You're between appointments, scrolling an abstract on your phone, and a number jumps out: 91.8% survival. It's tempting to file that away as a fact you now know. The retrospective study behind that number earns five more minutes than the abstract gives it.
Retrospective chart review is the workhorse of veterinary emergency and critical care research. Nobody is randomizing GDV dogs to "treat" versus "don't treat." So the field runs on pulling records after the fact: count the cases, report the pattern. That design is genuinely useful, and it is also full of traps that don't show up in the abstract. Three real papers from the Journal of Veterinary Emergency and Critical Care show three different versions of the trap.
This matters because the number in the abstract is usually the number you repeat to an owner at the kitchen table, or the number you cite when you're deciding how hard to push for a referral. If the number is answering a different question than the one you think it's answering, that conversation starts on the wrong foot.
The 64.1% That Was Actually 83.5%
Start with Sharp and colleagues' 2020 review of gastric dilatation-volvulus in JVECC, a single-center, 10-year chart pull from Tufts' Foster Hospital for Small Animals. It's a retrospective case series, n = 498 dogs. The headline number: 319 dogs, or 64.1%, survived to discharge.
Keep reading and the number moves. Of the 179 nonsurvivors, 149 were euthanized. Of those, 116 were euthanized at presentation, before surgery, with no intent to treat. Pull those dogs out of the denominator and survival among dogs who were actually taken to surgery climbs into the low 80s.
Neither number is wrong. They're answers to two different questions. "What happens to a dog that walks in the door with GDV" and "what happens to a dog whose owner elects surgery" are not the same question, and one line in an abstract will not tell you which one you're reading. It's also a single-center study at a referral hospital in a specific decade, so the case mix, the owners' financial picture, and the surgeons on call are specific to that building. A busy three-doctor practice in another state has no guarantee its GDV dogs look the same on paper.
The Group That Skipped Epinephrine
Confounding by indication is the second trap, and a 2026 JVECC paper on canine anaphylaxis shows it clean. Kadowaki and colleagues at Louisiana State University pulled records on 49 dogs with severe anaphylaxis, retrospective, single center, January 2019 through December 2023. The dogs in the study are the ones that did not get epinephrine as part of early treatment. Overall survival in that group was 91.8%.
Read that as "epinephrine is optional" and you've made a leap the data can't support. These 49 dogs didn't get epinephrine because a clinician, in the room, made a judgment call about severity or presentation. The paper's own numbers hint at why: shock index at presentation was significantly higher in the dogs that did not survive (p = 0.022), and lower body temperature on presentation tracked with the need for pressor support (p = 0.021). Dogs who looked sicker were likely managed differently than dogs who looked milder, and that decision is baked into the outcome before you ever reach the survival percentage. A retrospective study can tell you what happened to a group of animals whose treatment wasn't randomized. It can't tell you what would happen if a clinician chose, on purpose, to withhold a drug.
Fourteen Cases, No Denominator
The third trap is the case series that reads like a rate. A 2025 JVECC paper on norepinephrine extravasation, authored by Vasquez and colleagues, pulled together 14 patients (13 dogs, one cat) across four university hospitals and two private specialty groups in the US, Canada, and Australia between 2015 and 2021, 16 extravasation events total.
That paper is genuinely useful. It describes what norepinephrine extravasation can look like and what clinicians did about it, down to individual interventions. What it can't tell you is how often it happens. Nobody reported how many dogs and cats received norepinephrine through peripheral or central catheters over those same six years at those same six sites. Fourteen events is a numerator with no denominator. Skim it fast and it reads like an incidence figure. Read it slow and it's a writeup of fourteen cases someone remembered to report.
Six sites across three countries also means six different infusion protocols, six different sets of nurses checking catheter sites, and six different thresholds for what counts as a reportable event. A multicenter case series widens the geography. It doesn't turn a list into a rate.
Reading the Next Abstract
None of this makes retrospective studies worthless. In an emergency and critical care literature where true randomized trials are rare and expensive, chart review is often the only data the specialty has. It does mean the abstract is not the whole paper, and the whole paper is one data point, not a verdict on the disease.
Three questions travel well from these three papers. What counts as a case, and did the denominator quietly shrink somewhere between the introduction and the results table. Why did one group of animals get different treatment than another, and is that reason something the paper can rule out or something it just glosses past. Is there an actual rate here, or is this a list of memorable cases with no count of how many similar cases never made it into the writeup.
We're not asking you to reread the methods section of every JVECC paper between appointments. We are asking you to hold the topline number a little more loosely than the headline wants you to, especially before it changes how you talk to the next owner in exam room two.