Simplifying carb counting: A randomized controlled study – Feasibility and efficacy of an individualized, simple, patient‐centred carb counting tool

Abstract Introduction The purpose of this study was to introduce and test a simple, individualized carbohydrate counting tool designed for persons with Type 1 Diabetes Mellitus (T1DM) in order to determine whether the tool improved A1C levels for participants with age, education or language barriers. Methods In a randomized controlled trial, 85 participants were offered six diabetes instructional sessions free of charge over a six‐month period. Forty‐one received guidance using the regular carbohydrate counting (RCC) method. Forty‐four received guidance using an individualized ‘Simple Carb Counting’ (SCC), involving two customized tables prepared for participants. Results The simple, individualized SCC tool for carbohydrate counting was non‐inferior to the standard method of RCC. The SCC tool was more effective among participants aged 40 and older, while no differences were found when comparing participants by education level. Irrespective of intervention group, all participants improved their A1C level (9.9% = 13.2 mmol/L vs 8.6% = 11.1 mmol/L, p = .001). A greater improvement in A1C level was seen in newly diagnosed participants (−6.1 vs −0.7, p = .005, −3.4 vs 0.9, p = .032) in both the RCC and SCC groups. All participants expressed improved emotional level per their PAID5 questionnaires (Problem Areas in Diabetes Scale‐PAID), (10.6 (±5.7) vs 9.5 (±5.7), p = .023), with women reporting greater improvement than men. Conclusions SCC is a simple, individualized, feasible, low‐tech tool for carbohydrate counting, which promotes and enables accurate insulin dosing in people with T1DM. It was found more effective among participants aged 40 and older. Additional studies are needed to corroborate these findings.

In recent years, new technologies have been developed to simplify CC. 7  The aim of this randomized controlled study was to test the feasibility and efficacy of SCC, compared to the regular CC method.
Our Hypothesis is that this simple tool will enable all people with T1DM improve diabetes control.

| ME THODS
We performed an open-label randomized controlled trial at the Diabetes Clinic of SUMC, a tertiary 1200-bed hospital that treats a diverse population, including Bedouin and other Arabs, and Jews from the general, ultra-Orthodox and Ethiopian sectors. Patients with T1DM were eligible for this study if they were (i) over 18 years old, (ii) treated either with an insulin pump or with multiple daily injections of insulin and (iii) had a hemoglobinA1c (A1C) level equal to, or >8.5. The study excluded pregnant or lactating women, patients with severe renal failure, heart failure or under active treatment for cancer. All participants signed an informed consent statement. The study was approved by the SUMC Helsinki Committee on 8 October 2015, approval number 0320-15, and is registered at Clini calTr ials. gov, ID NCT04132128, and conducted from November 2015 to July 2017.
Using simple randomization, we assigned participants to one of the two groups, (RCC) Regular carbohydrate counting, or (SCC) simplified individualized tool for carbohydrate counting. The selection process was purely random for every assignment made by Diabetes Clinic staff.
All participants were allowed up to six instructional sessions with a registered dietitian who is also a diabetes care and education specialist during a period of about 6 months. All sessions were free of charge and lasted at least 60 min. During the first session, participants were introduced to the carbohydrate counting method to which they were assigned at randomization. Subsequent sessions were dedicated to reinforcing and practicing their method and changes in insulin dosage parameters were made if needed. All patients treated with insulin pumps were encouraged to use the bolus calculator. The primary end-point was level of A1C 6 months after the intervention. Additional data parameters were collected including those of demographics, education and duration of diabetes.
Weight was measured and blood studies were conducted before and after the intervention to determine baseline A1C and lipid levels. In order to identify depression and diabetes-related distress, the participants were asked to complete the PAID5 questionnaire (Problem Areas in Diabetes Scale-PAID) 8 at baseline and post-intervention.

| Regular carbohydrate counting (CC)
During the instructional sessions, the rationale for carbohydrate counting was explained. Commercial booklets containing a list of the carbohydrate content of foods were provided, and participants were introduced to websites or cellular phone applications designed to assist the public with determining the carbohydrate content of various foods. The participants were taught to calculate the amount of insulin needed using their personal I:C ratio, IS, correction factor, and the glucose target goals prescribed by the Diabetes Clinic team.
Participants were encouraged to keep a food diary to assist them with carbohydrate counting.

| Simplified individualized carbohydrate counting (SCC)
The SCC tool consisted of two tables written in the participant's native language and adjusted to the participant's specific requirements. Insulin doses were calculated by professional staff using personalized I:C ratios and IS.
First Table: The first table, derived from patients' personal IS, listed the number of units that participants needed to administer in order to correct every pre-meal blood glucose level so as to reach their target glucose.
Second Table: The second table contained a list of food items derived from participants' personal eating habits, as recorded in their food diaries. The list consisted mainly of the most common foods they regularly consumed, the carbohydrate content of those foods and the number of insulin units needed, as calculated by the personal I:C ratio per usual portion of each food item. High carbohydrate content foods that participants included in their diet were listed, not for healthy nutrition education, but for purposes of facilitating carb counting. Foods that did not contain carbohydrates, such as protein or fat items, were also listed to ensure that the patient realized that these foods contained no carbs. Patients treated with insulin pumps received a personalized table containing the carbohydrates in their food list as grams to accurately calculate the amount of carbs entered to the calculator (Figures 1 and 2).
At each instructional session, participants' tables were reviewed, personal dosing was tested, food items were added to, or deleted from participants' lists as warranted, and the logic of the method was reiterated for the purpose of reinforcing participants' understanding and compliance.

| Statistical analysis
On the basis of the expected A1C difference of 0.8% between groups with 90% power providing an SD of 0.4% and a significance level of 0.05% we estimated that the sample size should consist at least 37 participants in each group. We used descriptive statistics to summarize the data, reporting results as means and standard deviations. Categorical variables were summarized as counts and percentages. Paired t-test was used to examine within changes in A1C between baseline and follow-up, and Student's t-test was used to examine between differences as to the two intervention groups. We further analysed the data after stratifying the study population by sex, education level (above and below 12 years of school), age (above and below age 40) and duration of diabetes (more or <5 years since diagnosis). p values are shown.
All analyses were performed with IBM SPSS.

| RE SULTS
In total, 107 men and women were recruited for the study, of whom 22 were excluded, as shown in Figure 3. Of the 85 people who were deemed eligible, 48.2%, n = 41, of participants (23 women, 18 men) were assigned to RCC method group, and 51.8%, n = 44, of participants, (17 women, 27 men) to the SCC method group. The mean age of participants was 43.1 . About 43% of the RCC method group were treated with an insulin pump versus 36% of the SCC group (p = .48). All patients monitored their glucose by selfmonitoring of blood glucose (SMBG) (Figure 1 and Table 1).
The participants were followed for a mean period of 6 months, during which all participants in the study improved their A1C level between baseline and follow-up (9.9% = 13.2 mmol/L vs 8.6% = 11.1 mmol/L, p = .001). Other biomarkers did not change from baseline in either of the groups.
We stratified the study population by participant age (older and younger than age 40). Among older participants (mean age 55. We stratified the study population by level of education (above and below 12 school years). Both higher and lower educated participants in the SCC group demonstrated a significant improvement in A1C level. The results for those with above and below 12 school years were −1.4% (±2.0) vs −1.3% (±1.9) respectively, p = .7.
We found a greater improvement in A1C levels when we compared participants with more recently diagnosed diabetes (<5 years from diagnosis, n = 13) to those whose diabetes was of longer duration (n = 72). (−6.1 vs −0.7, span >−3.4 vs 0.9, p = .032) in both the RCC and SCC groups.
We measured the degree of patient compliance by the number of instructional sessions attended during the study period. Compliance was fairly good for all participants, with a mean of 4.8 visits (out of the six allowed). No differences were found in compliance between participants in the RCC group and those in the SCC group, but women were more compliant than men, with a mean of 5. 3 (1.3) visits, compared to 4.4 (1.9) for the men (p = .01). Compliance tended to be higher among participants with more than 12 years of education, compared to those with fewer years of education (4.  sex, they found a significant improvement among women, who reported a decrease in diabetes-related emotional distress from 11.5 to 9.9, p = .04, compared to men, 9.9 to 9.18, p = .27 (Table 2).
During the 6 month follow-up, two patients from the RCC group were hospitalized with diabetic ketoacidosis (DKA). There were no hospitalizations or ER visits due to hypoglycaemia in both groups.

| DISCUSS ION
The findings showed that the SCC simple, individualized tool for carbohydrate counting was non-inferior to the standard method of RCC.
The SCC tool was more effective among participants aged 40 and older, while no differences were found when comparing participants above and below 12 school years. However, significant improvement in A1C level was observed in all participants. Participants in both RCC and SCC groups who were diagnosed with diabetes within the previous 5 years exhibited significantly greater improvement in A1C level, compared to participants with diabetes of longer duration.
The SCC method presented in our study was developed in order to overcome difficulties and barriers that the diabetes clinic patients encountered in implementing CC, as described in several studies.  In this study, we present an option to overcome some of the barriers described in the above studies. Through personalization, flexibility, and a departure from a restrictive diet paradigm, SCC affords persons with diabetes an opportunity to continue eating their usual diet, including the customary dishes of their culture, and to go on with their life-long social dining habits.
The study showed superiority in reaching glycaemic control in participants older than age 40 who used the SCC method, compared to those in the RCC group. Treating older patients with T1DM is complicated by the combined challenges of insulindependent diabetes, age-related complications, and possible comorbidities, all of which negatively affect the older population's ability to self-manage diabetes. 11 The SCC tool presented in the study simplifies the tasks needed for carbohydrate control, and consequently leads to better glycaemic control especially for older age group.
Contrary to these findings, the study demonstrates that SCC was non-inferior in people with various levels of education. In a cross- Recently diagnosed participants in both RCC and SCC groups showed the greatest benefit in improving glycaemic control. This may be explained in two possible ways, one psychological and one physiological. When first diagnosed, many patients are highly motivated to do well. In addition, in the early period after diagnosis, sometimes called the 'honeymoon period,' the pancreas seems to do better and secretes more insulin, although this phenomenon decreases with time and differs with each patient.

F I G U R E 4 HbA1c results SCC vs RCC.
Baseline PAID Score (±SD)
All participants exhibited significant decreases in their PAID5 scores. Studies have shown 13,14 that people with diabetes suffer from higher levels of psychological distress than does the healthy population. People with T1DM are three times more likely to develop depression than those without diabetes. 15 Moreover, psychological distress has been shown to be associated with hyperglycaemia, complications and higher mortality rate. 16,17 Thus, there is a consensus that treating psychological stress and achieving psychological wellbeing ought to be one of the treatment goals of diabetes care. 18 A study by Zagarins et al 19 revealed a correlation between improvement of glycaemic control and alleviation of overall psychological stress, but not in depression. Our study corroborates these findings, and underscores the need for on-going diabetes education, better understanding and treatment of diabetes and promoting a greater sense of self-efficacy among patients in controlling the disease, as means of improving not only metabolic control, but also mental health.

| Limitations
The intervention tool was introduced at a single diabetes clinic in a tertiary teaching hospital with one registered dietitian/diabetes care and education specialist. The method was not tested on paediatric patients, a population that has more difficulties with glycaemic control than others. A larger population of people from the lower socio-economic and more culturally diverse backgrounds should be studied in order to corroborate the results and establish generalizability across populations. Harman Boehm: Conceptualization (equal); data curation (equal); investigation (equal); methodology (equal); resources (equal).

ACK N O WLE D G E M ENTS
The Authors declare there are no conflicts of interest and no financial support for this study. We thank all the diabetes unit staff for all technical assistance.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The Authors declare that there are no conflicts of interest.