Jump to comment:

• Change along the time
  Wenbin Liang
  Published on: 27 April 2016
• Author's Reply
  Annette J Dobson
  Published on: 27 April 2016
• Re: 'Excess in cardiovascular events on Mondays: a meta-analysis and prospective study'
  Wenbin Liang
  Published on: 27 April 2016

• Published on: 27 April 2016

  Change along the time

  • Wenbin Liang, taking master of public health in Curtin University

  Dear Editor,

  I do agree that there is an excess of events on Mondays, but I believe it would be affected by the changes in the components of the population at risk, if one or both of the two hypotheses: higher level of stress on Mondays in employed people sudden increase in alcohol consumption at weekends are true, as there two factors are more likely to affect certain age groups.

  For example, there a...

  Show More

  Dear Editor,

  I do agree that there is an excess of events on Mondays, but I believe it would be affected by the changes in the components of the population at risk, if one or both of the two hypotheses: higher level of stress on Mondays in employed people sudden increase in alcohol consumption at weekends are true, as there two factors are more likely to affect certain age groups.

  For example, there are two populations at risk, with 1000, 2000 people in working group respectively, and 10000, 9000 in other group (most of them in retired group) respectively, and the two hypothesizes factors increase both of the incident rate: I1 I2 by 20% on Mondays, compare Monday and a week day one would be 1000*1.2*I1+10000*I1/11000*I1=1.0181.

  Another one would be 2000*1.2*I2+9000*I2/11000*I2=1.0363. As can be seen, the components of the population at risk affect the results. Moreover incident rate in working groups are usually lower then retired groups (especially in fatal events), making the excess less obvious, however the components of the population at risk changed a lot over long time period even if it was observed in the same study.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 27 April 2016

  Author's Reply

  • Annette J Dobson, Professor of Biostatistics
  • Other Contributors:
    • Adrian G Barnett

  Dear Editor,

  We obtained estimates of the excess of events on Mondays from the published reports of each study. Then we meta-analysed the results using random effects models specifically to allow for differences between studies. A measure of heterogeneity was calculated and is reported in Table 3, together with point and interv...

  Show More

  Dear Editor,

  We obtained estimates of the excess of events on Mondays from the published reports of each study. Then we meta-analysed the results using random effects models specifically to allow for differences between studies. A measure of heterogeneity was calculated and is reported in Table 3, together with point and interval estimates of the Monday excess, for all studies and by various subgroups. The results demonstrate considerable variability, as expected for reasons such as those identified by Wenbin Liang. The width of the interval estimates quantifies the precision of the results. After taking the variability into account, there remains evidence of an excess of cardiovascular events on Mondays.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 27 April 2016

  Re: 'Excess in cardiovascular events on Mondays: a meta-analysis and prospective study'

  • Wenbin Liang, taking master of public health

  Dear Editor,

  In this research, the researchers have used ^m%=100(m/N-1/7)/1/7 to analyse the results, however, there are great differences among the studies which were used, and within many of the studies, the events happened over a large time period, thus using ^m% to analyse would lead to great bias. For example in a model only include two studies, each of which only include one year period of observation, now,...

  Show More

  Dear Editor,

  In this research, the researchers have used ^m%=100(m/N-1/7)/1/7 to analyse the results, however, there are great differences among the studies which were used, and within many of the studies, the events happened over a large time period, thus using ^m% to analyse would lead to great bias. For example in a model only include two studies, each of which only include one year period of observation, now, let incident rate of the year be I1 I2 respectively, population at risk be Po1 Po2 respectively, probability of the event happen on Mondays be P1 and P2, probabilities of errors which led to record an event that happened on Monday, as happened on any other days, be E1, E2, probabilities of errors which led to record an event that didn't happen on Mondays as happened on Monday, be F1, F2. Now neglect different standards of diagnosis and precision of diagnosis, changes through out the years. In this example, the core of the equation for ^m%:

  m/N= {I1*Po1*1year*(1-E1)*P1+I1*P01*1year*(1-P1)*F1}+ {I2*Po2*1year*(1-E2)*P2+I2*Po2*1year*(1-P2)*F2}/ (I1*Po1*1year+I2*Po2*1year).

  As can be seen that using m/N to measure (the probability of coronary events happened on Monday)/(the average of probability of coronary events happened throughout the week), is strongly affectively by the differences among the component (for example the proportion of old age groups in the risk population would increase the number of coronary events, but events happened in older age group in which most of the people are retired are much less likely be have high alcohol consumption in weekends, and to have stress from employment, thus to illustrate the two hypothesis effects on the coronary events among the whole population, without a standard population transformation) and size of the population, as the hypothesis suggested that Monday would be more dangerous to certain groups, while if some big studies with higher errors in date recording the results would be even more bias, moreover added changes in standards of diagnosis and precision of diagnosis over long period and over countries, would certainly affect the precision of the result.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top