Phototherapy Units

Table of Contents

Some biomedical background

It’s useful to review a little bit about why we do phototherapy. All neonates have some elevation in bilirubin, due to a variety of physiologic factors. Bilirubin comes from the breakdown of the heme in red blood cells, which creates unconjugated bilirubin. Unconjugated bilirubin is fat soluble and can cross the blood brain barrier, and if present at high levels can cause acute and chronic neurologic sequelae. Bilirubin must be made water soluble before it can be effectively excreted in the bile and urine. Unconjugated bilirubin can be conjugated in the liver, or we can photo-isomerize it in the skin, either of which allow it to be eliminated.

Photo-isomerization happens most effectively when the light is around 460-470nm in wavelength, which is a blue color. Anything in the 420-500nm range works to a degree. Phototherapy irradiance is measured in units of microwatts per square centimeter of exposed skin per nanometer of blue light spectrum. Sunlight is typically in the range of 5-10 μW/cm²/nm; the issue with sunlight is that it also contains light in the ultraviolet (<400nm) wavelengths, and so must be filtered. “Standard” phototherapy should be around 8-10, whereas “intensive” phototherapy is defined as >30 μW/cm²/nm. In general, for term children, there is no reason not to do intensive phototherapy, but very small infants (<1000g) should probably be limited to < 25 μW/cm²/nm as higher levels have the potential to cause neurologic injury (although there is still no consensus on the safe or appropriate level for premature infants).

My personal experience is that most phototherapy units in circulation in low and middle income countries do not even hit standard irradiance thresholds, let alone intensive.

Some electrical background

While you can use halogens, white compact fluorescents, or blue compact fluorescents, LEDs are the best choice, in my opinion. All of these can make an effective phototherapy unit, but LEDs are by far the least expensive and require the least maintenance. To put some numbers to it, the compact fluorescent units that my mission hospital had been using were built with 6 tubes, each costing USD $30. When you add in the lamp holders and the carpentry, the units cost around $230 to build. These would reach an irradiance of 10-14 μW/cm²/nm with fresh bulbs, but would rapidly degrade with heavy use, and the bulbs would need to be replaced every 3-6 months. By contrast, an LED unit costs about $35 to build, and the power supplies seem to go bad every 1-2 years, which cost about $15 to replace, although typically they can be fixed by swapping out the capacitors.

LEDs can be powered in one of two ways. The more efficient way is to use a constant current DC power supply (also called a constant current driver); this will vary the voltage within a range to maintain the target amperage through the LEDs. Constant voltage power supplies, by contrast, supply a fixed voltage but require a current limiting device on each LED to prevent overcurrent. This typically is in the form of a paired resistor, which reduces efficiency and generates more heat. While a poorer technical choice for these reasons, constant voltage power supplies are much more available and are usually less expensive, which is why our design uses them. LED bulbs that insert into a socket have the driver (typically constant current) built into the base.

Now for the LEDs themselves – thankfully, commercial blue LEDs are typically of the iridium-gallium type which produces light peaking in the 450-470nm wavelength range, which is just where we want it. “White” LEDs are made by combining red, blue, and green LEDs – so they will work, as they contain blue light, but are less efficient for our purposes. The same is true for color changing LEDs. Due to cost and availability, we use LEDs strips for most of our phototherapy units. These usually come in 5 meter rolls with 300 LEDs per roll. Each LED has a paired resistor so they work with constant voltage drivers – the standard is 12V. You will usually see them listed as SMD5050 or SMD3528 or some other number. SMD stands for “surface mounted device,” and the numbers following are the dimensions of the LED in 0.1mm. The ones I usually use are 5050, which have LEDs that are 5mm x 5mm, as these put out more light per LED, but any will do – you’ll just need to adjust the number of strips. Look for “blue,” not “icy blue.” Some manufacturers will list the wavelength but most won’t.

For 300 LED SMD5050 rolls, each 5 meter roll usually uses around 72W. So you’ll need a 12 volt DC driver that will supply 72W/12V = 6A per roll. The units we build use 2 rolls, so I typically use a 180W 15A power supply.

Shopping List

You’ll need the following. These may be available locally; if not, I have had good results ordering from aliexpress.com.

• Two 5 meter long (300 LED) 12V SMD5050 strips (usually less than $5 each); if using LEDs smaller than 5050 you’ll need more.

• One 180W 12V 15A power supply (usually around $15-20). Any DC power supply will do, and it’s OK if the amperage/wattage is higher, it just needs to be 12V. If you can only find smaller ones that’s OK too, you’ll just need to combine them.

• Basic soldering supplies (less than $1 of materials). The electrical work is well within the skillset of an electrical appliance repairman, so if you don’t solder yourself often and don’t feel like learning from YouTube then I’d recommend hiring someone for the soldering work.

• A cord and plug ($1-2), some 12-14 gauge (1.5-2mm) wire ($1-2), and a switch (< $1) to toggle between high intensity and lower intensity. If your plug is not fused I would recommend adding a fuse to the AC side of the circuit.

• Wood, screws, and paint or polyurethane ($5).

• Optional but recommended: surge protector and thermal fuse (I use 77C but the exact number is not critical, I would recommend in the 70-100C range)

• Optional but strongly recommended: a blue light irradiance meter ($250). I ordered from a Solarmeter brand unit from Amazon.com and had it shipped to my country. I know it is expensive but I think it is a worthwhile investment. I’ve tried using the luxmeter built into my phone but could not get consistent results. I think you might be able to make a very inexpensive meter with a photovoltaic cell and a multimeter but I haven’t had time to experiment to see how reproducible it is.

Total cost is around $35, time required to build a unit if you’ve never build one before is about 2 hours.

Instructions

Measurement and Maintenance

If you have a irradiance meter, which is an investment I strongly recommend if you can afford it, you should regularly test your units to make sure they are continuing to hit target levels. If you do not have a meter, I have some general conservative recommendations about replacement schedule, but these are more theoretical than tested.

We test our units monthly, and as needed if they appear less "bright" to the eye. To test the device, place the irradiance meter at the distance the baby's skin will be from the phototherapy unit (keeping in mind that premature babies will be further on average) and move it around the testing field to get a feel for what the average reading will be. Test "in-situ," using the incubators or bassinets that are used in your setting. In our setup the range is about 3 μW/cm²/nm difference between the brightest and dimmest locations in the incubator. I tend to use the low average value, as in general I would rather be overpowered than underpowered. We have been using the units for several years now and the only component that has needed repair or replacement has been the power supplies, but if the power starts to decrease it would be easy to add or replace LED strips.

If you do not have an irradiance meter, I would check the current output of the power supply monthly with a multimeter, and replace it if it drops by >10%. I would also routinely replace the LEDs every 2 years of typical use. This is very conservative: LEDs are probably good for around 10,000-20,000 hours; if we assume around 12h use on average per day this would mean we replace well before the 10,000 hour mark.

Here is the maintenance sheet we use:

Bonus: how to modify an irradiance meter to increase its upper limit of detection

As mentioned above, our irradiance meter was not able to read high enough to measure our intensive units, pegging out at 26μW/cm²/nm. To modify it, I used a nylon tea strainer mesh and doubled it up, forming it into a dome shape. This is taped to a cardboard sleeve which fits snugly around the meter. The mesh is then slipped over the sensor. Make sure that the mesh is always a fixed distance from the sensor or else it may make the attenuation less consistent. I tested the meter in a number of different settings with and without the mesh to determine the percent attenuation, and it was consistently measuring between 39-41% of the unfiltered reading.

Thanks

Thank you to Dr Tina Slusher and Dr Hendrik Vreman for lending their expertise (and spectrometer).

Soli Deo gloria.