Terry STEELE, Plaintiff-Appellant, v. Jo Anne B. BARNHART, Commissioner of Social Security, Defendant-Appellee, 290 F.3d 936

ILANA DIAMOND ROVNER, Circuit Judge. Terry Steele, a former City of Chicago firefighter, applied for supplemental security income at the age of forty-five asserting that he could not work because of debilitating epileptic seizures and low back pain. An administrative law judge found Steele not disabled and denied benefits, and the Social Security Administration’s appeals council declined review. Steele then commenced this action in the district court, which upheld the agency’s decision, and Steele now appeals.

Although our review of the Commissioner’s decision is deferential, we hold that a remand is necessary because the ALJ misconstrued the medical evidence of Steele’s epilepsy, inadequately explained why she discredited Steele’s own testimony, and relied on incomplete vocational testimony. Steele’s documented medical problems began with a back injury and assumed larger proportions when Steele was later diagnosed with epilepsy and depression. Steele’s first reported back problems occurred after a fall in 1991 that caused him to miss six months of work.

Steele further aggravated his back in 1993, and pain and difficulty walking again forced him to miss work. Dr. Nasim Rana, an orthopedist who examined Steele in January 1994, observed decreased range of motion in Steele’s lower back and also determined from x-rays that Steele had degenerative disk disease. Two months later, Steele was fighting a fire when he passed out from smoke inhalation, fell down a flight of stairs, and struck his head. After the fall Steele suffered an acute lumbar sprain along with chronic low back pain.

Soon after, Steele also began to experience what his doctors described as “absences” — periods of incoherence marked by lapses in consciousness and occasional hallucinations — which Dr. Steven Brint, a neurologist, thought might be related to the fall. The absences were diagnosed as epileptic seizures, and electroencephalograms (EEGs) performed in May confirmed that diagnosis by documenting “ictal rhythms” (electrical patterns in the brain that occur during a seizure). To treat the seizures, doctors prescribed anticonvulsant medication. Since placing Steele on the medication, Dr.

Brint reported in November that the “spells have stopped and he feels much better.” An EEG administered in November also revealed no evidence of clear ictal patterns, but the EEG did reflect a “slight neurophysiological disturbance” in the temporal areas of Steele’s brain. After Dr. Brint examined Steele in April 1995, he stated in an epilepsy report that the seizures appeared “well controlled” on anticonvulsant medication. But in the same report, Dr. Brint notably went on to observe that Steele continued to have several seizures a month, despite taking the medication as prescribed.

By January 19, 1996, when Steele applied for benefits, the frequency of his seizures allegedly had increased. During an assessment of Steele’s physical capacity to work conducted that April, for example, Dr. Julius Villaflor reported that Steele complained of “frequent” seizures and suggested that they might be better controlled if Steele followed up with his treating physician. According to Dr.

Villaflor, Steele could lift up to sixty pounds and sit or stand (with breaks) for between six and eight hours a day, but his seizures prevented him from operating machinery, driving motorized vehicles, or working at heights. Dr. Rochelle Hawkins, a specialist in internal medicine, also examined Steele about two months later and prepared a report for the Illinois Department of Public .Aid. According to Dr. Hawkins, Steele complained of daily seizures, though Dr.

Hawkins also wrote that Steele “admits he is not compliant with medication or follow up.” After his epilepsy diagnosis, Steele became depressed and began attending individual psychotherapy sessions. In connection with his application for benefits, Steele underwent psychiatric and psychological evaluations, and both concluded that he had “depressive neurosis.” According to Dr. J. Chen’s psychiatric evaluation, Steele’s depression would spoil his appetite, trigger insomnia, agitate him, and lead him to isolate himself. Although Steele’s depression was not impairing, Dr.

Thomas Low’s psychological report further concluded that depression restricted Steele’s daily activities and social functioning and also affected Steele’s ability to timely complete tasks by interfering with his concentration, persistence, and pace. At a hearing in August 1997, the ALJ received Steele’s medical records (many of which we have not discussed because they are either duplicative or inconsequential) and heard testimony from Steele. He testified that he suffered from depression, had difficulty walking, and had trouble lifting heavy objects because of his back.

Steele also testified that despite taking his medication, he continued to have seizures, typically twice a day and lasting five seconds. Steele added that two months earlier he had suffered a five-minute blackout, during which he fell and injured his hand. As recently as the morning of the hearing, Steele continued, he experienced a seizure that lasted five or more seconds. The ALJ also arranged for a psychiatrist and a vocational expert to testify at the hearing. Based on Dr.

Chen’s evaluation, the psychiatrist opined that he did not think Steele’s depression was impairing, and beyond that he lacked the expertise to-evaluate Steele’s neurological condition. The vocational expert concluded that a hypothetical person of Steele’s age and vocational background — who could not operate equipment or work at heights and who required a daily fifteen-minute break (at an unspecified time) to recover from a seizure — could still work light and sedentary jobs as a security guard, cashier, interviewer, or housekeeper.

But when asked to assume all of the functional limitations reported by Steele, the vocational expert could- not suggest any available jobs. The ALJ issued a- decision denying Steele’s application for benefits. In her decision the ALJ concluded that Steele’s May EEG was “generally unremarkable,” that his November EEG was “unremarkable,” that Steele’s seizures were controlled by medication, and that the seizures occurred only when Steele failed to follow prescribed treatment.

The ALJ also determined that despite Steele’s back problems, seizures, and depression, he did not have a conclusively disabling impairment and he retained the capacity for light work. After finding Steele’s own testimony not credible to the extent it suggested he could not perform light work, the ALJ determined from the vocational expert’s testimony that Steele could hold 15,000 jobs as a security guard and 2,000 jobs as a housekeeper.

Engaging in the familiar five-step analysis used to evaluate disability claims, 20 C.F.R. § 416.920, the ALJ concluded that Steele (1) did not have a job, (2) had a severe impairment, (3) did not have an impairment or combination of impairments listed in the agency’s regulations, (4) could not return to his job as a firefighter, (5) but could work a significant number of jobs as either a security guard or a housekeeper. Steele then appealed to the agency’s appeals council and submitted another EEG taken in September 1998 that suggested ictal rhythms.

But Steele’s request for review was denied, making the ALJ’s decision the final decision of the Commissioner. 20 C.F.R. § 416.1481. We will uphold the Commissioner’s decision if it is supported by substantial evidence and is free of legal error. 42 U.S.C. § 405(g). Although this standard is generous, it is not entirely uncritical, Clifford v. Apfel, 227 F.3d 863, 869 (7th Cir.2000), and where the Commissioner’s decision lacks evidentiary support or is so poorly articulated as to prevent meaningful review, the case must be remanded.

On appeal Steele advances three principal contentions: (1) the ALJ’s findings at step three of the sequential evaluation process do not command the support of substantial evidence; (2) the ALJ insufficiently explained why she discredited Steele’s own testimony; and (3) the ALJ’s findings at step five of the sequential evaluation process were based on incomplete vocational testimony. We consider these arguments in turn. At step three the ALJ needed to determine whether Steele was conclusively disabled based on one of the agency’s listed impairments.

One relevant provision is listing 11.03, which deals with “minor motor seizures.” It provides for a disability finding where the applicant has documented seizures “occurring more frequently than once weekly in spite of at least 3 months of prescribed treatment.” 20 C.F.R. Pt. 404, Subpt. P, App. 1, § 11.03; see Lewis v. Apfel, 236 F.3d 503, 512-13 & n. 10 (9th Cir.2001); Brown v. Apfel, 174 F.3d 59, 64 (2d Cir.1999); Flanery v. Chater, 112 F.3d 346, 349 (8th Cir.1997); Brown v. Bowen, 845 F.2d 1211, 1214 (3d Cir.1988).

Although Steele’s medical records plainly documented his history of seizures, the ALJ altogether failed to discuss, or even cite, listing 11.03. In at least two circuits, omitting a key listed impairment like listing 11.03 — coupled with the otherwise perfunctory analysis provided by the ALJ at step three — alone would'require a remand. See Burnett v. Commissioner, 220 F.3d 112, 119-20 (3d Cir.2000); Clifton v. Chater, 79 F.3d 1007, 1009 (10th Cir.1996). The Eighth Circuit, on the other hand, has concluded that a cursory discussion at step three is not automatically fatal. See Senne v.

Apfel, 198 F.3d 1065, 1067 (8th Cir.1999). But we need not explore the possible tension in these cases, for the ALJ’s decision could not stand even if she cited the correct rule. The chief problem lies in the ALJ’s mis-characterization of the medical evidence of Steele’s epilepsy. The ALJ’s determinations that Steele’s May 1994 EEG was “generally unremarkable” and that his November EEG was “unremarkable,” for example, are untenable on the current record. There are in fact two EEG reports from May.

One demonstrates twenty episodes of ictal rhythms during an over night EEG, and the other shows a one-minute episode during a regular EEG. The ALJ did not specify which report she was referring to, but even the Commissioner admits in her appellate brief that a report demonstrating seizure episodes “should be regarded as remarkable.” Likewise, although the November report does not show any ictal patterns, it does show a “slight neurophysiological disturbance” in Steele’s brain.

Nothing in the ALJ’s decision or the medical records explains the insignificance of this disturbance, and without further explanation, we are left to wonder how a report documenting a neu-rophysiological disturbance, however “slight,” could be characterized as “unremarkable.” The ALJ also depreciated the medical evidence of Steele’s epilepsy by concluding that his seizures were controlled by medication. According to Social Security Ruling 87-6, advances in clinical approaches to epilepsy have rendered most epileptic seizures controllable through drug therapy.

Consequently, before granting an application for benefits under listing 11.03, the ALJ must have current evidence showing a therapeutic level of medication in the applicant’s blood. Lewis, 236 F.3d at 513; SSR 87-6. Here the ALJ did not receive evidence of the level of medication in Steele’s blood — even though the partly adversarial, partly inquisito rial, procedure for adjudicating social security claims requires the ALJ to order additional tests if necessary to render an informed disability determination. See Smith v.

Apfel, 231 F.3d 433, 437-38 (7th Cir.2000); SSR 87-6 (requiring ALJs to solicit further evidence upon a treating physician’s ambiguous report of ongoing seizures); see also Green v. Apfel, 204 F.3d 780, 781 (7th Cir.2000) (requiring the ALJ to summon a medical expert to interpret clinical records). Instead the ALJ concluded that Dr. Brint’s epilepsy report establishes that Steele’s seizures were controlled by anti-convulsant medication and that Dr. Hawkins’s report for the Illinois Department of Public Aid demonstrates that Steele’s seizures were uncontrolled only because he failed to follow prescribed treatment.

But neither report supports the inference drawn by the ALJ. Dr. Brint’s report, for instance, is at best ambiguous: It observes that by April 1995 Steele’s seizures were both “well controlled” by anticonvulsant medication and persisting at a rate of “several/month.” And Dr. Hawkins’s report relays merely that Steele “admits he is not compliant with medication.” It fails to address the effect of Steele’s alleged noncompliance, if any, on the frequency of his seizures.

The cases recognize that evidence of noncompliance by itself proves nothing under listing 11.03; what matters instead is whether the record contains evidence of a causal link between the noncompliance and the ongoing seizure episodes. See Brown, 174 F.3d at 63; Lucas v. Sullivan, 918 F.2d 1567, 1572 (11th Cir.1990). To answer this question, the ALJ must have some reason to- think that the applicant’s seizures would not continue at their current rate were the prescribed treatment followed. The Commissioner insists that “the record as a whole” fills the gaps in the ALJ’s analysis left by the reports of Dr.

Brint and Dr. Hawkins. But regardless whether there is enough evidence in the record to support the ALJ’s decision, principles of administrative law require the ALJ to rationally articulate the grounds for her decision and confine our review to the reasons supplied by the ALJ. See SEC v. Chenery Corp., 318 U.S. 80, 93-95, 63 S.Ct. 454, 87 L.Ed. 626 (1943); Johnson v. Apfel, 189 F.3d 561, 564 (7th Cir.1999); Sarchet v. Chater, 78 F.3d 305, 307 (7th Cir.1996). That is why the ALJ (not the Commissioner’s lawyers) must “build an accurate and logical bridge from the evidence to her conclusion.” Dixon v.

Massanari, 270 F.3d 1171, 1176 (7th Cir.2001). Here the ALJ referred to no evidence other than the reports of Dr. Brint and Dr. Hawkins to support her view that medication controlled Steele’s seizures and that Steele’s noncompliance with treatment caused his seizures to become uncontrolled. Because these reports do not provide a rational basis for the ALJ’s conclusion, the ALJ’s decision at step three cannot be upheld.

Along with the ALJ’s tenuous discussion of the medical evidence, we also agree with Steele that the ALJ provided insufficient reasons for discounting his own testimony — including his assertions of depression, difficulty walking, daily seizures, and trouble lifting heavy objects because of his bad back. According to Social Security Ruling 96-7p, which applies to the ALJ’s evaluation of an applicant’s description of symptoms, the evaluation must contain “specific reasons” for a credibility finding; the ALJ may not simply “recite the factors that are described in the regulations.” SSR 96-7p.

Without an adequate explanation, neither the applicant nor subsequent reviewers will have a fair sense of how the applicant’s testimony is weighed. Zurawski v. Halter, 245 F.3d 881, 887 (7th Cir.2001); Briggs v. Massanari, 248 F.3d 1235, 1239 (10th Cir.2001); Schaudeck v. Commissioner, 181 F.3d 429, 433 (3d Cir.1999); SSR 96-7p. The ALJ dismissed Steele’s description of his limitations in a single sentence.

The assessment reads in its entirety: “The claimant’s subjective complaints and alleged limitations were considered under the criteria of Social Security Ruling 96-7p and found credible only to the extent of precluding the claimant from performing work in excess of light level.” This statement does not suggest how Steele could still perform light and sedentary work despite his reported problems with depression, walking and lifting, and seizures— which when considered cumulatively left the vocational expert unable to identify any jobs for Steele to perform.

And the ALJ’s evaluation does not seek to apply the factors for evaluating symptoms set forth in Social Security Ruling 96-7p, such as the degree to which Steele’s stated limitations were consistent with the medical evidence or the ALJ’s own observations. See Zurawski, 245 F.3d at 887-88; Schaudeck, 181 F.3d at 433. Invoking a legal rule does not substitute for complying with the requirements of that rule, and here the ALJ’s evaluation of Steele’s credibility does no more than cite ruling 96-7p without supplying any of the details demanded by that provision.

Moreover, Steele is also correct that the ALJ appears to have elicited incomplete testimony from the vocational expert. In her hypothetical questions to the vocational expert, the ALJ included many of Steele’s impairments. But she addressed neither how Steele’s depression restricted his daily activities and social functioning, nor how depression affected his ability to timely complete tasks by interfering with his concentration, persistence, and pace. Hypothetical questions posed to vocational experts ordinarily must include all limitations supported by medical evidence in the record. Cass v.

Shalala, 8 F.3d 552, 555-56 (7th Cir.1993); Gilbert v. Apfel, 175 F.3d 602, 604 (8th Cir.1999); Winfrey v. Chater, 92 F.3d 1017, 1024 n. 5 (10th Cir.1996). The reason for the rule is to ensure .that the vocational expert does not refer to jobs that the applicant cannot work because the expert did not know the full range of the applicant’s limitations. An exception therefore exists for cases in which the vocational expert independently learned of the limitations (through other questioning at the hearing or outside review of the medical records,"for example) and presumably accounted for them. See Ragsdale v.

Shalala, 53 F.3d 816, 818-21 (7th Cir.1996). Yet nothing in the record reflects that the vocational expert independently knew of all the limitations related to Steele’s depression that were omitted by the ALJ. True, the vocational expert testified at length about how bouts of hostility brought on by Steele’s depression might affect his vocational profile. And it is also true that the jobs identified for Steele to work (such as housekeeper and security guard) might not demand levels of sociability or concentration beyond his capabilities. See Donahue v. Barnhart, 279 F.3d 441, 444 (7th Cir.2002).

So our misgivings about this problem are not acute. But given the other difficulties in the case, we mention the ALJ’s incompletely formed hypothetical questions as well. The judgment of the district court is Vaoated, and the case is Remanded to the district court with instructions to remand the case to the Commissioner for further proceedings consistent with this opinion.