Estimating portions without a scale: fist and palm

You rely on fixed guides that you always have on hand: your hand, one particular cup, one particular bowl. The guide only works, though, if you have calibrated it against the scale [1]. Estimating carbohydrate then takes two steps, not one. First you judge how much the portion weighs, then you apply the carbohydrate percentage you already know for that food [2].

The eye is trained by checking, not by tables. Weigh the foods you eat most often for a few weeks. Estimate visually first and only then read the display on the scale [3]. Always use the same vessels: a 250 ml cup, a ladle or a bowl. Product packs help a great deal, because the weight is printed on them [4].

Your hand is the scale you always carry with you, and it gives you four guides:

• the closed fist — about a cup, that is 250 ml, suitable for rice, pasta, potatoes or cereals;
• the cupped hand — a volume of 100–120 ml, half a cup;
• the thumb, from base to tip — roughly a tablespoon, the guide for oil, butter, honey or jam;
• the palm without the fingers — a portion of meat or fish, which does not count toward carbohydrate [5].

The small additions to a main meal, such as oil or honey, are easy to forget in the count [6]. Hands do differ quite a lot from one person to another, though. Check your own fist against a measuring cup and you will then know exactly what to count it as, as the figure below shows [5].

A 250 ml cup of boiled rice or pasta comes to around 45 g of carbohydrate. Half a cup reaches 20–25 g. The two foods behave almost the same, so you do not need separate guides. The cup has to be leveled off, not heaped [1].

One important trap is the difference between raw and cooked. Pasta and rice absorb water and roughly triple their weight, so one cup of raw rice comes to about three cups cooked. Boil a weighed dry quantity once and watch how your usual bowl fills up. This test helps you grasp what is happening, because rice stays hard to estimate even for experienced patients [7].

About 15 g of carbohydrate, if the slice weighs 30 g. The problem is that real slices vary enormously. Homemade bread, sourdough bread and dense bread are cut into thicker slices. A slice of that kind reaches 50–60 g, that is 25–30 g of carbohydrate. Sliced sandwich bread is lighter.

Dense bread looks smaller at the same weight, so the eye underestimates it almost every time [8]. The simplest trick is to weigh a whole loaf and then count the slices it yields. Divide, and you find out what an ordinary slice weighs. Rolls, baguettes and bagels carry more carbohydrate at the same weight.

A fruit the size of your fist means roughly 15 g of carbohydrate. An apple, a pear or an orange of 120–140 g fits this rule well. Small fruits are measured differently. A cupped handful of grapes or cherries also comes to about 15 g of carbohydrate. Berries bring less carbohydrate. The hand estimates fruit better, incidentally, than any other food group [5].

Be careful with the word "medium", because shop fruit is often much larger than that. A big apple or a big banana often weighs 250–300 g and counts as two portions. Count only the part you eat, without peel, core or stones [9]. With dried fruit, volume guides stop working.

The guide is still the fist. A potato that size weighs around 180 g and brings 25–30 g of carbohydrate. Two or three small new potatoes make up one fist together, so you count the total, not the pieces. With mash you again work by volume. A mound the size of your fist comes close to the same figure of 25–30 g of carbohydrate. Potato is, incidentally, one of the foods where estimation goes wrong most often [10].

Frying changes the arithmetic, because it drives out water and concentrates the starch [11]. At the same weight, fries bring about twice as much carbohydrate as boiled potatoes, while oven-baked ones sit somewhere in between. Do not forget the potato in soups, stews and salads. There it sits under a sauce, looks like a mere side dish and is often lost in the count [6].

You start from the idea that the plate holds two or three ordinary portions, not one [12]. The plate itself is bigger, so a double quantity does not look all that large [13]. A restaurant portion of pasta often means two or three cups cooked, that is 90–135 g of carbohydrate, not just 45.

Count what you cannot see as well:

• the bread on the table, often brought without your asking;
• the sauce thickened with flour or starch;
• the breadcrumb coating on meat or vegetables;
• the sweet dressing on the salad;
• the drink, if it is not water [6].

Together these sometimes add up to a third of a meal's carbohydrate. The large chains publish nutrition values, so use them when they are available [14]. Otherwise, compare the plate with your own dishes at home and adjust the portion mentally before you eat, because in restaurants people systematically underestimate what is in front of them [15].

Three things are useful to you:

• a carbohydrate counting app, which keeps the count for you;
• a food composition database, which is where the percentages come from;
• barcode scanning, for packaged products [16].

The largest databases on the composition of foods are the American one and the European national ones. Values can differ between countries, because of varieties, local recipes and analysis methods [9].

Bear in mind that every tool has its weak points too. Foods entered by users are often wrong or incomplete, and diaries kept in an app frequently underestimate real intake [17]. The app's default portions often do not resemble your plate. Remember to always use the same unit of measurement for carbohydrate, grams for instance.

On the European label, the "Carbohydrate" line does not include fiber, unlike the American one [4]. So there is no fiber for you to subtract when you use a European label. Enter your home recipes into the app you prefer. Then adjust the carbohydrate estimate according to what your glucose does in practice.

Yes, but mainly as a diary, not as a measuring instrument. The photo shows you what actually produced the rise you see on the sensor two hours later. Photographing greatly reduces memory errors, which spoil a food diary more often than anything else [18]. It is also the most objective material you can take to a dietitian.

Automatic estimation from an image stays approximate, for simple reasons [19]. A photograph has no depth, so the volume below the surface has to be guessed [20]. Sauce and cheese cover what lies underneath, and the sugar, oil and flour used in cooking cannot be seen at all. It can sometimes help to put a fork in the frame and to photograph at an angle rather than from above [18].

It depends on the meal. With a simple, familiar plate the error is small and rarely affects the insulin dose significantly. With large, dense, mixed meals the error can grow a great deal [7] [8]. The direction is almost always toward underestimating [21]. An error the size of one or two slices of bread is common with a complicated plate, and that is enough to change the glucose level after the meal [22].

The explanation lies with volume, which grows much faster than it looks to the eye. A slightly wider bowl holds a great deal more food. Mixed dishes have to be broken down mentally into components. Each component estimated in your head adds an error of its own. When you are not sure, check your glucose two hours later, with the sensor or the meter [23]. Splitting the insulin dose in two is only done according to the plan agreed beforehand with your doctor.