A functional medicine approach to diabetes control and quality of life: an open-label randomized pilot study

Highlights

• •First randomized trial comparing a functional medicine (FM) approach to just usual care (UC) in insulin-dependent type 2 diabetes (T2D).
• •FM added to UC did not significantly improve the primary outcome of insulin discontinuation without increasing HbA1c or HbA1c <7% at 12 months.
• •Other outcomes including BMI, quality of life scores, hypoglycemia, and medication burden, showed no significant difference between FM and UC.
• •Study highlights real-world adherence challenges to complex FM regimens and their relevance in T2D management.

Abstract

Background

There has been increasing public interest in a functional medicine (FM) approach to the care of type 2 diabetes (T2D). However, it is unclear if this approach can further improve diabetes management beyond conventional diabetes care. This study compared the addition of a FM approach (intensive lifestyle modifications and tailored dietary supplements) plus usual care (FM) to conventional endocrinology management alone (usual care, UC).

Methods

In a 24-month unblinded randomized trial, participants with type 2 diabetes (T2D) were randomized to receive either FM or UC. The primary endpoint was discontinuation of insulin with no increase in HbA1c or HbA1c < 7% at 12 months. Secondary endpoints included changes in HbA1c, insulin dosage by weight, cardiometabolic biomarkers, hypoglycemic episodes, total number of medications and quality-of-life scores.

Results

A total of 107 individuals were screened − 34 were randomized into FM and 37 into UC. Participants were 46% female and median age was 62 years. Median baseline HbA1c was 7.75%. Of those randomized, 47 participants completed the study (FM: 20; UC: 27); 15.8% of patients in FM arm achieved the primary endpoint compared to 11.1% in the UC arm (p = 0.68). Secondary exploratory endpoints including change in HbA1c, insulin dosage by weight, hypoglycemia, BMI, weight and quality-of-life did not differ significantly between arms.

Conclusion

The addition of FM did not lead to significant improvement in glycemic, metabolic or quality-of-life outcomes compared to usual care for individuals with type 2 diabetes.

Introduction

The International Diabetes Federation (IDF) estimates that 547 million adults (aged 20–79) worldwide have diabetes, with over 90% being type 2 diabetes (T2D) [1]. In the US, diabetes cost was estimated at $412.9 billion in 2022, driven mainly by care for T2D and related complications. This represents a significant burden on the US healthcare system [2].

The conventional medical approach, or usual care (UC), for T2D management consists of lifestyle modifications, such as diet, exercise, and weight management, combined with pharmacologic therapy to achieve glycemic control and prevent complications [3], [4], [5]. Insulin therapy remains a common component of pharmacological management, especially as the disease progresses [6], but it is a suboptimal treatment as it often exacerbates weight gain and can cause hypoglycemia. Lifestyle recommendations are foundational to the management of many chronic conditions including T2D. The American Diabetes Association (ADA) has specifically highlighted the importance of encouraging patients with diabetes to participate in diabetes self-management education and support (DSMES) and engage in guided lifestyle modifications [4]. Lifestyle interventions, which support weight loss and increase insulin sensitivity, have been shown to reduce the dose and number of medications for T2D [7]. While there is consensus on the importance of lifestyle modification, the use of dietary supplements is not currently recommended by the ADA or other guideline bodies due to insufficient evidence [4], [8]. Despite this, some recent meta-analysis and clinical trials have suggested that certain supplements may improve glycemic and lipid outcomes [9], [10], [11], [12], [13], [14].

There is increasing interest amongst the public in using a functional medicine (FM) approach to address various health issues, including diabetes. The FM model of care works to promote health and optimize function by addressing underlying causes, symptoms, and imbalances in biological networks [15], [16]. Specifically, the FM model of care focuses on the use of food as medicine to prevent, treat, and reverse chronic disease [16], while integrating dietary supplements as part of the care provided. In addition to nutrition and supplements, the FM model of care incorporates other elements of lifestyle modification including improving sleep hygiene, stress management and movement. Used together, and tailored to a patient’s unique biology and environment, these elements of FM might successfully reduce or eliminate the need for insulin. At this time, however, it is unclear how combining FM with conventional T2D care can improve clinical and patient-reported outcomes in patients with T2D. This current study aims to determine if a FM approach with its unique model of lifestyle intervention could improve glycemic outcomes, reduce insulin use, and improve quality of life for patients with T2D.

Methods

Study design and population

A prospective, unblinded, randomized-controlled study was conducted at a tertiary care center in the United States. The trial was approved by the institutional review board. All participants provided written informed consent. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline for randomized controlled trials [17].

An overview of study enrollment, randomization and follow-up is provided in Fig. 1. Participants were eligible if they were ≥18 years, diagnosed with T2D and taking insulin treatment for at least 12 months. Exclusion criteria were positive glutamate decarboxylase antibody, HbA1c > 12%, history of diabetic ketoacidosis (DKA), pregnancy, known diagnosis of cognitive impairment or dementia, estimated glomerular filtration rate < 45 ml/min/1.73 m. Full eligibility criteria are in Appendix A.

Fig. 1.

Flow chart summarizing participant enrollment, allocation, follow-up, and analysis in the study. Participants were assigned to either the Functional Medicine (FM) group or the Usual Care (UC) group. FM = Functional Medicine; UC = Usual Care.

Study visits

After an initial screening/baseline visit, a stratified randomization procedure was used to allocate participants according to total daily insulin dose (≤75 units versus > 75 units). All participants continued to receive usual care from endocrinologists throughout the study. Participants were randomized to either continue UC delivered by the endocrinologists (UC arm) or to receive additional care from a functional medicine physician and nutritionist (FM arm). Following the screening/baseline visit, both arms had in-person study visits at 6, 12, and 24 months, during which physical exams, laboratory tests, and questionnaires were administered. In addition to these standard visits, participants in the FM arm had extra in-person visits with a physician at months 1, 2, 3, and 9.

Both groups also received nutrition-therapy visits. Participants in the UC arm had a scheduled dietitian visit at 1 month, whereas those in the FM arm had scheduled nutrition visits at months 1, 2, 3, 6, and 9. FM participants also had access to three additional visits with a Health Coach and three with a dietitian, which could be scheduled as needed to provide further individualized support. This higher visit frequency in the FM arm with physicians and other members of the health team was designed to reflect the real-world functional medicine practice, which incorporates a higher number of visits.

Monthly follow-up phone calls were conducted by a research coordinator throughout the first year and again at month 18 in the second year. For a detailed overview of study schedule see Appendix B.

The UC approach

The UC arm consisted of participants receiving nationally accepted, guideline-directed medical therapy for T2D. Care was delivered in accordance with the most recent ADA guidelines [18], [3].

The FM approach

The FM arm consisted of 2 phases: a diagnostic phase and a detoxification phase. The diagnostic phase consisted of symptom-based diagnostic testing and delivery of more general treatments including dietary modifications (i.e. elimination of foods associated with adverse reactions, etc.), a tailored supplement regimen, and other lifestyle modifications addressing exercise, stress reduction, sleep hygiene, and prevention of hypoglycemia. Commonly prescribed supplements included multivitamins, omega-3 fatty acids, alpha-lipoic acid, magnesium, vitamin D3, and berberine. The detoxification phase consisted of treating gut dysbiosis/leaky gut as well as providing treatment for heavy metal burden, if indicated. Details regarding the FM approach are available in Appendix C.

Primary and secondary outcomes

The primary endpoint was discontinuation of insulin with no increase in HbA1c or HbA1c < 7% at 12 months. This endpoint was chosen because it was agreed that it was unlikely to occur in the UC arm, as once patients begin insulin, they rarely discontinue it. Because it was believed that FM would be quite effective in getting patients of insulin, only a small number of patients would be needed to demonstrate effectiveness. Key secondary end points, for which multiplicity correction was applied, included change in BMI and the diabetes distress scale 17 (DDS17) at 12 months. Other secondary endpoints analyzed as exploratory outcomes, included changes in HbA1c, insulin dosage by weight, decrease in the presence of hypoglycemic episodes compared with baseline, total number of medications used and quality-of-life scores.

Quality-of-life was measured using a variety of questionnaires. DDS17 [19], [20], [21] focuses on diabetes-specific distress, while the Sleep disturbance, Pain, Anxiety, Depression, and Energy/Fatigue (SPADE) [22] questionnaire provides a general quality of life overview, the Patient Health Questionnaire (PHQ-9) [23] captures depression severity, and the Medical Symptoms Questionnaire (MSQ) [24] offers insight into a wide range of physical symptoms.

Statistical analyses

The estimated sample size of 45 patients per group was derived to achieve 80% power, based on preliminary internal data indicating that approximately 5% of patients in the usual care arm versus 30% in the functional medicine arm were expected to discontinue insulin at 6 or 12 months. Due to limitations of funding and challenges with recruitment during the pandemic, the study was terminated with 71 patients randomized.

Demographic and clinical characteristics were summarized using frequency and proportion (%) for categorical variables, and median with interquartile range (IQR) for continuous variables. A per-protocol approach was adopted in the analysis framework. The per-protocol population included all randomized patients who adhered to the study treatment and completed at least one post-baseline efficacy assessment. Patients who did not return for any of the follow-up visits were considered non-adherent to the study protocol and were excluded from the per-protocol population.

A complete-case descriptive analysis reported observed discontinuation rates of primary endpoint at 6, 12, and 24 months, with Clopper–Pearson exact 95% confidence intervals to reflect the precision of estimates given the small sample size and low event rates. Group comparisons at each time point used Fisher’s exact test. The primary analysis focused on the 12-month insulin discontinuation endpoint, with statistical significance defined as a two-sided α < 0.05. Discontinuation rates at 6 and 24 months were considered exploratory.

Binary secondary endpoint was analyzed using mixed-effects logistic regression with time-by-treatment interaction terms, reporting estimated probabilities and odds ratios (UC vs. FM) corresponding confidence intervals at each prespecified follow-up. Continuous secondary endpoints were analyzed using linear mixed-effect models adjusting for baseline values and including time-by-treatment interactions [25], with results expressed as estimated mean changes and between-group differences with confidence intervals. We did not adjust for multiple comparisons due to small sample size.

Models were assessed using standard residual diagnostics, and no violations of model assumptions were identified. Two key secondary endpoints (change in BMI and DDS17) were evaluated with Bonferroni-adjusted significance (α = 0.025) and corresponding 97.5% confidence intervals. All other secondary endpoints were summarized as exploratory point estimates with 95% confidence intervals, without multiplicity adjustment, and were not used for formal hypothesis testing. Analyses were conducted using R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Of 107 patients who were screened, 71 participants were randomized (FM: 34 and UC: 37). Twenty-four (33.8%) participants were lost to follow-up at 12 months, with 20 (58.8%) remaining in the FM arm and 27 (72.9%) in the UC arm. Patient demographics for the FM and UC arms are provided in Table 1. Baseline demographics and clinical characteristics of the study population were similar between both arms with a median age of 62 years (IQR 58, 66 years), median baseline HbA1c of 7.8%, (IQR 6.9%, 9.0%), and median insulin dose per weight of 0.39 unit/kg/day (IQR 0.25, 0.64 unit/kg/day). Table S1 details the various supplements utilized in the study and their frequency of use in the FM arm.

Table 1.

Comparison of demographic and baseline patient characteristics in UC vs. FM treatment groups.

Characteristic	N	Overalla	FM Arma	UC Arma
N		71	34	37
Age at enrollment	70	62 (58, 66)	60 (57.25, 64)	64.00 (58, 68.25)
Sex	71
Male		37 (52%)	18 (53%)	19 (51%)
Female		33 (46%)	16 (47%)	17 (46%)
Unknown		1 (1.4%)	0 (0%)	1 (2.7%)
Ethnicity	70
Hispanic or Latino		1 (1.4%)	1 (2.9%)	0 (0%)
Not Hispanic or Latino		68 (97%)	32 (94%)	36 (100%)
Declined		0 (0%)	0 (0%)	0 (0%)
Unavailable		1 (1.4%)	1 (2.9%)	0 (0%)
Race	71
White		50 (70%)	22 (65%)	28 (76%)
Black		18 (25%)	9 (26%)	9 (24%)
Native American		0 (0%)	0 (0%)	0 (0%)
Asian		0 (0%)	0 (0%)	0 (0%)
Unavailable		1 (1.4%)	1 (2.9%)	0 (0%)
Declined		2 (2.8%)	2 (5.9%)	0 (0%)
Weight (kg)	71	101.60 (87.50, 113.75)	105.10 (92.05, 113.10)	100.50 (87.00, 114.80)
BMI (kg/m2)	71	34.19 (30.13, 37.76)	34.45 (29.98, 39.76)	33.58 (30.23, 36.10)
HbA1C (%)	68	7.75 (6.88, 8.93)	7.50 (6.83, 9.08)	8.00 (6.88, 8.90)
LDL (mmol/L)	68	2.27 (1.83, 2.69)	2.33 (1.91, 2.60)	2.18 (1.68, 2.93)
Medications
Insulin dose per weight (unit/kg/day)	71	0.39 (0.25, 0.64)	0.42 (0.30, 0.56)	0.33 (0.24, 0.66)
Total medications number	71	11.00 (7.00, 16.00)	11.50 (7.00, 16.75)	9.00 (8.00, 15.00)
Having hypoglycemic events
Self-reported hypoglycemia episodes	71	31 (44%)	13 (38%)	18 (49%)
Hypoglycemia episodes confirmed by glucometer readings	71	24 (34%)	9 (26%)	15 (41%)
Any ED visits or hospitalizations for hypoglycemia	71	1 (1.4%)	1 (2.9%)	0 (0%)
Quality of Life Questionnaire
DDS17	71	34.00 (24.50, 48.00)	34.50 (25.50, 50.75)	34.00 (24.00, 43.00)
SPADE emotional distress	71	11.00 (8.00, 17.00)	12.00 (8.00, 19.75)	10.00 (8.00, 14.00)
SPADE fatigue	71	16.00 (11.00, 23.00)	16.50 (12.00, 29.00)	13.00 (10.00, 19.00)
SPADE pain interference	71	11.00 (8.00, 19.00)	12.00 (10.00, 22.25)	8.00 (8.00, 17.00)
SPADE sleep disturbance	70	19.00 (12.00, 24.00)	18.50 (11.00, 23.00)	19.00 (13.75, 24.25)
PHQ-9	71	4.00 (1.00, 7.00)	4.00 (1.00, 8.75)	3.00 (1.00, 7.00)
MSQ	71	29.00 (14.00, 51.50)	29.00 (17.25, 67.25)	28.00 (13.00, 42.00)
Abbreviations: FM = Functional Medicine plus Usual Care; UC = Usual Care; ED = Emergency Department; DDS17 = The diabetes-distress scale Questionnaire; SPADE = the Sleep disturbance, Pain, Anxiety, Depression, and Energy/Fatigue Questionnaire; PHQ-9 = the Patient Health Questionnaire; MSQ = the Medical Symptoms Questionnaire aMedian (IQR); n (%) Note that p-values are not presented per the CONSORT statement guidelines for baseline characteristics comparisons in clinical trials.

Primary and secondary endpoints

At 12 months, 3 (15%) participants in FM arm achieved the primary endpoint of insulin discontinuation with no increase in HbA1c or HbA1c < 7%, compared to 3 (11.1%) participants in the UC arm (p = 0.68). Similarly, no significant differences were observed at 6 or 24 months. (Table 2).

Table 2.

Descriptive, complete-case discontinuation rates of insulin with no HbA1c increase or HbA1c < 7%) at 6, 12, and 24 months.

Follow-up Time	Treatment Group	na	Eventsa	% Discontinued (95%Cl)b	p-valueb
6 months	UC	34	1	2.94 (0.07, 15.33)	0.34
	FM	31	3	9.68 (2.04, 25.75)
12 months	UC	27	3	11.11 (2.35, 29.16)	0.68
	FM	20	3	15 (3.38, 39.58)
24 months	UC	26	2	7.69 (0.95, 25.13)	> 0.99
	FM	19	1	5.26 (0.13, 26.03)
Abbreviations: FM = Functional Medicine plus Usual Care; UC = Usual Care; CI = Confidence Interval a1n: number of patients with non-missing discontinuation status at each time point. Events: number of events b% Discontinued = events/n * 100. Exact (Clopper–Pearson) 95% confidence intervals (CIs). p-values from two-sided Fisher’s exact test comparing UC vs. UC + FM. Primary endpoint at 12 months, tested at two-sided α = 0.05; 6- and 24-month comparisons are exploratory.

Change in BMI

Both arms demonstrated a total increase in BMI by 24 months compared to baseline (Table 3). At 6 months, the FM arm demonstrated a decrease of −1.49 kg/m², but this finding was attenuated at 12 months (−1.3 kg/m²) with regain at 24 months (1.5 kg/m²) compared with baseline. In contrast, the UC arm had an initial increase in BMI of 0.76 kg/m² at 6 months followed by a small improvement −0.64 kg/m² at 12 months, and weight gain of 0.49 kg/m² at 24 months. Both arms had similar BMI at every time point.

Table 3.

Weight trends over time.

Endpoint	Time Points	FM Armb	UC Armb	Differencec	p-valued
BMI (kg/m2)a	6 Months	−1.3 (−2.66, −1.01)	0.4(−0.88,0.78)	1.79 (0.61,2.96)	0.003
	12 Months	−2.4 (−2.08, −0.06)	−0.4(−1.37,0.42)	0.6 (−0.75,1.94)	0.38
	24 Months	2. 1 (−1.79,0.33)	0.5(−2.01, −0.22)	−0.39 (−1.78,1)	0.58

Endpoint	Time Points	FM Arme	UC Arme	Differencef
Weight Change (Kg)a	6 Months	−4.05 (−8.43,0.32)	2.33 (−2.11,6.78)	6.39(0.15,12.63)
	12 Months	−3.93 (−9.49,1.63)	−2.1 (−6.96,2.75)	1.83 (−5.55,9.21)
	24 Months	4.14 (−1.73,10.01)	0.95 (−3.81,5.72)	−3.19 (−10.76,4.38)
Abbreviations: FM = Functional Medicine plus Usual Care; UC = Usual Care; OR=Odds Ratio; CI = Confidence Interval aLinear mixed effect model was conducted to assess mean change at each time point, with time, treatment arms, the interaction of treatment arms and time, and baseline value of the outcome variable adjusted in the model. bEstimated mean change (Lower 97.5%CI, Upper 97.5%CI) cMean change difference between patients in the FM and UC arms (UC-FM) (Lower 95%CI, Upper 95%CI) dp value of contrast difference compared from UC to FM arm at specific time point, testing whether contrast difference is equal to 0, which means there is no difference between two arms. The determination of statistical significance was made using Bonferroni-adjusted p-value, setting the significance threshold at 0.025 after multiplicity correction. eEstimated mean change (Lower 95% CI, Upper 95% CI) fEstimated mean change difference between participants in the FM and UC arms (UC-FM) (Lower 97.5%CI, Upper 97.5%CI)

Quality of life

DDS17 scoring showed improvement (i.e. absolute score decreased) in both arms at 6 months (FM −5.65, UC −5.16) but that improvement was attenuated over the 24 months in the FM arm (−2.16) compared with UC arm (−7.44) where there appeared to be sustained improvement compared with baseline (Table 4). There was no statistically significant difference between the improvement in both arms at any point in time.

Table 4.

Changes in quality of life questionnaires over time.

Questionnairea	Time Points	FM Armb	UC Armb	Differencec	p-valued
DDS17	6 Months	−5.65 (−9.74, −1.56)	−5.16 (−8.69, −1.63)	−0.49 (−4.93,5.91)	0.86
	12 Months	−5.16 (−9.65, −0.67)	−7.36 (−11.09, −3.63)	−2.2 (−8.05,3.64)	0.46
	24 Months	−2.16(−6.66,2.33)	−7.44 (−11.22, −3.66)	−5.28 (−11.15,0.6)	0.078
Abbreviations: FM = Functional Medicine plus Usual Care; UC = Usual Care; DDS17 = The diabetes-distress scale Questionnaire; CI = Confidence Interval aLinear mixed effect model was conducted to assess mean change at each time point, with time, treatment arms, the interaction of treatment arms and time, and baseline value of the outcome variable adjusted in the model. bEstimated mean change (Lower 97.5%CI, Upper 97.5%CI) cEstimated Mean change difference between patients in the FM and UC arms (UC-FM) (Lower 97.5%CI, Upper 97.5%CI) dp value of contrast difference compared from UC to FM arm at specific time point, testing whether contrast difference is equal to 0, which means there is no difference between two arms. The determination of statistical significance was made using Bonferroni-adjusted p-value, setting the significance threshold at 0.025 after multiplicity correction.

Exploratory endpoints

Insulin use and hypoglycemia

The FM arm at 12 months demonstrated a larger decrease in HbA1c from baseline compared to the UC arm (FM: −0.31% vs. UC: −0.23%). By 24 months, only the FM arm maintained a decrease in HbA1c from baseline compared to the UC arm (FM: −0.29% vs. UC: 0.06%) (Table 5). Both arms demonstrated a decrease in insulin units used at 12 months (FM: −0.11 vs UC: −0.09 unit/kg/day). However, by 24 months the FM arm experienced an increase of 0.05 unit/kg/day compared with baseline, while the UC arm continued to have a decrease in insulin use compared with baseline of −0.11 unit/kg/day. At baseline (the 6 months prior to initiation of study), the FM arm exhibited lower probability of hypoglycemia compared to the UC arm (FM: 0.33 vs UC: 0.5). However, between 6–12 months the probability of hypoglycemia in the FM arm increased while it decreased in the UC arm (FM: 0.52 vs. UC: 0.32). This trend continued at 18–24 months, when the FM arm continued to have a higher probability of hypoglycemia compared to the UC arm (FM: 0.53 vs. UC: 0.2).

Table 5.

HbA1c, insulin dose, and hypoglycemia rates trends over time.

Endpoint	Time Points	FM Armc	UC Armc	Differenced
HbA1c (%)a	6 Months	−0.44 (−0.86, −0.01)	−0.07 (−0.43,0.29)	0.37 (−0.19,0.93)
	12 Months	−0.31 (−0.81,0.18)	−0.23 (−0.63,0.16)	0.08 (−0.55,0.71)
	24 Months	−0.29 (−0.77,0.19)	0.06 (−0.35,0.46)	0.34 (−0.28,0.97)
Insulin Dose (unit/kg/day) a	6 Months	−0.12 (−0.2,-0.04)	−0.07 (−0.15,0.01)	0.05 (−0.06,0.17)
	12 Months	−0.11 (−0.21,-0.02)	−0.09 (−0.17,0)	0.03 (−0.1,0.16)
	24 Months	0.05 (−0.05,0.14)	−0.11 (−0.19,-0.02)	−0.15 (−0.28,-0.02)
Endpoint	Time Points	FM Arme	UC Arme	ORf
Hypoglycemia Occurrenceb	Baseline 6 Months 12 Months24 Months	0.33 (0.15,0.59) 0.3 (0.12,0.57) 0.52 (0.23,0.79) 0.53 (0.25,0.8)	0.5 (0.28,0.72) 0.48 (0.26,0.71) 0.32 (0.13,0.59) 0.2 (0.07,0.46)	1.99 (0.49,8.09) 2.11 (0.48,9.4) 0.44 (0.08,2.38) 0.22 (0.04,1.29)
Abbreviations: FM = Functional Medicine plus Usual Care; UC = Usual Care; OR=Odds Ratio; CI = Confidence Interval aLinear mixed effect model was conducted to assess mean change at each time point, with time, treatment arms, the interaction of treatment arms and time, and baseline value of the outcome variable adjusted in the model. bMixed model with logit link was conducted to assess the probability of having the event and odds ratios, with time, treatment arms, the interaction of treatment arms and time included in the model cEstimated mean change (Lower 95%CI, Upper 95%CI) dMean change difference between patients in UC and FM arms (UC-FM) (Lower 95%CI, Upper 95%CI) eEstimated probability of having hypoglycemia (Lower 95%CI, Upper 95%CI) fEstimated odds ratio: The odds of having hypoglycemia in UC arm relative to FM arm (UC/FM) (Lower 95%CI, Upper 95%CI)

Change in diabetes medications and dietary supplementation

The change in diabetes medications and dietary supplementation is summarized in Table 6. At 6 months, the FM arm had a greater increase in the total medications compared to the UC arm (mean difference: −5.94), primarily driven by an increase in dietary supplements. This between-group difference persisted at 12 months (mean difference: −3.52) and became attenuated by 24 months (mean difference: −1.07).

Table 6.

Changes in total number of medications over time.

	Time Points	FM Arma	UC Arma	Differenceb
Total Number of Medications	6 Months	5.88 (4.04,7.72)	−0.06 (−1.91,1.78)	−5.94 (−8.55, −3.34)
	12 Months	4.34 (1.93,6.75)	0.82 (−1.25,2.89)	−3.52 (−6.69, −0.34)
	24 Months	1.47 (−0.94,3.88)	0.4 (−1.63,2.43)	−1.07 (−4.22,2.09)
Abbreviations: FM = Functional Medicine plus Usual Care; UC = Usual Care; OR=Odds Ratio; CI = Confidence Interval Linear mixed effect model was conducted to assess mean change at each time point, with time, treatment arms, the interaction of treatment arms and time, and baseline value of the outcome variable adjusted in the model. aEstimated mean change (Lower 95%CI, Upper 95%CI) bMean change between patients with UC and patients with FM (UC-FM) (Lower 95%CI, Upper 95%CI)

Quality of life

MSQ scores improved modestly in both arms over the course of the study; however, there was no statistically significant difference between groups at any time point. At 24 months, MSQ scores had decreased 11.4 points compared with UC 3.37 points. SPADE and PHQ9 scores for both arms were similar between baseline and 24 months (Table S2).

Discussion

This study evaluated whether incorporating a functional medicine (FM) approach into conventional diabetes care would lead to better glycemic control and improved quality of life in type 2 diabetes. The addition of the FM model didn’t lead to a statistically significant improvement in the primary outcome of insulin discontinuation with no increase in HbA1c or HbA1c < 7% at 12 months. In fact, the FM arm saw an increase in insulin dosage relative to UC by 24 months.

There have been no prior randomized controlled studies examining this question of how functional medicine can impact glycemic control and quality of life in those with type 2 diabetes. Very few prior studies have even been done on this subject at all. One of the only prior studies at an integrative medicine clinic evaluated data from 11 participants with T2D on at least 1 antidiabetic medication engaged in FM treatment over 2–10 months [26]. In the study, six patients discontinued all antidiabetic medications while on treatment and the other 5 reduced medication burden by 50%, collectively lowering HbA1c by an average of 2.71%. However, the population in this pre-post study was likely not representative of true FM outcomes as only around a third of the eligible patients with T2D allowed for their information to be included in the study, likely biasing the study towards those with more success. Furthermore, those in our study had more advanced diabetes requiring insulin utilization.

While more traditional lifestyle interventions have been shown to improve glycemic control in type 2 diabetes, maintaining intensive interventions over any amount of time is challenging and the addition of the FM approach, which included specific diagnostic testing, intervention and supplements, failed to demonstrated additional benefit [27]. Patients in the FM arm initially adhered to their supplements as indicated by the increase in number of medications at 6 and 12 months, but this did not result in less need for insulin. Patients in the FM arm initially lost more weight than those in the UC arm, but this weight loss was not sustained. Difficulty with adherence to complex regimens including lifestyle, dietary, and medication changes remains a crucial barrier to overall improvements in health and remains an area for future study [28], [29]. There were no other significant differences between the 2 arms in exploratory secondary aims.

At 24 months, patients in the FM arm received more insulin than in the UC arm, and that was associated with more hypoglycemic events and a reversal of the initial weight loss. This was the opposite of what we expected, based on the goals of the study. It is unclear why this occurred, but supports our finding the FM approach, as implemented here showed did not show additional benefit in getting patients off insulin. It may be related to a loss of initial adherence with FM interventions, as this change coincides with the reduction in medications back to close to baseline at 24 months in the FM arm.

Observational studies suggest that FM improves patient reported quality of life, even if it does not directly impact typical measures of health [16], [30]. We found no difference between the two arms for multiple patient-reported measures, including DDS17, SPADE, PHQ-9. The reasons for this are unclear, but it could relate to the rigor of a randomized trial versus observational studies. We also found no significant differences between groups in patient-reported outcomes, including DDS-17, SPADE, PHQ-9, and MSQ.

The FM approach as detailed in Appendix C and Table S1 contains elements of lifestyle counseling that have been previously demonstrated in trials to be successful [3], [4]. However, the supplements which were most utilized were intriguing. Most patients were provided with vitamins, omega and alpha lipoic acid supplements which have less data supporting their use for glycemic control [31]. Berberine, which does have moderate quality evidence for improvement of glycemic outcomes in type 2 diabetes, was used in less than 50% of the patients [32], [33]. This may have influenced the findings of this study but also reflects the real-world nature of the study.

Limitations

There are few limitations to the current study. Firstly, the sample size at baseline is small and there was significant loss of follow-up which limited the statistical power to detect differences in the outcomes. Given the pilot sample size and multiple correlated endpoints, our analyses are best interpreted as estimation of potential treatment-associated effects rather than definitive hypothesis testing. Non-significant results therefore should not be interpreted as evidence of no treatment difference. The original power calculation required 45 patients in each arm, and ultimately there were 34 in the FM arm and 37 in the UC arm due to challenges with recruitment and retention. However, given how few patients were able to discontinue insulin, even with full enrollment, we would have been unlikely to reach 30% discontinuation of insulin in the FM arm. Other studies have shown the difficulty in adherence to various lifestyle interventions and do not support the additional utility of FM specific diagnostics and interventions such as supplements [7], [27]. Secondly, the study largely coincided with the years of the COVID pandemic during which T2D outcomes generally worsened [29]. This may have altered the outcomes in both arms, although directionality of the effect is unclear. Further targeted studies should be done to better understand the specific diagnostics and interventions within FM that are most likely to be helpful for glycemic outcomes. In this study, the FM arm did not follow a specific protocol of care and patients received a wide range of testing and supplements. This highlights the difficulty of doing research on FM, which stresses the importance of individualized treatment. As such, the study can only be considered a test of the FM approach per se, without being able to invalidate any specific element of that approach. Additionally, achieving the primary outcome of the study could have been accomplished by change in diabetes medication rather than differences in care driven by the unique aspects of FM. However, we would expect transitions to more potent oral or injectable therapies to occur in both arms, reducing differential bias. Finally, drop out from both arms was high. This demonstrates the difficulty of getting patients to adhere to complex FM regimens, even in the context of a randomized trial with highly motivated patients. It is possible that if patients were to adhere better to these recommendations that their outcomes would have been better. However, an approach that cannot be widely adhered to is of limited utility.

Conclusion

In summary, our study demonstrated that addition of an FM approach to usual care did not result in a statistically significant improvement in discontinuation of insulin with no increase in HbA1c or HbA1c < 7% (53 mmol/mol) at any point in a 24-month follow-up.

CRediT authorship contribution statement

Mohamed Eldib: Writing – review & editing, Writing – original draft, Investigation, Formal analysis. Betul Hatipoglu: Writing – review & editing, Supervision, Project administration, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Michelle Beidelschies: Writing – review & editing, Supervision, Project administration, Investigation, Data curation. Marilyn Alejandro-Rodriguez: Writing – review & editing, Supervision, Project administration, Investigation, Data curation. Huijun Xiao: Writing – review & editing, Writing – original draft, Validation, Formal analysis. Michael Rothberg: Writing – review & editing, Methodology, Funding acquisition, Formal analysis, Conceptualization. Mark Hyman: Writing – review & editing, Investigation, Funding acquisition, Conceptualization. Keren Zhou: Writing – review & editing, Writing – original draft, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Funding

This work was supported by the Cleveland Clinic Foundation.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Supplementary Appendix and Tables. Supplementary data

The following are the Supplementary data to this article: