INTRODUCTION
Plaintiff Cheryl Ann Cordeiro brings this action under 42 U.S.C. § 405(g) for judicial review of a final decision denying her application for Social Security Disability Insurance ("SSDI"). She suffers from major depressive and anxiety disorders and seeks SSDI benefits for September 1, 2000 to December 31, 2005. She claims that the Administrative Law Judge ("ALJ") ignored and misconstrued key evidence, failed to develop the record, and misapplied the vocational expert's testimony.
For the following reasons, the Court ALLOWS Plaintiff's motion to vacate *173(Docket No. 45) and remands. The Court DENIES the Commissioner's motion to affirm (Docket No. 26).
FACTUAL BACKGROUND
The following facts are taken from the administrative record. Plaintiff is a 59-year-old woman who lives with her husband of 38 years in Somerset, Massachusetts.
I. Educational and Work History
Plaintiff has an eleventh-grade education and has not earned a GED. She volunteered at a women's center from 1994 to 1996. Between 1995 and 2000, she worked as a babysitter at a counseling office. She resigned from that position due to severe depression and panic attacks. She worked as a caregiver part-time from 2008 to 2014. She was insured for the purposes of SSDI through December 31, 2005.
II. Medical History
At age 13, Plaintiff was admitted to a psychiatric facility after overdosing on her mother's medications in an attempted suicide. She has a family history of mental illness but not suicide attempts or substance abuse. She has consistently reported being a victim of sexual, physical, and domestic abuse.
On August 8, 2000, at age 40, Plaintiff sought mental health treatment at Child and Family Services of Fall River ("CFS") following a severe depressive incident that she called a "nervous breakdown." Soon after, Dr. Marshall Wold, a psychiatrist who saw her four times, diagnosed her with "Major Depressive Disorder, Recurrent, Severe With Psychotic Features." See Am. Psychiatric Ass'n, Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev. 2000) code 296.34 (DSM-IV-TR). Dr. Wold prescribed her Zoloft, which improved but did not eliminate her symptoms. Plaintiff also saw John McMullen, a licensed independent clinical social worker at CFS. He noted slight improvements in her symptoms, some of which were followed by recurrences. Mr. McMullen also continuously noted that she felt unable to return to work. Plaintiff was forced to stop treatment at CFS at the end of 2000 due to a lack of insurance. On February 4, 2002, CFS formally terminated its relationship with Plaintiff. In that document, CFS noted that her GAF had improved as a result of treatment.
Around the time of her nervous breakdown in 2000, Plaintiff applied for disability benefits (though it is unclear from the record which state or federal disability program she applied to). A hearing on her application was scheduled for November 7, 2000, but she was erroneously sent a denial letter and did not attend. She did not appeal the erroneous denial because she was debilitated following her breakdown.
After her 2000 treatment at CFS, there is a gap in the medical records regarding Plaintiff's psychiatric condition. Except for one visit to a doctor for lower back pain after a motor vehicle accident in 2002, the next medical care reflected in the records is Plaintiff's August 2007 visit to Dr. Gloria Mercado, a physician at Healthfirst Family Care Center. Plaintiff saw Dr. Mercado more than thirty times over the next ten years for treatment of a variety of physical issues. Throughout those ten years, Dr. Mercado consistently referenced Plaintiff's severe depression and anxiety in her progress notes and prescribed Plaintiff Celexa and Xanax to treat these symptoms. In 2008, Dr. Mercado wrote a note excusing Plaintiff from jury duty based on the attention deficits caused by her anxiety and depression. In March 2017, Dr. Mercado wrote that Plaintiff "has [had] anxiety and depression since 2000."
*174In January 2016, Plaintiff suffered a severe depressive incident and went to the emergency department of Corrigan Mental Health Center in Fall River, Massachusetts. Alison Hathaway, a licensed independent clinical social worker, described Plaintiff as in "crisis" and referred her to urgent care. Plaintiff was regularly evaluated at Corrigan for several months, during which she made slow and inconsistent progress. In June 2016, Dr. Roger Boshes, a psychiatrist whom Plaintiff had seen multiple times at Corrigan, filed out two assessments for her. He diagnosed her with chronic PTSD related to late onset psychosis, depression, debilitating anxiety, and agoraphobia. He found Plaintiff to be disabled and unable to hold a job.
In August 2016, Plaintiff first saw Mary Cruz, a licensed independent clinical social worker at East Side Counseling. Ms. Cruz made a primary diagnosis of chronic PTSD and a secondary diagnosis of generalized anxiety disorder. Ms. Cruz saw Plaintiff at least twenty times between August 2016 and May 2017. During that time, she did not note any consistent amelioration of Plaintiff's symptoms.
III. Medical Opinions
In the fall of 2000, Dr. Wold, Plaintiff's treating psychiatrist at CFS, assigned Plaintiff a Global Assessment of Functioning ("GAF") score of 41. A GAF score rates a person's overall level of functioning. See DSM-IV-TR at 34. A GAF score of 41-50 indicates serious symptoms or any serious impairment in social or occupational functioning, such as an inability to have personal relationships or keep a job. Id.
On October 13, 2000, Disability Evaluation Services ("DES") at the University of Massachusetts Medical School conducted a review of Plaintiff's medical records in connection with her application for disability benefits. Dr. Paul Kaufman, a board-certified psychiatrist, and Pat Gaucher, a registered nurse, determined that Plaintiff suffered from sleep disturbance, decreased energy, feelings of guilt and worthlessness, difficulty concentrating, and paranoid thinking. They also found that she suffered from panic attacks and fatigue. They noted her marked difficulties in maintaining social functioning and deficiencies in concentration, persistence, and pace. Accordingly, they concluded that she was disabled for the purposes of Supplemental Security Income ("SSI") benefits because she met the criteria for the listed impairment for depressive disorders (12.04).
In 2016, state agency consultants reviewed Plaintiff's medical records and determined that her impairments were not severe. In the April 2016 reconsideration opinion, the state agency consultant noted that Plaintiff was treated from August 2000 to May 2001 for depression that the medical evidence indicated improved with medication. It is not clear that these consultants had Dr. Wold's opinion or the DES findings. Apparently, Dr. Wold was deceased by then.
Dr. Boshes found in June 2016 that Plaintiff had "extreme" impairments in a number of areas, including remembering locations and work-like procedures; understanding and remembering instructions; maintaining attention and concentration; interacting with others; accepting instructions and criticism at work; interacting with coworkers and peers without exhibiting behavioral extremes; and tolerating normal levels of stress. He also found "marked" deficits in areas such as carrying out instructions; working within schedules; functioning without special supervision; asking simple questions; and making plans. He further found a "moderate" impairment in her ability to behave appropriately and adhere to basic standards of cleanliness. Dr. Boshes determined that Plaintiff's *175impairments would interfere with her ability to work at least 20 percent of the time and require her to miss work about five times per month for treatment. He concluded that Plaintiff could not work on a regular and sustained basis.
In September 2016, Ms. Cruz wrote a letter to the Social Security Administration on Plaintiff's behalf. She described Plaintiff's medical and employment history and noted that Plaintiff had repeatedly stopped treatment due to a lack of insurance. She concluded that Plaintiff's "prolonged periods of depression and anxiety ... impaired her ability to function in a working environment."
IV. Vocational Expert's Testimony
Diane Durr, a vocational expert, testified before the ALJ about the work ability of hypothetical people of the same age, education, and vocational background as Plaintiff. First, Ms. Durr testified that an individual who could perform simple tasks at all exertion levels, tolerate occasional interpersonal interactions at work, and adapt to routine changes in the work environment could perform the work of a hand packager, dishwasher, or cleaner but could not do Plaintiff's past work. Second, she stated that the same individual who was either off-task for 20 percent of the workday due to symptoms of depression, fatigue, and other mental impairments or had to be absent from work two days per month would be unemployable.
V. Plaintiff's and Husband's Testimony
As part of her SSDI application, Plaintiff completed a self-diagnostic "Function Report" in October 2015. She claimed to be consistently anxious, depressed, impaired in both her memory and attention, socially isolated, unable to do housework, and bedridden for much of the day. She checked boxes indicating she could drive and cook but described having severe difficulty with both tasks in written responses.
Plaintiff testified before the ALJ that she suffers from severe depression, PTSD, anxiety, and agoraphobia. She has no relationship with her family other than her husband and son and has no friends. She almost never leaves her home. Due to extreme fatigue and lightheadedness, she sleeps for at least six hours each day during waking hours and cannot be active for more than a few hours at a time. She is only able to engage in very minimal housework and cannot consistently cook for herself, do laundry, clean, or shop for groceries. Her deficits in memory and attention prevent her from reading a newspaper or watching television. She testified that her nervous breakdown in August 2000 was caused by emotional abuse from her father.
Plaintiff's husband testified before the ALJ that Plaintiff was severely depressed between 2000 and 2005. He confirmed she suffered a nervous breakdown in August 2000 but was not hospitalized because they lacked medical insurance. He does most of the housework. He said Plaintiff has nonexistent relationships with friends and family. He confirmed Plaintiff sleeps for most of the day and often cannot get out of bed. He testified that Plaintiff is completely incapable of maintaining full-time employment.
LEGAL STANDARD
Under the Social Security Act, a claimant seeking SSI must prove that she is disabled, i.e., "unable to engage in any substantial gainful activity by reason of any medically determinable physical or mental impairment ... for a continuous period of not less than twelve months." 42 U.S.C. § 1382c(a)(3)(A). The Commissioner uses a five-step sequential evaluation process *176to assess a claim for disability benefits. See 20 C.F.R. § 404.1520(a)(4) ; Purdy v. Berryhill, 887 F.3d 7, 9-10 (1st Cir. 2018). The evaluation ends at any step if the Commissioner finds that the claimant is or is not disabled. 20 C.F.R. § 404.1520(a)(4). The steps are as follows:
1) if the applicant is engaged in substantial gainful work activity, the application is denied; 2) if the claimant does not have, or has not had within the relevant time period, a severe impairment or combination of impairments, the application is denied; 3) if the impairment meets the conditions for one of the "listed" impairments in the Social Security regulations, then the application is granted; 4) if the applicant's "residual functional capacity" [RFC] is such that he or she can still perform past relevant work, the application is denied; and 5) if the applicant, given his or her [RFC], education, work experience, and age, is unable to do any other work, the application is granted.
Purdy, 887 F.3d at 10 (quoting Seavey v. Barnhart, 276 F.3d 1, 5 (1st Cir. 2001) ). A claimant's RFC is "the most [the claimant] can still do despite [her] limitations." 20 C.F.R. § 404.1545(a)(1). Past relevant work encompasses "work that [the claimant has] done within the past 15 years, that was substantial gainful activity, and that lasted long enough for [her] to learn to do it." Id. § 404.1560(b)(1). If a claimant cannot still perform her past relevant work, the ALJ will assess whether there is any other work the claimant "can adjust to" that "exist[s] in significant numbers in the national economy." Id. § 416.1560(c)(1).
The claimant bears the burden of proof for steps one through four. Purdy, 887 F.3d at 9. If the analysis proceeds to step five, the Government bears the burden of proof to present evidence of specific jobs the claimant can perform. Id. at 10.
AGENCY DECISION
On October 9, 2015, Plaintiff applied for SSDI benefits for the period from September 1, 2000 through December 31, 2005.1 The claim was denied initially on February 12, 2016 and upon reconsideration on May 5, 2016. An ALJ held a hearing on April 12, 2017 and denied Plaintiff's claim on July 6, 2017.
At step one of the five-step evaluation process, the ALJ found that Plaintiff did not engage in substantial gainful activity between September 1, 2000 and December 31, 2005. At step two, she concluded that Plaintiff's major depressive disorder was a severe impairment. At step three, the ALJ found that Plaintiff's impairment did not meet or medically equal any impairments listed under 20 C.F.R. Part 404, Subpart P, Appendix 1, specifically listing 12.04 for depressive disorders. She noted that Plaintiff did not experience any "marked" or "extreme" limitations as required to satisfy the "paragraph B" criteria because she only had "mild" limitations understanding, remembering, applying information, and managing herself and "moderate" limitations interacting with others and concentrating. She also found that Plaintiff had "more than minimal" capacity to handle change and therefore did not satisfy the "paragraph C" criteria.
The ALJ then determined Plaintiff's RFC. She held that, while Plaintiff's impairment could reasonably be expected to cause her symptoms, Plaintiff's statements about the severity of her depression and anxiety were not supported by the medical evidence. Relying on treatment notes taken by Mr. McMullen and Dr. Wold in 2000, the ALJ described Plaintiff's symptoms as *177moderate and improving with treatment. She explained that the absence of inpatient psychiatric admissions and other mental health treatment during the relevant timeframe undercut the alleged severity of Plaintiff's symptoms and suggested that medication was effective. She arrived at the following RFC:
[Plaintiff could] perform a full range of work at all exertional levels but with the following nonexertional limitations - [Plaintiff] was limited to the performance of simple tasks; could tolerate occasional interaction with supervisors, coworkers, and the general public; and could adapt to routine changes in a work environment.
The ALJ gave little weight to Dr. Boshes's assessment because it came more than ten years after Plaintiff's last insured date. She also gave little weight to the October 2000 DES evaluation because it was based on "a handful of counseling and medication management records."
At step four, the ALJ noted that Plaintiff could not perform her past relevant work as a babysitter. Finally, at step five, the ALJ concluded that, as the vocational expert testified, Plaintiff could perform the work of a hand packager, dishwasher, or cleaner and was therefore not disabled.
The Appeals Council denied Plaintiff's request for review on January 11, 2018, making the ALJ's decision the final decision of the Commissioner. The case is now ripe for review under 42 U.S.C. §§ 405(g) and 1383(c)(3).
STANDARD OF REVIEW
A district court reviews an ALJ's decision "to determine 'whether the final decision is supported by substantial evidence and whether the correct legal standard was used.' " Coskery v. Berryhill, 892 F.3d 1, 3 (1st Cir. 2018) (quoting Seavey, 276 F.3d at 9 ). The substantial evidence standard is "not high" and requires only "such relevant evidence as a reasonable mind might accept as adequate to support a conclusion." Biestek v. Berryhill, --- U.S. ----, 139 S. Ct. 1148, 1154, 203 L.Ed.2d 504 (2019) (quoting Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938) ). In applying this standard, a court "must bear in mind that it is the province of the ALJ ... to find facts, decide issues of credibility, draw inferences from the record, and resolve conflicts in the evidence." Johnson v. Colvin, 204 F. Supp. 3d 396, 407 (D. Mass. 2016) (citing Ortiz v. Sec'y of Health & Human Servs., 955 F.2d 765, 769 (1st Cir. 1991) ).
In reviewing for legal error, "[f]ailure of the [ALJ] to apply the correct legal standards as promulgated by the regulations or failure to provide the reviewing court with the sufficient basis to determine that the [ALJ] applied the correct legal standards are grounds for reversal." Weiler v. Shalala, 922 F. Supp. 689, 694 (D. Mass. 1996). Where application of the correct legal standard could lead to a different conclusion, a court must remand to the agency. See Da Rosa v. Sec'y of Health & Human Serv., 803 F.2d 24, 26 (1st Cir. 1986). However, remand is not necessary if it "will amount to no more than an empty exercise." Ward v. Comm'r of Soc. Sec., 211 F.3d 652, 656 (1st Cir. 2000).
DISCUSSION
I. Listed Impairment
Plaintiff first challenges the ALJ's conclusion at step three that she did not meet the criteria for listed impairment 12.04 governing depressive, bipolar, and related disorders. To meet listing 12.04, a claimant must satisfy the criteria of paragraph A and the criteria of either paragraph B or paragraph C.
*17820 C.F.R. pt. 404, subpt. P, app. 1. Paragraph A is satisfied if the claimant provides medical documentation of at least five symptoms of depressive disorder. Id. Under paragraph B, the claimant must show at least one "extreme" or two "marked" limitations in the following areas: understanding, remembering, or applying information; interacting with others; concentrating, persisting, or maintaining pace; and adapting or managing herself. Id. The paragraph C criteria require the claimant to show a "serious and persistent" mental disorder, meaning medical documentation over at least two years and evidence of ongoing treatment that improves the symptoms and minimal capacity to adapt to change. Id. The ALJ found that Plaintiff failed to satisfy either the paragraph B or paragraph C criteria, but neither conclusion was supported by substantial evidence.
As an initial matter, the ALJ did not discuss the medical opinions in the record, or any medical evidence whatsoever. The record contains evidence from Plaintiff's treatment at CFS in 2000 and from her medical and psychiatric care starting in 2007. The record also contains medical opinions from two treating sources (Dr. Wold and Dr. Boshes) and DES consultants, all of whom stated that Plaintiff was disabled. In 2000, Dr. Wold reported that Plaintiff suffered from major depressive disorder and had a GAF of 41. In 2016, Dr. Boshes opined that Plaintiff had extreme and marked impairments that precluded her from working. The DES consultants, one of whom was a board-certified psychiatrist, even specifically found that Plaintiff satisfied the criteria for listing 12.04. Yet the ALJ ignored all of the medical evidence in concluding that Plaintiff did not satisfy paragraph B or paragraph C for listing 12.04. See Nguyen v. Chater, 172 F.3d 31, 35 (1st Cir. 1999) ("The ALJ was not at liberty to ignore medical evidence ....").
Instead, the ALJ found that Plaintiff did not meet the paragraph B criteria primarily based on her 2015 self-diagnostic "Function Report." Plaintiff points out that the ALJ's reading of this report was selective. For example, to support her finding that Plaintiff had "mild" limitations in understanding, remembering and applying information, the ALJ noted that the report indicated that Plaintiff could cook, manage money, and drive. However, Plaintiff also indicated in the report that her memory loss made cooking "difficult" and that she struggled "retaining info." To support her finding that Plaintiff had "moderate" social impairments, the ALJ relied on the check-box response that Plaintiff could shop and go out alone. Later in the 2015 report, though, Plaintiff wrote that she rarely shops, does not like driving alone, and isolates herself from people as much as possible. The ALJ's finding that Plaintiff had "moderate" deficits in concentration relied on check-box responses indicating she could drive, travel alone, shop, and do housework. But Plaintiff also described difficulty in all those areas. Most importantly, Plaintiff filled out the self-report in 2015, a decade after the end of the insured period. While the ALJ gave little weight to Dr. Boshes's opinion in determining Plaintiff's RFC because it was outside the insured period, she relied exclusively on Plaintiff's self-report from the exact same time period at step three.
In determining that Plaintiff failed to satisfy the criteria of paragraphs B and C, the ALJ also focused on the lack of inpatient admissions during the relevant time period. While in some circumstances a failure to pursue or comply with treatment may support a finding that a claimant is not disabled, an ALJ must consider whether the claimant had good reasons for her failure to do so. See *179Alcantara v. Astrue, 257 F. App'x 333, 333-36 (1st Cir. 2007) (per curiam). The record here indicates that Plaintiff lacked medical insurance during the relevant timeframe and halted medical care for this reason. Plaintiff's treatment history and lack of inpatient admissions therefore do not provide substantive evidence of improvement. See Sincavage v. Barnhart, 171 F. App'x 924, 927 (3d Cir. 2006) (faulting the ALJ for drawing a negative inference from the claimant's failure to seek counseling because he ignored the claimant's testimony that she lacked adequate insurance coverage); Perry v. Colvin, 91 F. Supp. 3d 139, 149-51 (D. Mass. 2015) (finding that the ALJ drew an impermissible negative inference based on the claimant's inconsistent treatment history where good causes, including the inability to pay for medications, explained gaps in treatment).
In sum, the ALJ did not provide substantial evidence to conclude that Plaintiff's condition did not meet listing 12.04 because she ignored the medical evidence, selectively relied on a self-report from a decade later, and drew unsupported conclusions from Plaintiff's lack of treatment history.
II. Residual Functional Capacity
Plaintiff also contends that the ALJ improperly weighed the medical opinions in determining her RFC. Under the applicable regulations, a "medical source" is "an individual who is licensed as a healthcare worker by a State and working within the scope of practice permitted under State or Federal law." 20 C.F.R. § 404.1502(d). An "acceptable medical source" includes a "licensed physician." Id. § 404.1502(a)(1). A "treating source" is an "acceptable medical source who provides [the claimant] with medical treatment or evaluation and who has, or has had, an ongoing treatment relationship with [the claimant]." Id. § 404.1527(a)(2). Both Dr. Wold and Dr. Boshes qualify as treating sources because they each saw Plaintiff multiple times as part of an ongoing treatment relationship.
The ALJ must give "[c]ontrolling weight ... to a treating physician's opinion on the nature and severity of a claimant's impairments if the opinion is 'well-supported by medically acceptable clinical and laboratory diagnostic techniques and is not inconsistent with the other substantial evidence' in the record."2 Johnson, 204 F. Supp. 3d at 409 (quoting 20 C.F.R. § 404.1527(c)(2) ). Even if not given controlling weight, a treating source's medical opinion generally receives more weight than opinions from other medical sources. 20 C.F.R. § 416.1527(c)(2); Purdy, 887 F.3d at 13.
However, a number of factors determine the appropriate weight to give to the opinions of treating and other medical sources. 20 C.F.R. § 416.1527(c). For all sources, the ALJ must consider whether the source examined the claimant, the support the source provides for her opinion, the consistency of the opinion with the record as a whole, and the specialty of the source. Id. For a treating source, the length, nature, and extent of the treatment relationship and frequency of examination are also relevant considerations. Id. An ALJ need not expressly address each factor identified by the regulations but must provide "good reasons" for the weight assigned to the opinion of a treating source. Bourinot v. Colvin, 95 F. Supp. 3d 161, 177 (D. Mass. 2015) (quoting 20 C.F.R. § 404.1527(c)(2) ).
*180In evaluating Plaintiff's RFC, the ALJ gave little weight to Dr. Wold's opinion because she found that Plaintiff's symptoms had improved with medication. The 2000 treatment notes from CFS (which, to be fair, are frequently illegible) do describe some improvements in Plaintiff's symptoms (like sleeping better) while she was taking Zoloft. Moreover, CFS's termination statement says that Plaintiff's condition improved. However, Dr. Wold assigned Plaintiff a GAF score of 41 in September 2000, indicating she was severely impaired and unable to hold a job. The notes do not describe a significant enough improvement in Plaintiff's symptoms to allow her to return to work. See Hagan v. Colvin, 52 F. Supp. 3d 167, 174-76 (D. Mass. 2014) (finding that treatment notes describing the improvement of symptoms with medication could not support a finding of "not disabled" where the notes also indicated the recurrence of symptoms during treatment). Most significantly, the ALJ did not account for the low GAF score of 41.
The ALJ also gave little weight to the opinion of Dr. Boshes because he treated Plaintiff in 2016, long after the relevant period ended in 2005. The ALJ may properly consider that an opinion came more than a decade after the insured period in determining the weight the opinion should receive. However, the opinion of Dr. Boshes, a treating physician, which he provided after she suffered a nervous breakdown, does shed some light on Plaintiff's mental illness and corroborates the medical evidence from 2000 showing severe symptoms, including the 2000 DES disability determination, Dr. Wold's treatment notes, and his GAF score of 41. At the very least, the ALJ erred in not considering Dr. Boshes's opinion in determining what weight, if any, to give Plaintiff's 2015 self-report, which she viewed through such a rosy lens at step three.
The ALJ therefore erred in her assessment of the proper weight to give to the treating source medical opinions in determining Plaintiff's RFC.3
ORDER
Accordingly, Plaintiff's motion to vacate (Docket No. 45) is ALLOWED and the Commissioner's motion to affirm (Docket No. 26) is DENIED. The Court remands to the Social Security Administration for further proceedings consistent with this memorandum and order.
SO ORDERED.
The period at issue is between Plaintiff's alleged onset date and her date last insured.
20 C.F.R. § 416.920c contains new rules regarding the weight given to treating sources that apply to claims filed on March 27, 2017 or later. See Purdy, 887 F.3d at 13 n.8. Because Plaintiff filed her claim on October 9, 2015, the old rules govern this appeal.
Given the ALJ's errors at step three and in determining Plaintiff's RFC, the Court need not address Plaintiff's additional arguments that the ALJ failed to adequately develop the record in light of the gaps in medical evidence and misapplied the vocational expert's testimony at step five.