Development of a detection algorithm for statin-induced myopathy using electronic medical records

Development and Validation of the Pharmacological Statin-Associated Muscle Symptoms Risk Stratification Score Using Electronic Health Record Data

Statin-associated muscle symptoms (SAMS) can lead to statin nonadherence. This paper aims to develop a pharmacological SAMS risk stratification (PSAMS-RS) score using a previously developed PSAMS phenotyping algorithm that distinguishes objective vs. nocebo SAMS using electronic health record (EHR) data. Using our PSAMS phenotyping algorithm, SAMS cases and controls were identified from Minnesota Fairview EHR, with the statin user cohort divided into derivation (January 1, 2010, to December 31, 2018) and validation (January 1, 2019, to December 31, 2020) cohorts. A Least Absolute Shrinkage and Selection Operator regression model was applied to identify significant features for PSAMS. PSAMS-RS scores were calculated and the clinical utility of stratifying PSAMS risk was assessed by comparing hazard ratios (HRs) between fourth vs. first score quartiles. PSAMS cases were identified in 1.9% (310/16,128) of the derivation and 1.5% (64/4,182) of the validation cohorts. Sixteen out of 38 clinical features were determined to be significant predictors for PSAMS risk. Patients within the fourth quartile of the PSAMS scores had an over sevenfold (HR: 7.1, 95% confidence interval (CI): 4.03-12.45, derivation cohort) or sixfold (HR: 6.1, 95% CI: 2.15-17.45, validation cohort) higher hazard of developing PSAMS vs. those in their respective first quartile. The PSAMS-RS score is a simple tool to stratify patients' risk of developing PSAMS after statin initiation which could inform clinician-guided pre-emptive measures to prevent PSAMS-related statin nonadherence.

Development and application of pharmacological statin-associated muscle symptoms phenotyping algorithms using structured and unstructured electronic health records data

Importance

Statins are widely prescribed cholesterol-lowering medications in the United States, but their clinical benefits can be diminished by statin-associated muscle symptoms (SAMS), leading to discontinuation.

Objectives

In this study, we aimed to develop and validate a pharmacological SAMS clinical phenotyping algorithm using electronic health records (EHRs) data from Minnesota Fairview.

Materials and Methods

We retrieved structured and unstructured EHR data of statin users and manually ascertained a gold standard set of SAMS cases and controls using the published SAMS-Clinical Index tool from clinical notes in 200 patients. We developed machine learning algorithms and rule-based algorithms that incorporated various criteria, including ICD codes, statin allergy, creatine kinase elevation, and keyword mentions in clinical notes. We applied the best-performing algorithm to the statin cohort to identify SAMS.

Results

We identified 16 889 patients who started statins in the Fairview EHR system from 2010 to 2020. The combined rule-based (CRB) algorithm, which utilized both clinical notes and structured data criteria, achieved similar performance compared to machine learning algorithms with a precision of 0.85, recall of 0.71, and F1 score of 0.77 against the gold standard set. Applying the CRB algorithm to the statin cohort, we identified the pharmacological SAMS prevalence to be 1.9% and selective risk factors which included female gender, coronary artery disease, hypothyroidism, and use of immunosuppressants or fibrates.

Discussion and Conclusion

Our study developed and validated a simple pharmacological SAMS phenotyping algorithm that can be used to create SAMS case/control cohort to enable further analysis which can lead to the development of a SAMS risk prediction model.

Development and validation of the pharmacological statin-associated muscle symptoms risk stratification (PSAMS-RS) score using real-world electronic health record data

Introduction

Statin-associated muscle symptoms (SAMS) contribute to the nonadherence to statin therapy. In a previous study, we successfully developed a pharmacological SAMS (PSAMS) phenotyping algorithm that distinguishes objective versus nocebo SAMS using structured and unstructured electronic health records (EHRs) data. Our aim in this paper was to develop a pharmacological SAMS risk stratification (PSAMS-RS) score using these same EHR data.

Method

Using our PSAMS phenotyping algorithm, SAMS cases and controls were identified using University of Minnesota (UMN) Fairview EHR data. The statin user cohort was temporally divided into derivation (1/1/2010 to 12/31/2018) and validation (1/1/2019 to 12/31/2020) cohorts. First, from a feature set of 38 variables, a Least Absolute Shrinkage and Selection Operator (LASSO) regression model was fitted to identify important features for PSAMS cases and their coefficients. A PSAMS-RS score was calculated by multiplying these coefficients by 100 and then adding together for individual integer scores. The clinical utility of PSAMS-RS in stratifying PSAMS risk was assessed by comparing the hazard ratio (HR) between 4th vs 1st score quartile.

Results

PSAMS cases were identified in 1.9% (310/16128) of the derivation and 1.5% (64/4182) of the validation cohort. After fitting LASSO regression, 16 out of 38 clinical features were determined to be significant predictors for PSAMS risk. These factors are male gender, chronic pulmonary disease, neurological disease, tobacco use, renal disease, alcohol use, ACE inhibitors, polypharmacy, cerebrovascular disease, hypothyroidism, lymphoma, peripheral vascular disease, coronary artery disease and concurrent uses of fibrates, beta blockers or ezetimibe. After adjusting for statin intensity, patients in the PSAMS score 4th quartile had an over seven-fold (derivation) (HR, 7.1; 95% CI, 4.03-12.45) and six-fold (validation) (HR, 6.1; 95% CI, 2.15-17.45) higher hazard of developing PSAMS versus those in 1st score quartile.

Conclusion

The PSAMS-RS score can be a simple tool to stratify patients' risk of developing PSAMS after statin initiation which can facilitate clinician-guided preemptive measures that may prevent potential PSAMS-related statin non-adherence.

Development and Application of Pharmacological Statin-Associated Muscle Symptoms Phenotyping Algorithms Using Structured and Unstructured Electronic Health Records Data

Background

Statins are widely prescribed cholesterol-lowering medications in the US, but their clinical benefits can be diminished by statin-associated muscle symptoms (SAMS), leading to discontinuation. In this study, we aimed to develop and validate a pharmacological SAMS clinical phenotyping algorithm using electronic health records (EHRs) data from Minnesota Fairview.

Methods

We retrieved structured and unstructured EHR data of statin users and manually ascertained a gold standard set of SAMS cases and controls using the SAMS-CI tool from clinical notes in 200 patients. We developed machine learning algorithms and rule-based algorithms that incorporated various criteria, including ICD codes, statin allergy, creatine kinase elevation, and keyword mentions in clinical notes. We applied the best performing algorithm to the statin cohort to identify SAMS.

Results

We identified 16,889 patients who started statins in the Fairview EHR system from 2010-2020. The combined rule-based (CRB) algorithm, which utilized both clinical notes and structured data criteria, achieved similar performance compared to machine learning algorithms with a precision of 0.85, recall of 0.71, and F1 score of 0.77 against the gold standard set. Applying the CRB algorithm to the statin cohort, we identified the pharmacological SAMS prevalence to be 1.9% and selective risk factors which included female gender, coronary artery disease, hypothyroidism, use of immunosuppressants or fibrates.

Conclusion

Our study developed and validated a simple pharmacological SAMS phenotyping algorithm that can be used to create SAMS case/control cohort for further analysis such as developing SAMS risk prediction model.

Development and application of an algorithm for statin-induced myopathy based on electronic health record-derived structured elements

Objective

Considering the non-specific nature of muscle symptoms, studies of statin-induced myopathy (SIM) in electronic health records require accurate algortihms that can reliably identify true statin-related cases. However, prior algorithms have been constructed in study populations that preclude broad applicability. Here we developed and validated an algorithm that accurately defines SIM from electronic health records using structured data elements and conducted a study of determinants of SIM after applying the algorithm.

Materials and Methods

We used electronic records from an integrated health care delivery system (including comprehensive pharmacy dispensing records) and defined SIM as elevated creatine kinase (CK) ≥4 x upper limit of normal. A diverse cohort of participants receiving a variety of statin regimens met the criteria for study inclusion.

Results

We identified multiple conditions strongly associated with elevated CK independent of statin use. A 2-step algorithm was developed using these all-cause conditions as secondary causes (step 1) along with evidence of a statin regimen change (step 2). We identified 1,262 algorithm-derived statin-induced elevated CK cases. Gold standard SIM cases determined from manual chart reviews on a random subset of the all-cause elevated CK cases were used to validate the algorithm, which had a 76% sensitivity and 77% specificity for detecting the most certain cases. Pravastatin use was associated with a 2.18 odds (95% confidence interval 1.39-3.40, P=0.0007) for statin-induced CK elevation compared to lovastatin use after adjusting for dose and other factors.

Conclusions

We have produced an efficient, easy-to-apply methodological tool that can improve the quality of future research on statin-induced myopathy.

An overview of statin-induced myopathy and perspectives for the future

Introduction: Statins remain the most commonly prescribed lipid-lowering drug class for the treatment of atherosclerotic cardiovascular disease. Their well-recognized side effects are known as statin-associated muscle symptom (SAMS). Some advances in this field have been made in recent years, but the understanding of the mechanisms has lagged. Investigating the specific role of the anti-HMGCR autoantibody, pharmacokinetic genetic variants, characterization of the known phenotypes of statin toxicity, in relation to clinical markers of disease, is of high importance.Areas covered: We summarized currently available findings (on PubMed) related to SAMS and discussed the therapeutic approaches, risk factors, drug interactions, potential novel systems, algorithms and biomarkers for SAMS detection. CoQ10 supplementation has been suggested as a complementary approach to manage SAMS, while vitamin D levels may be useful for both the diagnosis and management.Expert Opinion/Commentary: Further studies might help to understand the easiest way to diagnose SAMS, suitable prevention and an effective non-statin therapy. This review sheds new light on the future directions in both research and clinical practice, which will help with rapid risk assessment, identification of the SAMS risk factors in order to decrease the incidence of statins' adverse effects, and the most effective therapy.

Pilot Algorithm Designed to Help Early Detection of HMG-CoA Reductase Inhibitor-Induced Hepatotoxicity

Objectives

To enable early detection of adverse drug reactions (ADRs) in patients using HMG-CoA reductase inhibitors (statins), we developed an algorithm that automatically detects liver injury caused by statins from Electronic Medical Record (EMR) data. We verified the performance of our algorithm through manual ADR assessment and a direct chart review.

Methods

The subjects in this study were patients who had been prescribed a statin for the first time among outpatients in Seoul St. Mary's Hospital in Korea between January 2009 and December 2012. We extracted basic information about the patients, including laboratory information, underlying disease, diagnosis information, prescription information, and concomitant drugs. We developed an automatic ADR detection algorithm by using EMR data. We validated the results of the algorithm through a chart review.

Results

We developed the algorithm to assess ADR occurrences based on alanine transaminase (ALT) and alkaline phosphatase (ALP) levels. According to the proposed algorithm, any of these result options could be attained: ADR-free, little association, strong association, and weak association or indeterminable. The results of the ADR assessments obtained using the proposed algorithm showed that the data of 126 patients (1.4% of all 9,241 patients) included suspicious figures, thus indicating the possibility of an ADR. In the EMR chart review for verifying the algorithm, ADRs of 33 patients were not associated with statin use; therefore, the ADR occurrence rate was found to be 1.0% (93/9,241). Therefore, the positive predictive value was calculated to be 73.8% (93/126; 95% confidence interval, 69.2%–77.6%). No differences were observed between statin types (p = 0.472).

Conclusions

For early detection of statin-induced liver injury, we developed an automatic ADR assessment algorithm. We expect that algorithms that are more reliable can be developed if we conduct supplement clinical studies with a focus on adverse drug effects.

A rule-based electronic phenotyping algorithm for detecting clinically relevant cardiovascular disease cases

Background

The implementation of electronic medical records (EMR) is becoming increasingly common. Error and data loss reduction, patient-care efficiency increase, decision-making assistance and facilitation of event surveillance, are some of the many processes that EMRs help improve. In addition, they show a lot of promise in terms of data collection to facilitate observational epidemiological studies and their use for this purpose has increased significantly over the recent years. Even though the quantity and availability of the data are clearly improved thanks to EMRs, still, the problem of the quality of the data remains. This is especially important when attempting to determine if an event has actually occurred or not. We sought to assess the sensitivity, specificity, and agreement level of a codes-based algorithm for the detection of clinically relevant cardiovascular (CaVD) and cerebrovascular (CeVD) disease cases, using data from EMRs.

Methods

Three family physicians from the research group selected clinically relevant CaVD and CeVD terms from the international classification of primary care, Second Edition (ICPC-2), the ICD 10 version 2015 and SNOMED-CT 2015 Edition. These terms included both signs, symptoms, diagnoses and procedures associated with CaVD and CeVD. Terms not related to symptoms, signs, diagnoses or procedures of CaVD or CeVD and also those describing incidental findings without clinical relevance were excluded. The algorithm yielded a positive result if the patient had at least one of the selected terms in their medical records, as long as it was not recorded as an error. Else, if no terms were found, the patient was classified as negative. This algorithm was applied to a randomly selected sample of the active patients within the hospital’s HMO by 1/1/2005 that were 40–79 years old, had at least one year of seniority in the HMO and at least one clinical encounter. Thus, patients were classified into four groups: (1) Negative patients (2) Patients with CaVD but without CeVD; (3) Patients with CeVD but without disease CaVD; (4) Patients with both diseases. To facilitate the validation process, a stratified sample was taken so that each of the groups represented approximately 25% of the sample. Manual chart review was used as the gold standard for assessing the algorithm’s performance. One-third of the patients were assigned randomly to each reviewer (Cohen’s kappa 0.91). Both coded and un-coded (free text) sections of the EMR were reviewed. This was done from the first present clinical note in the patients chart to the last one registered prior to 1/1/2005.

Results

The performance of the algorithm was compared against manual chart review. It yielded high sensitivity (0.99, 95% CI 0.938–0.9971) and acceptable specificity (0.86, 95% CI 0.818–0.895) for detecting cases of CaVD and CeVD combined. A qualitative analysis of the false positives and false negatives was performed.

Conclusions

We developed a simple algorithm, using only standardized and non-standardized coded terms within an EMR that can properly detect clinically relevant events and symptoms of CaVD and CeVD. We believe that combining it with an analysis of the free text using an NLP approach would yield even better results.

Electronic supplementary material

The online version of this article (doi:10.1186/s13104-017-2600-2) contains supplementary material, which is available to authorized users.

Leveraging Food and Drug Administration Adverse Event Reports for the Automated Monitoring of Electronic Health Records in a Pediatric Hospital

The objective of this study was to determine whether the Food and Drug Administration’s Adverse Event Reporting System (FAERS) data set could serve as the basis of automated electronic health record (EHR) monitoring for the adverse drug reaction (ADR) subset of adverse drug events. We retrospectively collected EHR entries for 71 909 pediatric inpatient visits at Cincinnati Children’s Hospital Medical Center. Natural language processing (NLP) techniques were used to identify positive diseases/disorders and signs/symptoms (DDSSs) from the patients’ clinical narratives. We downloaded all FAERS reports submitted by medical providers and extracted the reported drug-DDSS pairs. For each patient, we aligned the drug-DDSS pairs extracted from their clinical notes with the corresponding drug-DDSS pairs from the FAERS data set to identify Drug-Reaction Pair Sentences (DRPSs). The DRPSs were processed by NLP techniques to identify ADR-related DRPSs. We used clinician annotated, real-world EHR data as reference standard to evaluate the proposed algorithm. During evaluation, the algorithm achieved promising performance and showed great potential in identifying ADRs accurately for pediatric patients.

Development and validation of algorithms for the detection of statin myopathy signals from electronic medical records

The purpose of this study was to develop and validate sensitive algorithms to detect hospitalized statin-induced myopathy (SIM) cases from electronic medical records (EMRs). We developed four algorithms on a training set of 31,211 patient records from a large tertiary hospital. We determined the performance of these algorithms against manually curated records. The best algorithm used a combination of elevated creatine kinase (>4× the upper limit of normal (ULN)), discharge summary, diagnosis, and absence of statin in discharge medications. This algorithm achieved a positive predictive value of 52-71% and a sensitivity of 72-78% on two validation sets of >30,000 records each. Using this algorithm, the incidence of SIM was estimated at 0.18%. This algorithm captured three times more rhabdomyolysis cases than spontaneous reports (95% vs. 30% of manually curated gold standard cases). Our results show the potential power of utilizing data and text mining of EMRs to enhance pharmacovigilance activities.

Development of a novel algorithm for detecting glucocorticoid-induced diabetes mellitus using a medical information database

WHAT IS KNOWN AND OBJECTIVE

Glucocorticoid-induced diabetes mellitus (GIDM) increases the risk of diabetes mellitus (DM)-related complications but is generally difficult to detect in clinical settings. The criteria for diagnosing GIDM have not been established. Recently, medical information databases (MIDs) have been used in post-marketing surveillance (PMS) studies. We conducted a pharmacoepidemiological study to develop an algorithm for detecting GIDM using MID.

METHODS

We selected 1214 inpatients who were newly prescribed with a typical glucocorticoid, prednisolone, during hospitalization from 2008 to 2014 from an MID of Hamamatsu University Hospital in Japan. GIDM was screened based on fasting blood glucose (FBG) and haemoglobin A1c (HbA1c) levels according to the current Japan Diabetes Society (JDS) DM criteria, and its predictability was evaluated by an expert's review of medical records. We investigated further candidate screening factors using receiver operating characteristics analysis.

RESULTS

Sixty-three inpatients were identified by the JDS DM criteria. Of these, 33 patients were definitely diagnosed as having GIDM by expert's review (positive predictive value = 52·4%). To develop a highly predictive algorithm, we compared the characteristics of inpatients diagnosed with definite GIDM and those diagnosed as non-GIDM. The maximum levels of HbA1c in patients with GIDM were significantly higher than those of patients with non-GIDM (66·9 mmol/mol vs. 58·7 mmol/mol, P < 0·001). The patients with GIDM had significantly higher relative increase in maximum level of HbA1c (RIM-HbA1c) than those with non-GIDM (0·3 vs. 0·03, P < 0·001). However, we did not observe a significant difference in those of fasting blood glucose (FBG) levels. We applied the RIM-HbA1c as a second screening factor to improve the detection of GIDM. It showed that a 13% increase in RIM-HbA1c separated patients with from patients without GIDM.

WHAT IS NEW AND CONCLUSIONS

Patients with GIDM had significantly higher RIM-HbA1c than patients with non-GIDM. There was a 13% increase in RIM-HbA1c in patients with GIDM compared to the others. Our detection algorithm for GIDM using an MID achieved high sensitivity and specificity, and was superior to one based only on the current JDS DM criteria. Our results suggest that monitoring changes in HbA1c levels is important for detecting GIDM and adds to current diagnostic criteria for type 2 DM.

Validity of diagnostic codes and laboratory measurements to identify patients with idiopathic acute liver injury in a hospital database

PURPOSE

The development and validation of algorithms to identify cases of idiopathic acute liver injury (ALI) are essential to facilitate epidemiologic studies on drug-induced liver injury. The aim of this study is to determine the ability of diagnostic codes and laboratory measurements to identify idiopathic ALI cases.

METHODS

In this cross-sectional validation study, patients were selected from the hospital-based Utrecht Patient Oriented Database between 2008 and 2010. Patients were identified using (I) algorithms based on ICD-9-CM codes indicative of idiopathic ALI combined with sets of liver enzyme values (ALT > 2× upper limit of normal (ULN); AST > 1ULN + AP > 1ULN + bilirubin > 1ULN; ALT > 3ULN; ALT > 3ULN + bilirubin > 2ULN; ALT > 10ULN) and (II) algorithms based on solely liver enzyme values (ALT > 3ULN + bilirubin > 2ULN; ALT > 10ULN). Hospital medical records were reviewed to confirm final diagnosis. The positive predictive value (PPV) of each algorithm was calculated.

RESULTS

A total of 707 cases of ALI were identified. After medical review 194 (27%) patients had confirmed idiopathic ALI. The PPV for (I) algorithms with an ICD-9-CM code as well as abnormal tests ranged from 32% (13/41) to 48% (43/90) with the highest PPV found with ALT > 2ULN. The PPV for (II) algorithms with liver test abnormalities was maximally 26% (150/571).

CONCLUSIONS

The algorithm based on ICD-9-CM codes indicative of ALI combined with abnormal liver-related laboratory tests is the most efficient algorithm for identifying idiopathic ALI cases. However, cases were missed using this algorithm, because not all ALI cases had been assigned the relevant diagnostic codes in daily practice.

Phenotype standardization for statin-induced myotoxicity

Statins are widely used lipid-lowering drugs that are effective in reducing cardiovascular disease risk. Although they are generally well tolerated, they can cause muscle toxicity, which can lead to severe rhabdomyolysis. Research in this area has been hampered to some extent by the lack of standardized nomenclature and phenotypic definitions. We have used numerical and descriptive classifications and developed an algorithm to define statin-related myotoxicity phenotypes, including myalgia, myopathy, rhabdomyolysis, and necrotizing autoimmune myopathy.