Other than the too-simple headline, this story about four dogs in Europe finding 71 out of 100 lung cancers is pretty responsible. For readers who get all the way to the end, it will be clear that the dog-detection system is far from available now and might require technology to replicate it into a device that works accurately almost all the time. But we wish the story had looked at some other valid questions about the research: They don’t provide any data about test-retest over time, or the other challenges that might exist with training large numbers of dogs over time and space.
The US spends an estimated $10 billion on treatment for patients with lung cancer. It is the second most common form of cancer in the US, and the leading cause of cancer death. An estimated 220,000 people will be diagnosed this year and about 157,000 will die of lung cancer, according to the National Cancer Institute. If there were a way to detect it earlier, it might add years to those lives and save the cost of more invasive testing for those who don’t have the disease. This story does not include any of this context for the reader. Of course, preventing smoking addiction in the population also helps save lives.
The story does not mention costs. At the very least, what does it cost to train dogs? If you can let someone call this the “holy grail,” you can tell people how much the search for the grail costs.
The study was not designed to compare CT with dog-sniffing, yet the researcher is allowed to say that the dog-sniffing “even surpasses the combination of chest computed tomography (CT) scan and bronchoscopy.”
The story does not explicitly discuss harms from a potential “sniff” test by dogs or machines imitating dogs. It does discuss contradictory research on whether cancer-sniffing dogs can achieve high accuracy. The obvious “harm” of inaccuracy is that a patient goes untreated for cancer, or conversely, is exposed to invasive testing because of a false positive. Without early detection by dogs, patients are already getting exposed to tests such as computer tomography and bronchoscopy.
The story shares that the European study involved only 100 samples from patients, and the dogs identified lung cancer accurately in only 71 samples. The story also points out that the length of training of the dogs, and the variability between dogs, makes this technique largely useless for clinical work. A machine equivalent of the sensing system in the dogs is necessary before this can be useful as a clinical tool, one expert says.
There is no disease mongering of lung cancer here.
The story quotes independent sources and identifies that one expert, who is not tied to the European research, is also working in a related area – developing technology to mimic the canine sniff test.
The story does a good job of explaining how lung cancer is detected by existing CT scan and bronchoscopy, and discusses the exposure of the patients to some risk during these procedures. Where it fails is in giving us numbers for how the earlier detection might show an increase in survival for patients.
The story explains that this very preliminary research procedure with dogs is not available yet.
Yes, the story explains that this new research on canine cancer detection is for lung cancer, which is a new application of a sniff-testing that has already been done for colon and skin cancers.
The story shows reporting beyond a news release.