When my son Drew was diagnosed with type 1 diabetes, we were lucky to be supported by the clinic staff at the Children's Hospital of Michigan. I've heard terrible stories from families about their clinic experience at other hospitals, involving a check list of video tapes to watch, or information sheets to take home and read on your own. Instead, we were given the best education and support from a staff of dietitians and CDE's and endocrinologists, many of whom have type 1 diabetes themselves. I'll never forget when one CDE came in to the hospital room, sat down on Drew's bed with him and said, "I'm sorry to hear that you have type 1 diabetes. You will learn how to take care of yourself with insulin and never need to feel this sick with DKA again. I've had type 1 diabetes too, for 30 years. Do I look sick to you?"

We were probably better equipped than many "newly diagnosed families" because my wife has a college background in medical technology before switching to business systems development, and I have a background in biomedical engineering before switching to military electronics applications. We understood what was being taught to us by the clinic team: blood glucose testing, basal insulin needs met by Lantus insulin and its day to day adjustments, carb counting and Novolog insulin to cover meals eaten, and correction doses. We learned about glucagon administration and rights to a public education and 504 plans, how to keep a diabetes log book, and countless other things.

At first, things went quite well. We knew what to do, we knew how to do it, and our younger son was happy to let us do it for him. He was up to the challenge and attended to his diabetic needs with a positive attitude. His HbA1c went from an 11 plus at diagnosis to a 7 in the first month, and was found to be less than 5.0 (in other words, it was completely normal) on his first regular 3-month endo appointment. Our CDE's then actually said to ease up a bit, not try to run so low, think about the long term, aim for A1c in the 6's, try for a better balance between the burden of diabetes care and living life to it's fullest. We should for example get a better nights sleep when we can, and let him play on the school football team if that's what he wants to do.

That's how it continued in the first year. It wasn't perfect by any means, but we were making the best of it and achieving A1c's in the 6's time and again. We transitioned to insulin pump therapy with little change in the program. The biggest problems we had were actually small problems, such as some resistance to test and keep a log especially when at school for example, or to test blood sugar before football practice. There were a few secret slurpies (but at least they were bolused for correctly), and some food hidden under the bed (we never denied him food except to limit large snacks at bedtime, but he felt it was necessary to hide it anyway).

However as he became a teenager, Drew wanted to have more independence and generally had the attitude that nothing bad could ever happen to him, just like his other teenage friends. If he tested his blood sugar and found it to be good, then he felt he didn't really need to test again for a while. He skipped lunch at school to play basketball with his friends. He rode his bike with his friends up to the ice cream stand in town and had whatever we wanted, just like them. Little by little, he was testing less often, against our wishes and the clinics recommendation. He started to forget to "bolus" more often when he ate something, especially snacks when he was active because they would be partially needed to "feed his activities" to some extent anyway. He stopped measuring carbs and stopped bolusing "food in mouth -- hand on pump" as was the clinic rule. Instead he bolused for 100 grams, and then ate as he desired and adjusted with sugar snacks after eating to strike the balance.

Eventually, he became so confident in his abilities to go it alone and so nonchalant with his diabetes care that he was no longer getting away with it unnoticed. A1c's were typically in the 7's now, and that seemed to be good enough for the clinic -- not optimum perhaps, but not unreasonable for a teenager they always say. There was one A1c test in the 8's, and we were very concerned. He would forget to do more and more of his care, skip infusion set changes on time, lie about things when he didn't feel like doing them. He refused to keep a log anymore, which didn't seem entirely necessary with the pump because the pump software does it for you, but that attitude carried over into pump vacations when he preferred to use injections for a short while. He would give himself a Novolog injection so that he could skip a set change until later. He would hide candy in bed and give himself an injection of Novolog from a syringe so that it wouldn't show up in the pump log. He would enter a fake BG reading in his pump, and then give a meal bolus based on it (which bypasses the bolus correction calculation that the pump would have done to suggest the best insulin dose). Things were going downhill, and it wasn't the lack of training that was our stumbling block: rather it was the continual burden of therapy.

Recently, we had our worst day with diabetes so far. Drew is 15 years old now. My wife and I were away for the day, across the state to visit my daughter at college. Both our teenage sons were home alone, playing video games and playing football outside with their friends. Drew did not eat much all day, or test his blood sugar. We called to ask, "How's it going and whats your blood-sugar level" (that's not a run-on sentence, but a single complete thought for all parents of children with diabetes) but he gave us a fake number rather than test, and we believed him. Later he went to bed with his glucose level and food and insulin out of balance. He went to bed without testing. As usual I tested him that night at 2 am, and found him to be very low (48). I turned off his pump to deny insulin for 1 hour (taking away a half unit which should raise his blood sugar usually about 40 counts) and also I woke him to give him a juice box. He sipped the juice and said he felt fine, so I thought he had not been "low" for very long then. Typically we give 15 grams of carbohydrate with a small juice box, but that week the store did not have the small ones, only those larger boxes with 22 grams of sugar per box. This would often be equivalent to over treating a low, but with this very low number of 48 I was happy for the extra few grams of carbs. I mentally calculated that the total correction should push his BG level from 48 to about 130.

The rules we were taught from the clinic are to retest a low 15 minutes after treating it to make sure that the treatment was sufficient. We've done that many times, but I did not that night. I was exhausted from the long drive, and had extra confidence in the big correction. I also felt confident because I knew my wife would be testing him again early in the morning as I slept in.

A funny thing happened at 5 in the morning. He woke up and went downstairs for breakfast. I woke up and asked him what he was doing because it was Saturday, but he replied that his alarm was set and when he woke up he prebolused for his breakfast cereal, and now had to eat it regardless. I asked him why he was up at 5am instead of the usual 6am, and he replied that he must have changed the hour accidentally when he set his alarm. I was annoyed at these stupid circumstances, but his answers seemed coherent, and I knew that he must indeed eat what he had already bolused for. I did not know however that he had bolused 10 units, which for him is enough for 200 grams of breakfast carbs.

At 8 in the morning, my wife tested Drew again and found his blood sugar level to be 130. He was also hot and sweaty, but that can be normal for him sometimes when he sleeps with too many blankets. She did not know that he was battling the lows at night, nor of the huge breakfast bolus, and so she did not recognize his adrenaline response to a low blood sugar and the rebound.

When I got up at 10am and asked about him, she said is glucose level was 130, which seemed to me to be just as expected. She did a correction bolus to 110, which was a very small dose of 0.3 units. We do that to get him close to 100 in the morning so that we can "strike the spike," even if he eats his favorite breakfast cereal.

At 11:30 we woke him up. It was Saturday and he likes to sleep in, but that was late enough. He came down to eat, grabbed a cereal bowl and a spoon from the cabinets, went to the pantry to get a cereal box, and dropped his spoon on the wood floor with a clang. He picked it up but immediately dropped it again. He picked it up yet again, and when he dropped it the third time, my wife noticed that his hand was spasming and so he could not grip the spoon. She called for me and we sat him down. His arms were now twitching at the elbow, up and down like bird wings. I ran for his testing supplies upstairs while she gave him some juice right away. Just as I was ready with strip in meter to poke him for a blood sample, Drew went into a full seizure.

His eyes rolled back into his head. The legs were now kicking and twitching like his arms. The worst part was the chest convulsion, with a grunting wheezing sound that was terrible. He could not breath. We got him laid down on the floor as we prepared the glucagon injection. He was foaming at the mouth by then, and there was a tinge of pink in the foam because he had bit his cheek or tongue. It was very scary as we gave the glucagon injection, fumbling with the kit that had seemed so easy when practiced but now had to do for the first time in an emergency.

His seizure lasted only 1 minute before he was fully unconscious and still, and was breathing normally again. His blood sugar level was found to be in the low 60's, not severely low, but this was a few minutes after the juice and the glucagon. We will never know what his level really was.

Drew was now perspiring heavily and turned white while we waited for the glucagon to take effect. His blood sugar was up to 80 very soon so we decided not to call 911 yet but to wait. In 10 minutes he was showing signs of recovering, opening his eyes when we called his name. Blood sugar was 130. We had him lying on his side with a bucket ready in case he threw up, and then he starting groaning and holding his stomach. He sat up and vomited violently, laid back groaning some more, sat up and vomited again, and then laid down and passed out again. In a few more minutes, he began to wake up in earnest, was opening he eyes, listening to our explanations, and responding to questions. It was a very long 15 minutes.

He felt ill for about 6 hours as we kept feeding him and testing his blood sugars. His muscles were sore from the convulsions, especially his neck and back. He also had a headache. He did not remember anything after getting the bowl and spoon, not even vomiting. We let him "ride a little high" for a few hours, with his BG test numbers reading in the 200's.

This was the first time (and only time so far) that we needed to use the glucagon kit. It worked. It was injected into the front of his thigh, left no particular mark or bruise or even soreness from what seems such a large needle compared to the little insulin injections we are used to giving. We wished it would have worked faster (more glucagon?) but also wished it has fewer side effects (less glucagon?). The final result is that he was back to normal faster than paramedics could respond, and that we kept him out of the hospital that day. It was a bad day for the whole family, but we are grateful that we had the glucagon kit and were able to use it immediately.

Now, consider all of the things that went wrong. First he was out of balance. That happens all of the time when his measuring and bolusing are not carefully balanced, and when his activities and basal rates are not balanced. He was also not testing vigilantly like he used to. Second he did not adequately recognize his own systems well enough to treat his low blood sugar, or even to avoid making the problem worse by bolusing inappropriately. Third, his support system let him down even though we were testing him manually at night and early morning as we do every day: his Dad did not retest after a bad low, or recognize a bizarre breakfast pattern at 5 in the morning on a Saturday, his Mom did not know what had happened at night and so did not recognize the waking sweats of the adrenaline response or the rebounding blood sugar or Somogyi response. Then in the morning, severely low, he was up and walking around right in front of us, behaving normally right up to the moments before the seizure.

By the way, the insulin pump was continuing to pump out the basal insulin all night long, exactly as we had programmed it to do. After the 2am temporary reduction for one hour, it automatically began at 3am to pump a little more than half a unit per hour. Between 3am and noon, it had therefore delivered an additional 5 units into our very hypoglycemic kid.

God, I hate diabetes.

I consider myself to be an expert caregiver using the best tools and techniques available. I've had the best training I can imagine, and then supplemented it with information from several well written books and seminars. I am an active member of what I consider to be the best support group of parents of children with diabetes. I have changed my career to work fewer hours and concentrate on the task of taking better care of my son. I have not slept through the night in years. I have done my best -- and still I have been measured and found lacking. Perfection is required, day in and day out, and I failed to be perfect. If only I had some better information about what was happening that night so I could have made better decisions, or some help from the equipment such as automatically reducing the basal rate to deny insulin. If only he had more help to get the insulin in balance in the first place. This is where an artificial pancreas can help. We don't need a perfect closed loop system from perfect sensors. We can already do very well with an open loop insulin pump on our "good days." It's the bad days where we need help the most, when there are huge gaps between where we want to be and where we really are. Do those of you who are working on the closed-loop solution now really understand this?

Frankly, I am rather tired of waiting while the medical researchers discuss if the CGMS systems are mature enough yet to be worthy of closing the loop. I have certain respect for them, but it is obvious they have insufficient control systems engineering experience in this matter. As an electrical engineer, I have closed the loop many times. Let me tell you, the sensors you need to close the loop never seem to be good enough. They always add delay. They always alter the data and need to be compensated. They all have failure modes that you must understand and prepare for. Sometimes even then there is a seemingly random dynamic response of the closed loop because of the nature of the system itself inside the loop, so you must understand these things and watch out for them. This is normal in the real world of control systems. Our job is not to judge the sensors and wait for one that is "mature." Rather our job is to make the best use of data available right now, and then make better use of new improved data when it becomes available in the future from even better continuous blood glucose sensors. In other words, the medical team is going to have to trust the systems engineers here. We can do it to the benefit of the pump users with the CGMS systems already out there now: we don't need to wait any more.

Let's get started.

Community content is available under CC-BY-SA unless otherwise noted.