Counting carbs

October 2012

I am still on my own at work, and it's not going too badly. I like the job more every day, and my steep learning curve continues. I feel privileged to read and hear the accounts that patients share with their doctors,  nurses and other healthcare professionals, even if they are not complimentary. Patients are angry, sad, depressed, anxious, overwhelmed or struggling, and luckily for my state of mind they are occasionally happy, positive, grateful or just demonstrate a very welcome sense of humour. They are always interesting, and sometimes interested.

I am now working with several patients on carbohydrate (carb) counting, which means that the patient estimates the carbohydrate content of everything they eat. I try to start with a brief description of digestion, then ideally find out what the patient tends to eat on a typical day. This allows me to understand their choices a bit better, and tailor the rest of the consultation to suit that person - for example, if the usual diet contains couscous and tofu, I might go about things differently compared with pie and chips.

Carb counting is usually something that adults with diabetes tend to learn after they've been diagnosed for a little while, a few months at least (although there are exceptions). So they ought to know already which foods do and don't contain carbohydrate, although it always pays to check their knowledge at this stage. Even within my short experience, I've found a surprising number of people who have been choosing their insulin dose according to the quantity of food on the plate, rather than considering only the carb-containing foods. For example, a very large cooked breakfast including bacon, eggs, sausage, black pudding, beans, mushrooms and tomatoes with a mug of tea with a splash of milk contains very little carbohydrate, and the amount of fat and fibre in the meal means that no quick-acting insulin needs to be injected.

So we establish appropriate knowledge of carb-containing foods, then move on to quantifying the amount they contain. I can provide different types of written information, and we have food models, pictures of food on plates and packets of food complete with nutritional labels. Then I can refer back to the typical day's diet, so the patient can have a go at estimating the carbs in their typical day. That's the first stage, and sometimes that's all that happens to start with. The patient goes home, and for a period of time just estimates the amount of carbs in their food without changing anything else.

The next stage is to apply this knowledge to insulin dosage. Essentially, to maintain good levels of blood glucose, carbohydrate intake and rapid-acting injected insulin need to be closely matched. This is a simple statement, but it's never that simple. There are different insulins, people react differently to the same food. Other factors impinge - stress, activity, alcohol, hormones, medications, illness, previous dietary intake, dosage of long-acting background insulin, the sex and size of the patient, previous blood glucose readings, injection sites, quality of insulin, quality of injection devices - there are a myriad of possible factors that will mess things up. But to start with, we work with just a few numbers.

At this second stage, having checked whether the patient's carb estimation is reasonably accurate, they might choose to start adjusting insulin dosage. This is done using two ratios: the amount of carbohydrate that is matched to a unit of insulin (or vice versa), and the change in blood glucose level that can be brought about by a unit of insulin. There is also a target range of blood glucose that people aim for, which is chosen individually and may depend upon the time of day - let's say for the sake of argument that it is between 5 and 8 mmol/L. We will further assume that the background insulin is at the right level, although this is an assumption that can rarely be made in the real world.

The patient tests blood glucose before a meal, and estimates the carb content of the meal. If the pre-meal test is within range, then the insulin to carb ratio is used to calculate the amount of rapid-acting insulin to be injected. For example, if the ratio is 1 unit to 10g and the meal contains 50g of carbohydrate, then 5 units of rapid-acting insulin are needed.

If the pre-meal test isn't within range, then a correction can be applied. If the pre-meal test shows blood glucose is high, then extra insulin can be given, and if low then an amount can be deducted. For example, if the pre-meal test is 15.2 mmol/L and the correction dose is 1 unit to 2 mmol/L, then 3 or 4 extra units of insulin will be needed along with the insulin to match the carbohydrate in the meal. If the pre-meal test is 4.3 mmol/L, then one fewer unit of insulin might be given - 4 instead of 5 units for a meal containing 50g of carbohydrate.

In an ideal world, this would result in a relatively steady blood glucose level that may rise immediately after a meal, but would return to within the target range by the time the next meal is due - no higher, and no lower. A higher blood glucose may make the patient feel ill, and increases the risk of diabetic ketoacidosis and long-term complications. A blood glucose below 4 mmol/L may make the patient feel ill, and needs immediate treatment to mitigate the risk of hypoglycaemic coma and, in the worse case scenario, death.

Together with a specialist nurse, this week for the first time I helped a patient to start carb counting. The nurse prescribed the insulin, but I suggested the ratio and correction dose. Ever since that consultation I have been worrying that my advice was flawed, to the extent that I have been trying to contact the patient to check that all is well. So far, I haven't managed to get in touch, and all I can hope is that if anything had gone wrong, the patient would have contacted us.