The Oncologist’s Perspective on Statistics: Part Two

In last week’s post, Statistics Part One, I discussed why statistics can be very helpful to the pet Guardian. And while stats are an important part on oncology, my years in practice highlight their limitations. So before we dive into some common statistical terms (in my next blog), I think it is important to remember that stats can be useful to give you reasonable expectations … but they will never predict every individual or outcome.

For example, I tell pet Guardians that about 5% percent of chemotherapy patients get hospitalized due to its side effects. But I will also remind them, that if your dog gets admitted, that 5% chance is less important to you, because your pet is now 100% sick and needs hospitalization for supportive care.

Similarly, I sometimes get frustrated with statistics, because there are no guarantees that your dog will live up to the statistics. I think many pet Guardians and even my colleagues get too caught up in the numbers. They get a biopsy report, and ask me: How long will the dog live? It’s just not that predictable, even with the more common cancers.

For example, canine lymphoma is relatively common, well-studied and well-reported. Without treatment, statistics tell us the “average” dog lives 1 month, but with a multi-agent chemotherapy protocol, the median survival times are 13 to 14 months. Statistical analyses provide prognostic factors, like phenotype (B vs T cell), substage (sick at diagnosis), stage (how advanced), and prior use of prednisone.

For a recent patient, Stella, I was happy to report that she had had many positive predictors: the cancer was in her B-cells (which has a better prognosis than when it is in the T-cells), she was asymptomatic at diagnosis, and was in stage III. Her remission rates should have been 90-95% with a survival of over a year. But Stella never went in remission, even though I tried multiple protocols and drugs, and she only lived about 4 months.

But then there was Guinness. He was very sick, with advanced stage V GI T-cell lymphoma. He lived over 18 months with chemotherapy. He outlived the B-cell dog. The statistics did not predict these two dogs well.  This does not mean the statistics are wrong – it means they do not predict every response or every survival time.

On the other hand, I often deal with uncommon cancers, ones you don’t find in the Dog Cancer Survival Guide. The statistics and clinical studies on less common tumors are often lacking. There may be one, two, or sometimes three published papers on that tumor type. In many papers, the dogs were not all treated the same – different chemotherapy protocols, different surgeries, and different radiation protocols.  In these tumors, good stats are scarce, and drawing conclusions may be difficult. If your pet was diagnosed with one of these, I strongly encourage you to seek out the help of an oncologist. Check out www.acvim.org to find an oncologist in your area, or ask your family vet for a referral.

I am happy when I am “wrong” and your dog is outliving the statistics. Stats can be helpful if you understand the limitations, and remember there are no guarantees. So look for information from stats, and remain cautiously optimistic.

In my next blog, we will look at some of the terms relating to stats and dog cancer. For more information on lymphoma and statistics, check out the Dog Cancer Survival Guide.