N/A in nutrition data: why a missing value is not a zero

N/A stands for not available, not for absent

In food composition data, N/A is short for not available: no value has been measured or published for that food and nutrient pair. In some tables the same letters read as not applicable, for a nutrient that makes no sense for the product. Either way, the note describes the state of knowledge, not the food itself.

A zero means measured and not detected at the chosen threshold; an empty cell means never measured. On screen the two look alike, but their evidential weight differs.

Notations vary: the Swiss database prefers a wording close to not stated, others use a dash, n. b. or k. A. The same dash can mean no data here and not present elsewhere: a legend only holds for its own table.

A, E, C: where each figure comes from

Every value shown on a datafood page carries an origin flag, taken from the FSVO database:

• A: analysed value, from a laboratory measurement.
• E: estimated value, taken from a very similar food or from a documentary source.
• C: calculated value, derived from the ingredients of a recipe or by conversion.

Out of the 26570 published values, 34% are analysed, 14% estimated and 53% calculated. A cooked vegetable shows many C flags, since cooking is modelled from the raw food; an unprocessed food concentrates the A flags. The flag ranks nothing: it states how much certainty comes with each reading.

Traces, less than, N/A: three different kinds of blank

Three notations coexist in the database:

• traces: the nutrient is detected, but the amount is too small to be quantified. 235 values fall into this case.
• less than: the content sits below a limit of quantification, without being nil. 144 values carry this marker.
• N/A: no data at all for that nutrient. 182 values remain open.

The almond page shows the second case for vitamin C, raw apricot the first for sodium, and Emmentaler, full fat the second again for vitamin C. The original notation is kept, never turned into a number.

Missing values: what datafood refuses to fill in

Food composition databases handle missing values differently. Some impute them: a family average, a figure borrowed from a similar food or a value taken from another table, delivering complete looking tables built on numbers no laboratory measured. Other systems quietly treat every missing value as zero.

All of that is ruled out here: no family average, no borrowed value, no missing cell read as zero. The reason is statistical: in analytical chemistry, a result below the detection limit is censored data, and counting it as zero pulls averages down and leads to underestimated intake totals.

The rule is visible in the rankings by vitamin D content: a food with no known value does not appear there, for want of a comparable figure, not because its content is nil. The comparison tool flags cells without data instead of counting them as zero.

Nutrition Facts labels and databases: two separate systems

The Nutrition Facts panel on a package follows food law, not database practice: a closed list of mandatory nutrients, tolerances around average values, rounding down to zero below a threshold, vitamins and minerals declarable only above a significant amount. No legal text requires N/A there; it remains a table publisher’s convention.

Hence a frequent question: why do the lines never add up to the weight of the food? Because water, total minerals, organic acids and other compounds are not all listed. A table is a selection of documented nutrients, not an exhaustive breakdown.

Reading a page together with its gaps

The database covers 1216 generic foods, spread across 19 families, with 22 nutrients per page. No page is complete across the board, and the display shows it.

Three reading points: the A, E or C flag before any figure is quoted; the difference between an empty cell and a displayed zero; the coverage of a nutrient before any comparison. The portion calculator follows the same rule: a missing value is flagged and left out of the total, never filled in.

Sources, notation conventions and the handling of missing values are set out in the methodology; the pages are listed in the food index.