Comparing the Centers for Medicare and Medicaid Services (CMS) enrollment data death dates to the National Death Index (NDI)

PURPOSE

In the United States, the National Death Index (NDI) is the most complete source of death information, while epidemiologic studies with mortality outcomes often rely on U.S. Medicare data for outcome ascertainment. The purpose of this study was to assess the agreement of death information between the Centers for Medicare & Medicaid Services (CMS) Medicare enrolment data and NDI.

METHODS

Using Medicare and NDI data from 1999 through 2016, we identified Medicare beneficiaries who were reported dead in the CMS Medicare enrolment database (EDB) and Common Medicare Environment (CME), linked these beneficiaries to the NDI using CMS Health Insurance Claim number, and compared death dates between the two data sources. To assess agreement between our data sources, we calculated kappa scores; where a kappa of 1 indicates perfect agreement and a kappa of 0 indicates agreement equivalent to chance. We also examined CMS to NDI linkage and death date matching for stability over time.

RESULTS

Of the 36 785 640, Medicare beneficiaries reported dead in CMS enrollment data from 1999 to 2016, 97.5% were linked to the NDI. A kappa score of 0.98 showed a near perfect agreement between NDI and CMS reported deaths. The percentage of linked cases exactly matching on death dates increased from 94.8% in 1999 to 99.4% in 2016.

CONCLUSIONS

Our findings suggest strong concordance between death dates as recorded by CMS enrollment data and the NDI in the entire Medicare population.

Search Engines to Capture Missing Deaths From Institutional Data Warehouse

INTRODUCTION

Clinical publications use mortality as a hard end point. It is unknown how many patient deaths are under-reported in institutional databases. The objective of this study was to query mortality in our patient cohort from our data warehouse and compare these deaths to those identified in different databases.

METHODS

We passed the first/last name and date of birth of 134 patients through online mortality search engines (Find a Grave Index, US Cemetery and Funeral Home Collection, etc.) to assess their ability to capture patient deaths and compared that to deaths recorded from our institutional data warehouse.

RESULTS

Our institutional data warehouse found approximately one-third of the total patient mortalities. After the Social Security Death Index, we found that the Find a Grave Index captured the most mortalities missed by the institutional data warehouse. These results highlight the advantages of incorporating readily available search engines into institutional data warehouses for the accurate collection of patient mortalities, particularly those that occur outside of index operative admission.

CONCLUSIONS

The incorporation of the mortality search engines significantly augmented the capture of patient deaths. Our approach may be useful for tailored patient outreach and reporting mortalities with institutional data.

L-Shaped Association of Serum Chloride Level With All-Cause and Cause-Specific Mortality in American Adults: Population-Based Prospective Cohort Study

BACKGROUND

Chloride is the most abundant anion in the human extracellular fluid and plays a crucial role in maintaining homeostasis. Previous studies have demonstrated that hypochloremia can act as an independent risk factor for adverse outcomes in various clinical settings. However, the association of variances of serum chloride with long-term mortality risk in general populations has been rarely investigated.

OBJECTIVE

This study aims to assess the association of serum chloride with all-cause and cause-specific mortality in the general American adult population.

METHODS

Data were collected from 10 survey cycles (1999-2018) of the National Health and Nutrition Examination Survey. All-cause mortality, cardiovascular disease (CVD) mortality, cancer mortality, and respiratory disease mortality data were obtained by linkage to the National Death Index through December 31, 2019. After adjusting for demographic factors and relevant lifestyle, laboratory items, and comorbid factors, weighted Cox proportional risk models were constructed to estimate hazard ratios and 95% CIs for all-cause and cause-specific mortality.

RESULTS

A total of 51,060 adult participants were included, and during a median follow-up of 111 months, 7582 deaths were documented, 2388 of CVD, 1639 of cancer, and 567 of respiratory disease. The weighted Kaplan-Meier survival analyses showed consistent highest mortality risk in individuals with the lowest quartiles of serum chloride. The multivariate-adjusted hazard ratios from lowest to highest quartiles of serum chloride (≤101.2, 101.3-103.2, 103.2-105.0, and ≥105.1 mmol/L) were 1.00 (95% CI reference), 0.77 (95% CI 0.67-0.89), 0.72 (95% CI 0.63-0.82), and 0.77 (95% CI 0.65-0.90), respectively, for all-cause mortality (P for linear trend<.001); 1.00 (95% CI reference), 0.63 (95% CI 0.51-0.79), 0.56 (95% CI 0.43-0.73), and 0.67 (95% CI 0.50-0.89) for CVD mortality (P for linear trend=.004); 1.00 (95% CI reference), 0.67 (95% CI 0.54-0.84), 0.65 (95% CI 0.50-0.85), and 0.65 (95% CI 0.48-0.87) for cancer mortality (P for linear trend=.004); and 1.00 (95% CI reference), 0.68 (95% CI 0.41-1.13), 0.59 (95% CI 0.40-0.88), and 0.51 (95% CI 0.31-0.84) for respiratory disease mortality (P for linear trend=.004). The restricted cubic spline analyses revealed the nonlinear and L-shaped associations of serum chloride with all-cause and cause-specific mortality (all P for nonlinearity<.05), in which lower serum chloride was prominently associated with higher mortality risk. The associations of serum chloride with mortality risk were robust, and no significant additional interaction effect was detected for all-cause mortality and CVD mortality (P for interaction>.05).

CONCLUSIONS

In American adults, decreased serum chloride concentrations were independently associated with increased all-cause mortality, CVD mortality, cancer mortality, and respiratory disease mortality. Our findings suggested that serum chloride may serve as a promising cost-effective health indicator in the general adult population. Further studies are warranted to explore the potential pathophysiological mechanisms underlying the association between serum chloride and mortality.

Psychological distress and the risk of drug overdose death

BACKGROUND

Previous research has shown an association between psychological distress and overdose death among specific populations. However, few studies have examined this relationship in a large US population-based cohort.

METHODS

Data from the 2010-2018 NHIS were linked to mortality data from the National Death Index through 2019. Psychological distress was measured using the Kessler 6 scale. Drug overdose deaths were examined, and deaths from all other causes were included as a comparison group. Cox proportional hazards regression was used to estimate mortality risk by psychological distress level.

RESULTS

The study population included 272,561 adults. Adjusting for demographic covariates and using no psychological distress as the reference, distress level was positively associated with the risk of overdose death: low (HR = 1.8, 95 % CI = 1.1-2.8), moderate (HR = 4.1, 95 % CI = 2.5-6.7), high (HR = 10.3, 95 % CI = 6.5-16.1). A similar pattern was observed for deaths from all other causes: low (HR = 1.2, 95 % CI = 1.1-1.2), moderate (HR = 1.9, 95 % CI = 1.7-2.0), high (HR = 2.6, 95 % CI = 2.4-2.8).

LIMITATIONS

Limited substance use information prevented adjustment for this potentially important covariate.

DISCUSSION

Adults with psychological distress were at greater risk of drug overdose death, relative to those without psychological distress. Adults with psychological distress were also at increased risk of death due to other causes, though the association was not as strong.

Ascertaining Death Events in a Pragmatic Clinical Trial: Insights From the TRANSFORM-HF Trial

BACKGROUND

Death ascertainment can be challenging for pragmatic clinical trials that limit site follow-up activities to usual clinical care.

METHODS AND RESULTS

We used blinded aggregate data from the ongoing ToRsemide comparison with furoSemide FOR Management of Heart Failure (TRANSFORM-HF) pragmatic clinical trial in patients with heart failure to evaluate the agreement between centralized call center death event identification and the United States National Death Index (NDI). Of 2284 total patients randomized through April 12, 2021, 1480 were randomized in 2018-2019 and 804 in 2020-2021. The call center identified 416 total death events (177 in 2018-2019 and 239 in 2020-2021). The NDI 2018-2019 final file identified 178 death events, 165 of which were also identified by the call center. The study's inter-rater reliability metric (Cohen's kappa coefficient, 0.920; 95% confidence interval, 0.889-0.951) demonstrates a high level of agreement. The time between a death event and its identification was less for the call center (median, 47 days; interquartile range, 11-103 days) than for the NDI (median, 270 days; interquartile range, 186-391 days).

CONCLUSIONS

There is substantial agreement between deaths identified by a centralized call center and the NDI. However, the time between a death event and its identification is significantly less for the call center.

A methodological assessment of privacy preserving record linkage using survey and administrative data

BACKGROUND

The National Center for Health Statistics (NCHS) links data from surveys to administrative data sources, but privacy concerns make accessing new data sources difficult. Privacy-preserving record linkage (PPRL) is an alternative to traditional linkage approaches that may overcome this barrier. However, prior to implementing PPRL techniques it is important to understand their effect on data quality.

METHODS

Results from PPRL were compared to results from an established linkage method, which uses unencrypted (plain text) identifiers and both deterministic and probabilistic techniques. The established method was used as the gold standard. Links performed with PPRL were evaluated for precision and recall. An initial assessment and a refined approach were implemented. The impact of PPRL on secondary data analysis, including match and mortality rates, was assessed.

RESULTS

The match rates for all approaches were similar, 5.1% for the gold standard, 5.4% for the initial PPRL and 5.0% for the refined PPRL approach. Precision ranged from 93.8% to 98.9% and recall ranged from 98.7% to 97.8%, depending on the selection of tokens from PPRL. The impact of PPRL on secondary data analysis was minimal.

DISCUSSION

The findings suggest PPRL works well to link patient records to the National Death Index (NDI) since both sources have a high level of non-missing personally identifiable information, especially among adults 65 and older who may also have a higher likelihood of linking to the NDI.

CONCLUSION

The results from this study are encouraging for first steps for a statistical agency in the implementation of PPRL approaches, however, future research is still needed.

A Proinflammatory Diet May Increase Mortality Risk in Patients with Diabetes Mellitus

This was an observational study based on the National Health and Nutrition Examination Survey (NHANES) and National Death Index (NDI) 2009–2014 which aimed to validate whether a proinflammatory diet may increase mortality risk in patients with diabetes mellitus. Dietary inflammatory potential was assessed by dietary inflammatory index (DII) based on 24 h dietary recall. Mortality follow-up information was accessed from NDI, which was then merged with NHANES data following the National Center for Health Statistics (NCHS) protocols. For 15,291 participants from the general population, the average DII was 0.37 ± 1.76 and the prevalence rate of diabetes was 13.26%. DII was positively associated with fasting glucose (β = 0.83, 95% CI: 0.30, 1.36, p = 0.0022), glycohemoglobin (β = 0.02, 95% CI: 0.01, 0.03, p = 0.0009), and the risk of diabetes (OR = 1.05, 95% CI: 1.01, 1.09, p = 0.0139). For 1904 participants with diabetes and a median follow-up of 45 person-months, a total of 178 participants with diabetes died from all causes (mortality rate = 9.34%). People with diabetes who adhered to a proinflammatory diet showed a higher risk of all-cause mortality (HR = 1.71, 95%CI: 1.13, 2.58, p = 0.0108). In summary, DII was positively associated with diabetes prevalence and a proinflammatory diet may increase mortality risk in patients with diabetes mellitus.

Using Synthetic Data to Replace Linkage Derived Elements: A Case Study

While record linkage can expand analyses performable from survey microdata, it also incurs greater risk of privacy-encroaching disclosure. One way to mitigate this risk is to replace some of the information added through linkage with synthetic data elements. This paper describes a case study using the National Hospital Care Survey (NHCS), which collects patient records under a pledge of protecting patient privacy from a sample of U.S. hospitals for statistical analysis purposes. The NHCS data were linked to the National Death Index (NDI) to enhance the survey with mortality information. The added information from NDI linkage enables survival analyses related to hospitalization, but as the death information includes dates of death and detailed causes of death, having it joined with the patient records increases the risk of patient re-identification (albeit only for deceased persons). For this reason, an approach was tested to develop synthetic data that uses models from survival analysis to replace vital status and actual dates-of-death with synthetic values and uses classification tree analysis to replace actual causes of death with synthesized causes of death. The degree to which analyses performed on the synthetic data replicate results from analysis on the actual data is measured by comparing survival analysis parameter estimates from both data files. Because synthetic data only have value to the degree that they can be used to produce statistical estimates that are like those based on the actual data, this evaluation is an essential first step in assessing the potential utility of synthetic mortality data.

The National Death Index as a Source of Homicide Data: A Methodological Exposition of Promises and Pitfalls for Criminologists

Criminologists largely rely on national de-identified data sources to study homicide in the United States. The National Death Index (NDI), a comprehensive and well-established database compiled by the National Center for Health Statistics, is an untapped source of homicide data that offers identifiable linkages to other data sources while retaining national coverage. This study's five aims follow. First, we review the data sources in articles published in Homicide Studies over the past decade. Second, we describe the NDI, including its origins, procedures, and uses. Third, we outline the procedures for linking a police gang intelligence database to the NDI. Fourth, we introduce the St. Louis Gang Member-Linked Mortality Files database, which is composed of 3,120 police-identified male gang members in the St. Louis area linked to NDI records. Finally, we report on preliminary cause-of-death findings. We conclude by outlining the benefits and drawbacks of the NDI as a source of homicide data for criminologists.

Exceptional mortality risk among police-identified young black male gang members

Gang membership is associated with many risky behaviors but is often overlooked as a source of mortality among young Americans. Gang Member-Linked Mortality Files (GM-LMFs) match St. Louis, Missouri gang members listed in a law enforcement gang database to mortality records in the National Death Index. We created three analytic samples composed of black males aged 15-35 years by merging cases of the GM-LMFs with National Vital Statistics System and Census data in years 1993-2016. Mortality rates standardized to the 15-35-year-old 2010 U.S. male population were estimated for all-cause (1477.4, 99% CI = 1451.5-1503.3), homicide (950.1, 99% CI = 932.2-967.9), non-homicide injury (314.0, 99% CI = 308.8-319.2), and non-injury (213.3, 99% CI = 202.3-224.4) deaths in the GM-LMFs. We fitted Poisson rate models to estimate mortality rate ratios (RR) between gang members and demographically-matched comparison groups. Black male gang members in St. Louis were at an elevated mortality risk from all causes of death, and homicides contributed substantially to this risk. Compared to black males in St. Louis, gang members experienced greater relative risk of all-cause (RR = 2.9, 99% CI = 2.4-3.5), homicide (RR = 3.2, 99% CI = 2.5-4.1), and non-homicide injury (RR = 4.0, 99% CI = 2.8-5.8) mortality between 1993 and 2016. Relative risk was greater when compared to black males in St. Louis MSA, Missouri, and the USA. These results identify a key source of excess mortality among young black Americans. Health policies and interventions may be most efficacious when they acknowledge, address, and incorporate information about and target high-risk populations, including gang members, that contribute to relatively high mortality risk in the USA.

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.

Confirming mortality in a longitudinal exposure cohort: optimizing National Death Index search result processing

PURPOSE

The National Death Index (NDI) is an important resource for mortality ascertainment. Methods selected to process NDI search results are rarely described in studies using linked data and can have an impact on resources and mortality ascertainment. We evaluate methods to process NDI search results among a 9/11-exposed cohort-the World Trade Center Health Registry (Registry).

METHODS

We describe three approaches to process search results (NDI-recommended cutoff points [NDIc]; National Program of Cancer Registries [NPCR] algorithm, and modified National Institute of Occupational Safety and Health algorithm [mNIOSH]). We calculate percent agreement, positive predictive value, sensitivity, specificity, and quantify the burden of manual review to compare the approaches.

RESULTS

Of 51,158 Registry enrollees submitted for linkage, 9449 enrollee-level and 17,909 record-level matches were identified. NPCR and mNIOSH were highly concordant (97.1%); more record pairs required manual review for mNIOSH (mNIOSH: 2.7% and NPCR: 1.8%). NDIc sensitivity was 82.9%, with differences observed by race and ethnicity (Asian: 74.4% and White: 86.1%).

CONCLUSIONS

NPCR algorithm minimized false matches and reduced the manual review burden. NDIc had nonrandom distribution of missed matches and low sensitivity. NDI search processing methods have important implications for resulting linked data; measures of linkage quality should be available to data users.

Comparability of Mortality Estimates from Social Surveys and Vital Statistics Data in the United States

Social surveys prospectively linked with death records provide invaluable opportunities for the study of the relationship between social and economic circumstances and mortality. Although survey-linked mortality files play a prominent role in U.S. health disparities research, it is unclear how well mortality estimates from these datasets align with one another and whether they are comparable with U.S. vital statistics data. We conduct the first study that systematically compares mortality estimates from several widely-used survey-linked mortality files and U.S. vital statistics data. Our results show that mortality rates and life expectancies from the National Health Interview Survey Linked Mortality Files, Health and Retirement Study, Americans' Changing Lives study, and U.S. vital statistics data are similar. Mortality rates are slightly lower and life expectancies are slightly higher in these linked datasets relative to vital statistics data. Compared with vital statistics and other survey-linked datasets, General Social Survey-National Death Index life expectancy estimates are much lower at younger adult ages and much higher at older adult ages. Cox proportional hazard models regressing all-cause mortality risk on age, gender, race, educational attainment, and marital status conceal the issues with the General Social Survey-National Death Index that are observed in our comparison of absolute measures of mortality risk. We provide recommendations for researchers who use survey-linked mortality files.

service members

PURPOSE

Accurate knowledge of the vital status of individuals is critical to the validity of mortality research. National Death Index (NDI) and NDI-Plus are comprehensive epidemiological resources for mortality ascertainment and cause of death data that require additional user validation. Currently, there is a gap in methods to guide validation of NDI search results rendered for active duty service members. The purpose of this research was to adapt and evaluate the CDC National Program of Cancer Registries (NPCR) algorithm for mortality ascertainment in a large military cohort.

METHODS

We adapted and applied the NPCR algorithm to a cohort of 7088 service members on active duty at the time of death at some point between 2001 and 2009. We evaluated NDI validity and NDI-Plus diagnostic agreement against the Department of Defense's Armed Forces Medical Examiner System (AFMES).

RESULTS

The overall sensitivity of the NDI to AFMES records after the application of the NPCR algorithm was 97.1%. Diagnostic estimates of measurement agreement between the NDI-Plus and the AFMES cause of death groups were high.

CONCLUSIONS

The NDI and NDI-Plus can be successfully used with the NPCR algorithm to identify mortality and cause of death among active duty military cohort members who die in the United States.

Comparison of Expert Adjudicated Coronary Heart Disease and Cardiovascular Disease Mortality With the National Death Index: Results From the REasons for Geographic And Racial Differences in Stroke (REGARDS) Study

BACKGROUND

The National Death Index (NDI) is widely used to detect coronary heart disease (CHD) and cardiovascular disease (CVD) deaths, but its reliability has not been examined recently.

METHODS AND RESULTS

We compared CHD and CVD deaths detected by NDI with expert adjudication of 4010 deaths that occurred between 2003 and 2013 among participants in the REGARDS (REasons for Geographic And Racial Differences in Stroke) cohort of black and white adults in the United States. NDI derived CHD mortality had sensitivity 53.6%, specificity 90.3%, positive predictive value 54.2%, and negative predictive value 90.1%. NDI-derived CVD mortality had sensitivity 73.4%, specificity 84.5%, positive predictive value 70.6%, and negative predictive value 86.2%. Among NDI-derived CHD and CVD deaths, older age (odds ratios, 1.06 and 1.04 per 1-year increase) was associated with a higher probability of disagreement with the adjudicated cause of death, whereas among REGARDS adjudicated CHD and CVD deaths a history of CHD or CVD was associated with a lower probability of disagreement with the NDI-derived causes of death (odds ratios, 0.59 and 0.67, respectively).

CONCLUSIONS

The modest accuracy and differential performance of NDI-derived cause of death may impact CHD and CVD mortality statistics.

Observed long-term mortality after 18,000 person-years among survivors in a large regional drowning registry

AIM

Long-term outcomes beyond one year after non-fatal drowning are uncharacterized. We estimated long-term mortality and identified prognostic factors in a large, population-based cohort.

METHODS

Population-based prospective cohort study (1974-1996) of Western Washington Drowning Registry (WWDR) subjects surviving the index drowning through hospital discharge. Primary outcome was all-cause mortality through 2012. We tabulated Utstein-style exposure variables, estimated Kaplan-Meier curves, and identified prognostic factors with Cox proportional hazard modeling. We also compared 5-, 10-, and 15-year mortality estimates of the primary cohort to age-specific mortality estimates from United States Life Tables.

RESULTS

Of 2824 WWDR cases, 776 subjects (5[IQR 2-17] years, 68% male) were included. Only 63 (8%) non-fatal drowning subjects died during 18,331 person-years of follow-up. Long-term mortality differed by Utstein variables (age, precipitating alcohol use, submersion interval, GCS, CPR, intubation, defibrillation, initial vital signs, neurologic status at hospital discharge) and inpatient markers of illness severity (mechanical ventilation, vasopressor use, seizure, pneumothorax). Survival differed by age (HR 1.04;95%CI 1.03-1.05), drowning-related cardiac arrest (HR 3.47;95%CI 1.97-6.13), and neurologic impairment at hospital discharge (HR 5.10;95% CI 2.70-9.62). In adjusted analysis, age (HR 1.05;95%CI 1.03-1.06) and severe neurologic impairment at discharge (HR 2.31;95%CI 1.01-5.28) were associated with long-term mortality. Subjects aged 5-15 years had higher mortality risks than those calculated from Life Tables.

CONCLUSION

Most drownings were fatal, but survivors of non-fatal drowning had low risk of subsequent long-term mortality similar to the general population that was independently associated with age and neurologic status at hospital discharge.

Record matching between the National Hospital Care Survey and the National Death Index

Linking the National Hospital Care Survey (NHCS) with the National Death Index (NDI) provides information on the outcomes of hospitalizations and allows for analysis of individual and provider characteristics associated with in-hospital and post-discharge mortality. We test the viability of confirming hospital mortality through the linkage of preliminary 2011 NHCS data for "known dead" inpatient discharges (i.e., patients that died during a hospitalization) with the NDI, assessing the true match rate and the quality of the match. We then expand the analysis to identify patients with a 30-, 60-, and 90-day post-discharge mortality. The true match rate for the "known dead" is 94 percent.