CARDAMONE, Circuit Judge:

Albert Shaw (claimant or plaintiff) appeals from two judgments of the United States District Court for the Eastern District of New York (Spatt, J.), entered on March 21, 1996 (1996 judgment) and on March 27, 1999 (1999 judgment), Shaw v. Apfel, No. CV 97-3820 (E.D.N.Y. March 27, 1999). The 1996 judgment denied claimant mandamus review of an administrative decision refusing to consolidate his claim for Social Security Disability (SSD) benefits with his claim for Supplemental Security Income (SSI) benefits, Shaw v. Shalala, No. CV 94-3928 (E.D.N.Y. March 21, 1996). Because the two applications were not filed at the same time, the district court ruled that the Administrative Law Judge's (ALJ) refusal to consolidate the claims was not an abuse of his discretion sufficient to invoke mandamus. The 1999 judgment affirmed the ALJ's conclusion that claimant did not qualify for SSD benefits because he was not "disabled" during the relevant time period. Shaw appeals from both judgments.

To put the circumstances of this appeal in clear perspective, we analogize its facts to those at a gaming table in a scenario where the SSA is dealer and complainant is playing against the house. In this scenario the dealer did not put all its cards on the table and the player, seeing the cards the dealer showed and thinking he had a winning hand, bet accordingly, only to learn later that the dealer had another, unseen card up its sleeve which, when it was played, gave the house a winning hand and the player a losing hand. Essentially, that is what befell claimant in the case before us.

AT BACKGROUND A. Prior Legal Proceedings

Claimant applied for SSI and SSD benefits after suffering a series of automobile accidents that he asserted disabled him. The application process resulted in his spending many years navigating through the Social Security Administration's labyrinthine application procedures. He first filed concurrent applications for SSD and SSI benefits with the Department of Health and Human Services on October 8, 1985. Both applications were denied. He reapplied on June 30, 1992 thinking that a single application form for SSI benefits would also serve as an application for SSD benefits because of the following caption printed in large, bold-face type on the SSI application

I am . . . applying for Supplemental Security Income and any federally administered State supplementation under title XVI of the Social Security Act, for benefits under the other programs administered by the Social Security Administration, and where applicable, for medical assistance under title XIX of the Social Security Act.

The Social Services claims representative who assisted Shaw with his application did not inform him otherwise. As a result, he did not fill out a separate application for SSD benefits when he filed for SSI benefits.

Upon Shaw's counsel becoming aware — shortly before his October 15, 1993 hearing before an ALJ — that the only form on his June 30, 1992 reapplication was for SSI benefits, he promptly submitted an application for SSD benefits to the ALJ on the day of the hearing. The ALJ awarded SSI benefits to Shaw, but refused to consolidate the two claims or even consider his application for SSD benefits. The Appeals Council affirmed the ALJ's decision on July 11, 1994.

Shaw then sought mandamus review in district court of the administrative refusal to consolidate asserting that the SSI application was an actual and/or protective filing for SSD benefits, and that the ALJ had a non-discretionary duty to decide his claims concurrently. This challenge to the administrative action was premised not only on the referred-to caption that appeared in bold print on the SSI application, but also on a policy statement contained in the Social Security Administration's (SSA) Program Operations Manual Systems (POMS) SI 00510.005B:

Policy — Social Security Benefits

. . .

Because all applications for title XVI payments [SSI benefits] are also applications for title II benefits [SSD benefits], adjudicators must take steps to determine whether the claimant is eligible for title II [benefits]. Unless an individual files a title II application concurrently with the title XVI application, the title II aspects of the title XVI application must always be closed out. (emphasis added)

The district court at first granted the Secretary's motion to dismiss pursuant to Fed.R.Civ.P. 12(b)(1) for lack of subject matter jurisdiction on the ground that plaintiff had not yet exhausted his administrative remedies in that no final decision had been reached on Shaw's still-pending disability claim. On June 2, 1997 the Appeals Council denied Shaw's request for review of the ALJ's finding that he was not "disabled" during the relevant time period to qualify for SSD benefits. When plaintiff again sought review in federal court, the district court in 1999 granted the Commissioner's motion for judgment on the pleadings, ruling plaintiff not "disabled" for the purposes of SSD benefits.

B. Medical Record Pertaining to Disability

We set forth the facts relating to the 1999 judgment that held Shaw not disabled. Formerly employed as a machinist with a high school education, Shaw was seriously injured in four separate auto accidents in 1977, 1981, 1986, and 1989. His injuries and impairments have been extensive, including pinched nerves, memory loss, cognitive defects, a herniated disc, a club foot, difficulty in climbing stairs without pain, muscle spasms, headaches, and dizziness. He has been treated primarily by one physician, Dr. Arminius Cassvan, who concluded that claimant was totally disabled. From time to time he has also been seen by Dr. Aubrey Lewis and Dr. Milford Blackwell, a board-certified neurologist.

Shaw last worked regularly in 1977 before his auto accident of that year. After that he was no longer able to work as a machinist. When he visited Dr. Cassvan for the first time it was found that his injuries limited his flexibility, stair climbing and ability to sit or stand for prolonged periods. Later physical examination revealed S1 root compression and extensive paraspinal muscle spasms. The treatment prescribed was physical therapy. After his 1981 accident, EMG testing found mild to moderate neuropathy and mild carpal tunnel syndrome. The treatment again was physical therapy. From March 1982 to March 31, 1985 claimant received no medical treatment. He testified that his no-fault insurance had run out and he could not afford to pay for further treatment. After the 1986 accident he was diagnosed as having a frozen shoulder and weakness in his right upper arm. Soft tissue calcification, limiting the motion of his right hip, was observed. An elastic knee brace and physical therapy was ordered. Following his 1989 accident, Shaw was treated by Dr. Meyer, an orthopedist, and referred by an auto insurance carrier to Dr. Blackwell, who determined that plaintiff had been disabled since April 12, 1989, the date of his accident.

At the time Shaw commenced the instant action in federal court, the Social Security Administration had already adjudged him "disabled" and therefore entitled to SSI benefits under Title XVI of the Social Security Act as of June 30, 1992, the date of his application. But the SSA ruled against plaintiff with respect to his claim for disability benefits because it found he was not disabled as of March 31, 1985, the date he was last insured for SSD insurance purposes. Plaintiff maintains he was disabled as of March 1982, three years earlier and, pursuant to the Social Security Act, 42 U.S.C. § 405(g), sought review of this final administrative determination. Thus, the specific issue the district court decided against plaintiff was that he was not "disabled" on March 31, 1985, the last date of his fully insured status.

In this consolidated appeal, Shaw appeals from the 1999 judgment and reinstates his earlier appeal from the 1996 judgment. Shaw's earlier appeal is relevant because if we reverse the 1999 judgment and determine that plaintiff is "disabled" for disability purposes, then the ALJ's refusal to decide both claims concurrently affects the date from which Shaw is entitled to receive such benefits.

DISCUSSION I The 1999 Judgment A. Eligibility for SSD Benefits

An applicant must be insured for disability insurance benefits to be eligible for SSD benefits. 42 U.S.C. § 423(a)(1)(A), (c)(1); 20 C.F.R. 404.101(a), 404.110 to 404.130, 404.130 to 404.133; Arnone v. Bowen, 882 F.2d 34, 37 (2d Cir. 1989). This translates into two requirements: the applicant must have 1) adequate social security earnings to be "fully insured," 20 C.F.R. 404.110 through 404.115; and 2) "disability insured status" in the quarter he became disabled or in a later quarter in which he was disabled, C.F.R. 404.131(a).

A district court may set aside the Commissioner's determination that a claimant is not disabled only if the factual findings are not supported by "substantial evidence" or if the decision is based on legal error. 42 U.S.C. § 405(g); Bubnis v. Apfel, 150 F.3d 177, 181 (2d Cir. 1998). Substantial evidence "means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion." Richardson v. Perales, 402 U.S. 389, 401 (1971) (quoting Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229 (1938)). The ALJ has an obligation to develop the record in light of the non-adversarial nature of the benefits proceedings, regardless of whether the claimant is represented by counsel. See Schaal v. Apfel, 134 F.3d 496, 505 (2d Cir. 1998); Echevarria v. Secretary of Health Human Servs., 685 F.2d 751, 755 (2d Cir. 1982). On appeal, we conduct a plenary review of the administrative record to determine if there is substantial evidence, considering the record as a whole, to support the Commissioner's decision and if the correct legal standards have been applied. See Schaal, 134 F.3d at 500-01.

The ALJ's decision must be guided by the relevant legal standards. To receive federal disability benefits, an applicant must be "disabled" within the meaning of the Social Security Act. See 42 U.S.C. § 423(a), (d). To show "disabled" status a claimant must establish "inability to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment . . . which has lasted or can be expected to last for a continuous period of not less than 12 months." Id. 423(d)(1)(A). The impairment must be of "such severity that he is not only unable to do his previous work but cannot, considering his age, education, and work experience, engage in any other kind of substantial gainful work which exists in the national economy." Id. 423(d)(2)(A).

Agency rules promulgated under the Act outline a five-step analysis to determine disability. See 20 C.F.R. 404.1520, 416.920. We tracked this methodology in DeChirico v. Callahan, 134 F.3d 1177, 1179-80 (2d Cir. 1998), as follows:

1. The Commissioner considers whether the claimant is currently engaged in substantial gainful activity.

2. If not, the Commissioner considers whether the claimant has a "severe impairment" which limits his or her mental or physical ability to do basic work activities.

3. If the claimant has a "severe impairment," the Commissioner must ask whether, based solely on medical evidence, claimant has an impairment listed in Appendix 1 of the regulations. If the claimant has one of these enumerated impairments, the Commissioner will automatically consider him disabled, without considering vocational factors such as age, education, and work experience.

4. If the impairment is not "listed" in the regulations, the Commissioner then asks whether, despite the claimant's severe impairment, he or she has residual functional capacity to perform his or her past work.

5. If the claimant is unable to perform his or her past work, the Commissioner then determines whether there is other work which the claimant could perform. The Commissioner bears the burden of proof on this last step, while the claimant has the burden on the first four steps.

Thus, to create an irrebuttable presumption of disability under the regulations, the claimant must either have a "listed impairment," 20 C.F.R. 404.1520(d), 416.920(d), or one that is "equal to" a listed impairment, id. 404.1520(d), 416.920(d) ("If you have an impairment(s) which . . . is listed in appendix 1 or is equal to a listed impairment(s), we will find you disabled without considering your age, education, and work experience.").

Accordingly, plaintiff maintained that his condition as of March 31, 1985 met or equaled the criteria for vertebrogenic disorders as defined by Listing 1.05(C), 20 C.F.R. § 404, Sbpt. P, App. 1, Pt. A, 1.05(C), which describes these impairments as disorders "persisting for at least 3 months despite prescribed therapy and expected to last 12 months," with

1. Pain, muscle spasm, and significant limitation of motion in the spine; and

2. Appropriate radicular distribution of significant motor loss with muscle weakness and sensory and reflex loss.

In performing the five-step analysis to determine whether Shaw was disabled the ALJ did not refer to the Listing that enumerates "severe impairments," yet expressly found that claimant did not "meet or equal the medical criteria of a listed impairment" and therefore failed to establish disability on March 31, 1985. Concluding that Shaw did not have a disability per se under step three of the analysis, the ALJ proceeded to step four, and found him capable of performing his past relevant work as a machinist, described as "light work," and "other light work."

Claimant contended on appeal that the ALJ erred in concluding that his impairment did not equal a listed impairment, and this occurred because the ALJ discounted his functional limitations and failed to follow the "treating physician rule" that required him to give controlling or at least greater weight to the opinion of Dr. Cassvan. See 20 C.F.R. 416.927(d)(2).

B. The Ruling of the District Court

The district court rejected both of these arguments. Although the ALJ did not specifically refer to Listing 1.05(c), the district court looked at other portions of the record before the ALJ and found his determination supported by substantial evidence. As evidence of his disabled condition, plaintiff had maintained that his treating physician, Dr. Cassvan, found "all other elements of 1.05(C), including a reduced ankle jerk on the right, paraspinal muscle spasm, reduced range of motion in the back, positive straight leg raising, weakness in the toes, gait irregularity and on several occasions, positive Lasague and Patrick tests." But the district court disposed of Dr. Cassvan's medical findings by reasoning that the three-year period during which plaintiff obtained no medical treatment — from January 1982 through December 1985 — undermined his contention of disability. It further found that Dr. Cassvan's observations of Shaw's disability were too remote in time to be dispositive, and therefore provided the ALJ with "substantial evidence" to conclude that plaintiff was not disabled in 1985.

We find this logic troubling for several reasons. While it is conceivable that a three-year gap in plaintiff's medical treatment might be part of a more extensive inquiry into whether he was in fact disabled, the fact of this time lapse does not negate the compelling evidence in the record as a whole that plaintiff was completely disabled. Given the many times plaintiff was treated between his 1977 accident and March 1982, and that his condition did not improve, it was not unreasonable for him to discontinue those treatments, particularly in light of his testimony that he could not afford further medical care. See Gamble v. Chater, 68 F.3d 319, 321 (9th Cir. 1995) (citing cases from other circuits) (disabled claimant cannot be denied benefits for failing to obtain treatment that he cannot afford); Gordon v. Schweiker, 725 F.2d 231, 237 (4th Cir. 1984) ("[SSD] and SSI benefits exist to give financial assistance to disabled persons because they are without the ability to sustain themselves."). It would fly in the face of the plain purposes of the Social Security Act to deny claimant benefits because he is too poor to obtain additional treatment that had proved unhelpful.

The record shows that Dr. Cassvan was unable to recommend treatment other than physical therapy, and that because Shaw was not experiencing a significant reduction in pain from the physical therapy sessions, he essentially abandoned them. Just because plaintiff's disability went untreated does not mean he was not disabled. The facts that his condition did not improve, and that there was no suitable treatment other than physical therapy, bolster the argument that plaintiff's impairments were permanent and that he was unlikely to recover from them.

In addition to Dr. Cassvan's pre-1985 observations that the district court discounted, Shaw also offered his December 1985 X-rays that revealed a narrowing of the interspace area at L5-S1 as proof of his disability in 1985. See Perez v. Chater, 77 F.3d 41, 48 (2d Cir. 1996) ("A treating physician's retrospective medical assessment of a patient may be probative when based upon clinically acceptable diagnostic techniques."). Dr. Lewis noted that these X-rays could indicate a deteriorating problem in the spine, which is an example of a vertebrogenic disorder. The district court refused to give weight to this evidence because it concluded that Dr. Lewis' observations did not constitute a true retrospective diagnosis and therefore was not probative of plaintiff's disability.

It is not clear to us why Dr. Lewis' December 1985 observations fail to provide a "true retrospective diagnosis." The district court relied on cases where the utility of retrospective diagnoses was limited because the record already contained ample evidence of the claimant's condition at the relevant time period, see Pratts v. Chater, 94 F.3d 34, 36 (2d Cir. 1996), or when the claimant sought the introduction of new evidence after the ALJ had reached a decision, see Jones v. Sullivan, 949 F.2d 57, 60 (2d Cir. 1991). Since neither of these circumstances is present in the instant case, there was no basis to discount Dr. Lewis' observation of plaintiff's condition.

C. ALJ's Refusal to Follow Treating Physician Rule

Plaintiff appears to rely on Dr. Lewis' observations, in part, because the ALJ gave little weight to the probative value of plaintiff's own treating physician's diagnosis. By failing to give Dr. Cassvan's opinions controlling or at least greater weight, the ALJ violated the SSA regulation's "treating physician" rule. That rule mandates that the medical opinion of a claimant's treating physician is given controlling weight if it is well supported by medical findings and not inconsistent with other substantial record evidence. It provides:

Generally, we give more weight to opinions from your treating sources . . . [i]f we find that a treating source's opinion on the issue(s) of the nature and severity of your impairment(s) is well-supported by medically acceptable clinical and laboratory diagnostic techniques and is not inconsistent with the other substantial evidence in your case record, we will give it controlling weight.

20 C.F.R. 416.927(d)(2); see Rosa v. Callahan, 168 F.3d 72, 78-79 (2d Cir. 1999); Clark v. Commissioner of Soc. Sec., 143 F.3d 115, 118 (2d Cir. 1998).

In deciding whether a treating physician's view should be accorded controlling weight, the ALJ must weigh: "(i) the frequency of examination and the length, nature, and extent of the treatment relationship; (ii) the evidence in support of the opinion; (iii) the opinion's consistency with the record as a whole; and (iv) whether the opinion is from a specialist." Clark, 143 F.3d at 118. The regulations also require the ALJ to set forth her reasons for the weight she assigns to the treating physician's opinion. See id.

The district court believed that the ALJ properly decided not to give controlling weight to Dr. Cassvan's medical opinion because his "limited findings and the intermittent nature of his treatment" did not support a finding of disability. Such falls far short of the standard for contradictory evidence required to override the weight normally assigned the treating physician's opinion. Dr. Cassvan had treated plaintiff for at least seven years and had made medical observations far more extensive than those of any other consulting physician. The record contains no indication that Dr. Cassvan's observations were unsupported by medical evidence or that his opinion is inconsistent with the record as a whole. In fact, the bulk of the record in this case is drawn exclusively from Dr. Cassvan's medical assessments.

For the ALJ to conclude that plaintiff presented no evidence of disability at the relevant time period, yet to simultaneously discount the medical opinion of his treating physician, violates his duty to develop the factual record, regardless of whether the claimant is represented by legal counsel. See Schaal, 134 F.3d at 505 ("[E]ven if the clinical findings were inadequate, it was the ALJ's duty to seek additional information from [the treating physician] sua sponte."); Pratts, 94 F.3d at 37 ("[T]he rule in our circuit [is] that `the ALJ, unlike a judge in a trial, must [her]self affirmatively develop the record' in light of `the essentially non-adversarial nature of a benefits proceeding'. . . . [E]ven when, as here, claimant is represented by counsel.") (quoting Echevarria, 685 F.2d at 755).

In addition, the district court improperly characterized the fact that Dr. Cassvan recommended only conservative physical therapy, hot packs, EMG testing — not surgery or prescription drugs — as substantial evidence that plaintiff was not physically disabled during the relevant period. Neither the trial judge nor the ALJ is permitted to substitute his own expertise or view of the medical proof for the treating physician's opinion. See Balsamo v. Chater, 142 F.3d 75, 81 (2d Cir. 1998). Essentially, the ALJ and trial court imposed their notion that the severity of a physical impairment directly correlates with the intrusiveness of the medical treatment ordered. This is not the overwhelmingly compelling type of critique that would permit the Commissioner to overcome an otherwise valid medical opinion. See Wagner v. Secretary of Health and Human Servs., 906 F.2d 856, 862 (2d Cir. 1990) (explaining that while a physician's opinion might contain inconsistencies and be subject to attack, "a circumstantial critique by non-physicians, however thorough or responsible, must be overwhelmingly compelling in order to overcome a medical opinion").

Significantly, both the ALJ and the district court rejected Dr. Cassvan's medical opinion when it supported a finding that plaintiff was disabled; yet at the same time relied on Dr. Cassvan's January 1986 and 1987 observations, that plaintiff's condition was improving to provide proof that plaintiff was not disabled in 1985. Such an inconsistent use of the medical evidence undermines any argument that Dr. Cassvan's opinion was so unreliable that it should not have been assigned controlling weight.

Given these deficiencies in the ALJ's analysis the record lacked substantial evidence to support its finding that Shaw was not disabled on March 31, 1985. A remand for a step five analysis that places the burden of proof on the Commissioner to show that the claimant could perform other work in the economy, even if he could not perform his past work, is appropriate in cases where there is more uncertainty regarding the claimant's condition. But here because the record provides overwhelming proof that Shaw suffered from a vertebrogenic disorder — one of the "severe impairments" enumerated by Listing 1.05(C) — as of March 31, 1985, rendering him "disabled" for the purposes of SSD benefits, a remand for further proceedings is unnecessary.

II The 1996 Judgment

Having determined to reverse the 1999 judgment, we pass to the 1996 judgment which denied Shaw mandamus relief. Shaw raises a claim of mandamus under 28 U.S.C. § 1361, requesting us "to compel the ALJ to treat the SSI application of June 30, 1992 as a concurrent application for SSD." Although the date on which plaintiff filed for SSD benefits does not affect our earlier inquiry into whether plaintiff was disabled in 1985, it does bear on the amount of retroactive benefits he will receive as a result of being disabled.

The thrust of Shaw's argument is either the caption on the SSI application form stating that it was an application for benefits under other Social Security programs, or the POMS excerpt discussed above, or both together, obligated the ALJ (1) to consider Shaw's June 30, 1992 application as both an SSI application and an SSD application, and (2) to consolidate the two applications for hearing and decision. Because the applications have already been heard and decided at the ALJ level and we are reversing the denial of SSD benefits, only the effective date of filing the SSD application remains to be resolved.

On appeal, the government concedes that "if the Commissioner had awarded plaintiff [SSD] benefits, the [SSD] application would have been given the earlier filing date of the SSI application." (emphasis added). There therefore appears to be no dispute between the parties as to the correct date.

The correct filing date is indeed June 30, 1992. The government argues that the relevant social security regulation, 20 C.F.R. 416.1452(a), grants the ALJ discretion as to whether to hold a consolidated hearing on the SSI and SSD claims, but says nothing about the effective date of filing for SSD. The government does not dispute that POMS governs the filing date. The manual states that "applications for [SSI] payments are also applications for [SSD] payments." POMS SI 00510.005(B)(2)(b).

Even if POMS did not govern this question, the caption on the SSI form would entitle Shaw to relief under 42 U.S.C. § 402(j)(5). That statute provides:

In any case in which it is determined to the satisfaction of the Commissioner of Social Security that an individual failed as of any date to apply for monthly insurance benefits under this title by reason of misinformation provided to such individual by any officer or employee of the Social Security Administration relating to such individual's eligibility for benefits under this title, such individual shall be deemed to have applied for such benefits on the later of [the date the misinformation was given or the date the applicant became eligible for benefits apart from the application requirement].

Shaw did not file a separate SSD application on June 30, 1992 because he relied on the information provided by the Social Security Administration on its SSI application form. If the information that the SSA provided Shaw on that form was incorrect, he would be deemed to have applied for the benefits on the date he received the form.

The government's concession is therefore sound. Shaw is entitled to SSD benefits from his filing date of June 30, 1992. Because we are granting plaintiff the relief he seeks, we need not reach the question of mandamus.

III Stieberger Class Settlement

Finally, plaintiff argues on appeal that if we decide he is entitled to SSD benefits, his earlier claims for benefits filed in 1985 must be reopened pursuant to the class settlement agreement reached in Stieberger v. Sullivan, 792 F. Supp. 1376, modified, 801 F. Supp. 1079 (S.D.N.Y. 1992). We reject this argument.

The Stieberger settlement agreement was developed as a remedy for the effects of the SSA's "non-acquiescence policy," under which ALJs were essentially told to disregard the law of this Circuit on certain issues when adjudicating disability claims. The settlement agreement required the Commissioner to provide a New York statewide class with notice and opportunity to request the reopening of, and new decision on, disability claims that were denied or terminated at any administrative level between October 1, 1981 and October 17, 1985, and were decided at the administrative law judge or Appeals Council level between October 7, 1981 and July 2, 1992.

The denial of disability benefits by a federal district court is not reopenable under Stieberger under the rationale that even if the SSA had failed to apply Second Circuit law the district court would have done so. Plaintiff concedes this point, yet argues that our decision to reverse the district court's 1999 decision, which found that plaintiff was not "disabled" for the purposes of SSD benefits, would effectively revoke the result contemplated in Stieberger. This argument is unavailing because our decision to reverse the district court's 1999 decision — a district court decision that plaintiff was not "disabled" — no longer constitutes a denial of disability benefits. Instead, we are granting those benefits. Additionally, no Stieberger relief is necessary because we are awarding plaintiff the SSD benefits he seeks.

CONCLUSION

For the reasons stated, we reverse the 1999 judgment insofar as it failed to find Shaw disabled as of March 1985, and we instead find him disabled as of that date. We affirm the 1996 judgment insofar as it denied mandamus relief, but hold that the Secretary gave claimant misinformation that entitles him to relief under 42 U.S.C. § 402(j)(5) for SSD benefits retroactive to June 30, 1992, a date the government concedes is correct.