Regulatory Open Forum

The tools have a variety of applications, so selecting the right tool for a particular situation can be a problem. In product safety risk management, as in ISO 14971:2007 or ISO 14971:2019, the problem two parts. The first half finds the ways your device can harm the patient, while the second half keeps it from happening. The first half is the hazard, sequence of events, hazardous situation, harm categorization. The second half is risk reduction.

The hazards include both normal and fault conditions. Some people try to apply FMECA, but that only deals with failures. Hazard analysis is more comprehensive.

One powerful application of an FMECA is in a production process. By analyzing the probability and detectability of production process failures, the analysis can determine if a nonconforming product escapes. If it can, evaluate the effect on the patient or use. Nonconforming products do not always result in patient or user harm, e.g. the color of the case is the wrong shade of blue.

Another FMECA application is reliability. The tool helps determine the probability that the device fails "during the mission". However, not all failures result in patient harm.


FMECA, by nature, deals with single point failures. It does not provide information in a normal condition or include a sequence of events.

However, hazard analysis uses a sequence of events, which is not a single point failure. The appropriate tool here is, in my opinion, FTA. It starts with a top event and breaks it down into all the ways the top event can occur. In reliability, the top event is a failure. However, in hazard analysis the top event can be the patient or user harm. It can be in a normal condition or a fault condition. The FTA analysis provides all the paths by which the top event can happen – the sequences of events. If you were able to assign frequency of occurrence to the basic events you could calculate the frequency of occurrence of the top event.

There is a new tool that shows up sometimes call the bow tie model. It combines aspect of the FTA and the FMECA to provide a more comprehensive picture.

My recommendation is to use FMECA as an input to hazard analysis, not in place of hazard analysis. FTA, in which the top event is a patient or user harm, is a powerful hazard analysis input. Bow tie analysis is a very close match to the 14971 methodology and can handle both normal and fault conditions.

One small point, Dan. We do not use the term "hazard analysis" as we are not analyzing the hazard in ISO 14971. This is explained in the Preamble to 21 CFR 820 from 1996. The proper term is risk analysis as we are analyzing the risk in ISO 14971. 

Fault Tree is a powerful tool that can give lots of help in identifying such points as the probability of harm occurring. A slightly different permutation of Fault Tree is Event Tree Analysis. These also make up for one of the major limitations of FME(C)A, that if single fault. ISO 14971 does not limit itself to single fault, but requires all known and foreseeable hazards to be identified in both normal and fault conditions. 

As you mention, some companies attempt to use FME(C)A as their only risk analysis tool. This does not meet the requirements of risk management. The reference I made to ISO/IEC 31010 lists many useful risk analysis tools, and if I remember correctly, bow tie analysis is one of them. It is a useful tool. But like all tools we must take into account their strengths and weaknesses and use them correctly. 

in a product liability case I am presently involved in, the company only used FMEA and did not look at normal condition hazards. Probably the result will be,  "How big is the check?" And in the 8 figure area based on some of the lawsuits these days. A well done risk management process can keep a company out of these problems. After all, we are supposed to prevent or reduce harm and provide a benefit to the patient and the user. 

As I start my training for risk management for the practitioners, "You are going to be the first patient using this device. Do the risk management accordingly!"  Wakes up my classes when they ponder that statement. 

The tools have a variety of applications, so selecting the right tool for a particular situation can be a problem. In product safety risk management, as in ISO 14971:2007 or ISO 14971:2019, the problem two parts. The first half finds the ways your device can harm the patient, while the second half keeps it from happening. The first half is the hazard, sequence of events, hazardous situation, harm categorization. The second half is risk reduction.

The hazards include both normal and fault conditions. Some people try to apply FMECA, but that only deals with failures. Hazard analysis is more comprehensive.

One powerful application of an FMECA is in a production process. By analyzing the probability and detectability of production process failures, the analysis can determine if a nonconforming product escapes. If it can, evaluate the effect on the patient or use. Nonconforming products do not always result in patient or user harm, e.g. the color of the case is the wrong shade of blue.

Another FMECA application is reliability. The tool helps determine the probability that the device fails "during the mission". However, not all failures result in patient harm.


FMECA, by nature, deals with single point failures. It does not provide information in a normal condition or include a sequence of events.

However, hazard analysis uses a sequence of events, which is not a single point failure. The appropriate tool here is, in my opinion, FTA. It starts with a top event and breaks it down into all the ways the top event can occur. In reliability, the top event is a failure. However, in hazard analysis the top event can be the patient or user harm. It can be in a normal condition or a fault condition. The FTA analysis provides all the paths by which the top event can happen – the sequences of events. If you were able to assign frequency of occurrence to the basic events you could calculate the frequency of occurrence of the top event.

There is a new tool that shows up sometimes call the bow tie model. It combines aspect of the FTA and the FMECA to provide a more comprehensive picture.

My recommendation is to use FMECA as an input to hazard analysis, not in place of hazard analysis. FTA, in which the top event is a patient or user harm, is a powerful hazard analysis input. Bow tie analysis is a very close match to the 14971 methodology and can handle both normal and fault conditions.