Software failures were behind 24 percent of all the medical device recalls in 2011, according to data from the U.S. Food and Drug Administration, which said it is gearing up its labs to spend more time analyzing the quality and security of software-based medical instruments and equipment.
The FDA’s Office of Science and Engineering Laboratories (OSEL) released the data in its 2011 Annual Report on June 15, amid reports of a compromise of a Web site used to distribute software updates for hospital respirators. The absence of solid architecture and “principled engineering practices” in software development affects a wide range of medical devices, with potentially life-threatening consequences, the Agency said. In response, FDA told Threatpost that it is developing tools to disassemble and test medical device software and locate security problems and weak design.
In an e-mail statement, the agency said that it is developing “techniques and laboratory expertise to assist our review staff in identifying potential vulnerabilities and evaluating risk mitigation measures” similar to those used in “regulated industry.”
The Agency is also acquiring expertise in areas like “detecting malware inside device designs…(and) reverse engineering certain types of malware to best identify the specific protective practices which manufacturers should be employing,” the report reads. (PDF)
The statement is the clearest indication to date that FDA is shifting focus to make software quality an area of interest. The agency has come under fire in recent years for not holding manufacturers’ accountable for insecure or poorly written software. There is growing evidence that software security and integrity is a growing problem in the medical field. In October, for example, security researcher Barnaby Jack demonstrated a remote, wireless attack on an implantable insulin pump from the firm Medtronic. The attack could have enabled a remote assailant to command the pump to release a fatal dose of insulin to a diabetic. That presentation was similar to one in August, 2011, at the DEFCON hacking conference in Las Vegas. In that case, researcher Jerome Radcliffe – a diabetic, himself – demonstrated how he could remotely manipulate the dosage levels delivered by this insulin pump from up to 300 feet away. The demonstration prompted calls for a review of wireless medical device security from two U.S. Congressmen.
FDA says it “shares the concern” of security researchers about the security and privacy of medical devices. FDA “emphasizes security as a key element in device design. Any system with wireless communication can be subject to interception of data and compromised privacy as well as interference with performance that can compromise the safety and effectiveness of the device,” according to an e-mail statement from the Agency.
“We continue to closely monitor for safety or security problems,” FDA wrote.
Security researchers point out that many medical devices now rely on some form of embedded software for management and diagnostics. However, that software is often rife with vulnerabilities and other exploitable security holes. Moreover, medical device makers don’t deploy tools and strategies long used by commercial software developers to ensure the integrity of software and communications. Those include the use of cryptographic hashes to verify the authenticity of software updates.
Recent research done on the security of medical devices by a team of researchers identified software security vulnerabilities in software that controlled an Automated External Defibrillator (AED) which is used to treat cardiac arrhythmias (PDF). The researchers also found that the device would accept unsigned, counterfeit software updates.
Research by University of Massachusetts researchers Shane Clark and Kevin Fu found that the problem of securing the software that runs medical devices is complicated by the tension between the need for devices to operate securely and the need for many to operate swiftly and reliably in the event of an emergency. Software coding problems and fuzzier “human factors” – such as our tendency to ignore or incorrectly use security features were also identified as confounding factors in medical device security.
Those questions were brought into sharp relief this week with news that a Web site operated by the firm CareFusion Inc. was hacked and infected with malicious code. The Web sites in question were used to distribute software updates for CareFusion’sAVEA and VELA brand respirators. In a statement, CareFusion said that the infection lasted around two weeks and did not affect any downloadable software on the site.
Fu, of The University of Massachusetts, said that device manufacturers operate in a manner that’s almost “medieval” compared to mainstream technology vendors.
“There are some very basic problems right now, like merely recognizing that it’s not safe to distribute software without cryptographic protection for authenticity, integrity, and freshness. It’s not part of the the manufacturing vocabulary,” he wrote. “Some companies might do a good job, but I have yet to find a single company that issues digital signatures or hashes of medical device downloads.”
And that doesn’t even address the (bigger) problem of software quality. However, there’s evidence that the FDA is beginning to turn its attention to that topic. In the latest OSEL report, FDA describes efforts by its Division of Electrical and Software Engineering to identify deficiencies in a medical device manufacturer’s software quality inspection processes. In one instance, a DESE-employed biomedical software engineer identified a pattern of customer complaints about incorrect or missing notifications to clinicians when test results were out of range. The DESE employee linked the reports to what FDA said were “several coding defects which directly caused many of (the) customer complaints. “Some defects were basic violations of software coding practices while others were new defects that were introduced during correction of the previous defects.” The audit resulted in the manufacturer issuing two FDA-mandated “Correction and Removal” notices and 11 Class II recalls of its products.
Still, the FDA maintains that it is the job of manufacturers to secure their products and make sure they stay secure.
“Manufacturers are responsible for identifying risks and hazards associated with medical device software (or) firmware, including risks related to security, and are responsible for putting appropriate mitigations in place to address patient safety,” the agency said in an e-mail statement. “Information related to theoretical device security problems is helpful. However, it is very important that the agency receive reports of devices that have had security breaches.”




You’d have to be a real low-life scum-bag to hack someone’s insuline pump and deliver a fatal dose. Too bad we can’t assume there isn’t human trash out there that would do such a thing.
Doesn’t always have to be malicious intent. What about the embedded software programmer whose code was not thoroughly tested (at least from a secure coding perspective) before it was loaded onto the devices?
I wrote some info about this from things I’d seen when I was going to school a few months. They had this scenario where we were to advice a doctor about what to do with his Linux network that contained confidential data, and I told him to disconnect it from the web because they had already violated hypocratic oathby having them on networks that could be hacked in the first place. I kind of got a chuckle of this being in print.
The statements that this article “quotes” from the referenced OSEL report aren’t in that report, if you follow the link and download it. It points to a file “osel2011.pdf” – if you search that for malware, 24% – they are just not present. Even the term “recall” only appears 2 times, neither in the context of a summary of 2011′s results.
I am presuming that the reference report is the wrong one, or else the author of this article really stretched what is in the OSEL report.
with all the tools, education, and cool laptops available to developers.. they still can’t manage to generate secure code. i think we’re pampering this obese coders too much.
I worked for a year at a med device company, after working several years in defense (where I am working again). I asked a coworker who was developing the new wireless comm what kind of encryption they were planning to use, and he stared at me blankly. I immediately thought of potential assassination, but the people who worked there just didn’t think that way. I gently warned them that they might want to build robust security into the development plan!
Worse yet, the plan then was for a patient to have an external transmitter/receiver by the bed or favorite chair that would passively download data or upload new settings from the doctor periodically so the patient wouldn’t have to come in to get checked up. So THAT device, the external home transmitter is ANOTHER thing that could easily be targeted, because it would connect via internet to the doc’s office.
Made me verrrry nervous. I wonder what has been done for security since then (2006)?
I have reason to believe my sister was killed by a hospital nurse who turned the morphine pump to its maximum rate in order to steal all the morphine from the machine prior to its transport to a nursing home. When the same machine was refilled and reconnected (by low-end nurses at a nursing home) my sister quickly became “unresponsive” and died.
but you gotta admit, it is kinda lulzworthy.
Not necessarily a “low-life scumbag”, NSA or CIA or MOSSAD directorate could decide such expediency is for the good of humanity.
You’re still describing a low-life scumbag, by the way.
It’s in the graph, Figure 5.
You choice of lists is interestingly revealing
From that section with Fig 5:
In collaboration with the inspection team, Ms.Simone identified a trend in customer complaints that the firm had not been aware of. This trend involved incorrect or missing patient results in a laboratory information system, and incorrect or missing notifications to clinicians that test results were out of range.
Basically, problems with LIMS or other systems. It’s not that the device released the incorrect dose, it’s that the record keeping system had an incorrect dose history or test result, causing a doctor or nurse to administer an incorrect dose.
Similar to the issue discussed in Economist’s science and technology blog on May 23, and nothing to do with “malware”:
The quote above with the term “malware” obviously did not come from the linked report.