Data Science for State-of-the-Art Blood Banking (BloodStart)
There are around 300,000 people in the Netherlands who donate blood on a regular basis. Women can give blood up to three times a year and men up to five times, resulting in approximately one million blood donations each year. Patients that receive this donated blood are already in a vulnerable condition, which is why quality control is very important to Sanquin, the blood bank. However, it is also important to keep the overall health of blood donors in good condition. Before each donation, a donor is given a medical screening.
- 2018 - 2022
- Marieke Vinkenoog
If you've ever given blood, you know that your iron levels are checked before donation. Using a single drop of blood, produced with a small finger prick, your hemoglobin (Hb) level in your blood is measured. Hb is a molecule that contains iron, which binds oxygen and transports it through your body. If there is not enough iron in your body, you can become anemic, and your health will deteriorate.
There is not that much iron stored in your body: only around 6 grams in total. Most of it is stored in your liver and bone marrow in a large molecule called ferritin. A small part is circulating in your blood, bound to hemoglobin. From your body's point of view, donating blood is just losing 500 mL of blood, which is around 10% of your total blood volume. Restoring your blood volume happens in several steps. First, the volume itself is restored, as your body increases blood plasma production. Then, new red blood cells are created. This requires the synthesis of new hemoglobin. The iron necessary for this hemoglobin is taken from ferritin stores. The last step is the replenishing of ferritin stores by increasing iron absorption from the food you eat. This process happens over the span of a few weeks or months.
To ensure that donors do not become iron deficient through blood donation, your hemoglobin has to be above a certain threshold before donation: 7.8 mmol/L or higher for women, and 8.4 mmol/L or higher for men. In November 2017, Sanquin also started to monitor ferritin levels in donors. Ferritin is thought to be a better indicator for iron deficiency than hemoglobin, because your hemoglobin starts to drop only when your ferritin reserves are already depleted. Monitoring ferritin allows a risk of iron deficiency to be detected before many symptoms arise. Ferritin is tested in all new donors, and then in every fifth donation. If your ferritin is below 15 ng/mL, you are deferred from donating blood for a year; below 30 ng/mL, you are deferred for six months. These thresholds are the same for men and women.
The first part of my PhD research is about hemoglobin levels in blood donors. Data on all donations from 2006 onward are available: over 12 million donations in total. Not many variables are available for each data point, but when donors visit the blood bank frequently, longitudinal analyses can be done on the data. By taking the hemoglobin level at each donation, I can visualize a hemoglobin trajectory for each donor throughout their donation history. Different trends are clearly visible in these trajectories: some donors have stable Hb levels well above the thresholds, others have stable Hb levels right around the threshold value, and again others show a declining Hb level over time.
Hemoglobin levels vary for several reasons, but an important factor that is not so easily measured is the lifestyle of the donor: how iron-rich their diet is, how active they are, etc. Taking into account factors that we do know (e.g. sex, age, donation frequency) and correcting for those, I assume (part of) the remaining variation in Hb levels to be a proxy for the unobserved donor characteristics mentioned before. I plan to cluster donors by Hb trajectory, identify optimal donation intervals per cluster, and then predict the cluster to which a donor belongs as early as possible. This way, donors can be invited to donate only when their Hb level is well above the threshold, lowering the deferral rate. This will make Sanquin's process more efficient, and keep donors from getting demotivated when they get deferred at the blood bank.
The second project I'm working on now has to do with the new ferritin policy at Sanquin. Several research questions arise: what is the ferritin distribution in the 'normal' population? Are there differences in ferritin distribution between new and repeat donors? And: what is the impact of frequent blood donation on ferritin levels?
The first results show that in new donors, that have never donated blood before, 27% of women already have a ferritin level below 30 ng/mL. In men, this is only 1.7%. In regular donors, these percentages are much higher: 52% of women and 43% of men have a ferritin level below the donation threshold. These percentages have far-reaching consequences for the blood supply, as almost half of the donor population will be deferred from donating for at least six months.
I am currently researching how this new ferritin policy affects donors, and whether deferral from donation for six or twelve months is sufficient to restore ferritin to desirable levels.