Time series analysis of injuries

Rising gasoline prices increase new motorcycle sales and fatalities

Background

We examined whether sales of new motorcycles was a mechanism to explain the relationship between motorcycle fatalities and gasoline prices.

Methods

The data came from the Motorcycle Industry Council, Energy Information Administration and Fatality Analysis Reporting System for 1984–2009. Autoregressive integrated moving average (ARIMA) regressions estimated the effect of inflation-adjusted gasoline price on motorcycle sales and logistic regressions estimated odds ratios (ORs) between new and old motorcycle fatalities when gasoline prices increase.

Results

New motorcycle sales were positively correlated with gasoline prices (r = 0.78) and new motorcycle fatalities (r = 0.92). ARIMA analysis estimated that a US$1 increase in gasoline prices would result in 295,000 new motorcycle sales and, consequently, 233 new motorcycle fatalities. Compared to crashes on older motorcycle models, those on new motorcycles were more likely to be young riders, occur in the afternoon, in clear weather, with a large engine displacement, and without alcohol involvement. Riders on new motorcycles were more likely to be in fatal crashes relative to older motorcycles (OR 1.14, 95 % confidence interval (CI) 1.02–1.28) when gasoline prices increase.

Conclusions

Our findings suggest that, in response to increasing gasoline prices, people tend to purchase new motorcycles, and this is accompanied with significantly increased crash risk. There are several policy mechanisms that can be used to lower the risk of motorcycle crash injuries through the mechanism of gas prices and motorcycle sales such as raising awareness of motorcycling risks, enhancing licensing and testing requirements, limiting motorcycle power-to-weight ratios for inexperienced riders, and developing mandatory training programs for new riders.

Electronic supplementary material

The online version of this article (doi:10.1186/s40621-015-0054-3) contains supplementary material, which is available to authorized users.

Major Trauma in Pregnant Women During the Summer

BACKGROUND

Pregnant women represent a major challenge in trauma care because of the risks to both mother and child and because of the difficulties in following standard protocols.

METHODS

We analyzed data for all pregnant women admitted to the hospital in Canada over 7 years to test whether major trauma still clustered in the summer despite their aversion toward alcohol, recklessness, and extreme sports.

RESULTS

A total of 2,618 pregnant women sustained major trauma. The prevalence of pregnancy was marginally lower in summer than in winter (decrease, 3%; 95% confidence interval, 2-4%), whereas the incidence of major trauma in pregnant women was significantly higher in summer than in winter (increase, 12%; 95% confidence interval, 3-21%; p = 0.005). No evidence of offsetting decreases in severity appeared in analyses of length of stay, number of surgical procedures, or mortality.

CONCLUSION

We suggest that normal lifestyle choices contribute to an increased risk of major trauma during pregnancy and merit greater awareness throughout the year.

Evaluating Disease Management Program Effectiveness: An Introduction to Time-Series Analysis

Currently, the most widely used method in the disease management (DM) industry for evaluating program effectiveness is referred to as the "total population approach." This model is a pretest-posttest design, with the most basic limitation being that without a control group, there may be sources of bias and/or competing extraneous confounding factors that offer a plausible rationale explaining the change from baseline. Furthermore, with the current inclination of DM programs to use financial indicators rather than program-specific utilization indicators as the principal measure of program success, additional biases are introduced that may cloud evaluation results. This paper presents a non-technical introduction to time-series analysis (using disease-specific utilization measures) as an alternative, and more appropriate, approach to evaluating DM program effectiveness than the current total population approach.

Secular trends of motor vehicle mortality in the United States, 1910-1994

To examine the secular trends of mortality from motor vehicle crashes, the authors compiled annual population and mortality data for the United States from 1910 to 1994 and performed an age-period-cohort analysis through graphical presentation, median polish, and Poisson regression modeling. During the 85-year study period, death rates from motor vehicle crashes showed two peaks, first in 1935-39 and then in 1965-69. Age and period effects accounted for 94% of the variation in motor vehicle mortality in men and 84% of the variation in women. Age patterns of motor vehicle mortality varied greatly with birth cohorts: for those who were born before 1910, death rates increased with age; for those born after 1910, death rates peaked at age 20-24 years for men and at age 15-19 years for women. A crossover characterized by a downward trend in death rates among the elderly and an upward trend among adolescents and young adults was observed in both sexes. The complex age, period, and cohort patterns of motor vehicle mortality are likely to have been shaped by changes in traffic patterns and driver behavior, and by improvements in safety design and emergency medical service systems.

Monitoring epidemiologic surveillance data using hidden Markov models

The analysis of routinely collected surveillance data is an important challenge in public health practice. We present a method based on a hidden Markov model for monitoring such time series. The model characterizes the sequence of measurements by assuming that its probability density function depends on the state of an underlying Markov chain. The parameter vector includes distribution parameters and transition probabilities between the states. Maximum likelihood estimates are obtained with a modified EM algorithm. Extensions are provided to take into account trend and seasonality in the data. The method is demonstrated on two examples: the first seeks to characterize influenza-like illness incidence rates with a mixture of Gaussian distributions, and the other, poliomyelitis counts with mixture of Poisson distributions. The results justify a wider use of this method for analysing surveillance data.

An evaluation of the general effect of the New Zealand graduated driver licensing system on motorcycle traffic crash hospitalisations

In 1987, New Zealand introduced a comprehensive Graduated Driver Licensing System (GDLS) which was designed to allow new drivers to gain driving experience while being excluded from high risk situations. This study sought to evaluate the impact of the GDLS on motorcycle traffic crashes that resulted in serious injury. Injury crash data were obtained from the New Zealand Health Information Services national public hospital inpatient data files for the years 1978-1994, inclusive. Cases were disaggregated into three age groups, 15-19 years, 20-24 years, and 25 years or older, for which injury trends were obtained. In order to determine whether trends in motorcycle traffic crashes simply followed national trends in other injury events, two non-traffic comparison groups for the main target group (15-19 years) were included. Using time series analyses, the introduction of the GDLS was found to be closely followed by a significant reduction (22%) in motorcycle traffic crash hospitalizations for the 15-19 year age group. An examination of vehicle registration and driver licensure data suggests that the reduction in injury crashes may, largely, be attributable to an overall reduction in exposure to motorcycle riding.

Box-Jenkins modelling in medical research

Notifications of diseases, entries in a hospital, injuries due to accidents, etc., are frequently collected in fixed equally spaced intervals. Such observations are likely to be dependent. In environmental medicine, where series such as daily concentrations of pollutants are collected and analysed, it is evident that dependence of consecutive measurements may be important. A high concentration of a pollutant today has a certain 'inertia', i.e. a tendency to be high tomorrow as well. Dependence of consecutive observations may be equally important when data such as blood glucose are recorded within a single patient. ARIMA models (autoregressive integrated moving average models, Box-Jenkins models), which allow the stochastic dependence of consecutive data to be modelled, have become well established in such fields as economics. This article reviews basic concepts of Box-Jenkins modelling. The methods are illustrated by applications. In particular, the following topics are presented: the ARIMA model, transfer function models (assessment of relations between time series) and intervention analysis (assessment of changes of time series).

Forecasting areawide hospital utilization: a comparison of five univariate time series techniques

Time series analysis is one of the methods health services researchers, managers and planners have to examine and predict utilization over time. The focus of this study is univariate time series techniques, which model the change in a dependent variable over time, using time as the only independent variable. These techniques can be used with administrative healthcare databases, which typically contain reliable, time-specific utilization variables, but may lack adequate numbers of variables needed for behavioral or economic modeling. The inpatient discharge database of the Department of Veterans Affairs, the Patient Treatment File, was used to calculate monthly time series over a six-year period for the nation and across US Census Bureau regions for three hospital utilization indicators: average length of stay, discharge rate, and multiple stay ratio, a measure of readmissions. The first purpose of this study was to determine the accuracy of forecasting these indicators 24 months into the future using five univariate time series techniques. In almost all cases, techniques were able to forecast the magnitude and direction of future utilization within a 10% mean monthly error. The second purpose of the study was to describe time series of the three hospital utilization indicators. This approach raised several questions concerning Department of Veterans Affairs hospital utilization.

Geophysical variables and behavior: LXVIII. Distal and lunar variables and traffic accidents in Saskatchewan 1984-1989

The daily total number of traffic accidents and traffic accident severity for the years 1984 to 1989 in Saskatchewan were analyzed to estimate relationships with day of the week, season of the year, and various lunar variables (lunar phase, distance of the moon from the earth, and moon-sun tidal index). A regression model with both a linear term and a trigonometric term was fitted to the data. No relationship between lunar cycles and total accidents or severity of accident was found. A larger number of total accidents was found during the winter months and weekends; however, more severe accidents tended to occur during the summer months. A slight linear trend was uncovered for total accidents over the time period under consideration.

The environmental accident at 'Schweizerhalle' and respiratory diseases in children: a time series analysis

During an investigation concerned with the relationship between air pollution and respiratory diseases in children, the 'Schweizerhalle' accident occurred when unknown amounts of pollutants were discharged into the environment. In that investigation, two series of medical data were collected during one year: (a) The daily relative number of preschool children, exhibiting diseases of the respiratory tract, who either came to the outpatients' clinic of the Children's Hospital or were reported by paediatricians in Basle; (b) The daily number of respiratory symptoms per child, observed in a group of randomly selected preschool children. The purpose of the present time series analysis is the assessment of possible change in these series after the environmental accident. The nature of the change is studied by complementary approaches. First, a forecast arising from models identified in the preaccident period is compared with the actual data. Thereafter, intervention models which adequately and parsimoniously represent the change are identified. Finally, an identification of a change-point is performed.