Validation of an international classification of disease, tenth revision, clinical modification (ICD-10-CM) algorithm in identifying severe hypoglycaemia events for real-world studies.

AIM: The transition to the ICD-10-CM coding system has reduced the utility of hypoglycaemia algorithms based on ICD-9-CM diagnosis codes in real-world studies of antidiabetic drugs. We mapped a validated ICD-9-CM hypoglycaemia algorithm to ICD-10-CM codes to create an ICD-10-CM hypoglycaemia algorithm and assessed its performance in identifying severe hypoglycaemia. MATERIALS AND METHODS: We assembled a cohort of Medicare patients with DM and linked electronic health record (EHR) data to the University of North Carolina Health System and identified candidate severe hypoglycaemia events from their Medicare claims using the ICD-10-CM hypoglycaemia algorithm. We confirmed severe hypoglycaemia by EHR review and computed a positive predictive value (PPV) of the algorithm to assess its performance. We refined the algorithm by removing poor performing codes (PPV ≤0.5) and computed a Cohen's κ statistic to evaluate the agreement of the EHR reviews. RESULTS: The algorithm identified 642 candidate severe hypoglycaemia events, and we confirmed 455 as true severe hypoglycaemia events, PPV of 0.709 (95% confidence interval: 0.672, 0.744). When we refined the algorithm, the PPV increased to 0.893 (0.862, 0.918) and missed <2.42% (<11) true severe hypoglycaemia events. Agreement between reviewers was high, κ = 0.93 (0.89, 0.97). CONCLUSIONS: We translated an ICD-9-CM hypoglycaemia algorithm to an ICD-10-CM version and found its performance was modest. The performance of the algorithm improved by removing poor performing codes at the trade-off of missing very few severe hypoglycaemia events. The algorithm has the potential to be used to identify severe hypoglycaemia in real-world studies of antidiabetic drugs.

Prescriber Uncertainty as Opportunity to Improve Care of Type 2 Diabetes with Chronic Kidney Disease: Mixed Methods Study.

BACKGROUND: Over 5 million patients in the United States have type 2 diabetes mellitus (T2D) with chronic kidney disease (CKD); antidiabetic drug selection for this population is complex and has important implications for outcomes. OBJECTIVE: To better understand how providers choose antidiabetic drugs in T2D with CKD DESIGN: Mixed methods. Interviews with providers underwent qualitative analysis using grounded theory to identify themes related to antidiabetic drug prescribing. A provider survey used vignettes and direct questions to quantitatively assess prescribers' knowledge and preferences. A retrospective cohort analysis of real-world prescribing data assessed the external validity of the interview and survey findings. PARTICIPANTS: Primary care physicians, endocrinologists, nurse-practitioners, and physicians' assistants were eligible for interviews; primary care physicians and endocrinologists were eligible for the survey; prescribing data were derived from adult patients with serum creatinine data. MAIN MEASURES: Interviews were qualitative; for the survey and retrospective cohort, proportion of patients receiving metformin was the primary outcome. KEY RESULTS: Interviews with 9 providers identified a theme of uncertainty about guidelines for prescribing antidiabetic drugs in patients with T2D and CKD. The survey had 105 respondents: 74 primary care providers and 31 endocrinologists. Metformin was the most common choice for patients with T2D and CKD. Compared to primary care providers, endocrinologists were less likely to prescribe metformin at levels of kidney function at which it is contraindicated and more likely to correctly answer a question about metformin's contraindications (71% versus 41%) (p < .05). Real-world data were consistent with survey findings, and further showed low rates of use of sodium-glucose cotransporter 2 inhibitors and glucagon-like peptide 1 receptor agonists (<10%) in patients with eGFR below 60 ml/min/1.73m2. CONCLUSIONS: Providers are unsure how to treat T2D with CKD and incompletely informed as to existing guidelines. This suggests opportunities to improve care.

Baseline haemoglobin A1c and the risk of COVID-19 hospitalization among patients with diabetes in the INSIGHT Clinical Research Network.

AIMS: To examine the association between baseline glucose control and risk of COVID-19 hospitalization and in-hospital death among patients with diabetes. METHODS: We performed a retrospective cohort study of adult patients in the INSIGHT Clinical Research Network with a diabetes diagnosis and haemoglobin A1c (HbA1c) measurement in the year prior to an index date of March 15, 2020. Patients were divided into four exposure groups based on their most recent HbA1c measurement (in mmol/mol): 39-46 (5.7%-6.4%), 48-57 (6.5%-7.4%), 58-85 (7.5%-9.9%), and ≥86 (10%). Time to COVID-19 hospitalization was compared in the four groups in a propensity score-weighted Cox proportional hazards model adjusting for potential confounders. Patients were followed until June 15, 2020. In-hospital death was examined as a secondary outcome. RESULTS: Of 168,803 patients who met inclusion criteria; 50,016 patients had baseline HbA1c 39-46 (5.7%-6.4%); 54,729 had HbA1c 48-57 (6.5-7.4%); 47,640 had HbA1c 58-85 (7.5^%-9.9%) and 16,418 had HbA1c ≥86 (10%). Compared with patients with HbA1c 48-57 (6.5%-7.4%), the risk of hospitalization was incrementally greater for those with HbA1c 58-85 (7.5%-9.9%) (adjusted hazard ratio [aHR] 1.19, 95% confidence interval [CI] 1.06-1.34) and HbA1c ≥86 (10%) (aHR 1.40, 95% CI 1.19-1.64). The risk of COVID-19 in-hospital death was increased only in patients with HbA1c 58-85 (7.5%-9.9%) (aHR 1.29, 95% CI 1.06, 1.61). CONCLUSIONS: Diabetes patients with high baseline HbA1c had a greater risk of COVID-19 hospitalization, although association between HbA1c and in-hospital death was less consistent. Preventive efforts for COVID-19 should be focused on diabetes patients with poor glucose control.

Medication selection, health services outcomes, and cost trajectories for Medicaid beneficiaries with infantile spasms.

OBJECTIVE: There are three recommended first-line treatments for infantile spasms, adrenocorticotropic hormone (ACTH), oral corticosteroids, and vigabatrin, though non-standard treatments such as topiramate are sometimes selected. Is it uncertain how treatment selection influences health services outcomes. METHODS: We conducted a retrospective cohort study of Medicaid beneficiaries newly diagnosed with infantile spasms from 2009-2010. We included infants with a new diagnosis of infantile spasms between age 2-9 months who filled ACTH (reference), prednisolone, vigabatrin, or topiramate prescriptions. Multivariable Cox proportional hazards regression compared time to first emergency department (ED) visit or hospitalization across treatment groups during 2 years of follow-up. Monthly costs for each treatment were examined in 6-month intervals and compared in a multivariable generalized linear model. RESULTS: Among 256 children with infantile spasms, 116 received ACTH, 62 prednisolone, 15 vigabatrin, and 63 topiramate. The rate of ED visit or hospitalization per person-year did not differ significantly for prednisolone (0.9 [95 % CI 0.7-1.2]; adjusted hazard ratio [aHR] 0.84, 95 % CI 0.57-1.24), vigabatrin (0.8 [95 % CI 0.4-1.5]; aHR 0.91, 95% CI 0.45-1.84), or topiramate (1.7 [95 % CI 1.3-2.3]; aHR 1.15, 95 % CI 0.80-1.65), when compared to ACTH (1.1 [95 % CI 0.9-1.3]). The median payment for ACTH was $96,406 (interquartile range 70,742-138,476) during the first 6 months. The adjusted mean total payment in the first 6 months was 73% lower for prednisolone (95% CI -82, -61), 69% lower for vigabatrin (95% CI -84, -40), and 73% lower for topiramate (95% CI -82, -59). However, in subsequent 6-month intervals, costs associated with ACTH were not significantly different compared to other treatments. SIGNIFICANCE: Compared to other treatments for infantile spasms, use of ACTH was associated with greater cost in the first 6 months of treatment, but not with reduced ED visits or hospitalizations. The cost effectiveness of ACTH depends on its relative clinical efficacy, and merits additional research.

Safety and effectiveness of metformin in patients with reduced renal function: A systematic review.

AIM: To examine clinical and safety outcomes associated with metformin use in patients with impaired renal function. MATERIALS AND METHODS: We searched PubMed and Embase databases from inception to August 2020, supplementing our search with a review of investigator files and reference lists of included studies. Any study reporting original data on metformin and patient-centred outcomes in patients with impaired renal function, defined as an estimated glomerular filtration rate (eGFR) of less than 60 mL/min/1.73m2 , was included. Post hoc meta-analysis was performed for the outcomes of mortality, cardiovascular events and acidosis. RESULTS: Nine small prospective studies enrolling patients with significantly impaired renal function identified only one case of clinically apparent lactic acidosis. Among 13 larger retrospective studies, seven examined the risk of mortality across patient subgroups; meta-analysis showed reductions in overall mortality at an eGFR of 45 mL/min/1.73m2 or higher but not at an eGFR of less than 45 mL/min/1.73m2 . Eight retrospective studies evaluated acidosis as an outcome; meta-analysis showed no increase in risk of acidosis except at an eGFR of less than 30 mL/min/1.73m2 , in which group the HR was 1.97 (95% CI 1.03-3.77). CONCLUSIONS: The literature shows metformin to be associated with reduced mortality and no increased risk of acidosis at an eGFR of 45 mL/min/1.73m2 or higher. Metformin appears to be associated with fewer benefits and possible increases in the risk of acidosis at an eGFR of less than 30 mL/min/1.73m2 . Consistent with US Food and Drug Administration guidelines, metformin should not be used at an eGFR less than 30 mL/min/1.73m2 , and further research on its risk-benefit profile at eGFR values approaching 30 mL/min/1.73m2 is warranted.

Building a Data Linkage Foundation for Mother-Child Pharmacoepidemiology Research.

BACKGROUND: Expanding our understanding of the effects of maternal medication exposure through research is a public health priority and will help inform both clinical and policy decision making, ultimately improving outcomes for pregnant women and their children. OBJECTIVE: Our objective was to describe a linked-data research platform that facilitates studies of pregnancy medication exposures and policy changes on maternal and child health outcomes. METHODS: Mothers receiving Medicaid benefits were probabilistically linked with newborns in the Tennessee Medicaid program (TennCare) through three distinct linkage processes. Medicaid claims data and state birth and fetal death certificate records (vital records) were used to identify and link potential mothers, deliveries, and newborn children. The linkage process started with the creation of a merged pool of potential mothers and eligible deliveries, which was linked to vital records and to children's records. In the last step, linked records from the preceding steps were combined into the final Mother-child linked records. For each data linkage step, rubrics and scoring systems for exact and partial matches and mismatches among key linkage fields were applied and used to examine the strength of the probabilistic linkages. Summary linkage yields for year 2013 are reported for illustration purposes. RESULTS: Among the 84,253 potential deliveries, 1,761,557 eligible potential mothers, and 51,400 eligible children identified in Tennessee Medicaid records in 2013, a total of 60,265 of these records were uniquely linked to vital records, including 46,172 (77%) with linked mother-child-vital records. Among the 51,400 eligible children records identified in Tennessee Medicaid for that year, 97% (50,053) had at least one link to vital records or a mother-delivery record. In linked records, the median maternal age was 24 years, and the median gestational age was 39 weeks. About 33% of pregnant women underwent cesarean birth, and 1% of births were classified as complicated deliveries. CONCLUSIONS: Supplementing existing Medicaid claims data with birth certificate records complements administrative claims information and allows for detailed assessments of pregnancy exposures and policy changes on mother and child outcomes.

Evaluation of a Care Management Program for Pediatric Epilepsy Patients.

OBJECTIVE: To evaluate the impact of a pediatric epilepsy care management intervention on emergency department visits, hospitalizations, and seizure freedom. METHODS: We conducted a prospective observational study at a single academic medical center. Children with epilepsy with high risk of frequent emergency department use were enrolled in the intervention from January through May 2015, which included a baseline visit and follow-up support from a care management team. Controls selected from the same institution received standard of care. Baseline and follow-up information were collected from electronic health records and surveys (Family Impact Scale, Pediatric Epilepsy Medication Self-Management Questionnaire). Propensity score-weighted logistic regression compared emergency department visits, unplanned hospitalizations, and 3-month seizure freedom after 1 year in the intervention vs control groups. RESULTS: A total of 56 children were enrolled in the intervention and 359 received standard of care. The intervention group was younger and had greater use of health services at baseline. When comparing the intervention to standard of care after 1 year, we found no significant difference in the risk of any emergency department visit (adjusted odds ratio [OR] 2.2, 95% confidence interval [CI] 0.6-8.5) or seizure freedom (adjusted OR 2.5, 95% CI 0.3-21.5). However, the risk of unplanned hospital admissions remained higher in the intervention group (adjusted OR 23.1, 95% CI 5.1-104). CONCLUSION: We did not find that children with epilepsy who received a care management intervention had less use of health services or better clinical outcomes after a year compared with controls. The study is limited by small sample size and nonrandomized study design.

Developing a Standardized and Reusable Method to Link Distributed Health Plan Databases to the National Death Index: Methods Development Study Protocol.

BACKGROUND: Certain medications may increase the risk of death or death from specific causes (eg, sudden cardiac death), but these risks may not be identified in premarket randomized trials. Having the capacity to examine death in postmarket safety surveillance activities is important to the US Food and Drug Administration's (FDA) mission to protect public health. Distributed networks of electronic health plan databases used by the FDA to conduct multicenter research or medical product safety surveillance studies often do not systematically include death or cause-of-death information. OBJECTIVE: This study aims to develop reusable, generalizable methods for linking multiple health plan databases with the Centers for Disease Control and Prevention's National Death Index Plus (NDI+) data. METHODS: We will develop efficient administrative workflows to facilitate multicenter institutional review board (IRB) review and approval within a distributed network of 6 health plans. The study will create a distributed NDI+ linkage process that avoids sharing of identifiable patient information between health plans or with a central coordinating center. We will develop standardized criteria for selecting and retaining NDI+ matches and methods for harmonizing linked information across multiple health plans. We will test our processes within a use case comprising users and nonusers of antiarrhythmic medications. RESULTS: We will use the linked health plan and NDI+ data sets to estimate the incidences and incidence rates of mortality and specific causes of death within the study use case and compare the results with reported estimates. These comparisons provide an opportunity to assess the performance of the developed NDI+ linkage approach and lessons for future studies requiring NDI+ linkage in distributed database settings. This study is approved by the IRB at Harvard Pilgrim Health Care in Boston, MA. Results will be presented to the FDA at academic conferences and published in peer-reviewed journals. CONCLUSIONS: This study will develop and test a reusable distributed NDI+ linkage approach with the goal of providing tested NDI+ linkage methods for use in future studies within distributed data networks. Having standardized and reusable methods for systematically obtaining death and cause-of-death information from NDI+ would enhance the FDA's ability to assess mortality-related safety questions in the postmarket, real-world setting. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/21811.

Association of Treatment With Metformin vs Sulfonylurea With Major Adverse Cardiovascular Events Among Patients With Diabetes and Reduced Kidney Function.

IMPORTANCE: Before 2016, safety concerns limited metformin use in patients with kidney disease; however, the effectiveness of metformin on clinical outcomes in patients with reduced kidney function remains unknown. OBJECTIVE: To compare major adverse cardiovascular events (MACE) among patients with diabetes and reduced kidney function who continued treatment with metformin or a sulfonylurea. DESIGN, SETTING, AND PARTICIPANTS: Retrospective cohort study of US veterans receiving care within the national Veterans Health Administration, with data supplemented by linkage to Medicare, Medicaid, and National Death Index data from 2001 through 2016. There were 174 882 persistent new users of metformin and sulfonylureas who reached a reduced kidney function threshold (estimated glomerular filtration rate <60 mL/min/1.73 m2 or creatinine ≥1.4 mg/dL for women or ≥1.5 mg/dL for men). Patients were followed up from reduced kidney function threshold until MACE, treatment change, loss to follow-up, death, or study end (December 2016). EXPOSURES: New users of metformin or sulfonylurea monotherapy who continued treatment with their glucose-lowering medication after reaching reduced kidney function. MAIN OUTCOMES AND MEASURES: MACE included hospitalization for acute myocardial infarction, stroke, transient ischemic attack, or cardiovascular death. The analyses used propensity score weighting to compare the cause-specific hazard of MACE between treatments and estimate cumulative risk accounting for the competing risks of changing therapy or noncardiovascular death. RESULTS: There were 67 749 metformin and 28 976 sulfonylurea persistent monotherapy users; the weighted cohort included 24 679 metformin and 24 799 sulfonylurea users (median age, 70 years [interquartile range {IQR}, 62.8-77.8]; 48 497 men [98%]; and 40 476 white individuals [82%], with median estimated glomerular filtration rate of 55.8 mL/min/1.73 m2 [IQR, 51.6-58.2] and hemoglobin A1c level of 6.6% [IQR, 6.1%-7.2%] at cohort entry). During follow-up (median, 1.0 year for metformin vs 1.2 years for sulfonylurea), there were 1048 MACE outcomes (23.0 per 1000 person-years) among metformin users and 1394 events (29.2 per 1000 person-years) among sulfonylurea users. The cause-specific adjusted hazard ratio of MACE for metformin was 0.80 (95% CI, 0.75-0.86) compared with sulfonylureas, yielding an adjusted rate difference of 5.8 (95% CI, 4.1-7.3) fewer events per 1000 person-years of metformin use compared with sulfonylurea use. CONCLUSIONS AND RELEVANCE: Among patients with diabetes and reduced kidney function persisting with monotherapy, treatment with metformin, compared with a sulfonylurea, was associated with a lower risk of MACE.

Evaluation of weight change and hypoglycaemia as mediators in the association between insulin use and death.

AIM: To evaluate whether weight change or hypoglycaemia mediates the association between insulin use and death. MATERIALS AND METHODS: In a retrospective cohort of veterans who filled a new prescription for metformin and added insulin or sulphonylurea (2001-2012), we assessed change in body mass index (BMI) and hypoglycaemia during the first 12 months of treatment intensification. Cox proportional hazards models compared the risk of death between treatment groups. Using the difference method, we estimated the indirect effect and proportion mediated through each mediator. A sensitivity analysis assessed mediators in the first 6 months of intensified therapy. RESULTS: Among 28 892 patients surviving 12 months, deaths per 1000 person-years were 15.4 for insulin users and 12.9 for sulphonylurea users (HR 1.20, 95% CI 0.87, 1.64). Change in BMI and hypoglycaemia mediated 13% (-98, 98) and -1% (-37, 71) of this association, respectively. Among 30 214 patients surviving 6 months, deaths per 1000 person-years were 34.8 for insulin users and 21.3 for sulphonylurea users (HR 1.66, 95% CI 1.28, 2.15). Change in BMI and hypoglycaemia mediated 9% (1, 23) and 0% (-9, 4) of this association, respectively. CONCLUSIONS: We observed an increased risk of death among metformin users intensifying treatment with insulin versus sulphonylurea and surviving 6 months of intensified therapy, but not among those surviving 12 months. This association was mediated in part by weight change.

A comparison of two algorithms to identify sudden cardiac deaths in computerized databases.

PURPOSE: Two previously validated algorithms to identify sudden cardiac death using administrative data showed high positive predictive value. We evaluated the agreement between the algorithms using data from a common source population. METHODS: We conducted a cross-sectional study to assess the percent agreement between deaths identified by two sudden cardiac death algorithms using Tennessee Medicaid and death certificate data from 2007 through 2014. The source population included all deceased patients aged 18 to 64 years with Medicaid enrollment in the 6 months prior to death. To identify sudden cardiac deaths, algorithm 1 used only hospital/emergency department (ED) claims from encounters at the time of death, and algorithm 2 required death certificates and used claims data for specific exclusion criteria. RESULTS: We identified 34 107 deaths in the source population over the study period. The two algorithms identified 4372 potential sudden cardiac deaths: Algorithm 1 identified 3117 (71.3%) and algorithm 2 identified 1715 (39.2%), with 460 (10.5%) deaths identified by both algorithms. Of the deaths identified by algorithm 1, 1943 (62.3%) had an underlying cause of death not specified in algorithm 2. Of the deaths identified by algorithm 2, 1053 (61.4%) had no record of a hospital or ED encounter at the time of death, and 202 (11.8%) had a discharge diagnosis code not specified in algorithm 1. CONCLUSIONS: We found low agreement between the two algorithms for identification of sudden cardiac deaths because of differences in sudden cardiac death definitions and data sources.

Accuracy of a composite event definition for hypoglycemia.

PURPOSE: To evaluate the accuracy of a composite definition for the identification of hypoglycemia events that used both administrative claims and laboratory data in a cohort of patients. METHODS: We reviewed medical records in a sample of presumed hypoglycemia events among patients who received care at the Veterans Health Administration Tennessee Valley Healthcare System in 2001 to 2012. A hypoglycemia event was defined as a hospitalization or emergency department visit judged by the treating clinician to be due to hypoglycemia, or an outpatient laboratory or point-of-care blood glucose measurement <60 mg/dL. Based on medical record review, each event was classified as true positive (severe, documented symptomatic, documented asymptomatic) or false positive (probable symptomatic, not hypoglycemia). The positive predictive values (PPV) of the individual event types (hospitalization, emergency department, and outpatient) were estimated. RESULTS: Of 2250 events identified through the composite definition, 321 events (15 hospitalizations, 103 emergency department visits, and 203 outpatient events) were reviewed. The PPVs were 80% for hospitalization events, 48% for emergency department events, and 96% for outpatient events. The emergency department definition included a nonspecific diagnosis code for diabetic complications which captured many false positive events. Excluding this code from the definition improved the PPV for emergency department events to 70% and missed one true event. CONCLUSIONS: Our composite definition for hypoglycemia performed moderately well in a cohort of Veterans. Further evaluation of the emergency department events may be needed.

Opioid prescribing after childbirth: overprescribing and chronic use.

PURPOSE OF REVIEW: Overprescribing opioids contributes to the epidemic of drug overdoses and deaths in the United States. Opioids are commonly prescribed after childbirth especially after caesarean, the most common major surgery. This review summarizes recent literature on patterns of opioid overprescribing and consumption after childbirth, the relationship between opioid prescribing and chronic opioid use, and interventions that can help reduce overprescribing. RECENT FINDINGS: It is estimated that more than 80% of women fill opioid prescriptions after caesarean birth and about 54% of women after vaginal birth, although these figures vary greatly by geographical location and setting. After opioid prescriptions are filled, the median number of tablets used after caesarean is roughly 10 tablets and the majority of opioids dispensed (median 30 tablets) go unused. The quantity of opioid prescribed influences the quantity of opioid used. The risk of chronic opioid use related to opioid prescribing after birth may seem not high (annual risk: 0.12-0.65%), but the absolute number of women who are exposed to opioids after childbirth and become chronic opioid users every year is very large. Tobacco use, public insurance and depression are associated with chronic opioid use after childbirth. The risk of chronic opioid use among women who underwent caesarean and received opioids after birth is not different from the risk of women who received opioids after vaginal delivery. SUMMARY: Women are commonly exposed to opioids after birth. This exposure leads to an increased risk of chronic opioid use. Physician and providers should judiciously reduce the amount of opioids prescribed after childbirth, although more research is needed to identify the optimal method to reduce opioid exposure without adversely affecting pain management.

Evaluation of Frailty as an Unmeasured Confounder in Observational Studies of Antidiabetic Medications.

BACKGROUND: It is unknown whether observational studies evaluating the association between antidiabetic medications and mortality adequately account for frailty. Our objectives were to evaluate if frailty was a potential confounder in the relationship between antidiabetic medication regimen and mortality and how well administrative and clinical electronic health record (EHR) data account for frailty. METHODS: We conducted a retrospective cohort study in a single Veterans Health Administration (VHA) healthcare system of 500 hospitalizations-the majority due to heart failure-of Veterans who received regular VHA care and initiated type 2 diabetes treatment from 2001 to 2008. We measured frailty using a modified frailty index (FI, >0.21 frail). We obtained antidiabetic medication regimen and time-to-death from administrative sources. We compared FI among patients on different antidiabetic regimens. Stepwise Cox proportional hazards regression estimated time-to-death by demographic, administrative, clinical EHR, and FI data. RESULTS: Median FI was 0.22 (interquartile range 0.18, 0.27). Frailty differed across antidiabetic regimens (p < .001). An FI increase of 0.05 was associated with an increased risk of death (hazard ratio 1.45, 95% confidence interval 1.32, 1.60). Cox proportional hazards model for time-to-death including demographic, administrative, and clinical EHR data had a c-statistic of 0.70; adding FI showed marginal improvement (c-statistic 0.72). CONCLUSIONS: Frailty was associated with antidiabetic regimen and death, and may confound that relationship. Demographic, administrative, and clinical EHR data, commonly used to balance differences among exposure groups, performed moderately well in assessing risk of death, with minimal gain from adding frailty. Study design and analytic techniques can help minimize potential confounding by frailty in observational studies.

Diagnostic Algorithms for Cardiovascular Death in Administrative Claims Databases: A Systematic Review.

INTRODUCTION: Valid algorithms for identification of cardiovascular (CV) deaths allow researchers to reliably assess the CV safety of medications, which is of importance to regulatory science, patient safety, and public health. OBJECTIVE: The aim was to conduct a systematic review of algorithms to identify CV death in administrative health plan claims databases. METHODS: We searched MEDLINE, EMBASE, and Cochrane Library for English-language studies published between January 1, 2012 and October 17, 2017. We examined references in systematic reviews to identify earlier studies. Selection included any observational study using electronic health care data to evaluate the sensitivity, specificity, positive predictive value (PPV), or negative predictive value (NPV) of algorithms for CV death (sudden cardiac death [SCD], myocardial infarction [MI]-related death, or stroke-related death) among adults aged ≥ 18 years in the United States. Data were extracted by two independent reviewers, with disagreements resolved through further discussion and consensus. The Quality Assessment of Diagnostic Accuracy Studies-2 instrument was used to assess the risk of bias. RESULTS: Five studies (n = 4 on SCD, n = 1 on MI- and stroke-related death) were included after a review of 2053 citations. All studies reported algorithm PPVs, with incomplete reporting on other accuracy parameters. One study was at low risk of bias, three studies were at moderate risk of bias, and one study was at unclear risk of bias. Two studies identified community-occurring SCD: one identified events using International Classification of Disease, Ninth Revision (ICD-9) codes on death certificates and other criteria from medical claims (PPV = 86.8%) and the other identified events resulting in hospital presentation using first-listed ICD-9 codes on emergency department or inpatient medical claims (PPV = 92.3%). Two studies used death certificates alone to identify SCD (PPV = 27% and 32%, respectively). One study used medical claims to identify CV death (PPV = 36.4%), coronary heart disease mortality (PPV = 28.3%), and stroke mortality (PPV = 34.5%). CONCLUSION: Two existing algorithms based on medical claims diagnoses with or without death certificates can accurately identify SCD to support pharmacoepidemiologic studies. Developing valid algorithms identifying MI- and stroke-related death should be a research priority. PROSPERO 2017 CRD42017078745.

Validation of an algorithm to identify heart failure hospitalisations in patients with diabetes within the veterans health administration.

OBJECTIVES: We aimed to validate an algorithm using both primary discharge diagnosis (International Classification of Diseases Ninth Revision (ICD-9)) and diagnosis-related group (DRG) codes to identify hospitalisations due to decompensated heart failure (HF) in a population of patients with diabetes within the Veterans Health Administration (VHA) system. DESIGN: Validation study. SETTING: Veterans Health Administration-Tennessee Valley Healthcare System PARTICIPANTS: We identified and reviewed a stratified, random sample of hospitalisations between 2001 and 2012 within a single VHA healthcare system of adults who received regular VHA care and were initiated on an antidiabetic medication between 2001 and 2008. We sampled 500 hospitalisations; 400 hospitalisations that fulfilled algorithm criteria, 100 that did not. Of these, 497 had adequate information for inclusion. The mean patient age was 66.1 years (SD 11.4). Majority of patients were male (98.8%); 75% were white and 20% were black. PRIMARY AND SECONDARY OUTCOME MEASURES: To determine if a hospitalisation was due to HF, we performed chart abstraction using Framingham criteria as the referent standard. We calculated the positive predictive value (PPV), negative predictive value (NPV), sensitivity and specificity for the overall algorithm and each component (primary diagnosis code (ICD-9), DRG code or both). RESULTS: The algorithm had a PPV of 89.7% (95% CI 86.8 to 92.7), NPV of 93.9% (89.1 to 98.6), sensitivity of 45.1% (25.1 to 65.1) and specificity of 99.4% (99.2 to 99.6). The PPV was highest for hospitalisations that fulfilled both the ICD-9 and DRG algorithm criteria (92.1% (89.1 to 95.1)) and lowest for hospitalisations that fulfilled only DRG algorithm criteria (62.5% (28.4 to 96.6)). CONCLUSIONS: Our algorithm, which included primary discharge diagnosis and DRG codes, demonstrated excellent PPV for identification of hospitalisations due to decompensated HF among patients with diabetes in the VHA system.