations or both clinics having two coaches. A structured protocol (or checklist) could also have been implemented so that coaches at site A and B intervened in a more similar manner. Culturally matching coaches with the patient demographic of each clinic may have also impacted outcomes and should be considered in future studies. Also, health coaches receiving regular supervision may have also addressed potential problems with implementing coaching principles. The authors found the expected results except for changes in blood pressure adding to the mixed findings in the literature. It may take longer to reduce hypertension and perhaps a long term study would yield the expected results.

Overall, we found this paper very interesting. This paper described the significant impact health coaches have on long term care of chronic diseases. As pressures of time and productivity increase on the physician, a niche for health coaches in today's health care model is emerging and needed. As the authors suggest, further research should focus on what characteristics of health coaching are most effective and if it differs by type of disease management. Lastly, future studies may include physicians and nurses' experiences regarding the utility of health coaches in family medicine.

Competing interests: None declared

To the Editor,

We thank Dr. Crowley and Dr. Bosworth for their comments and insights regarding health coaching in response to our recent article in this journal and appreciate the opportunity to respond. Drs. Crowley and Bosworth note that health coaching appeared to be most effective for improving control of diabetes and had no effect of control of hypertension. While we also noted this in the article it does bear emphasizing, particularly as our a priori primary and secondary outcomes were composite measures of control of any, or all, of the three cardiovascular risk factors (diabetes, hypertension or hyperlipidemia). The fact that nearly half (49%) of the patients in the coached group achieved reached the target for glycemic control (defined as A1C< 8.0%), compared to 28% of the usual care arm, seems clinically important, especially given that these were patients already receiving care at clinics that provided them with access to language concordant diabetes support, including nurse educators and dieticians.

Drs. Crowley and Bosworth ask if health coaching "would be best suited for patients who are only minimally above their glycemic target". We think the answer is clearly no. In fact we found a somewhat larger decrease in A1C for patients with higher baseline A1C, though as expected, patients with the highest A1C were less likely to reach a level < 8.0%. Health coaching is not meant to substitute for, but to supplement, other available services; therefore it does not make sense to limit it to patients minimally above their A1C threshold.

The finding, noted by Drs. Crowley and Bosworth, that coaching was successful at one site but not the other complicates our results, but also provides a 'reality check' that the effectiveness of coaching cannot be taken for granted, but will likely vary by characteristics of the coaches, patients, and the clinical environment. As discussed in the paper, the coach at this site struggled with the role of health coach and failed to deliver the intended intensity of intervention. Unfortunately it is not possible from our data to isolate the reason or reasons for the difference between clinics, and we believe it would be premature to require cultural concordance for health coaching based on our results.

Competing interests: None declared

To the Editor,

We read with interest "Health Coaching by Medical Assistants to Improve Control of Diabetes, Hypertension, and Hyperlipidemia in Low-Income Patients: A Randomized Controlled Trial" by Willard-Grace, et al.[1] Health coaching by medical assistants is an appealing adjunct to chronic disease management because it can improve patient understanding of and adherence to care plans,[2] can enhance the cultural concordance of medical care, and is potentially cost-effective.[3]

Though health coaching holds much promise, this study highlights areas of uncertainty that may hamper widespread implementation. For example, though this study primarily examined composite outcomes, the impact of health coaching varied by disease state. Per Tables 4 and 5, health coaching significantly improved only diabetes control relative to usual care - it appeared to have less relative impact on hyperlipidemia, and even less on hypertension. In order to inform appropriate targeting of health coaching, further work should explore these findings. It may be that health coaching should focus specifically on diabetes rather than global cardiovascular disease risk reduction.

The magnitude of the gains associated with health coaching also warrants consideration. Leaving aside the relative lack of effect on blood pressure and cholesterol, health coaching in this study lowered mean hemoglobin A1c by 0.7%, and less than half of intervention patients achieved their hemoglobin A1c goal. Given this moderate impact, we need to better understand which patients would be best served by health coaching versus alternative, more intensive approaches for diabetes, like pharmacist medication and behavioral management. Though any improvement in hemoglobin A1c lowers complications,[4] would health coaching be best suited for patients who are only minimally above their glycemic target (particularly if more intensive, effective options are available)?

Finally - and importantly - this study indicated that health coaching was beneficial only in the 'Clinic A' setting, where there was cultural concordance between medical assistants and patients. This finding supports cultural concordance as an important aspect of the medical assistant-patient relationship. Should cultural concordance therefore be a requirement of health coaching? To what extent was the lack of impact in the 'Clinic B' population an interventionist effect? These questions are highly relevant to the implementation of health coaching in practice.

In addition to formally evaluating the cost-effectiveness of health coaching (an analysis the investigators indicate is forthcoming), further research addressing these remaining areas of uncertainty will help this strategy can fulfill its promise.

Sincerely, Matthew J. Crowley, MD and Hayden B. Bosworth, PhD

References
1. Willard-Grace R, Chen EH, Hessler D, DeVore D, Prado C, Bodenheimer T, Thom DH. Health coaching by medical assistants to improve control of diabetes, hypertension, and hyperlipidemia in low-income patients: a randomized controlled trial. Ann Fam Med. 2015;13:130-8.
2. Ghorob A. Health coaching: teaching patients how to fish. Fam Pract Manag. 2013;20:40-2.
3. Jonk Y, Lawson K, O'Connor H, Riise KS, Eisenberg D, Dowd B, Kreitzer MJ. How effective is health coaching in reducing health services expenditures? Med Care. 2015;53:133-40.
4. Stratton IM, Adler AI, Neil HA, Matthews DR, Manley SE, Cull CA, et al. Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. 2000;321:405-12.

Competing interests: None declared