DUGGAN, J.

          The petitioner, James Margeson (employee), appeals a decision of the New Hampshire Compensation Appeals Board (CAB) denying him reimbursement for medical treatment and workers' compensation benefits. The parties dispute whether his injury arose out of his employment as required by RSA 281-A:2, XI (2010). We vacate and remand.

         The CAB found or the record supports the following facts. The employee injured his right knee on April 18, 2009, while working for the respondent, the New Hampshire Department of Health and Human Services (employer), as a youth counselor at the John Sununu Youth Center (Youth Center). The employee was working the third shift and performing a routine bed check to ensure the residents were in bed at the required time. While conducting this bed check, he descended the stairs in one of the Youth Center's buildings, and his left foot landed awkwardly, causing him to lose his balance and twist his right knee. The stairs were in new condition and were not defective. Additionally, they were not wet or otherwise hazardous. While the employer alleged that the injury was precipitated by a pre-existing war wound to the employee's foot, the CAB rejected this as a cause of his injury. As a result of the knee injury, the employee sought treatment and incurred medical bills at four different medical centers and hospitals and missed work from April 19 to June 5.

         The employee subsequently sought reimbursement for his medical expenses and disability and indemnity benefits. The employer denied his claim because it determined that his injury did not arise out of his employment. A department of labor hearings officer upheld the employer's denial of benefits and the employee appealed to the CAB. The CAB upheld the decision because the employee " did not encounter any greater risk of his employment than in his everyday life and the stairs were merely an incident or an occasion that accompanied the injury and the employment was not a contributory or additional risk in bringing his injury about." The CAB denied the employee's motion for reconsideration. This appeal followed.

         We will not disturb the CAB's decision absent an error of law, or unless, by a clear preponderance of the evidence, we find it to be unjust or unreasonable. Appeal of Belair, 158 N.H. 273, 276, 965 A.2d 1006 (2009); RSA 541:13 (2007). The appealing party has the burden of demonstrating that the CAB's decision was erroneous. Appeal of Belair, 158 N.H. at 276, 965 A.2d 1006.

          To make out a claim for workers' compensation, the employee had to show that his injuries arose " out of and in the course of his employment." RSA 281-A:2, XI. The phrase " in the course of" employment refers to whether the injury " occurred within the boundaries of time and space created by the terms of employment" and " occurred in the performance of an activity related to employment." Murphy v. Town of Atkinson, 128 N.H. 641, 645, 517 A.2d 1170 (1986). The phrase " arising out of" employment refers to the causal connection between the injury and risks of employment, and requires proof that the injury " resulted from a risk created by the employment." Id.; see also Rio All Suite Hotel and Casino v. Phillips, 240 P.3d 2, 4-5 (Nev.2010); Odyssey/Americare of Oklahoma v. Worden, 948 P.2d 309, 311 (Okla.1997); 1 A. Larson, Larson's Workers' Compensation Law § 3.01, at 3-4 (Matthew Bender ed. rev. 2011).

          The parties agree that the employee was injured while at work, i.e., " in the course of" employment. Instead, they dispute whether he suffered an injury " arising out of" his employment. Courts throughout the country have utilized three different tests, the increased-risk test, actual-risk test, and positional-risk test, to determine whether a claimant has met this requirement. We have yet to adopt any one of these tests, and the parties disagree regarding which standard should govern our inquiry. Accordingly, to provide guidance and clarity to the department of labor, the bench and the bar, we adopt a single test to determine whether the type of injury that occurred here arises out of employment.

         To aid us in characterizing how the injury in this case should be analyzed, we begin by outlining four types of injury-causing risks commonly faced by an employee at work. These categories of injury-causing risks include: (1) risks directly associated with employment; (2) risks personal to the claimant; (3) mixed risks; and (4) neutral risks. 1 Larson, supra §§ 4.01-4.03, at 4-2 to 4-3.

          Employment-related risks include " all the obvious kinds of injur[ies] that one thinks of at once as industrial injur[ies]" and are almost always compensable. Id. § 4.01, at 4-2. These risks include falling objects, explosives, and fingers being caught in gears. Id. Typically, a slip and fall is only attributable to an employment-related risk if it results from tripping on a defect or falling on an uneven or slippery surface on an employer's premises. Rio All Suite Hotel and Casino, 240 P.3d at 5. This category of risks always arises out of employment. 1 Larson, supra § 4.01, at 4-2.

          The next category of risks, personal risks, are " so clearly personal that, even if they take effect while the employee is on the job, they could not possibly be attributed to the employment." Id.Rio All Suite Hotel and Casino, § 4.02, at 4-2. A fall caused solely by an employee's personal condition, such as a bad knee, epilepsy, or multiple sclerosis, falls into this category. 240 P.3d at 5. Injuries falling squarely into this category are never compensable. Id.

          The third category of risks, mixed risks, involve a personal risk and an employment risk combining to produce injury. 1 Larson, supra § 4.04, at 4-3. A common example of a mixed-risk injury is when a person with heart disease dies because of employment-related strain on his heart. Id. While not all injuries resulting from mixed risks are compensable, the concurrence of a personal risk does not necessarily defeat compensability if the claimant's employment was also a substantial contributing factor to the injury. Id.; see also New Hampshire Supply Co. v. Steinberg, 119 N.H. 223, 231, 400 A.2d 1163 (1979)( Steinberg I ).

          Finally, neutral risks are " of neither distinctly employment nor distinctly personal character." 1 Larson, supra § 4.03, at 4-2. This middle-ground category is the most controversial in modern compensation law. Determining whether an injury resulting from a neutral risk arises out of employment is a question of fact to be decided in each case. See Odyssey/Americare of Oklahoma, 948 P.2d at 311. These risks include being hit by a stray bullet, being struck by lightning, or being bitten by a poisonous insect. 1 Larson, supra § 4.03, at 4-3. They also include cases in which " the cause itself, or the character of the cause, is simply unknown." Id. An unexplained fall is considered a neutral risk. Mitchell v. Clark County Sch. Dist., 121 Nev. 179, 111 P.3d 1104, 1106 n. 7 (2005); cf. Odyssey/Americare of Oklahoma, 948 P.2d at 313 (explaining that wet and slippery grass because of rain is considered a neutral risk). Here, the employee slipped and was injured while descending a staircase that was free of defects, and the employer does not contend on appeal that a personal risk caused the employee's injury. Accordingly, we conclude that the employee's injury resulted from a neutral risk.           Injuries caused by neutral risks are by definition not clearly personal or employment-related in nature. Courts predominantly apply one of three tests to determine whether such an injury arises out of employment. 1 Larson, supra § 3.01, at 3-4; Rio All Suite Hotel and Casino, 240 P.3d at 6. The first is the increased-risk test, which is the most widely utilized of these tests. See 1 Larson, supra § 3.03, at 3-4.1. It " examines whether the employment exposed the claimant to a risk greater than that to which the general public was exposed." Rio All Suite Hotel and Casino, 240 P.3d at 6 (quotation omitted). The second, the actual-risk test, ignores whether the risk faced by the employee was also common to the public. 1 Larson, supra § 3.04, at 3-5. A claimant may recover so long as the employment subjects him to the actual risk that causes the injury. Id. The final test is the positional-risk test. In jurisdictions that have adopted this test, an injury arises out of employment " if it would not have occurred but for the fact that the conditions and obligations of the employment placed [the] claimant in the position where he was injured." Id. § 3.05, at 3-6. In other words, under the positional risk test, an injury arises out of employment so long as the obligations of employment place the employee in the particular place at the particular time that he suffers an injury. Id.

         The employer, relying upon Dustin v. Lewis, 99 N.H. 404, 112 A.2d 54 (1955), and Appeal of Lockheed Martin, 147 N.H. 322, 786 A.2d 872 (2001), contends that New Hampshire has adopted the increased-risk test. The employee, however, points to our decisions in Appeal of Redimix Cos., 158 N.H. 494, 969 A.2d 474 (2009), and Appeal of Kehoe, 141 N.H. 412, 686 A.2d 749 (1996), and argues that New Hampshire has adopted either the positional risk or actual risk test. While we recognize some support in our case law for each party's position, we have never before addressed the compensability of an injury caused by a neutral risk.

          Our most recent line of cases addressing the " arising out of" requirement explained that a claimant must prove by a preponderance of the evidence that his work-related activities probably caused or contributed to his disability. Appeal of Redimix, 158 N.H. at 496, 969 A.2d 474. This entails meeting a two-part causation test, which requires proof of legal and medical causation. Id. Legal causation entails a showing that the claimant's injury is in some way work-related, while medical causation requires a showing that the injury was actually caused by the work-related event. Id.

          In applying the legal causation portion of the test, we have explained that the test used depends upon the prior health of the claimant. Id. " When a claimant has a pre-existing disease or condition, he must show by a preponderance of the evidence that his employment contributed something substantial to his medical condition by demonstrating that the work-related conditions presented greater risks than those encountered in his non-employment activities." Id. at 496-97, 969 A.2d 474 (quotation and brackets omitted). " When the claimant does not have a pre[-]existing condition, any work-related activity connected with the injury as a matter of medical fact is sufficient to show legal causation." Id. at 497, 969 A.2d 474.

          Because the employee here did not have a pre-existing condition, he relies upon this final sentence of the legal causation test to assert he is entitled to compensation. Taken at face value, the employee's argument appears to have some merit. However, we originally adopted this test in the context of a compensation claim resulting from a heart attack. See Cheshire Toyota/Volvo, Inc. v. O'Sullivan, 129 N.H. 698, 700-01, 531 A.2d 714 (1987); Steinberg I, 119 N.H. at 230-31, 400 A.2d 1163. In Steinberg I, because of the difficulties associated with proving that work-related stress or anxiety can cause a heart attack in the legal sense, we explained that a heart attack is only compensable if a claimant proves that " the work-related stresses in the particular case at issue were a causal factor in the heart attack which ensued." Steinberg I, 119 N.H. at 230, 400 A.2d 1163 (quotation omitted).

         In order to prove this causal connection, we adopted the contours of the test upon which the employee now relies. We determined that to establish work-related causation for the heart attack, a claimant had to prove both medical and legal causation. Id. We adopted Professor Larson's approach to proving causation in heart attack cases and explained:

The legal causation test defines the degree of exertion that is necessary to make the injury work-connected. The test to be used depends upon the previous health of the employee. Where there is a prior weakness in the form of a previously weakened or diseased heart, then the employment must contribute something substantial to the heart attack. That is, the employment-connected stress or strain must be greater than is encountered in normal non-employment life. Thus, heart attacks that actually result from work-related stress are distinguished from those that occur at work merely as a result of natural physiological process. If there is no prior weakness or disease of the heart, any exertion connected with the heart attack as a matter of medical fact is adequate to satisfy the legal test of causation so as to make the injury or death compensable.

Id. at 230-31, 400 A.2d 1163 (citations omitted). We later noted that differing tests for legal causation take " into account the ‘ personal risks' contributed by the worker and the ‘ employment risks' present in his work." New Hampshire Supply Co. v. Steinberg, 121 N.H. 506, 508, 433 A.2d 1247 (1981) (quotation omitted) ( Steinberg II ).

         In the immediate aftermath of Steinberg I, we repeatedly emphasized that the test we had adopted should be limited to determining compensability for heart attacks and other heavy exertion or stress-related injuries. See Bartlett Tree Experts Co. v. Johnson, 129 N.H. 703, 708, 532 A.2d 1373 (1987) (explaining that Steinberg I " adopted Professor Larson's suggested analysis for the difficult problem of determining the compensability for heart attacks" ); Rogers v. Town of Newton, 121 N.H. 702, 705, 433 A.2d 1303 (1981) (explaining that the Steinberg I test is applicable to all types of injuries resulting from stress or exertion); O'Sullivan, 129 N.H. at 700, 531 A.2d 714 (noting the adoption of Professor Larson's approach to allow recovery for injuries resulting from work-related stress).

          However, since the original adoption of Professor Larson's causation test in Steinberg I, both the wording of the test and its application have gradually broadened. For example, in Wheeler v. School Administrative Unit 21, 130 N.H. 666, 550 A.2d 980 (1988), the plaintiff suffered from manic-depression, but was denied benefits because the trial court determined that he failed to meet his burden of proof under Steinberg I . Wheeler, 130 N.H. at 669-71, 550 A.2d 980. On appeal, we first noted that Steinberg I set forth " a two-part analysis for obtaining workers' compensation benefits in heart attack cases." Id. at 671, 550 A.2d 980. While the plaintiff's claimed injury was depression, and not a heart attack, we nonetheless applied the Steinberg I framework without explanation and stated that " [w]here there is no prior weakness, any exertion connected as a matter of medical fact is adequate to satisfy" the legal causation test. Id. (quotation and ellipsis omitted). Because the evidence supported the master's finding that the plaintiff previously suffered from depression, we concluded that the plaintiff had to prove that his employment contributed " something substantial" to his injury. Id. at 671, 550 A.2d 980 (quotation omitted); see also Appeal of Cote, 139 N.H. 575, 579, 660 A.2d 1090 (1995) (reversing board's denial of compensation for claimant's back injury where evidence established that employee had not suffered from back pain until his repeated exertion at work); Averill v. Dreher-Holloway, 134 N.H. 469, 472, 593 A.2d 1149 (1991) (applying same legal causation standard and determining that plaintiff proved causation where evidence showed he was not predisposed to depression).

         Application of this test was further broadened in Appeal of Briggs, 138 N.H. 623, 645 A.2d 655 (1994), in which the plaintiff conceded a pre-existing injury to his knee, but sought coverage for exacerbation of that injury suffered while at work. Appeal of Briggs, 138 N.H. at 628, 645 A.2d 655. We again applied a variation of the Steinberg I test and stated that the employee had the burden to prove that " the cumulative work-related stress to the [employee]'s knees probably caused or contributed to his disability under a two-pronged test." Id. With regard to the legal causation portion of the test, we reversed the board's denial of benefits because the employee proved that his " work-related activities required more exertion than his non-work-related activities." Id.

         More recently, in Appeal of Kehoe, the Steinberg I test was reduced to its current formulation. The employee in Appeal of Kehoe was repeatedly exposed to chemicals while performing her job and, for the first time, suffered from headaches and respiratory problems. Appeal of Kehoe, 141 N.H. at 414, 686 A.2d 749. We stated that an employee with no pre-existing condition could prove legal causation by establishing " any work-related activity connected with the injury as a matter of medical fact." Id.See at 416, 686 A.2d 749 (emphasis added). Prior to this decision, as explained above, the test had focused upon the employee's degree of stress or exertion because it was intended to apply only to heart attacks and other stress-related injuries. Averill, 134 N.H. at 472, 593 A.2d 1149 (" Where no prior weakness is found, any exertion connected as a matter of medical fact is adequate ...." (quotations and ellipsis omitted; emphasis added)). However, in Appeal of Kehoe, the test was expanded to apply to " any work-related activity. " Appeal of Kehoe, 141 N.H. at 416, 686 A.2d 749 (emphasis added). It is this altered language, which represents a significant expansion of our workers' compensation law, upon which the employee here now relies.

          We have never before been confronted with applying the Steinberg I test to the type of injury at issue here: an unexplained fall that is not attributable to either the employer or employee. As noted, we originally adopted this test to resolve one of the most difficult issues in workers' compensation law: the compensability of heart attacks. See Steinberg II, 121 N.H. at 509, 433 A.2d 1247 (explaining the difficulty of adjudicating heart attack cases because of the " ambiguous relationship of emotional stress and physical exertion to the development and aggravation of coronary heart disease" ). Although initially intended only for heart attack injuries, the test's application was logically extended to those injuries likely to be caused by a pre- existing condition because of the similar difficulty in determining whether such injuries result from an employment-related risk or some other cause. See Appeal of Briand, 138 N.H. 555, 560, 644 A.2d 47 (1994) (explaining that legal and medical causation tests apply in cases involving pre-existing conditions).

         Such difficulty does not arise with regard to neutral risk injuries. Moreover, our case law and the workers' compensation statute do not support further extension of Steinberg I or the adoption of the actual or positional risk theories. While we construe the Workers' Compensation Law liberally in favor of the injured employee, Appeal of N.H. Dep't of Health and Human Servs., 145 N.H. 211, 213, 761 A.2d 431 (2000), we cannot expand the law beyond the legislature's intent. See Dustin, 99 N.H. at 408, 112 A.2d 54 (" Compensation acts have not been designed to place the entire burden of employees' losses upon industry and the right of the workman is no greater than the Legislature has provided it shall be." (quotation and brackets omitted)).

         We thus conclude that further extension of the Steinberg I test to the injury at issue here, which resulted from a neutral risk attributable neither to the employer nor employee, would have the effect of eliminating the statutory " arising out of" requirement, would be contrary to our prior workers' compensation case law, and would be tantamount to strict liability.

         We first reject the further extension of the Steinberg IMurphy, test because doing so would directly contravene the workers' compensation statute, which requires both an injury " arising out of and in the course of employment." RSA 281-A:2, XI. Allowing recovery for an injury simply because it occurred at work would read the " arising out of" requirement out of the statute because a claimant could recover without proving that the injury " resulted from a risk created by the employment." 128 N.H. at 645, 517 A.2d 1170; see also Rio All Suite Hotel and Casino, 240 P.3d at 6. Stated differently, such a reading of the statute would effectively reduce two separate and distinct statutory requirements— an injury " arising out of" and " in the course of" employment— to one, contrary to our long-standing interpretation of the workers' compensation statute. See Murphy, 128 N.H. at 645, 517 A.2d 1170; see also Odyssey/Americare of Oklahoma, 948 P.2d at 311 (emphasizing that " arising out of" and " in the course of" are two distinct and not synonymous statutory requirements).

          Additionally, our case law emphasizes a more stringent causal nexus than would be required if the second prong of the Steinberg I legal causation test were applied to injuries caused by a neutral risk. Not every person suffering an injury at work is entitled to compensation. Dustin, 99 N.H. at 408, 112 A.2d 54. Instead, the claimant must prove that the injury actually resulted from a hazard of the employment. Heinz v. Concord Union School Dist., 117 N.H. 214, 218, 371 A.2d 1161 (1977). Moreover, an " injury must result from the conditions and obligations of the employment and not merely from the bare existence of the employment" to be compensable. Appeal of Lockheed Martin, 147 N.H. at 325-26, 786 A.2d 872 (quotation omitted). The Steinberg I test, when applied to non-neutral risk injuries, is entirely consistent with these principles. Indeed, when applied to non-neutral risk injuries, the test ensures that an injury is caused by an actual hazard of employment rather than some risk that is personal to the employee.

          We therefore conclude that both the workers' compensation statute and our prior case law support the adoption of the increased-risk test when the claimant's injury results from a neutral risk. Because neutral risk injuries result from some unexplained cause not directly attributable to either the employee or employer, this test strikes an important balance between the employee's right to receive compensation and the employer's right not to be held liable for every injury that occurs in the workplace. See Dustin, 99 N.H. at 408, 112 A.2d 54; Rio All Suite Hotel and Casino, 240 P.3d at 7.

          Under the increased-risk test, an employee may recover if his injury results from " a risk greater than that to which the general public is exposed." Rio All Suite Hotel and Casino, 240 P.3d at 7 (quotation and brackets omitted). Importantly, even if the risk faced by the employee " is not qualitatively peculiar to the employment, the injury may be compensable as long as [he] faces an increased quantity of a risk." Id.Id.; (quotations and ellipsis omitted). The act of descending a staircase is an everyday, commonplace activity, which most people must undertake on a daily basis, whether at home, work, or in a shopping mall. Accordingly, we cannot say that the act of descending a staircase at work presents a greater risk than that faced by the general public. see also Nascote Industries v. Industrial Com'n, 353 Ill.App.3d 1056, 289 Ill.Dec. 755, 820 N.E.2d 531, 535 (2004); Southside Va. Training Center v. Shell, 20 Va.App. 199, 455 S.E.2d 761, 763 (1995).

         However, while the act of descending stairs in and of itself does not warrant compensation under the increased-risk test, the employee may still be entitled to compensation under certain circumstances. For example, an employee who must use stairs more frequently than a member of the general public as part of his job faces an increased risk of injury. Rio All Suite Hotel and Casino, 240 P.3d at 7 (determining that claimant's injury arose out of her employment under the increased-risk test because her frequent use of the stairs subjected her to a significantly greater risk than the general public). Additionally, stairs of an unusual height or the manner in which an employee is required to perform his job may also increase the risk of injury. Id.

         Here, the CAB found that the employee " had to ascend and descend stairs on a regular basis during his work day," but also determined that his use of the stairs did not cause him to encounter any greater risk of injury than he incurred in his everyday life. The employee also testified that he used the stairs four times per hour at various times during his shift. Because today we adopt the increased-risk test for the first time and because the CAB did not make explicit findings regarding whether the employee used the stairs more frequently than a member of the general public, we remand to the CAB to make factual findings and apply the increased-risk test to those facts.

         We emphasize that our adoption of the increased-risk test applies only to those injuries attributable to neutral risks such as the unexplained fall at issue here. While the original formulation of the Steinberg I test was intended to be used only in heart attack or stress and exertion cases, that test, because of its long-standing application and continued workability in non-neutral risk cases, should continue to be used for determining compensability for non-neutral risk injuries. Cf. Kalil v. Town of Dummer Zoning Bd. of Adjustment, 159 N.H. 725, 731, 992 A.2d 725 (2010) (explaining that the stability of the law calls for " settlement of principle and consistency of ruling when due consideration has been given and error is not clearly apparent" (quotation omitted)).           Accordingly, in determining the appropriate test in future cases, the CAB should first make a finding regarding the cause of the claimant's injury. If the CAB finds that the injury was caused by a neutral risk, the increased-risk test will apply. However, should the CAB find that the injury resulted from a non-neutral risk, the claimant must prove both legal and medical causation and the test to be used for legal causation depends upon the previous health of the employee. Steinberg I, 119 N.H. at 231, 400 A.2d 1163. If the employee suffers from a prior weakness, the employment-connected stress or strain must be greater than that encountered in normal non-employment life. Id. If there is no prior weakness, any work-related stress or strain connected with the injury as a matter of medical fact satisfies the legal causation test. Id.

         As noted above, the application of the increased-risk test to injuries caused by neutral risks and the use of Steinberg I for adjudicating all other injuries will ensure that a claimant receives compensation only for an injury " arising out of employment." In other words, the injury must actually result from the hazards of employment and " not merely from the bare existence of employment." Heinz, 117 N.H. at 217, 371 A.2d 1161 (quotations omitted); Appeal of Lockheed Martin, 147 N.H. at 325-26, 786 A.2d 872.

          Vacated and remanded.

          DALIANIS, C.J., and HICKS, CONBOY and LYNN, JJ., concurred.