Plaintiff appeals the district court's affirmance of the Secretary's decision denying him disability insurance benefits.1 Plaintiff claims to have been disabled since April 1991 as a result of pain in his neck, shoulder, upper and lower back, and right leg, headaches, a catch in the middle finger of his left hand, tingling in his right hand, asthma, liver problems, hypoglycemia, hiatal hernia, depression, general anxiety disorder, and somatoform disorder. After conducting two hearings and a de novo review of the record, the administrative law judge (ALJ) determined that plaintiff could return to his past relevant work as a truck driver, as that work is generally performed in the national economy. When the Appeals Council denied review, the ALJ's decision became the final decision of the Secretary.

"We review the Secretary's decision to determine whether it is supported by substantial evidence and whether the Secretary applied the correct legal standards." Washington v. Shalala, 37 F.3d 1437, 1439 (10th Cir.1994). We must examine the record closely to determine whether substantial evidence supports the Secretary's determination. Id. Substantial evidence is "such relevant evidence as a reasonable mind might accept as adequate to support a conclusion." Richardson v. Perales, 402 U.S. 389, 401, 91 S. Ct. 1420, 1427, 28 L. Ed. 2d 842 (1971)(quotation omitted). In addition to a lack of substantial evidence, the Secretary's failure to apply the correct legal standards, or to show us that she has done so, are also grounds for reversal. Washington, 37 F.3d at 1439.

Plaintiff, who was sixty years old at the time the ALJ issued the present decision, worked as a truck driver for Roadway Express for twenty-one years. Plaintiff's job required him not only to drive a truck, but to load and unload cargo. His testimony reflected that he performed the job at an exertional level between medium and heavy. In 1988, plaintiff suffered a work-related injury to his neck and shoulder, for which he received worker's compensation. As a result of alleged increasing difficulty in performing his job, plaintiff quit in April 1991. In a decision issued August 1992, the ALJ determined that plaintiff could perform a full range of medium work, but that, in light of his treating physician's opinion that "the claimant's loading of trucks and pulling down and latching doors may represent a difficulty for the claimant given his limitations," Tr. at 241, plaintiff could not perform his past relevant work. The ALJ, therefore, proceeded to step five of the sequential analysis, see 20 C.F.R. § 404.1520 (setting forth the five steps), where he concluded that plaintiff was not disabled based on the grids.

The Appeals Council reversed the ALJ's decision and remanded the action to obtain further evidence concerning plaintiff's physical and mental limitations and to reevaluate the evidence under proper legal principles. On remand, the ALJ found that plaintiff had the physical ability to perform a full range of medium work, but that he had nonexertional limitations resulting from his mental impairments. The ALJ concluded that, despite these limitations, plaintiff could return to his past work as a truck driver--as distinguished from his past work as a loader and unloader--as that work is generally performed in the national economy. See Social Security Ruling (SSR) 82-61, Soc. Sec. Rep. Serv., Rulings 1975-1982, 836, 838 (West 1983); Andrade v. Secretary of Health & Human Servs., 985 F.2d 1045, 1050-51 (10th Cir.1993)(holding that "past relevant work" includes not only claimant's particular former job, but also claimant's former occupation as it is generally performed in the national economy).

Plaintiff alleges five categories of error in the Secretary's decision: improper evaluation of plaintiff's subjective complaints; improper evaluation of plaintiff's mental impairments; improper evaluation of plaintiff's residual functional capacity; improper evaluation of plaintiff's ability to return to his past relevant work; and improper application of the vocational expert's testimony. We will address each in turn.

"A claimant's subjective allegation of pain is not sufficient in itself to establish disability. Before the ALJ need even consider any subjective evidence of pain, the claimant must first prove by objective medical evidence the existence of a pain-producing impairment that could reasonably be expected to produce the alleged disabling pain." Thompson v. Sullivan, 987 F.2d 1482, 1488 (10th Cir.1993) (citations omitted). Plaintiff met this initial burden here. X-rays of plaintiff's cervical spine taken in 1991 showed marked degenerative changes, and x-rays of his left shoulder showed degenerative changes and a narrowing of the AC joint. A CAT scan of plaintiff's cervical spine in the fall of 1991 revealed "extensive osteoarthritis changes of the facet joints without evidence of [a] ruptured disk," Tr. at 206, and x-rays of plaintiff's lumbosacral spine showed "moderate lipping of osteoarthritis," but "no degenerative changes," id. at 213. The ALJ was then required to consider all the relevant objective and subjective evidence and "decide whether he believe[d] the claimant's assertions of severe pain," Luna v. Bowen, 834 F.2d 161, 163 (10th Cir.1987). The ALJ found that plaintiff's subjective complaints of pain were not credible to the extent they suggested he could not perform a full range of medium work.

"Findings as to credibility should be closely and affirmatively linked to substantial evidence...." Huston v. Bowen, 838 F.2d 1125, 1133 (10th Cir.1988). "Credibility determinations are peculiarly the province of the finder of fact, [however,] and we will not upset such determinations when supported by substantial evidence." Diaz v. Secretary of Health & Human Servs., 898 F.2d 774, 777 (10th Cir.1990).

Here, some of the reasons advanced by the ALJ for finding plaintiff's subjective complaints of pain incredible were not supported by substantial evidence. The most notable of these was the ALJ's determination that plaintiff had an incentive not to work. This finding was based entirely on the ALJ's speculation that the terms of plaintiff's pension might prohibit plaintiff from working. Conversely, the ALJ also failed to consider relevant factors that were supported by the record. The most notable of these was "the possibility that psychological disorders combine with physical problems," Luna, 834 F.2d at 166. Dr. Spray, a clinical psychologist, diagnosed plaintiff with somatoform disorder, not otherwise specified. He found that plaintiff manifested an "[u]nrealistic interpretation of physical signs or sensations associated with the preoccupation or belief that one has a serious disease or injury." Tr. at 266. Dr. Spray also noted that plaintiff was "extremely somatically preoccupied," id. at 308, and that, even if plaintiff were given only simple job instructions, he would be "likely to complain about somatic concerns," id. at 310. The ALJ, however, did not treat plaintiff's somatic preoccupation as a manifestation of his mental impairment or consider how that impairment, along with plaintiff's depression and anxiety disorder, affected his perception of pain. Instead, the ALJ discounted plaintiff's credibility based on his determination that "claimant is bored and whiny and has [an] incentive not to work and to complain about problems." Id. at 37. Thus, the ALJ's evaluation of plaintiff's subjective complaints was flawed by his reliance on factors that were not supported by the record and by his failure to consider other factors that were supported by the record.

Dr. Spray first examined plaintiff in October 1992, at which time he performed a mental status exam and administered several tests, including the Minnesota Multiphasic Personality Inventory-2 (MMPI-2). Dr. Spray detailed his findings in a report and completed a Psychiatric Review Technique (PRT) form. Dr. Spray diagnosed plaintiff with the following conditions: (1) somatoform disorder, not otherwise specified, (2) dysthymia, (3) history of alcohol abuse, and (4) personality disorder, not otherwise specified, with dependent and antisocial features.

In May 1993, plaintiff was examined by Dr. Dean, a psychiatrist, at the Secretary's request. Dr. Dean concluded that plaintiff suffered from a generalized anxiety disorder and major depression. In response to an inquiry from plaintiff's attorney as to the presence of a somatoform disorder, Dr. Dean wrote that he "was unable to adequately assess the presence or absence of objective findings which would either corroborate or eliminate the presence of a somatoform disorder," because he was not given complete data on plaintiff's medical care and he neither obtained a complete medical history from plaintiff nor conducted his own physical examination of plaintiff. Tr. at 314.

In connection with his examination of plaintiff, Dr. Dean completed a Medical Assessment of Ability to Do Work-Related Activities form, on which he indicated that plaintiff's abilities in seven areas were only "fair"2 and his abilities in eight areas were "good".3 Dr. Dean noted that "chronic anxiety and depression limits patient to more solitary vocations and limits his ability to relate to other people." Id. at 283.

Dr. Spray examined plaintiff again in September 1993. He found plaintiff "extremely somatically preoccupied," with "occult ideas about not only the causes of his perceived pain, but also treatment." Id. at 308. He also noted that plaintiff "easily got off on somatic tangents that were not related to the question asked of him," id. at 309, and found that plaintiff "has difficulty concentrating due to his somatic preoccupation," id. at 310. With one exception, Dr. Spray's diagnoses in 1993 were the same as those he made in 1992. The only change concerned plaintiff's alcohol abuse. In the 1992 report, Dr. Spray noted that plaintiff's statements about his alcohol use were inconsistent and, therefore, he was unclear when plaintiff had stopped using alcohol or if plaintiff had stopped completely. At the time of the exam in 1993, however, plaintiff "admit[ted] to continued use of alcoholic beverages, the last time approximately two days ago when he drank eight to ten beers," id. at 309. Therefore, Dr. Spray changed his diagnosis from a "history of alcohol abuse" to "alcohol abuse." In connection with his examination in 1993, Dr. Spray also completed a Medical Assessment of Ability to Do Work-Related Activities form, on which he concluded that plaintiff's abilities were "poor or none" in eight areas, "fair" in six areas, and "good" in only one area: maintaining personal appearance.

Dr. Goodman, a psychiatrist, acted as a medical adviser at the supplemental hearing. Because he had not examined plaintiff, his comments were based solely on his review of the record and plaintiff's testimony at the supplemental hearing. Dr. Goodman questioned Dr. Spray's diagnosis of somatoform disorder, because he thought there appeared to be physical reasons for plaintiff's somatic complaints. Otherwise, Dr. Goodman testified that the evidence revealed an individual who abuses alcohol and has somatic complaints, as well as dysthymia, a generalized anxiety disorder, and personality disorders that may meet the A criteria under the listings, see 20 C.F.R., Pt. 404, Subpt. P, App. 1, § 12.08. Dr. Goodman noted that plaintiff had not been treated for his dysthymia, which he described as mild chronic depression, but that the condition is "very treatable with medication," id. at 128. Dr. Goodman did not say, however, whether treatment could be expected to eliminate all effects of the condition on plaintiff's ability to do work-related activities.

Despite Dr. Spray's diagnosis, the ALJ found that plaintiff did not have a somatoform disorder. The ALJ gave three reasons for this finding: (1) Dr. Goodman's opinion that there were physical reasons for plaintiff's somatic complaints; (2) his own interpretation of "Dr. Dean's initial failure to consider this diagnosis [as] a reflection of his belief at the time that such diagnosis was not warranted," id. at 28; and (3) his own opinion that Dr. Spray improperly used the MMPI-2 as a basis for the diagnosis. None of these grounds justifies rejecting Dr. Spray's diagnosis.

First, because Dr. Goodman did not examine plaintiff, his opinion was not entitled to as much weight as that of Dr. Spray. 20 C.F.R. § 404.1527(d)(1); Talbot v. Heckler, 814 F.2d 1456, 1463 (10th Cir.1987). Dr. Spray's diagnosis was not inconsistent with the opinion of Dr. Dean, the only other mental health professional who examined plaintiff. Second, the ALJ's attempt to infer from Dr. Dean's failure to diagnose a somatoform disorder a determination by Dr. Dean that the diagnosis was not warranted is belied by Dr. Dean's own explanation for the absence of a diagnosis. Finally, the ALJ clearly overstepped his bounds when he substituted his medical judgment for that of Dr. Spray, by determining that the results of the MMPI-2 test were not an adequate basis on which to make a diagnosis. See Kemp v. Bowen, 816 F.2d 1469, 1476 (10th Cir.1987)(holding that an ALJ "can not interpose his own 'medical expertise' over that of a physician"). We note that Dr. Goodman never suggested that Dr. Spray's reliance on the results of the MMPI-2 test as a diagnostic tool was improper.

The ALJ rejected not only Dr. Spray's diagnosis of a somatoform disorder, but also Dr. Spray's conclusions about the effect of plaintiff's various mental impairments on his ability to work. Other than those discussed above, the only additional reason the ALJ advanced for rejecting Dr. Spray's opinion was that he did not think the evidence supported Dr. Spray's change of diagnosis from a history of alcohol abuse to alcohol abuse. The ALJ found that "Dr. Spray's knee-jerk diagnosis of alcohol abuse based on [plaintiff's report of having drunk eight to ten beers two days before the exam] appears to be somewhat flimsy in light of the remainder of the record." Tr. at 31. The ALJ found that the rate of consumption to which plaintiff testified at the hearing ("a few beers a week or two prior to the supplemental hearing") was not "excessive or abusive." Id.

Once again, the ALJ appears to have second-guessed a medical expert's judgment, by determining what rate of alcohol consumption should be considered excessive or abusive for plaintiff. Further, although plaintiff testified at the supplemental hearing that he had quit drinking four years ago and had drunk only a few beers in recent weeks, he had previously told Dr. Spray otherwise. In making his diagnosis, Dr. Spray was entitled to rely on the information plaintiff provided him, including that plaintiff continued to drink and had recently drunk as many as ten beers in one day. Moreover, rejecting Dr. Spray's diagnosis of current alcohol abuse would not justify rejecting the remainder of his opinion. Not only did Dr. Dean and Dr. Goodman corroborate other diagnoses by Dr. Spray, but all the explanations Dr. Spray provided for his answers on the Medical Assessment of Ability to Do Work-Related Activities form were based on plaintiff's other mental impairments, not his alcohol abuse.

In sum, the ALJ erred in rejecting Dr. Spray's opinions without adequate justification and in substituting his own medical judgment for that of mental health professionals. This, in turn, led the ALJ to make conclusions about the effects of plaintiff's mental impairments on his ability to work that did not fairly take into account all the evidence. If, on remand, the ALJ determines that the evidence about plaintiff's mental impairments is not sufficient to support a decision on plaintiff's claim, the ALJ can order another consultative exam, see 20 C.F.R. § 404.1519a(b), and, this time, provide the examining physician with all the medical records necessary to make an informed diagnosis.

Step four of the sequential analysis, at which the ALJ found plaintiff not disabled, is comprised of three phases. In the first phase, the ALJ must evaluate a claimant's physical and mental residual functional capacity (RFC), see SSR 86-8, Soc. Sec. Rep. Serv., Rulings 1983-1991, 423, 427 (West 1992), and in the second phase, he must determine the physical and mental demands of the claimant's past relevant work. 20 C.F.R. § 404.1520(e). In the final phase, the ALJ determines whether the claimant has the ability to meet the job demands found in phase two despite the mental and/or physical limitations found in phase one. See SSR 82-62, Soc. Sec. Rep. Serv., Rulings 1975-1982, 809; Henrie v. United States Dep't of Health & Human Servs., 13 F.3d 359, 361 (10th Cir.1993). At each of these phases, the ALJ must make specific findings. See Henrie, 13 F.3d at 361.

A. Phase One: Plaintiff's Residual Functional Capacity

In determining a claimant's physical abilities, the ALJ should "first assess the nature and extent of [the claimant's] physical limitations and then determine [the claimant's] residual functional capacity for work activity on a regular and continuing basis." 20 C.F.R. § 404.1545(b). Here, the ALJ concluded that plaintiff had the physical ability to do a full range of medium work. SSR 83-10 defines "full range of work" as "[a]ll or substantially all occupations existing at an exertional level," and provides that "[t]he considerable lifting required for the full range of medium work usually requires frequent bending-stooping ... [f]lexibility of the knees as well as the torso is important for this activity." Soc. Sec. Rep. Serv., Rulings 1983-1991, at 30.4 Dr. Dandridge, an orthopedic surgeon who examined plaintiff in 1993 at the request of the Secretary, concluded that plaintiff could climb, stoop, kneel, balance, crouch, and crawl only occasionally, commenting that, "[a]t this patient's age, flexibility and elasticity of tissues prevent frequency of activities." Tr. at 294. The ALJ credited Dr. Dandridge's opinion, but determined that it did not conflict with a determination that plaintiff could do a full range of medium work. The ALJ reasoned that, because kneeling and crouching are forms of bending, plaintiff's ability to stoop, kneel, and crawl occasionally meant that he could frequently bend-stoop. In her brief on appeal, the Secretary does not even attempt to justify this fallacious reasoning.

The ALJ's RFC determination also failed to take into account the physical limitations found by Dr. Combs, plaintiff's treating physician. Although the ALJ stated that he found entirely credible Dr. Combs' opinion that plaintiff would have difficulty loading cargo, pulling down doors and latching them, the ALJ did not include any of these limitations in his determination of plaintiff's RFC. Thus, the RFC finding the ALJ made at the first phase of the step four analysis did not accurately reflect plaintiff's exertional limitations.5

The ALJ also erred in evaluating plaintiff's mental RFC. "When there is evidence of a mental impairment that allegedly prevents a claimant from working, the Secretary must follow the procedure for evaluating mental impairments set forth in 20 C.F.R. § 404.1520a and the Listing of Impairments and document the procedure accordingly." Cruse v. United States Dep't of Health & Human Servs., 49 F.3d 614, 617 (10th Cir.1995). Documentation is made by completing a PRT form, which the ALJ must attach to his written decision. "[T]he record must contain substantial competent evidence to support the conclusions reached on the PRT form[, and] if the ALJ prepares the form himself, he must 'discuss in his opinion the evidence he considered in reaching the conclusions expressed on the form.' " Id. at 617-18 (quoting Washington, 37 F.3d at 1442). Here, the ALJ repeated in the body of his written opinion the conclusions he reached on the PRT, but he did not relate these conclusions to the evidence. The ALJ also recited his conclusions as to those abilities that are the subject of the Medical Assessment of Ability to Perform Work-Related Activities (Mental) form. Again, however, the ALJ did not relate his conclusions to the evidence, other than to state that they were based on "a reasonable understanding of the medical records," Tr. at 31. We note that the ALJ's conclusions as to plaintiff's abilities differed dramatically from Dr. Spray's conclusions; a difference which the ALJ did not explain.

B. Phase Two: Demands of Plaintiff's Past Relevant Work

At the second phase of the step four analysis, the ALJ must make findings regarding the physical and mental demands of the claimant's past relevant work. See Henrie, 13 F.3d at 361. To make the necessary findings, the ALJ must obtain adequate "factual information about those work demands which have a bearing on the medically established limitations." SSR 82-62, Soc. Sec. Rep. Serv., Rulings 1975-1982, at 812. When the claimant has a mental impairment,

care must be taken to obtain a precise description of the particular job duties which are likely to produce tension and anxiety, e.g., speed, precision, complexity of tasks, independent judgments, working with other people, etc., in order to determine if the claimant's mental impairment is compatible with the performance of such work.

Id. Here, the ALJ made no inquiry into, or any findings specifying, the mental demands of plaintiff's past relevant work, either as plaintiff actually performed the work or as it is customarily performed in the national economy.

On appeal, the Secretary argues, relying on cases from outside this circuit, that plaintiff bore the responsibility for developing the record as to the demands of his past relevant work. Tenth Circuit law concerning the ALJ's duty of inquiry and factual development is, however, to the contrary. See, e.g., Washington, 37 F.3d at 1442; Henrie, 13 F.3d at 361. Further, the Secretary's own rule dictates that the ALJ make the necessary findings at phases two and three of the step four inquiry. See SSR 82-62, Soc. Sec. Rep. Serv., Rulings 1975-1982, at 813.

C. Phase Three: Plaintiff's Ability to Perform His Past Relevant Work

Having failed to complete phase two appropriately, the ALJ was unable to make the necessary findings at phase three about plaintiff's ability to meet the mental demands of his past relevant work despite his mental impairments. The Secretary glosses over the absence of the required ALJ findings, by relying on the testimony of the VE that plaintiff could meet the mental demands of his past relevant work, given the mental limitations found by the ALJ. This practice of delegating to a VE many of the ALJ's fact finding responsibilities at step four appears to be of increasing prevalence and is to be discouraged.

At step five of the sequential analysis, an ALJ may relate the claimant's impairments to a VE and then ask the VE whether, in his opinion, there are any jobs in the national economy that the claimant can perform. This approach, which requires the VE to make his own evaluation of the mental and physical demands of various jobs and of the claimant's ability to meet those demands despite the enumerated limitations, is acceptable at step five because the scope of potential jobs is so broad.

At step four, however, the scope of jobs is limited to those that qualify as the claimant's past relevant work. Therefore, it is feasible at this step for the ALJ to make specific findings about the mental and physical demands of the jobs at issue and to evaluate the claimant's ability to meet those demands. Requiring the ALJ to make specific findings on the record at each phase of the step four analysis provides for meaningful judicial review. When, as here, the ALJ makes findings only about the claimant's limitations, and the remainder of the step four assessment takes place in the VE's head, we are left with nothing to review.

We are not suggesting, as has the Fourth Circuit, see Smith v. Bowen, 837 F.2d 635, 637 (4th Cir.1987), that the ALJ may not rely on VE testimony in making the necessary findings at step four. As SSR 82-62, and SSR 82-61, Soc. Sec. Rep. Serv., Rulings 1975-1982, 836, indicate, a VE may supply information to the ALJ at step four about the demands of the claimant's past relevant work. Id. at 811-12, 838. For example, if the ALJ determines that the claimant's mental impairment affects his ability to concentrate, the ALJ may ask the VE for information about the level of concentration necessary to perform the claimant's past relevant work. The VE's role in supplying vocational information at step four is much more limited than his role at step five, where he is called upon to give his expert opinion about the claimant's ability to perform work in the national economy. Therefore, while the ALJ may rely on information supplied by the VE at step four, the ALJ himself must make the required findings on the record, including his own evaluation of the claimant's ability to perform his past relevant work.

Here, the ALJ's error in relying on the VE to give an opinion about plaintiff's ability to do his past relevant work was compounded by the fact that the VE's testimony did not establish that plaintiff could perform his past relevant work as it is generally performed in the national economy. Although the VE initially testified that plaintiff could do his past relevant work, even given the mental limitations found by the ALJ, he later qualified this opinion when questioned by plaintiff's counsel.

Specifically, the VE concluded that, in light of plaintiff's successful employment with Roadway Express for twenty-one years, plaintiff's mental impairments would not negatively affect his ability to return to work at Roadway Express, because he would be "working with people that he's used to being with as well as supervisors, co-workers, and things like this, [and] it would not be like he's going into a new situation to where he would have new work adjustments to make." Id. at 135. If, however, plaintiff had to change work locations, the VE was of the opinion that his mental impairments would have a negative effect, such that plaintiff "would probably be discharged from the job if he could not maintain a certain production level.... It wouldn't mean he couldn't work, it would mean--[h]e wouldn't be able to keep the job." Id. at 137.

" 'A finding that a claimant is able to engage in substantial gainful activity requires more than a simple determination that the claimant can find employment and that he can physically perform certain jobs; it also requires a determination that the claimant can hold whatever job he finds for a significant period of time.' " Washington, 37 F.3d at 1442 (quoting Singletary v. Bowen, 798 F.2d 818, 822 (5th Cir.1986)). Thus, the VE's opinion that plaintiff's mental impairments would negatively impact his ability to hold a job somewhere other than at Roadway Express is significant, because the ALJ specifically found that plaintiff could return to his past relevant work only as it is generally performed in the national economy, not as plaintiff performed it at Roadway Express. The ALJ, however, did not ask the VE whether the customary demands of the occupation of truck driver include meeting a production level. In the absence of evidence that truck drivers are not customarily expected to meet a production level, the VE's testimony cannot constitute substantial evidence supporting the ALJ's conclusion that plaintiff can return to his past relevant work as it is generally performed in the national economy.

In sum, the ALJ's analysis at step four was flawed in several respects. In the first phase of the analysis, the ALJ failed to include all plaintiff's exertional limitations in his RFC finding and failed to relate the conclusions he recorded on the PRT about plaintiff's mental RFC to the evidence. In the second phase, the ALJ failed to develop the record on, and to make the required findings about, the mental demands of plaintiff's past relevant work. This failure infected the third phase, where the ALJ abdicated his fact finding and evaluation responsibilities to the VE. This error, in turn, was compounded by the ALJ's failure to elicit sufficient information from the VE to support the ALJ's ultimate conclusion that plaintiff could return to his past relevant work as it is generally performed in the national economy.

Because the ALJ committed numerous legal errors and his ultimate conclusion was not supported by substantial evidence, we must reverse the decision and remand for further proceedings. Both the tone of the ALJ's opinion and the manner in which the ALJ evaluated the evidence, including substituting his medical judgment for that of experts, suggest that the ALJ may not be able to provide plaintiff with a fair hearing on remand. Therefore, we recommend that the Secretary assign this case to a different ALJ for all further proceedings.

The judgment of the United States District Court for the Eastern District of Oklahoma is REVERSED, and the case is REMANDED with directions to remand the action to the Secretary for further proceedings consistent with this opinion.