The Farmington Country Club, Inc., and its insurer, Great American Insurance Company/Ohio Casualty Group (hereinafter collectively “employer”), appeal a decision of the Workers’ Compensation Commission holding that employer is responsible for payment for the prescription medication Nexium for Frank Wayne Marshall (claimant) to treat his gastro-esophageal reflux disease (GERD). On appeal, employer contends the evidence is insufficient to prove claimant’s GERD is causally related to his compensable industrial injury. It contends further that, even if the GERD is causally related, it is a consequence of a compensable consequence and, thus, that the *19causal connection is too attenuated to be covered under the Workers’ Compensation Act. For the reasons stated, we find the commission erred in finding the necessary causal connection between claimant’s industrial injury and his GERD. Thus, we reverse the commission’s award of medical benefits.
I.
BACKGROUND
On September 6, 2000, while working as a painter for employer, claimant fell twelve feet from a ladder onto a wall, injuring his back.1 He was diagnosed with “T8 through LI fractures” and “thoracic spinal cord contusion with an ASIA level B paraplegia.” He received emergency treatment and spent five weeks in a rehabilitation facility.
While in the rehabilitation facility, claimant received numerous medications for pain, muscle spasms, and insomnia. He also was prescribed various medications for indigestion, nausea, bloating, and gas, including Famotidine, Maalox, Simethicone, and a “GI cocktail” of “lidocaine viscous” and “donnatal elixir.” On claimant’s discharge from the rehabilitation hospital on October 17, 2000, he was instructed to continue taking numerous medications for pain. He was also prescribed Protonix, a drug similar to Nexium, to be taken as needed. Finally, he was “instructed ... to wear a soft corset” and “to use his wheelchair as provided with no unassisted ambulation.”
Pursuant to a commission award entered May 21, 2001, claimant received medical benefits and ongoing temporary total disability benefits beginning September 7, 2000. In September 2002, claimant began treatment with Dr. Robert L. Muller, a licensed clinical psychologist. Dr. Muller noted claimant experienced chronic pain and had significant depression and anxiety secondary to his medical problems and permanent disability. Claimant obtained treatment from Dr. Muller intermittently through at least January 2004.
*20In 2004, nearly 4 years after the injury, claimant was initially diagnosed with GERD. On June 18, 2004, family physician Dr. William R. Dandridge prescribed Nexium for the treatment of claimant’s GERD. Claimant sought payment from the employer for the prescription, and when employer denied coverage, claimant sought a ruling by the commission.
Medical evidence before the commission included statements from several of claimant’s treating physicians. Dr. Alan Alfano, one of the physicians who treated claimant at the rehabilitation facility shortly following his injury, was asked, “[Pjlease advise of your opinion as to whether or not Mr. Marshall’s GERD is a result of his back injury____” In response, Dr. Alfano opined on July 15, 2004, that “I don’t believe that [claimant’s] GERD is [djirectly caused by his [spinal cord injury]. However, I do believe that it is indirectly related to his injury in that it is likely exacerbated by stress. I have referred the treatment of his GERD to his [primary care physician].” (Underlining in original.)
On July 27, 2004, Dr. Martin D. Katz, a partner of Dr. Dandridge, opined as follows: “[Claimant] had no history of gastric reflux or gastric ulcer prior to his fall in September of 2000. The emotional and physical stress he endures everyday as a result of his injuries from that fall may have caused his present gastrointestinal problems.”
Claimant also obtained an opinion from Dr. Muller, his treating licensed clinical psychologist. On August 2, 2004, Dr. Muller wrote: ■
Mr. Marshall asked me to comment on his recent diagnosis of an acid reflux disorder. The past few years Mr. Marshall has been experiencing significant anxiety & depression stemming from his physical disabilities and change of life. His stress level certainly must be considered as an etiological factor for the acid reflux & also considered to significantly exacerbate his symptoms.
In a determination made on the record, the deputy commissioner found the medical opinions of Drs. Muller and Alfano “persuasive in implicating stress caused by the work accident *21as a partial cause or exacerbating cause of the claimant’s GERD.” The deputy also rejected employer’s argument that claimant’s GERD nevertheless was not covered because it was a consequence of a compensable consequence. Because claimant’s compensable back injury caused symptoms of emotional and physical stress, which, in turn, caused or aggravated claimant’s GERD, the deputy held the GERD was a compensable consequence directly linked to the original injury. The deputy further opined:
Our decision in the present matter turns upon a rather subtle distinction. Had we found that the GERD condition causing the need for the disputed medication was caused by depression resulting from physical disabilities and pain resulting from the work accident, we would have ruled against the claimant. This chain of causation is similar to the brain injury resulting from depression caused by the physical injury to the claimant’s wrist caused by the work accident in Paul Johnson Plastering [v. Johnson, 265 Va. 237, 576 S.E.2d 447 (2003)]. In the present matter, however, we find that the GERD condition is directly caused or aggravated, at least in part, by emotional and physical stress resulting from the work injury. Put another way, this is not a case where stress from the physical injury causes depression which, in turn, causes GERD. Rather, the stress directly causes (or aggravates or exacerbates) the GERD.
On employer’s request for review, the commission unanimously affirmed, agreeing with the deputy’s findings of fact and conclusions of law. It relied on the decision in Berglund Chevrolet, Inc. v. Landrum, 43 Va.App. 742, 601 S.E.2d 693 (2004), in which an employer was held responsible for a prescription to treat erectile dysfunction caused, at least in part, by pain and medicines related to a compensable work accident. It also relied on case law holding that where stress from a work accident causes depression, that depression is covered as a compensable consequence.
Employer noted this appeal.
*22II.
ANALYSIS
An accidental injury that arises out of and in the course of a claimant’s employment is compensable under the Workers’ Compensation Act. See, e.g., Amoco Foam Prods. Co. v. Johnson, 257 Va. 29, 32, 510 S.E.2d 443, 444 (1999). The “causal connection” necessary for compensability “is established when it is shown that an employee has received a compensable injury which materially aggravates or accelerates a pre-existing latent disease.” Justice v. Panther Coal Co., 173 Va. 1, 6-7, 2 S.E.2d 333, 336 (1939), cited with approval in Ohio Valley Constr. Co. v. Jackson, 230 Va. 56, 58, 334 S.E.2d 554, 555 (1985). The disease caused or aggravated may be physical or emotional. For example, when the original injury produces stress, thereby causing depression or aggravating a pre-existing case of depression or some other emotional or mental condition, the resulting disability is compensable. Yellow Freight Sys. v. Robinson, 37 Va.App. 436, 441, 559 S.E.2d 381, 383 (2002). Generally speaking, an employer’s liability for an industrial injury extends to “ ‘all the medical consequences and sequelae that flow from the primary injury.’” American Filtrona Co. v. Hanford, 16 Va.App. 159, 163, 428 S.E.2d 511, 513 (1993) (quoting 1 Arthur Larson, The Law of Workmen’s Compensation § 13.11 (1992)).
In keeping with this principle, the Supreme Court has adopted the doctrine of compensable consequences. Immer & Co. v. Brosnahan, 207 Va. 720, 152 S.E.2d 254 (1967). Under this doctrine, a covered employee who suffers a second injury that bears a sufficient “causal connection” to the first injury is entitled to the protections of the Act for the second injury, as well. Amoco Foam Prods., 257 Va. at 32, 510 S.E.2d at 444. “ When the primary injury is shown to have arisen out of and in the course of employment, every natural consequence that flows from the injury likewise arises out of the employment, unless it is the result of an independent intervening cause attributable to claimant’s own intentional conduct.’ ” Imperial Trash Serv. v. Dotson, 18 Va.App. 600, *23606-07, 445 S.E.2d 716, 720 (1994) (quoting Morris v. Badger Powhatan/Figgie Int’l, Inc., 3 Va.App. 276, 283, 348 S.E.2d 876, 879 (1986)). Under these circumstances, “the second injury is treated as if it occurred in the course of and arising out of the employee’s employment.” Leonard v. Arnold, 218 Va. 210, 214, 237 S.E.2d 97,100 (1977).
For example, where a claimant with a compensable industrial injury suffers an additional injury in an automobile accident while traveling “to or from a doctor’s office for the treatment of [the] work-connected injury,” the second injury is “sufficiently causally related to the employment by the mere fact that [the] work-connected injury was the cause of the journey.” Brosnahan, 207 Va. at 722-23, 152 S.E.2d at 255-56. This is so even where the direct cause of the automobile accident is a pre-existing medical condition of the employee that is unrelated to the original industrial injury. Id. at 726, 152 S.E.2d at 258. A second injury also is considered a compensable consequence of the original industrial injury when weakness from the original injury precipitates a fall that causes the second injury. Bartholow Drywall Co. v. Hill, 12 Va.App. 790, 407 S.E.2d 1 (1991) (holding wrist injury compensable consequence when claimant with compensable back injury fell as result of residual back weakness).
The Supreme Court has prescribed certain limits to the doctrine of compensable consequences. In Amoco Foam Prods., the Court considered whether a claimant may recover for a “compensable consequence of a compensable consequence,” and decided a claimant could not. 257 Va. at 31, 510 S.E.2d at 444. The claimant in Amoco Foam Prods, suffered a compensable left ankle injury in 1992, and in 1994, while the claimant was recovering from surgery on her left ankle, the ankle gave way, causing injury to her right knee. Id. at 30-31, 510 S.E.2d at 443. The commission determined the 1994 right knee injury was a compensable consequence of the 1992 left ankle injury, and the employer did not appeal that determination. Id. at 31, 510 S.E.2d at 443. In 1995, due to ongoing pain from the 1994 right knee injury, the claimant fell again, causing a new injury to her right knee. Id. On these facts, the *24Supreme Court held “[the 1995] injury did not arise out of the employment because there is absence of a causal connection between the incidents of claimant’s employment and the 1995 injury.” Id. at 33, 510 S.E.2d at 445. The occurrence of the 1994 injury, although causally related to the 1992 injury, was sufficient to break the causal connection between the 1992 and 1995 injuries. Id. Essentially, Amoco Foam Prods., found that if “A,” being the compensable injury, caused “B,” and “B” caused “C,” then “C” is not a compensable consequence. Only if “A” caused “C” can claimant recover.
The Court applied these same limitations in Paul Johnson Plastering v. Johnson, 265 Va. 237, 576 S.E.2d 447 (2003). The claimant in that case suffered a compensable injury to his wrist and back in 1990 and subsequently was diagnosed with “chronic depression” and “a major depressive disorder.” Id. at 239, 576 S.E.2d at 449. In 1995 he was diagnosed with a brain injury, and in 1998, he was found to be permanently and totally disabled as a result of his wrist, back, and brain injuries. Id. at 240, 576 S.E.2d at 449. Johnson contended his “disability resulted from a structural change in his brain or a brain injury that developed from his depression which in turn was caused by the injury to his wrist.”2 Id. at 241, 576 S.E.2d at 450. The Court held that, under Amoco Foam Prods., if the claimant’s wrist injury caused depression and the depression, in turn, caused the claimant’s brain injury, “the brain injury is not compensable because, under this theory, there is no direct causal link between that brain injury and the original industrial accident injury, the wrist injury.” Id. at 244, 576 S.E.2d at 452. The original injury to claimant’s back and wrist was “A.” The depression was “B,” and the brain injury was “C.” The Supreme Court found that while “A” caused “B” and “B” caused “C,” no evidence indicated “A” *25caused “C.” As a result, the Court held the claimant was not entitled to compensation for the brain injury. Id. at 245, 576 S.E.2d at 452.3
Our decision in Berglund Chevrolet, Inc. v. Landrum, 43 Va.App. 742, 601 S.E.2d 693 (2004), can be explained with the “A, B, C” analysis. Landrum suffered a compensable back injury (“A”). As a direct result of that back injury, Johnson suffered from “major depression” (“B”) and was prescribed anti-depression medications, which, in part caused erectile dysfunction (“C”). Id. at 747, 601 S.E.2d at 695. His doctor concluded:
Mr. Landrum is regularly followed through our office for complications of a closed head injury that occurred on June 8, 1990 with associated chronic headaches. He, in addition, has a history of back injuries associated with both the headaches and back pain, a chronic pain syndrome/complex regional pain disorder. As a result of the combination of these problems and the medications used to treat them, he has difficulty with erectile dysfunction, improved with Viagra. It is difficult to blame a specific medication or injury and it is likely that it is the combination of both injuries and his medications that have contributed to his sexual dysfunction.
Id. at 747, 601 S.E.2d at 696. We held the doctor “attributed Landrum’s sexual dysfunction (and, thus, his need for Viagra) to the original workplace injury.” Id. at 752, 601 S.E.2d at 698. We further concluded:
Dr. Daugherty focused on the combination of the “chronic pain syndrome/complex regional pain disorder” caused in part by the workplace accident and “the medications used to treat them.” In his view, the causal chain was not interrupted by any intervening accidents, unrelated injuries, *26idiopathic conditions, or causal influences for which Land-rum should bear responsibility. Finding Dr. Daugherty’s opinion persuasive, the commission concluded “the evidence predominates in establishing that the claimant has developed sexual/erectile dysfunction (and a need for Viagra), at least in part, as a direct consequence of the workplace injury to his back.”
Id. (emphasis in original).
Thus, in Berglund, the back injury (“A”) was a direct cause of the erectile dysfunction (“C”). It was not that the back injury caused depression, which in turn caused the dysfunction.
III.
CAUSATION AND COMPENSABLE CONSEQUENCE
The commission’s determination regarding causation is a finding of fact. Marcus v. Arlington County Bd. of Supers., 15 Va.App. 544, 551, 425 S.E.2d 525, 580 (1993). A finding of causation need not be based exclusively on medical evidence, and a claimant is not required to produce a physician’s medical opinion in order to establish causation. Dollar Gen. Store v. Cridlin, 22 Va.App. 171, 176-77, 468 S.E.2d 152, 154r-55 (1996). Causation of a medical condition may be proved by either direct or circumstantial evidence, including medical evidence or “the testimony of a claimant.”4 Id. at 176, 468 S.E.2d at 154; see Lee County Sch. Bd. v. Miller, 38 Va.App. 253, 260, 563 S.E.2d 374, 378 (2002). Any medical opinion offered into evidence “is not necessarily conclusive, but is subject to the commission’s consideration and weighing.” Hungerford Mech. Corp. v. Hobson, 11 Va.App. 675, 677, 401 S.E.2d 213, 215 (1991). In determining whether credible evidence exists to support the commission’s findings of fact, “the appellate court does not retry the facts, reweigh ... the evidence, or make its own determination of the credibility of *27the witnesses.” Wagner Enters. v. Brooks, 12 Va.App. 890, 894, 407 S.E.2d 32, 35 (1991). Thus, unless we can say as a matter of law that claimant failed to sustain her burden of proving causation, the commission’s findings are binding and conclusive upon us. Marcus, 15 Va.App. at 551, 425 S.E.2d at 530.
Employer contends that, based on the Supreme Court’s holdings in Amoco Foam Prods, and Paul Johnson Plastering, the connection between claimant’s back injury and his subsequent development of GERD is too attenuated to provide the necessary causal link. We agree.
We first must review the medical testimony to determine whether the facts support the commission’s finding. It should be noted neither the deputy nor the commission found the original back injury caused the GERD condition. The deputy stated, “[W]e find that the GERD condition is directly caused or aggravated, at least in part, by emotional and physical stress resulting from the work injury.” The commission, in awarding benefits, concluded, “Further as the deputy commissioner noted, where stress from the work accident causes depression, case law uniformly holds the [employer] responsible in that situation.”
This is not the correct legal standard as set forth in Amoco Foam Prods, and Paul Johnson Plastering. The commission only related stress/depression to the GERD condition. Nowhere in the opinions is a finding that the original injury (“A”) caused the GERD condition (“C”). The record before the commission only showed that the original injury (“A”) caused the depression/stress (“B”), which in turn caused the GERD condition (“C”). This is the very causation pattern rejected by Amoco Foam Prods, and Paul Johnson Plastering.
The evidence from the treating physicians, rather than proving causation, as the dissent suggests, indicates the error of the commission’s finding.
When invited to comment on the relationship between claimant’s GERD and his back injury, Dr. Alfano specifically concluded that there was no direct causation between the injury *28and the GERD. Thus, Dr. Alfano’s medical opinion affords no basis for causation.
Dr. Katz indicated, “[T]he emotional and physical stress [claimant] endures everyday as a result of his injuries from the fall may have caused his present gastrointestinal problems.” (Emphasis added). The word “may” is insufficient to prove causation. “May” is defined as “to express possibility.” Webster’s College Dictionary 811 (2d ed.1997). “A medical opinion based on a ‘possibility’ is irrelevant, purely speculative and hence, inadmissible.” Spruill v. Commonwealth, 221 Va. 475, 479, 271 S.E.2d 419, 421 (1980). See also Westmoreland Coal Co. v. Campbell, 7 Va.App. 217, 224, 872 S.E.2d 411, 416 (1988) (“[I]f the expert medical witnesses cannot testify that it is at least more probable than not that the disease arose out of and in the course of employment, compensation must and should be denied, not because the law requires more of medicine than it can produce, but because the law requires more than simply proof that the disease ‘might’ have been caused by a particular result.”). Again, Dr. Katz’s finding establishes no causation between the original injury and the GERD condition.
Dr. Muller found claimant “has been experiencing significant anxiety and depression stemming from his physical disabilities and his stress level certainly must be considered as an etiological factor for the acid reflux and also considered to significantly exacerbate his symptoms.” Dr. Muller, at best, opines the original injury (“A”) caused the stress/depression (“B”), which in turn caused the GERD condition (“C”). Nothing in Dr. Muller’s opinion can be interpreted to indicate the original back injury was a direct cause of the GERD condition.5
*29The dissent points to the fact that during claimant’s initial hospitalization for the original injury, he was prescribed medication for indigestion, gas, nausea, and bloating.6 The discharge summary indicates claimant was also prescribed Protonix, an acid blocker medication similar to Nexium. However, nothing in the record indicates that indigestion, nausea, bloating, and gas are symptoms of GERD. Nothing in the discharge summary indicates any symptoms of GERD, nor a diagnosis of GERD, nor any referral for treatment of a GERD condition. While Protonix is an acid blocker, as is Nexium, there is no medical evidence indicating that Protonix was prescribed for a GERD condition. It is pure speculation that claimant suffered any symptoms of GERD during the initial hospitalization.
We conclude no evidence supports the commission’s determination of causation and farther conclude the commission applied an incorrect analysis concerning compensable consequences. We reverse the commission’s award of medical benefits.
Reversed.
. Employer does not contest the compensability of the original injury.
. Johnson contended in the alternative that the brain injury was directly caused by the original industrial accident. Johnson, 265 Va. at 243, 576 S.E.2d at 451. The Court noted that the deputy commissioner concluded that the medical evidence failed to support such a finding. Id. The Court further noted that, because "[tjhis factual finding was never set aside,” Johnson could not prevail on that basis. Id.
. While Food Distributors v. Estate of Ball, 24 Va.App. 692, 485 S.E.2d 155 (1997), seems to support claimant’s position, Food Distributors has been implicitly overruled by Amoco Foam Prods., 257 Va. 29, 510 S.E.2d 443, and Paul Johnson Plastering, 265 Va. 237, 576 S.E.2d 447, which set the parameters for application of the compensable consequences doctrine.
. The record does not show claimant related the original injury to the GERJD condition.
. We note that the uncompensable consequences in Amoco Foam Prods. and Paul Johnson Plastering involve subsequent injuries, while the claimant here complains of an ensuing symptom of an illness. We find no distinction between a consequential injury and consequential symptom for purposes of an "A-B-C” analysis, as long as the evidence links the original injury to the subsequent consequence.
. Neither the deputy, nor the full commission, considered the gastrointestinal symptoms in their determination of causation.