Exploring in-hospital adverse drug events using ICD-10 codes

 Abstract
Objective Adverse drug events (ADEs) during hospital admissions are a widespread problem associated with adverse patient outcomes. The ‘external cause’ codes in the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) provide opportunities for identifying the incidence of ADEs acquired during hospital stays that may assist in targeting interventions to decrease their occurrence. The aim of the present study was to use routine administrative data to identify ADEs acquired during hospital admissions in a suburban healthcare network in Melbourne, Australia.

Methods Thirty-nine secondary diagnosis fields of hospital discharge data for a 1-year period were reviewed for ‘diagnoses not present on admission’ and assigned to the Classification of Hospital Acquired Diagnoses (CHADx) subclasses. Discharges with one or more ADE subclass were extracted for retrospective analysis.

Results From 57 205 hospital discharges, 7891 discharges (13.8%) had at least one CHADx, and 402 discharges (0.7%) had an ADE recorded. The highest proportion of ADEs was due to administration of analgesics (27%) and systemic antibiotics (23%). Other major contributors were anticoagulation (13%), anaesthesia (9%) and medications with cardiovascular side-effects (9%).

Conclusion Hospital data coded in ICD-10 can be used to identify ADEs that occur during hospital stays and also clinical conditions, therapeutic drug classes and treating units where these occur. Using the CHADx algorithm on administrative datasets provides a consistent and economical method for such ADE monitoring.

What is known about the topic? Adverse drug events (ADEs) can result in several different physical consequences, ranging from allergic reactions to death, thereby posing a significant burden on patients and the health system. Numerous studies have compared manual, written incident reporting systems used by hospital staff with computerised automated systems to identify ADEs acquired during hospital admissions. Despite various approaches aimed at improving the detection of ADEs, they remain under-reported, as a result of which interventions to mitigate the effect of ADEs cannot be initiated effectively.

What does this paper add? This research article demonstrates major methodological advances over comparable published studies looking at the effectiveness of using routine administrative data to monitor rates of ADEs that occur during a hospital stay and reviews the type of ADEs and their frequency patterns during patient admission. It also provides an insight into the effect of ADEs that occur within different hospital treating units. The method implemented in this study is unique because it uses a grouping algorithm developed for the Australian Commission on Safety and Quality in Health Care (ACSQHC) to identify ADEs not present on admission from patient data coded in ICD-10. This algorithm links the coded external causes of ADEs with their consequences or manifestations. ADEs identified through the use of programmed code based on this algorithm have not been studied in the past and therefore this paper adds to previous knowledge in this subject area.

What are the implications for health professionals? Although not all ADEs can be prevented with current medical knowledge, this study can assist health professionals in targeting interventions that can efficiently reduce the rate of ADEs that occur during a hospital stay, and improve information available for future medication management decisions.

Predicting risk of suicide attempt using history of physical illnesses from electronic medical records

BACKGROUND: Although physical illnesses, routinely documented in electronic medical records (EMR), have been found to be a contributing factor to suicides, no automated systems use this information to predict suicide risk.

OBJECTIVE: The aim of this study is to quantify the impact of physical illnesses on suicide risk, and develop a predictive model that captures this relationship using EMR data.

METHODS: We used history of physical illnesses (except chapter V: Mental and behavioral disorders) from EMR data over different time-periods to build a lookup table that contains the probability of suicide risk for each chapter of the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) codes. The lookup table was then used to predict the probability of suicide risk for any new assessment. Based on the different lengths of history of physical illnesses, we developed six different models to predict suicide risk. We tested the performance of developed models to predict 90-day risk using historical data over differing time-periods ranging from 3 to 48 months. A total of 16,858 assessments from 7399 mental health patients with at least one risk assessment was used for the validation of the developed model. The performance was measured using area under the receiver operating characteristic curve (AUC).

RESULTS: The best predictive results were derived (AUC=0.71) using combined data across all time-periods, which significantly outperformed the clinical baseline derived from routine risk assessment (AUC=0.56). The proposed approach thus shows potential to be incorporated in the broader risk assessment processes used by clinicians.

CONCLUSIONS: This study provides a novel approach to exploit the history of physical illnesses extracted from EMR (ICD-10 codes without chapter V-mental and behavioral disorders) to predict suicide risk, and this model outperforms existing clinical assessments of suicide risk.

Stable feature selection for clinical prediction: Exploiting ICD tree structure using Tree-Lasso

Modern healthcare is getting reshaped by growing Electronic Medical Records (EMR). Recently, these records have been shown of great value towards building clinical prediction models. In EMR data, patients' diseases and hospital interventions are captured through a set of diagnoses and procedures codes. These codes are usually represented in a tree form (e.g. ICD-10 tree) and the codes within a tree branch may be highly correlated. These codes can be used as features to build a prediction model and an appropriate feature selection can inform a clinician about important risk factors for a disease. Traditional feature selection methods (e.g. Information Gain, T-test, etc.) consider each variable independently and usually end up having a long feature list. Recently, Lasso and related l1-penalty based feature selection methods have become popular due to their joint feature selection property. However, Lasso is known to have problems of selecting one feature of many correlated features randomly. This hinders the clinicians to arrive at a stable feature set, which is crucial for clinical decision making process. In this paper, we solve this problem by using a recently proposed Tree-Lasso model. Since, the stability behavior of Tree-Lasso is not well understood, we study the stability behavior of Tree-Lasso and compare it with other feature selection methods. Using a synthetic and two real-world datasets (Cancer and Acute Myocardial Infarction), we show that Tree-Lasso based feature selection is significantly more stable than Lasso and comparable to other methods e.g. Information Gain, ReliefF and T-test. We further show that, using different types of classifiers such as logistic regression, naive Bayes, support vector machines, decision trees and Random Forest, the classification performance of Tree-Lasso is comparable to Lasso and better than other methods. Our result has implications in identifying stable risk factors for many healthcare problems and therefore can potentially assist clinical decision making for accurate medical prognosis.

Association between abnormal psychosocial situations in childhood, generalised anxiety disorder and oppositional defiant disorder

Objective: Psychosocial stressors are important in the pathogenesis of most mental disorders. However, little is known about the way psychosocial stressors uniquely combine to create risk for different expressions of child and adolescent psychopathology. The purpose of this study was to determine whether core dimensions of stressful psychosocial situations are differentially associated with childhood generalized anxiety disorder and oppositional defi ant disorder.

Method: A case-control design conducted in Trondheim (Norway) from 2002 to 2004 comparing exposure to ICD-10-defi ned abnormal psychosocial situations (Z-codes) among 21 children with oppositional defi ant disorder (ODD) and 22 children with generalized anxiety disorder (GAD) recruited from a university outpatient clinic with 42 non-patient school controls.

Results: Multigroup discriminant analysis extracted two signifi cant dimensions within the psychosocial variables assessed. Function 1 was characterized by overprotection, parental pressures and acute life events and was associated with GAD. Function 2 was characterized by parental abuse/hostility and interpersonal stress and was associated with ODD. Both dimensions were able to correctly classify 89.7% of the cases, compared to 35.9% by chance.

Conclusions: The results indicate that specifi c psychosocial dimensions are differentially related to childhood GAD and ODD. This may be useful in targeting at-risk populations for preventive intervention as well as informing more accurate alignment of psychosocial resources for treatment.

Identification of incident fractures : the Geelong Osteoporosis Study

Background: Complete fracture ascertainment is critical for fracture cost estimates and planning for future health care facilities. Virtually complete ascertainment is possible for hip fractures because they nearly always require hospitalisation.

Aims: To validate the use of radiological reports as a resource for ascertaining fracture cases, using hip fracture as a model.

Methods: Hip fracture rates obtained from radiological reports were compared with rates obtained from hospital discharge summaries of medical records using International Classification of Diseases-9 (ICD-9) codes 820.0–820.9 and 733.1 over a three-year period.

Results: Hip fracture cases numbered 589 using radiological reports and 585 using medical records. Discharge summaries failed to identify 15 cases ascertained through radiology reports whereas 11 cases ascertained through medical records were not identified from X-ray reports. The age-specific incidence rates for radiological ascertainment were within the 95% confidence limits of the rates derived from medical records.

Conclusions: Among a population of patients generally admitted to hospital for treatment of their fracture, we were able to identify more cases from radiological reports than from medical records. Incidence rates for hip fracture were comparable using the two methods. Radiological reports provide a valuable resource for identifying incident fractures. This method of case ascertainment would be suitable for identifying both major and minor fractures in regions with self-contained health services where access to all radiological reports is possible.

The effect of community interventions on alcohol-related assault in Geelong, Australia

Alcohol has consistently been demonstrated to increase levels of aggression and violence, particularly in late night licensed venues. Since 2005, the City of Geelong in Australia has implemented a substantial number of interventions to reduce alcohol related violence, including a liquor accord, increased police surveillance, ID scanners, CCTV, a radio network and an alcohol industry sponsored social marketing campaign. The aim of the current study is to assess the individual and collective impact of community interventions on indicators of alcohol-related assaults in the Geelong region. This paper reports stage one findings from the Dealing with Alcohol-related problems in the Night-time Economy project (DANTE) and specifically examines assault rate data from both emergency department presentations, ICD-10 classification codes, and police records of assaults. None of the interventions were associated with reductions in alcohol-related as-sault or intoxication in Geelong, either individually or when combined. However, the alcohol industry sponsored social marketing campaign ‘Just Think’ was associated with an increase in assault rates. Community level interventions appeared to have had little effect on assault rates during high alcohol times. It is also possible that social marketing campaigns without practical strategies are associated with increased assault rates. The findings also raise questions about whether interventions should be targeted at reducing whole-of-community alcohol consumption.

Hierarchical Bayesian nonparametric models for knowledge discovery from electronic medical records

Electronic Medical Record (EMR) has established itself as a valuable resource for large scale analysis of health data. A hospital EMR dataset typically consists of medical records of hospitalized patients. A medical record contains diagnostic information (diagnosis codes), procedures performed (procedure codes) and admission details. Traditional topic models, such as latent Dirichlet allocation (LDA) and hierarchical Dirichlet process (HDP), can be employed to discover disease topics from EMR data by treating patients as documents and diagnosis codes as words. This topic modeling helps to understand the constitution of patient diseases and offers a tool for better planning of treatment. In this paper, we propose a novel and flexible hierarchical Bayesian nonparametric model, the word distance dependent Chinese restaurant franchise (wddCRF), which incorporates word-to-word distances to discover semantically-coherent disease topics. We are motivated by the fact that diagnosis codes are connected in the form of ICD-10 tree structure which presents semantic relationships between codes. We exploit a decay function to incorporate distances between words at the bottom level of wddCRF. Efficient inference is derived for the wddCRF by using MCMC technique. Furthermore, since procedure codes are often correlated with diagnosis codes, we develop the correspondence wddCRF (Corr-wddCRF) to explore conditional relationships of procedure codes for a given disease pattern. Efficient collapsed Gibbs sampling is derived for the Corr-wddCRF. We evaluate the proposed models on two real-world medical datasets - PolyVascular disease and Acute Myocardial Infarction disease. We demonstrate that the Corr-wddCRF model discovers more coherent topics than the Corr-HDP. We also use disease topic proportions as new features and show that using features from the Corr-wddCRF outperforms the baselines on 14-days readmission prediction. Beside these, the prediction for procedure codes based on the Corr-wddCRF also shows considerable accuracy.

Religios/spirituelles befinden bei psychisch kranken: Angstlich/depressive und suchtkranke im vergleich zu gesunden kontrollpersonen

Objective
The aim of this study is to investigate different dimensions of religiosity and spirituality among psychiatric in-patients. The study examines differences between addictive (ICD 10: F1x) and anxious/depressive (ICD 10: F3x/F4x) patients and considers the main implications for treatment.

Method
Differences in dimensions of religious/spiritual well-being (RSWB) between addictive (n = 389) and anxious/depressive patients (n = 200) are investigated, also by comparison to a control group (n = 1,500). Furthermore dimensions of RSWB are related to personality factors and different psychiatric parameters within the psychiatric groups.

Results
The psychiatric groups show a lower amount of overall RSWB (p < 0.001) than the healthy controls. Furthermore, dimensions of RSWB turned out to be negatively correlated with several psychiatric symptoms.

Conclusions
Based on these results we emphasize religious/spiritual issues within psychiatric treatment. Moreover, there may be a strong potential of the RSWB dimensions such as “Hope” or “Forgiveness” as positive therapeutic factors in psychiatric treatment.

Life satisfaction and morbidity among postmenopausal women

OBJECTIVE: To investigate associations between morbidity and global life satisfaction in postmenopausal women taking into account type and number of diseases.

MATERIALS AND METHODS: A total of 11,084 women (age range 57-66 years) from a population-based cohort of Finnish women (OSTPRE Study) responded to a postal enquiry in 1999. Life satisfaction was measured with a 4-item scale. Self-reported diseases diagnosed by a physician and categorized according to ICD-10 main classes were used as a measure of morbidity. Enquiry data on health and lifestyle were used as covariates in the multivariate logistic models.

RESULTS: Morbidity was strongly associated with life dissatisfaction. Every additional disease increased the risk of life dissatisfaction by 21.1% (p < .001). The risk of dissatisfaction was strongest among women with mental disorders (OR = 5.26; 95%CI 3.84-7.20) and neurological disorders (OR = 3.62; 95%CI 2.60-5.02) compared to the healthy (each p < .001). Smoking, physical inactivity and marital status were also associated with life dissatisfaction (each p < .001) but their introduction to the multivariate model did not attenuate the pattern of associations.

CONCLUSIONS: Morbidity and life dissatisfaction have a disease-specific and dose-dependent relationship. Even if women with mental and neurological disorders have the highest risk for life dissatisfaction, monitoring life satisfaction among aging women regardless of disorders should be undertaken in order to intervene the joint adverse effects of poor health and poor well-being.

On the validity of measuring change over time in routine clinical assessment: a close examination of item-level response shifts in psychosomatic inpatients

Objective: Significant life events such as severe health status changes or intensive medical treatment often trigger response shifts in individuals that may hamper the comparison of measurements over time. Drawing from the Oort model, this study aims at detecting response shift at the item level in psychosomatic inpatients and evaluating its impact on the validity of comparing repeated measurements. Study design and setting: Complete pretest and posttest data were available from 1188 patients who had filled out the ICD-10 Symptom Rating (ISR) scale at admission and discharge, on average 24 days after intake. Reconceptualization, reprioritization, and recalibration response shifts were explored applying tests of measurement invariance. In the item-level approach, all model parameters were constrained to be equal between pretest and posttest. If non-invariance was detected, these were linked to the different types of response shift. Results: When constraining across-occasion model parameters, model fit worsened as indicated by a significant Satorra–Bentler Chi-square difference test suggesting potential presence of response shifts. A close examination revealed presence of two types of response shift, i.e., (non)uniform recalibration and both higher- and lower-level reconceptualization response shifts leading to four model adjustments. Conclusions: Our analyses suggest that psychosomatic inpatients experienced some response shifts during their hospital stay. According to the hierarchy of measurement invariance, however, only one of the detected non-invariances is critical for unbiased mean comparisons over time, which did not have a substantial impact on estimating change. Hence, the use of the ISR can be recommended for outcomes assessment in clinical routine, as change score estimates do not seem hampered by response shift effects.

Accuracy of national mortality codes in identifying adjudicated cardiovascular deaths

Objective: This study investigated the sensitivity and specificity of the national mortality codes in identifying cardiovascular disease (CVD) deaths and documents methods of verification.

Methods: A 12-year retrospective case ascertainment of all ICD-coded CVD deaths was performed for deaths between 1990 and 2002 in the Melbourne Collaborative Cohort Study, comprising 41,528 subjects. Categories of non-CVD codes were also examined. Stratified samples of 750 deaths were adjudicated from a total of 2,230 deaths. Expert panels of cardiologists and neurologists adjudicated deaths.

Results: Of the 750 deaths adjudicated, 582 were verified as CVD [392 coronary heart disease (CHD) and 92 stroke] and 168 non-CVD. Estimated sensitivity and specificity of national mortality codes for identifying specific causes of death were: CHD 74.2% (95% CI: 69.8–78.5%) and 97.6% (96.0–99.2%), respectively; myocardial infarction 59.9% (50.9–69.0%) and 94.2% (92.4–96.0%), respectively; haemorrhagic stroke 58.9% (46.0–71.7%) and 99.8% (99.4–100.0%), respectively and; ischaemic stroke 38.7% (20.5–56.9%) and 99.9% (99.6–100.0%), respectively. Misclassification was most common for deaths with primary ICD codes for endocrine-metabolic and genito-urinary diseases.

Conclusions: National mortality coding under-estimated the true proportion of CHD and stroke deaths in the cohort by 13.6% and 50.8%, respectively.

Communicating the ethos of codes of ethics in corporate Australia, 1995-2001: whose rights, whose responsibilities?

This paper considers the commitment to business ethics of the top 500 companies operating in the Australian private sector and communicates the results of a longitudinal study conducted from 1995 to 2001. Primary data was obtained (in 1995 and again in 2001) via a self-administered mail  questionnaire distributed to a census of these top 500 Australian  companies. This commitment of each company to their code of ethics was indicated and measured via a range of methods used by organizations to communicate the ethos of their codes to employees. Just as they were in  1995, it would appear that companies in 2001 still are good at ensuring that their rights are protected, but at the same time they do not seem to take on the responsibility to ensure that employees'' rights are just as well protected. This double standard leads to cynicism towards the current business ethics processes inherent in Australian companies.

Codes of ethics: what are they really and what should they be?

Codes of ethics are prevalent in major corporations around the world. They are seen as the first tangible commitment to being ethical. This paper examines codes of ethics and tries to establish what they are, how they are developed and their net utility. We then proffer the idea of codes as the first part of a five-stage process that leads to an overall corporate commitment to being ethical in one's business dealings.

Combinatorial aspects of jump codes

We develop an approach to jump codes concentrating on their combinatorial and symmetry properties. The main result is a generalization of a theorem previously proved in the context of isodual codes. We show that several previously constructed jump codes are instances of this theorem.