Page 219 - Safety Risk Management for Medical Devices
P. 219

198   Safety Risk Management for Medical Devices


                   • If qualitative methods are used, are the estimates of risks still valid?
                   • Are there reports of misuse which were not foreseen in the original risk man-
                      agement process?
                   • Are the Risk Controls proving to be effective in reducing/maintaining risk levels?
                   • Is there any evidence that the actual market experience of risk acceptability for
                      the received benefit has changed?

                   One of the key benefits of production and postproduction monitoring is the deri-
                vation of P 1 and P 2 numbers from field data. P 1 and P 2 estimates that are based on
                actual field data are far more credible than expert opinion. However, the key to suc-
                cessful mining of field data is thoughtful, well-planned and well-executed collection,
                coding and cataloging of the field data.
                   P 1 is the probability of occurrence of a Hazardous Situation. P 1 is prevalence. P 1
                has units. The units of P 1 are dependent on the application and are determined by the
                manufacturer. For example, for an insulin pump a sensible unit could be patient-hours
                of operation. For sphygmomanometer (blood pressure monitor) a sensible unit could
                be number of uses. To measure P 1 , a certain time interval needs to be selected.
                   A hypothetical example: 50,000 units of insulin pump model X have been in
                operation between Jan 1, 2017 and Jun 30, 2018. During this period, the average
                length of service for model X has been 4000 hours. In the same time period, there
                have been 250 reported and confirmed cases of over-infusion.
                                                     26
                   P 1 5 250/(50,000 3 4000) 5 1.25 3 10  per patient-hour
                   This information should be fed back to preproduction risk management so that if
                                                                         26
                the prediction of P 1 for over-infusion was different from 1.25 3 10  it can be updated.
                   P 2 is the probability of sustaining Harm, given that the Hazardous Situation has
                already occurred. In the BXM method, risk is computed in five severity classes, as
                defined in Table 26.1.
                   P 2 is outcome-based. To derive P 2 from filed data, we need to know after the
                Hazardous Situation happened, what was the outcome for the patient. For example,
                in the insulin pump example earlier, we would ask, after the over-infusion event what
                was the Harm and what was the outcome to the patient? A potential Harm of insulin
                over-infusion is hypoglycemia. Potential answers to the above question could be:

                   • Patient died (catastrophic)
                   • Patient became unconscious and suffered brain damage (critical)
                   • Patient fainted and was taken to the emergency room, but has recovered now
                      (serious)
                   • Patient felt a little light headed but ate a piece of candy and was fine (minor)
                   • Patient reported feeling a little strange, but it passed (negligible)
   214   215   216   217   218   219   220   221   222   223   224