[¶ 1.] Connie Hanten appeals the circuit court’s order reversing a Department of Labor decision that her tendonitis condition is work-related and therefore compensable. We affirm.
FACTS & PROCEDURE
[¶2.] Connie Hanten (Hanten) was employed by Palace Builders, Inc. (Palace) in 1979 as a clerical worker, which, beginning in 1982, involved the use of computers. In 1989, Hanten began developing pain, numbness, and a tingling sensation in her right hand and arm. She reported her injury to Palace and began treatment with a chiropractor.
[¶ 3.] Hanten’s symptoms grew worse. In September 1990, she was seen by Dr. Robert VanDemark, an orthopedist, who recommended carpal tunnel release surgery. Palace’s insurer hired a medical rehabilitation consultant to manage Hanten’s medical services. Surgery was delayed due to Hanten’s pregnancy, but was eventually performed in March 1991. Although Dr. VanDemark recommended she remain off work for four to six weeks following surgery, Hanten returned to work in three weeks. Palace paid Hanten worker’s compensation benefits for her carpal tunnel injury without dispute as to causation or the nature of the subsequent medical treatment.
[¶4.] Hanten continued to have problems with her right hand following surgery and continued to see an occupational therapist. In fact, following treatment at a center for hand therapy in Sioux Falls, she continued therapeutic treatment in Mitchell from April 1991 through November 1992. In October 1991, Palace’s medical rehabilitation consultant described Hanten’s condition as “tendonitis.” Three days later, Palace terminated her employment.
[¶ 5.] Approximately five months later, in March 1992, Hanten reported she felt a grinding or popping sensation after attempting to open a jar in her home. In June 1992, Dr. VanDemark diagnosed Hanten’s condition as de Quervain’s tenosynovitis, a type of tendonitis, after performing a Finkelstein’s test, a medically recognized test for this condition. Surgery was recommended.
[¶ 6.] Hanten sought worker’s compensation benefits from her former employer Palace, for her condition of de Quervain’s tendonitis. The Department of Labor determined this condition was work-related and awarded benefits. The circuit court re*78versed the Department’s decision, finding the Department erred in its reliance on the orthopedist’s opinion and that Hanten failed to establish the cause of the de Quervain’s was work-related.
[¶ 7.] Hanten appeals to this Court raising the following issue:
Whether the Department of Labor’s finding that the de Quervain’s condition is work-related and therefore compensable is supported by substantial evidence in the record?
STANDARD OF REVIEW
[¶ 8.] We recently set forth our well-settled standard of review of worker’s compensation decisions in Helms v. Lynn’s Inc., 1996 SD 8, ¶¶ 9-10, 542 N.W.2d 764, 766. In Helms, we noted:
Our standard of review from decisions of administrative agencies is governed by SDCL 1-26-37. This statute provides:
An aggrieved party or the agency may obtain a review of any final judgment of the circuit court under this chapter by appeal to the Supreme Court. The appeal shall be taken as in other civil cases. The Supreme Court shall give the same deference to the findings of fact, conclusions of law and final judgment of the circuit court as it does to other appeals from the circuit court. Such appeal may not be considered de novo.
However, when the issue is a question of law, the agency’s actions are fully renewable. Caldwell v. John Morrell & Co., 489 N.W.2d 353, 357 (S.D.1992); Egemo v. Flores, 470 N.W.2d 817, 820 (S.D.1991). Further, we review the findings based on deposition testimony and documentary evidence de novo. Caldwell, 489 N.W.2d at 357.
The issue we must determine is whether the record contains substantial evidence to support the agency’s determination. In re Establishing Certain Territorial Elec. Boundaries, 318 N.W.2d 118, 121 (S.D.1982); Nehlich v. SD Comprehensive Health, 290 N.W.2d 477, 478 (SD 1980); Dail v. SD Real Estate Comm’n, 257 N.W.2d 709, 712 (SD 1977).
ANALYSIS AND DECISION
[¶ 9.] In a worker’s compensation case, the “[cjlaimant has the burden of establishing the causal connection between the employment and the injury by a preponderance of the evidence, and ‘a possibility is insufficient and a probability is necessary.’ ” Helms, 1996 SD ¶20, 542 N.W.2d at 768; Caldwell, 489 N.W.2d at 358; Deuschle v. Bak Constr. Co., 443 N.W.2d 5, 6 (S.D.1989). Claimant’s burden is not met when the probabilities are equal. Helms, 1996 SD 8, ¶ 20, 542 N.W.2d at 768; Caldwell, 489 N.W.2d at 358; King v. Johnson Bros. Constr. Co., 83 S.D. 69, 155 N.W.2d 183 (1967).
[¶ 10.] Expert witness testimony must be used to establish the causal connection between one’s employment and subsequent injury where “the field is one in which laymen are not qualified to express an opinion.” Helms, 1996 SD 8, ¶ 13, 542 N.W.2d at 767; Deuschle, 443 N.W.2d at 6. See also Day v. John Morrell & Co., 490 N.W.2d 720, 724 (S.D.1992); Howe v. Farmers Coop. Creamery, 81 S.D. 207, 212, 132 N.W.2d 844, 846 (1965) (stating that if the relationship between an injury and work is unclear, a medical expert may be needed to establish this relationship). We have noted, however, that “expert testimony is entitled to no more weight than the facts upon which it is predicated.” Westergren v. Baptist Hospital of Winner, 1996 SD 69, ¶25, 549 N.W.2d 390, 397 (1996); Helms, 1996 SD 8, ¶21, 542 N.W.2d at 768; Podio v. American Colloid, 83 S.D. 528, 532, 162 N.W.2d 385, 387 (1968). We have also long held that while the worker’s compensation act is to be liberally construed in favor of the claimant, this rule applies to the law and not to the evidence offered to support the claim. Egemo, 470 N.W.2d at 824; Wold v. Meilman Food Industries, Inc., 269 N.W.2d 112, 116 (S.D.1978); Podio, 83 S.D. at 534, 162 N.W.2d at 388.
[¶ 11.] In the present ease, the Department of Labor stated it relied upon the testimony of Dr. Robert VanDemark in determining that Hanten’s de Quervain’s tendonitis condition was causally connected to *79her employment at Palace Builders, Inc. Dr. VanDemark had been Hanten’s treating physician since her referral to him in September 1990 regarding her carpal tunnel injury.
[¶ 12.] Dr. VanDemark was deposed twice regarding Hanten’s de Quervain’s tendonitis condition and whether it was causally connected to her employment at Palace. In his first deposition, taken October 8, 1992, the doctor’s testimony did not establish a causal connection with the de Quervain’s condition and Hanten’s employment. In a letter dated March 8,1998 to an attorney involved in this case, Dr. VanDemark, responding to an-inquiry, stated the de Quervain’s was not related to any work injury.1 In his second deposition taken October 11, 1994, however, Dr. VanDemark changed his opinion and testified the de Quervain’s condition “probably is work related.” He testified that he based his change of opinion on reports prepared by the employer/insurer’s medical rehabilitation consultant, a registered nurse, and that he assumed the information contained therein had been based on medical records:
Q: What has changed your opinion?
A: ... I think the fact that it [the reports of the rehabilitation nurse] said this diagnosis of tendinitis made by whoever is certainly suggestive of a de Quervain’s.
[[Image here]]
Q: Is a nurse capable of making a report referring to tendinitis?
A: Sure. I assume she has some medical record to review to do that.
(emphasis added).
[¶ 13.] As it was later discovered, Dr. VanDemark’s assumption was incorrect in that the rehabilitation consultant referred to no medical record when she stated in her October 1991 report that “the continuing pain may be due to tendonitis and can linger for prolonged periods of time.” (emphasis added). Hanten’s employment with Palace was terminated October 22, 1991, three days after this report was written. Regardless of the fact that no medical diagnosis of de Quer-vain’s tendonitis was made until June 1992, the rehabilitation consultant’s reports continued to refer to Hanten’s condition as “tendonitis.” Dr. VanDemark made clear in his deposition testimony that he relied upon several of these reports in changing his opinion. An expert’s opinion is entitled to no greater weight than the facts upon which it is based. Westergren, 1996 SD 69, ¶25, 549 N.W.2d at 397. The factual basis for Dr. VanDemark’s change of opinion was incorrect.
[¶ 14.] The question of the correct diagnosis of Hanten’s condition aside, a claimant must further establish the injury is causally related to her employment to be awarded worker’s compensation benefits. Westergren, 1996 SD 69, ¶ 24, 549 N.W.2d at 396-97. As to the degree of medical certainty regarding causation, Dr. VanDemark testified to the following:
Q: Doctor, based upon reasonable medical probability, do you have an opinion as to whether or not this patient’s present symptoms of tendinitis, for which you have recommended surgery, are consistent with an injury occurring at work?
[[Image here]]
A: Yes.
Q: And what is your opinion?
[[Image here]]
A: That it probably is work related.
Had this been the extent of Dr. VanDe-mark’s testimony, it would have provided an evidentiary basis to establish the necessary work-related causation. However, later, on cross examination during this same deposition, Dr. VanDemark departed from his above-quoted direct testimony on the causation question.
A: I’m not sure if I would say that her job caused it. I guess I can’t say that *80I don’t know that it did cause it, if that makes sense.
[[Image here]]
Q: You can’t say that you don’t know that it did cause it?
A: Yeah. I guess the fact that she developed it — what I’m trying to say is that her job may well have caused it, but I can’t say that for sure, but I’m not going to say that it didn’t cause it because I don’t know that, if that makes sense.
[[Image here]]
Q: Is it fair to say, Doctor, that your opinion would be that you cannot rule out the work environment as being the cause of her de Quervain’s?
A: I think that’s a fair statement.
Q: Would you be willing to go beyond that in terms of certainty or the causal relationship between the de Quer-vain’s and the workplace?
[[Image here]]
A: No.
[¶ 15.] The totality of such testimony does not establish causation to the level of proof required for Hanten to prevail on her claim. Examining his testimony as a whole, we note that Dr. VanDemark could not state with the required degree of certainty that Hanten’s employment was the cause of Han-ten’s de Quervain’s tendonitis. Proof need not arise to a degree of absolute certainty, but an award may not be based upon mere possibility or speculative evidence. Mehlum v. Nunda Coop. Ass’n, 74 S.D. 545, 551, 56 N.W.2d 282, 285 (1952). Causation must be established to a reasonable medical probability, not just a possibility. Harden v. S.D. Credit Union League, Inc., 87 S.D. 433, 438, 209 N.W.2d 665, 667 (1973). Dr. VanDe-mark’s opinion rose only to the level of a possibility, not probability, in favor of Han-ten’s claim. As noted above, where the probabilities are equal, the claim fails.
[¶ 16.] Following advisement by Hanten’s counsel of Dr. VanDemark’s revised opinion, but prior to taking his second deposition, employer/insurer scheduled an independent medical examination of Hanten by Dr. Chris P. Tountas. Based on Hanten’s medical history, his review of the records and his own experience as an orthopedic surgeon with a practice limited to the hand and upper extremity, Dr. Tountas concurred with the diagnosis of de Quervain’s but opined that this condition was not causally related to Han-ten’s employment with Palace Builders. He further noted there was no evidence in the record of de Quervain’s prior to Hanten’s complaints in the early part of 1992, prior to Dr. VanDemark’s diagnosis of de Quervain’s and some months following Hanten’s termination from Palace.2
[¶ 17.] All of the evidence in this ease points to the fact that Connie Hanten is a dedicated worker who has worked hard all of her life, taking minimal time off to recover from surgery or from delivery of her children. She worked for Palace for thirteen years, assisting them as a “one-woman office” in getting their new-found business operating. There is no evidence to indicate she is a malingerer. See Welch v. Automotive Co., 528 N.W.2d 406, 408 (S.D.1995) (malingerer defined as someone who feigns or intentionally exaggerates disability in order to draw compensation benefits). Regardless of her admirable work ethic, however, she has the burden of establishing that her injury is work related before she may receive an award of worker’s compensation.
[¶ 18.] “ ‘A [worker’s] compensation award cannot be based on possibilities or probabilities, but must be based on sufficient evidence that the claimant incurred a disability arising out of and in the course of his employment.’” Day, 490 N.W.2d at 724 (quoting Wold, 269 N.W.2d at 116). As noted above, where the relationship between the work and the injury is not clear, medical expert testimony is required to establish the causal connection. Howe, 81 S.D. 207, 212, 132 N.W.2d 844, 846. “Medical testimony to the effect that it is possible that a given *81injury caused a subsequent disability is insufficient, standing alone, to establish the causal relation under worker’s compensation statutes.” Day, 490 N.W.2d at 724. See also Thomas v. St. Mary’s Roman Catholic Church, 283 N.W.2d 254, 258 (S.D.1979) (medical experts are qualified to express their opinions based upon medical certainty or medical probability, but not upon possibility). In accord with the testimony offered by the medical experts in this case and under the authority cited above, we do not find substantial evidence to support the Department of Labor’s determination of causation. We affirm the circuit court.
[¶ 19.] MILLER, C.J., and AMUNDSON and KONENKAMP, JJ., concur. [¶ 20.] SABERS, J., dissents.. .In this letter, written eighteen months after Palace terminated Hanten's employment and nine months after his medical diagnosis of de Quervain’s, Dr. VanDemark wrote:
Ms. Hanten has a tendinitis of the wrist which is called de Quervain's tenosynovitis. She relates to me that this began in March 1992 when she twisted a jar and heard her wrist 'pop.'
It is my opinion that this is not related to her Workmen’s Compensation injury.
. Likewise, Hanten's testimony before the Department of Labor, which we have reviewed in deciding this appeal, does not satisfy her burden of establishing causation to a reasonable medical probability.