Effectiveness of SARS-CoV-2 primary vaccines and boosters in patients with type 2 diabetes mellitus in Hungary (HUN-VE 4 Study)

INTRODUCTION

Type 2 diabetes mellitus is a risk factor for severe COVID-19 infection and is associated with increased risk of complications. The present study aimed to investigate effectiveness and persistence of different COVID vaccines in persons with or without diabetes during the Delta wave in Hungary.

RESEARCH DESIGN AND METHODS

Data sources were the national COVID-19 registry data from the National Public Health Center and the National Health Insurance Fund on the total Hungarian population. The adjusted incidence rate ratios and corresponding 95% CIs were derived from a mixed-effect negative binomial regression model.

RESULTS

A population of 672 240 cases with type 2 diabetes and a control group of 2 974 102 non-diabetic persons free from chronic diseases participated. Unvaccinated elderly persons with diabetes had 2.68 (95% CI 2.47 to 2.91) times higher COVID-19-related mortality rate as the 'healthy' controls. Primary immunization effectively equalized the risk of COVID-19 mortality between the two groups. Vaccine effectiveness declined over time, but the booster restored the effectiveness against mortality to over 90%. The adjusted vaccine effectiveness of the primary Pfizer-BioNTech against infection in the 14-120 days of postvaccination period was 71.6 (95% CI 66.3 to 76.1)% in patients aged 65-100 years with type 2 diabetes and 64.52 (95% CI 59.2 to 69.2)% in the controls. Overall, the effectiveness tended to be higher in individuals with diabetes than in controls. The booster vaccines could restore vaccine effectiveness to over 80% concerning risk of infection (eg, patients with diabetes aged 65-100 years: 89.1 (88.1-89.9)% with Pfizer-on-Pfizer, controls 65-100 years old: 86.9 (85.8-88.0)% with Pfizer-on-Pfizer, or patients with diabetes aged 65-100 years: 88.3 (87.2-89.2)% with Pfizer-on-Sinopharm, controls 65-100 years old: 87.8 (86.8-88.7)% with Pfizer-on-Sinopharm).

CONCLUSIONS

Our data suggest that people with type 2 diabetes may have even higher health gain when getting vaccinated as compared with non-diabetic persons, eliminating the marked, COVID-19-related excess risk of this population. Boosters could restore protection.

Opposite trends in incidence of breast cancer in young and old female cohorts in Hungary and the impact of the Covid-19 pandemic: a nationwide study between 2011-2020

Background

This nationwide study examined breast cancer (BC) incidence and mortality rates in Hungary between 2011-2019, and the impact of the Covid-19 pandemic on the incidence and mortality rates in 2020 using the databases of the National Health Insurance Fund (NHIF) and Central Statistical Office (CSO) of Hungary.

Methods

Our nationwide, retrospective study included patients who were newly diagnosed with breast cancer (International Codes of Diseases ICD)-10 C50) between Jan 1, 2011 and Dec 31, 2020. Age-standardized incidence and mortality rates (ASRs) were calculated using European Standard Populations (ESP).

Results

7,729 to 8,233 new breast cancer cases were recorded in the NHIF database annually, and 3,550 to 4,909 all-cause deaths occurred within BC population per year during 2011-2019 period, while 2,096 to 2,223 breast cancer cause-specific death was recorded (CSO). Age-standardized incidence rates varied between 116.73 and 106.16/100,000 PYs, showing a mean annual change of -0.7% (95% CI: -1.21%-0.16%) and a total change of -5.41% (95% CI: -9.24 to -1.32). Age-standardized mortality rates varied between 26.65-24.97/100,000 PYs (mean annual change: -0.58%; 95% CI: -1.31-0.27%; p=0.101; total change: -5.98%; 95% CI: -13.36-2.66). Age-specific incidence rates significantly decreased between 2011 and 2019 in women aged 50-59, 60-69, 80-89, and ≥90 years (-8.22%, -14.28%, -9.14%, and -36.22%, respectively), while it increased in young females by 30.02% (95%CI 17,01%- 51,97%) during the same period. From 2019 to 2020 (in first COVID-19 pandemic year), breast cancer incidence nominally decreased by 12% (incidence rate ratio [RR]: 0.88; 95% CI: 0.69-1.13; 2020 vs. 2019), all-cause mortality nominally increased by 6% (RR: 1.06; 95% CI: 0.79-1.43) among breast cancer patients, and cause-specific mortality did not change (RR: 1.00; 95%CI: 0.86-1.15).

Conclusion

The incidence of breast cancer significantly decreased in older age groups (≥50 years), oppositely increased among young females between 2011 and 2019, while cause-specific mortality in breast cancer patients showed a non-significant decrease. In 2020, the Covid-19 pandemic resulted in a nominal, but not statistically significant, 12% decrease in breast cancer incidence, with no significant increase in cause-specific breast cancer mortality observed during 2020.

Wastewater-based modeling, reconstruction, and prediction for COVID-19 outbreaks in Hungary caused by highly immune evasive variants

(MOTIVATION)

Wastewater-based epidemiology (WBE) has emerged as a promising approach for monitoring the COVID-19 pandemic, since the measurement process is cost-effective and is exposed to fewer potential errors compared to other indicators like hospitalization data or the number of detected cases. Consequently, WBE was gradually becoming a key tool for epidemic surveillance and often the most reliable data source, as the intensity of clinical testing for COVID-19 drastically decreased by the third year of the pandemic. Recent results suggests that the model-based fusion of wastewater measurements with clinical data and other indicators is essential in future epidemic surveillance.

(METHOD)

In this work, we developed a wastewater-based compartmental epidemic model with a two-phase vaccination dynamics and immune evasion. We proposed a multi-step optimization-based data assimilation method for epidemic state reconstruction, parameter estimation, and prediction. The computations make use of the measured viral load in wastewater, the available clinical data (hospital occupancy, delivered vaccine doses, and deaths), the stringency index of the official social distancing rules, and other measures. The current state assessment and the estimation of the current transmission rate and immunity loss allow a plausible prediction of the future progression of the pandemic.

(RESULTS)

Qualitative and quantitative evaluations revealed that the contribution of wastewater data in our computational epidemiological framework makes predictions more reliable. Predictions suggest that at least half of the Hungarian population has lost immunity during the epidemic outbreak caused by the BA.1 and BA.2 subvariants of Omicron in the first half of 2022. We obtained a similar result for the outbreaks caused by the subvariant BA.5 in the second half of 2022.

(APPLICABILITY)

The proposed approach has been used to support COVID management in Hungary and could be customized for other countries as well.

Underlying reasons for post-mortem diagnosed lung cancer cases - A robust retrospective comparative study from Hungary (HULC study)

Objective

The Hungarian Undiagnosed Lung Cancer (HULC) study aimed to explore the potential reasons for missed LC (lung cancer) diagnosis by comparing healthcare and socio-economic data among patients with post-mortem diagnosed LC with those who were diagnosed with LC during their lives.

Methods

This nationwide, retrospective study used the databases of the Hungarian Central Statistical Office (HCSO) and National Health Insurance Fund (NHIF) to identify patients who died between January 1, 2019 and December 31, 2019 and were diagnosed with lung cancer post-mortem (population A) or during their lifetime (population B). Patient characteristics, socio-economic factors, and healthcare resource utilization (HCRU) data were compared between the diagnosed and undiagnosed patient population.

Results

During the study period, 8,435 patients were identified from the HCSO database with LC as the cause of death, of whom 1,203 (14.24%) had no LC-related ICD (International Classification of Diseases) code records in the NHIF database during their lives (post-mortem diagnosed LC population). Post-mortem diagnosed LC patients were significantly older than patients diagnosed while still alive (mean age 71.20 vs. 68.69 years, p<0.001), with a more pronounced age difference among female patients (difference: 4.57 years, p<0.001), and had significantly fewer GP (General Practitioner) and specialist visits, X-ray and CT scans within 7 to 24 months and 6 months before death, although the differences in GP and specialist visits within 7-24 months did not seem clinically relevant. Patients diagnosed with LC while still alive were more likely to be married (47.62% vs. 33.49%), had higher educational attainment, and had more children, than patients diagnosed with LC post-mortem.

Conclusions

Post-mortem diagnosed lung cancer accounts for 14.24% of total lung cancer mortality in Hungary. This study provides valuable insights into patient characteristics, socio-economic factors, and HCRU data potentially associated with a high risk of lung cancer misdiagnosis.

Effectiveness and Waning of Protection With Different SARS-CoV-2 Primary and Booster Vaccines During the Delta Pandemic Wave in 2021 in Hungary (HUN-VE 3 Study)

Background

In late 2021, the pandemic wave was dominated by the Delta SARS-CoV-2 variant in Hungary. Booster vaccines were offered for the vulnerable population starting from August 2021.

Methods

The nationwide HUN-VE 3 study examined the effectiveness and durability of primary immunization and single booster vaccinations in the prevention of SARS-CoV-2 infection, Covid-19 related hospitalization and mortality during the Delta wave, compared to an unvaccinated control population without prior SARS-CoV-2 infection.

Results

The study population included 8,087,988 individuals who were 18-100 years old at the beginning of the pandemic. During the Delta wave, after adjusting for age, sex, calendar day, and chronic diseases, vaccine effectiveness (VE) of primary vaccination against registered SARS-CoV-2 infection was between 11% to 77% and 18% to 79% 14-120 days after primary immunization in the 16-64 and 65-100 years age cohort respectively, while it decreased to close to zero in the younger age group and around 40% or somewhat less in the elderly after 6 months for almost all vaccine types. In the population aged 65-100 years, we found high, 88.1%-92.5% adjusted effectiveness against Covid-19 infection after the Pfizer-BioNTech, and 92.2%-95.6% after the Moderna booster dose, while Sinopharm and Janssen booster doses provided 26.5%-75.3% and 72.9%-100.0% adjusted VE, respectively. Adjusted VE against Covid-19 related hospitalization was high within 14-120 days for Pfizer-BioNTech: 76.6%, Moderna: 83.8%, Sputnik-V: 78.3%, AstraZeneca: 73.8%, while modest for Sinopharm: 45.7% and Janssen: 26.4%. The waning of protection against Covid-19 related hospitalization was modest and booster vaccination with mRNA vaccines or the Janssen vaccine increased adjusted VE up to almost 100%, while the Sinopharm booster dose proved to be less effective. VE against Covid-19 related death after primary immunization was high or moderate: for Pfizer-BioNTech: 81.5%, Moderna: 93.2%, Sputnik-V: 100.0%, AstraZeneca: 84.8%, Sinopharm: 58.6%, Janssen: 53.3%). VE against this outcome also showed a moderate decline over time, while booster vaccine types restored effectiveness up to almost 100%, except for the Sinopharm booster.

Conclusions

The HUN-VE 3 study demonstrated waning VE with all vaccine types for all examined outcomes during the Delta wave and confirmed the outstanding benefit of booster vaccination with the mRNA or Janssen vaccines, and this is the first study to provide clear and comparable effectiveness results for six different vaccine types after primary immunization against severe during the Delta pandemic wave.

Nationwide Effectiveness of First and Second SARS-CoV2 Booster Vaccines During the Delta and Omicron Pandemic Waves in Hungary (HUN-VE 2 Study)

Background

In Hungary, the pandemic waves in late 2021 and early 2022 were dominated by the Delta and Omicron SARS-CoV-2 variants, respectively. Booster vaccines were offered with one or two doses for the vulnerable population during these periods.

Methods and Findings

The nationwide HUN-VE 2 study examined the effectiveness of primary immunization, single booster, and double booster vaccination in the prevention of Covid-19 related mortality during the Delta and Omicron waves, compared to an unvaccinated control population without prior SARS-CoV-2 infection during the same study periods. The risk of Covid-19 related death was 55% lower during the Omicron vs. Delta wave in the whole study population (n=9,569,648 and n=9,581,927, respectively; rate ratio [RR]: 0.45, 95% confidence interval [CI]: 0.44-0.48). During the Delta wave, the risk of Covid-19 related death was 74% lower in the primary immunized population (RR: 0.26; 95% CI: 0.25-0.28) and 96% lower in the booster immunized population (RR: 0.04; 95% CI: 0.04-0.05), vs. the unvaccinated control group. During the Omicron wave, the risk of Covid-19 related death was 40% lower in the primary immunized population (RR: 0.60; 95% CI: 0.55-0.65) and 82% lower in the booster immunized population (RR: 0.18; 95% CI: 0.16-0.2) vs. the unvaccinated control group. The double booster immunized population had a 93% lower risk of Covid-19 related death compared to those with only one booster dose (RR: 0.07; 95% CI. 0.01-0.46). The benefit of the second booster was slightly more pronounced in older age groups.

Conclusions

The HUN-VE 2 study demonstrated the significantly lower risk of Covid-19 related mortality associated with the Omicron vs. Delta variant and confirmed the benefit of single and double booster vaccination against Covid-19 related death. Furthermore, the results showed the additional benefit of a second booster dose in terms of SARS-CoV-2 infection and Covid-19 related mortality.

Booster Vaccination Decreases 28-Day All-Cause Mortality of the Elderly Hospitalized Due to SARS-CoV-2 Delta Variant

(1) Background: SARS-CoV-2 infections are associated with an increased risk of hospital admissions especially in the elderly (age ≥ 65 years) and people with multiple comorbid conditions. (2) Methods: We investigated the effect of additional booster vaccinations following the primary vaccination series of mRNA, inactivated whole virus, or vector vaccines on infections with the SARS-CoV-2 delta variant in the total Hungarian elderly population. The infection, hospital admission, and 28-day all-cause mortality of elderly population was assessed. (3) Results: A total of 1,984,176 people fulfilled the criteria of elderly including 299,216 unvaccinated individuals, while 1,037,069 had completed primary vaccination and 587,150 had obtained an additional booster. The primary vaccination series reduced the risk of infection by 48.88%, the risk of hospital admission by 71.55%, and mortality by 79.87%. The booster vaccination had an additional benefit, as the risk of infection, hospital admission, and all-cause mortality were even lower (82.95%; 92.71%; and 94.24%, respectively). Vaccinated patients needing hospitalization suffered significantly more comorbid conditions, indicating a more vulnerable population. (4) Conclusions: Our data confirmed that the primary vaccination series and especially the booster vaccination significantly reduced the risk of the SARS-CoV-2 delta-variant-associated hospital admission and 28-day all-cause mortality in the elderly despite significantly more severe comorbid conditions.

Nationwide effectiveness of five SARS-CoV-2 vaccines in Hungary-the HUN-VE study

Objectives

The Hungarian vaccination campaign was conducted with five different vaccines during the third wave of the coronavirus disease 2019 (COVID-19) pandemic in 2021. This observational study (HUN-VE: Hungarian Vaccine Effectiveness) estimated vaccine effectiveness against severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and COVID-19-related mortality in 3.7 million vaccinated individuals.

Methods

Incidence rates of SARS-CoV-2 infection and COVID-19-related mortality were calculated using data from the National Public Health Centre surveillance database. Estimated vaccine effectiveness was calculated as 1 – incidence rate ratio ≥7 days after the second dose for each available vaccine versus an unvaccinated control group using mixed-effect negative binomial regression controlling for age, sex and calendar day.

Results

Between 22 January 2021 and 10 June 2021, 3 740 066 Hungarian individuals received two doses of the BNT162b2 (Pfizer-BioNTech), HB02 (Sinopharm), Gam-COVID-Vac (Sputnik-V), AZD1222 (AstraZeneca), or mRNA-1273 (Moderna) vaccines. Incidence rates of SARS-CoV-2 infection and COVID-19-related death were 1.73–9.3/100 000 person-days and 0.04–0.65/100 000 person-days in the fully vaccinated population, respectively. Estimated adjusted effectiveness varied between 68.7% (95% CI 67.2%–70.1%) and 88.7% (95% CI 86.6%–90.4%) against SARS-CoV-2 infection, and between 87.8% (95% CI 86.1%–89.4%) and 97.5% (95% CI 95.6%–98.6%) against COVID-19-related death, with 100% effectiveness in individuals aged 16–44 years for all vaccines.

Conclusions

Our observational study demonstrated the high or very high effectiveness of five different vaccines in the prevention SARS-CoV-2 infection and COVID-19-related death.

A nationwide study of the epidemiology of relapsing polychondritis

OBJECTIVE

Relapsing polychondritis (RP) is a rare autoimmune inflammatory disease that attacks mainly cartilaginous structures or causes serious damage in proteoglycan-rich structures (the eyes, heart, blood vessels, inner ear). This study shows results regarding the epidemiology, progression, and associations of this highly variable disease by collecting all cases from a 124-million-person-year Central European nationwide cohort.

METHODS

We used the Hungarian Health Care Database to identify all persons with possible RP infection. We followed patients who had International Classification of Diseases 10th edition code M94.1 at least once in their inpatient or outpatient records between January 1, 2002 and December 31, 2013 in Hungary. We classified these patients into disease severity groups by their drug consumption patterns between January 1, 2010 and December 31, 2013. We analyzed the regional distribution of RP incidences as well. Overall maps of comorbidity are presented with network layouts.

RESULTS

We identified 256 patients with RP among cumulatively 11.5 million registered inhabitants. We classified these patients into four severity classes as "extremely mild" (n=144), "mild" (n=22), "moderate" (n=41), and "severe" (n=4). Two additional groups were defined for patients without available drug data as "suspected only" (n=23) and "confirmed but unknown treatment" (n=22). The age and sex distributions of patients were similar to worldwide statistics. Indeed, the overall survival was good (95% confidence interval for 5 years was 83.6%-92.9% and for 10 years was 75.0%-88.3% which corresponds to the overall survival of the general population in Hungary), and the associations with other autoimmune disorders were high (56%) in Hungary. Almost any disease can occur with RP; however, the symptoms of chromosomal abnormalities are only incidental. Spondylosis can be a sign of the activation of RP, while Sjögren syndrome is the most frequent autoimmune association. Regional distribution of incidences suggests arsenic drinking water and sunlight exposure as possible triggering factors.

CONCLUSION

The good survival rate of RP in Hungary is probably associated with the early diagnosis of the disease.

Ontological analysis of SNOMED CT

Background

SNOMED CT is the most comprehensive medical terminology. However, its use for intelligent services based on formal reasoning is questionable.

Methods

The analysis of the structure of SNOMED CT is based on the formal top-level ontology DOLCE.

Results

The analysis revealed several ontological and knowledge-engineering errors, the most important are errors in the hierarchy (mostly from an ontological point of view, but also regarding medical aspects) and the mixing of subsumption relations with other types (mostly 'part of').

Conclusion

The found errors impede formal reasoning. The paper presents a possible way to correct these problems.

Dichotomy - a forgotten ancient principle

Dichotomy is an ancient principle of categorisation, where a class is divided into two jointly exhaustive and mutually disjoint categories. The principle as a general requirement was abandoned during the middle. The recent inquiry shows that studying this principle is still worthwhile and in some cases it can be used as a quality assessment tool. The paper presents algorithms that can transform any kind of categorial structures into dichotomy. The resulting representation sometimes can make apparent the problematic parts of the source. Problems often result from stating Is_a relations without differentiating criteria. A simple experiment of dichotomous transformation of the high level categories of the first chapter of ICD was carried out. The problem of "other" and "not elsewhere classified" categories is discussed.

CONCLUSION

we should not strive to build dichotomous structures but sometimes a dichotomous transformation of an existing structure can be helpful to detect critical parts of a system of categories.

Design principles of DOLCE-based formal representation of ICD10

The authors present the framework of formal representation of ICD10 based on DOLCE. The goal of the work is to represent the meaning of the categories of the classification systems based on a formal top-level ontology. The ICD categories are described in the space of atomic disease concepts, while the diseases themselves are defined on a pathological basis. The anatomical entities are taken from FMA, functions, morphological abnormalities and procedures from SNOMED International, while organisms are taken form the biological taxonomy. The ontology is represented in OWL 1.0 in a modularised way. The transformation of the previous GALEN-based representation to the same conceptual system is also under way.

GALEN based formal representation of ICD10

OBJECTIVES

The main objective is to create a knowledge-intensive coding support tool for the International Classification of Diseases (ICD10), which is based on formal representation of ICD10 categories. Beyond this task the resulting ontology could be reused in various ways. Decidability is an important issue for computer-assisted coding; consequently the ontology should be represented in description logic.

METHODS

The meaning of the ICD10 categories is represented using the GALEN Core Reference Model. Due to the deficiencies of its representation language (GRAIL) the ontology is transformed to the quasi-standard OWL. A test system which extracts disease concepts and classifies them to ICD10 categories has been implemented in Prolog to verify the feasibility of the approach.

RESULTS

The formal representation of the first two chapters of ICD10 (infectious diseases and neoplasms) has been almost completed. The constructed ontology has been converted to OWL DL. The test system successfully identified diseases in medical records from gastrointestinal oncology (84% recall, however precision is only 45%). The classifier module is still under development. Due to the experiences gained during the modelling, in the future work FMA is going to be used as anatomical reference ontology.

Non Aristotelian categories in medicine

This paper discusses the representation of medical categories that can not be defined in Aristotelian sense. Two kinds of these categories are mentioned: the prototype and the family resemblance categories. Such categories obviously do exist in medical domain. Search on the Net was performed for free text definition for some commonly used medical categories, like 'autism', 'Burkitt lymphoma' and 'disease'. Most of the found often contradicting definitions do not follow the Aristotelian rules of definition. Many definitions describe statistical properties of the category that are often useless in individual cases. A simple way is suggested that makes possible to represent such categories in biomedical ontologies and treat them separate from better formed categories. This makes possible to revise these categories at any later stage of ontology development.

Restructuring the foundational model of anatomy

The authors present a method to convert the FMA to a description logic-based representation in OWL. The concepts denoting anatomical structures are aligned to the DOLCE formal top-level ontology, and converted to a compact core ontology in the spirit of GALEN. The paper presents the identified problems in the FMA and the main aspects of the re-modelling.

A pilot ontological model of public health indicators

There are various public health databases in the world aiming to provide data to compare health conditions in different countries. Their data sets are more or less overlapping but data from different databases and different countries are hard to compare due to different definitions and interpretations. Our aim was to create a core ontological model that is able to represent public health indicators. We assumed, that by such representation comparability and quality of data could be improved.

METHOD

Three sets of indicators were taken, and a core ontology was built from information objects describing their top level entities. The Protégé ontology editor with RDF backend was used for building the ontology. The used indicator sets were the indicators of the Health for All Database of the World Health Organisation (HFA), the OECD Health Data, and the set of indicators proposed by the European Community Health Indicators (ECHI) European project. Then 19 indicators selected from HFA was represented using the core ontology. Strength and weaknesses of the descriptive capability of the model was studied.

RESULT

The drafted core model seems to be useful in representing many of the public health indicators. In some cases it really helps improve comparability. However, some of the semantic details cannot be sufficiently expressed by the used ontology representation language. There is a need of merging other domain ontologies to represent indicators related to other domains, such as economy, social and environmental sciences.

Ontological definition of population for public health databases

This paper discusses available definitions of population and criticise them against some basic rules of ontology. None of the found definition satisfies the requirement of an ontology that supports building consistent public heath databases. Most definitions define population on territorial bases or as reproductive communities. The author argues that populations are systems (its members must be in connection to each other) and individuals forming the population must share a common resource. Proper usage of the definitions may help to build consistent ontology for public health indicators and consistent databases.

GALEN Based Formal Representation of ICD10

The authors present a formal representation of ICD10 based on GALEN CRM. The goal of the work is to create a coding support tool for coding clinical diagnoses to ICD10. The formal representation of the first two chapters of ICD10 has been almost completed. The paper presents the main aspects of the modelling, and the experienced problems. The constructed ontology has been converted to OWL, and a test system has been implemented in Prolog to verify the feasibility of the approach. The system successfully identified diseases in medical records from gastrointestinal oncology. The classifier module is still under development.

Conceptual framework of health indicators: the IDA model

This paper presents a flexible conceptual framework for pub-lic health indicator databases. The model is a multidimensional-hierarchical representation of statistical data describing health and health influencing factors. The main characteristics of the IDA model are the strong discrimination of concepts (categories of entities enumerated in statistical systems) and dimensions (aspects that divide categories). Top level concepts of known data sources (WHO HFA database, OECD Health Data and ECHI) have been compared and a generalized structure had been created which can represent easily and consistently all the top level concepts of the known data sources. The model has been implemented in a prototype system, which demonstrates the feasibility of the approach.

Using n-gram method in the decomposition of compound medical diagnoses

OBJECTIVE

Our goal in this study was to find an easy to implement method to detect compound medical diagnosis in Hungarian medical language and decompose them into expressions referring to a single disease.

METHODS

A corpus of clinical diagnoses extracted form discharge reports (3,079 expressions, each of them referring to only one disease) was represented in an n-gram tree (a series of n consecutive word). A matching algorithm was implemented in a software, which is able to identify sensible n-grams existing both in test expressions and in the n-gram tree. A test sample of another 92 diagnoses was decomposed by two independent humans and by the software. The decompositions were compared with measure the recall and the precision of the method.

RESULTS

There was not full agreement between the decompositions of the humans, (which underlines the relevance of the problem). A consensus was arrived in all disagreed point by a third opinion and open discussion. The resulting decomposition was used as a gold standard and compared with the decomposition produced by the computer. The recall was 82.6% the precision 37.2%. After correction of spelling errors in the test sample the recall increased to 88.6% while the precision slightly decreased to 36.7%.

CONCLUSION

The proposed method seems to be useful in decomposition of compound diagnostic expressions and can improve quality of diagnostic coding of clinical cases. Other statistical methods (like vector space methods or neural networks) usually offer a ranked list of candidate codes either for single or compound expressions, and do not warn the user how many codes should be chosen. We propose our method especially in a situation where formal NLP techniques are not available, as it is the case with scarcely spoken languages like Hungarian.

About the language of Hungarian discharge reports

Concerning use of terminological systems translated to any particular language or computer assisted translation of patient records the daily medical language has to be taken into account rather than the pure academic language. The aim of this study to investigate the language of the diagnosis of the Hungarian hospital discharge reports. Word of collection of discharge diagnoses were categorised according to the language and form (normal words, abbreviations, acronyms). It was found that Latin is still a dominant, but the language is rather a mixture, where local jargon, non standardised abbreviations occur frequently. Some signs show, that discharge reports are not used primarily as a tool of communication of physicians, rather serve local administrative purposes.

Indexing of medical diagnoses by word affinity method

Automated coding of medical diagnoses is still an unsolved problem. Our goal in recent work was to find efficient, cheap and easy to implement method to assist the work of human encoders in hospitals. The proposed method is based on a vector-space model especially adapted to deal with short expressions, like clinical diagnoses. Using a set of coded diagnoses the co-occurrence of codes and words is more or less characteristic. The method describes these characteristics mathematically, by introduction of the so-called word adhesion. Two human encoders were asked to code the same set of 92 clinical diagnoses. Their results were compared to the ranked list of codes, produced by the computer. The results were better where the two human encoders agreed, and the overall results demonstrate the feasibility of the approach.

Using n-gram method in the decomposition of compound medical diagnoses

Compound diagnoses are often assigned to just one disease code. This is a known cause of coding error. Our paper outlines an efficient, cheap and easy to implement method for semi-atutomatic decomposition of such diagnostic expressions. The proposed method is based on n-grams. To verify the method two human encoders were asked to analyse the same set of 92 clinical diagnoses. Their results were compared to the analysis produced by the method. The results demonstrate the reasonability of the approach.

Typing of diseases in the Hungarian minimal basic data set for hospital treatment episodes

For reimbursement and epidemiological statistics the coded diagnoses should be classified in types like principal diagnosis, complication, co-morbidity etc. Computer assisted semi-automatic coding systems usually does not pay attention to this problem. After the description of pertaining national regulation, we present a logical framework of an algorithm, which minimise the clerical work for physicians and presumably will satisfy the need of epidemiology and reimbursement. The algorithm makes difference between diagnosis and disease, various diagnostic statement types, and uses causal chains among conditions expressed by diagnoses.

Maintenance of self-consistency of coding tables by statistical analysis of word co-occurrences

The author presents a method for maintaining the internal consistency of coding tables. The method was tested on a table used for assisting the daily work of indexing clinical cases to International Classification of Diseases. 300 item were tested selected randomly form a corpus of 3082 clinical diagnoses. The method discovered potential consistency problems in 39 cases, out of which 10 were false positive.

Questions on validity of International Classification of Diseases-coded diagnoses

International Classification of Diseases (ICD) codes are used for indexing medical diagnoses for various purposes and in various contexts. According to the literature and our personal experience, the validity of the coded information is unsatisfactory in general, however the 'correctness' is purpose and environment dependent. For detecting potential error sources, this paper gives a general framework of the coding process. The key elements of this framework are: (1) the formulation of the established diagnoses in medical language; (2) the induction from diagnoses to diseases; (3) indexing the diseases to ICD categories; (4) labelling of the coded entries (e.g. principal disease, complications, etc.). Each step is a potential source of errors. The most typical types of error are: (1) overlooking of diagnoses; (2) incorrect or skipped induction; (3) indexing errors; (4) violation of ICD rules and external regulations. The main reasons of the errors are the physician's errors in the primary documentation, the insufficient knowledge of the encoders (different steps of the coding process require different kind of knowledge), the internal inconsistency of the ICD, and some psychological factors. Computer systems can facilitate the coding process, but attention has to be paid to the entire coding process, not only to the indexing phase.

TSMI: a CEN/TC251 standard for time specific problems in healthcare informatics and telematics

Time is the most important variable in healthcare, and standards are needed about how to represent information with explicit references to time. In this paper, the European Prestandard 'TSMI: time standards for healthcare specific problems' (CEN/TC251 preENV 12381) is presented which aims to be the first contribution to this harmonisation process, focusing on 'representation' and 'explicit reference' of temporal information in healthcare. The prestandard is mainly composed of two parts. First, the basic building blocks for modelling time-related information are introduced, and a formal representation scheme proposed. In a second part, conformance rules and principles for Healthcare Data and Information as well as for Healthcare Information Systems, are covered.

Quality of healthcare related software applications--setting up an accreditation system in Hungary

Meeting expectations of high quality health care, the safe and secure operation of medical information systems is a "must". However for healthcare software nationwide quality control systems are not widely used. A quality control project of health care applications in Hungary has been launched in 1996 by the Hungarian Society of Healthcare Informatics (MEIT) and Medico-Biological Section of Johann Neumann Society of Computing (NJSZT) by establishing a joint Healthcare Informatics Applications Accreditation Board (Board ESAB). The Board developed an evaluation methodology and a legal procedure to test health care software application modules. The evaluation method is based on international standards as ISO-9126 and on emerging European standards of CEN/TC 251. First rounds of accreditation already proved that there is a need among providers and users for the accreditation process. The authors hope that establishing an accreditation system will lead to a more balanced health care software market where users have an opportunity to inform themselves by the opinion of independent experts on the product they intend to purchase.

On different roles of natural language information in medicine

In this paper the authors analyze the main different function types of language in medical environment in different communicative situations and descriptive tasks. These functions are categorized as knowledge transfer, documentation, directive function, expression of emotions. The computer representation of the information have to be different according to the different tasks. The paper highlights the most important differences and concludes that further research is necessary in the details.

Theoretical considerations on medical concept representation

Concepts are seen as building elements of more compositional information objects in medicine. Consequently, representation of medical concepts plays a critical role in any information system. The authors present some of the key problems which need solution, such as definition of medical domain, analysis of internal structure, different subsets of medical concepts, and the problem of the 'elementary' concepts. As a result of these considerations we can conclude that the domain of medical concepts can hardly be delimited and consists of different subsets. These subsets need different representation methods according to their different nature. Some of these subsets have a hierarchic structure. In those one can find sometimes multiple hierarchies. We suggest avoiding multiple hierarchies in concept systems by the introduction of new dimensions.

[Method for the determination of the functional degree of esophageal stricture and the effectiveness of dilatation]

Based on their experiences of 494 oesophagus dilatation in 98 patients, authors developed a stadium system for defining the severity of dysphagia caused by oesophagus stenosis of stricture. This system is usable for follow-up the course of the disease, for the establishing the necessity of the dilatation, and for the comparison of results achieved by different dilatation methods, as well. The method is based on simple, well-defined clinical parameters, special skills or instrumentations not required, and fits the clinical experiences of many years of the authors. For this reasons it is useful both general practitioners and for the specialists performing the esophagus dilatation, as well.

Towards a quantitative approach of medical information. Part 2. Comparative measurement of medical information objects

One of the key problems in medical knowledge representation is that we usually have no idea about the largeness of the knowledge to be represented. Underestimation of this largeness may occur frequently. Considering this situation, the authors have applied a method of measuring or estimating the largeness of a certain medical knowledge area modelled in a theoretical information space. The method is tested on two sets of terms, the OMED terminology of digestive endoscopy and on the second and third version of the SNOMED nomenclature. The assessment of largeness of a knowledge area does not seem possible by a single measure. The 'volume', 'density' and 'complexity' must be addressed separately. In the present study different medical knowledge-representation systems are compared according to their volume. Possible ways to estimate their complexity and density are mentioned.

Towards a quantitative approach of medical information. Part 1. Measures of a multidimensional medical information space

Despite the importance of quantitative analysis of medical information, it is a rare subject in informatics literature. To stimulate more interest the authors propose a multidimensional medical information space, in which measures can be defined to compare different medical information objects or knowledge areas. To describe the measures, classical information theory, the general database theory of Sundgren and the Blois model of medical thinking are used. A space model of (at least) three dimensions is offered, where information objects might have a normalized length, a total depth (measuring embedded knowledge levels) and a complexity width.