Development and validation of a Medicines Comorbidity Index for older people

Interpretable clinical phenotypes among patients hospitalized with COVID-19 using cluster analysis

Background

Multiple clinical phenotypes have been proposed for coronavirus disease (COVID-19), but few have used multimodal data. Using clinical and imaging data, we aimed to identify distinct clinical phenotypes in patients admitted with COVID-19 and to assess their clinical outcomes. Our secondary objective was to demonstrate the clinical applicability of this method by developing an interpretable model for phenotype assignment.

Methods

We analyzed data from 547 patients hospitalized with COVID-19 at a Canadian academic hospital. We processed the data by applying a factor analysis of mixed data (FAMD) and compared four clustering algorithms: k-means, partitioning around medoids (PAM), and divisive and agglomerative hierarchical clustering. We used imaging data and 34 clinical variables collected within the first 24 h of admission to train our algorithm. We conducted a survival analysis to compare the clinical outcomes across phenotypes. With the data split into training and validation sets (75/25 ratio), we developed a decision-tree-based model to facilitate the interpretation and assignment of the observed phenotypes.

Results

Agglomerative hierarchical clustering was the most robust algorithm. We identified three clinical phenotypes: 79 patients (14%) in Cluster 1, 275 patients (50%) in Cluster 2, and 203 (37%) in Cluster 3. Cluster 2 and Cluster 3 were both characterized by a low-risk respiratory and inflammatory profile but differed in terms of demographics. Compared with Cluster 3, Cluster 2 comprised older patients with more comorbidities. Cluster 1 represented the group with the most severe clinical presentation, as inferred by the highest rate of hypoxemia and the highest radiological burden. Intensive care unit (ICU) admission and mechanical ventilation risks were the highest in Cluster 1. Using only two to four decision rules, the classification and regression tree (CART) phenotype assignment model achieved an AUC of 84% (81.5-86.5%, 95 CI) on the validation set.

Conclusions

We conducted a multidimensional phenotypic analysis of adult inpatients with COVID-19 and identified three distinct phenotypes associated with different clinical outcomes. We also demonstrated the clinical usability of this approach, as phenotypes can be accurately assigned using a simple decision tree. Further research is still needed to properly incorporate these phenotypes in the management of patients with COVID-19.

Predictivity of the comorbidity indices for geriatric syndromes

Background

The aging population and increasing chronic diseases make a tremendous burden on the health care system. The study evaluated the relationship between comorbidity indices and common geriatric syndromes.

Methods

A total of 366 patients who were hospitalized in a university geriatric inpatient service were included in the study. Sociodemographic characteristics, laboratory findings, and comprehensive geriatric assessment(CGA) parameters were recorded. Malnutrition, urinary incontinence, frailty, polypharmacy, falls, orthostatic hypotension, depression, and cognitive performance were evaluated. Comorbidities were ranked using the Charlson Comorbidity Index(CCI), Elixhauser Comorbidity Index(ECM), Geriatric Index of Comorbidity(GIC), and Medicine Comorbidity Index(MCI). Because, the CCI is a valid and reliable tool used in different clinical settings and diseases, patients with CCI score higher than four was accepted as multimorbid. Additionally, the relationship between geriatric syndromes and comorbidity indices was assessed with regression analysis.

Results

Patients’ mean age was 76.2 ± 7.25 years(67.8% female). The age and sex of multimorbid patients according to the CCI were not different compared to others. The multimorbid group had a higher rate of dementia and polypharmacy among geriatric syndromes. All four indices were associated with frailty and polypharmacy(p < 0.05). CCI and ECM scores were related to dementia, polypharmacy, and frailty. Moreover, CCI was also associated with separately slow walking speed and low muscle strength. On the other hand, unlike CCI, ECM was associated with malnutrition.

Conclusions

In the study comparing the four comorbidity indices, it is revealed that none of the indices is sufficient to use alone in geriatric practice. New indices should be developed considering the complexity of the geriatric cases and the limitations of the existing indices.

Contribution of an extensive medication-based comorbidity index (Rx-Risk) in explaining the excess mortality after hip fracture in older Norwegians: a NOREPOS cohort study

OBJECTIVES

Patients with hip fracture are typically characterised by extensive comorbidities and excess mortality. Methods that account for a wide range of comorbidities are needed when attempting to identify causal associations in registry-based studies. We aimed to study the association between the prescription-based Rx-Risk Comorbidity Index (abbreviated Rx-Risk) and mortality by history of hip fracture, and to quantify the contribution of Rx-Risk in explaining the excess mortality after hip fracture.

SETTING

In this prospective study, we used nationwide registry data from outpatient care. Rx-Risk was based on filled prescriptions recorded in the Norwegian Prescription Database. Medications were mapped to 46 comorbidity categories by Anatomical Therapeutic Chemical code. Information on hip fractures during 1994-2013 was available from the Norwegian Epidemiologic Osteoporosis Studies hip fracture database, and year of death was obtained from Statistics Norway. We estimated 1-year mortality risk (January through December 2014) according to Rx-Risk score based on dispensed prescriptions in 2013, history of hip fracture, age and sex using Poisson regression.

PARTICIPANTS

All individuals aged 65 years and older who were alive by the end of 2013 and had filled at least one prescription in an outpatient pharmacy in Norway in 2013 (n=735 968).

RESULTS

Mortality increased exponentially with increasing Rx-Risk scores, and it was highest in persons with a history of hip fracture across the major range of Rx-Risk scores. Age- and sex-adjusted mortality risk difference according to history of hip fracture (yes vs no) was 4.4 percentage points (7.8% vs 3.4%). Adjustment for Rx-Risk score further attenuated this risk difference to 3.3 percentage points.

CONCLUSIONS

History of hip fracture and comorbidity assessed by Rx-Risk are independent risk factors for mortality in the community-dwelling older population in Norway. Comorbidity explained a quarter of the excess mortality in persons with a history of hip fracture.

Examining variation in the measurement of multimorbidity in research: a systematic review of 566 studies

BACKGROUND

A systematic understanding of how multimorbidity has been constructed and measured is unavailable. This review aimed to examine the definition and measurement of multimorbidity in peer-reviewed studies internationally.

METHODS

We systematically reviewed studies on multimorbidity, via a search of nine bibliographic databases (Ovid [PsycINFO, Embase, Global Health, and MEDLINE], Web of Science, the Cochrane Library, CINAHL Plus, Scopus, and ProQuest Dissertations & Theses Global), from inception to Jan 21, 2020. Reference lists and tracked citations of retrieved articles were hand-searched. Eligible studies were full-text articles measuring multimorbidity for any purpose in community, primary care, care home, or hospital populations receiving a non-specialist service. Abstracts, qualitative research, and case series were excluded. Two reviewers independently reviewed the retrieved studies with conflicts resolved by discussion or a third reviewer, and a single researcher extracted data from published papers. To assess our objectives of how multimorbidity has been measured and examine variation in the chronic conditions included (in terms of number and type), we used descriptive analysis (frequencies, cross-tabulation, and negative binomial regression) to summarise the characteristics of multimorbidity studies and measures (study setting, source of morbidity data, study population, primary study purpose, and multimorbidity measure type). This systematic review is registered with PROSPERO, CRD420201724090.

FINDINGS

566 studies were included in our review, of which 206 (36·4%) did not report a reference definition for multimorbidity and 73 (12·9%) did not report the conditions their measure included. The number of conditions included in measures ranged from two to 285 (median 17 [IQR 11-23). 452 (79·9%) studies reported types of condition within a single multimorbidity measure; most included at least one cardiovascular condition (441 [97·6%] of 452 studies), metabolic and endocrine condition (440 [97·3%]), respiratory condition (422 [93·4%]), musculoskeletal condition (396 [87·6%]), or mental health condition (355 [78·5%]) in their measure of multimorbidity. Chronic infections (123 [27·2%]), haematological conditions (110 [24·3%]), ear, nose, and throat conditions (107 [23·7%]), skin conditions (70 [15·5%]), oral conditions (19 [4·2%]), and congenital conditions (14 [3·1%]) were uncommonly included. Only eight individual conditions were included by more than half of studies in the multimorbidity measure used (diabetes, stroke, cancer, chronic obstructive pulmonary disease, hypertension, coronary heart disease, chronic kidney disease, and heart failure), with individual mental health conditions under-represented. Of the 566 studies, 419 were rated to be of moderate risk of bias, 107 of high risk of bias, and 40 of low risk of bias according to the Effective Public Health Practice Project quality assessment tool.

INTERPRETATION

Measurement of multimorbidity is poorly reported and highly variable. Consistent reporting of measure definitions should be required by journals, and consensus studies are needed to define core and study-dependent conditions to include in measures of multimorbidity.

FUNDING

Health Data Research UK.

Comparison between drug therapy-based comorbidity indices and the Charlson Comorbidity Index for the detection of severe multimorbidity in older subjects

BACKGROUND

To know burden disease of a patient is a key point for clinical practice and research, especially in the elderly. Charlson's Comorbidity Index (CCI) is the most widely used rating system, but when diagnoses are not available therapy-based comorbidity indices (TBCI) are an alternative. However, their performance is debated. This study compares the relations between Drug Derived Complexity Index (DDCI), Medicines Comorbidity Index (MCI), Chronic Disease Score (CDS), and severe multimorbidity, according to the CCI classification, in the elderly.

METHODS

Logistic regression and Receiver Operating Characteristic (ROC) analysis were conducted on two samples from Italy: 2579 nursing home residents (Korian sample) and 7505 older adults admitted acutely to geriatric or internal medicine wards (REPOSI sample).

RESULTS

The proportion of subjects with severe comorbidity rose with TBCI score increment, but the Area Under the Curve (AUC) for the CDS (Korian: 0.70, REPOSI: 0.79) and MCI (Korian: 0.69, REPOSI: 0.81) were definitely better than the DDCI (Korian: 0.66, REPOSI: 0.74). All TBCIs showed low Positive Predictive Values (maximum: 0.066 in REPOSI and 0.317 in Korian) for the detection of severe multimorbidity.

CONCLUSION

CDS and MCI were better predictors of severe multimorbidity in older adults than DDCI, according to the CCI classification. A high CCI score was related to a high TBCI. However, the opposite is not necessarily true probably because of non-evidence-based prescriptions or physicians' prescribing attitudes. TBCIs did not appear selective for detecting of severe multimorbidity, though they could be used as a measure of disease burden, in the absence of other solutions.

A narrative review of using prescription drug databases for comorbidity adjustment: A less effective remedy or a prescription for improved model fit?

BACKGROUND

The use of claims data for identifying comorbid conditions in patients for research purposes has been widely explored. Traditional measures of comorbid adjustment included diagnostic data (e.g., ICD-9-CM or ICD-10-CM codes), with the Charlson and Elixhauser methodology being the two most common approaches. Prescription data has also been explored for use in comorbidity adjustment, however early methodologies were disappointing when compared to diagnostic measures.

OBJECTIVE

The objective of this methodological review is to compare results from newer studies using prescription-based data with more traditional diagnostic measures.

METHODS

A review of studies found on PubMed, Medline, Embase or CINAHL published between January 1990 and December 2020 using prescription data for comorbidity adjustment. A total of 50 studies using prescription drug measures for comorbidity adjustment were found.

CONCLUSIONS

Newer prescription-based measures show promise fitting models, as measured by predictive ability, for research, especially when the primary outcomes are utilization or drug expenditure rather than diagnostic measures. More traditional diagnostic-based measures still appear most appropriate if the primary outcome is mortality or inpatient readmissions.

Development of a specialty intensity score to estimate a patient's need for care coordination across physician specialties

BACKGROUNDS AND AIMS

This article develops a Specialty Intensity Score, which uses patient diagnosis codes to estimate the number of specialist physicians a patient will need to access. Conceptually, the score can serve as a proxy for a patient's need for care coordination across doctors. Such a measure may be valuable to researchers studying care coordination practices for complex patients. In contrast with previous comorbidity scores, which focus primarily on mortality and utilization, this comorbidity score approximates the complexity of a patient's the interaction with the health care system.

METHODS

We use 2015 inpatient claims data from the Centers for Medicare and Medicaid Services to model the relationship between a patient's diagnoses and physician specialty usage. We estimate usage of specialist doctors by using a least absolute shrinkage and selection operator Poisson model. The Specialty Intensity Score is then constructed using this predicted specialty usage. To validate our score, we test its power to predict the occurrence of patient safety incidents and compare that with the predictive power of the Charlson comorbidity index.

RESULTS

Our model uses 127 of the 279 International Classification of Disease, 10th Revision, Clinical Modification (ICD-10-CM) diagnosis subchapters to predict specialty usage, thus creating the Specialty Intensity Score. This score has significantly greater power in predicting patient safety complications than the widely used Charlson comorbidity index.

CONCLUSION

The Specialty Intensity Score developed in this article can be used by health services researchers and administrators to approximate a patient's need for care coordination across multiple specialist doctors. It, therefore, can help with evaluation of care coordination practices by allowing researchers to restrict their analysis of outcomes to the patients most impacted by those practices.

More than two-dozen prescription drug-based risk scores are available for risk adjustment: A systematic review

OBJECTIVE

While several prescription drug-based risk indices have been developed, their design, performance, and application has not previously been synthesized.

STUDY DESIGN AND SETTING

We searched Ovid MEDLINE, CINAHL and Embase from inception through March 3, 2020 and included studies that developed or updated a prescription drug-based risk index. Two reviewers independently performed screening and extracted information on data source, study population, cohort sizes, outcomes, study methodology and performance. Predictive performance was evaluated using C statistics for binary outcomes and R² for continuous outcomes. The PROSPERO ID for this review is CRD42020165498.

RESULTS

Of 19,112 articles that were retrieved, 124 were full-text screened and 25 were included, each of which represented a de novo or updated drug-based index. The indices were customized to varied age groups and clinical populations and most commonly evaluated outcomes including mortality (36%), hospitalization (24%) and healthcare costs (24%). C statistics ranged from 0.62 to 0.92 for mortality and 0.59 to 0.72 for hospitalization, while adjusted R² for healthcare costs ranged from 0.06 to 0.62. Seven of the 25 risk indices included used global drug classification algorithms.

CONCLUSIONS

More than two-dozen prescription drug-based risk indices have been developed and they differ significantly in design, performance and application.

Proactive inter-disciplinary CME to improve medication management in the elderly population

BACKGROUND

The absence of collaboration between health professionals is known to influence prescriptions' quality, also disadvantaging elderly frail patients' polytherapies.

OBJECTIVES

This study aims to improve the adherence to medications of elderly patients suffering from multiple diseases through interpersonal continuing medical education (CME). The CME was organized for general practitioners (GPs) by hospital pharmacists (HPs) from a Territorial Pharmaceutical Centre of Piedmont, in collaboration with pharmacists from the Drug Science and Technology Department of the University of Turin, to enhance awareness on the management of chronic therapies and de-prescription.

METHODS

Pharmacists set face-to-face lessons for GPs between April 2018 and November 2018, while therapies' reconciliation and delivery of the Illustrated Therapy Schedules (ITS) lasted until September 2019. Polytherapies were evaluated by pharmacists and GPs in terms of appropriateness (number of potentially inappropriate prescriptions - PIPs according to 2019 Beers Criteria) and number of drug-drug interactions (DDIs), using a clinical decision support system (CDSS - NavFarma©) to help health professionals dealing with the process of review, reconciliation and individuation of possible adverse reactions.

RESULTS

From the CME organization it emerged that the collaboration between health professionals supported by a CDSS could improve the quality of elderly patients polytherapies. Two-hundred fifteen patients were enrolled by GPs; patients included were aged - results reported as average (sd) - 76.4 (6.3), mostly men (54.9%), number of daily medications per patient was 8.1 (2.4); 2.1 (1.8) DDIs per patient were individuated, 12.9% of which were solved thanks to the CME. Average number of PIPs found was 2.5 (1.4) per patient.

CONCLUSIONS

The CME represented a proactive approach by HPs to the management of elderly patients' polytherapies. Moreover, clinicians' engagement is a mean to enhance quality, safety, professionalism and communication in health processes.

Discontinuation of antipsychotics treatment for elderly patients within a specialized behavioural unit: a retrospective review

Background Best practice guidelines recommend regular evaluation of antipsychotics in managing behaviours for dementia patients with a view to de-prescribing, given its significant mortality and adverse outcomes (Reus et al. in Am J Psychiatry 173(5):543-546, 2016, Deprescribing Guidelines and Algorithms in https://deprescribing.org/resources/deprescribing-guidelines-algorithms/ , 2019). The relationship between the dose of antipsychotic and the probability of discontinuation remains unknown in hospitalized dementia patients. Objectives This study aims to examine the relationship between high dose antipsychotic (greater than 62 mg chlorpromazine equivalent daily dose) and antipsychotics discontinuation in hospitalized dementia patients. Setting Specialized Dementia Behavioral Health Program in Hamilton, Ontario, Canada. Method A retrospective chart review was completed from August to December of 2019. A univariate logistic regression model was applied to antipsychotic doss (in chlorpromazine equivalent) and antipsychotic discontinuation outcome at 60 days (Narayan and Nishtala in Eur J Clin Pharmacol 73(12):1665-1672, 2017). A multivariant model was used to assess potential confounders, including other psychiatric medication exposure and Medicines Comorbidity Index (Luthra in J Gerontol Geriatr Res 4(260):2, 2015). Regression and dose-response models were utilized to identify the threshold dose (maximum daily dose). Main outcome measures Antipsychotic discontinuation at 60 days after the last dose. Results A total of 42 patients were eligible for outcome analysis. High dose antipsychotic was associated with worse discontinuation outcomes in both unadjusted (odds ratio, 0.09; 95% confidence interval, 0.02-0.37; p < 0.01) and adjusted generalized estimation equation models (odds ratio 0.65; 95% confidence interval, 0.59-0.72; p = 0.01). There were no statistically significant associations between baseline comorbidities (Medicines Comorbidity Index) (p = 0.68), mood stabilizer (p = 0.14), benzodiazepines (p = 0.93) and antidepressant exposure (p = 0.68) with antipsychotic discontinuation. The logistic regression model identified 40.7 mg of quetiapine, 1.7 mg of olanzapine and 0.51 mg of risperidone as the threshold dose, balancing sensitivity and specificity. The dose-response model also identified similar doses of 42 mg of quetiapine, 1.76 mg of olanzapine and 0.53 mg of risperidone. Conclusion The use of high dose antipsychotics is associated with worse discontinuation outcomes in hospitalized dementia patients. Therefore, our results suggest not exceeding a daily dose of 50 mg of quetiapine, 1.75 mg of olanzapine and 0.5 mg of risperidone when used for responsive behaviours and reassess the benefits and risks for each patient regularly.

Comparative risk of Parkinsonism associated with olanzapine, risperidone and quetiapine in older adults-a propensity score matched cohort study

PURPOSE

The purpose of this study was to examine the incidence of Parkinsonism in new users of second-generation antipsychotics (SGAs) in older adults (≥65 years). In the secondary analyses, we examined the risk of Parkinsonism by type and dose of SGA and conducted age-sex interactions.

METHOD

This population-based study included older adults who had a new-onset diagnosis of Parkinsonism and who started taking olanzapine, risperidone or quetiapine between 1 January 2005, and 30 December 2016. The Cox proportional hazard (COXPH) model with inverse probability treatment weighted (IPTW) covariates was used to evaluate the risk of new-onset Parkinsonism associated with SGAs, using quetiapine as the reference. We used the Generalized Propensity Score method to evaluate the dose-response risk of Parkinsonism associated with SGAs.

RESULTS

After IPTW adjustment for covariates, the COXPH model showed that compared to quetiapine, the use of olanzapine and risperidone were associated with an increased risk of Parkinsonism. The IPTW-hazard ratios are 1.76 (95% confidence interval 1.57-1.97) and 1.31 (95%CI 1.16-1.49), respectively. The dose-response risk of Parkinsonism was highest for olanzapine with a hazard ratio of 1.69 (95%CI 1.40-2.05) and the least for quetiapine with a hazard ratio of 1.22 (95%CI 1.14-1.31). The risk of Parkinsonism in the 65 to 74-year age group was higher for both sexes with risperidone compared to olanzapine, but the risk increased with olanzapine for both sexes in the 85+ age group.

CONCLUSION

The study found that the risk of new-onset Parkinsonism in older adults is 31% and 76% higher with risperidone and olanzapine respectively compared to quetiapine.

GheOP3 S tool and START/STOPP criteria version 2 for screening of potentially inappropriate medications and omissions in nursing home residents

RATIONALE, AIMS, AND OBJECTIVE

There is limited information about the comparative effectiveness of the START/STOPP (Screening Tool of Older Person's Prescriptions/Screening Tool to Alert doctors to Right Treatment) criteria and the Ghent Older People's Prescriptions community Pharmacy Screening tool (GheOP³ S tool) for the screening of potentially inappropriate prescribing (PIP) in the geriatric population. Considering this, the aim of this study was to compare the ability of the START/STOPP criteria and GheOP³ S tool to identify the PIP and potential prescribing omissions (PPOs) among elderly patients visiting their primary care physician.

METHODS

This is a retrospective observational study where a total of 422 subjects were included. The Charlson Co-morbidity Index (CCI) and the Medicines Co-morbidity Index (MCI) for older people were used to determine the co-morbidity status. The user's diagnosis and medications prescribed were analysed with the START/STOPP criteria and GheOP³ S tool. The Wilcoxon signed rank test was used to compare these criteria. The statistical relationship between the occurrence of PIP and users' age, the number of medication prescribed, the number of diagnoses, CCI, and MCI was determined with one-tailed bivariate correlation.

RESULTS

The START/STOPP criteria detected 843 PIPs and 1067 PPOs, while the GheOP³ S tool detected 936 PIPs and 202 PPOs. The GheOP³ S tool detected significantly more PIPs than did the STOPP criteria (P = 0.003). A significantly higher number of PPOs were detected with the START criterion (P < 0.0001). The results obtained with the START/STOPP criteria positively correlated with mentioned variables. Oppositely, there is a negative correlation between the results obtained with the GheOP³ S tool and age. Still, the positive correlation could be found with the rest of the variables.

CONCLUSION

The results of this study indicate that both tested tools demonstrated efficiency to detect PIPs and PPOs. The GheOP³ S tool detected significantly more PIPs than did the STOPP criteria. On the other hand, the START criteria performed much better for the screening of PPOs.

Assessment of Medication Safety Using Only Dispensing Data

PURPOSE OF REVIEW

The purpose of this review is to provide an overview of the published studies that have been used to generate evidence on the safety of medicine use when only medication dispensing data are available.

RECENT FINDINGS

Medication dispensing databases are increasingly available for research on large populations, particularly in countries that provide universal coverage for medicines. These data are often used for drug utilisation studies to identify inappropriate medicine use at the population level that may be associated with known safety issues. Lack of coded diagnoses, to identify outcomes, and lack of data on confounders can limit use of these data in practice for medication safety assessment. To overcome these issues, studies have exploited the fact that symptoms of adverse effects of medications can be treated with other medications, for example antidepressants to treat depression or oxybutynin to treat urinary incontinence. The challenge of unmeasured confounding has been addressed by implementing self-controlled study designs that use within-person comparisons and provide inherent control for confounding. Prescription sequence symmetry analysis (SSA) is a within-person study design that has been demonstrated as a useful tool for safety signal generation in dispensing data.

SUMMARY

Using medicine initiation as a proxy for the development of adverse events can help to generate evidence of the safety of medicines when only medication dispensing data are available. Careful consideration, however, should be given to the sensitivity and specificity of the proxy medicine for the adverse event and potential for time-varying confounding due to trends in medicine utilisation. Data-mining approaches using dispensing data have the potential to improve safety assessments; however, the challenge of unmeasured confounding with these methods remains to be investigated.