Risk-Adjusting Mortality in the Nationwide Veterans Affairs Healthcare System

Background The US Veterans Affairs (VA) healthcare system began reporting risk-adjusted mortality for intensive care (ICU) admissions in 2005. However, while the VA’s mortality model has been updated and adapted for risk-adjustment of all inpatient hospitalizations, recent model performance has not been published. We sought to assess the current performance of VA’s 4 standardized mortality models: acute care 30-day mortality (acute care SMR-30); ICU 30-day mortality (ICU SMR-30); acute care in-hospital mortality (acute care SMR); and ICU in-hospital mortality (ICU SMR). Methods Retrospective cohort study with split derivation and validation samples. Standardized mortality models were fit using derivation data, with coefficients applied to the validation sample. Nationwide VA hospitalizations that met model inclusion criteria during fiscal years 2017–2018(derivation) and 2019 (validation) were included. Model performance was evaluated using c-statistics to assess discrimination and comparison of observed versus predicted deaths to assess calibration. Results Among 1,143,351 hospitalizations eligible for the acute care SMR-30 during 2017–2019, in-hospital mortality was 1.8%, and 30-day mortality was 4.3%. C-statistics for the SMR models in validation data were 0.870 (acute care SMR-30); 0.864 (ICU SMR-30); 0.914 (acute care SMR); and 0.887 (ICU SMR). There were 16,036 deaths (4.29% mortality) in the SMR-30 validation cohort versus 17,458 predicted deaths (4.67%), reflecting 0.38% over-prediction. Across deciles of predicted risk, the absolute difference in observed versus predicted percent mortality was a mean of 0.38%, with a maximum error of 1.81% seen in the highest-risk decile. Conclusions and Relevance The VA’s SMR models, which incorporate patient physiology on presentation, are highly predictive and demonstrate good calibration both overall and across risk deciles. The current SMR models perform similarly to the initial ICU SMR model, indicating appropriate adaption and re-calibration.

Allowing for shocks in portfolio mortality models

The COVID-19 pandemic creates a challenge for actuaries analysing experience data that include mortality shocks. Without sufficient local flexibility in the time dimension, any analysis based on the most recent data will be biased by the temporarily higher mortality. Also, depending on where the shocks sit in the exposure period, any attempt to identify mortality trends will be distorted. We present a methodology for analysing portfolio mortality data that offer local flexibility in the time dimension. The approach permits the identification of seasonal variation, mortality shocks and occurred-but-not reported deaths (OBNR). The methodology also allows actuaries to measure portfolio-specific mortality improvements. Finally, the method assists actuaries in determining a representative mortality level for long-term applications like reserving and pricing, even in the presence of mortality shocks. Results are given for a mature annuity portfolio in the UK, which suggest that the Bayesian information criterion is better for actuarial model selection in this application than Akaike’s information criterion.

A COMPARISON OF PARAMETRIC MORTALITY MODELS TO GRADUATE URBAN AND RURAL MORTALITY

This study examine twenty-nine parametric mortality models and assess their suitability for graduating mortality rates of urban and rural areas in Pakistan. Grouped age specific mortality rates of rural and urban populations for the year 2019 are used. The data is collected from the website of National Institute of Population Studies which conduct Maternal Mortality Survey in Pakistan on regular basis. The parametric mortality models were applied to rural and urban mortality data. We used R software to estimate the model’s parameters and assess their suitability for urban and rural populations. The suitability of these models was assessed by using 3 different loss functions. Our analyses found that the fourth type of Heligman-Polard’s model with loss function 3 provides reliable results for graduating the mortality of rural population while second type of Carriere model with loss function 3 produce best results for graduating the urban mortality of Pakistan. Based on two models, mortality rates of urban and rural population have been graduated over age range 0-85. We suggest the use the graduated mortality rates of urban and rural areas for pricing life insurance products in rural and urban areas respectively. In addition, graduated mortality rates are also suggested for use in calculation of life insurance liabilities.

Natural population movement and COVID-19: data from Russia

The COVID-19 pandemic is highly infectious, so it paralyzed the health systems of many countries causing a high mortality rate. Official data on COVID-19 deaths at many sites are questioned, and the figures are considered several times higher than official data. In this sense, the objective of the study was to determine the impact of the COVID-19 pandemic on the natural movement of the population and, in addition, to evaluate the real mortality rate from COVID-19 in Russia from the construction of predictive mortality models. The study used data from the World Health Organization and the Statistical Service of the Federal State of Russia; se used linear and polynomial models to construct mortality models. The study revealed an underestimation of the official COVID-19 death rate by 2.4 to 6.8 times, depending on the data source. There was a sharp increase in mortality in Russia in 2020 among people over 50 years of age, and with the increase in age, mortality increased. The main reasons for the sharp increase in mortality were coronary heart disease, cerebrovascular diseases, and respiratory diseases, among others.

Common Factor Cause-Specific Mortality Model

Recent pension reforms in Europe have implemented a link between retirement age and life expectancy. The accurate forecast of life tables and life expectancy is hence paramount for governmental policy and financial institutions. We developed a multi-population mortality model which includes a cause-specific environment using Archimedean copulae to model dependence between various groups of causes of death. For this, Dutch data on cause-of-death mortality and cause-specific mortality data from 14 comparable European countries were used. We find that the inclusion of a common factor to a cause-specific mortality context increases the robustness of the forecast and we underline that cause-specific mortality forecasts foresee a more pessimistic mortality future than general mortality models. Overall, we find that this non-trivial extension is robust to the copula specification for commonly chosen dependence parameters.

Association of Sedentary and Active Bout Frequency With Mortality in Older Men Using Accelerometry

Background Time spent sedentary increases with age and has several negative health consequences. We sought to examine associations between daily sedentary and active bout frequency with all-cause mortality. Methods Data are from 2,918 men in the Osteoporotic Fractures in Men (MrOS) study (mean age at Visit 3±SD: 79.0±5.1 years) with valid activity monitor data (5.1±0.3 days worn>90%) at Year 7 visit (Visit 3, 2007-2009). Sedentary and active bout frequencies are defined as the daily transition frequency from a sedentary bout lasting 5+ minutes to activity of any intensity, and the transition frequency from an active bout lasting 5+ minutes to sedentary. Deaths were centrally adjudicated using death certificates. Cox proportional hazard models were used to examine associations between quartiles of sedentary (Q1 referent, <13.6 bouts/day) or active (Q1 referent, <5 bouts/day) bout frequency and mortality. Models were repeated, stratifying by median daily total time spent sedentary and active. Results After 9.4±3.7 years of follow-up, 1,487 (51.0%) men died. Men averaged 16.9±5.1 and 8.2±4.2 sedentary and active bouts/day, respectively. After full covariate adjustment, each quartile reflecting a higher sedentary (Q4 vs Q1 HR: 0.68, 95%CI: 0.58-0.81, p-trend<0.001) and active bout (Q4 vs Q1 HR: 0.57, 95%CI: 0.48-0.68, p-trend<0.001) frequency was associated with lower mortality risk. There was no evidence that effects differed by total sedentary time (p-interaction for sedentary bout frequency and total sedentary time>0.05). Conclusions More frequent, prolonged sedentary and active bouts are associated with a lower mortality risk in older men and is not moderated by total sedentary time.

Modeling natural mortality for different plant densities in dendroenergetic trials

Mortality was modelled for three species (Acacia melanoxylon, Eucalyptus camaldulensis, Eucalyptus nitens) at three plantation densities (5000, 7500, and 10000 trees ha-1) in an trial of biomass production for purposes of dendroenergetic. One modelling based on individual tree level and two mortality modelling alternatives were evaluated: four survival probability equations and eight difference equations. The individual tree survival modelling considered a logistic model, is a linear combination of variables to individual tree at current time and the previous time as estimator, being the main variables the variation of the competition index and the variation of basal area growth between the current growth period and the previous growth period. The survival probability alternative used state variables of the stand (age, dominant height, average square diameter) as predictors, whereas the difference equations were adjusted according to age-based changes only. The models to stand levels showed better result than individual tree models, and in general, the mortality models based on difference equations presented better indicators of precision and parsimony. The rate of relative mortality was constant, i.e., (dN/dE)/N, and varied between species, revealing greater mortality, consecutively, in E. nitens, A. melanoxylon, and E. camaldulensis. Although mortality tended to be higher at greater plantation densities, stand density did not significantly affect the parameters of the adjusted models. Highlights The mortality stand level models showed better results than the individual tree models for dendroenergetic crops, and in general, the mortality models based on difference equations presented better precision indicators and parsimony. The survival probability alternative involved state variables of the stand like age, dominant height, and average square diameter as predictors, while the difference equations were fitted according to age-based changes only. Mortality tended to be higher at greater plantation densities, however stand density did not significantly affect the parameters of the mortality equations.

A GROUP REGULARISATION APPROACH FOR CONSTRUCTING GENERALISED AGE-PERIOD-COHORT MORTALITY PROJECTION MODELS

Given the rapid reductions in human mortality observed over recent decades and the uncertainty associated with their future evolution, there have been a large number of mortality projection models proposed by actuaries and demographers in recent years. Many of these, however, suffer from being overly complex, thereby producing spurious forecasts, particularly over long horizons and for small, noisy data sets. In this paper, we exploit statistical learning tools, namely group regularisation and cross-validation, to provide a robust framework to construct discrete-time mortality models by automatically selecting the most appropriate functions to best describe and forecast particular data sets. Most importantly, this approach produces bespoke models using a trade-off between complexity (to draw as much insight as possible from limited data sets) and parsimony (to prevent over-fitting to noise), with this trade-off designed to have specific regard to the forecasting horizon of interest. This is illustrated using both empirical data from the Human Mortality Database and simulated data, using code that has been made available within a user-friendly open-source R package StMoMo.

Novel COVID-19 Mortality Rate Prediction (MRP) Model for India Using Regression Model With Optimized Hyperparameter

The main objective of this study is to estimate the future COVID-19 mortality rate for India using COVID-19 mortality rate models from different countries. Here, the regression method with the optimal hyperparameter is used to build these models. In the literature, numerous mortality models for infectious diseases have been proposed, most of which predict future mortality by extending one or more disease-related attributes or parameters. But most of these models predict mortality rates from historical data. In this paper, the Gaussian process regression model with the optimal hyperparameter is used to develop the COVID-19 mortality rate prediction (MRP) model. Five different MRP models have been built for the U.S., Italy, Germany, Japan, and India. The results show that Germany has the lowest death rate in 2000 plus COVID-19 confirmed cases. Therefore, if India follows the strategy pursued by Germany, India will control the COVID-19 mortality rate even in the increase of confirmed cases.