scholarly journals Evaluation of statistical methods used in the analysis of interrupted time series studies: a simulation study

2020 ◽  
Author(s):  
Simon L Turner ◽  
Andrew B Forbes ◽  
Amalia Karahalios ◽  
Monica Taljaard ◽  
Joanne E McKenzie
Keyword(s):  
Time Series ◽  
Statistical Methods ◽  
Simulated Data ◽  
Population Level ◽  
Interrupted Time Series ◽  
Ordinary Least Squares ◽  
In Series ◽  
Time Series Studies ◽  
Unbiased Estimates ◽  
Biased Estimates

AbstractInterrupted time series (ITS) studies are frequently used to evaluate the effects of population-level interventions or exposures. To our knowledge, no studies have compared the performance of different statistical methods for this design. We simulated data to compare the performance of a set of statistical methods under a range of scenarios which included different level and slope changes, varying lengths of series and magnitudes of autocorrelation. We also examined the performance of the Durbin-Watson (DW) test for detecting autocorrelation. All methods yielded unbiased estimates of the level and slope changes over all scenarios. The magnitude of autocorrelation was underestimated by all methods, however, restricted maximum likelihood (REML) yielded the least biased estimates. Underestimation of autocorrelation led to standard errors that were too small and coverage less than the nominal 95%. All methods performed better with longer time series, except for ordinary least squares (OLS) in the presence of autocorrelation and Newey-West for high values of autocorrelation. The DW test for the presence of autocorrelation performed poorly except for long series and large autocorrelation. From the methods evaluated, OLS was the preferred method in series with fewer than 12 points, while in longer series, REML was preferred. The DW test should not be relied upon to detect autocorrelation, except when the series is long. Care is needed when interpreting results from all methods, given confidence intervals will generally be too narrow. Further research is required to develop better performing methods for ITS, especially for short series.

scholarly journals Evaluation of statistical methods used in the analysis of interrupted time series studies: a simulation study

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Simon L. Turner ◽  
Andrew B. Forbes ◽  
Amalia Karahalios ◽  
Monica Taljaard ◽  
Joanne E. McKenzie
Keyword(s):  
Time Series ◽  
Statistical Methods ◽  
Population Level ◽  
Interrupted Time Series ◽  
Ordinary Least Squares ◽  
Standard Errors ◽  
In Series ◽  
Time Series Studies ◽  
Unbiased Estimates ◽  
Biased Estimates

Abstract Background Interrupted time series (ITS) studies are frequently used to evaluate the effects of population-level interventions or exposures. However, examination of the performance of statistical methods for this design has received relatively little attention. Methods We simulated continuous data to compare the performance of a set of statistical methods under a range of scenarios which included different level and slope changes, varying lengths of series and magnitudes of lag-1 autocorrelation. We also examined the performance of the Durbin-Watson (DW) test for detecting autocorrelation. Results All methods yielded unbiased estimates of the level and slope changes over all scenarios. The magnitude of autocorrelation was underestimated by all methods, however, restricted maximum likelihood (REML) yielded the least biased estimates. Underestimation of autocorrelation led to standard errors that were too small and coverage less than the nominal 95%. All methods performed better with longer time series, except for ordinary least squares (OLS) in the presence of autocorrelation and Newey-West for high values of autocorrelation. The DW test for the presence of autocorrelation performed poorly except for long series and large autocorrelation. Conclusions From the methods evaluated, OLS was the preferred method in series with fewer than 12 points, while in longer series, REML was preferred. The DW test should not be relied upon to detect autocorrelation, except when the series is long. Care is needed when interpreting results from all methods, given confidence intervals will generally be too narrow. Further research is required to develop better performing methods for ITS, especially for short series.

Change-Points Detections for Interrupted Time Series Analysis: A Literature Review

2013 ◽  
Vol 462-463 ◽  
pp. 187-192
Author(s):  
Jing Bo Chen ◽  
Jun Bao Zheng ◽  
Lei Yang ◽  
Ya Ming Wang
Keyword(s):  
Time Series ◽  
Literature Review ◽  
Time Series Analysis ◽  
Statistical Methods ◽  
Interrupted Time Series ◽  
Change Points ◽  
Detection Methods ◽  
Interrupted Time Series Analysis ◽  
Series Analysis ◽  
Common Purpose

General review of Change-Points detection methods applied in Interrupted Time Series Analysis for recent years. Articles from domains like meteorology, hydrology, stock analysis, sequences mining et al. are compared together. The literatures range from the 1980s to 2013. The methods are generally classified in Parametric, Semi-Parametric, and Nonparametric. Some non-statistical methods are also mentioned in this review. Characters of each method are briefly summarized. As all methods mentioned in this review share a common purpose that to detect change-points, most of them can be used in other domains after some proper adjustment.

scholarly journals The impact of China’s lockdown policy on the incidence of CoVID-19: An Interrupted time series analysis

2020 ◽  
Author(s):  
Mooketsi Molefi ◽  
John Tlhakanelo ◽  
Thabo Phologolo ◽  
Shimeles G. Hamda ◽  
Tiny Masupe ◽  
...  
Keyword(s):  
Time Series ◽  
Linear Trend ◽  
Effective Means ◽  
Interrupted Time Series ◽  
Ordinary Least Squares ◽  
World Health ◽  
Least Squares Regression ◽  
Interrupted Time Series Analysis ◽  
Post Intervention ◽  
The Impact

Abstract BackgroundPolicy changes are often necessary to contain the detrimental impact of epidemics such as the coronavirus disease (COVID-19). China imposed strict restrictions on movement on January 23rd, 2020.Interrupted time series methods were used to study the impact of the lockdown on the incidence of COVID-19. MethodsThe number of cases of COVID-19 reported daily from January 12thto March 30th, 2020 were extracted from the World Health Organization (WHO) COVID-19 dashboard ArcGIS® and matched to China’s projected population of 1 408 526 449 for 2020 in order to estimate daily incidences. Data were plotted to reflect daily incidences as data points in the series. A deferred interruption point of 6thFebruary was used to allow a 14-day period of diffusion. The magnitude of change and linear trend analyses were evaluated using the itsafunction with ordinary least-squares regression coefficients in Stata® yielding Newey-West standard errors.ResultsSeventy-eight (78) daily incidence points were used for the analysis, with 11(14.10%) before the intervention. There was a daily increase of 163 cases (β=1.16*10-07, p=0.00) in the pre-intervention period. Although there was no statistically significant drop in the number of cases reported daily in the immediate period following 6thFebruary 2020 when compared to the counterfactual (p=0.832), there was a 241 decrease (β=-1.71*10-07, p=0.00) in cases reported daily when comparing the pre-intervention and post-intervention periods. A deceleration of 78(47%) cases reported daily. ConclusionThe lockdown policy managed to significantly decrease the incidence of CoVID-19 in China. Lockdown provides an effective means of curtailing the incidence of COVID-19.

scholarly journals Changes in the dollar value of per capita alcohol, essential, and non-essential retail sales in Canada during COVID-19

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Daniel T. Myran ◽  
Brendan T. Smith ◽  
Nathan Cantor ◽  
Lennon Li ◽  
Sudipta Saha ◽  
...  
Keyword(s):  
Time Series ◽  
Alcohol Use ◽  
Population Level ◽  
Interrupted Time Series ◽  
Interrupted Time Series Analysis ◽  
Retail Sales ◽  
Ongoing Research ◽  
Per Capita ◽  
Statistics Canada ◽  
Alcohol Sales

Abstract Background Multiple survey reports suggest that alcohol use has increased in Canada during the COVID-19 pandemic. However, less is known about how per capita alcohol sales, which predict population-level alcohol use, have changed and whether changes in alcohol sales differ from changes in sales of other products due to pandemic factors. Methods We obtained monthly retail sales data by industry from Statistics Canada, for the six largest provinces in Canada (containing 93% of the national population), between January 2010 and November 2020, representing time before and 9 months after the start of the pandemic in Canada. We used an interrupted time series analysis to estimate pandemic impacts on the dollar value of monthly per capita (per individuals 15+ years) alcohol, essential and non-essential retail sales. We adjusted our analyses for pre-pandemic sales trends, inflation, seasonality and changing population demographics over time. Results During the first 9 months of the pandemic, the values of per capita alcohol, essential and non-essential sales were, respectively, 13.2% higher, 3.6% higher and 13.1% lower than the average values during the same period in the prior 3 years. Interrupted time series models showed significant level change for the value of monthly per capita alcohol sales (+$4.86, 95% CIs: 2.88, 6.83), essential sales (−$59.80, 95% CIs: − 78.47, − 41.03) and non-essential sales (−$308.70, 95% CIs: − $326.60, − 290.79) during the pandemic. Alcohol sales were consistently elevated during the pandemic, and the pre- and post-pandemic slopes were comparable. In contrast, essential and non-essential retail sales declined in the early months of the pandemic before returning to regular spending levels. Conclusion During the first 9 months of the pandemic, per capita alcohol sales were moderately elevated in Canada. In contrast, non-essential sales were lower than prior years, driven by large decreases during the initial months of the pandemic. These findings suggest that the pandemic was associated with increased population-level alcohol consumption, which may lead to increased alcohol-related harms. Ongoing research is needed to examine how factors, including pandemic-related stressors and specific alcohol sales-related policies, may have influenced changes in alcohol use and harms.

scholarly journals Comparison of Six Statistical Methods for Interrupted time Series Studies: Empirical Evaluation of 190 Published Series

2020 ◽  
Author(s):  
Simon Turner ◽  
Amalia Karahalios ◽  
Andrew Forbes ◽  
Monica Taljaard ◽  
Jeremy Grimshaw ◽  
...  
Keyword(s):  
Time Series ◽  
Statistical Method ◽  
Statistical Methods ◽  
Statistical Significance ◽  
Empirical Evaluation ◽  
Interrupted Time Series ◽  
Series Data ◽  
Standard Errors ◽  
P Values ◽  
The Impact

Abstract Background The Interrupted Time Series (ITS) is a quasi-experimental design commonly used in public health to evaluate the impact of interventions or exposures. Multiple statistical methods are available to analyse data from ITS studies, but no empirical investigation has examined how the different methods compare when applied to real-world datasets. MethodsA random sample of 200 ITS studies identified in a previous methods review were included. Time series data from each of these studies was sought. Each dataset was re-analysed using six statistical methods. Point and confidence interval estimates for level and slope changes, standard errors, p-values and estimates of autocorrelation were compared between methods. ResultsFrom the 200 ITS studies, including 230 time series, 190 datasets were obtained. We found that the choice of statistical method can importantly affect the level and slope change point estimates, their standard errors, width of confidence intervals and p-values. Statistical significance (categorised at the 5% level) often differed across the pairwise comparisons of methods, ranging from 4% to 25% disagreement. Estimates of autocorrelation differed depending on the method used and the length of the series. ConclusionsThe choice of statistical method in ITS studies can lead to substantially different conclusions about the impact of the interruption. Pre-specification of the statistical method is encouraged, and naive conclusions based on statistical significance should be avoided.

scholarly journals Assessment of the Impact of COVID-19 pandemic on population level interest in Skincare: Evidence from a google trends-based Infodemiology study

2020 ◽  
Author(s):  
Hasan Symum ◽  
Md. F. Islam ◽  
Habsa K. Hiya ◽  
Kh M. Ali Sagor
Keyword(s):  
Time Series ◽  
Moving Average ◽  
Arima Model ◽  
Population Level ◽  
Interrupted Time Series ◽  
Interrupted Time Series Analysis ◽  
Search Volume ◽  
Quasi Experimental ◽  
Relative Search Volume ◽  
The Impact

AbstractBackgroundCOVID-19 pandemic created an unprecedented disruption of daily life including the pattern of skin related treatments in healthcare settings by issuing stay-at-home orders and newly coronaphobia around the world.ObjectiveThis study aimed to evaluate whether there are any significant changes in population interest for skincare during the COVID-19 pandemic.MethodsFor the skincare, weekly RSV data were extracted for worldwide and 23 counties between August 1, 2016, and August 31, 2020. Interrupted time-series analysis was conducted as the quasi-experimental approach to evaluate the longitudinal effects of COVID-19 skincare related search queries. For each country, autoregressive integrated moving average (ARIMA) model relative search volume (RSV) time series and then testing multiple periods simultaneously to examine the magnitude of the interruption. Multivariate linear regression was used to estimate the correlation between countries’ relative changes in RSV with COVID-19 confirmed cases/ per 10000 patients and lockdown measures.ResultsOut of 23 included countries in our study, 17 showed significantly increased (p<0.01) RSVs during the lockdown period compared with the ARIMA forecasted data. The highest percentage of increments occurs in May and June 2020 in most countries. There was also a significant correlation between lockdown measures and the number of COVID-19 cases with relatives changes in population interests for skincare.ConclusionUnderstanding the trend and changes in skincare public interest during COVID-19 may assist health authorities to promote accessible educational information and preventive initiatives regarding skin problems.

Sign in / Sign up

Export Citation Format

Share Document