The results reported here provide the first prospectively collected estimates of change in outcomes during and following the delivery of ND care to people with type 2 diabetes. Provision of ANC was associated with a variety of improvements in diabetes self-management including: increased self-monitoring of blood glucose, improved diet, increased physical activity, greater self-efficacy, improved mood and reduced problem areas in diabetes. Glucose control also improved in the ANC cohort, while remaining unchanged in the usual care cohort. Although we attempted to minimize confounding between groups by performing adjusted analyses for our primary clinical comparisons, our study was observational and therefore these encouraging findings cannot be causally attributed to ANC. Nevertheless, several factors suggest probable causation. Foremost, causation is supported by the temporality of the observations with the greater improvements corresponding to the period of greater utilization of ANC, i.e., the first 6-months. Secondly, the treatment recommendations delivered by the NDs closely correspond to the types of changes reported by participants. Finally, the majority of participants reported they made self-care changes as a result of their ANC.
Although there may have been overlap in the content of ANC visits with that of visits to primary care providers, diabetes educators, and nutritionists during the typical GH care for diabetes, any benefits from the delivery of these services to participants during the study period would have been included in our estimates of change in both groups.. However, even if ANC was responsible for the improvements measured in this study, the simultaneous increases in utilization of usual care medical services in the ANC group make it impossible to distinguish between direct effects of the ANC intervention and indirect effects of ANC, which may have stimulated patient re-engagement in their health and health care more generally.
Permanently changing behaviors is notoriously difficult for patients. Comparable to several large clinical trials that have tested behavior-targeted interventions, our results suggest the greatest changes in behavior and clinical risk occurred during the most intensive phase of the intervention (months 1 through 6) and then decreased thereafter [21
]. Yet, unlike most previous trials, our intervention was delivered within the context of routine care by physician-level providers and not according to a standardized protocol. This lack of treatment standardization or inclusion of special incentives for patients to adhere to a fixed visit schedule likely resulted in lower utilization, and possibly fewer benefits, than may have occurred with a more standardized intervention. Typical barriers for patients to adopt longer-lasting behaviors include depression, reduced self-efficacy and low patient motivation [15
]. Therefore, the persistence in improvements in mood, self-efficacy and motivation to change lifestyle beyond the period of greatest utilization of ANC, , is a promising sign that longer lasting changes in patient behaviors may be possible through optimization of ND visit content and frequency.
The pragmatic delivery of care, combined with collection of patient-reported and clinical outcomes following a CAM intervention, are unique features and considerable strengths of the study design because they provide a multi-faceted view of the potential real world effects of ANC implemented within usual care in a real world clinical setting (vs. the very structured, but often untranslatable, components of a clinical trial protocol). Inviting patients from a managed care setting is also a unique feature of this study compared to past research on ND care in diabetes because the patients were not explicitly seeking additional care, and therefore our results may be more generalizable than past results from samples of self-selecting patients [7
]. However, these differences also make it difficult to compare the changes in HbA1c we observed in this study to those reported by earlier evaluations of ND care. Prior retrospective assessments of ND care reported a mean change of −0.65 % in patients at an academic clinic [7
]. However, patients may have been more motivated to change self-care than the participants in the current study because they self-selected for ND care. Also, on average, the patients included in that study completed eleven visits over a twenty-seven month observation period compared to only four visits over a twelve-month period in the current study. A recent uncontrolled trial of a ND nutritional program also found a reduction in HbA1c (−0.4 %) after just 3 months [10
], but this study included a well-developed protocol in contrast to the pragmatic, real-world approach we used in the current study. The generally positive findings of all of the observational studies of ND care support the need for carefully designed randomized trials.
One important limitation in the generalizability of our findings is the ND care in this study was applied as an adjunct to usual care, which may have limited its benefits. It remains unknown if the changes in PROs or clinical risk could have been increased or extended had the co-utilization of usual care and ND care been formally coordinated (or “integrated”) or if ND care had been offered as a primary care option. Because the delivery of behavioral change counseling is infrequent in typical primary care [27
], very limited data are available regarding the potential impact of an ongoing, routine emphasis on behavioral change coupled to routine clinical services. Future research should investigate the formal integration of usual and ND care, and extend the treatment duration of ND care, in order to further evaluate its potential for promoting lasting behavior change.
There are several additional limitations to our study: 1) because our study population was mainly Caucasian and relatively well educated, the results may not apply to regions with different demographic characteristics, 2) our study was conducted in a managed care setting that imposed some constraints on ND practices (e.g., NDs could not directly order additional laboratory tests for GH patients or prescribe medications without approval of the patient’s GH primary care provider), and 3) because dietary supplements are not covered by insurance, many participants reported they did not use the supplements recommended by the ND, which may have reduced the potential clinical benefits of ANC. For example, although the clinical effectiveness of many dietary supplements remains unknown, omega-3 fatty acids have strong evidence for improving outcomes in people with high cardiovascular risk and small, randomized trials suggest chromium, cinnamon and coenzyme Q10 reduce blood glucose and improve other risk factors for developing complications of diabetes [29
Based on our observations of improved glucose control, self-care, self-efficacy, and mood plus reduced problem areas in diabetes after initiating ANC, future research on ANC is justified and should employ a randomized, controlled trial design that permits determination of causality. To fully evaluate the potential value of ANC, randomized comparative effectiveness trials should compare unrestrained, “whole-system” ANC including dietary supplements and a full scope of practice with usual care. Additional pragmatic trials should compare “best practice” ND protocols to unconstrained ND care as it is practiced in the community. Finally, because of the high frequency of recommendations for dietary supplements in ND practice, and emerging evidence of potent clinical effects from placebos [34
], careful consideration is needed for how to best evaluate the effectiveness of the highly variable dietary supplement recommendations made by NDs in practice.
Despite the need for continued research on ND approaches, we believe the results of this study have important implications for patients’ diabetes care, especially for patients interested in trying ANC. Our findings suggest numerous possible benefits and minimal risk for patients willing to use ANC. Our findings also suggest that ANC does not disrupt patients’ engagement in usual care and, in fact, may increase it. Finally, the observed increases in use of primary care services and medications for diabetes indicate that ANC was used as a complement to usual care and not as an alternative. These findings should be reassuring to usual care providers concerned that ANC may negatively impact medical care for type 2 diabetes—our findings suggest just the opposite.