There are a few things to note about the story. First, though Dosa's piece about Oscar in The New England Journal of Medicine is sometimes described as a "study," it was nothing of the sort. No scientific experiments or medical research was conducted on the feline's alleged death-detecting abilities; the piece was instead a personal essay.
There's nothing wrong with essays, but they are essentially stories and anecdotes, which don't necessarily carry any scientific or evidential weight.
In an analysis of Dosa's book about Oscar published in Skeptical Inquirer magazine, researcher Joe Nickell found a troubling lack of scientific rigor surrounding the cat's surprising abilities:
[Dosa's] evidence is the kind disparaged in science as anecdotal. That is, it is based on personal narratives that may be affected by mistaken perceptions, faulty memory, folkloric influences, and many other faults.... Biased selection is a very real problem: there is a natural tendency for believers in some phenomenon to collect stories supporting it, just as there is for disbelievers to collect stories discrediting it.
And because Dosa admits to making up some events and creating fictional characters in his book, Nickell concludes "there is no point in trying to evaluate the anecdotal evidence: it has been manipulated - in the interest of telling a good story, of course - so it is scientifically worthless."
So there's some question about how reliable the data behind Oscar's legend really is: Was Oscar there at the moment of death, or within a few minutes, or a few hours? Was he there for all deaths, or just some of them? Did he meow at the bedside alerting nurses, or did he just wander by the room at some point earlier?
We have no hard data on these questions. But assuming that Oscar really does visit imminently terminal patients more often than others, there may be scientific explanations.
It's possible that Oscar may simply be responding to nurses and staff activity related to the patient's condition. For example, patients who are near death are likely to have more activity in their rooms (for the obvious reason that nurses pay extra attention to those in danger of dying than they do to medically stable patients). Oscar may visit those beds more often simply because there's more going on there, or the patients seem especially unwell.
It's well known that animals often sense and react to subtle cues around them. This was demonstrated in the famous case of Clever Hans, a horse said to have amazing abilities including reading, writing, and doing math calculations in the 1800s.
Hans baffled huge crowds until finally two psychologists discovered that the horse was responding to unconscious cues from his trainer - even movements as subtle as a smile or a lean forward. The Clever Hans story and its discovery of unconscious cueing is still discussed among psychologists and animal communication experts. In fact, a study last year found that dogs follow cues from their owners' facial expressions.
In some cases the cat was physically brought to the bedsides of people who were known to be dying. This is understandable, since pets can comfort the dying, but is hardly mysterious, and reflects the staff's - not Oscar's - medical knowledge about the patient's prognosis.
It's also important to note that the floor Oscar wanders is not just any hospital wing, but the advanced dementia unit - a 41-bed unit where, as Dosa acknowledges, "passing of some residents was not unexpected due to their advanced condition."
If Oscar wanders by enough rooms for long enough, he will likely be in or near rooms with dying patients just by random chance every now and then. Still, Dosa and others insist, Oscar is uncannily accurate.
Could it be something else? There's interesting research suggesting that dogs can detect cancer and other diseases. The most likely explanation is not some psychic ability but instead that the dogs' super-sensitive noses can detect faint biochemical traces given off by the body's reaction to the disease.
While studies show that canine disease detection is better than chance, it is still not as accurate as formal medical tests in many cases. A 2004 British Medical Journal study found that dogs could detect bladder cancer in patients at a rate 25 percent better than chance, but at a lower rate than standard tests.
Dosa offers a similar mechanism as a potential explanation for Oscar's powers, though it's not clear what specific chemicals released just before death would trigger a reaction. In addition, there are countless confounding odors in a hospital setting that might mask any animal-detected "death scent," including flowers, food, cleaning sanitizers, drugs and perfumes.
The story of Oscar the death-detecting cat is an interesting, lighthearted story, but falls short of being scientifically unexplainable. However that interpretation is more comforting than another explanation: that the lovable Oscar is really a feline serial killer, bringing death wherever he goes.