Assessing the quality of cause of death data in six high-income countries: Australia, Canada, Denmark, Germany, Japan and Switzerland.

OBJECTIVES: To assess the policy utility of national cause of death (COD) data of six high-income countries with highly developed health information systems. METHODS: National COD data sets from Australia, Canada, Denmark, Germany, Japan and Switzerland for 2015 or 2016 were assessed by applying the ANACONDA software tool. Levels, patterns and distributions of unusable and insufficiently specified "garbage" codes were analysed. RESULTS: The average proportion of unusable COD was 18% across the six countries, ranging from 14% in Australia and Canada to 25% in Japan. Insufficiently specified codes accounted for a further 8% of deaths, on average, varying from 6% in Switzerland to 11% in Japan. The most commonly used garbage codes were Other ill-defined and unspecified deaths (R99), Heart failure (I50.9) and Senility (R54). CONCLUSIONS: COD certification errors are common, even in countries with very advanced health information systems, greatly reducing the policy value of mortality data. All countries should routinely provide certification training for hospital interns and raise awareness among doctors of their public health responsibility to certify deaths correctly and usefully for public health policy.

Investigating predictors of eating: is resting metabolic rate really the strongest proxy of energy intake?

Background: Evidence suggests that fat-free mass and resting metabolic rate (RMR), but not fat mass, are strong predictors of energy intake (EI). However, body composition and RMR do not explain the entire variance in EI, suggesting that other factors may contribute to this variance.Objective: We aimed to investigate the associations between body mass index (in kg/m2), fat mass, fat-free mass, and RMR with acute (1 meal) and daily (24-h) EI and between fasting appetite ratings and certain eating behavior traits with daily EI. We also evaluated whether RMR is a predictor of the error variance in acute and daily EI.Design: Data collected during the control condition of 7 studies conducted in Ottawa, Ontario, Canada, were included in these analyses (n = 191 and 55 for acute and daily EI, respectively). These data include RMR (indirect calorimetry), body composition (dual-energy X-ray absorptiometry), fasting appetite ratings (visual analog scales), eating behavior traits (Three-Factor Eating Questionnaire), and EI (food buffet or menu).Results: Fat-free mass was the best predictor of acute EI (R2 = 0.46; P < 0.0001). The combination of fasting prospective food consumption ratings and RMR was the best predictor of daily EI (R2 = 0.44; P < 0.0001). RMR was a statistically significant positive predictor of the error variance for acute (R2 = 0.20; P < 0.0001) and daily (R2 = 0.23; P < 0.0001) EI. RMR did, however, remain a statistically significant predictor of acute (R2 = 0.32; P < 0.0001) and daily (R2 = 0.30; P < 0.0001) EI after controlling for this error variance.Conclusions: Our findings suggest that combined measurements of appetite ratings and RMR could be used to estimate EI in weight-stable individuals. However, greater error variance in acute and daily EI with increasing RMR values was observed. Future studies are needed to identify whether greater fluctuations in daily EI over time occur with increasing RMR values. This trial was registered at clinicaltrials.gov as NCT02653378.