How accurate are Continuous Glucose Monitoring Devices?

I have been a diabetic since 1973. Endocrinologists initially started me on Metformin, diet, and exercise. As the symptoms slowly worsened, the Endocrinologists tried out a large number of other diabetic medications available in the market from time to time. About 10 years back the Endocrinologist decided that the Islets of Langerhans in the Pancreas could not be activated to produce more insulin and put me on a 50/50 mixture of Fast-Acting and Long-Acting Insulin. This product came in the form of a premixed pen. The quantity of this mixture I used gradually went down with advancing age. When the insurance companies removed this insulin pen from their approved list, I was prescribed 2 separate vials of quick acting or R insulin and slow acting or N insulin. The effective number of units of each and the proportion of the two was to be determined by the 4 or 5 Finger Prick blood glucose levels taken daily. In fact, these insulins are available at Walmart without a written prescription from a doctor at a fraction of the cost of the premixed pen. My current intake of insulin is 2 units of Fast-Acting insulin and 4 units of Long-Acting insulin before breakfast and 3 units and 6 units respectively before dinner. Unless I walk 3 miles after each of these two meals, I need to take more insulin to retain my blood sugar level between 80 and 140 mg/dl. With all these variations, I have maintained my A1C between 6 and 7 right through my life.
A few months back my Endocrinologist who has diligently taken care of me for the last 17 years recommended that I use a Continuous Glucose Monitoring (CGM) Device to avoid the 4 or 5 finger pricks a day that was necessary to keep my blood sugar levels within the range specified above. There was a long delay in getting this device due to Medicare and Insurance Company approval that was necessary. As soon as I received my CGM device I went on a tour to Honolulu, Hawaii confident that I was well equipped to monitor my blood sugar without finger pricks. One evening when the CGM sensor showed a reading of 82 mg/dl I went walking to a restaurant 1 ½ miles away from the hotel. On the way I nearly fainted with a hypoglycemic attack. I had to take a Uber Ride to get to the restaurant. Even after the meal I had to take another Uber Ride to get back to the hotel as I was still feeling hypoglycemic. I was shocked that the new measuring device had mislead me completely.
When I came home from my vacation, I decided to check the accuracy of the CGM. As the first step, I checked of the accuracy of the Accu-Check blood glucose meter I had with the test solutions they provide. It was accurate. I then started to take Finger Prick readings at the same time as CGM Sensor readings. I was surprised that the CGM Sensor readings were much higher than those given by the Finger Prick blood glucose readings. I investigated for possible causes for this difference but could not find any information with the American Diabetic Association or with the National Institute of Diabetes and Digestion, and Kidney Diseases or with my Endocrinologist. Surprisingly, even the manufacturer of the CGM, Free Style Libre did not have any explanation. Further investigation revealed that the CGM was reading the glucose level of the interstitial fluid which is a fluid that fills the small spaces between cells not contained within blood vessels. It now became clear to me why the CGM manufacturer covers himself by inserting the sentence that states “before changing insulin intake, measure the blood glucose level by the finger prick method”. This clause has been added as all regular medical advice specifies the safe blood glucose levels and not the glucose level of the interstitial fluid.
I then decided to investigate whether there was a relationship between the CGM sensor reading and the finger prick blood glucose readings. I first tested for accuracy of the Accu-Check Blood Glucose meter with the standard solutions they provide and found it was accurate. I proceeded to plot blood glucose readings taken on the Accu-Check Blood Glucose meter against those taken on the CGM sensors at the same time. I plotted eighty pairs of readings taken in March and April 2022 on seven different CGM sensors and the Accu-Check meter on a graph. The graph obtained is shown in the chart below. The X axis shows the CGM Sensor reading, and the Y axis shows the Accu-Check Finger Prick blood glucose levels. The equation to the best fit straight line is Y=0.8264X with a confidence level of 99.27%.
I then developed the table shown below based on the above equation. I assigned multiples of 5 to the CGM Sensor readings in the first column shown in red and calculated the corresponding Finger Prick blood glucose reading for each assigned value using the above equation and entered them in the second column in green. I now use this table to convert the CGM Sensor reading I obtain into actual Finger Prick blood glucose reading regularly to control my diabetes. I have had no hypoglycemic incidents since I started using this chart.
Alert! These give only AVERAGE reads. When the sensor shows 85 it could be as low as 60 like what happened to me in Honolulu.


Sensor Finger Sensor Finger Sensor Finger
70 58 150 124 230 190
75 62 155 128 235 194
80 66 160 132 240 198
85 70 165 136 245 202
90 74 170 140 250 207
95 79 175 145 255 211
100 83 180 149 260 215
105 87 185 153 265 219
110 91 190 157 270 223
115 95 195 161 275 227
120 99 200 165 280 231
125 103 205 169 285 236
130 107 210 174 290 240
135 112 215 178 295 244
140 116 220 182 300 248
145 120 225 186