Wanda S. FLATEN, Plaintiff-Appellant, v. SECRETARY OF HEALTH & HUMAN SERVICES, Defendant-Appellee

D.W. NELSON, Circuit Judge: Appellant Wanda S. Flaten appeals the ruling of the district court affirming the decision of the Secretary of Health & Human Services (the “Secretary”) to deny disability benefit payments under section 223 of the Social Security Act (the “Act”), 42 U.S.C. § 423(a)(1). Flaten argues that the district court erred because her current disability, resulting from degenerative disc disease, should relate back to a prior period when she *1456 required surgery and was unable to work because of effects of the same disease. Because the prior period of disability occurred before the expiration of her insured status, Flaten contends that she need not prove that her current total disability has existed continuously since sometime on or before the date that her insured status expired. In the alternative, Flaten argues that the district court erred in upholding the Secretary’s determination that Flaten has not been continuously disabled since her original surgery before the expiration of her insured status, because the Secretary did not present clear and convincing reasons for refusing to accept the treating physician’s retrospective diagnosis of continuing disability from the time of the original surgery. We have jurisdiction under 28 U.S.C. § 1291. We affirm. I. Facts and Proceedings Below Flaten is a sixty-year-old woman with an eleventh grade education who has worked as a bartender and as a manager of a cafe and lounge. Because of a fall on the job in 1969, she sustained a back injury that resulted in severe lower back pain. Although she continued working, she was hospitalized in April 1976 because of the severity of her back pain. In August 1977, Flaten underwent surgery, a laminectomy and discectomy at the L-4, 5 level, which provided some relief. Flaten’s surgeon, Dr. Ingraham, noted at the time that the surgery had been successful in relieving the neurological symptomatol-ogy and suggested that Flaten should be able to return to work within a few months. Fla-ten did not apply for disability benefits to cover the period of time that she had been incapacitated. She returned to work as a bartender during the next two years, posting total earnings of $1,726.63 in 1979 and $2,877.55 in 1980. She has not worked since then, and her insured status under the Act lapsed on March 31, 1982. In 1985, Flaten fell on a boat dock, which may have aggravated her back condition, according to Dr. Ingraham. A short time later, she suffered a recurrence of incapacitating pain that necessitated a second laminectomy in 1986, performed by Dr. Joern. Flaten and her husband testified that she now spends a considerable portion of each day in bed and that her most strenuous activity is earing for house plants in the couple’s mobile home. She contends that she has been continuously disabled because of her degenerative disc disease since the acute incident requiring hospitalization in April 1976. To substantiate her claim, she relies on Dr. Joern’s opinion that Flaten’s current disability is a typical progression of her disease and is directly traceable to the original lumbar injury. On July 2, 1986, after the second laminec-tomy, Flaten filed an application for disability benefits in which she requested benefit payments beginning July 1985. The Social Security Administration (the “SSA”) denied her application initially and upon reconsideration, informing her that she was ineligible for benefits because her current condition had not become disabling on or before the expiration of her insured status. After Flaten supplemented the record with medical evidence relating to her first surgery, she requested a hearing before an Administrative Law Judge (“ALJ”). On October 27,1988, the ALJ denied her claim. The Appeals Council vacated that decision and remanded the case for further development of the record and use of the appropriate medical-vocational rules. On October 20, 1989, a second ALJ denied Flaten’s application. The Appeals Council again vacated the decision because the ALJ failed to request reports of earnings for 1979 and 1980, which were necessary for the appropriate calculation of the final date of insured status, and because the ALJ inappropriately made a determination that Flaten had been disabled since July 1985, a date indisputably after the expiration of her insured status. On February 6, 1991, a third ALJ found that Flaten’s insured status extended until March 31,1982, but he denied her claim for benefits. Although finding that Flaten had not engaged in any substantial gainful activity since the initial period of disability in April 1976, the third ALJ found that she retained sufficient residual functional capacity to perform sedentary work after her recuperation from the 1977 surgery through the date that her insured status lapsed. The ALJ thus decided that Flaten was not disabled from a time on *1457 or before the lapse of her insured status. On July 17, 1991, the Appeals Council denied review, and the third ALJ’s decision became the Secretary’s final ruling, pursuant to 20 C.F.R. § 404.981. Flaten then brought an action for review of the Secretary’s decision under section 205(g) of the Act, 42 U.S.C. § 405(g). After Flaten filed a motion for summary judgment, the case was referred to a magistrate judge who applied a relation-back doctrine that links a recurrence of a disability to an earlier occurrence, during insured status, of a period of disability stemming from the same underlying cause. Because the relation-back doctrine provides automatic eligibility for benefits for a second period of disability after remission, even though the onset of the second period does not meet the insured status requirement, the magistrate judge recommended reversal of the Secretary’s denial of benefits. The Secretary objected to the report and recommendation, and the district court sustained those objections, holding that the relation-back doctrine was not applicable. After remand to the magistrate judge, the district court entered a final order on March 3, 1993 affirming the Secretary’s decision. Flaten timely appealed to this court on April 2, 1993. II. Standard of Review This court reviews a district court’s order upholding the Secretary’s denial of benefits de novo. Matthews v. Shalala, 10 F.3d 678, 679 (9th Cir.1993). The scope of our review, however, is limited: we may set aside a denial of benefits only if it is not supported by substantial evidence or if it is based on legal error. Gonzalez v. Sullivan, 914 F.2d 1197, 1200 (9th Cir.1990). Substantial evidence is relevant evidence which, considering the record as a whole, a reasonable person might accept as adequate to support a conclusion. Tylitzki v. Shalala, 999 F.2d 1411, 1413 (9th Cir.1993). If the evidence can reasonably support either affirming or reversing the Secretary’s conclusion, the court may not substitute its judgment for that of the Secretary. Richardson v. Perales, 402 U.S. 389, 401, 91 S.Ct. 1420, 1427, 28 L.Ed.2d 842 (1971); Matney ex rel. Matney v. Sullivan, 981 F.2d 1016, 1019 (9th Cir.1992). III. Discussion A. Relation Back and the Continuity Requirement Flaten bases her claim for disability benefits on a novel legal theory that a subsequent period of disability should “relate back” to a prior period of disability from the same cause if the prior disability period occurred while the claimant was insured under the Act, even though the claimant did not work for a sufficient period of time to continue her insured status after the end of the earlier period of disability. Flaten argues that her disability satisfies the statutory definition in Section 423(d), because it is undisputed that Flaten was physically disabled for a period of more than twelve months before the expiration of her insured status. Although the ALJ found that Flaten no longer suffered from a disability because of medical improvement as of February 3, 1978, the records also support the conclusion that she has been physically disabled as a result of the same degenerative disc disease processes since July 1985. Flaten claims that her current disability is undisputed because the first two ALJs made an explicit finding to the effect that she has been disabled since July 1985. The Appeals Council vacated those findings, however, with an explicit statement that any finding of a second disability commencing after the expiration of insured status is beyond the jurisdiction of the ALJ in assessing eligibility for benefits arising from an alleged onset of disability beginning prior to the expiration of insured status. At each stage of this case, the Secretary has noted that the first two ALJ decisions were vacated and remanded by the Appeals Council. The magistrate judge erred, therefore, in accepting the vacated ALJ factual findings as undisputed fact, because the district court’s review is necessarily limited to the final decision of the Secretary (in this ease, the third ALJ decision and the affirming decision of the Appeals Council), although it may review the record as a whole to decide whether that final decision is supported by substantial evi *1458 dence. See Taylor v. Heckler, 765 F.2d 872, 875 (9th Cir.1985). Because the Secretary did not object to the magistrate judge’s recommendation on the specific grounds that the judge had accepted a vacated finding as undisputed fact, however, we deem that the Secretary has waived that argument for purposes of this appeal. Thus, we find it necessary to clarify the law regarding the applicability of the relation-back doctrine to a disability that recurs after the expiration of insured status. For the purposes of this appeal, we accept that Flaten is currently suffering from a second period of disability that commenced in July 1985. We further assume that this second period of disability is a recurrence of the same disabling disease from which Flaten suffered prior to the expiration of her insured status. Flaten asks that we interpret the disability provisions of the Act to incorporate a relation-back doctrine that would allow a claimant in her circumstances, who is now unable to work due to a physical impairment and whose insured status lapsed prior to the onset of this period of disability, to demonstrate statutory eligibility by linking the current disability to an earlier period of disability that began and ended before the expiration of her insured status. We reject the relation-back interpretation of the statutory eligibility provisions; instead, we hold that an individual cannot receive disability benefits for a recurrence of a disability, after a period of medical improvement when the individual was no longer disabled under the Act, unless the individual can establish that the current period of disability began on or prior to the expiration of insured status. Thus, we find that individuals who apply for benefits under the Act after the expiration of their insured status, for a disability that prevents substantial gainful activity at the time of the application, must show that the current disability has existed continuously since some time on or before the date that their insured status lapsed. If the claimant was not disabled on the last day that the claimant was insured under the Act, then the claimant cannot recover disability benefits for a new period of disability with an onset date after expiration of insured status, even if the claimant suffers a recurrence of the same disability from which the claimant suffered during a time when the claimant was eligible for disability benefits under the Act. Our decision follows directly from the language and structure of the Act and the purposes for which it was enacted, as well as from the relevant case law from this circuit and persuasive authority from other jurisdictions. The statutory provision of a transitional work period to enable those who suffer from disabilities to return to work when their disability ceases is especially relevant to cases involving recurring disabilities. 1. The Statutory Scheme The Social Security Act and amendments established a complex schema of statutory provisions and regulations that were designed both to provide protection for individuals who can no longer work because of disability and to encourage individuals who have previously suffered from a disability to return to substantial gainful employment when their medical condition improves sufficiently to allow them to do so. See, e.g., S.Rep. No. 1856, 86th Cong., 2d Sess. (1960), reprinted in 1960 U.S.C.C.A.N. 3608, 3623-25 (discussing the trial work period, rehabilitative services, and the interrelationship between work attempts and reentitlement to benefits for a recurring period of disability); Richard C. Ruskell, Social Security Disability Claims: Prac. & Proc. § 2-3, at 36 (3d ed. 1993) (“The term ‘disability’ can be interpreted to have two parts: a ‘medical’ part which is concerned with the nature and severity of a claimant’s impairment, and a ‘vocational’ part which is concerned with the availability of suitable work.”). For example, the vocational rehabilitation sections of the Act are expressly designed to return formerly disabled beneficiaries to the workforce. See 42 U.S.C. § 422 (rehabilitation and trial work provisions); Arthur Abraham & David L. Kopelman, Federal Social Security 103 (1979) (“Congress has expressly stated that its policy is the rehabilitation of the maximum number of disabled applicants for productive activity.”). *1459 The relation-back theory reflects a fundamental misunderstanding of this statutory schema, because it would entitle an individual to benefits for recurring periods of disability even though the individual did not work again after a first period of disability ended. This promise of perpetual benefits would serve as a disincentive to work even during long periods of medical improvement, thus undermining the statutory connection between work and disability. See, e.g., Cruise v. Harris, 510 F.Supp. 534, 535-36 (W.D.Mo.1981) (finding that a claimant with recurring mental illness who had periods of remission which lasted as long as a year or more was capable of employment during those periods and not disabled). Thus, although we note the frequently cited principle that the Act should be construed broadly to accomplish its remedial purpose of protecting disabled workers, see, e.g., Beecher v. Heckler, 756 F.2d 693, 696 n. 2 (9th Cir.1985) (citing Dorsey v. Heckler, 702 F.2d 597, 605 (5th Cir.1983)); Rivera v. Schweicker, 717 F.2d 719, 723 (2d Cir.1983), this court cannot ignore the important complementary purpose of providing an incentive for individuals to return to work at the end of a period of disability. Section 423 of the Act establishes the statutory eligibility requirements which an individual must satisfy to receive a disability insurance benefit. An individual is eligible who: (A) is insured for disability insurance benefits (as determined under subsection (c)(1) of this section), (B) has not attained retirement age ..., (C) has filed application for disability insurance benefits, and (D) is under a disability (as defined in subsection (d) of this section). inability to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment which can be expected to result in death or which has lasted or can be expected to last for a continuous period of not less than 12 months. 42 U.S.C. § 423(a)(l)(A)-(D). The meaning of the text is clear: these are conjunctive conditions that must be satisfied simultaneously for the period of disability currently claimed. Thus, the individual must be insured at the time that the individual suffers from the disability in order to receive benefit payments. Disability is further defined as an 42 U.S.C. § 423(d)(1)(A). This provision establishes a severity and durational requirement for recognition of disabilities under the Act that accords with the remedial purpose of the Act to protect workers whose disability prevents them from earning a living for a significant period of time. Flaten, however, contends that Section 423(d) merely establishes a measure of severity for recognition of a period of disability. She argues that, because “continuous” modifies the “12 month period” to create a minimum durational requirement, there is no other continuity requirement. By considering this provision outside of its context as a part of the overall statutory scheme, however, Flaten misconstrues the importance of the continuous “period of disability” that functions as the basic measure of disability benefit entitlements. Because individuals who suffer from a disability may later improve to the point that they no longer are disabled within the meaning of the Act and are again able to engage in substantial gainful activity, the statute defines a “period of disability” as the basis for the award of disability benefits. 42 U.S.C. § 416(i)(2). A period of disability begins on the day that a disability begins, if the individual is fully insured at that time. 42 U.S.C. §§ 416(i)(2)(C), 416(i)(3)(A); 20 C.F.R. § 404.131. See, e.g., Sprow v. Bowen, 865 F.2d 207, 208 (9th Cir.1989) (noting that the phrase “period of disability” is a “term of art” that means that the person is disabled, cannot work, and is insured). To have disability insured status during any quarter, an individual must be fully insured in that quarter and have at least twenty quarters of coverage in the last forty-quarter period ending with that quarter. 20 C.F.R. § 404.130(b). 1 In order to receive disability *1460 benefits, a person must apply for a period of disability. 42 U.S.C. § 416(i)(2)(B); 20 C.F.R. § 404.603. 2 Establishment of a period of disability also allows an individual to exclude that time from the number of years for which quarters of coverage are required, Sprow, 865 F.2d at 209, making it easier for an individual to satisfy the insured status requirement for a second period of disability at a later time. See, e.g., Abraham & Kopel-man, supra at 101-02 (discussing the disability freeze). The application must be filed within a specific statutory limitation period— either at any time while the period of disability continues 3 or within 12 months of the end of the period of disability. 20 C.F.R. § 404.320(b)(3). Disability benefits cease when the period of disability ends due to medical improvement (or for other reasons) and the individual is again able to do substantial work. 42 U.S.C. § 423(f); 20 C.F.R. §§ 404.321, 404.325, 404.1545(a), 404.1471. The SSA conducts “continuing disability reviews,” which are of varying frequency depending on the likelihood of improvement, to determine whether a period of disability has ended. 20 C.F.R. §§ 404.1588-1598. Individuals who earlier established a period of disability for which the Secretary terminated benefits because of a cessation of the disability may become reentitled to disability payments at a later date if they again satisfy all of the requirements of the Act. See, e.g., 20 C.F.R. §§ 404.250-404.242 (special computation rales for people who had an earlier period of disability and become reentitled to payments at a later date); 20 C.F.R. § 404.315 (individuals who were previously entitled to disability benefits need not satisfy the waiting period requirement when they apply for disability benefits based on a current disability for which they are insured); 20 C.F.R. § 404.1592(2)(ii) (no trial work periods for individuals receiving disability benefits in a second period of disability for which there was no waiting period). The statutory schema thus requires that a disability be continuously disabling from the time of onset during insured status to the time of application for benefits, if an individual applies for benefits for a current disability after the expiration of insured status. This is consistent with the Secretary’s interpretation of the statute, to which we defer if, as in this case, there is not a clear congressional intent to the contrary and the Secretary’s interpretation is a permissible construction of the statute. Chevron, U.S.A., Inc. v. Natural Resources Defense Council, 467 U.S. 837, 842-43, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984). 2. The Case Law Flaten’s novel theory for the interpretation of this statutory scheme finds little support in the case law. Flaten primarily relies upon an opinion of the Colorado district court, which explicitly adopted the relation-back theory. Cassel v. Harris, 493 F.Supp. 1055 (D.Colo.1980). Flaten contends that the Ninth Circuit impliedly adopted the Cassel approach in Goerg v. Schweiker, 643 F.2d 582 (9th Cir.1981). In Goerg, we remanded to the district court for consideration of new evidence that the claimant suffered from an inoperable carcinoma, noting that “[t]here is little question that the plaintiff would be eligible for benefits if his carcinoma had existed prior to the date on which his eligibility for disability insurance expired.” Id. at 583. Similarly, Flaten notes that the district court in Valentine v. Schweiker, 559 F.Supp. 644 (D.Mont.1983), suggested in dictum that Goerg permitted a current disability to relate back to an earlier period of disability. See id. at 649. We find these precedents unpersuasive. The Valentine court did not reach the relation-back issue, because it found that the claimant had established a continuous disability through retrospective diagnosis. Id. *1461 The Goerg statement is consistent with a requirement of continuous disability, because it dealt with the presence of an inoperable carcinoma prior to the lapse of the claimant’s insured status and in combination with other impairments. In any event, the Ninth Circuit has never explicitly adopted the relation-back theory. Nor has the Tenth Circuit adopted the Cas-sel holding. See, e.g., Potter v. Secretary of Health & Human Servs., 905 F.2d 1346, 1348-49 (10th Cir.1990) (finding that a retrospective diagnosis must establish actual disability prior to expiration of insured status in order to entitle a claimant to benefits). Although one commentator has endorsed the relation-back theory for latent disabilities, see Harvey L. McCormick, Social Security Claims and Procedures § 401, at 333 (4th ed. 1991), that authority relies solely on an early New York district court case, Mann v. Richardson, 323 F.Supp. 175 (S.D.N.Y.1971), which considered the latent condition in combination with other ailments. The Second Circuit later rejected the relation-back theory to find that a claimant who had failed to demonstrate continuous disability from a time prior to the expiration of his insured status was ineligible for disability benefits. See Arnone v. Bowen, 882 F.2d 34 (2d Cir.1989). Nor are the many cases making disability determinations on the basis of a retrospective diagnosis 4 dispositive oh this issue, as Flaten claims. Because the SSA and courts focus on the onset date of the current period of disability in order to decide whether an individual is eligible for benefits for a period of disability that existed on or before the. expiration of insured status, Flaten interprets these cases to buttress her theory that later disabilities may relate back to earlier ones, suggesting that an earlier period of disability satisfies the requirement as stated in the cases. Although these eases could be read to support Flaten’s “prior disability suffices” theory by ignoring the overall statutory schema, these cases are more appropriately interpreted to impose a continuity requirement when a claimant applies for benefits for a current disability after expiration of insured status. The claimant is eligible for coverage only if the current period of disability extends back continuously to an onset date prior to the expiration of insured status. The claimant may establish such continuous disabling severity by means of a retrospective diagnosis. 5 Our interpretation of the statute' to include a continuity requirement is in agreement with all other appellate courts that have considered the issue of recurring disabilities. *1462 As the Third Circuit noted, these courts have concluded that “an individual who recovered from a prior disability is not entitled to disability benefits unless he is still insured under the Social Security Act.” Golofski v. Califano, 607 F.2d 1063, 1068 n. 11 (3d Cir.1979); see also Arnone v. Bowen, 882 F.2d 3'4, 41 (2d Cir.1989) (requiring a claimant to “demonstrate that [the disabling] condition had been continuous from the time his insured status lapsed” because “suffering from a ‘relapse’ of an earlier disability ... does not entitle one to disability insurance benefits”); Poulin v. Bowen, 817 F.2d 865, 872, 875 (D.C.Cir.1987) (finding that a schizophrenic claimant must show either that symptom-free intervals were so likely to result in relapse from stress that he remained disabled and could not work or, if he did not remain disabled, that he became disabled again for a continuous period that began before the expiration of his insured status); Henry v. Gardner, 381 F.2d 191, 196 (6th Cir.) (finding that a claimant must “prove that he was continuously under some disability from the date that his insured status terminated” in order to “seek benefits from an alleged present disability”), cert. denied, 389 U.S. 993, 88 S.Ct. 492, 19 L.Ed.2d 487 (1967). Interpreting analogous provisions for child disability benefits, the Seventh Circuit reached the same result, holding that a claimant is entitled to benefits only if the disability is one “which began prior to age 22 [the time at which insured status lapsed] and continued up to the time of application.” Reading v. Mathews, 542 F.2d 993, 997 (7th Cir.1976). 3. Relapse and the Trial Work Period Provisions Flaten contends that interpreting the Act to require continuity places individuals with recurring disabilities such as hers at a disadvantage if they unsuccessfully attempt to return to work but must quit before they have again satisfied the special earnings requirement. Again, this argument represents a fundamental misunderstanding of the interaction of the various provisions of the Act. The 1980 amendments to the Act, for example, provide a trial work period as an incentive for individuals with disabling impairments to try to return to work. See Pub.L. No. 96-265, 94 Stat. 453, 457 (1980) (codified as amended at 42 U.S.C. § 422(c) (1991)). Work and earnings during the trial work period, which may last up to nine nonconsecutive months, are disregarded in determining whether the individual’s disability ceased during the trial work period. See 42 U.S.C. § 422(e); McCormick, supra § 409. An extended period of reentitlement begins immediately following the completion of nine months of trial work and continues for an additional thirty-six months: during that time, a person would be reentitled to benefits, without filing a new application for a new disability determination, for those months for which his or her earnings fell below the level of substantial gainful activity. 42 U.S.C. § 423(e)(1); 20 C.F.R. § 404.1592a. Because the regulations provide for assessments of both ability to work and medical improvement, individuals whose disabilities include periods of remission will lose their disabled status only if they are able to engage in substantial gainful activity. See 20 C.F.R. § 404.1594; see, e.g., Hemphill v. Weinberger, 483 F.2d 1137, 1139 (5th Cir.1973) (finding that a claimant who suffers temporary attacks of acute arthritis is not disabled when capable of gainful employment during periods of remission). Unpredictable temporary remissions generally will not be considered indicative of medical improvement. 20 C.F.R. § 404.1594(c)(3)(iv); see, e.g., Wilcox v. Sullivan, 917 F.2d 272 (6th Cir.1990) (finding that an individual diagnosed with multiple sclerosis was continuously disabled from the onset date despite intermittent periods of remission that were likely to end under the additional stress of work). We hold, therefore, that claimants who apply for benefits for a current disability after the expiration of their insured status must prove that the current disability has existed continuously since a date on or before the date that their insurance coverage lapsed. Accordingly, we affirm the district court’s refusal to apply the relation-back doctrine to Flaten’s case. B. The Continuous Disability Claim Flaten also contends that she is entitled to disability benefits because she has been con *1463 tinuously disabled since April 1976. Flaten acknowledges that she carries the burden to prove the existence of a disability within the meaning of the Act, see Swanson v. Secretary of Health and Human Servs., 763 F.2d 1061, 1065 (9th Cir.1985), but she claims that the report from Dr. Joem satisfies this burden. She argues that the Secretary failed to provide clear and convincing reasons for disregarding the treating physician’s opinion letter. See Montijo v. Secretary of Health & Human Servs., 729 F.2d 599, 601 (9th Cir.1984) (although the ALJ “is not bound by the uncontroverted opinions of the claimant’s physicians on the ultimate issue of disability, ... he cannot reject them without presenting clear and convincing reasons for doing so”) (citing Rhodes v. Schweiker, 660 F.2d 722, 723 (9th Cir.1981) and Day v. Weinberger, 522 F.2d 1154, 1156 (9th Cir.1975)). Flaten contends that Rhodes, which requires the court to overturn an AL J’s decision when the ALJ rejects a treating physician’s opinion solely because of internal contradictions in the opinion’s reference to some improvement in the claimant’s condition, is dispositive in this case. The Secretary counters that Dr. Ingraham’s records do not support a finding of continuous disability and Dr. Joern’s earlier medical reports contradict his June 1989 disability opinion. When there is conflicting medical evidence, the Secretary need only set forth “specific, legitimate reasons” constituting substantial evidence for disregarding a treating physician’s opinion. Magallanes v. Bowen, 881 F.2d 747, 751 (9th Cir.1989) (finding that the ALJ must provide a summary of the facts and conflicting medical evidence forming the basis for the decision); accord Swanson, 763 F.2d at 1064; Murray v. Heckler, 722 F.2d 499, 502 (9th Cir.1983). “ ‘[Questions of credibility and resolutions of conflicts in the testimony are functions solely of the Secretary.’ ” Sample v. Schweiker, 694 F.2d 639, 642 (9th Cir.1982) (quoting Waters v. Gardner, 452 F.2d 855, 858 n. 7 (9th Cir.1971)). The ALJ found that Flaten suffered a severe impairment due to her degenerative disc disease which resulted in a period of disability that continued from April 1976 through February 3, 1978; however, he concluded that Flaten was not disabled as defined by the Act after her 1978 recovery and prior to the July 1985 recurrence. The ALJ provided a detailed written opinion that summarized the facts, his interpretation of the facts, and his reasons for those interpretations. We find these conclusions persuasive, for the following reasons. First, the ALJ pointed to substantial conflicting medical evidence supporting his decision to disregard Dr. Joern’s report regarding Flaten’s condition prior to the July 1985 onset of a second period of disability. See Swanson, 763 F.2d at 1064-65 (finding an ALJ justified in rejecting the treating physician’s opinion in a case “replete with conflicting evidence as to the appellant’s condition prior to the ... onset date”). The ALJ credited the medical evidence provided by Dr. Ingraham, the treating physician during the period of the initial surgery and post-surgery recuperative period, whose notes several months after the surgery indicated that Flaten was doing well and should be able to return to work in three months. 6 The ALJ noted that Dr. Joern’s 1989 opinion was solicited after Flaten had been denied benefits in 1988, and thus was “all the less persuasive.” Weetman v. Sullivan, 877 F.2d 20, 23 (9th Cir.1989) (rejecting a treating physician’s opinion because it was inconsistent with contemporaneous medical notes and because the claimant obtained the opinion after an adverse ALJ determination). Although the 1989 letter summarized the progressive nature of the disease and provided substantial evidence that the condition had become disabling by the time of the second surgery, the ALJ found that the retrospective opinion was inconsistent with medical records that Dr. Joern had prepared at the time of his initial examinations in 1986, in which he noted that Flaten had good results following the first surgery but had suffered a recent onset of increased back and leg pain. It is also significant that Dr. Joern’s assessments were based on Flaten’s *1464 own reports of the severity of her condition. See Fair v. Bowen, 885 F.2d 597, 605 (9th Cir.1989) (finding that an opinion of disability “premised to a large extent upon the claimant’s own accounts of his symptoms and limitations” may be disregarded where those complaints have been “properly discounted”); Brawner v. Secretary of Health & Human Servs., 839 F.2d 432, 433-34 (9th Cir.1988). Second, the ALJ considered it significant that there were only two occasions for which Flaten found it necessary to seek medical assistance for brief episodes of back pain during the years after the first surgery and before the second laminectomy: a consultation with a chiropractor in 1980 following a week of back pain, and consultations over a period of one week in 1982. Flaten sought no further care for complaints related to her back problem until mid-1985 when a recurrence of her back pain led to multiple physician visits and, ultimately, to the second surgery in mid-1986. The ALJ discounted Fla-ten’s claim that lack of money prevented her from seeking help for ongoing problems, because she sought appropriate medical care from a Dr. Matthews for other medical symptoms, such as abdominal pain and gall bladder problems, during the intervening years. Reports from those visits do not include any complaints related to back pain or to her surgery. Although the absence of references to a claimant’s condition are not as important as medical and lay testimony, Smith v. Bowen, 849 F.2d 1222, 1226 (9th Cir.1988), the ALJ was entitled to draw an inference from the general lack of medical care for back problems during the intervening years between the two surgeries, Fair, 885 F.2d at 603 (finding it appropriate to consider “an unexplained, or inadequately explained, failure to seek treatment”). Third, the ALJ did not find Flaten’s testimony credible for the period from 1978 to 1985 when there was no supporting medical evidence for a continuing disability. Flaten’s subjective testimony of incapacitation and the progressive nature of degenerative disc disease indicate that she suffered some impairments prior to the July 1985 onset of the second period of disability. The ALJ, however, may disregard unsupported, self-serving statements, Taylor v. Heckler, 765 F.2d 872, 876 (9th Cir.1985), and this court will defer to the ALJ’s credibility determinations when they are appropriately supported in the record by “specific findings justifying that decision,” Cotton v. Bowen, 799 F.2d 1403, 1407 (9th Cir.1986); see also Bunnell v. Sullivan, 947 F.2d 341, 342 (9th Cir.1991) (en banc) (reaffirming the Cotton standard); Magallanes, 881 F.2d at 755 (finding that thé ALJ must make specific findings justifying the decision to discredit a claimant’s testimony that excess pain rendered her unable to engage in substantial gainful activity). Here, the relative absence of medical problems during the years between surgeries, Dr. Ingraham’s records indicating that Flaten had a reasonable result from the surgery, and Dr. Joern’s medical histories indicating that Flaten considered herself to have had a good response to the surgery and that she managed without medication for her back pain until the 1985 recurrence fully support the ALJ’s decision to discredit Flaten’s testimony and find that her impairment during the intervening period had not rendered her disabled before the expiration of her insured status in March 1982. See Swanson, 763 F.2d at 1065 (finding that the ALJ properly rejected the claimant’s subjective complaints based on negative clinical findings and the fact that the claimant stopped taking medication, and properly disregarded the treating physician’s report because of consistently negative clinical findings and findings that the claimant’s ailment became apparent after expiration of insured status). Fourth, the ALJ appropriately discharged his burden of proof that Flaten could engage in other substantial gainful activity that exists in the national economy even though she was unable to perform her previous work. See 42 U.S.C. § 423(d)(2)(A); 20 C.F.R. § 404.1520(e); accord Bowen v. Yuckert, 482 U.S. 137, 141, 107 S.Ct. 2287, 2291, 96 L.Ed.2d 119 (1987). Because the ALJ had determined that Flaten retained the residual functional capacity to engage in sedentary work, he applied the medical-vocational guidelines set forth in the regulations to find *1465 that Flaten was not disabled as defined in the Social Security Act before the expiration of her insured status. See 20 C.F.R. §§ 404.1564, 404.1569 and 404, subpt. P, app. 2, tbl. 1, rule 201.19. Use of the guidelines was appropriate here. See Blacknall v. Heckler, 721 F.2d 1179, 1180-81 (9th Cir.1983). Therefore, we find that substantial evidence supports the Secretary’s conclusion that Flaten, though physically impaired, was not continuously disabled from February 1978 to March 31, 1982, when her insured status lapsed. IY. Conclusion Although we sympathize with individuals like Flaten who earnestly believe that their disability prevented their return to work, we cannot disregard the Social Security Act’s eligibility requirements by adopting a theory such as relation-back that essentially undoes the insured-status link between periods of work and periods of disability benefits. We conclude that the Secretary appropriately interpreted the statute to require onset of the current continuous period of disability prior to expiration of insured status. We also conclude that the Secretary properly discounted the treating physician’s retrospective diagnosis, and that substantial evidence supports the Secretary’s decision that Flaten was not disabled under the Act at the time her insured status expired. Thus, we affirm the district court’s decision upholding the Secretary’s ruling that Flaten is ineligible for disability benefits. Footnotes [1] The statute also provides special rules for determining insured status for those who become *1460 disabled before the age of 31, or who have already had a period of disability before age 31, or who are statutorily blind. 20 C.F.R. § 404.130(c)-(e). [2] Note that Flaten did not apply to establish a period of disability for 1976-77. [3] If an application is filed after the onset of disability and the person is eligible for benefits, the person will be entitled to benefits for a period beginning at most 12 months before the application was filed. 20 C.F.R. § 404.603(b). [4] In a retrospective diagnosis case, a "[c]laimant is not entitled to disability benefits unless he can demonstrate that his disability existed prior to the expiration of his insured status.” Cruz Rivera v. Secretary of Health & Human Servs., 818 F.2d 96, 97 (1st Cir.1986) (citations omitted), cert. denied, 479 U.S. 1042, 107 S.Ct. 903, 93 L.Ed.2d 854 (1987); accord Ivy v. Sullivan, 898 F.2d 1045, 1048 (5th Cir.1990); Higgs v. Bowen, 880 F.2d 860, 862 (6th Cir.1988); Martonik v. Heckler, 773 F.2d 236, 238 (8th Cir.1985); Jeralds v. Richardson, 445 F.2d 36, 38 (7th Cir.1971); Flack v. Cohen, 413 F.2d 278, 279 (4th Cir.1969). The Ninth Circuit has followed this same approach in a long line of cases beginning with Waters v. Gardner, 452 F.2d 855 (9th Cir.1971), in which the court held that a claimant must establish a disability as of the date when the claimant last met the Act’s prior employment criteria. The court noted that “[a]ny deterioration in her condition subsequent to that time is, of course, irrelevant.” Id. at 858; see, e.g., Penny v. Sullivan, 2 F.3d 953, 956 (9th Cir.1993) (finding that claimant, who suffered from degenerative disc disease, must show disability prior to the last date that he met the special earnings requirement); Morgan v. Sullivan, 945 F.2d 1079, 1080 (9th Cir.1991) (finding that the claimant must demonstrate disability prior to his last insured date); Weetman v. Sullivan, 877 F.2d 20, 22 (9th Cir.1989) (affirming the Secretary’s finding that the claimant was not disabled before the expiration of her insured status); Vincent ex rel. Vincent v. Heckler, 739 F.2d 1393, 1394 (9th Cir.1984) (finding that "only disabilities existing before that time [expiration of insured status] can trigger insurance benefits"); Harmon v. Finch, 460 F.2d 1229, 1231 (9th Cir.) (requiring establishment of disability on or before the last date of insured status), cert. denied, 409 U.S. 1063, 93 S.Ct. 571, 34 L.Ed.2d 516 (1972). [5] Retrospective diagnoses by treating physicians and medical experts, contemporaneous medical records, and testimony from family, friends, and neighbors are all relevant to the determination of a continuously existing disability with onset prior to expiration of insured status. See, e.g., Hartman v. Bowen, 636 F.Supp. 129 (N.D.Cal.1986) (discussing contemporaneous medical reports, retrospective diagnosis, and testimony from family and friends). [6] On October 13, 1977, Dr. Ingraham noted that "[t]he patient is doing quite well.... I think she may return to work in three months. We would like to see her again in one year.”