In a delayed diagnosis case, the breach is often the easy part. The hard part is proving what would have happened instead: at a hospital the patient never reached, on a day that never came, in the hands of physicians who never saw her. The defendant did not create that evidentiary gap in any conventional sense, but her negligence is the reason it exists. How much inference is a trial judge allowed to do to close it?
Ibrahimova v Cavanagh, 2026 ONCA 573, released on August 7, 2026, is the Court of Appeal for Ontario’s most direct answer to that question in some time. Dr. Angela Cavanagh did not challenge the finding that she was negligent. Her appeal was entirely about causation, and her central complaint was that the trial judge had filled the gaps with speculation dressed up as common sense. Gomery J.A., writing for a panel that included Copeland and Monahan JJ.A., dismissed the appeal. I wrote about Justice Sproat’s trial decision when it was released, and the appeal leaves it intact.
Four emergency visits in four days
Tamara Ibrahimova was 28 years old and 17 weeks into her first pregnancy when she went to the emergency department at the Kincardine site of South Bruce Grey Health Centre on Friday, May 3, 2019, reporting vaginal bleeding, lower abdominal cramps and a pinkish discharge. Dr. Cavanagh, a family physician who occasionally worked in the emergency department, reviewed her ultrasound and bloodwork, examined her, consulted an obstetrician by telephone, diagnosed a threatened miscarriage and told her to follow up with her obstetrical care provider.
She returned about 12 hours later. She had woken to a gush of fluid, and she showed Dr. Cavanagh a pad with straw coloured fluid on it. Dr. Cavanagh’s own evidence was that she thought this was likely amniotic fluid. She did not order a white blood cell count or an ultrasound, and she did not consult or urgently refer to an obstetrician. She recorded a final diagnosis of threatened abortion, advised watch and wait, and sent Ms. Ibrahimova home.
Two more physicians saw her over the next two days. Dr. Sivanesan Kalaichandran, at Alexandra Marine & General Hospital on May 5, recorded a heart rate of 100 and a self-reported low-grade fever, found white blood cells in her urine, cancelled a complete blood count he had ordered, and made a semi-urgent obstetrical referral. Dr. David Ponesse, at the Walkerton site on May 6, diagnosed a second trimester bleed and told her to come back the next morning. Her husband, Oleksii Gubenko, called an ambulance at 2:08 a.m. on May 7, roughly eight hours after that discharge. She was in septic shock. She was airlifted to a London hospital, where the pregnancy was evacuated at 9:30 a.m.
The consequences were catastrophic: partial amputation of her right foot, amputation of her left leg below the knee, kidney failure, a stroke, seizures and compromised use of her right arm. She was in hospital from May 7, 2019 until March 9, 2020, and on home dialysis for more than 17 months before a kidney transplant in July 2021. The trial judge described her prognosis as dire.
After a 28 day trial, Justice Sproat found that Dr. Cavanagh breached the standard of care on May 4 and that the breach caused the injuries. He also found Drs. Kalaichandran and Ponesse negligent, but dismissed the action against them because Ms. Ibrahimova would have developed severe infection and sepsis even absent their negligence. That allocation is worth pausing on. On Dr. David Talan’s evidence, accepted at trial, a dilation and evacuation had to occur by 2:00 a.m. on May 6 to avert septic shock. The May 5 and May 6 encounters fell outside the window, so a narrow causal window concentrated the entire loss on the earliest breach. Whether that reasoning holds up is not something this appeal explores, because only Dr. Cavanagh appealed.
Where inference ends and speculation begins
The trial judge used the delayed diagnosis framework from Sacks v Ross, 2017 ONCA 773, 417 D.L.R. (4th) 387, reaffirmed in Hasan v Trillium Health Centre (Mississauga), 2024 ONCA 586, 499 D.L.R. (4th) 178: establish what actually happened, then determine what would have happened had the standard of care been met. I have written before about the way Hasan handles an evidentiary gap the defendant’s own negligence produced, and the same tension runs through this case.
He also cited Goodman v Viljoen, 2012 ONCA 896, 299 O.A.C. 257, at para. 76, which cuts in both directions. It authorizes a robust and pragmatic approach to causation, decided on the balance of probabilities rather than to scientific certainty. It also warns that resorting to speculation, or to common sense on a question that requires expert knowledge, is a misuse of that approach rather than an application of it.
Dr. Cavanagh’s four grounds of appeal, distilled from nine alleged errors, were essentially variations on the theme that the trial judge had crossed that line. The most substantial was the first: that he made four findings about what would have happened at the London hospital without evidence capable of grounding them. There was no witness from the hospital, no maternal-foetal medicine specialist, no neonatologist, and no evidence about specialist availability or how operating room time was allocated there in May 2019.
The Court of Appeal held that the findings were open to the trial judge on the evidence he had, and that the evidence he had was of two kinds: the opinion of an obstetrician who actually worked at the receiving hospital, and the documented sequence of what happened when Ms. Ibrahimova did arrive there on May 7.
Dr. Robert Di Cecco, an obstetrician and gynaecologist at the London hospital, testified that a patient with a previable, prelabour, preterm rupture of membranes (pPPROM) would be told that the pregnancy was unlikely to reach viability, that a surviving foetus would likely have significant complications, and that continuing carried a risk of life-threatening sepsis. He testified that infection is the primary risk in pPPROM, which is why it is a medical emergency, and that any evidence of infection would mean antibiotics and no realistic option but to terminate. The defence called two experts on liability, in emergency medicine and infectious diseases. It called no obstetrician. Dr. Di Cecco’s evidence on the obstetrical counterfactual was therefore uncontradicted, and the trial judge had ruled that the defence emergency medicine expert could not testify about what obstetricians would do.
That absence did most of the work on appeal. The Court of Appeal treated the argument that an obstetrical team would have overlooked classic signs of possible infection, or would have needed an infectious diseases specialist to confirm them, as untenable. On the timing of the procedure, the trial judge had drawn an inference that a large tertiary hospital keeps specialists available around the clock and prioritizes emergent procedures over deferrable ones, and had found that inference corroborated by what happened on May 7, when the evacuation was performed within two hours of arrival. The Court of Appeal saw no basis to interfere.
There is also a fallback in the reasons that deserves attention, because it may matter more than the headline holdings. Even if the two hour inference had been wrong, it would not have been an overriding error. Dr. Talan’s deadline of 2:00 a.m. on May 6 left roughly 36 hours from the counterfactual admission on the afternoon of May 4, so the case never actually depended on a two hour turnaround.
Assuming the next team would have met the standard of care
The most transferable holding concerns Ediger v Johnston, 2013 SCC 18, [2013] 2 S.C.R. 98. Dr. Cavanagh argued that relying on Ediger to assume the London obstetrical team would have taken the steps needed to avoid sepsis, without direct evidence from that hospital, overturned decades of jurisprudence. Gomery J.A. found nothing inappropriate in the trial judge’s assumption, consistent with Ediger, that “the obstetrics team would have provided treatment consistent with the standard of care.”
That sentence will be quoted, and it should be read carefully. It is not a licence to assume the counterfactual into existence. The assumption did work only because the content of the counterfactual standard of care had itself been proved: the only obstetrical expert at trial established what that standard required, which was counselling about the life-threatening risks of continuing a likely non-viable pregnancy and, once termination was elected, an evacuation arranged on an emergency basis. Take away the specialty expert and there is nothing for the assumption to attach to. The reasoning is closer to the counterfactual analysis in Gumbley v Vasiliou than to anything that dispenses with proof.
The adverse inference arguments turned against the defence
Grounds three and four both concerned adverse inferences, and both failed. The defence had asked the trial judge to infer that the evacuation would not have happened in time because the plaintiffs never asked Dr. Di Cecco how quickly it could be done, or whether Ms. Ibrahimova had an intrauterine infection on May 5. The Court of Appeal held there was no reason to put the first question to an obstetrical expert, who was not an infectious diseases expert, and no reason to put the second in chief, since it mattered only to the defence theory of causation. On the timing question, the trial judge had addressed the request directly, applied the principles in R. v. Ellis, 2013 ONCA 9, 113 O.R. (3d) 641, and declined the inference because there was other evidence on the point.
The fourth ground, that the trial judge had reversed the onus of proof, foundered on the same terrain. He had observed that a witness from the London hospital would not be favourably disposed to the plaintiffs or under their control, which explained why they had not called one, while defence counsel had ready access to that hospital’s physicians and its practices. If there had been a knockout blow available in the form of direct evidence that the procedure could not have been performed on May 5, he expected the defence would have delivered it rather than arguing about inferences. The Court of Appeal read that as an observation about an evidentiary vacuum, not a shifted burden.
The practical lesson is uncomfortable but clear. Where the counterfactual runs through an institution the defence can access and the plaintiff cannot, silence is not neutral.
What the decision settles, and what it does not
Three things are now considerably harder to argue in Ontario. A plaintiff does not need institutional witnesses to prove a counterfactual treatment path, if a specialist from the relevant discipline can describe what the standard of care required. What actually happened when the patient finally did reach the receiving hospital is admissible evidence of what that hospital could have done earlier. And a defendant who declines to call expert evidence in the specialty through which the counterfactual runs is unlikely to succeed on appeal by characterizing the resulting findings as speculation.
The defence position was stronger than the outcome suggests, and it is worth stating at its strongest. The finding that Ms. Ibrahimova would have elected immediate termination on May 4 carries a great deal of weight, and it rests on retrospective evidence from the plaintiffs themselves about a hypothetical decision, given by witnesses the trial judge found to have gaps in recollection and inaccuracies on other points. Dr. Di Cecco acknowledged that many patients with pPPROM continue their pregnancies. The corroboration was real but modest: she had sought and followed medical advice throughout the pregnancy, and the couple had pursued prenatal screening. The trial judge cross-checked the finding against what a reasonable person in her circumstances would have done, as Arndt v Smith, [1997] 2 S.C.R. 539 requires, weighing her age, educational goals, uncertain immigration status and limited financial resources. That is a legitimate analysis, but it is also an invitation. Nothing in these reasons prevents a defendant in the next case from leading evidence about what patients in the plaintiff’s position actually elect to do. In this case, no one did.
The standard of review carried much of the rest. Causation findings that rest on weighing a large body of complex expert evidence attract deference absent an error of law or principle or a palpable and overriding error of fact: Benhaim v St-Germain, 2016 SCC 48, [2016] 2 S.C.R. 352, at para. 37, and Hacopian-Armen Estate v Mahmoud, 2021 ONCA 545, 77 C.C.L.T. (4th) 206. That deference is symmetrical, as the Court of Appeal recently illustrated in Brown v Meaney, 2026 ONCA 445, which supplied the framing for the disposition here: the plaintiffs’ burden was to show on a common-sense basis that the injuries would not have occurred but for the negligence, and the trial judge found it met. The appeal was dismissed with costs of $60,000, the amount the parties had agreed on.
None of this changes what has to be proved in a case like this one. The standard of care in any medical negligence claim is a matter of expert evidence, and so is the clinical sequence by which an untreated rupture of membranes becomes sepsis. What Ibrahimova clarifies is that once a plaintiff has proved what the receiving specialists should have done, the court is not required to pretend it cannot infer that they would have done it. For patients whose injuries follow a missed or delayed diagnosis of sepsis, that is a meaningful narrowing of a defence that has often been available.
Ibrahimova v Cavanagh, 2026 ONCA 573, on appeal from Ibrahimova v Cavanagh, 2025 ONSC 4808.
Ibrahimova v Cavanagh: The Court of Appeal Upholds a Causation Finding Built on Inference
The Court of Appeal for Ontario dismisses Dr. Cavanagh's causation appeal in Ibrahimova v Cavanagh, upholding findings about what would have happened at a tertiary hospital the patient never reached on the relevant day. The decision confirms that a plaintiff can prove a counterfactual treatment path through a specialist from the receiving discipline, without institutional witnesses, and that a defendant who calls no expert in that specialty is unlikely to recast the resulting findings as speculation.
In a delayed diagnosis case, the breach is often the easy part. The hard part is proving what would have happened instead: at a hospital the patient never reached, on a day that never came, in the hands of physicians who never saw her. The defendant did not create that evidentiary gap in any conventional sense, but her negligence is the reason it exists. How much inference is a trial judge allowed to do to close it?
Ibrahimova v Cavanagh, 2026 ONCA 573, released on August 7, 2026, is the Court of Appeal for Ontario’s most direct answer to that question in some time. Dr. Angela Cavanagh did not challenge the finding that she was negligent. Her appeal was entirely about causation, and her central complaint was that the trial judge had filled the gaps with speculation dressed up as common sense. Gomery J.A., writing for a panel that included Copeland and Monahan JJ.A., dismissed the appeal. I wrote about Justice Sproat’s trial decision when it was released, and the appeal leaves it intact.
Four emergency visits in four days
Tamara Ibrahimova was 28 years old and 17 weeks into her first pregnancy when she went to the emergency department at the Kincardine site of South Bruce Grey Health Centre on Friday, May 3, 2019, reporting vaginal bleeding, lower abdominal cramps and a pinkish discharge. Dr. Cavanagh, a family physician who occasionally worked in the emergency department, reviewed her ultrasound and bloodwork, examined her, consulted an obstetrician by telephone, diagnosed a threatened miscarriage and told her to follow up with her obstetrical care provider.
She returned about 12 hours later. She had woken to a gush of fluid, and she showed Dr. Cavanagh a pad with straw coloured fluid on it. Dr. Cavanagh’s own evidence was that she thought this was likely amniotic fluid. She did not order a white blood cell count or an ultrasound, and she did not consult or urgently refer to an obstetrician. She recorded a final diagnosis of threatened abortion, advised watch and wait, and sent Ms. Ibrahimova home.
Two more physicians saw her over the next two days. Dr. Sivanesan Kalaichandran, at Alexandra Marine & General Hospital on May 5, recorded a heart rate of 100 and a self-reported low-grade fever, found white blood cells in her urine, cancelled a complete blood count he had ordered, and made a semi-urgent obstetrical referral. Dr. David Ponesse, at the Walkerton site on May 6, diagnosed a second trimester bleed and told her to come back the next morning. Her husband, Oleksii Gubenko, called an ambulance at 2:08 a.m. on May 7, roughly eight hours after that discharge. She was in septic shock. She was airlifted to a London hospital, where the pregnancy was evacuated at 9:30 a.m.
The consequences were catastrophic: partial amputation of her right foot, amputation of her left leg below the knee, kidney failure, a stroke, seizures and compromised use of her right arm. She was in hospital from May 7, 2019 until March 9, 2020, and on home dialysis for more than 17 months before a kidney transplant in July 2021. The trial judge described her prognosis as dire.
After a 28 day trial, Justice Sproat found that Dr. Cavanagh breached the standard of care on May 4 and that the breach caused the injuries. He also found Drs. Kalaichandran and Ponesse negligent, but dismissed the action against them because Ms. Ibrahimova would have developed severe infection and sepsis even absent their negligence. That allocation is worth pausing on. On Dr. David Talan’s evidence, accepted at trial, a dilation and evacuation had to occur by 2:00 a.m. on May 6 to avert septic shock. The May 5 and May 6 encounters fell outside the window, so a narrow causal window concentrated the entire loss on the earliest breach. Whether that reasoning holds up is not something this appeal explores, because only Dr. Cavanagh appealed.
Where inference ends and speculation begins
The trial judge used the delayed diagnosis framework from Sacks v Ross, 2017 ONCA 773, 417 D.L.R. (4th) 387, reaffirmed in Hasan v Trillium Health Centre (Mississauga), 2024 ONCA 586, 499 D.L.R. (4th) 178: establish what actually happened, then determine what would have happened had the standard of care been met. I have written before about the way Hasan handles an evidentiary gap the defendant’s own negligence produced, and the same tension runs through this case.
He also cited Goodman v Viljoen, 2012 ONCA 896, 299 O.A.C. 257, at para. 76, which cuts in both directions. It authorizes a robust and pragmatic approach to causation, decided on the balance of probabilities rather than to scientific certainty. It also warns that resorting to speculation, or to common sense on a question that requires expert knowledge, is a misuse of that approach rather than an application of it.
Dr. Cavanagh’s four grounds of appeal, distilled from nine alleged errors, were essentially variations on the theme that the trial judge had crossed that line. The most substantial was the first: that he made four findings about what would have happened at the London hospital without evidence capable of grounding them. There was no witness from the hospital, no maternal-foetal medicine specialist, no neonatologist, and no evidence about specialist availability or how operating room time was allocated there in May 2019.
The Court of Appeal held that the findings were open to the trial judge on the evidence he had, and that the evidence he had was of two kinds: the opinion of an obstetrician who actually worked at the receiving hospital, and the documented sequence of what happened when Ms. Ibrahimova did arrive there on May 7.
Dr. Robert Di Cecco, an obstetrician and gynaecologist at the London hospital, testified that a patient with a previable, prelabour, preterm rupture of membranes (pPPROM) would be told that the pregnancy was unlikely to reach viability, that a surviving foetus would likely have significant complications, and that continuing carried a risk of life-threatening sepsis. He testified that infection is the primary risk in pPPROM, which is why it is a medical emergency, and that any evidence of infection would mean antibiotics and no realistic option but to terminate. The defence called two experts on liability, in emergency medicine and infectious diseases. It called no obstetrician. Dr. Di Cecco’s evidence on the obstetrical counterfactual was therefore uncontradicted, and the trial judge had ruled that the defence emergency medicine expert could not testify about what obstetricians would do.
That absence did most of the work on appeal. The Court of Appeal treated the argument that an obstetrical team would have overlooked classic signs of possible infection, or would have needed an infectious diseases specialist to confirm them, as untenable. On the timing of the procedure, the trial judge had drawn an inference that a large tertiary hospital keeps specialists available around the clock and prioritizes emergent procedures over deferrable ones, and had found that inference corroborated by what happened on May 7, when the evacuation was performed within two hours of arrival. The Court of Appeal saw no basis to interfere.
There is also a fallback in the reasons that deserves attention, because it may matter more than the headline holdings. Even if the two hour inference had been wrong, it would not have been an overriding error. Dr. Talan’s deadline of 2:00 a.m. on May 6 left roughly 36 hours from the counterfactual admission on the afternoon of May 4, so the case never actually depended on a two hour turnaround.
Assuming the next team would have met the standard of care
The most transferable holding concerns Ediger v Johnston, 2013 SCC 18, [2013] 2 S.C.R. 98. Dr. Cavanagh argued that relying on Ediger to assume the London obstetrical team would have taken the steps needed to avoid sepsis, without direct evidence from that hospital, overturned decades of jurisprudence. Gomery J.A. found nothing inappropriate in the trial judge’s assumption, consistent with Ediger, that “the obstetrics team would have provided treatment consistent with the standard of care.”
That sentence will be quoted, and it should be read carefully. It is not a licence to assume the counterfactual into existence. The assumption did work only because the content of the counterfactual standard of care had itself been proved: the only obstetrical expert at trial established what that standard required, which was counselling about the life-threatening risks of continuing a likely non-viable pregnancy and, once termination was elected, an evacuation arranged on an emergency basis. Take away the specialty expert and there is nothing for the assumption to attach to. The reasoning is closer to the counterfactual analysis in Gumbley v Vasiliou than to anything that dispenses with proof.
The adverse inference arguments turned against the defence
Grounds three and four both concerned adverse inferences, and both failed. The defence had asked the trial judge to infer that the evacuation would not have happened in time because the plaintiffs never asked Dr. Di Cecco how quickly it could be done, or whether Ms. Ibrahimova had an intrauterine infection on May 5. The Court of Appeal held there was no reason to put the first question to an obstetrical expert, who was not an infectious diseases expert, and no reason to put the second in chief, since it mattered only to the defence theory of causation. On the timing question, the trial judge had addressed the request directly, applied the principles in R. v. Ellis, 2013 ONCA 9, 113 O.R. (3d) 641, and declined the inference because there was other evidence on the point.
The fourth ground, that the trial judge had reversed the onus of proof, foundered on the same terrain. He had observed that a witness from the London hospital would not be favourably disposed to the plaintiffs or under their control, which explained why they had not called one, while defence counsel had ready access to that hospital’s physicians and its practices. If there had been a knockout blow available in the form of direct evidence that the procedure could not have been performed on May 5, he expected the defence would have delivered it rather than arguing about inferences. The Court of Appeal read that as an observation about an evidentiary vacuum, not a shifted burden.
The practical lesson is uncomfortable but clear. Where the counterfactual runs through an institution the defence can access and the plaintiff cannot, silence is not neutral.
What the decision settles, and what it does not
Three things are now considerably harder to argue in Ontario. A plaintiff does not need institutional witnesses to prove a counterfactual treatment path, if a specialist from the relevant discipline can describe what the standard of care required. What actually happened when the patient finally did reach the receiving hospital is admissible evidence of what that hospital could have done earlier. And a defendant who declines to call expert evidence in the specialty through which the counterfactual runs is unlikely to succeed on appeal by characterizing the resulting findings as speculation.
The defence position was stronger than the outcome suggests, and it is worth stating at its strongest. The finding that Ms. Ibrahimova would have elected immediate termination on May 4 carries a great deal of weight, and it rests on retrospective evidence from the plaintiffs themselves about a hypothetical decision, given by witnesses the trial judge found to have gaps in recollection and inaccuracies on other points. Dr. Di Cecco acknowledged that many patients with pPPROM continue their pregnancies. The corroboration was real but modest: she had sought and followed medical advice throughout the pregnancy, and the couple had pursued prenatal screening. The trial judge cross-checked the finding against what a reasonable person in her circumstances would have done, as Arndt v Smith, [1997] 2 S.C.R. 539 requires, weighing her age, educational goals, uncertain immigration status and limited financial resources. That is a legitimate analysis, but it is also an invitation. Nothing in these reasons prevents a defendant in the next case from leading evidence about what patients in the plaintiff’s position actually elect to do. In this case, no one did.
The standard of review carried much of the rest. Causation findings that rest on weighing a large body of complex expert evidence attract deference absent an error of law or principle or a palpable and overriding error of fact: Benhaim v St-Germain, 2016 SCC 48, [2016] 2 S.C.R. 352, at para. 37, and Hacopian-Armen Estate v Mahmoud, 2021 ONCA 545, 77 C.C.L.T. (4th) 206. That deference is symmetrical, as the Court of Appeal recently illustrated in Brown v Meaney, 2026 ONCA 445, which supplied the framing for the disposition here: the plaintiffs’ burden was to show on a common-sense basis that the injuries would not have occurred but for the negligence, and the trial judge found it met. The appeal was dismissed with costs of $60,000, the amount the parties had agreed on.
None of this changes what has to be proved in a case like this one. The standard of care in any medical negligence claim is a matter of expert evidence, and so is the clinical sequence by which an untreated rupture of membranes becomes sepsis. What Ibrahimova clarifies is that once a plaintiff has proved what the receiving specialists should have done, the court is not required to pretend it cannot infer that they would have done it. For patients whose injuries follow a missed or delayed diagnosis of sepsis, that is a meaningful narrowing of a defence that has often been available.
Ibrahimova v Cavanagh, 2026 ONCA 573, on appeal from Ibrahimova v Cavanagh, 2025 ONSC 4808.
Paul Cahill
Partner, Davidson Cahill Morrison LLP | LSO Certified Specialist in Civil Litigation
Paul represents victims of medical malpractice across Ontario, with trial experience including a $11.5M jury verdict in a birth injury case. He is recognized in Best Lawyers in Canada and serves as trial counsel to other lawyers on complex medical negligence matters.
About PaulMore on medical malpractice in Ontario.
Other articles by Paul exploring the conditions, decisions, and systems behind preventable medical harm.
Finding the Right Expert in a Medical Malpractice Case
Why expert selection drives outcomes in Ontario medical malpractice cases. Paul Cahill on finding the right expert from an OTLA webinar.
Common Misunderstandings About Medical Malpractice in Ontario
Some of the most common beliefs about medical malpractice in Ontario are simply wrong, and they cut both ways: pushing some people toward hopeless claims and others away from good ones. Here are the misunderstandings I see most often, set against how the law actually works, from what counts as negligence to limitation periods, College complaints, causation, and the cap on damages.
Bloody Stools, a Family History of Colon Cancer, and a Three-Year Diagnostic Delay
A settlement on behalf of a woman in her 60s whose family physician dismissed three years of GI symptoms despite a known family history of colon cancer.