1^**+,,- Nutrition Data Quality page  Nutrition Data Quality- What To Look For by David Dunetz SweetWARE 1994 by SweetWARE Introduction The Nutrition Labeling and Education Act (NLEA), which became effective May 8, 1994, has created a large demand for accurate nutrition data for the purpose of providing Nutrition Facts labels on food products. In this article I will explain how nutrition data is organized, and I will show you how to evaluate the quality of data that you receive from ingredient manufacturers or software vendors. Ingredients vs. Recipes When talking about nutrition data, it is important to distinguish between ingredients and recipes. A recipe is any item that is made from a combination of ingredients or other recipes. An ingredient is any item that is not a recipe. (In other words, an ingredient is an item that has no ingredients.) The only direct source of nutritional information for an ingredient is a chemical analysis performed by a certified laboratory. Nutrition information for a recipe may be obtained by chemical analysis, or by computer analysis. Since laboratory chemical analysis is so expensive ($300 to $600 per item tested), it is often desirable to use a computer generated nutritional analysis instead. This type of analysis calculates the nutritional content of a recipe or formula based on the nutritional content of the ingredients. There are several commercial software packages on the market capable of generating such an analysis. Obviously, the validity of the analysis depends on the quality of the ingredient data used. Nutrition Data Quality - What does it mean? When I refer to data quality, I am referring to completeness, consistency and detail. By completeness I mean that information is provided for all of the nutrients we need to track for our application. When data is provided for some nutrients and not others, our calculated analysis will clearly be incomplete. By consistency I mean that when summary and detail information is provided, the values of the detail information must agree with the summary values. By detail, I mean the number of nutrients that are reported. I will describe the nutrient sets and amount of detail required for different applications later in this article. I will also show you how to precisely measure completeness and consistency, and I will describe a rating system that we use to measure the overall quality of the data we receive. Note that I am talking about nutrition data quality, not accuracy. Quality means that the nutrition data makes sense, that it doesnt contradict itself. I leave the accuracy issues to the professional chemists, and trust that they are providing us with reliable data from established analytical procedures. Data Data Everywhere - But none of its the same We are engaged in an ongoing effort to collect and maintain ingredient nutrition information from a wide variety of sources. As part of this process we have examined a large assortment of nutrition data. We have found that there is no standardization of the format for nutrition data reporting, and that the quality and consistency of much of the data is less than adequate. In addition, statistical information that is essential to an accurate computer analysis of the data is rarely provided. (One shining exception to this is USDA Handbook 8, which includes the statistical information required.) In this article I suggest a format for providing nutrition data (including several forms which you may duplicate), and I describe guidelines for evaluating the quality of the data provided. First I will describe some common problems we have encountered, and then I will offer a list of suggestions for improved nutrition information reporting. Finally, I will describe a set of tests that can be used to evaluate the consistency of the data, culminating with a simple rating system for the overall quality of the nutrition data. Common problems Here are some of the common problems with the nutrition data we have collected: No standardization of the set of nutrients. No indication of the source of the nutrition information. Data that cannot be traced to its source has little validity. Incomplete nutrition information - the nutrients in a 100g sample do not add up to 100g. Other data consistency problems as described below. No information about the range of values to be expected. This is usually expressed as the standard deviation of the samples tested. Information about the number of samples tested and the standard deviation is essential to a computer analysis of the nutrient content of a recipe. Suggestions for Ingredient Manufacturers. Nutrition information for recipes is usually reported for a standard serving size. However, ingredient nutrition information should be reported for a 100g sample. Please do not provide values for a specific serving size, or values expressed as a percentage of daily requirements. Do not provide values expressed as a percentage of daily requirements. The daily standards have been revised several times over the years, and such percentages have no meaning by themselves without reference to a specific dated set of standards and a specific age and sex group. Attempt to completely categorize the nutrient content of your product. This means that the nutrients for a 100g sample should total 100g. Otherwise the nutrient information is incomplete. (Consequently, the nutrition analysis of any recipe containing that ingredient will be incomplete.) We have included several Other nutrient categories to account for detail information that may be missing or incomplete. Nutrition Facts panels are not adequate sources of ingredient nutrition information for the following reasons: The values listed are rounded off to make them easier to read for the consumer. Information is provided for only some of the nutrients in your products. The NLEA Nutrition Facts panel lists Sugars and Fiber as components of Carbohydrates. Until recently, many nutrition analyses did not include data for sugar content. Needless to say, this information is crucial for NLEA labels. Many nutrition analyses list nutrient content as less than a certain value. Data presented in this way is useless for a nutrition analysis. The proper way to report nutrient values is to list the mean, standard deviation and number of samples tested. This gives a range of expected values that can be used to accurately predict the nutrient content of a recipe. Report Vitamin A, Retinol and Carotene values as Retinol Equivalents (RE), rather than the previously standard International Units (IU). When you have a chemical analysis done, make sure the data is consistent as described in this article, and make sure that the standard deviation and number of samples tested for each nutrient are reported. Use one of the Nutrition Data forms included with this article as a guideline for the nutrients to report. List of Nutrients - Summary and Detail Values We are collecting data for 100 nutrients. The nutrients are listed in figure 1, which is a sample of a data collection form you may duplicate. Suggestions for additional nutrients are appreciated. The nutrition data is organized as summary and detail information. For example, proteins are listed as Protein (summary) and Amino Acids (detail). If all of the amino acid constituents are listed, the total weight of the amino acids should equal the total protein summary value. If the two values are not equal, it indicates that the detail information is incomplete, or that the summary value is inaccurate. Carbohydrates and fats have two levels of detail. For example, Total Fat (summary) is listed as Saturated Fat, Monounsaturated Fat and Polyunsaturated Fat (detail). These nutrients are further broken down into constituent fatty acids (further detail). Again, the total weight of the detail values should equal the weight of the summary value. If the two values are not equal, something is missing or wrong. Total Carbohydrate (summary) consists of Total Sugar, Total Fiber and Other Carbohydrates (detail). Total Sugar is further broken down into mono- and polysaccharides, and sugar alcohols (further detail). Total Fiber is further broken down into Soluble Fiber and Insoluble Fiber (further detail). Vitamins and Minerals do not have detail information, except for Vitamin A (summary), which consists of Retinol and/or Carotene (detail). Additionally, there is a relationship between calories (summary) and protein, fat and carbohydrate (detail). Protein provides 4 calories per gram, fat has 9 calories per gram, and carbohydrates have 4 calories per gram. To be consistent, the number of calories listed should agree with the calories calculated from protein, fat and carbohydrate. Consistency - How close should it be? How well should the detail information match the summary value? Ideally, the two values will match exactly, but we do not live in an ideal world. It is standard practice in statistical analysis to use accuracy levels of 1% or 5%. In our consistency tests we categorize data consistency as being accurate to 1% or better (excellent matching), or 5% or better (good matching). Any values that disagree by more than 5% are considered out of tolerance. Note that we are not talking about the accuracy of the nutrition information itself, but only of the consistency of the detail information with respect to the summary information. The Nutrition Data Consistency Report The Consistency Report tabulates the quality of the nutrition information for one or more ingredients (items). It can be used to rate a single ingredient, a list of ingredients, a recipe or an ingredient library. It is divided into 5 sections. A sample Consistency Report is shown in figure 2: Proximate Consistency Analysis This section reports the number of items that are consistent within 1%, 5% or out of tolerance for total nutrient weight, calorie consistency, fat consistency and carbohydrate consistency. Detail Summary This section lists the number of items for which detail information is provided for fat, protein (amino acid), sugar, and fiber. It also indicates the number of items for which total sugar is reported. (Some ingredients naturally do not have sugar; they are included as total sugar not reported.) Detail Consistency When further detail information is provided, this section indicates the number of items that are consistent within 1%, 5% or out of tolerance. Data Source We list the number of items from each data source. Ideally, all of the information would come from a lab analysis. Data Quality Data quality is rated for each item. Our quality rating depends on the results of the Proximate Consistency Analysis. The rating is as follows: Excellent All 4 Proximate Consistencies within 1% Good All 4 Proximate Consistencies within 5% Fair 2 or more Proximate Consistencies within 5% Poor Fewer than 2 Proximate Consistencies within 5% Nutrients for Different Applications How much detail should be reported? The answer depends on what the nutrition information is to be used for. There are three levels of detail that are commonly used: 1. NLEA required nutrients. 2. NLEA required and optional nutrients. 3. Complete nutrition data, including all detail. The following tables list the required and optional nutrients for NLEA Nutrition Facts labels. NLEA Required Nutrients Total Calories Calories from Fat Total Fat Saturated Fat Cholesterol Sodium Total Carbohydrate Dietary Fiber Sugars Protein Vitamin A Vitamin C Calcium Iron NLEA Optional Nutrients Calories from Saturated Fat Polyunsaturated Fat Monounsaturated Fat Potassium Soluble Fiber Insoluble Fiber Sugar Alcohol Other Carbohydrate Vitamin D Vitamin E Thiamin Riboflavin Niacin Folate Vitamin B12 Biotin Pantothenic Acid Phosphorus Iodine Magnesium Zinc Copper We recommend that ingredient manufacturers provide information about at least the required and optional NLEA nutrients, and if possible provide complete nutrient information. In addition, water and ash often comprise a significant portion of the 100g sample, and should be included in all nutrition reports. We also include Alcohol, Caffeine and Theobromine as miscellaneous items that don't fit in any of the other categories. ___________________________________________________________________________ Figure 1 - Nutrition Request Form Nutrition Data Manufacturer ________________________________________________________________ Product Description ___________________________________________________________ Part Number ______________ Data Source ___________________ Date ________________ Nutrition Values per 100 grams Standard # samples Value Deviation tested Calories ___________________ kCal _______ ________ Protein ____________________ g _______ ________ Total Carbohydrates __________ g _______ ________ Total Sugar _______________ g _______ ________ Total Fiber ________________ g _______ ________ Soluble Fiber_____________ g _______ ________ Insoluble Fiber ___________ g _______ ________ Other Carbohydrates _________ g _______ ________ Total Fat ___________________ g _______ ________ Saturated Fat ______________ g _______ ________ Monounsaturated Fat ________ g _______ ________ Polyunsaturated Fat _________ g _______ ________ Cholesterol ________________ mg _______ ________ Other Fats _________________ mg _______ ________ Water _____________________ g _______ ________ Ash _______________________ g _______ ________ Alcohol ____________________ g _______ ________ Caffeine____________________ mg _______ ________ Theobromine ________________ mg _______ ________ Vitamins Minerals Standard # samples Standard # samples Value Deviation tested Value Deviation tested Total Vit. A ________ RE _______ _______ Calcium _______ mg _______ _______ Carotene _________ RE _______ _______ Chlorine_______ mg _______ _______ Retinol __________ RE _______ _______ Copper ________ mg _______ _______ Thiamin (B1) _______ mg _______ _______ Iodine _________ mg _______ _______ Riboflavin (B2)______ mg _______ _______ Iron __________ mg _______ _______ Niacin (B3) ________ mg _______ _______ Magnesium_____ mg _______ _______ Vitamin B6_________ mg _______ _______ Manganese_____ mg _______ _______ Vitamin B12________ mcg _______ _______ Phosphorus_____ mg _______ _______ Biotin_____________ mcg _______ _______ Potassium______ mg _______ _______ Choline____________ mcg _______ _______ Selenium_______ mcg ______ _______ Folacin ____________ mcg _______ _______ Silicon_________ mg _______ _______ Inositol ____________ mg _______ _______ Sodium_________ mg _______ _______ Pantothenic Acid_____ mg _______ _______ Sulfur _________ mg _______ _______ Vitamin C __________ mg _______ _______ Zinc___________ mg _______ _______ Vitamin D __________ mcg _______ _______ Other Minerals__ mg _______ _______ Other Vitamins ______mg _______ _______ Nutrition Data - Page 2 Further Detail Manufacturer ________________________________________________________________ Product Description ___________________________________________________________ Part Number _________________________________________________________________ Sugar Details Nutrition Values per 100 grams Value Standard # samples Deviation tested Fructose ___________ g _______ _______ Galactose __________ g _______ _______ Glucose ___________ g _______ _______ Lactose __________ g _______ _______ Starch Maltose____________ g _______ _______ Standard # samples Sucrose____________ g _______ _______ Deviation tested Other Sugar ________ g _______ _______ Starch and other Value Sugar Alcohols ______ g _______ _______ Carbohydrates _____ g ________ _______ Amino Acids Standard #samples Standard # samples Value Deviation tested Value Deviation tested Alanine ___________ g _______ ______ Methionine _________ g _______ _______ Arginine __________ g _______ ______ Phenylalanine_______ g _______ _______ Aspartic Acid_______ g _______ ______ Proline____________ g _______ _______ Cystine____________ g _______ ______ Serine____________ g _______ _______ Glutamic Acid_______ g _______ ______ Threonine__________ g _______ _______ Glycine____________ g _______ ______ Tryptophan_________ g _______ _______ Histidine___________ g _______ ______ Tyrosine___________ g _______ _______ Isoleucine__________ g _______ ______ Valine_____________ g _______ _______ Lysine_____________ g _______ ______ Other Amino Acids___ g _______ _______ Leucine____________ g _______ ______ Fatty Acid Details Standard #samples Standard # samples Value Deviation tested Value Deviation tested Fatty Acid 4:0 ________ g _______ ______ Fatty Acid 14:1 _____ g _______ ______ Fatty Acid 6:0 ________ g _______ ______ Fatty Acid 16:1 _____ g _______ ______ Fatty Acid 8:0 ________ g _______ ______ Fatty Acid 18:1 _____ g _______ ______ Fatty Acid 10:0 _______ g _______ ______ Fatty Acid 20:1 _____ g _______ ______ Fatty Acid 12:0 _______ g _______ ______ Fatty Acid 22:1 _____ g _______ ______ Fatty Acid 14:0 _______ g _______ ______ Fatty Acid 18:2 _____ g _______ ______ Fatty Acid 15:0 _______ g _______ ______ Fatty Acid 18:3 _____ g _______ ______ Fatty Acid 16:0 _______ g _______ ______ Fatty Acid 18:4 _____ g _______ ______ Fatty Acid 17:0 _______ g _______ ______ Fatty Acid 20:4 _____ g _______ ______ Fatty Acid 18:0 _______ g _______ ______ Fatty Acid 20:5 _____ g _______ ______ Fatty Acid 20:0 _______ g _______ ______ Fatty Acid 22:5 _____ g _______ ______ Fatty Acid 22:0 _______ g _______ ______ Fatty Acid 22:6 _____ g _______ ______ Fatty Acid 24:0 _______ g _______ ______ ______________________________________________________________________________ Figure 2 - Sample Nutrition Data Quality Report Nutrition Data Summary USDA Handbook 8 Number of Items Included: 3851 10:55am 10/27/94 Page 1 Proximate Consistency Analysis Number of Items: within 1% within 5% out of tolerance Total Nutrient Weight = 100g 3230 580 41 Calorie Consistency 242 2220 1389 Fat Consistency 71 496 3284 Carbohydrate Consistency 3230 580 41 Detail Summary # Items Fatty Acid Detail 3392 Amino Acid Detail 3253 Total Sugar Data 316 Sugar Detail 299 Fiber Detail 0 Detail Consistency Number of Items: within 1% within 5% out of tolerance Saturated Fatty Acid Consistency 1410 1212 770 Monounsat. Fatty Acid Consistency 2227 908 257 Polyunsat. Fatty Acid Consistency 2290 840 262 Amino Acid Consistency 380 1523 1350 Sugar Detail Consistency 181 21 97 Fiber Detail Consistency 0 0 0 Source of Nutrition Data Data Quality # Items # Items Lab Analysis 0 Excellent 5 USDA Handbook 8 3851 Good 76 Calculated Analysis 0 Fair 2729 NLEA label 0 Poor 1041 Unknown 0 _____ g _______ ______ Fatty Acid 17:0 _______ g _______ ______ Fatty Acid 20:4 _____ g _______ ______ Fatty Acid 1wrnjfb^ZVRNJZ` 0@ wso kg(c4_6[AWFSLOTKZ`  T`w7sBokga ch _ [ Wm S OKZ`  )w~sokg2c9_[WSOKZ` wsok g'cA_H[lWxSOKZ`   @wGs_ofkgc_x[WSO"KZ` ")wNsUoXk^gpc}_[WSOKZ` wso^kgzc_[W#S#O%KZ`   %%w7%s;%oO&kv&g'c'`(\(Y)U)Rj*NZ`  j*x*x*t*q -m3-i.e.a.].Y1UO1Q6MZ`  66w6s6o7k]7g_7c7_7[7W7S7P8LZ`  $ 88w9s#9o$9kJ9gW9c9_9[9W9S9O9KZ` 9":w4:sZ:of:k:g:c:_;[+;W:;Sd;Or;KZ` r;;w;s;o;k<gG<cL<_M<[l<Wy<S<O<KZ` <<w<s=o=k>=gL=cT=`d=\=X=U=Q=MZ` =>w>s?o?kd?gf?c?_?[@W@SO@OQ@KZ` Q@@w@s@o@k4Ag6AcA_A[AWASBO!BKZ` !BoBwqBsBoBk CgCc7C_=C[dCWfCSDODKZ`$ DUDwYDsDpDlEhEdE`E\EXET%FP+FMZ` +F.Fw0FsFoFkGgGcSG_UG[GWGSGOGKZ` GGxItIpIlIh0Jd2J`J\JXJTJPKLZ` KKwKsKoLk LgLdL`AM\CMXMTMPMLZ` MMwCNsENoNkNgNcN_EO[GOWOSOOHPKZ` HPJPwPsPoPkPg"QcQ_Q[QWQSQOQKZ`$$ QQwQs Ro RkRgRc5R_9R[YUWiUSaWOWKZ`    W[w[s^o RkRgRc5R_9R[YUWiUSaWOWKZ`    44^o RkRgRc5R_9R[YUF BP^!44^o RkRgRc5R_9R[YUF BP^!44^o RkRgRc5R_9R[YUF= BP^!444 RkRgRc5R_9R[YUF BP^!444 RkRgRc5R_9R[YUF= BP^!44444Rc5R_9R[YUF BP^!44444Rc5R_9R[YUF= BP^!44444444YUF BP^!44444444YUF BP^!44444444YUF= BP^!4O4Q4244444} 4YUF BP^!}  43 45 4i 4k 4m 44} 4YUF BP^!m 43 45 4i 4k 4m 44} 4YUF= BP^! 4n 4p 4 4 4444YUF BP^!44)4+4 4444YUF= BP^!+|44)4+4 4444YUF BP^!|~44)4+4 4444YUF BP^!~424444444YUF8 BP^!444 44444YUF= BP^! @4_44x44444YUF8 BP^!x4"44x44444YUF BP^!"4444\4444YUF8 BP^!\^4444\4444YUF= BP^!U4W444 4 4!4!4YUF BP^!!C"4E"4#44 4 4!4!4YUF BP^!##4#4#4#4 4 4!4!4YUF= BP^!#$4$4%4%4K&4M&4!4!4YUF BP^!M&O&4v&4x&4%4K&4M&4!4!4YUF= BP^!x&'4'4'4'4(4(4(4(4YUF BP^!()4)4)4)4h*4j*4x*4z*4YUF BP^!z**4*4*4+4+4+4x*4z*4YUF BP^!++4F,4,4 -4+4+4x*4z*4YUF= BP^! - -4 -4,4 -4+4+4x*4z*4YUF BP^! -3-4 -4,4 -4+4+4x*4z*4YUF= BP^!3-5-4Z-4\-4-4-4+4x*4z*4YUF BP^!-.41.4i.4-4-4+4x*4z*4YUF BP^!i.k.4.4.4-4-4+4x*4z*4YUF BP^!..4.4.4-4-4+4x*4z*4YUF BP^!.&/4X/4{/4/4/4/4!04O04YUF= BP^!O0v04040404041414 14YUF= BP^! 1Q1414141424*24S24z24YUF= BP^!z2242424 34134Q34q3434YUF= BP^!3343444$44G44e444434YUF= BP^!444754954764964646464YUF BP^!664754954764964646464YUF BP^!664754954764964646464YUF BP^!66474_74747474b8484YUF BP^!8#94V94949494-:4e:4:4YUF BP^!::4;46;4n;4;4;4<4F<4YUF BP^!F<x<4<4<4=4J=4L=4=4>4YUF BP^!>|>4f?4?4@4Q@4@4@46A4YUF BP^!6AA4A4!B4qB4B4C49C4;C4YUF BP^!;C=C4A4!B4qB4B4C49C4;C4YUF BP^!=CVC4fC4!B4qB4B4C49C4;C4YUF BP^!fChC4C4D4WD4YD4D4"E4hE4YUF BP^!hEE4E4E40F4F4G4UG4G4YUF BP^!GG4G4E40F4F4G4UG4G4YUF BP^!GH4G4E40F4F4G4UG4G4YUF BP^!H4I4I4I42J4J4J40K4K4YUF BP^!KK4L4L4 L4tL4L4CM4M4YUF BP^!MM4EN4N4N4GO4O4O4JP4YUF BP^!JPP4P4$Q4&Q4vQ4xQ4Q4Q4YUF BP^!QQ4Q4$Q4&Q4vQ4xQ4Q4Q4YUF BP^!QQ4Q4$Q4&Q4vQ4xQ4Q4Q4YUF= BP^!Q R4R4$Q4&Q4vQ4xQ4Q4Q4YUF BP^!RR47R4$Q4&Q4vQ4xQ4Q4Q4YUF BP^!7R9R47R4$Q4&Q4vQ4xQ4Q4Q4YUF= BP^!9R;R47R4$Q4&Q4vQ4xQ4Q4Q4YUF4 BP^!;RR47R4$Q4&Q4vQ4xQ4Q4Q4YUF BP^!R*S4S4:T4&Q4vQ4xQ4Q4Q4YUF BP^!:TT4TU4:T4&Q4vQ4xQ4Q4Q4YUF= BP^!TUWU4YU4iU4&Q4vQ4xQ4Q4Q4YUF BP^!iUkU4YU4iU4&Q4vQ4xQ4Q4Q4YUF BP^!kUU4YU4iU4&Q4vQ4xQ4Q4Q4YUF BP^!UU47V4V4V40W42W4Q4Q4YUF< BP^!2W4W47V4V4V40W42W4Q4Q4YUF= BP^!4WW4X4V4V40W42W4Q4Q4YUF< BP^!X1X4X4V4V40W42W4Q4Q4YUF< Z BP^!1XX44Y4Y4Z4Z4[4[4Q4YUF< BP^![[4[4[4[4^\4\4[4Q4YUF= BP^!\V]4[4[4[4^\4\4[4Q4YUF< BP^!V]]4M^4^4^4^\4\4[4Q4YUF= BP^!f=/2!;(BP^!^^4B4C49C4;C4YUF BP^!Times New Roman ArialSymbol4hE4YUF BP^!