Validation Data Based on a Web-Based Photographic Questionnaire for Assessment of Skills in Estimating Carbohydrate Portion Sizes in Adults with Type 1 Diabetes
Keywords: Type 1 diabetes mellitus; Basic carbohydrate counting; Carbohydrate estimation accuracy; Food photographs; Patient
Department of Nutrition, Exercise and Sports, Faculty of Science, University of Copenhagen, Frederiksberg, Denmark Bettina Ewers, Steno Diabetes Center Copenhagen, Gentofte, Denmark, Schouw, N., Skouboe, AG., Bruun, JM., Ewers, B. (2019) Validation Data Based On A Web-Based Photographic Questionnaire For Assessment Of Skills In Estimating Carbohydrate Portion Sizes In Adults With Type 1 Diabetes. J . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The data in this article present differences and agreements in estimating the carbohydrate content in foods based on web-based photographs vs. real foods in adults with type 1 diabetes (T1D). The Carbohydrate Photographic Questionnaire (CPQ) consists of photograph series of 20 commonly eaten high-carbohydrate foods in the Nordic countries. The CPQ was developed as a simple tool for assessing skills in accurate estimation of the carbohydrate content of foods. Ninety-six adults (52% females) on insulin pump therapy attended the study. Error estimates for the true carbohydrate content were calculated based on the participants’ estimations from the CPQ and from real foods. Cross-classifications indicated an overall moderate agreement in estimation errors between the CPQ and real foods. The median percentage of participants classified into the same error category when estimating the carbohydrate content in real foods vs. the CPQ was 39% (interquartile range (IQR) 35–50%). The highest correct classification was obtained for grapes (79%) while lasagne (15%) had the lowest Department of Medicine, Randers Regional Hospital, Denmark Steno Diabetes Center Aarhus, Aarhus, Denmark Department of Clinical Medicine, University of Aarhus, Aarhus, Denmark
E-mail: bettina.ewers@regionh.dk Received: 12 August 2019; Accepted: 19 September 2019; Published:
Clin Nutr Food Sci, 2(2): 054-056.
Copyright: © 2019 Schouw, N, et al correct classification.
education.
• The data can provide new insights on how well adults with type 1 diabetes estimate carbohydrate content in real foods versus two-dimensional learning materials i.e. photos on paper, web-based computer graphics, or photos on smart phone applications • The data can be used to identify food groups or food items which are particularly difficult to estimate the carbohydrate content of using two-dimensional learning materials • The data provide important new insight in relation to nutrition 20 September 2019 education supporting the use of real foods when people with diabetes practice their skills in estimating the carbohydrate content in grams of foods correctly in a controlled learning environment • The data can support health care professionals within the field of nutrition education by providing access to use or be inspired by a simple tool to assess carbohydrate estimation accuracy in people with diabetes • The data can be used by clinicians and academia for further research and as reference Questionnaire For Assessment Of Skills In Estimating Carbohydrate Portion Sizes In Adults With Type 1 Diabetes. J Clin Nutr Food Validation of a web-based photographic questionnaire for assessment of skills in estimating carbohydrate From a web-based questionnaire compared with the study dietitian’s recordings of the estimated values of The CPQ was developed as a tool to evaluate the effect of learning how to estimate dietary carbohydrate content so we included patients with T1D who had previously received education in this. The participants were not allowed to use aids i.e. smartphone applications with information on the carbohydrate content during the examination. The examination of the participants’ skills was done with only one participant at a Participants (n=96) were presented to 20 different foods sequentially and asked to estimate the carbohydrate content in grams. Subsequently the participants were asked to choose one photograph from a set of six photographs, depicting different amounts of the same food (from the web-based CPQ), which most closely resembled the carbohydrate content in the presented real food. Inclusion criteria were adults (18–75 years) with T1D and on insulin pump therapy, having attended at least one consultation with a dietician within the last two years. Exclusion criteria were carbohydrate counting naive, visually impaired, or people diagnosed with T1D less than 12 months before beginning of the study Skills lacking in estimating carbohydrate content: A need for continual education of adults with type 1 Sci, 2(2): 54-056.
the carbohydrate content in the real foods presented
time in the examination room to reduce bias.
Data source location Institution: Steno Diabetes Center Copenhagen
Data on error estimates of carbohydrate portion sizes when adults with T1D assess web-based photographs as compared with real foods, the latter considered a ”gold standard” for validation of photographs, were collected in a quality assurance project. Patient characteristics have been presented in the related research article [1].
Participants were presented to a serving of real food and asked to estimate the carbohydrate content in grams. Subsequently, the participants were asked to choose the photograph from a set of six photographs depicting different amounts of the same food which most closely resembled the carbohydrate content in the real food. This procedure was repeated for 20 commonly eaten high-carbohydrate foods in Denmark. The 20 photograph series were presented in the newly developed web-based Carbohydrate Photographic Questionnaire (CPQ). Figure 1 shows a diagram of the experimental design.
The web-based photograph series were photographed according to practical guidelines from the National Food Institute at the Technical University of Denmark [2]. Error estimates for carbohydrate portion sizes were calculated for both real food and the CPQ based on (Equation 1):
Error estimate (g) = Estimated carbohydrate content (g) – Actual amount of carbohydrate (g) (1) A negative difference was chosen to indicate underestimation of the presented food. To compare the estimation errors in real food and the CPQ, the percentages of participants within each error category Nutrition portion sizes in adults with type 1 diabetes Table Analysed City: Gentofte Country: Denmark With the article Schouw N, Skouboe AG, Bruun JM, Ewers B diabetes were determined. The error categories were defined as estimation errors within ±10g of the actual carbohydrate content (accurate), above 10g of the actual carbohydrate content (overestimation), and below -10g of the actual carbohydrate content (underestimation). The agreement in estimation errors between real food and the CPQ was determined by cross-classification of error estimates which was used to determine the cumulative percentage of participants who were classified into the same error category in real food and the CPQ (Equation 2):
Cumualative percentage = n classified into the same estimation category/Total n x 100% (2) The cumulative percentage of participants, who were classified into the same error category, was calculated for each food. For all statistical tests a two-sided significance level of p <0.05 was used. All statistical analyses were performed with the SPSS software for Windows, version 22.0 (IBM Corp, Armonk, NY, USA).
The study was carried out in accordance with the Declaration of Helsinki. Informed consent to participate in the study was obtained from all participants prior to study inclusion according to Danish regulations. The study was approved by the local ethics committee and the Danish Data Protection Agency and included an approval of an External Data Manager Agreement with the Capital Region of Denmark for data storing.
The article presents data on the agreement in estimation errors when adults with T1D are asked to estimate the carbohydrate Questionnaire For Assessment Of Skills In Estimating Carbohydrate Portion Sizes In Adults With Type 1 Diabetes. J Clin Nutr Food cipant was asked to chose one photograph out of six photographs of the same food which closely resembled the carbohydrate content in the real food serving. This was repeated for 20 erent photographic series.
Sci, 2(2): 54-056.
The parti The participant was presented to a serving of real food and asked to estimate the carbohydrate content. This was repeated for 20
Table 1: The cumulative percentage of participants classified in the same error category for real food and the CPQ.
Cumulative percentage (%) 1. Grapes 79 2. Milk chocolate 59 3. Raisins 54 4. Potato chips 52 5. Boiled pasta 52 6. Dried apricots 49 7. Pizza 49 8. Fruit pie 45 9. Melon 40 10. Mixed candy 39 11. Dark chocolate 38 12. Bun 38 13. Oatmeal 37 14. Pasta salad 36 15. Boiled rice 36 16. Boiled potatoes 33 17. Ice cream 31 18. Kidney beans 25 19. Potato slices 24 20. Lasagne 15 Median (IQR) 39 (35–50) We calculated the estimation errors in grams for real food vs. the chosen photograph. Cross-classifications between estimation errors in real food vs. photographs were used to examine the agreement between the two methods (real food vs. CPQ) in assessing carbohydrate portions.
content in different portion sizes of real foods versus portions sizes of the same foods presented as photographs based on a web-based questionnaire. Table 1 shows the percentage of participants classified in the same error category for real foods and the CPQ. The percentage was determined from cross-classifications between the percentages of participants who were classified into the same error category.
The authors would like to thank the patients from our outpatient diabetes clinic who participated in the survey and provided data for this article. We also thank our clinic at Steno Diabetes Center Copenhagen for funding.