Page 133 - Exploring the Potential of Self-Monitoring Kidney Function After Transplantation - Céline van Lint

Self-monitoring: the reliability of patient-reported data 131
Reliability: correctness and representativeness of registered data
The reliability of registered data is determined by both the correctness and the representativeness of registered values. To study correctness of the registered data, we investigated the one on one correspondence between a registered and measured value on a given day. Three different categories were distinguished:
1. Reliable SMSS registrations: in case a value that was registered in the SMSS corresponded to the
device logged value on a given date. Only days with an equal number of measurements logged in
the device and SMSS were taken into account.
2. Non-correspondence, in case an SMSS registered value did not correspond to the device logged
value on a given date. Only days with an equal number of measurements in the device logged and SMSS logged data were selected. All cases of non-correspondence were carefully checked for potential causes of the deviance (e.g. wrong combination of date and measured value, typo, rounding off). The cases where no potential cause was found were termed incorrect entries. For each patient, a mean level of creatinine was calculated for the values that were actually measured and for the values that were registered using cases of incorrect entry only. A paired t-test was performed to compare these means. Total and median number of non-corresponding values was calculated per patient. Patients with a high number of non-corresponding values were selected for further exploration.
3. Phantomvalues,incaseavaluewasregisteredintheSMSSonagivendatewhileaccordingtothe data stored in the device no measurement was performed on that specific date. All potential phantom values were thoroughly checked for alternative explanations before it was concluded that there was no relation with values that had been measured by the patient. A paired t-test was performed using the mean of the phantom values versus the mean of all measured creatinine values per patient.
Further, to get a reliable impression of a creatinine level over time (trend), the SMSS registered values need to represent what was actually measured. It is therefore important to know how often a measured value was not registered in the SMSS and whether the unregistered values differed in any way from the registered values. The measured values not being registered in the SMSS were split into two categories:
1. Omissions, in case one or more measurements were performed on a given date, but no value was registered in the SMSS. Total and median number of omissions per patient was calculated. For each patient, we calculated a mean level of creatinine for the values that were both measured and
6