UPC Cambridge

ISO 13485 | Medical Devices Quality Management | MD 69713

Technology Insights

Data Integrity

Evidence

Analysis

Transcription

Observations

Data Entry & Check

Conclusions

Numerical Analysis

Recommendations

#1 Evidence

Raw Data

E-Data

Log Book

Wet Ink Sheets

#2 Analysis

Data Entry

Numerical Analysis

Run graph check 

X,Y,Z Graphs

Trend Lines

Data integrity is a worthy subject, if dull to most. Evidence must be stored as raw data, ready for re-inspection.  UPC use wet ink as the preferred method, and raw e-data in a vault. The raw data becomes couplets, triplets or more; then graphed for trend analysis. To present data that is accurate and correct ‘beyond reasonable doubt’ is a hard task, that only a few do with full understanding and grace.

Door frame height readings recorded in a ‘log book’ in indelible ink over ten years.  A rule was used to transpose the marks into heights for the two growing children, date  

The measurement error in heights is ±1mm at best, depending on whether the sharpie mark was placed with the flat surface at 90° to the frame.  Transcription error ±0.5mm.

Checking: A date transcription mistake was found when graphing versus time on P trace.

Q: Compare the two children’s height with time?  Will P be taller than T when the same age?

O: T will not get much taller as his growth is slowing.  P is still growing taller over the last 12 months.

O: The data predicts that P will not catch up T over 502 days. 

C: Past performance does not predict future performance…  R: Keep taking data…