*606This case raises the question whether, when a driving under the influence (DUI) suspect is injured in an accident and taken to the hospital, the arresting officer needs to obtain a warrant before arranging for the suspect's blood to be drawn for blood alcohol content (BAC) testing. The trial court answered this question in the negative on the facts of this case, denying a motion to suppress BAC evidence. We think the trial court was too quick to find exigent circumstances here, and in the published portion of our opinion conclude that the blood draw was inconsistent with the Fourth Amendment. However, we find the error in this case to have been harmless, and in the unpublished portion of the opinion also reject appellant Matthew Enrique Meza's other challenges. We affirm Meza's convictions for driving under the influence causing injury, and for driving with a BAC of 0.08 percent or more causing injury, in violation of Vehicle Code section 23153, subdivisions (a) and (b).
FACTUAL AND PROCEDURAL BACKGROUND
On the afternoon of September 1, 2013, Meza called his girlfriend and told her he would finish his beer and his golf game and come see her. An hour or more later, about 5:30 p.m., he arrived at her house in Lafayette, driving his own car. After half an hour, they left in her car. Meza was driving. She asked him to drive because she was feeling unwell, and after initially declining he agreed. She saw no symptoms that he was under the influence, and he consumed no alcohol in her presence that evening.
Attempting to keep up with a fast car in the lane next to them, Meza began to speed. At one point where the posted speed limit was 45 miles per hour, he was driving at least twice the speed limit. His girlfriend told him to slow down. As he applied the brakes, the car began to fishtail and he lost control. The car catapulted across the median and on-coming traffic, and fell down an embankment. It was about 6:28 p.m. An off-duty California Highway Patrol sergeant who saw the crash went to check for survivors, and saw Meza emerge from the driver's side of the car.
Concord police officer David Kasid arrived and, with Meza's help, extracted the injured passenger, dragging her to safety ahead of a grass fire that had started under the car. Officer Kasid noticed a slight odor of alcohol on Meza, information that he relayed to Concord police officer Danielle Cruz when additional officers arrived on scene. Officer Kasid then continued with the accident investigation, while *897Officer Cruz pursued the intoxication investigation. A third police officer helped with traffic control, and a corporal in the Concord Police Department came to assess the scene. *607Officer Cruz had a brief conversation with Meza while Meza was waiting to be treated by emergency medical personnel. She asked him what had happened, and smelled "a moderate odor of alcoholic beverage coming from his mouth" as he responded. She noticed blood-shot and watery eyes, which she knew could be a symptom of alcohol consumption, or potentially a head injury. Because Meza was complaining of neck and back pain, Officer Cruz did not ask him to participate in field sobriety tests. She concluded, based on the evidence she had, that he should be arrested for driving under the influence. When an ambulance took Meza and his passenger to the John Muir Medical Center in Walnut Creek, she followed.
At the hospital, Meza had his blood drawn twice. Brenda Leatham, the hospital's lab operations manager, testified that at 7:08 p.m. the hospital drew blood, as they do for all trauma patients, so that doctors will know how to treat patients appropriately. Half an hour later, the hospital had test results. They measured Meza's BAC at 0.148 percent. Then at 8:25 p.m., a phlebotomist summoned by Officer Cruz drew Meza's blood. Joaquin Jimenez, a forensic alcohol expert from the county crime lab, tested this blood sample and measured its BAC at 0.11 percent.
Officer Cruz never attempted to get a warrant before directing the phlebotomist to draw Meza's blood for forensic purposes. She was aware of the United States Supreme Court's decision in Missouri v. McNeely (2013) 569 U.S. 141, 133 S.Ct. 1552, 185 L.Ed.2d 696 ( McNeely ), which held that an officer could not dispense with getting a warrant before having a motorist's blood drawn on grounds of exigent circumstances if the only exigency was that alcohol in the driver's body would dissipate with time. Shortly after this decision came down and before she investigated this case, Officer Cruz attended a training session where she learned how to obtain a warrant on short notice from an on-call judge. She learned she could fill out a simple affidavit (a template of which is posted at the jail), call the Sheriff's Department to contact the judge on duty, and then fax the warrant application to the judge for review. But this is a procedure Officer Cruz thought she needed only "if we have somebody that refuses to do a blood draw," and Meza did not refuse to have his blood drawn. At the hospital, Officer Cruz told Meza that because she had arrested him for driving under the influence he was subject to a blood draw, and that a phlebotomist was on his way. Meza responded "Okay," and allowed the phlebotomist to draw his blood.
The district attorney initiated criminal proceedings and, after a preliminary examination, filed a felony information against Meza. At the preliminary examination, the court heard evidence and ruled on Meza's motion to suppress the test results from the second, police-initiated blood draw. The trial court rejected the prosecutor's argument that Meza's actual or implied *608consent authorized the warrantless blood draw, but nonetheless denied Meza's motion. Citing Schmerber v. California (1966) 384 U.S. 757, 86 S.Ct. 1826, 16 L.Ed.2d 908 ( Schmerber ), the court found that exigent circumstances associated with the need to get medical care for Meza and his passenger justified having Meza's blood drawn without a warrant.
Jury trial began on October 26, 2015. The People presented the testimony of officers Kasid and Cruz, of laboratory scientists Leathem and Jimenez, and of Meza's passenger and other percipient witnesses *898to his driving. The defense cross-examined witnesses but presented no affirmative case.
The trial testimony of the laboratory scientists delved into several technical subjects. Leatham explained the procedures followed in drawing and testing Meza's blood to ensure accuracy in the hospital's blood test results, and Jimenez of the crime laboratory testified similarly regarding the forensic blood draw. Leatham testified that, although she did not know the extent to which the hospital's procedures complied with Title 17 regulations for forensic alcohol analysis, the hospital's procedures were sufficiently rigorous that doctors relied on the test results in making treatment decisions.
Jimenez testified to two factors that explain why the two BAC test results differed. First, the county crime laboratory tests whole blood, meaning a blood sample that includes both blood plasma and other cellular material, whereas medical testing is usually (and was in this case) performed on the blood plasma alone. When other cellular material is excluded, the test results from blood plasma usually yield slightly higher numbers, compared to testing from whole blood. Jimenez explained that a test performed from blood plasma produces a number that is higher by a ratio of about 1.1 to 1, or 1.3 to 1, relative to a test from whole blood, with the precise ratio depending on the person whose blood is being tested. Based on what he considered a standard ratio, 1.15 to 1, Jimenez converted Meza's 0.148 hospital test result from blood plasma to 0.128 percent, if the test had been from whole blood as is required for forensic alcohol testing.
Second, Jimenez explained that the human body eliminates alcohol over time, so one would expect the forensic test of Meza's BAC, from a sample taken an hour-and-a-quarter after the hospital's blood sample, to produce a lower reading. Calculating backwards from the 0.11 percent result obtained from Meza's blood drawn at 8:25 p.m., Jimenez estimated Meza's BAC at the time of the accident two hours earlier. He made some assumptions in performing this calculation: that Meza had consumed no alcohol since arriving at his girlfriend's home at 5:30 p.m., and that Meza's body eliminated alcohol at the average rate of 0.015 percent per hour. Based on these *609assumptions, he calculated Meza's BAC at the time of the accident as 0.14 percent, with a margin of error of 0.01 percent. This value is well above the legal limit of 0.08 percent.
Jimenez also testified that whenever an individual's BAC is at or above 0.08 percent, he or she is too impaired to drive safely. He testified that this is not only his personal opinion, but a consensus in the scientific literature that has been endorsed by the American Medical Association, the National Highway Safety Administration, and the National Safety Council.
On November 2, 2015, the jury convicted Meza on both felony counts. In count 1, he was convicted of driving under the influence causing injury. ( Veh. Code, § 23153, subd. (a).) In count 2, he was convicted of driving with a BAC of 0.08 percent or more causing injury. ( Id. , § 23153, subd. (b).) As to both counts, the jury found true the allegation that he personally inflicted great bodily injury upon the victim (the passenger in his car), and the court subsequently found true the allegation that he had two prior convictions for driving under the influence. (See Pen. Code, § 12022.7, subd. (a) ; id. , § 23566.)
The court sentenced Meza to six years in state prison, and Meza timely appealed.
DISCUSSION
I. Motion to Suppress Blood Test Results
In reviewing the trial court's denial of Meza's motion to suppress, we defer *899to the trial court's factual findings where they are supported by substantial evidence, but " ' "exercise our independent judgment in determining the legality of a search on the facts so found." ' " ( People v. Tully (2012) 54 Cal.4th 952, 979, 145 Cal.Rptr.3d 146, 282 P.3d 173.) Here, the trial court found exigent circumstances justifying the police-initiated blood draw because Officer Cruz "had limited information" as to whether Meza was under the influence,1 and had to choose between taking him into custody and going to the hospital, on the one hand, or going to the police station to get a warrant, on the other. We think the totality of the circumstances in this case requires a different result, and that the blood draw Officer Cruz ordered was a violation of Meza's Fourth Amendment rights. The error, however, was harmless under the standards of Chapman v. California (1967) 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 ( Chapman ).
Under the Fourth Amendment to the United States Constitution, a warrantless search is per se unreasonable unless the People prove that the *610search comes within a recognized exception to the warrant requirement. ( People v. Laiwa (1983) 34 Cal.3d 711, 725, 195 Cal.Rptr. 503, 669 P.2d 1278.) A blood draw is a search under the Fourth Amendment. ( Birchfield v. North Dakota (2016) --- U.S. ----, 136 S.Ct. 2160, 2173, 195 L.Ed.2d 560 ( Birchfield ).) Here, the People argue and the trial court found that the search was authorized by exigent circumstances, which is a recognized exception. (See Mincey v. Arizona (1978) 437 U.S. 385, 393-394, 98 S.Ct. 2408, 57 L.Ed.2d 290.) "To determine whether a law enforcement officer faced an emergency that justified acting without a warrant," we look "to the totality of the circumstances." ( McNeely , supra , 569 U.S. at p. 149, 133 S.Ct. 1552.)
On grounds of exigent circumstances, the United States Supreme Court more than 50 years ago approved of a warrantless blood draw at the direction of a police officer investigating a driving under the influence case. ( Schmerber , supra , 384 U.S. 757, 86 S.Ct. 1826.) Schmerber had been taken to a hospital after being injured in an automobile accident, and at the hospital he refused to consent to having his blood drawn. ( Id . at pp. 758-759, 86 S.Ct. 1826.) The court held that the blood draw was permissible because the officer "might reasonably have believed that he was confronted with an emergency, in which the delay necessary to obtain a warrant, under the circumstances, threatened 'the destruction of evidence.' ... Particularly in a case such as this, where time had to be taken to bring the accused to a hospital and to investigate the scene of the accident, there was no time to seek out a magistrate and secure a warrant." ( Id . at pp. 770-771, 86 S.Ct. 1826.) But the court noted that its "judgment that there had been no Fourth Amendment violation was strictly based 'on the facts of the present record.' " ( McNeely , supra , 569 U.S. at p. 152, 133 S.Ct. 1552 (quoting Schmerber , at p. 772, 86 S.Ct. 1826 ).)
Four months before the accident in this case, the United States Supreme Court revisited Schmerber , in another DUI case where the prosecution argued exigent circumstances. ( McNeely , supra , 569 U.S. at p. 145, 133 S.Ct. 1552.) McNeely's was a fairly standard DUI arrest, one not involving an automobile accident. The prosecution argued in McNeely for a per se rule that would have allowed law enforcement to have blood drawn from anyone arrested for DUI, on the theory that "the natural metabolization of alcohol in the bloodstream presents a per se exigency." ( Ibid . ) In McNeely , the court left Schmerber in *900place but declined to adopt such a per se rule. One reason was its concern that the State's approach "fails to account for advances in the 47 years since Schmerber was decided that allow for the more expeditious processing of warrant applications, particularly in contexts like drunk-driving investigations where the evidence offered to establish probable cause is simple." ( McNeely , at p. 154, 133 S.Ct. 1552.) In weighing the totality of circumstances, courts must consider such technological advances and not be quick to give up "the neutral magistrate judge's essential role as a check on police discretion." ( Id . at p. 155, 133 S.Ct. 1552.) The court also considered significant that BAC evidence disappears *611only "gradually and relatively predictably." ( Ibid . ) Each case must be decided on its facts, but McNeely articulates the constitutional standard: "In those drunk-driving investigations where police officers can reasonably obtain a warrant before a blood sample can be drawn without significantly undermining the efficacy of the search, the Fourth Amendment mandates that they do so." ( Id. at p. 152, 133 S.Ct. 1552.)
Because Meza's case and Schmerber both involved drivers injured in an accident and taken to the hospital for medical care, the trial court saw strong parallels between the two cases. We think the court overlooked dispositive differences. First, while the decision in Schmerber mentions only one police officer involved in the arrest and investigation of that case, in this case there were at least four officers who responded to the accident scene. (See Schmerber , supra , 384 U.S. at pp. 769-771, 86 S.Ct. 1826.) The record contains no information explaining why one or more of the other officers could not have assisted Officer Cruz, if interacting with Meza and his passenger at the hospital required too much of her attention to allow her also to obtain a warrant before Meza's blood was drawn. Second, as the high court emphasized in McNeely , advances in technology in the half century since Schmerber was decided have made it easier for law enforcement officers to get warrant applications before a judge when court is not in session. Nothing in the record explains why Officer Cruz could not have used the procedures she had been trained on to get a warrant application approved by the on-call judge in the two hours after Meza and his passenger went down the embankment.
On the record before us, the People have not established that exigent circumstances prevented the Concord Police Department from obtaining a warrant before having Meza's blood drawn. Officer Cruz did not even try to get a warrant, presumably because she thought Meza's acquiescence constituted consent.2 Instead of applying for a warrant, Officer Cruz spent time at the accident scene, after Meza left in an ambulance, interviewing a witness. At the hospital, she was communicating with the family of Meza's passenger, filling out paperwork, and engaging in casual conversation with Meza as he awaited medical care. The People offered no evidence to explain *901why she *612could not have sought a warrant during any of that time. Her activities are ones we expect her colleagues could have undertaken, or she could have put off until later, so that she had time to prepare an affidavit and use a fax machine at the hospital to submit a warrant application. But if this was not possible, then Officer Cruz could and should have enlisted assistance from another police officer. The forensic blood draw in this case did not occur until two hours after the accident, so if Officer Cruz and her colleagues had used diligent efforts to prepare and submit a brief warrant application, we have every reason to believe that they would have procured a warrant without causing any delay in the blood draw. (Cf. People v. Toure (2015) 232 Cal.App.4th 1096, 1104-1105, 181 Cal.Rptr.3d 857 ( Toure ) [violent defendant requiring physical restraints, and delays in obtaining warrants after court closings].) Had the officers tried to get a timely warrant but failed, then the People would have evidence to explain that failure, and our assessment of the totality of the circumstances would necessarily be different.
If the court were to conclude on this record that exigent circumstances excuse law enforcement from getting a warrant, it would be hard to imagine a case requiring a warrant for a blood draw when a DUI suspect is taken to the hospital. We would be creating, sub rosa , a rule that exempts accident cases from the totality-of-the-circumstances inquiry that McNeely requires, and we would be ignoring McNeely 's point about the availability of electronic warrants. ( McNeely , supra , 569 U.S. at pp. 154-156, 133 S.Ct. 1552.) This we decline to do. The People have not proven that the Concord Police Department could not obtain a warrant to have Meza's blood drawn for forensic purposes in a time frame that would not "undermin[e] the efficacy of the search." ( Id . at p. 156, 133 S.Ct. 1552.) We therefore hold that exigent circumstances do not excuse the police from getting a warrant in this case. The evidence of Meza's BAC from the 8:25 p.m. blood draw should have been suppressed because it was obtained in violation of his Fourth Amendment rights.
Where, as here, an error is of constitutional dimension, we excuse it as harmless only if we are persuaded beyond a reasonable doubt that it did not contribute to the guilty verdicts. ( Chapman , supra , 386 U.S. at p. 24, 87 S.Ct. 824 ; People v. Neal (2003) 31 Cal.4th 63, 86, 1 Cal.Rptr.3d 650, 72 P.3d 280.) In this case, because of the evidence from the hospital's blood draw, we find the error harmless even under this exacting standard. The hospital's blood test established that 40 minutes after the accident Meza's BAC was 0.128 percent, calculated as a proportion of his whole blood (down from 0.148 percent in blood plasma). This BAC is more than 50 percent above the legal limit of 0.08 percent. Leatham testified extensively about the hospital's procedures for drawing and testing blood in a manner designed to ensure accurate results. Although the hospital is not licensed for forensic testing, it is licensed by the Department of Public Health and accredited by the College of American Pathologists as a clinical laboratory, and treating physicians rely on the *613results of its tests to be accurate. Meza is correct that a jury could consider the hospital's failure to follow Title 17 standards for forensic blood testing as evidence tending to undermine the validity of the test results (see CALCRIM No. 2101 ), but in light of Leatham's testimony as to why the results were reliable and in the absence of any other reasons to mistrust their reliability, we conclude that no rational jury would have doubted that Meza's BAC at the time of the accident was at least 0.08 percent, even if the forensic blood test evidence had been suppressed. *902Meza's arguments about Title 17 would have more force if his BAC had been closer to the legal limit, but on this record we conclude the error was harmless beyond a reasonable doubt.
II. Other Challenges to the Convictions***
DISPOSITION
The judgment of the trial court is affirmed.
We concur:
Kline, P.J.
Richman, J.