United States Court of Appeals
                           FOR THE EIGHTH CIRCUIT
                                    ___________

                                    No. 08-3113
                                    ___________

Crystal D. Conklin,                     *
                                        *
            Appellant,                  *
                                        * Appeal from the United States
     v.                                 * District Court for the Western
                                        * District of Arkansas.
Michael J. Astrue, Social Security      *
Administration Commissioner,            * [UNPUBLISHED]
                                        *
            Appellee.                   *
                                   ___________

                              Submitted: October 27, 2009
                                 Filed: January 14, 2010
                                  ___________

Before BYE, BOWMAN, and BENTON, Circuit Judges.
                           ___________

PER CURIAM.

      Crystal D. Conklin appeals the district court’s order affirming the denial of
supplemental security income. Conklin alleged disability since January 2003 from
anxiety attacks and depression. Following a February 2007 hearing, where Conklin
was counseled, an administrative law judge (ALJ) determined that (1) Conklin’s
anxiety disorder, major depression, and alcohol/substance abuse were severe
impairments, but were not of listing-level severity, alone or combined; (2) her
subjective complaints were not entirely credible; (3) she could perform the physical
requirements of work at all exertional levels, but her mental residual function capacity
(RFC) precluded her past relevant work as a cashier and cook; and (4) based on the
testimony of a vocational expert (VE) in response to a hypothetical the ALJ posed,
Conklin could perform certain other jobs available in substantial numbers locally and
nationally. The Appeals Council denied review, and the district court affirmed. For
the reasons that follow, we conclude that certain aspects of the ALJ’s decision are not
supported by substantial evidence.1 See Pate-Fires v. Astrue, 

, 942 (8th
Cir. 2009) (standard of review).

       Conklin first challenges the ALJ’s credibility findings. The ALJ correctly noted
that the record reflected a history of noncompliance with medication and treatment,
but this court has recognized that a mentally ill claimant’s noncompliance with
treatment can be, and ordinarily is, the result of her mental impairment, and thus is not
willful or without a justifiable excuse. See id at 945-47 (collecting cases; noting
ALJ’s failure to make critical distinction between claimant’s awareness of need to take
her medication and question of whether her noncompliance was medically-
determinable symptom of her mental illness). Similarly, the ALJ’s reliance on
Conklin’s drug and alcohol abuse, and her failure to seek treatment for it, may not be
a valid reason for denying benefits given that she was in day treatment at Therapeutic
Family Services (TFS) for mental problems, and she testified that she drank because
of her depression, which is consistent with a TFS note indicating she was advised not
to treat her symptoms with alcohol. See Pettit v. Apfel, 

, 903 (8th Cir.
2000) (claimant has initial burden of showing that alcoholism or drug use is not
material to finding of disability; key factor is whether claimant would still be found
disabled if she stopped using drugs and alcohol). Conklin also testified she had not
used alcohol in thirteen to fourteen weeks and had used cocaine only once, and
although one of treating psychiatrist Robert Farrell’s 2005 diagnoses was marijuana
abuse, we find no other indications in the extensive treatment records of cocaine or
marijuana use.


      1
      We do not address Conklin’s meritless and undeveloped arguments concerning
nonexertional limitations or the listings under mental disorders.
                                          -2-
       We also question the ALJ’s other reasons for discounting Conklin’s subjective
complaints. First, Conklin’s failure to seek treatment earlier may be related to her
underlying mental illness and lack of insight. Second, we believe the ALJ
mischaracterized the TFS therapist’s August 2005 note concerning low-stress jobs:
we interpret it as suggesting Conklin should pursue such a job in the future, when she
was psychologically ready to look for work. We note that, earlier in 2005, Conklin’s
TFS therapist recommended she apply for supplemental security income and TFS staff
assisted her in completing the necessary forms. Third, as to Conklin’s activities, she
reportedly cared for her personal needs and did some housework, but she also
reportedly spent a lot of time in bed and occasionally did not bathe; and she and her
mother testified, consistent with the TFS treatment records, that she had periods of a
few days to two weeks when she could not get out of bed due to depression. As to
family relationships and attending church, the TFS records contain numerous
instances when poor family relationships were addressed, and Conklin testified she
had lost custody of her children. Conklin’s mother clarified that they attended only
a small class at church, as Conklin could not handle being around more people at
church services, and this testimony is also supported by the TFS records reflecting
Conklin’s withdrawal and discomfort around others. Finally, it seems the
determination by mental health professionals that Conklin required extended
rehabilitation in the TFS day program, a program she was still attending at the time
of the hearing, is consistent with her allegations of disabling mental impairments. We
thus find that the ALJ’s credibility findings are not entitled to deference. Cf. Finch
v. Astrue, 

, 935-36 (8th Cir. 2008) (if ALJ explicitly discredits claimant
and gives good reason for doing so, courts will normally defer to his judgment).

       Conklin also challenges the ALJ’s RFC findings. The ALJ gave great weight
to the RFC opinions of agency reviewing psychologists. He noted that while these
opinions were rendered before all the TFS records were received, there was no
evidence showing a subsequent decrease in Conklin’s mental functioning, other than

                                         -3-
a twenty-four-hour period when she was suicidal, and he further noted the lack of an
RFC opinion from a treating doctor. He thus found that Conklin was limited to work
that minimized social interaction, but that she retained the mental capacity to perform
the basic demands of competitive, remunerative, unskilled work, including the ability
to handle simple instructions; to respond appropriately to supervisors, coworkers, and
usual work situations; and to deal with changes in the routine work setting.

        We agree with Conklin that the RFC findings are not supported by substantial
evidence. See Davidson v. Astrue, 

, 844 (8th Cir. 2009) (ALJ should
determine claimant’s RFC based on all relevant evidence, including medical records,
observations of treating physicians and others, and claimant’s own description of
limitations). In February 2005 Dr. Farrell rated Conklin’s global assessment of
functioning (GAF) at 35, the same rating given by TFS staff a few weeks earlier.2 On
a 1-10 scale concerning the level of dysfunction (0-1 severe, 2-3 moderate, 4-6 mild,
and 7-8 average), Conklin was mostly assessed during 2005 at levels 2-4. In 2006 her
GAF was documented as 40 in August, and in 2006 and early 2007 her functional
level was assessed mostly from 1-3. See 

(RFC findings
were not supported by substantial evidence on whole record, in part due to ALJ’s
failure to discuss or consider numerous GAF scores below 50). Further, Conklin
testified that the depression, panic attacks, and paranoia and nervousness she
experienced when she was around others kept her from working and that while she
had enjoyed a past job in a chicken house, as it involved minimal interaction with
others, she lost the job because of attendance problems arising from her depression
and related inability to get out of bed for periods of time. Conklin’s testimony is
consistent with the TFS treatment records reflecting social withdrawal and periodic
attendance problems. In addition, the VE testified that the hypothetical claimant was



      2
        A GAF of 31-40 reflects a major impairment in several areas such as work,
family relations, judgment, or mood. See Am. Psychiatric Ass’n, Diagnostic &
Statistical Manual of Mental Disorders (DSM-IV-TR) 34 (4th ed., text rev. 2000).
                                         -4-
essentially unemployable if she was limited to working mostly alone, with minimum
contact with others, or if she missed two or more days of work a month.

       Conklin also argues that the ALJ failed to develop the record by not contacting
her treating psychiatrist regarding mental RFC. Given the GAF ratings and
assessment findings in the TFS records, the ALJ was arguably on notice of the need
to clarify Conklin’s mental RFC with her treating psychiatrists. See Smith v.
Barnhart, 

, 930 (8th Cir. 2006) (social security hearing is non-adversarial
proceeding, and ALJ has duty to develop record fully; duty may include seeking
clarification from treating physicians if crucial issue is undeveloped or
underdeveloped); Garza v. Barnhart, 

, 1089-90 (8th Cir. 2005) (per
curiam) (ALJ’s duty to develop record fully and fairly exists even when claimant is
represented by counsel). Based on the record before the ALJ, it is also not clear
whether Conklin’s depression, panic attacks, and other mental problems would be
disabling absent her alcoholism or drug abuse, and thus we find that further
development is also warranted on this issue. See 

.

       Accordingly, we reverse the judgment of the district court and remand for
further proceedings consistent with this opinion.
                      ______________________________




                                         -5-