In linear regression , I have been performing to identify the relation between risk ratio for outcome of tumors relapse and duration of follow up in months. The results were :
Slope :-0.00015. P value =0.8882
Intercept : -0.599 ,P value = 0.02
Tau-square = 0.118
See regression plot below
Beside the answer above, I would add some minor comments on what have been done:
- You have follow up time, is there any issue with lost-to-follow up or even censoring data here?
- Regarding your regression model: Have you performed model diagnostic (normal distribution of error terms, constant variance).
- Have you considered other possible variables in the model?
- If everything is checked and the model seems to be appropriate: given the high p-value (for the slope), it is said that there is no linear relationship between log(RR) and duration of follow up.
with the p-value being so high on the slope term and knowing nothing else about the data, objectives, etc. I would say that there is no linear relationship between log(risk Ratio) and duration of followup. The plot seems to support this. Best wishes.
Beside the answer above, I would add some minor comments on what have been done:
- You have follow up time, is there any issue with lost-to-follow up or even censoring data here?
- Regarding your regression model: Have you performed model diagnostic (normal distribution of error terms, constant variance).
- Have you considered other possible variables in the model?
- If everything is checked and the model seems to be appropriate: given the high p-value (for the slope), it is said that there is no linear relationship between log(RR) and duration of follow up.
In addition to the above, the plot makes me think that most of the obs were taken at the same length of time. Do the majority of patients have the same length of stay? By length of stay I meant duration of followup.
Dear David Eugene,
Thank you for your helpful answer. Please be advise that duration of follow up was based on the means (average). I have inputting the means of induced studies. Is this method valid or should be for patients ? this analysis was done for systematic review.
Thank you again
Dear Thao Mai Phuong Tran,
Thank you for your answer,
Yes there are data concerning that matter, but I didnot included in above analysis. Should I do that?
I have used comprehensive meta analysis with effect of moment (random effect model)
No, only duration of follow up was used as explanatory variable
I think you might want to look at the data for patients. I think you would be more likely to see an effect there. Best wishes
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression