Causal epidemiology for regulatory risk analysis seeks to evaluate how removing or reducing exposures would change disease occurrence rates. We define interventional probability of causation (IPoC) as the change in probability of a disease (or other harm) occurring over a lifetime or other specified time interval that would be caused by a specified change in exposure, as predicted by a fully specified causal model. We define the closely related concept of causal assigned share (CAS) as the predicted fraction of disease risk that would be removed or prevented by a specified reduction in exposure, holding other variables fixed. Traditional approaches used to evaluate the preventable risk implications of epidemiological associations, including population attributable fraction (PAF) and the Bradford Hill considerations, cannot reveal whether removing a risk factor would reduce disease incidence. We argue that modern formal causal models coupled with causal artificial intelligence (CAI) and realistically partial and imperfect knowledge of underlying disease mechanisms, show great promise for determining and quantifying IPoC and CAS for exposures and diseases of practical interest.
We briefly review key CAI concepts and terms and then apply them to define IPoC and CAS. We present steps to quantify IPoC using a fully specified causal Bayesian network (BN) model. Useful bounds for quantitative IPoC and CAS calculations are derived for a two-stage clonal expansion (TSCE) model for carcinogenesis and illustrated by applying them to benzene and formaldehyde based on available epidemiological and partial mechanistic evidence.
Causal BN models for benzene and risk of acute myeloid leukemia (AML) incorporating mechanistic, toxicological and epidemiological findings show that prolonged high-intensity exposure to benzene can increase risk of AML (IPoC of up to 7e-5, CAS of up to 54%). By contrast, no causal pathway leading from formaldehyde exposure to increased risk of AML was identified, consistent with much previous mechanistic, toxicological and epidemiological evidence; therefore, the IPoC and CAS for formaldehyde-induced AML are likely to be zero.
We conclude that the IPoC approach can differentiate between likely and unlikely causal factors and can provide useful upper bounds for IPoC and CAS for some exposures and diseases of practical importance. For causal factors, IPoC can help to estimate the quantitative impacts on health risks of reducing exposures, even in situations where mechanistic evidence is realistically incomplete and individual-level exposure-response parameters are uncertain. This illustrates the strength that can be gained for causal inference by using causal models to generate testable hypotheses and then obtaining toxicological data to test the hypotheses implied by the models-and, where necessary, refine the models. This virtuous cycle provides additional insight into causal determinations that may not be available from weight-of-evidence considerations alone.

Population attributable fraction (PAF), probability of causation, burden of disease, and related quantities derived from relative risk ratios are widely used in applied epidemiology and health risk analysis to quantify the extent to which reducing or eliminating exposures would reduce disease risks. This causal interpretation conflates association with causation. It has sometimes led to demonstrably mistaken predictions and ineffective risk management recommendations. Causal artificial intelligence (CAI) methods developed at the intersection of many scientific disciplines over the past century instead use quantitative high-level descriptions of networks of causal mechanisms (typically represented by conditional probability tables or structural equations) to predict the effects caused by interventions. We summarize these developments and discuss how CAI methods can be applied to realistically imperfect data and knowledge - e.g., with unobserved (latent) variables, missing data, measurement errors, interindividual heterogeneity in exposure-response functions, and model uncertainty. We recommend that CAI methods can help to improve the conceptual foundations and practical value of epidemiological calculations by replacing association-based attributions of risk to exposures or other risk factors with causal predictions of the changes in health effects caused by interventions.

We argue that population attributable fractions, probabilities of causation, burdens of disease, and similar association-based measures often do not provide valid estimates or surrogates for the fraction or number of disease cases that would be prevented by eliminating or reducing an exposure because their calculations do not include crucial mechanistic information. We use a thought experiment with a cascade of dominos to illustrate the need for mechanistic information when answering questions about how changing exposures changes risk. We suggest that modern methods of causal artificial intelligence (CAI) can fill this gap: they can complement and extend traditional epidemiological attribution calculations to provide information useful for risk management decisions.

A worldwide increase in the incidence of thyroid cancer during the last decades is largely due to papillary thyroid microcarcinomas (MPTCs), which are mostly low-risk tumors. In view of recent clinical recommendations to reduce the extent of surgery for low-risk thyroid cancer, and persisting uncertainty about the impact of radiation history, we set out to address whether clinicopathological characteristics and prognosis of post-Chornobyl MPTCs were changing with regard to: i) the latency period, ii) probability of causation (POC) of a tumor due to radiation, and iii) tumor size.
Patients (n = 465) aged up to 50 years at diagnosis who lived in April, 1986 in six northern, most radiocontaminated regions of Ukraine were studied.
Latency period was statistically significantly associated with the reduction of POC level, tumor size and the frequency of fully encapsulated MPTCs. In contrast, the frequency of oncocytic changes and the BRAFV600E mutation increased. Invasive properties and clinical follow-up results did not depend on latency except for a lower frequency of complete remission after postsurgical radioiodine therapy. The POC level was associated with more frequent extrathyroidal extension, and lymphatic/vascular invasion, less frequent oncocytic changes and BRAFV600E , and did not associate with any clinical indicator. Tumor size was negatively associated with the latency period and BRAFV600E , and had a statistically significant effect on invasive properties of MPTCs: both the integrative invasiveness score and its components such as lymphatic/vascular invasion, extrathyroidal extension and lymph node metastases increased. The frequency of total thyroidectomy, neck lymph node dissection and radioiodine therapy also increased with the larger tumor size. The duration of the latency period, POC level or tumor size did not associate with the chance of disease recurrence.
In summary, we did not observe overall worsening of the clinicopathological features or treatment results of radiogenic MPTCs that could be associated with the latency period or POC level, suggesting that radiation history did not strongly affect those in the analyzed MPTC patients. However, the increase in the invasive properties with tumor size indicates the need for individual risk stratification for each MPTC patient, regardless of radiation history, for treatment decision-making.

Objective: To analyze the diagnosis of 3 cases of leukemia applying for the diagnosis of occupational radiogenic neoplasms. Methods: Retrospective analysis the occupational history, the disease history and the probability of causation (PC value) information of 3 radiological workers. Results: Two cases\' PC value of 95% confidence limit of were >50%, and they were diagnosed as radiogenic neoplasms. One case was <50% and diagnosed as nonoccupational radiogenic neoplasms. Conclusion: The probability of causation analysis has important guiding significance for the diagnosis of occupational radiogenic neoplasms. Radiological workers should improve their awareness of self-protection and reduce the occurrence of occupational diseases.
目的： 对申请职业性放射性肿瘤诊断的3例白血病进行诊断情况分析。 方法： 回顾分析3例放射工作人员职业史、患病史及病因概率PC值（probability of causation）计算资料。 结果： 2例患者经计算得到95%可信限上限的PC值>50%，诊断为职业性放射性肿瘤。1例患者95%可信限上限PC值<50%，诊断为非职业性放射性肿瘤。 结论： 病因概率分析对职业性放射性肿瘤的诊断有重要的指导意义；放射工作人员需提高自身防护意识，减少职业病的发生。.

Personalized medicine asks if a new treatment will help a particular patient, rather than if it improves the average response in a population. Without a causal model to distinguish these questions, interpretational mistakes arise. These mistakes are seen in an article by Demidenko [2016] that recommends the \"D-value,\" which is the probability that a randomly chosen person from the new-treatment group has a higher value for the outcome than a randomly chosen person from the control-treatment group. The abstract states \"The D-value has a clear interpretation as the proportion of patients who get worse after the treatment\" with similar assertions appearing later. We show these statements are incorrect because they require assumptions about the potential outcomes which are neither testable in randomized experiments nor plausible in general. The D-value will not equal the proportion of patients who get worse after treatment if (as expected) those outcomes are correlated. Independence of potential outcomes is unrealistic and eliminates any personalized treatment effects; with dependence, the D-value can even imply treatment is better than control even though most patients are harmed by the treatment. Thus, D-values are misleading for personalized medicine. To prevent misunderstandings, we advise incorporating causal models into basic statistics education.

We show how experimental results can be generalized across diverse populations by leveraging knowledge of local mechanisms that produce the outcome of interest, only some of which may differ in the target domain. We use structural causal models and a refined version of selection diagrams to represent such knowledge, and to decide whether it entails the invariance of probabilities of causation across populations, which then enables generalization. We further provide: (i) bounds for the target effect when some of these conditions are violated; (ii) new identification results for probabilities of causation and the transported causal effect when trials from multiple source domains are available; as well as (iii) a Bayesian approach for estimating the transported causal effect from finite samples. We illustrate these methods both with simulated data and with a real example that transports the effects of Vitamin A supplementation on childhood mortality across different regions.

L\'applicazione del metodo della Probability of Causation (PC) è certamente ampio poiché è oggi lo strumento riconosciuto per la individuazione del nesso di causa non solo nelle richieste di indennizzo in ambito assicurativo (per il quale è stata utilizzato inizialmente) ma anche per dirimere contenziosi giuridici in ambito civilistico e penalistico.
Thanks to the Italian Association of Medical Radiation Protection (AIRM), PC method has been recently proposed as an aid for the radiation protection occupational physician in medical assessments involving both the mandatory actions that, in case of suspicion of occupational disease, the physician needs to perform (report / complaint / reporting) and the expression of the fitness evaluation in case of return to work after cancer and clinical recovery.
For all these uses PC value, calculated through the method, should be used in a flexible manner, and thus lead to different decisions, \"modulated\" on the basis of purposes and listed contexts; and this not only within the legal framework, but also in the strictly professional one.
According to different purposes, different PC values are proposed as a reference for the decisions to be taken.
L’applicazione del metodo della Probability of Causation (PC) è certamente ampio poiché è oggi lo strumento riconosciuto per la individuazione del nesso di causa non solo nelle richieste di indennizzo in ambito assicurativo (per il quale è stata utilizzato inizialmente) ma anche per dirimere contenziosi giuridici in ambito civilistico e penalistico. Recentemente, anche grazie alla spinta dell’Associazione Italiana di Radioprotezione Medica (AIRM), il metodo della PC è stato proposto come strumento di ausilio per il medico di radioprotezione nelle valutazioni che riguardano sia le azioni obbligatorie che, in caso di sospetto di malattia professionale, il medico deve compiere (referto/denuncia/segnalazione), sia per l’espressione del giudizio di idoneità in caso di rientro al lavoro dopo neoplasia e guarigione clinica. Per tutti questi utilizzi il valore di PC calcolato attraverso il metodo dovrà essere usato in maniera flessibile, e quindi dar luogo a decisioni diverse, “modulate” sulla base delle finalità e dei contesti elencati; e questo non solo all’interno del contesto giuridico, ma anche in quello strettamente professionale. Vengono quindi proposti, in funzione delle diverse finalità, dei valori della PC ai quali fare riferimento per le decisioni da adottare.

The Probability of Causation (PC) was introduced to compensate objectively and more possible legally the U.S. diseased subjects involved in the nuclear armament activities.
The methodology is related to the attributable risk concept, but it is widely different from it, since it doesn\'t evaluate the \"attributablity\" from a collective point of view, but from a \"personalistic\" point, that is from the particular exposure condition, from the specific physical parameters and from the biological individual features of the single exposed subject. So the PC become an evaluation of the harm probability \"tailored\" for \"that\" specific exposed person, on the basis of the epidemiological indications coming from an exposed group with very similar characteristics of the under investigation individual. This is clearly possible owing to the large and exhaustive amount of epidemiological studies in the specific field of radiation exposure. The process to reach the PC adoption took a long time, was plodding and politically thwarted and various reexaminations and bills during time were necessary to extended the laws to the different exposure categories.
Now in the U.S. three departments (Health, Energy and Labour) are involved in the evaluation processes; they gather the personal, dosimetric and clinical data and with a computer program (usable on line also) based on the updated knowledge, evaluate the eligibility for compensation on the basis of the \"more likely than not\" criterion.
The method meets the interest and the favor at international level and organizations in prominent positions in the pacific use of nuclear energy and in the radiation protection fields, like: NCRP, IAEA, WHO, ILO,... fight for it use. Now many institutional organism and the more enlightened justice courts utilize the PC to settle cases (increasing in frequency) in work and health activities, for which more often compensation claims are dealing with.
La probabilità di causa nasce per risarcire nel modo più obiettivo possibile e più giuridicamente corretto una serie di individui coinvolti negli Stati Uniti nello sviluppo dell’energia nucleare bellica. Questa metodologia si ispira al concetto di rischio attribuibile, ma si differenzia profondamente da esso in relazione al fatto che non valuta l’”attribuibilità” da un punto di vista collettivo, ma da un punto di vista “personalistico”, cioè con riferimento alle caratteristiche espositive, fisiche e biologiche dell’individuo in studio. Diviene così una probabilità “ritagliata” sul singolo soggetto esposto, sulla base delle indicazioni epidemiologiche di gruppi di individui dalle caratteristiche sovrapponibili a quelle del soggetto in studio. Questo è naturalmente possibile in grazia della grande quantità di studi epidemiologici e del dettaglio dei medesimi nel campo dell’esposizione a radiazioni ionizzanti. Il processo che ha portato all’adozione della PC è stato lungo, laborioso e politicamente contrastato e per estenderlo a tutte le categorie di esposti ha richiesto numerosi riesami e quindi varie leggi che si sono susseguite nel tempo. Oggi negli S.U. tre dipartimenti (salute, energia, lavoro) sono impegnati nel procedimento di valutazione che, dopo la raccolta dei dati personali, attraverso un programma di calcolo ad hoc, valutano il diritto all’indennizzabilità sulla base del criterio “più probabile che no”. Il metodo ha incontrato l’interesse ed il favore a livello mondiale e organismi di grande rilievo nel campo della radioprotezione e dell’utilizzo delle radiazioni, come: il NCRP, l’IAEA, il WHO, l’ILO ne propugnano l’utilizzo, utilizzo di cui si servono alcuni organismi istituzionali e le più illuminate coorti di giustizia per dirimere i contenziosi (purtroppo frequenti) nel contesto lavorativo ed in quello sanitario, nei quali più frequenti sono i casi di richiesta di indennizzo.

The continuous scientific advances against neoplastic diseases affecting all areas of oncology biomedical research. Age is an extremely important factor in cancer development, since the incidence of cancer increases significantly with age. Because of aging of the Italian population, although the incidence is kept constant, the number of cancer diagnosis is inevitably going to increase over time only to increasing age.
Survival after the diagnosis of cancer is one of the main indicators that allow to evaluate the effectiveness of the health system against the cancer disease. The 5-year survival after diagnosis is a widely used indicator. If we consider the relative survival data after 5 years of diagnosis, for cancer cases diagnosed in subsequent three-year periods, from 1990-1992 to 2005-2007, it shows that the 5-year survival has increased significantly over time for both men and women. Many so-called patients \"long-term survivors \"are of working age and should return to work. This aims to ensure both the mental and social well-being of the worker, both industrial production. For the oncogenic risk assessment by ionizing radiation, the ICRP Publication 60 has referred to the mortality and cancer data collected from 1950 to 1985 by the RERF, Japan-US bi-national institution with headquarters in Hiroshima that leads the research program called Life Span study (LSS), that is the study of the long-term effects on survivors of the bomb A. For the thyroid, instead, reference is made to the data from medical irradiations, as well as for liver and bone, using in this case adapted data relating to exposure to alpha rays (thorium and radio). The interpretation model is the traditional one: the linear dose-effect assumptions without a threshold even at small doses (LNT theory) when epidemiological data are not more informative for statistical uncertainty, although we resort to radiobiological studies.
In transferring the risk among different populations ICRP in Publication 103 accommodates the idea that for each type of cancer is more suitable, from time to time, the additive or multiplicative model or a combination of the two.
To study the oncogenic role of occupational exposure to ionizing radiation in the onset of neoplastic disease, the probability of cause (PC), is a \"reasonable way to address the problem of evaluation of the likelihood that previous exposure to ionizing radiation (IR) is responsible for an oncogenic event \"(Committee on Radiation Protection and Measurements - NCRP - Statement N. 7 of 30/09/92).
I continui progressi scientifici contro le malattie neoplastiche riguardano tutti i settori della ricerca biomedica oncologica. L’età è un fattore di estrema rilevanza nello sviluppo del cancro, dal momento che l’incidenza dei tumori aumenta significativamente con l’età. In virtù di un costante invecchiamento della popolazione italiana, anche se l’incidenza si mantenesse costante, il numero di diagnosi tumorali è inevitabilmente destinato ad aumentare nel corso del tempo solo per motivi anagrafici. La sopravvivenza dopo la diagnosi di tumore è uno dei principali indicatori che permette di valutare l’efficacia del sistema sanitario nei confronti della patologia tumorale. La sopravvivenza a 5 anni dalla diagnosi è un indicatore ampiamente entrato nell’uso comune. Se si considerano i dati di sopravvivenza relativa dopo 5 anni dalla diagnosi, per i casi di tumore diagnosticati in trienni successivi, dal 1990-1992 al 2005-2007 emerge che la sopravvivenza a 5 anni è aumentata notevolmente nel tempo sia per gli uomini che per le donne. Molti dei cosiddetti “lungopravviventi” sono in età lavorativa e quindi protagonisti attivi del processo di inserimento occupazionale che deve vedere il ritorno al lavoro del soggetto “oncologico” sia con la finalità di garantire il benessere psicofisco e sociale del lavoratore stesso, sia per finalità produttive. Per la valutazione del rischio oncogeno da radiazioni ionizzanti l’ICRP Pubblicazione 60 ha fatto riferimento ai dati di mortalità e neoplasia raccolti dal 1950 al 1985 dalla RERF, Istituzione binazionale nippostatunitense con sede ad Hiroshima che conduce il programma di ricerca denominato Life Span Study (LSS), cioè lo studio degli effetti a lungo termine sui sopravvissuti alla bomba A. Per la tiroide invece viene fatto riferimento ai dati da irradiazioni mediche, così come per fegato e osso, impiegando in questo caso dati opportunamente adattati relativi a sposizione a raggi alfa (torio e radio). Il modello interpretativo rimane quello tradizionale: l’ipotesi lineare dose-effetto, senza una soglia neppure alle piccole dosi (LNT theory) per le quali i dati epidemiologici non sono più informativi per ragioni di incertezza statistica, anche se si ricorre agli studi radiobiologici. Nel trasferire il rischio tra popolazioni L’ICRP nella Pubblicazione 103 accoglie l’idea che per ogni tipo di tumore sia più adatto, di volta in volta, il modello additivo o moltiplicativo o una combinazione tra i due. Per lo studio del ruolo oncogeno della esposizione occupazionale a radiazioni ionizzanti nell’insorgenza della malattia neoplastica, la Probabilità di causa (PC), rappresenta un “modo ragionevole per indirizzare il problema della valutazione della verosimiglianza che una precedente esposizione a radiazioni ionizzanti (RI) sia responsabile di un evento oncogeno” (Committee on Radiation Protection and Measurements - NCRP - Statement n° 7 del 30/09/92).