Cost-benefit considerations of the biased diagnostician.

OBJECTIVES: In the cognitive process of establishing a diagnosis, the performance of a diagnostician can be characterized in terms of sensitivity and specificity. The aims of the present study are to analyze in quantitative terms how cognitive bias affects the performance of a diagnostician, and how a diagnostician's biased decision making is further influenced by personal cost-benefit considerations. METHODS: The test matrices of two sequential diagnostic tests are manipulated according to the rules of linear algebra, using multiplication of the second with the first test matrix to calculate their joint test characteristics. The decision tree and receiver operating characteristic (ROC) of a biased and unbiased diagnostician are used to calculate which combination of test characteristics maximizes the expected utility value. RESULTS: Biased diagnosticians cannot establish a diagnosis beyond their own limited or distorted level of understanding. An unbiased and a biased diagnostician alike adjust their choice of test characteristics according to their different cost-benefit estimation of the various test outcomes. From the perspective of an unbiased diagnostician, the choices made by a biased diagnostician appear to invert reality. However, the same appearance of inverted reality is perceived by the biased diagnostician, judging the choices made by the unbiased diagnostician. CONCLUSIONS: As a general principle, human testers cannot test beyond their own level of understanding. They only see what they know. As they base their judgment on preconceived notions about the utilities associated with different test outcomes, human testers also tend to only know what they want to know.

Optimal timing of a colonoscopy screening schedule depends on adenoma detection, adenoma risk, adherence to screening and the screening objective: A microsimulation study.

Colonoscopy-based screening provides protection against colorectal cancer (CRC), but the optimal starting age and time intervals of screening colonoscopies are unknown. We aimed to determine an optimal screening schedule for the US population and its dependencies on the objective of screening (life years gained or incidence, mortality, or cost reduction) and the setting in which screening is performed. We used our established open-source microsimulation model CMOST to calculate optimized colonoscopy schedules with one, two, three or four screening colonoscopies between 20 and 90 years of age. A single screening colonoscopy was most effective in reducing life years lost from CRC when performed at 55 years of age. Two, three and four screening colonoscopy schedules saved a maximum number of life years when performed between 49-64 years; 44-69 years; and 40-72 years; respectively. However, for maximum incidence and mortality reduction, screening colonoscopies needed to be scheduled 4-8 years later in life. The optimum was also influenced by adenoma detection efficiency with lower values for these parameters favoring a later starting age of screening. Low adherence to screening consistently favored a later start and an earlier end of screening. In a personalized approach, optimal screening would start earlier for high-risk patients and later for low-risk individuals. In conclusion, our microsimulation-based approach supports colonoscopy screening schedule between 45 and 75 years of age but the precise timing depends on the objective of screening, as well as assumptions regarding individual CRC risk, efficiency of adenoma detection during colonoscopy and adherence to screening.

Similar geographic distribution of mortality from multiple sclerosis, Hodgkin lymphoma, and inflammatory bowel disease in the United States.

BACKGROUND: Previous epidemiologic studies have suggested a linkage between the occurrence of multiple sclerosis (MS), Hodgkin lymphoma (HL), Crohn's disease (CD), and ulcerative colitis (UC). It was hypothesized that the 4 diagnoses would be characterized by similar geographic distributions within the United States. AIMS: To compare the US geographic distributions of these 4 diagnoses in a cross-sectional study. METHODS: Using the US vital statistics, state-specific death rates and age-specific proportional mortality ratios (PMR) were calculated for each diagnosis. Similarities in the geographic distributions of the 4 diagnoses were tested by linear and Poisson regression analyses. The PMR values from different states were correlated among pairs of consecutive age-groups. RESULTS: The 6 linear correlation coefficients (r) among the geographic distributions of the 4 diseases were as follows: HL vs. MS (r = 0.28), HL vs. CD (r = 0.74), HL vs. UC (r = 0.64); MS vs. CD (r = 0.18), MS vs. UC (r = 0.66); CD vs. UC (r = 0.58). Using Poisson regression, the geographic distributions of MS, HL, CD, and UC were all found to be significantly correlated with each other. In MS, significant correlations between the PMR values of each two consecutive age-groups started with the age-group 25-44 years. In HL, such significant correlations started at age 10-14, in CD at age 20-24, and in UC at age 20-24 years. CONCLUSIONS: Within the United States, mortality from MS, HL, CD, and UC are characterized by similar geographic distributions. The environmental influences responsible for these resembling geographic distributions start exerting their influence during early lifetime.

Concurrence of inflammatory bowel disease with multiple sclerosis or Hodgkin lymphoma.

BACKGROUND AND AIMS: Epidemiologic evidence suggests that Hodgkin lymphoma (HL) and multiple sclerosis (MS) share a common set of risk factors with Crohn's disease (CD) and ulcerative colitis (UC). It was hypothesized that such shared risk factors would lead to clustering of the 4 diagnoses in the same patients. METHODS: All patients with HL, MS, CD, or UC were identified in the veterans population from 2016-2020 and the Medicare population from 1986 to 1989. In a case-control study, the observed concurrences amongst these 4 diagnoses were compared with their expected frequencies in the overall veterans or Medicare population during the same time period by calculating odds ratios (OR) with their 95% confidence intervals (CI). RESULTS: The study included 6 million veterans and 35 million Medicare patients. In the veterans population, inflammatory bowel disease (IBD) was significantly associated with a concurrent diagnosis of HL (OR: 1.40, 95% CI: 1.15-1.71) and MS (1.34, 1.19-1.50). In the Medicare population, IBD was also significantly associated with HL (1.84, 1.07-3.17) and MS (2.31, 1.59-3.35). Similar trends were observed in CD or UC when analyzed separately in both datasets. In the veterans population, adjustment for the potentially confounding influence of ethnicity, sex, and age left all OR values largely unaffected and statistically significant. CONCLUSION: The concurrence of IBD with HL or MS could reflect on a common pathway in the etiology or pathogenesis of these 4 diseases.

Similar geographic distributions of death rates from inflammatory bowel disease and Hodgkin lymphoma or multiple sclerosis.

BACKGROUND: A previous comparative analysis of the time trends of Hodgkin lymphoma (HL), multiple sclerosis (MS), Crohn's disease (CD), and ulcerative colitis (UC) suggested that the occurrence of all four diseases was precipitated by exposure to similar environmental risk factors during early lifetime. In the present cross-sectional study, it was hypothesized that besides their resembling temporal variations the four diseases would also show similar geographic distributions. METHODS: Using the vital statistics of 21 countries from 1951 to 2020, overall and age-specific death rates from the four diseases were calculated for each individual country. The death rates of different countries were compared using linear regression analysis. RESULTS: The data revealed strikingly similar geographic distributions of all four diseases. Their occurrence was common in Europe and relatively uncommon in countries outside Europe. Further stratification by consecutive age groups showed that for each disease analyzed separately, there were significant correlations amongst each two sequential age groups. In HL and UC, the inter-age correlations started at age 5 years or less. In MS and CD, the inter-age correlations only started at age 15 years. CONCLUSIONS: The similarities in the geographic distributions of death rates from HL, MS, CD, and UC suggest that these four diseases share a set of one or more common environmental risk factors. The data also support the contention that the exposure to such shared risk factors starts during an early period of lifetime.

Progressive changes in platelet counts and Fib-4 scores precede the diagnosis of advanced fibrosis in NASH patients.

BACKGROUND: Cirrhosis and its complications develop in a subgroup of patients with non-alcoholic fatty liver disease (NASH). Early detection of liver fibrosis represents an important goal of clinical care. AIM: To test the hypothesis that the development of cirrhosis in nonalcoholic fatty liver disease patients is preceded by the long-term trends of platelet counts and Fib-4 scores. METHODS: We identified all patients in our healthcare system who had undergone fibrosis staging by liver biopsy or magnetic resonance elastography (MRE) for non-alcoholic fatty liver disease during the past decade (n = 310). Platelet counts, serum glutamic-pyruvic transaminase and serum glutamic oxalacetic transaminase values preceding the staging tests were extracted from the electronic medical record system, and Fib-4 scores were calculated. Potential predictors of advanced fibrosis were evaluated using multivariate regression analysis. RESULTS: Significant decreases in platelet counts and increases in Fib-4 scores were observed in all fibrosis stages, particularly in patients with cirrhosis. In the liver biopsy group, the presence of cirrhosis was best predicted by the combination of the Fib-4 score at the time closest to staging (P < 0.0001), the presence of diabetes (P = 0.0001), and the correlation coefficient of the preceding time-dependent drop in platelet count (P = 0.044). In the MRE group, Fib4 score (P = 0.0025) and platelet drop (P = 0.0373) were significant predictors. In comparison, the time-dependent rise of the Fib-4 score did not contribute in a statistically significant way. CONCLUSION: Time-dependent changes in platelet counts and Fib-4 scores contribute to the prediction of cirrhosis in NASH patients with biopsy- or MRE-staged fibrosis. Their incorporation into predictive algorithms may assist in the earlier identification of high-risk patients.

Case-control study of the concurrence of coeliac disease with inflammatory bowel disease.

BACKGROUND AND AIMS: Previous studies suggested that inflammatory bowel disease (IBD) is associated with an increased prevalence of comorbid coeliac disease. Our case-control study aimed to test this association using a large histopathology database. METHODS: The Inform Diagnostics database is a repository of histopathologic records from patients distributed throughout the United States. In a case-control study among patients with bidirectional endoscopy, we compared the occurrence of coeliac disease in case subjects with IBD or microscopic colitis (MC) and control subjects without inflammatory colitis, calculating odds ratios (OR) and their 95% confidence intervals (CI) adjusted to the varying age, gender and ethnic distributions of case and control subjects. RESULTS: The study population was split into 12,816 IBD cases and 6486 MC cases, who were compared to 345,733 control subjects without colitis. A total of 2892 patients were diagnosed with coeliac disease. Of 12,816 IBD patients, 57 patients (0.4%) harboured coeliac disease compared to 0.7% (2548/345,733) in the control population. The prevalence of coeliac disease among MC patients was 4.4% (288/6486). The corresponding ORs were significantly decreased in IBD (OR: 0.50, CI: 0.38-0.64) and significantly increased in MC patients (6.78, 5.96-7.69). Further stratification of the case populations into subtypes of IBD (Crohn's disease or ulcerative colitis) and MC (collagenous or lymphocytic colitis) similarly revealed significantly decreased and increased ORs for each subtype. CONCLUSIONS: The previously reported positive association between coeliac disease and IBD may have been possibly biased by the inclusion of MC cases in the IBD patient population.

The Occurrence of Gastritis in Microscopic Colitis and Inflammatory Bowel Disease.

Multiple studies have shown that Helicobacter pylori infection is associated with a lower prevalence of inflammatory bowel disease (IBD).1,2 Besides chronic active gastritis (CAG) resulting from gastric infection with H pylori, pathologists have noticed another form of CAG, which is unrelated to H pylori infection and seems to cluster in patients with IBD.3-5 The aim of the present study was to compare the prevalence of H pylori-negative and H pylori-positive CAG in patients with IBD, and microscopic colitis (MC).

Cost-utility advantage of interventional endoscopy.

BACKGROUND: Gastroenterologists frequently face the dilemma of how to choose among different management options. AIM: To develop a tool of medical decision analysis that helps choosing between competing management options of interventional endoscopy and surgery. METHODS: Carcinoma-in-situ of the esophagus, large colonic polyps, and ampullary adenoma serve as three examples for disorders being managed by both techniques. A threshold analysis using a decision tree was modeled to compare the costs and utility values associated with managing the three examples. If the expected healing or success rate of interventional endoscopy exceeds a threshold calculated as the ratio of endoscopy costs over surgery costs, endoscopy becomes the preferred management option. A low threshold speaks in favor of endoscopic intervention as initial management strategy. RESULTS: If the decision in favor of surgery is focused exclusively on preventing death from a given disease, surgical intervention may seem to provide the best treatment option. However, interventional endoscopy becomes a viable alternative, if the comparison is based on a broader perspective that includes adverse events and long-term disability, as well as the healthcare costs of both procedures. For carcinoma-in-situ of the esophagus, the threshold for the expected success rate is 24% (range in the sensitivity analysis: 7-29%); for large colonic polyps it is 10% (5-12%), and for duodenal papillary adenoma it is 17% (5-21%). CONCLUSIONS: Even if a management strategy surpasses its alternative with respect to one important outcome parameter, there is often still room for the lesser alternative to be considered as viable option.

Similar Time Trends of Hodgkin Lymphoma, Multiple Sclerosis, and Inflammatory Bowel Disease.

BACKGROUND: The Epstein-Barr virus (EBV) plays a role in the causation of Hodgkin lymphoma (HL) and multiple sclerosis (MS). A previous study showed that the time trends of mortality from Crohn's disease (CD) and MS shared striking similarities. It was hypothesized that such similarities would also involve the time trends of ulcerative colitis and HL. AIMS: To compare the time trends of CD and UC with those of HL and MS in 6 different countries. METHODS: Using the vital statistics of England, Canada, Netherlands, Scotland, Switzerland, and United States from 1951 to 2020, the time trends of mortality from these 4 diseases were compared. The time-dependent changes of death rates were subjected to a birth-cohort analysis. RESULTS: Similar trends were observed in all 6 countries. UC mortality rose among generations born during the nineteenth century and decreased among all generations born subsequently during the twentieth century. CD mortality was similarly characterized by a birth-cohort pattern with a rise and fall that were shifted by 20-30 years towards more recent generations when compared to UC. The birth-cohort pattern of UC was matched by a similar pattern of HL, whereas the birth-cohort pattern of CD was matched by a similar pattern of MS. CONCLUSIONS: The similarities in the ubiquitous birth-cohort patterns of UC, CD, HL, and MS suggest that these 4 diseases share a common environmental risk factor. Such risk factor may be linked to EBV or its acquisition during an early period of a patient's lifetime.

Successful Hepatitis C Birth Cohort Screening and Linkage to Care in a US Community Health System.

CONTEXT: Birth cohort ("baby boomer") screening represents a well-validated strategy for the identification of asymptomatic hepatitis C-infected patients. However, successful linkage of newly diagnosed patients to antiviral therapy has been more difficult to accomplish. OBJECTIVE: To analyze the results of a systemwide birth cohort screening program in a US community health care system. DESIGN: We analyzed the data from an ongoing hepatitis C virus (HCV) screening and treatment program that was established at NorthShore University Health System in 2015. Hepatitis C virus screening by primary care providers was prompted through automated Best Practice and Health Maintenance alerts. Patient visits and screening orders were tracked using a customized HCV dashboard. Virologic, demographic, and treatment data were assessed and compared with those of a cohort of patients with previously established HCV infection. RESULTS: Since program inception, 61 8161 (64.3%) of the entire NorthShore baby boomer population of 96 001 patients have completed HCV antibody testing, and 160 patients (0.26%) were antibody positive. Of 152 antibody-positive patients who underwent HCV RNA testing, 53 (34.2%) were viremic. A total of 39 of 53 patients (73.6%) underwent antiviral therapy and achieved a sustained virologic response. Compared with patients identified through screening, a comparison cohort of patients with previously established HCV had more advanced fibrosis and significantly lower dropout rates. The COVID-19 pandemic was associated with a decrease in the number of outpatient visits of screening-eligible patients and with a reduction in HCV screening rates. CONCLUSION: Our data demonstrate the electronic medical records-assisted systemwide implementation of HCV birth cohort screening and successful linkage to antiviral therapy in a community-based US multihospital system.

The test characteristics of a biased or ignorant diagnostician.

BACKGROUND: A human diagnostician may harbour a special bias towards favourable positive or negative test results. The aim of the present analysis is to describe in quantitative terms how bias can affect the test characteristics of a human tester. METHODS: Whereas an unbiased tester would give absolute (100%) preference to true positive or true negative test results, and no (0%) preference to any false positive or false negative test results, a biased tester may harbour some preferences towards false positive or false negative tests. Such bias can be phrased in terms of a separate sensitivity-specificity matrix. The bias matrix multiplied with the original test matrix yields the biased test matrix. Similarly, the extent of ignorance by a human tester about the diagnosis is modelled as a separate sensitivity-specificity matrix, which captures the concordance between positive and negative diagnoses made by an ignorant and expert diagnostician. RESULTS: Increasing bias or ignorance result in decreasing test performance with decreasing positive predictive values until the test completely loses its discriminatory power. With more pronounced bias towards false test results, any positive test outcome may even become misinterpreted as predicting the non-existence of a given diagnosis. CONCLUSIONS: The proposed model helps to understand in quantitative terms, how bias and ignorance can alter a diagnostician's interpretation of test outcomes and result in diagnostic errors.

Synchronous occurrence of different polyp types during colonoscopy.

BACKGROUND AND AIMS: The general probabilities of finding different types of hyperplastic or neoplastic colon lesions clustering in the same patient are unknown. We hypothesised that a systematic analysis of polyp clustering would reveal a high frequency of different concurrent types and that such clustering would be influenced by the underlying histopathology of the individual polyps. METHODS: Using an electronic database of histopathologic records, a cross-sectional study of 2,910,174 colonoscopies evaluated the concordant occurrence of hyperplastic and neoplastic mucosal lesions. RESULTS: Amongst patients harbouring any colon polyp, one can expect to find another histopathologic type in about 22%. Being the most prevalent type of polyp, tubular adenoma was the most common type of colonic polyp to be associated with other lesions. For instance, 31% of all colonoscopies with sessile serrated adenomas or hyperplastic polyps also revealed tubular adenomas. Hyperplastic polyp, the second most prevalent type of colonic polyp, was also the second most common type of polyp to be found associated with other types of neoplasia. For instance, 25% of all colonoscopies with sessile serrated adenomas and 31% of those with tubular adenomas also revealed hyperplastic polyps. CONCLUSIONS: Different types of colon polyps commonly coincide in individual patients. The present set of data from a large nationwide database may provide guidance for the endoscopist of what variety in colon polyps to expect during colonoscopy.

Temporal changes in the histology of microscopic colitis.

BACKGROUND: In a subgroup of patients with microscopic colitis [MC], its histopathology changed from lymphocytic [LC] to collagenous colitis [CC] and vice versa. Previous studies have also observed histopathological transitions between MC and inflammatory bowel disease [IBD]. AIMS: The aim of the present study was to analyse the prevalence of such transitions in a large population of MC patients. METHODS: The Inform Diagnostics database is an electronic repository of histopathology records of patients distributed throughout the USA. In a cross-sectional study, we analysed the prevalence of changes in MC histology. Each prevalence was expressed as the rate per 100 MC patients with its 95% Poisson confidence interval. RESULTS: In a total population of 29 307 MC patients, our cross-sectional study focused on a subgroup of 4363 patients who underwent two or more consecutive colonoscopies between December 2008 and March 2020. Overall, 1.6% [95% CI 1.2-2.0%] of patients changed their MC phenotype from LC to CC, and 0.5% [0.3-0.7%] from CC to LC. Of 4363 MC patients, 414 [9.5%] were also diagnosed with IBD. In 2.9% [2.4-3.5%], MC and IBD were diagnosed as synchronous mucosal lesions. In 2.1% [1.7-2.6%], MC changed to IBD, and in 4.5% [3.9-5.2%] IBD changed to MC. CONCLUSIONS: The analysis confirmed the synchronous occurrence of MC and IBD and transitions between the two diagnoses. In patients who fail therapy for either one of the two diseases, the gastroenterologist should search for changes in the underlying phenotype as a possible explanation.

Trends of Colonic Neoplasia in US Outpatient Endoscopy Centers.

BACKGROUND: A variety of studies have shown rising trends in the occurrence of colorectal cancer in younger patients as opposed to falling trends among older patients aged 55 years or more. We hypothesized that the time trends of benign colonic precursor lesions would reveal similar patterns. AIMS: The present study was designed to test this hypothesis in a large nationwide sample of the US population undergoing colonoscopy in community-based endoscopy centers. METHODS: The Inform Diagnostics database is an electronic repository of histopathologic records of patients distributed throughout the USA. A cross-sectional study analyzed the detection rates of sessile serrated adenomas (SSA), hyperplastic polyps (HP), tubular adenomas (TA), traditional serrated adenomas (TSA), or adenocarcinomas (colorectal cancer, CRC) in 2,910,174 colonoscopies done 2008-2020. RESULTS: During the 13-year time period, the rate of SSA showed a significant rise, both in patients younger and older than 55 years. HP and TA both showed a significant decline during the same time period. The trends of CRC in the older age group decreased significantly between 2008 (or its peak in 2012) and 2020. The trends of CRC in the younger age group increased significantly between 2008 and its peak in 2017. CONCLUSIONS: The age-specific time trends of benign and malignant colonic neoplasia are characterized by dissimilar temporal patterns. Such dissimilarity could suggest that besides a set of shared risk factors that affect all types of colonic neoplasia alike, there is yet another set of environmental risk factors that specifically influence malignant transformation.