In cases where we found faulty data, a thorough analysis of the variables involved yielded the following results:

(1)   4% of all cases were caused by sensors that had not been completely implanted. This resulted in a low electric current at the beginning, which meant that all data was entirely void. The primary reason for this error was that the patient did not give proper attention at the beginning to instructions for use.

(2) 1% of all cases were caused because the sensor was left in too long. In some cases, that period was more than six days, resulting in record cards that were overloaded and disorderly.

(3) In less than 1% of cases, the polarizing periods of the electrode were too long: the primary polarizing and stable periods of the sensors exceeded 3 hours. The main reason was that the sensory membrane was not thick enough, but this has been improved.

(4) In 0.5% of cases there was no data on the record cards either because the card was not fixed correctly, or the patients had not been told to remove the sensors after hearing the alarm sound.

(5) In a few cases, the sensors had low sensitivity and deviations in hyperglycemia and hypoglycemia were relatively large. The reasons were the same as #3.

(6) In 0.3% of cases, the cards were removed to download the data intermediately, but were not installed again in half an hour, resulted in missing data.

(7) In 0.5% of cases the Sensor was only implanted halfway.