Nonbacterial thrombotic endocarditis

Article written and reviewed by Dr Luca Moretti

Nonbacterial thrombotic endocarditis, also known by the acronym NBTE, is characterized by endocardial vegetations composed predominantly of platelets and fibrin, without an infection responsible for their formation. It mainly involves the mitral and aortic valves and represents a cardiac manifestation of systemic conditions capable of combining coagulation activation, endothelial dysfunction, and inflammation. The historical term marantic endocarditis reflects the association with advanced neoplastic disease and cachectic states, but does not encompass the entire spectrum of disease: sterile thrombotic lesions may also accompany antiphospholipid syndrome and other immune-mediated conditions. Libman-Sacks endocarditis belongs to this general field, although it has immunopathologic and structural features that justify separate discussion.
The dominant clinical consequence is arterial embolization, sometimes recurrent and multiterritorial, whereas valvular dysfunction may be mild. A patient may therefore present with stroke or visceral ischemia without a cardiac history and without a significant murmur. This dissociation between the extent of the valvular finding and the severity of extracardiac injury is one of the main reasons for underdiagnosis.

Incidence in the general population is not precisely defined. Autopsy series identify lesions that were never recognized during life, whereas echocardiographic studies frequently include patients already selected for malignancy, emboli, or suspected endocarditis: their results are not interchangeable. The most recurrent oncologic associations include adenocarcinomas, particularly of the lung, pancreas, and gastrointestinal tract, often at a disseminated stage; NBTE may nevertheless precede recognition of the malignancy and does not, by itself, prove metastatic disease. Contemporary series also show a clinically relevant autoimmune component. Early recognition therefore serves both to identify a potentially treatable embolic source and to detect the systemic condition sustaining it, while avoiding attribution of every stroke in a patient with cancer to a generic hypercoagulable state.

Etiology, pathogenesis, and pathophysiology

The normal valve has an endothelial surface that hinders platelet adhesion and modulates coagulation. In NBTE, transition to a proadhesive and procoagulant phenotype allows deposition of thrombotic material, especially in regions exposed to mechanical stress, including leaflet closure lines. Endothelial activation promotes platelet adhesion; thrombin generation converts fibrinogen to fibrin and stabilizes the deposit. Hemodynamically significant valvular disease need not be present: an alteration of the endothelial surface, even if not recognizable on echocardiography, may be sufficient when blood is strongly procoagulant. The lesion therefore results from the interaction between the valvular substrate and systemic thrombogenicity, rather than from the mere presence of an intracardiac mass.

In cancer, hypercoagulability results from several concomitant processes. Tumor cells and extracellular vesicles may express tissue factor, activating the coagulation cascade; inflammatory mediators alter endothelial function and increase interactions among leukocytes, platelets, and the vascular wall. Mucins produced by some adenocarcinomas can participate in selectin-mediated interactions, while neutrophil activation and formation of extracellular traps contribute to the prothrombotic milieu. The relative weight of these mechanisms varies among tumors and patients and cannot be measured by a single clinical test. Immobility, intercurrent infection, surgery, and antineoplastic treatment can add further stimuli, but do not by themselves explain valvular predilection.
Disseminated intravascular coagulation may coexist with NBTE, particularly in advanced malignancy. Thrombosis and consumption of platelets and clotting factors may then develop simultaneously: the appearance of bleeding does not demonstrate that thrombotic risk has ceased. Elevated D-dimer and thrombocytopenia describe this setting, but do not identify a vegetation and do not distinguish a cardiac embolic source from thrombosis formed directly within vessels.

In antiphospholipid syndrome, antibodies directed against proteins that interact with phospholipids, especially beta2-glycoprotein I, contribute to activation of endothelium, platelets, monocytes, and complement. The consequence is greater availability of procoagulant surfaces and signals; valvular thickening and vegetations may develop. In lupus, immune-complex deposition, immune-inflammatory injury, and subsequent fibrous organization are added. These forms share a thrombotic component with neoplastic NBTE, but should not be considered biologically identical: a fibrotic and deformed leaflet is not equivalent to a recent fibrin deposit, and the response to therapy depends on the predominant component. Other severe systemic diseases have also been associated with sterile vegetations, but an association reported in a case series does not prove causality in an individual patient.

On gross examination, vegetations may be minute or confluent, sessile or irregular, and located on one or more leaflets. Histologically, platelet aggregates and fibrin predominate, with inflammatory cellularity often sparse in marantic forms and no demonstrable microorganisms after adequate assessment. Attachment to the substrate may be relatively weak; tissue invasion and suppurative destruction typical of many forms of infective endocarditis are usually absent. The limited local reaction contributes to fragility of the deposit and the possibility that a substantial portion of the vegetation may detach without leaving a large echocardiographic lesion. Small size or reduction of the mass after a neurologic event is therefore not necessarily a sign of low risk.

Vegetations on left-sided valves release material into the systemic circulation. Different fragments may successively reach the brain, spleen, kidneys, mesentery, or limbs; severity depends on the caliber of the occluded vessel, collateral reserve, and duration of ischemia. In the brain, dissemination may produce bilateral infarcts in multiple arterial territories and of different ages, but this pattern is not specific to NBTE. A similar distribution may occur in infective endocarditis or other embolic conditions. Venous thrombosis associated with malignancy does not result from detachment of a mitral or aortic vegetation: it is a parallel expression of the same procoagulant state, except in particular mechanisms such as paradoxical embolism through a shunt.
Hemodynamic damage is often less important than embolic risk. Small vegetations may not interfere with coaptation, whereas larger deposits, leaflet deformation, or pre-existing valve disease may cause regurgitation. Major acute regurgitation, perforation, or perivalvular extension requires especially careful reconsideration of the infectious hypothesis, without assuming that a known malignancy makes every lesion sterile.

The relationship between deposition and elimination of thrombotic material helps explain the course. A vegetation is not necessarily a continuously growing structure: new layers may be deposited, one portion may organize, and another may fragment. Its echocardiographic size represents the net result of these processes at a specific moment. Anticoagulation also primarily modifies thrombus generation and propagation, without being a fibrinolytic treatment capable of immediately dissolving every deposit. Failure of a mass to disappear early therefore does not equal treatment failure, just as disappearance after an embolus does not equal cure. Anatomic and clinical trajectories must be compared while keeping the two outcomes separate.

The relationship with Trousseau syndrome is one of partial inclusion, not synonymy. Cancer-associated hypercoagulability may produce migratory thrombophlebitis, venous thrombosis, consumptive coagulopathy, or arterial ischemia through different pathways; NBTE specifically identifies valvular involvement with vegetations. This distinction prevents two opposite errors: excluding a cardiac source because cancer already explains the thrombotic tendency, or diagnosing NBTE in every patient with malignancy, stroke, and elevated D-dimer. The presence of thrombosis in multiple compartments strengthens suspicion of a systemic procoagulant condition, but demonstration of the endocardial lesion requires its own diagnostic pathway.

Clinical manifestations

The history should first reconstruct the sequence of embolic events. A sudden focal deficit, transient episodes of aphasia or amaurosis, subacute cognitive decline, and fluctuating neurologic symptoms may represent different expressions of the same process. Date of onset, duration, recovery, and appearance of new symptoms after treatment began must be specified; the label “previous stroke” alone does not establish whether embolization remains active. Abdominal or flank pain, hematuria, sudden limb pain, and chest pain suggest visceral, renal, peripheral, or coronary ischemia respectively, although each always requires confirmation and comparison with alternative causes.

The search for a predisposing condition includes the type and stage of a known malignancy, response to treatment, recent progression, vascular access, procedures, and hematologic complications. If cancer is not known, weight loss, anorexia, bleeding, changes in bowel habits, persistent cough, or other organ-specific symptoms guide further investigation; none alone diagnoses malignancy. Previous arterial or venous thrombosis, recurrent pregnancy loss or other obstetric complications, nephritis, cytopenias, and mucocutaneous or articular manifestations instead suggest a possible autoimmune setting. Family history of thrombosis may be obtained, but in an adult with NBTE and an obvious acquired factor, indiscriminate testing for inherited thrombophilia rarely clarifies the origin of the valvular lesion.

An essential part of the history concerns treatments already received. Previous antibiotics, anticoagulant dose and administration, missed doses, interruptions for biopsies or procedures, vomiting or malabsorption, and drugs capable of altering anticoagulant exposure should be reconstructed. An embolus occurring while a drug was prescribed does not necessarily mean an event occurred during genuinely adequate therapy. In parallel, fever, chills, skin or dental infections, invasive procedures, prostheses, and intracardiac devices are sought. Absence of fever does not exclude infective endocarditis, particularly in older or immunosuppressed patients; fever may also be related to malignancy, an extracardiac infection, or organ infarction.

Cardiac symptoms may be absent. When dyspnea, orthopnea, or reduced exercise tolerance occurs, it should be determined whether they result from valvular regurgitation, ventricular dysfunction, anemia, pulmonary thromboembolism, effusions, or neoplastic respiratory disease. This attribution is necessary because finding a vegetation does not automatically make every episode of dyspnea cardiac. Palpitations and syncope also require rhythm evaluation: atrial fibrillation may constitute a competing embolic source even in a patient with documented NBTE.

Physical examination begins with level of consciousness, vital signs, oxygenation, and perfusion. Neurologic assessment should document the deficit rather than merely note confusion; examination of pulses, limb temperature, and color looks for peripheral ischemia. Abdominal tenderness, peritoneal signs, and pain out of proportion to early findings may require an urgent pathway for mesenteric ischemia. Cardiopulmonary examination assesses murmurs, signs of congestion, and venous pressure, remembering that an emboligenic vegetation may be silent on auscultation. General examination searches for lymphadenopathy, masses, skin lesions, signs of lupus, and possible infectious portals of entry. Petechiae and splinter hemorrhages, if present, do not by themselves distinguish the sterile from the infective nature of the lesion.

Investigations and diagnosis

Clinical diagnosis of NBTE requires convergence of three elements: a compatible valvular lesion, a plausible thrombotic context, and adequate assessment of the infectious alternative. There is no validated international score capable of certifying NBTE by summing echocardiographic and laboratory findings. Pathologic confirmation is possible when valve tissue is available, but does not justify an operation performed solely to obtain a diagnosis. In cases managed without surgery, the report and clinical discussion should explicitly state the degree of certainty and the reasons supporting the attribution, especially when previous antibiotics limit the value of cultures.

Transthoracic echocardiography is the first examination for identifying masses, assessing all valves, and measuring regurgitation, chamber dimensions, and ventricular function. The report should describe location, number, size, mobility, attachment surface, and relationship with coaptation; a generic phrase such as “possible vegetation” is insufficient for follow-up. Transesophageal echocardiography is indicated when suspicion remains high despite a negative or inconclusive transthoracic examination, particularly in the presence of systemic emboli and potentially small lesions. It provides better definition of the valve apparatus and alternative diagnoses, but its greater sensitivity in studies of selected patients is not an indication for invasive screening of every patient with cancer.

Echocardiographic morphology does not prove sterility. An infectious vegetation may appear similar to a thrombotic deposit, especially before destructive lesions develop. A fibroelastoma tends to appear as a circumscribed mass, often pedunculated and mobile; Lambl's excrescences are generally thin filamentous structures along closure lines; calcifications and degeneration produce additional pseudotumoral images. These characteristics guide reasoning but do not replace overall assessment. Three-dimensional reconstruction can better define the attachment base and extent, whereas cardiac CT and MRI have selective complementary roles and may not resolve minute vegetations. Metabolic uptake, if present, does not by itself equal infection, and lack of uptake does not certify NBTE.

Microbiology should be established before antibiotics when clinical conditions permit, obtaining multiple appropriately sampled blood cultures, usually three peripheral sets when endocarditis is suspected. Blood volume, timing in relation to therapy, and collection technique affect yield. In instability or sepsis, samples should be organized rapidly without delaying necessary therapy. Negativity after antibiotics does not have the same meaning as negativity in a never-treated patient. If suspicion of blood culture-negative endocarditis persists, serology and molecular methods are guided by exposures and the clinical picture, including pathogens such as Coxiella burnetii and Bartonella spp. when appropriate. Even established cancer can coexist with infective endocarditis.

Duke-ISCVID criteria support classification of infective endocarditis, but failure to meet criteria for definite endocarditis does not prove NBTE. Automatically converting a “possible” or incompletely investigated case into a diagnosis of sterile vegetation is particularly risky. Abscesses, fistulas, perforations, new prosthetic dehiscence, or persistent bacteremia strengthen the infectious hypothesis; their absence does not negate it. Clinical response to antibiotics is itself influenced by many variables and cannot be used as the sole retrospective criterion of etiology.

Complete blood count, platelet count, renal and hepatic function, urinalysis, inflammatory markers, and coagulation profile define the clinical setting and treatment safety. Fibrinogen and D-dimer may document activation and consumption, but do not measure vegetation burden or individual embolic risk with a validated threshold. When antiphospholipid syndrome is suspected, lupus anticoagulant, anticardiolipin, and anti-beta2-glycoprotein I antibodies are tested, with interpretation of isotype, titer, and persistence of positivity. Confirmation at least twelve weeks apart helps distinguish a persistent profile from transient positivity; anticoagulants may interfere with lupus anticoagulant testing, making discussion with the laboratory necessary. A prolonged aPTT in this setting does not necessarily indicate a bleeding tendency.

The search for occult malignancy should be proportionate to clinical probability and available findings. History, physical examination, blood tests, age-appropriate screening, and targeted imaging allow selection of further thoracoabdominal or organ-specific investigations and biopsy of suspicious lesions. An extensive panel of tumor markers does not replace this pathway and can generate false-positive findings without localizing disease. If no cause emerges, the diagnosis remains open to reassessment: the label idiopathic NBTE describes the absence of a recognized condition at that time, not definitive exclusion of malignancy or autoimmune disease.

Evaluation of embolic injury proceeds according to urgency. Brain CT and vascular imaging are part of the acute stroke pathway; diffusion-weighted MRI more sensitively defines small infarcts and multiterritorial distribution, while blood-sensitive sequences contribute to hemorrhage assessment. The pattern should be compared with rhythm monitoring, aortic atheroma, vascular disease, and other stroke mechanisms in cancer. For abdominal pain, limb ischemia, or suspected renal infarction, imaging is directed to the relevant territory and the possibility of revascularization. The presence of a vegetation makes a cardiac source plausible but does not prove that every ischemic lesion derives from it.

When surgical material is available, sample allocation should be agreed in advance: a portion intended for microbiology should be sent under appropriate conditions without fixing all tissue in formalin; another portion is used for histologic examination. The description of fibrin and platelets should be integrated with assessment of inflammation, organization, necrosis, and microorganisms, considering cultures, stains, and indicated molecular methods. A thrombus recovered during thrombectomy may provide useful information, but the presence of fibrin does not necessarily localize its origin to the valve. Clinicopathologic correlation therefore remains essential even when tissue is available.

The diagnostic yield of tests must be interpreted in relation to patient selection. In the retrospective Cleveland Clinic series, which included 42 cases recognized over twenty years, stroke was the most common presentation; transesophageal echocardiography identified the lesion in 33 of 34 patients in whom it was used, while transthoracic echocardiography allowed diagnosis in 19 cases. These results demonstrate the usefulness of transesophageal echocardiography in high clinical suspicion, but do not constitute a measure of sensitivity and specificity obtained in a population subjected to a uniform independent diagnostic standard. Selective verification, referral to a tertiary center, and nonidentical availability of the two tests limit extrapolation to general oncologic screening.

An initially negative echocardiogram therefore requires a decision based on residual probability. Inadequate technical quality, a recent embolus that reduced the deposit, or a lesion still too small may explain failure to detect it; conversely, weak suspicion and a convincing alternative explanation may make indefinite repetition of the examination unhelpful. Reassessment should compare the same segments and views while considering differences in resolution and acquisition. An artifact should be investigated by verifying whether the structure is recognizable in different views and whether its anatomic relationship is coherent. Better image definition improves description but does not by itself convert a mass into an etiologic diagnosis.

Negative cultures and positive antibodies can also be falsely reassuring when interpreted separately. Some infections are accompanied by autoantibodies, complement consumption, or immunologic manifestations, and a patient with lupus may receive antibiotics before cultures are obtained. The diagnostic problem is not solved by assigning a single finding to each of the two categories. One must ask which hypothesis best explains the entire sequence, which data remain incompatible, and which test could actually reduce uncertainty. This reassessment of probability is particularly useful before starting intense immunosuppression or withholding antimicrobial treatment in a still-suspicious clinical picture.

Treatment and prognosis

Treatment must act simultaneously on the embolic source, the predisposing condition, and organ damage already produced. In NBTE with embolization, anticoagulation is a central component, but no large dedicated randomized trials define the optimal drug, intensity, and duration for all phenotypes. Recommendations derive from case series, clinical experience, and guidance related to the underlying disease. This limitation does not mean that no strategy is available: it requires specification of why a treatment is chosen, which risks condition it, and through which clinical events its effectiveness will be judged.

In cancer-associated disease, available experience mainly supports unfractionated heparin or low-molecular-weight heparin. The former can be useful when rapid titration and reversibility are needed, for example around procedures or during major clinical changes; the latter provide more predictable administration but require attention to renal function, body weight, and bleeding risk. Choice and monitoring should consider platelet count, bleeding, tumor location, and upcoming procedures. Direct oral anticoagulants have evidence in cancer-associated venous thromboembolism, but those studies do not automatically demonstrate efficacy in preventing arterial emboli from NBTE. A prescription based solely on the fact that a patient has “thrombosis and cancer” therefore overlooks a clinically relevant distinction.

Oncologic treatment is part of controlling thrombogenicity. Reduction in tumor burden and activity may decrease the coagulation stimulus, whereas progression may sustain new deposits despite anticoagulation. Rapid cancer control is not always possible; in these cases, embolic prevention must be balanced against prognosis, bleeding risk, need for procedures, and the patient's goals. Echocardiographic disappearance of the vegetation alone is not a criterion for stopping treatment if active malignancy or another strong thrombotic indication persists. There is no fixed duration applicable to every NBTE that can automatically be equated with that of thrombosis provoked by a transient factor.

In thrombotic antiphospholipid syndrome, long-term strategy follows the syndrome profile. Vitamin K antagonists are the usual reference treatment; for venous thrombosis, an INR of 2–3 is generally used, whereas after an arterial event recommendations contemplate INR 2–3 or 3–4 according to the balance between recurrence and bleeding. These targets are not diagnostic criteria for NBTE and should not be transferred without assessment to an incidental vegetation. Rivaroxaban should not be considered equivalent to warfarin in high-risk triple-positive patients; direct oral anticoagulants are not the reference choice in forms with arterial events. Treatment of concomitant lupus controls the autoimmune component but does not replace anticoagulation when a specific thrombotic indication exists.

A recent stroke requires a specific decision regarding timing of anticoagulation. Infarct extent, hemorrhagic transformation, primary hemorrhage, possible brain metastases, thrombocytopenia, and residual probability of infection modify the balance between prevention of further emboli and intracranial risk. Applying a single interval to all patients or regarding every cerebral lesion as a permanent contraindication is inappropriate. In the acute phase, thrombolysis or thrombectomy follows neurologic and vascular assessment of the individual case; the possible infectious nature of the vegetation substantially changes the balance of reperfusion therapy and must be communicated to the stroke team.

A new event during therapy first requires verification of its nature. Neurologic worsening may result from a new embolus, hemorrhage, edema from a known infarct, or another cerebral disease. If recurrent ischemia is confirmed, drug exposure, interruptions, absorption, renal function, platelet trends, and activity of the systemic cause are reassessed, along with repeat evaluation of embolic sources when indicated. Only after this reconstruction can failure of the antithrombotic strategy be discussed and modification considered. Empirically adding an antiplatelet agent to an anticoagulant does not guarantee control of the valvular process and increases bleeding risk.

Surgery is considered mainly in the presence of severe valvular dysfunction with hemodynamic consequences or recurrent emboli despite adequate treatment, when the valvular source remains plausible and the expected benefit exceeds risk. There is no vegetation-size threshold validated specifically for NBTE that can be applied as an automatic operative indication. In a patient with cancer, feasibility of causal treatment, recovery from stroke, frailty, and time required to benefit from surgery must be considered. Repair or replacement depends on anatomy; removing the vegetation does not eliminate the procoagulant state, and antithrombotic planning must continue after the operation.

Antibiotics do not treat a vegetation demonstrated to be sterile, but initial empirical therapy may be necessary when infection has not yet been reasonably excluded and the clinical picture makes it dangerous. Procedural antibiotic prophylaxis likewise depends on the risk categories defined for infective endocarditis, not on the diagnosis of NBTE alone. Follow-up should monitor new embolic events, valvular function, response of the causative disease, and anticoagulant safety; serial echocardiography is scheduled according to evolution rather than a universal timetable. Prognosis is often severe in disseminated malignancy, but this cannot be extrapolated to autoimmune forms or a treatable tumor. Systemic prognosis and valvular prognosis should be described separately to prevent the cardiac finding from becoming an indiscriminate prediction of terminal illness.

Pharmacology helps make antithrombotic choice coherent. Heparins potentiate antithrombin activity; unfractionated heparin inhibits thrombin and factor Xa, whereas low-molecular-weight heparins have relatively greater anti-Xa activity. Vitamin K antagonists reduce functional synthesis of vitamin K-dependent factors and require time to reach a stable effect, in addition to being sensitive to dietary changes, interactions, and hepatic function. These differences explain different administration and monitoring requirements, but do not by themselves demonstrate the clinical superiority of one drug against NBTE. Choice must link available evidence with patient characteristics without turning a biologic rationale into comparative proof.

Monitoring must distinguish pharmacologic effect from the underlying coagulopathy. An abnormal baseline aPTT, for example in the presence of lupus anticoagulant, can complicate titration of unfractionated heparin; the laboratory can help select an appropriate method, avoiding interpretation of an abnormal value as automatic proof of over-anticoagulation. With vitamin K antagonists, quality of control over time matters, not only the INR measured on the day of an event. With low-molecular-weight heparins, renal function and body weight should be reassessed when they change substantially; a dose initially appropriate may no longer be so after acute renal failure or a major clinical change. These checks help identify a correctable cause before concluding that the disease is refractory.

Management of procedures requires an explicit plan for interruption and resumption. Biopsies, drainages, and operations may be essential to treat malignancy or a complication, but repeated interruptions can leave the patient exposed to new events. Duration of interruption depends on the drug, renal function, procedural bleeding risk, and clinical urgency; it cannot be inferred from the diagnosis of NBTE alone. When the expected benefit of anticoagulation becomes uncertain because of recurrent bleeding, advanced disease, or changed goals of care, the decision should be reassessed and documented. Reducing treatment burden in a palliative setting does not mean denying embolic risk, but redefining goals in relation to the actual situation.

Complications

Cerebrovascular complications include major infarcts, recurrent emboli, and multiple lesions that can cause cumulative disability even without a single devastating event. Cognitive impairment, dysphagia, and loss of independence also influence the ability to take medications and receive oncologic treatment. Hemorrhagic transformation of an infarct creates a conflict between bleeding risk and the risk of further emboli, requiring clinical and radiologic reassessment. Reduction of the vegetation does not imply recovery of already infarcted brain tissue; secondary prevention and rehabilitation therefore remain separate goals.

Visceral or peripheral embolization can cause renal and splenic infarctions, mesenteric ischemia, arterial occlusions of the limbs, and, more rarely, embolic coronary ischemia. Splenic or renal lesions may remain paucisymptomatic, whereas mesenteric and limb ischemia require rapid recognition to avoid necrosis and organ loss. Multiple renal infarctions can worsen renal function and alter anticoagulant management; abdominal pain in a patient with cancer should not automatically be attributed to the tumor. Associated deep-vein thrombosis and pulmonary embolism document systemic risk but, in left-sided valvular forms, are not the direct anatomic consequence of the vegetation.

Valvular regurgitation may cause volume overload and heart failure when coaptation becomes insufficient. Progression should be assessed by measuring the defect and its ventricular and pulmonary consequences, not merely by comparing the diameter of the mass. Rapidly destructive deterioration suggests reopening the hypothesis of superimposed infection. An already altered endocardial surface may in fact be colonized during bacteremia, especially in the presence of vascular access and immunosuppression; new fever, positive cultures, or new perivalvular lesions should not be dismissed on the basis of a previous diagnosis of sterility.

Hemorrhagic complications result from the interaction among anticoagulation, fragility of neoplastic lesions, thrombocytopenia, renal failure, and consumptive coagulopathy. Bleeding and thrombosis can coexist and cannot be managed by observing a single laboratory value. An unexpected fall in platelet count during heparin therapy also requires consideration of drug-related causes, in addition to tumor progression, infection, and consumption. After surgery, risks of embolism, prosthetic thrombosis, bleeding, and recurrence of deposits persist if the systemic cause remains active. Iatrogenic complications should therefore be sought alongside progression of NBTE, because intensifying treatment on the basis of an incorrect attribution can worsen the condition rather than correct it.