Halsey, lupus and a second diagnosis: who investigates what the first disease does not explain?

By Ramon Martins Andrade, attorney (OAB/RJ 188.374)

English translation of an article first published in Portuguese: read the original.

On 4 and 5 June 2024, Halsey published two posts on Instagram about her own health, cited in a statement of 5 June by the Lupus Research Alliance. In the second, according to a CNN report, she said that in 2022 she “was first diagnosed with Lupus SLE and then a rare T-cell lymphoproliferative disorder”. She said that both conditions are “being managed or in remission”, that she will likely have both for the rest of her life and that, “after a rocky start”, she got everything under control “with the help of amazing doctors”. In the earlier post, she had tagged the Lupus Research Alliance itself and the Leukemia & Lymphoma Society.

The account is hers, and nothing is stated here about the physicians, the services or the clinical choices of that journey. What it brings into view is the word “then”: two diseases can live in the same body, and the second arrives when the first already has a name. The thesis of the article is that an established diagnosis does not authorize anyone to stop investigating. When a finding appears that the known disease does not explain, the duty to diagnose begins again, and it is by that new beginning that liability is measured, both that of whoever saw the finding and that of whoever received it on referral.

The legal question, then, is not whether the second diagnosis was slow in coming, because sometimes it is slow through no one’s fault. It is who had the finding in hand, what they did with it and whether the patient learned of it in time.

Lupus and lymphoproliferation: what science knows about their coexistence

Systemic lupus erythematosus is an autoimmune disease: the immune system attacks healthy cells, with pain, fever, skin rashes and organ damage. Lymphoproliferative disorders, as the same report explains, are conditions in which the body produces lymphocytes, a type of white blood cell, in uncontrolled quantities. The two things are connected, because lupus itself also affects the blood: the European recommendations for the disease, cited further on, include specific guidance for the haematological disease of lupus. It is this overlap that makes it difficult to separate what is disease activity from what is a second disease.

The association has been measured. Bernatsky and colleagues assembled an international cohort of 9,547 lupus patients, at 23 centres, linked to regional cancer registries and followed for 76,948 patient-years, on average eight years (Arthritis & Rheumatism, 2005). There were 431 cancers. For all cancers combined, the standardized incidence ratio was 1.15, that is, slightly above the general population; for haematological malignancies, 2.75; for non-Hodgkin lymphoma, 3.64. The meta-analysis by Clarke and colleagues (Seminars in Arthritis and Rheumatism, 2021), with 41 observational studies, found a risk of non-Hodgkin lymphoma and of Hodgkin lymphoma more than three times higher in people with lupus. The studies are heterogeneous and the research had pharmaceutical funding, declared in the paper.

Why the risk rises is still unresolved. In the review by Bernatsky, Ramsey-Goldman and Clarke in Rheumatic Disease Clinics of North America (2005), the authors note that the possible mechanisms are numerous, but that the hypotheses remain “largely speculative”, and that the data are insufficient to know how the risk relates to exposure to medications. This matters because the treatment of lupus acts on the immune system. The recommendations of the European League Against Rheumatism (EULAR), updated in 2023 and published by Fanouriakis and colleagues in the Annals of the Rheumatic Diseases (2024), bring together 5 overarching principles and 13 recommendations: hydroxychloroquine for all patients, at a target dose of 5 mg per kilogram of actual body weight per day; glucocorticoids as a “bridge” during phases of activity, reduced in maintenance to 5 mg a day or less and withdrawn when possible; immunosuppressants and biologics to control the disease and allow that reduction.

None of this means that a blood abnormality in someone with lupus is a lymphoma, and the disorder Halsey mentioned is not necessarily one. The epidemiological data do not prove causation in any specific case. They do something else, more modest and more useful: they justify a persistent and unexplained finding being given a hypothesis of its own, instead of being read, by inertia, as more lupus.

Anchoring and premature closure: how the first name hides the second

The patient-safety literature has a name for the shortcut. Pat Croskerry, in Academic Medicine (2003), catalogued the “cognitive dispositions to respond”, biases that lead clinical reasoning astray, and proposed metacognition as the antidote: stepping back and examining one’s own thought process. Two biases in that family describe the problem of someone who has two diseases. Anchoring, as the name suggests, ties reasoning to the first impression; premature closure ends the investigation once a plausible diagnosis appears.

Graber, Franklin and Gordon studied one hundred cases of diagnostic error involving internists in the United States (Archives of Internal Medicine, 2005). Ninety caused harm, including 33 deaths. System factors contributed in 65% of the cases and cognitive factors in 74%, almost always in combination. The single most common cause was premature closure, which the authors define as the failure to continue considering reasonable alternatives after an initial diagnosis has been reached. Lack of medical knowledge was uncommon. Among the system factors, the most frequent involved procedures, teamwork and communication.

The systematic review by Saposnik and colleagues (BMC Medical Informatics and Decision Making, 2016), with 20 studies and 6,810 physicians, associated overconfidence, anchoring and information and availability biases with diagnostic inaccuracies in 36.5% to 77% of the scenarios examined; the authors themselves rated 60% of the studies as low quality, so the figures indicate a trend, not a measurement of the Brazilian consulting room. The frequency of the problem has also been estimated: Singh, Meyer and Thomas, in BMJ Quality & Safety (2014), calculated an outpatient diagnostic error rate of 5.08% among US adults, around 12 million people a year, half of them with the potential for harm.

The US National Academies of Sciences, Engineering, and Medicine, in the report Improving Diagnosis in Health Care (2015), defined diagnostic error as “the failure to (a) establish an accurate and timely explanation of the patient’s health problem(s) or (b) communicate that explanation to the patient”. The plural in parentheses describes exactly the risk faced by someone with two diseases, and item (b) is a reminder that a diagnosis that does not reach the patient is also a failure. The same report concluded that most people will experience at least one diagnostic error in their lifetime.

What Brazilian rules require of those who investigate

The Código de Ética Médica (the Code of Medical Ethics, Res. CFM 2.217/2018, issued by the CFM, the Federal Council of Medicine) forbids a physician to “fail to use all available means of health promotion and of prevention, diagnosis and treatment of disease, scientifically recognized and within his or her reach, in the patient’s favour” (art. 32). Faced with a finding that the known disease does not explain, the available means may be a test, a specialist opinion or a referral. The code also provides for the way back: the specialist who receives the patient for a specialized procedure must return him or her to the treating physician and “provide the latter with due information about what occurred during the period for which he or she was responsible” (art. 53); no physician may fail to provide another with information about the clinical condition, if the patient so authorizes (art. 54).

On the patient’s side, the new diagnosis has to be explained (art. 34) and a copy of the medical record has to be handed over, with the explanations needed to understand it (art. 88). The Estatuto dos Direitos do Paciente (the Patients’ Rights Statute), Lei 15.378/2026, reinforced each end of that circuit: information that is “accessible, up to date and sufficient” (art. 12, § 1), a second opinion “at any stage of treatment” (art. 18) and access to the medical record “without having to give any justification” (art. 19). The same law placed on the patient a duty that weighs in these cases: to inform professionals “of unexpected changes in his or her condition” (art. 22, sole paragraph, V). Coordination, in the text of the Brazilian law, is a two-way street.

In the SUS: comprehensive care is the way there and the way back

Lei 8.080/1990 (the Organic Health Law) defines comprehensive care (integralidade) as the “articulated and continuous set of preventive and curative, individual and collective actions and services required for each case at all levels of complexity of the system” (art. 7, II). Articulated and continuous are words that presuppose someone building the bridge between rheumatology and haematology. In the public network of the SUS (Brazil’s public health system), that bridge begins in primary care, and what happens when the patient gets lost between the health unit and the specialist is in the article on the family physician who referred and the patient who got lost (in Portuguese).

When the hypothesis becomes one of blood cancer, the clock starts. Lei 12.732/2012 (the law on time limits for cancer care), in § 3 of art. 2, provides: “In cases where the main diagnostic hypothesis is that of a malignant neoplasm, the tests necessary for its elucidation must be performed within a maximum of 30 (thirty) days, upon a reasoned request from the responsible physician”. The time limit depends on a written sentence: the hypothesis recorded in the request. An abnormal blood count noted merely as “lupus activity” does not start the clock; the same test with the hypothesis of lymphoproliferative disease does. Lei 13.460/2017 (the law on the rights of users of public services), in turn, guarantees users of public services “access to and obtaining of information concerning their person held in records or databases” (art. 6, III), which includes the results that were left in the system unread.

There is a specific trap in chronic follow-up. The patient with lupus returns for consultation every few months, has routine tests and gets used to hearing that they are “as always”. Routine protects, but it also numbs. The test that changes pattern, such as a lymphocytosis that does not subside or a lymph node that keeps growing, is the moment when routine should turn into investigation, and it is at that point that the question of the uncommunicated result, dealt with in the article on the critical value that nobody reported (in Portuguese), meets that of the result that was read and misinterpreted.

In the health plan: time limits for the consultation and for the test

In supplementary health (private health plans), the investigation of the second diagnosis has maximum time limits. Normative Resolution RN 566/2022 of the ANS (the health-plan regulator) sets fourteen working days for a consultation in medical specialties beyond the basic ones (art. 3, II), which includes haematology, and ten working days for outpatient diagnostic tests (art. 3, XI); each request generates a protocol number (art. 12). The time limits apply to the health-plan operator’s guarantee of care, not to the physician’s decision: if nobody requests the opinion, there is no time limit running. That is why, in cases of coexisting disease, the discussion almost never begins with coverage. It begins with the request that was not made.

When the health plan is slow, liability lies with the operator; when the physician does not make the request, the question is a different one and is directed at the physician, or at the clinic and the hospital for the services that are their own. Distinguishing the two situations is the first task of any analysis.

Comparative law: three ways of dealing with the finding that nobody investigated

1. Germany: the omitted test that reverses the burden of proof

The German Civil Code (BGB) deals with the medical treatment contract in provisions of its own. § 630c, paragraph 2, requires the treating party to explain to the patient, in an understandable way, at the start and, where necessary, over the course of treatment, all the essential circumstances, “in particular the diagnosis”. § 630h, paragraph 5, goes further: where there is a gross treatment error capable of causing the harm, causation is presumed, and the rule also applies when the professional “failed to obtain or record in time a medically indicated finding”, provided that the finding, with sufficient probability, would have produced a result that would have required further measures and that failing to take them would have been grossly wrong (author’s translation). German law shifted onto the physician, in such cases, the burden that in Brazil usually falls on the patient.

2. France: diagnosis with “the greatest care” and physicians who keep each other informed

The French Public Health Code incorporates the code of medical ethics. Art. R4127-33 provides that the physician must always make the diagnosis “with the greatest care, devoting the necessary time to it”, using the most appropriate scientific methods and, “where appropriate, with suitable assistance”. Art. R4127-64, in the wording in force since 30 July 2026, provides that, when several physicians collaborate in the examination or treatment of a patient, “they must keep one another informed”, each assumes his or her own responsibility and sees to it that the patient is informed (author’s translation). Civil liability remains based on fault: professionals are liable for the consequences of acts of diagnosis or care only “in the event of fault” (art. L1142-1, I). The French model places the duty of coordination in ethics and keeps fault as the yardstick.

3. United States: diagnostic error as a quality problem

The American path described by the 2015 National Academies report is that of quality policy. The report’s first goal is to facilitate teamwork in the diagnostic process “among health care professionals, patients, and their families”; the sixth proposes an accountability environment that learns from errors, with the recommendation that the states encourage communication-and-resolution programmes, with legal protection for disclosure of the error and for apology, and test alternative approaches to resolving medical injury. The report is not law. But it shows a choice different from the other two: treating diagnostic error as a system failure to be measured and corrected, in addition to fault to be proved.

Brazil is closer to France. There is no rule presuming causation when the test was not requested; there is the ethical duty to use the available means and to inform, and the procedural instrument of the Código de Defesa do Consumidor (the Consumer Protection Code), which allows the burden of proof to be reversed “when, in the judge’s discretion, the allegation is plausible or when the consumer is at a disadvantage” (art. 6, VIII). What Germany wrote into statute, the Brazilian judge may do case by case.

When delay becomes liability

Medical liability “is always personal and cannot be presumed” (Code of Medical Ethics, art. 1, sole paragraph), and that of self-employed professionals is determined “upon a finding of fault” (art. 14, § 4, of the Código de Defesa do Consumidor). The clinic and the hospital are liable regardless of fault, but only for defects in the services that are their own (art. 14, head provision). The guide to medical error (in Portuguese) explains that division.

The most delicate point is the timing of the assessment. The question is what the finding indicated on the day it appeared, not in the light of the second diagnosis, which only exists afterwards. Saposnik notes that hindsight bias, that of judging a decision with knowledge of the outcome, was not even investigated in the studies he gathered. An abnormal test that was seen, recorded and investigated within a clinically reasonable time is not an error, even if the diagnosis takes time. What usually weighs are three absences: the result nobody read, the request nobody made and the return visit nobody scheduled. It is the pattern examined in the article on the creatinine that rose with every test (in Portuguese), and it repeats itself with the blood count.

Once the failure is proved, what is discussed is almost never the whole disease. The STJ (Superior Court of Justice), in REsp 1.254.141/PR (Third Panel, reporting Justice Nancy Andrighi, judged on 4 December 2012), in a case of inadequate treatment of breast cancer, and not of delayed diagnosis, admitted loss of a chance and recorded in the headnote: “The chance, however, can never reach the value of the good that was lost”. Compensation measures the probability of a different course that was lost, and proof of that probability usually depends on expert evidence. The logic is in the article on delayed cancer diagnosis and loss of a chance (in Portuguese). When the first consultation was quick and generalist, the yardstick is that of the article on the diagnosis that slipped by in a ten-minute consultation (in Portuguese).

Before deciding

In my view, coordination becomes verifiable in the circuit that the Code of Medical Ethics already describes: referral, counter-referral and communication to the patient. In private practice no rule names “the coordinator”, but whoever had the finding in hand answers for what they did with it, and the expert must guard against hindsight bias as much as the physician against anchoring. Anyone living with a chronic disease who receives an abnormal result can ask for the hypothesis to be written on the test or referral request, keep the dates of each result and each referral, ask for a copy of the medical record and seek a second opinion when the explanation is unconvincing. If the test or the specialist opinion was denied or delayed by the health plan, if the investigation stalled in the SUS queue or if the second diagnosis seems to have come late, get in touch with the firm: the analysis starts from the tests, the requests and the dates, and does not replace medical assessment.

About the author

Ramon Martins Andrade (OAB/RJ 188.374) is a lawyer who graduated from UFRJ in 2011, with master’s degrees from Université Sorbonne Nouvelle and from the Naval War College (EGN, Brazilian Navy). He is currently a researcher in Human Rights and Health at ENSP/Fiocruz.

This content is for information and education only. It does not constitute legal advice, advertising of results or any guarantee of outcome, under Rule 205/2021 of the Brazilian Bar Association (OAB).