This happens to all of us.
When I was just getting started, I had a patient in the hospital with some inflammatory problem of some sort. I thought I was hot-stuff, thought this was GCA. Told myself I didn't need a biopsy, I was really sure and a negative biopsy wouldn't change my mind (which in GCA is an issue, since the biopsies are relatively low yield and this was long before we were using U/S to target sites). I put her on high dose steroids and BAM! her symptoms got all better. I was really proud of myself. The day before discharge she told me her vision was blurry. I examined her eyes as best I could, everything was fine. I assumed it was nothing. The next AM when I saw her, she was completely blind. I assumed this was GCA (although bilateral would be uncommon). Ophthal saw her urgently, and her eyes were completely WNL. A brain scan demonstrated the problem -- she had bilateral occiptal infarcts. She rapidly decompensated and died three days later. 6 months after that, I'm at M&M conference and as the case unfolds I realize 1) it's her, and 2) something really bad is going to be found that I missed. It's a really horrible feeling. By then, I had made enough mistakes that I was on my way down from Mt. Stupid and joined the discussion pointing out I was caring for her, and I clearly missed something/made an error here. Pathology from autopsy was very definitive -- she had invasive aspergillus everywhere. It's angioinvasive and infarcted her brain. No question that the steroids let it run rampant.
I had a husband and wife in my primary care practice, and my wife and I were somewhat social with them also. The husband got admitted late for something that seemed minor - mild abdominal / back pain. The resident saw them. This was back-in-the-day, so there were no hospitalists. When my patients were admitted, I was the inpatient attending of record. It was late in the PM, I had plans. I asked the resident if they thought they needed me to see them, they felt comfortable. So off to dinner I went. Dinner was interrupted by a page, that my patient was coding. I scrambled back, and he was dead. I took one look at the chart, and the documented marked difference in femoral pulses, and I didn't need an autopsy to tell me what happened. I still wonder if I would have noticed and saved his life had I come in. Telling her he had died and watching her collapse to the ground in tears still haunts me.
The stories go on and on. I expect there are more if you do procedures regularly.