Understanding bias when estimating life expectancy from age at death: a simulation approach applied to Morquio syndrome A

Objective Life expectancy can be estimated accurately from a cohort of individuals born in the same year and followed from birth to death. However, due to the resource-consuming nature of following a cohort prospectively, life expectancy is often assessed based upon retrospective death record reviews. This conventional approach may lead to potentially biased estimates, in particular when estimating life expectancy of rare diseases such as Morquio syndrome A. We investigated the accuracy of life expectancy estimation using death records by simulating the survival of individuals with Morquio syndrome A under four different scenarios. Results When life expectancy was constant during the entire period, using death data did not result in a biased estimate. However, when life expectancy increased over time, as is often expected to be the case in rare diseases, using only death data led to a substantial underestimation of life expectancy. We emphasize that it is therefore crucial to understand how estimates of life expectancy are obtained, to interpret them in an appropriate context, and to assess estimation methods within a sensitivity analysis framework, similar to the simulations performed herein.

Changing effect of the numerator–denominator bias in unlinked data on mortality differentials by education: evidence from Estonia, 2000–2015

BackgroundThis study highlights changing disagreement between census and death record information in the reporting of the education of the deceased and shows how these reporting differences influence a range of mortality inequality estimates.MethodsThis study uses a census-linked mortality data set for Estonia for the periods 2000–2003 and 2012–2015. The information on the education of the deceased was drawn from both the censuses and death records. Range-type, Gini-type and regression-based measures were applied to measure absolute and relative mortality inequality according to the two types of data on the education of the deceased.ResultsThe study found a small effect of the numerator–denominator bias on unlinked mortality estimates for the period 2000–2003. The effect of this bias became sizeable in the period 2012–2015: in high education group, mortality was overestimated by 23–28%, whereas the middle education group showed notable underestimation of mortality. The same effect was small for the lowest education group. These biases led to substantial distortions in range-type inequality measures, whereas unlinked and linked Gini-type measures showed somewhat closer agreement.ConclusionsThe changing distortions in the unlinked estimates reported in this study warn that this type of evidence cannot be readily used for monitoring changes in mortality inequalities.

Accuracy of identifying incident stroke cases from linked health care data in UK Biobank

ObjectiveIn UK Biobank (UKB), a large population-based prospective study, cases of many diseases are ascertained through linkage to routinely collected, coded national health datasets. We assessed the accuracy of these for identifying incident strokes.MethodsIn a regional UKB subpopulation (n = 17,249), we identified all participants with ≥1 code signifying a first stroke after recruitment (incident stroke-coded cases) in linked hospital admission, primary care, or death record data. Stroke physicians reviewed their full electronic patient records (EPRs) and generated reference standard diagnoses. We evaluated the number and proportion of cases that were true-positives (i.e., positive predictive value [PPV]) for all codes combined and by code source and type.ResultsOf 232 incident stroke-coded cases, 97% had EPR information available. Data sources were 30% hospital admission only, 39% primary care only, 28% hospital and primary care, and 3% death records only. While 42% of cases were coded as unspecified stroke type, review of EPRs enabled a pathologic type to be assigned in >99%. PPVs (95% confidence intervals) were 79% (73%–84%) for any stroke (89% for hospital admission codes, 80% for primary care codes) and 83% (74%–90%) for ischemic stroke. PPVs for small numbers of death record and hemorrhagic stroke codes were low but imprecise.ConclusionsStroke and ischemic stroke cases in UKB can be ascertained through linked health datasets with sufficient accuracy for many research studies. Further work is needed to understand the accuracy of death record and hemorrhagic stroke codes and to develop scalable approaches for better identifying stroke types.

Evaluation of Oklahoma’s Electronic Death Registration System and Event Fatality Markers for Disaster-Related Mortality Surveillance – Oklahoma USA, May 2013

Introduction:Official counts of deaths attributed to disasters are often under-reported, thus adversely affecting public health messaging designed to prevent further mortality. During the Oklahoma (USA) May 2013 tornadoes, Oklahoma State Health Department Division of Vital Records (VR; Oklahoma City, Oklahoma USA) piloted a flagging procedure to track tornado-attributed deaths within its Electronic Death Registration System (EDRS). To determine if the EDRS was capturing all tornado-attributed deaths, the Centers for Disease Control and Prevention (CDC; Atlanta, Georgia USA) evaluated three event fatality markers (EFM), which are used to collate information about deaths for immediate response and retrospective research efforts.Methods:Oklahoma identified 48 tornado-attributed deaths through a retrospective review of hospital morbidity and mortality records. The Centers for Disease Control and Prevention (CDC; Atlanta, Georgia USA) analyzed the sensitivity, timeliness, and validity for three EFMs, which included: (1) a tornado-specific flag on the death record; (2) a tornado-related term in the death certificate; and (3) X37, the International Classification of Diseases, 10th Revision (ICD-10) code in the death record for Victim of a Cataclysmic Storm, which includes tornadoes.Results:The flag was the most sensitive EFM (89.6%; 43/48), followed by the tornado term (75.0%; 36/48), and the X37 code (56.2%; 27/48). The most-timely EFM was the flag, which took 2.0 median days to report (range 0-10 days), followed by the tornado term (median 3.5 days; range 1-21), and the X37 code (median >10 days; range 2-122). Over one-half (52.1%; 25/48) of the tornado-attributed deaths were missing at least one EFM. Twenty-six percent (11/43) of flagged records had no tornado term, and 44.1% (19/43) had no X37 code. Eleven percent (4/36) of records with a tornado term did not have a flag.Conclusion:The tornado-specific flag was the most sensitive and timely EFM. Using the flag to collate death records and identify additional deaths without the tornado term and X37 code may improve immediate response and retrospective investigations. Moreover, each of the EFMs can serve as quality controls for the others to maximize capture of all disaster-attributed deaths from vital statistics records in the EDRS.Issa AN, Baker K, Pate D, Law R, Bayleyegn T, Noe RS. Evaluation of Oklahoma’s Electronic Death Registration System and event fatality markers for disaster-related mortality surveillance – Oklahoma USA, May 2013. Prehosp Disaster Med. 2019;34(2):125–131

Characterizing Fentanyl-Associated Mortality using the Literal Causes of Death

ObjectiveTo characterize fentanyl-associated mortality in Florida using freetext queries of the literal causes of death listed on death certificates.IntroductionIn October 2015, the Centers for Disease Control and Prevention(CDC) released health advisory #384 to inform people about increasesin fentanyl fatalities. Florida’s statewide syndromic surveillancesystem, Electronic Surveillance System for the Early Notification ofCommunity-based Epidemics (ESSENCE-FL), captures electronicdeath record data in near real time which allows for the monitoringof mortality trends across the state. One limitation of using deathrecord data for fentanyl surveillance is the lack of a fentanyl-specificoverdose ICD-10 code; however, the literal cause of death fields(“literals”) provide a level of detail that is rich enough to capturementions of fentanyl use. The “literals” are a free text field on thedeath certificate, recorded by a physician at the time of death anddetail the factors that led to the death. ESSENCE-FL has the benefitof not only receiving death record data in near real-time, but alsoreceiving the literal cause of death fields. This work analyzes trendsin fentanyl-associated mortality in Florida over time by using theliteral cause of death fields within death records data obtained fromESSENCE-FL.MethodsThe “literals” elements of Florida Vital Statistics mortality datafrom 2010 through 2015 accessed via ESSENCE-FL were queriedfor the term ^fent^. No necessary negations or extra term inclusionswere deemed necessary after looking at the records pulled with ^fent^alone. Deaths were analyzed by various demographic and geographicvariables to characterize this population in order to assess whichgroups are most heavily burdened by fentanyl-associated mortality.Population estimates by county for 2015 were obtained from the U.S.Census Bureau to calculate mortality rates. Language processing in RStudio was used to determine which other substances were commonlyreported when fentanyl was listed on the death certificate, in order toassess polydrug use and its impact on increased mortality.ResultsCompared to the number of fentanyl-associated mortalities in 2010(82), fentanyl-associated mortality in 2015 (599) was 6.5 times higherafter controlling for the natural increase in total mortality between2010 and 2015. Almost three-fourths of the deaths in 2015 were male(73%), which is higher than the proportion of male deaths in 2010(55%). The age group with the largest burden of fentanyl-associatedmortality was the 30 – 39 age group, with almost one-third of thedeaths in 2015 coming from this age group (31%) compared to only10% in 2010, a roughly 200% increase. Fentanyl-associated mortalitywas almost exclusive to people that are Caucasian, with 94% of thefentanyl-associated mortalities in 2015 occurring among Caucasians.Multi-drug use was also identified for those with fentanyl-associatedmortality. Mentions of other drugs were present in at least 10% of thedeaths. Some of the other drugs mentioned in the “literals” includedheroin, cocaine, and alprazolam. There was county variation in thenumber of fentanyl morality deaths ranging from 21.19 deaths per100,000 to 0.29 deaths per 100,000 residents. Two counties with thehighest rates were located adjacent to one another.ConclusionsHaving death record data readily available within the statesyndromic surveillance system is beneficial for rapid analysisof mortality trends and the analytic methods used for syndromicsurveillance can be applied to mortality data. Free text querying ofthe “literals” in the vital statistics death records data allowed forsurveillance of fentanyl-associated mortality, similar to methods usedfor querying emergency department chief complaint data. Althoughunderlying ICD-10 codes can lack detail about certain causes ofdeath, the “literals” provide a clearer picture as to what caused thedeath. The “literals” also make it possible to look at potential drugcombinations that may have increased risk of mortality, which willbe explored more thoroughly. Further work will explore other datasources for fentanyl usage and mortality trends, as well as examinepotential risk factors and confounders.

Hip Fracture-Related Mortality among Older Adults in the United States: Analysis of the CDC WONDER Multiple Cause of Death Data, 1999–2013

Objectives. To examine trends in hip fracture-related mortality among older adults in the United States between 1999 and 2013. Material and Methods. The Wide-Ranging Online Data for Epidemiological Research system was used to identify adults aged 65 years and older with a diagnosis of hip fracture reported in their multiple cause of death record. Joinpoint regression analyses were performed to estimate the average annual percent change in hip fracture-related mortality rates by selected characteristics. Results. A total of 204,254 older decedents listed a diagnosis of hip fracture on their death record. After age adjustment, hip fracture mortality rates decreased by −2.3% (95% CI, −2.7%, and −1.8%) in men and −1.5% (95% CI, −1.9%, and −1.1%) in women. Similarly, the proportion of in-hospital hip fracture deaths decreased annually by −2.1% (95% CI, −2.6%, and −1.5%). Of relevance, the proportion of cardiovascular diseases reported as the underlying cause of death decreased on average by −4.8% (95% CI, −5.5%, and −4.1%). Conclusions. Hip fracture-related mortality decreased among older adults in the United States. Downward trends in hip fracture-related mortality were predominantly attributed to decreased deaths among men and during hospitalization. Moreover, improvements in survival of hip fracture patients with greater number of comorbidities may have accounted for the present findings.

The growing contribution of hepatitis C virus infection to liver-related mortality in Scotland

The large number of individuals in Scotland who became infected with the hepatitis C virus (HCV) in the 1970s and 1980s leads us to expect liver-related morbidity and mortality to increase in the coming years. We investigated the contribution of HCV to liver-related mortality in the period January 1991 to June 2006. The study population consisted of 26,861 individuals whose death record mentioned a liver-related cause (underlying or contributing). Record-linkage to the national HCV Diagnosis database supplied HCV-diagnosed status for the study population. The proportion diagnosed with HCV among people dying from a liver-related cause rose from 2.8% (1995-1997) to 4.4% (2004-June 2006); the largest increase occurred in those aged 35-44 years at death (7% to 17%). Among all deaths from a liver-related cause, an HCV-positive diagnosis was more likely in those who died in 2001 or later than those who died in 1995-1997 (2001-2003: odds ratio=1.4, 95% confidence interval: 1.1-1.7; 2004-June 2006: 1.6, 1.3-2.0), and in those who died at under 55 compared with at least 55 years of age. HCV infection represents a significant, growing, public health burden in Scotland in terms of early deaths from liver disease.