Evaluation of stability of directly standardized rates for sparse data using simulation methods

• DOI: 10.1186/s12963-018-0177-1
• Publication Date: 2018-12-21T20:33:24Z
• Authors: Joan K. Morris; Joachim Tan; Paul Fryers; Jonathan Bestwick
• Keywords: Research; Computer applications to medicine. Medical informatics; R858-859.7; Reference Standards; 01 natural sciences; Dobson; Tiwari; Age Distribution; England; Data Interpretation, Statistical; Direct standardization; Confidence Intervals; Humans; Confidence interval coverage; Poisson Distribution; Public aspects of medicine; RA1-1270; 0101 mathematics; Epidemiologic Methods; Monte Carlo Method; Monte Carlo simulation

Directly standardized rates (DSRs) adjust for different age distributions in populations and enable, say, the of disease between to be directly compared. They are routinely published but there is concern that a DSR not valid when it based on "small" number events. The aim this study was determine value at which should analyzing real data England. Standard Monte Carlo simulation techniques were used assuming events 19 groups (i.e., 0–4, 5–9, ... 90+ years) follow independent Poisson distributions. total events, specific risks, population sizes each group varied. For 10,000 simulations (using 2013 European Population weights), together with coverage three methods (normal approximation, Dobson, Tiwari modified gamma) estimating 95% confidence intervals (CIs), calculated. normal approximation was, as expected, suitable use fewer than 100 occurred. method Dobson calculating produced similar estimates either expected or observed numbers 10 greater. accuracy CIs influenced by distribution across categories degree clustering, sampling populations, no occurring them). DSRs given less 10. might considered preferred due formulae being simpler slightly more accurate.