Medical device clinical trials are complex undertakings for many reasons, but most specifically due to the clinical safety and efficacy unknowns. The presence of these unknowns can make the process of informed consent difficult. Although there are three components of informed consent (information, comprehension, voluntariness), it is the information segment that can be most challenging. Based upon my experience as an institutional review board consultant, this challenge is often the compound result of several coexisting issues, namely the following: (1) performance data are limited, because investigational devices have not yet been tested in humans and the results of animal studies have to be extrapolated to potential human events; and (2) available product development data may not have been shared with clinical investigators. Additionally, there is the matter of the medical team transmitting to patients or research subjects information that may be of uncertain value or high complexity. In discussing this matter, I have selected cardiothoracic implant devices; however, the arguments could be exported to other devices as well.
FMEA (failure mode and effects analysis) documents are prepared by device manufacturers during product development and are generally part of the overall document submission package to regulatory agencies such as the FDA (United States Food and Drug Administration). FMEAs represent a structured way of evaluating the components of a device in an effort to determine the manner in which each component could fail and how the failure would affect the performance of the component itself, and potentially the device as a complete unit. By correlation, an FMEA could be a potentially useful tool in the identification of possible failures when a device is used in a clinical setting. 1 Furthermore, because FMEAs are living documents and are continually revised, even after the devices are in clinical use, the data coming from this clinical use (e.g., product complaint reports) correlates to risk, especially when reported failures are not user related (surgeon), but design or use related (human factors).
The exact structure of an FMEA is not rigidly defined by either engineers or regulatory agenices 2; however, I offer that, in the case of implantable medical devices, the following should be included: (1) identification of the individual components of the device; (2) determination of potential failure modes for each component; (3) the potential causes of each failure mode; (4) the effect of each failure mode on the patient; (5) the effect of each failure mode on the surgical procedure (e.g., lengthening the duration of the procedure, termination of the procedure, modification of the implantation technique); (6) the criticality of the failure; and (7) the probability of the occurrence of the failure mode. Criticality can be ranked on a numerical scale beginning with 1 (leads directly to severe injury or death) and ending with 4 (a minor defect such as a cosmetic blemish that will not impact device performance or patient safety). Probability can be similarly ranked on a scale of 1 to 4, with 1 representing a 1 in 100 chance of failure, and 4 representing a 1 in 1,000 chance of failure. Criticality and probability rankings do not have uniformly set identities; that is, there is no set industry or regulatory standard for the formulation of these rankings. Quality engineering handbooks and professional societies denote various strategies for both criticality and probability. In my own experience, I have found variations among strategies used by different manufacturers.
For a patient, probability data can be potentially frightening, for it is foreseeable that patients might automatically view themselves as being on the receiving end of a bad outcome, even if the probability data are not that dramatic. Risk communication has been empirically studied, and it has been determined that the complexity of the risk information, as well as time constraints, can affect a patient’s comprehension. 3 As a result, tools have been developed to help the public to better interpret health risk information and guide communicators to better present such information. The Department of Veterans Affairs Medical Center (White River Junction, VT) has created a tutorial entitled Understanding Numbers in Health that reviews basic concepts of probability and their application to medical studies to help people become better critical readers of health information. 4 It would be advantageous to use such beneficial tools during the informed consent process so as to help prevent miscommunication and misunderstanding of risk probabilities.
These probability data, analyzed in conjunction with data from “the effect on the patient” and “the effect on the surgical procedure” can be of benefit to the surgical team as they consider project risk and as they develop informed consent documentation for use of an investigational device. This information can also be useful to surgeons in their decision-making about whether or not to attempt repair or replacement of an implanted device in light of coexisting surgical risks. By balancing the clinical risks of a particular failure mode (e.g., leak) against the surgical risks for a patient who may have comorbidities, surgeons can strategize about the optimal time for device repair or replacement. It is possible that immediate re-operation might not be necessary, allowing time for some clinical parameters to stabilize or normalize without negatively affecting the outcome of re-operation. Not all device or component failures require immediate repair or replacement, and knowing when surgical delay is permissible can be facilitated by the criticality information provided by an FMEA, as criticality is reflective of the sense of urgency of each potential failure mode. Therefore, the information provided by an FMEA can be helpful in determining the timing of re-operation (e.g., immediate, delayed, elective), even though the patient’s overall clinical presentation is the primary driving force in decisions about if and when to re-operate. An example of an FMEA is shown in Table 1.
How does knowing the functional effect of a component failure translate to determining a clinical effect? In the case of a left ventricular assist device, several critical failures can occur 5–8 yet, for example, how does pump stoppage inform that backflow, regurgitation, pooling, thrombosis, and heart contractility problems can result? Having possession of the FMEA in advance of beginning a clinical trial can be a first step in the analytic process of identifying clinical risks. As a failure mode, pump stoppage would have been bench tested using a simulated cardiothoracic model equipped with blood flow, allowing observed mechanical events to be extrapolated to clinical events. The possible clinical manifestations that are predicted through animal testing can then be used to help determine the risk profile of the device (exclusive of associated surgical complications). Certainly not all failure events (and recovery procedures) can be foreseen by an FMEA; however, it is the best compilation and analysis of potential events.
The intricate engineering of each device is best understood by the device manufacturer and they have the responsibility of providing appropriate training and documentation for each product. Although surgeons might be generally aware of the potential failure modes of a particular investigational device, review of the FMEA will provide the best baseline of information for educating them as well as the potential clinical trial participants. Knowing this, the transfer of the FMEA to the clinical investigator requires collaboration with the device manufacturer. Design confidentiality can still be maintained through customary agreements between manufacturer and the clinical site; thus, these concerns should not limit the information shared with clinical investigators. Minimizing the potential harms to study participants can be aided by full disclosure of the FMEA to clinical investigators so that they themselves are fully informed of information that is critical to device safety and efficacy. Similarly, knowing the potential failure modes and their effects in advance gives surgeons the ability to devise procedures to deal with potential adverse events. The creation of trial participant informed consent documents such as diagrams, handbooks, workbooks, and forms can be greatly enhanced when surgeons have the up-front knowledge that the device FMEA provides. Patients can thus be presented with a fuller disclosure of risks facilitating a more informed choice.
The informed consent documentation used for a clinical trial is generally detailed and lengthy. Because of this, there can be a tendency to “thin” documentation in an effort to simplify the informed consent process once investigational devices are approved for routine clinical use. Experience cautions against this practice, 9 as it weakens the efforts of the FMEA process and it potentially short changes patients from the full disclosure they need for informed decision making (e.g., potential risks such as pump dependence, in the case of ventricular assist devices). I am not suggesting that FMEA documents be given to patients or research subjects, or that the FMEA document be an attachment to the informed consent form, as certainly voluminous information alone does not ensure informed consent. Furthermore, presenting a patient or research subject with an FMEA document would likely cause frustration or fear due to the inability to comprehend the information presented in such a technical format. The importance of an FMEA is found in its content, and this content is best reviewed by physicians and reflected upon in a nontechnical manner during drafting of informed consent documentation and discussions with potential patients or research subjects during the informed consent process. Particularly at the stage of clinical investigation (when therapeutic safety and efficacy data are lacking), FMEAs are potentially the best tool to provide information regarding foreseeable clinical risk. This is especially important when the devices used are invasive. 10
To this end, sharing accurate, up-to-date FMEAs with surgeons can be a valuable aid to the informed consent process for all involved parties. I pose that there be a shared responsibility between surgeons and manufacturers to ensure that each device’s FMEA is revised throughout the history of the product’s use. This should take the form of surgeons reporting adverse device experiences to the manufacturer (per Federal requirements) and the manufacturer initiating a timely investigation with FMEA revisions as warranted. Similar strategic concepts have been proposed for novel surgical procedures. 11 The FMEA should be understood as a living document that will change over the course of device use in an effort to provide the most accurate risk profile. Although there are no Federal requirements to use FMEAs in the manner that I have proposed, from an ethics standpoint, analytic strategies such as FMEA can be beneficent to the learning curve and informed consent dilemma presented by investigational devices.
1. Juran JM, Gryna FM: Quality Planning and Analysis from Product Development Through Use. New York, McGraw-Hill, 1980, pp. 182–184.
2. Bachmann J: How to conduct a failure mode and effect analysis: An interview with Ashweni Sahni. Med Device Diagn Ind
October 1991: pp. 18–24.
3. Schwartz LM, Woloshin S, Welch HG: Risk communication in clinical practice: Putting cancer in context. Natl Cancer Inst Monogr
4. Woloshin S, Schwartz LM: How can people make sense of medical data? Effective Clin Pract 2: 176–183, 1999.
5. Knosalla C, Weng Y, Buz S, Loebe M, Hetzer R: Pseudoaneurysm of the outflow graft in a patient with Novacor N100 LVAD system. Ann Thorac Surg 70: 1594–1596, 2000.
6. Moazami N, Argenziano M, Kohomoto T, et al: Inflow valve regurgitation during left ventricular assist device support may interfere with ventricular remodeling. Ann Thorac Surg 65: 628–631, 1998.
7. Piccione W, Kao W, Mattea A, Rodriguez ER, Trohman R: Failure of an implantable left ventricular assist device: A distinctive electrocardiographic pattern before malfunction. J Thorac Cardiovasc Surg 115: 1376–1378, 1998.
8. Scheld H, Soeparwata R, Schmid C, Loick M, Weyand M, Hammel D: Rupture of inflow conduits in the TCI-Heartmate system. J Thorac Cardiovasc Surg 114: 287–289, 1997.
9. Bramstedt KA, Wenger NS: When withdrawal of life-sustaining care does more than “allow death to take its course”: The dilemma of left ventricular assist devices. J Heart Lung Transplant 20: 544–548, 2001.
10. Syse A: Norway: Valid (as opposed to informed) consent. Lancet 356: 1347–1348, 2000.
11. Bull C, Yates R, Sarkar D, Deanfield J, de Leval M: Scientific, ethical, and logistical considerations in introducing a new operation: A retrospective cohort study from paediatric surgery. BMJ 320: 1168–1173, 2000.