Texas Appeals Court Overturns 'Shaken Baby' Conviction Ahead of Execution Date in Another Disputed Case

The court found scientific opinion about "shaken baby syndrome" has changed, and a man sentenced to 35 years in prison deserves a new trial.

As the state of Texas prepares for the first execution in the country based on evidence of what used to be called "shaken baby syndrome," a state appeals court has overturned a conviction in a separate, similar case.

The Texas Criminal Court of Appeals ruled on Wednesday that Andrew Roark, a Dallas man who was sentenced to 35 years in prison in 2000 in a shaken baby case, deserves a new trial because it's unlikely that a jury would convict him today if they heard current scientific testimony. Two of the state's expert witnesses at Roark's trial later recanted portions of their testimony about the unlikelihood of accidental or natural causes for the symptoms they observed.

"We find that scientific knowledge has evolved regarding SBS and its application in [Roark's] case," the court wrote. "Additionally, we find that given further study, the experts would have given a different opinion on several issues at a trial today—some already have. The admissible scientific testimony at trial today would likely yield an acquittal."

There is likely no one else in the country as interested in the case, besides Roark, than Robert Roberson, a Texas death row inmate scheduled to be executed on October 17. Roberson would be the first person in the U.S. to be executed based on evidence of what used to be called "shaken baby syndrome," now officially called abusive head trauma (AHT). 

Roberson's lawyers filed another appeal this week asking the Texas Criminal Court of Appeals to reconsider its previous decisions denying him relief in light of Roark's case, especially since both trials featured testimony from the same expert witness, Janet Squires—a licensed expert in child abuse pediatrics who has testified in hundreds of child-abuse trials.

"The same prosecution expert testified in both trials making many identical pronouncements about how shaking had to be the principal explanation for the child's brain condition, when science has since demonstrated that many things, including natural disease progression and accidental short falls with head impact, can cause the same conditions," Gretchen Sween, Roberson's attorney, said in a press release. "The flaws in the expert testimony are nearly identical."

As Reason detailed in a feature on Roberson's case in August, he was convicted in 2003 of murdering his two-year-old daughter Nikki and sentenced to death. He claimed that his daughter fell out of bed in the middle of the night, but he calmed her down and eventually got her back to sleep. He says he later woke up and discovered that she had stopped breathing and turned blue. Roberson took his daughter to the hospital, where she was declared brain dead and eventually taken off life support. Medical examiners found a trio of internal head conditions that were, at the time, considered to be conclusive evidence of violent shaking and impacts.

Roberson's lawyers argue that AHT amounts to junk science and that the scientific consensus around it has shifted dramatically since Roberson's conviction. His lawyers also argued in a petition for a stay of execution, filed this summer, that they had uncovered previously undisclosed medical records that proved Roberson's daughter died of severe viral and bacterial pneumonia, not trauma from shaking or abuse.

Among those who now believe that Roberson is innocent are the former detective who arrested him as well as novelist John Grisham, who wrote a feature on Roberson's case for D Magazine last month.

However, the Texas Court of Criminal Appeals rejected his last appeal. It found that Roberson's claims did not meet the standards for filing a subsequent petition for writ of habeas corpus—Roberson was previously granted a stay of execution and additional evidentiary hearings in 2016 but denied relief—and the court dismissed Roberson's new petition without reviewing it on the merits.

The State of Texas argues in a brief opposing Roberson's new motion to reconsider that Roark's case is "easily distinguishable," despite featuring testimony from the same expert witness. Roberson's conviction, the state argues, rested on findings of shaking and multiple blunt force impacts, in addition to witness testimony that he was short-tempered and violent.

"The experts—supported by the autopsy report—established that Nikki died because of the blunt force injuries [Roberson] inflicted upon her, combined with shaking," the state's brief says. "None of these experts' opinions have been recanted, in part or in whole."

However, the court opinion overturning Roark's conviction notes that the "state theorized that Applicant caused the infant victim serious bodily injury by violently shaking her and possibly striking her with or against something, respectively referred to as Shaken Baby Syndrome and Shaken Impact Syndrome." 

Squires testified in Roark's case that it was possible the victim "had a shake and an impact."

Roberson's execution would come amid intense courtroom debate over the reliability of AHT evidence. Prosecutors and pediatric abuse specialists say there's broad scientific consensus around AHT, but innocence groups and advocates for forensic reform have successfully persuaded several state courts otherwise.

According to the National Registry of Exonerations, at least 34 parents and caregivers in 18 states convicted based on AHT evidence have been exonerated.