In this appeal, we must decide if a conviction for the crime of operating a motor vehicle while having any amount of a controlled substance in a person as measured by the person's urine can be based on an initial laboratory test that was positive for controlled substances. We conclude an initial test is insufficient under the facts of this case to establish guilt beyond a reasonable doubt. We vacate the decision of the court of appeals, reverse the judgment and sentence of the district court, and remand the case to the district court for dismissal of the charge.

*825 I. Factual Background and Proceedings.

The facts of this case resulted in the prosecution and conviction of Jeffrey Myers for the crime broadly referred to as operating while intoxicated (OWI). On March 12, 2016, around 1 a.m., Myers was operating a motor vehicle in Charles City. A police officer, Cory Van Horn, observed the vehicle and noticed the taillights were not illuminated. Officer Van Horn stopped the vehicle. After he informed Myers of the reason for the stop, Myers flipped a switch in the car's interior, which illuminated the lights. Officer Van Horn also noticed Myers was sweating profusely.

Officer Van Horn placed Myers in the passenger seat of his patrol car. He checked Myers's eyes and noticed that they were watery and bloodshot and that he had difficulty keeping them open. Additionally, he noted that Myers's eyes dilated very little upon exposure to his flashlight and that the back of Myers's tongue was a brownish green color. Another officer arrived at the scene to assist Officer Van Horn with the remainder of the stop.

Officer Van Horn administered several field sobriety tests, including horizontal gaze nystagmus, lack of convergence, walk and turn, one leg stand, and the modified Romberg test. The test results prompted Officer Van Horn to ask Myers if he had "taken" anything that night. Myers replied he had taken cold medicine. The officers concluded Myers was under the influence of a drug and arrested him. Myers consented to the submission of a urine specimen for testing. An initial test of the urine sample by the Iowa Division of Criminal Investigation (DCI) laboratory revealed detectable levels of amphetamines and marijuana.

On March 30, 2016, the State charged Myers by trial information with OWI in violation of Iowa Code section 321J.2 (2016). 1 The minutes of testimony included the official toxicology report from the DCI laboratory. The positive screen for amphetamines was 589 ng/ml, and the positive screen for marijuana metabolites was 62 ng/ml. 2 The report stated the positive screens "indicate[ ] the possible presence" of substances at levels equal to or more than the levels established in the Iowa Administrative Code. The document concluded by indicating a report on the positive screens "to confirm the presence of specific drugs will follow." The minutes of testimony, however, did not include a follow-up report.

On June 6, Myers filed a motion to suppress. He argued his taillights were illuminated and there was no basis to justify the stop. At the suppression hearing, the State submitted a copy of Officer Van Horn's dash cam video recordings. The *826 district court denied the motion to suppress. It concluded the taillights were not illuminated and the stop was justified.

The case proceeded to a bench trial on the minutes of testimony. The district court found Myers guilty beyond a reasonable doubt. In making this finding on the record, the court explained,

All right. Mr. Myers, basically the State has two things that they have to prove in order to establish this offense. The first is that you were driving or operating a motor vehicle. ... That element has been established. The second element is at the time you were operating a motor vehicle, you had a detectable level of a controlled substance in your blood stream. They could also prove you were under the influence of something . In this case, you did agree to provide a urine sample. The urine sample was positive for both marijuana metabolites and for amphetamines; and so, those are the elements that the State has to establish, and I believe that the State has established those elements beyond a reasonable doubt.

It then entered a written finding that the minutes of testimony established beyond a reasonable doubt that Myers committed all the elements of OWI in violation of Iowa Code section 321J.2. It did not designate the specific subsection. The court imposed the mandatory minimum penalties, including two days in jail and a fine of $1250.

Myers appealed. He claimed (1) the district court erred by denying his motion to suppress because there was no probable cause to support the stop and (2) the evidence was insufficient to establish the presence of a controlled substance in his system. Specifically, he argued the initial screen test only found the "possible presence" of drugs not their actual presence. He argued a confirmatory test should have been done on his urine to verify the presence of controlled substances. Without a confirmatory test, he claims the evidence in the minutes of testimony was insufficient to support a finding of guilt.

In response, the State argued the results of the laboratory test included in the minutes of testimony were sufficient to support the conviction . Alternatively, it asserted the lab report did measure an amount of a controlled substance in the urine as required under the statute and, combined with other circumstantial evidence of impairment described in the minutes of testimony, constitutes sufficient evidence of guilt.

We transferred the case to the court of appeals. It found the district court properly denied Myers's suppression motion. It also found there was no legal requirement for a confirmatory test and concluded that the detectable amounts of controlled substances by the initial test provided sufficient evidence to support a conviction for OWI. Myers sought, and we granted, further review.

On further review, we only address the issue of whether or not the minutes of testimony in this case provided sufficient evidence to support a conviction for OWI. In re Marriage of Schenkelberg , 824 N.W.2d 481 , 483 (Iowa 2012) ("[W]e have the discretion to review all or part of the issues raised on appeal or in the application for further review."). We do not address the claim of error based on the denial of the motion to suppress.

II. Scope of Review.

We review a claim of insufficient evidence in a bench trial just as we do in a jury trial. State v. Weaver , 608 N.W.2d 797 , 803 (Iowa 2000). "If the verdict is supported by substantial evidence, we will affirm." Id. We determine whether substantial evidence supports the verdict by *827 reviewing "all the evidence and the record in the light most favorable to the trial court's decision." State v. Hearn , 797 N.W.2d 577 , 580 (Iowa 2011). Finally, our review of challenges to the sufficiency of evidence is for errors at law because "the question ... is simply whether the evidence was sufficient to support [the] conviction." State v. Petithory , 702 N.W.2d 854 , 856 (Iowa 2005).

III. Analysis.

A. District Court Findings. We first consider the context of the verdict and the finding of guilt by the district court. The trial information alleged the violation of Iowa Code sections 321J.2(1)( a ) and ( c ). The first alternative was based on the operation of a motor vehicle while under the influence of an alcoholic beverage or a drug. Iowa Code § 321J.2(1)( a ). The second alternative was based on the operation of a motor vehicle with any amount of a controlled substance as measured in the person's blood or urine. Id. § 321J.2(1)( c ). Yet, the written verdict entered by the district court only made reference to section 321J.2. It did not specify any particular subsection in finding Myers violated section 321J.2. The trial transcript, however, revealed the district court only found Myers guilty of violating section 321J.2(1)( c ). The district court parenthetically mentioned the State "could also prove" Myers was operating under the influence of a controlled substance but made no finding that Myers was "under the influence" to support guilt under section 321J.2(1)( a ). The only finding made was that Myers had "any amount" of a controlled substance as measured in his urine. 3 Thus, we only consider whether substantial evidence supported the verdict rendered.

Additionally, even if the pronouncement by the district court is considered a general verdict based on a crime with multiple bases for guilt, substantial evidence must support each alternative under the statute. See State v. Lukins , 846 N.W.2d 902 , 912 (Iowa 2014) ; State v. Smith , 739 N.W.2d 289 , 295 (Iowa 2007) ; State v. Heemstra , 721 N.W.2d 549 , 559 (Iowa 2006). Accordingly, in this case, if substantial evidence did not support guilt under Iowa Code section 321J.2(1)( c ), the conviction must be reversed. We therefore proceed to consider whether the minutes of testimony supported the finding that Myers operated a motor vehicle while any amount of a controlled substance was present in his person, as measured by his urine.

B. Overview of the Crime of Operating While Intoxicated. The crime of operating while intoxicated can be committed in three ways. First, the statute criminalizes operating a motor vehicle "[w]hile under the influence of an alcoholic beverage or other drug or a combination of such substances."

*828 Iowa Code § 321J.2(1)( a ). Second, the statute criminalizes operating a motor vehicle "[w]hile having an alcohol concentration of .08 or more." Id. § 321J.2(1)( b ). Third, the statute criminalizes operating a motor vehicle "[w]hile any amount of a controlled substance is present in the person, as measured by the person's blood or urine." Id. § 321J.2(1)( c ). Each prong uses a different theory and primarily relies on different evidence. The first prong primarily utilizes evidence of a person's conduct and demeanor. The second prong primarily utilizes evidence of the results of testing that measures a person's alcohol concentration level from a breath, blood, or urine specimen. The third prong primarily uses evidence of the results of testing that measures any amount of a controlled substance from a blood or urine specimen. While the last two prongs require evidence derived from a test, not conduct, the test under the third prong requires no specific threshold level of a prohibited substance.

C. Substantial Evidence of a Controlled Substance as Measured by a Urine Test. In this case, we only consider if substantial evidence supported the conviction under the third prong. The State claims the initial screening test conducted by the DCI laboratory constitutes sufficient evidence to support the conviction because it revealed the "possible presence" of a controlled substance or metabolites in Myers urine specimen in amounts that exceeded the standards for initial laboratory testing for controlled substances. See id. § 321J.2(12)( c ) (imposing a statutory requirement for the department of public safety to adopt "nationally accepted standards for determining detectable levels of controlled substances in the division of the criminal investigation's initial laboratory screening test for controlled substances"); Iowa Admin. Code r. 661-157.7 (adopting the federal guidelines for workplace testing in initial screenings). These standards established detectable levels for initial laboratory testing at fifty ng/ml for marijuana metabolites and 500 ng/ml for amphetamines. The initial test results in this case measured Myers's marijuana metabolites at sixty-two ng/ml and measured amphetamines at 589 ng/ml. Thus, we must first decide if an initial test alone is sufficient to establish guilt beyond a reasonable doubt when the laboratory who conducted the test indicates the results only reveal the "possible presence" of drugs.

D. Confirmatory Testing. Testing for controlled substances in urine specimens is generally recognized to consist of an initial test and a confirmatory test. 1 Kevin B. Zeese, Drug Testing Legal Manual § 2:2 (2d ed.), Westlaw (database updated June 2018) ("Drug testing, whether of blood, urine, or other body chemicals, should be considered to be at least a two-stage process. ... This initial test alone is generally insufficient as far as both the scientific and legal community are concerned."); Wis. State Crime Labs, Wis. Dep't of Justice, WSCL FAQs: Toxicology , https://www.doj.state.wi.us/section-faqs/wscl (last visited Mar. 1, 2019) ("An immunoassay screen does not test in enough detail for a drug to be identified or confirmed, so drugs or classes of drugs can only be indicated from this test."). The Iowa DCI described the process as follows:

The detection of drugs in a urine sample is determined by initial screening or presumptive tests. These tests target compounds in a drug group rather than specific drugs. Following a positive screening result on a sample, a second confirmatory test is performed. This second test uses a different analytical technique to identify a specific drug compound.

Div. of Criminal, Iowa Dep't of Pub. Safety, Urine Drug Analysis , http://www.dps.state.ia.us/DCI/lab/toxicology/Urine_Drug_Analysis.shtml *829 (last visited Mar. 1, 2019).

Iowa Code section 321J.2(1)( c ) does not require a specific drug to be identified. The statute only requires any measurable amount of "a controlled substance." However, the identification of a specific drug in the testing process serves to eliminate any errors in relying on the identification of known compounds of a drug group. Because most confirmatory testing "technique[s] provide[ ] information about the chemical structure of a substance, it is possible to definitively state the specific drug that is present." Wis. State Crime Labs, Wis. Dep't of Justice, WSCL FAQs: Toxicology , https://www.doj.state.wi.us/section-faqs/wscl (emphasis added) (last visited Mar. 1, 2019). Thus, confirmatory tests are seen as safeguards against the potential flaws associated with the initial drug test. See Karen E. Moeller et al., Clinical Interpretation of Urine Drug Tests: What Clinicians Need to Know About Urine Drug Screens , 92 Mayo Clinic Proceedings 774, 775 (2017) ("[I]mmunoassays will detect substances with similar characteristics [to drug metabolites or classes of drug metabolites], resulting in cross-reactivity leading to false-positive results.").

For employment drug testing purposes, urine samples are divided into two portions at the time of collection. Iowa Code § 730.5 (7)( b ). One portion is used for initial drug testing; a confirmatory test of this sample is required if the initial test reveals the presence of drugs. Id. § 730.5(7)( f )(1) ; see also Harrison v. Emp't Appeal Bd. , 659 N.W.2d 581 , 582 (Iowa 2003) (explaining this procedure with citation to a prior Iowa Code version). If the positive results are confirmed, an employee is then entitled to another confirmatory test of the second portion. Id. § 730.5(7)( i )(1) ; see also Harrison , 659 N.W.2d at 582 .

Confirmatory drug tests are also a staple in federal employment drug testing. The Mandatory Guidelines for Federal Workplace Drug Testing Programs meticulously detail the procedures required for federal employment drug testing. 73 Fed. Reg. 71,858 (Nov. 25, 2008). The regulations provide for both an initial drug test and confirmatory drug test. 4 The regulations establish specific measurement cutoff requirements for both initial drug tests and confirmatory drug tests and prescribe strict standards to reduce the risk of inaccurate test results. Id. at 71,858, 71 ,861 -62.

We too have recognized the existence of a confirmatory test in State v. Comried , 693 N.W.2d 773 , 774 (Iowa 2005). In Comried , the defendant was convicted of vehicular homicide while having a controlled substance in his system in violation of sections 321J.2(1)( c ) and 707.6A. Id. Comried challenged the "any amount of a controlled substance" language in section 321J.2(1)( c ). Id. at 774 . He argued we should apply the department of public safety rule that established cutoff levels for measurement of drug concentrations. Id. at 775 . The state argued, and we agreed, that " 'any amount' means what it says-if a test detects any amount of a controlled substance the any-amount element is satisfied." Id. at 775, 778 . Yet, we observed the distinction between initial and confirmatory tests:

[The state's criminalistics] laboratory performs two types of tests: an initial *830 screening test and a confirmatory test. The initial test is performed at certain "cutoff levels," meaning that only drug concentrations over the cutoff level will yield a "positive" test result. Any concentration below the cutoff level is reported "negative." If the initial screening test shows positive, a second test is performed on the sample. This second test, the confirmatory test , is presumably more expensive but is also more reliable and produces very accurate results.

Id. at 774 (emphasis added) (citation omitted). 5

This background provides context to the issue we confront and helps explain the force of two important statements contained in the written initial report by the DCI laboratory in this case. First, the report specifically stated the positive screens only indicated the "possible presence" of a controlled substance. Second, the report stated a second report to confirm the presence of specific drugs would follow.

We recognize the initial test is evidence of the presence of a controlled substance in the urine of a person. However, the lack of confidence in the results of the initial test has given rise to the common requirement for a confirmatory test in other areas of drug testing. If confirmatory testing is a part of workplace drug testing, it would be just as important, if not more important, in the criminal justice system. Significantly, the minutes of testimony in this case did not include any expert testimony or other evidence to explain the accuracy of the initial test beyond its admitted possibility of the presence of controlled substances. Thus, we are left with doubts about its accuracy, and those doubts mean the initial test falls short of establishing guilt beyond a reasonable doubt. See, e.g. , Hearn , 797 N.W.2d at 580 ("[E]vidence which merely raises suspicion, speculation, or conjecture is insufficient." (quoting State v. Casady , 491 N.W.2d 782 , 787 (Iowa 1992) ) ). We conclude the results of the initial testing of the urine specimen, alone, is insufficient to satisfy the burden of proof required of our criminal justice system. To support a conviction under the statute, the test must identify an amount of a controlled substance in a blood or urine sample beyond a reasonable doubt. Without other evidence, a test that only identifies the "possible presence" of a controlled substance falls short of satisfying the reasonable doubt standard.

E. Circumstantial Evidence of Impairment. Notwithstanding, the State argued that even if the test results only show a "possible presence" of a controlled substance, this evidence, together with the circumstantial evidence of impairment disclosed by the minutes of testimony in the case, is sufficient to support the conviction. It asserts that the initial test satisfied the requirement for the evidence of guilt be based on a test that measures a controlled substance in a person's urine. Additionally, it claims the observations of Myers at the *831 scene of the stop elevated the level of certainty in the test results to a level beyond reasonable doubt. This evidence included observations of his conduct and physical condition, as well as his performance on numerous field sobriety tests. Thus, the question is whether circumstantial evidence of impairment-bloodshot eyes, poor performance on field sobriety tests, drowsiness-provides substantial evidence to support a conviction under Iowa Code section 321J.2(1)( c ) in the absence of a confirmatory test.

The conduct and demeanor of a person are important considerations in determining whether a person is "under the influence" under section 321J.2(1)( a ). State v. Price , 692 N.W.2d 1 , 3 (Iowa 2005). Moreover, a witness is permitted to "state whether or not another was intoxicated at a particular time" and nonexperts may even "state how far another was affected by intoxication." State v. Davis , 196 N.W.2d 885 , 893 (Iowa 1972) (quoting State v. Cather , 121 Iowa 106 , 108, 96 N.W. 722 , 722 (1903) ). This evidence, in turn, can support a conviction. See State v. Truesdell , 679 N.W.2d 611 , 616 (Iowa 2004) (finding witnesses' and police officers' reports regarding defendant's erratic driving and behavior supported a finding that he was under the influence of alcohol when he operated his vehicle). Thus, evidence of impaired conduct and the demeanor of a person could help support a laboratory test indicating the presence of a controlled substance in that person. Yet, the only issue under section 321J.2(1)( c ) is whether a test shows the presence of a controlled substance, not conduct. Impaired conduct can be consistent with the presence of a controlled substance, but it can also result from a medical condition or other causes unrelated to the consumption of a controlled substance. Unlike a confirmatory test that validates a preliminary finding of a controlled substance, witness testimony of impairment does not serve to validate the presence of a controlled substance in a person, at least not without expert testimony that could eliminate causes for the conduct and demeanor other than the effects of a controlled substance or other evidence of drug consumption by the person sufficient to eliminate the reasonable doubt left by the preliminary test. Without this evidence, the reasonable doubt that emanates from the initial test of a "possible presence" of a controlled substance is not eliminated by the circumstantial evidence that a person is under the influence.

The reasonable doubt standard has a deep and important meaning within the American criminal justice system. 6 It is important this meaning always be observed in each case. In this case, it means the plain language of Iowa Code section 321J.2(1)( c ) cannot be satisfied by relying on the circumstantial evidence of impairment.

IV. Conclusion.

We conclude that the minutes of testimony were insufficient to establish that Myers violated Iowa Code section 321J.2(1)( c ) beyond a reasonable doubt. We reverse the judgment and sentence of the district court and remand the case to the district court to dismiss the charge.

REVERSED AND REMANDED.

Wiggins and Appel, JJ., join this opinion. Mansfield, Waterman, and Christensen, JJ., concur specially. McDonald, J., takes no part.

MANSFIELD, Justice (concurring specially).

*832 I join the court's well-reasoned opinion. I write separately to explain why nothing precludes the State from asking the district court to consider Jeffrey Myers's potential guilt under Iowa Code section 321J.2(1)( a ) (2016) on remand. As the court notes, the district court made no factual finding on that theory.

State v. Pexa is on point. See 574 N.W.2d 344 (Iowa 1998). There the state maintained that the defendant was guilty of unauthorized possession of an offensive weapon under two separate definitions of "offensive weapon." Id. at 345 . Following a bench trial, the district court found the defendant guilty under the Iowa Code section 724.1(3) (1995) alternative but did not consider the section 724.1(4) alternative. Id. On appeal, we found that the evidence did not support the section 724.1(3) alternative on which the defendant was convicted by the trial court. Id. at 346 . However, rather than acquitting the defendant entirely as he requested, we remanded the case for further proceedings so the court could consider and rule on the section 724.1(4) alternative. Id. at 346-47 . We explained, "A failure to consider an alternative definition of the offense charged does not constitute an acquittal of that offense for double jeopardy purposes." Id. at 347 .

The same analysis applies here. The district court clearly did not acquit Myers of the section 321J.2(1)( a ) alternative when it said, "They could also prove you [Myers] were under the influence of something." So that alternative remains fair game on remand.

The New Mexico Court of Appeals recently cited to Pexa in a case with facts quite similar to ours. See State v. Ben , 362 P.3d 180 , 183 (N.M. Ct. App. 2015). In Ben , the defendant was charged with driving while intoxicated (DWI) under New Mexico law. Id. at 181 . The criminal complaint alleged violations of both the "per se" and the "under the influence" subsections-alternatives that mirror the options available under Iowa law. Id. After a nonjury trial, the magistrate found the defendant guilty of per se DWI but "did not refer to the impaired DWI provision." Id. Following an appeal to the district court, a jury convicted the defendant of impaired DWI but found no violation of per se DWI. Id. The defendant appealed further, arguing that his district court conviction on the impaired theory of DWI violated double jeopardy. Id. As the New Mexico Court of Appeals explained,

Defendant divides the single offense of DWI into its alternative theories, contending that his conviction in the first trial on one theory of DWI (the per se theory) necessarily constitutes an implied acquittal on the alternative theory on which no conviction was entered (the impaired DWI theory).

Id. at 183 .

The court disagreed with this argument and elaborated as follows:

When a defendant is convicted based on one of two alternative means of committing a single crime, which is the situation presented in this case, the near uniform majority of jurisdictions that have considered the issue have refused to imply an acquittal on the other alternative. See United States v. Ham , 58 F.3d 78 , 84-86 (4th Cir. 1995) ; United States v. Wood , 958 F.2d 963 , 971-72 (10th Cir. 1992) ; United States ex rel. Jackson v. Follette , 462 F.2d 1041 , 1047, 1049-50 (2d Cir. 1972) ;

*833 Beebe v. Nelson , 37 F.Supp.2d 1304 , 1308 (D. Kan. 1999) ; Schiro v. State , 533 N.E.2d 1201 , 1207-08 (Ind. 1989) ; State v. Pexa , 574 N.W.2d 344 , 347 (Iowa 1998) ("A failure to consider an alternative definition of the offense charged does not constitute an acquittal of that offense for double jeopardy purposes."); State v. Wade , 284 Kan. 527 , 161 P.3d 704 , 715 (2007) ; Commonwealth v. Carlino , 449 Mass. 71 , 865 N.E.2d 767 , 774-75 (2007) ; People v. Jackson , 20 N.Y.2d 440 , 285 N.Y.S.2d 8 , 231 N.E.2d 722 , 728-30 (1967) ; State v. Wright , 165 Wash. 2d 783 , 203 P.3d 1027 , 1035 (2009) (en banc); State v. Kent , 223 W. Va. 520 , 678 S.E.2d 26 , 30-33 (2009) ; cf. State v. Terwilliger , 314 Conn. 618 , 104 A.3d 638 , 651-52 (2014) (refusing to imply an acquittal where a general verdict form made it impossible to know which theory supported the defendant's conviction); [State v.] Torrez, 2013-NMSC-034 , ¶¶ 10-14, 305 P.3d 944 (same). But see Terry v. Potter , 111 F.3d 454 , 458 (6th Cir. 1997) ; State v. Hescock , 98 Wash. App. 600 , 989 P.2d 1251 , 1256-57 (1999) (applying Terry ).

Id.

It is also worth noting that the official toxicology report from the Iowa Division of Criminal Investigation (DCI) lab is part of the trial record here. It was included in the minutes, and no one objected to its being considered. Thus, while I agree that this lab report is not enough to prove that Myers was guilty beyond a reasonable doubt under the Iowa Code section 321J.2(1)( c ) (2016) alternative, it is evidence that may be taken into account in determining Myers's guilt or innocence under section 321J.2(1)( a ).

Waterman and Christensen, JJ., join this special concurrence.