Dear Muhammad Adnan

If you are to follow up your cases for the two scoring methodologies over a certain period of time to see mortality then i guess you must have in mind certain number of patients being followed up. This will be your cohort and i interpret it will be a cohort study. It will be then prospective in nature.

Regards

Dear Muhammad Adnan

If you are to follow up your cases for the two scoring methodologies over a certain period of time to see mortality then i guess you must have in mind certain number of patients being followed up. This will be your cohort and i interpret it will be a cohort study. It will be then prospective in nature.

Regards

Agree with Sikandar Hayat Khan

Following

Dr. Sikandar Hayat Khan and Dr. Ajit kumar Roy, I need further explanation from you.

Cohort Study involves follow up and a non-exposed or matched control group. In my study, there is follow up but no control group. Moreover, I am not interested in mortality associated factors, as I will compare two mortality predicting scales.

Case series lacks control group, but it is descriptive design.

What you suggest now?

I am also considering cohort study.

Dear Muhammad Adnan

Thanks again. In fact you talked about mortality and comparison of two scoring systems:

FIRST: Will you be doing same scoring on same patient as I interpret?

SECOND: If not mortality, then define morbidity related end-points like stay in hospital or whatever is feasible to you

THIRD: You can't have controls, so just do a "case series study", though low in evidence but as I interpret you can't have mortality data in short-term, You can;t controls, You are to find some end-points

You can make groups like those leaving early from hospital and initial scores at assessment time AND those leaving late and their assessment scores by 2 methods but still will be termed as "comparative cross-sectional study".

Anyway defining OBJECTIVE is prerequisite here before you start anything

Regards

Dr. Sikandar Hayat Khan

Sir objective is to compare new scoring scale versus classical scale.

Both scales predict mortality.

Upon admission patients enrollment and scoring will be performed.

Then patients will be observed for mortality till discharge from hospital.

Dear discussants! @ MAdnan's last entry says very clearly what must and will be done. You may call is a (clinical) cohort study if you like. A consecutive prospective clinical sample, I would call it, comparing two prognostic (non-intervention) 'devices.'

Critical is: a clear def. of entry criteria; strict consecutivity; comparative data collection and scores not to be revealed to those in charge of treatment INCLUDING those deciding whether and when to discharge the patient [it's easy to overlook the last point].

- - Once the data are collected, the choice of an appropriate statistical analysis is another challenge.

Hope this is helpful! Jørgen

Dr. Jørgen Hilden, thank you so much for your kind instructions, definitely I will go with this design, as other colleagues have also recommended it. However I am a little bit confused about control group which will not be part of this proposed cohort study.

What would the purpose of a 'control group' be? What should a control group look like, i.e., how would its handling differ from what you already have? If you plan to test a difference from the control group, what would this test help making sure? -- -- No, as the purpose in described, everything rests on the INTERNAL comparisons in the data set [and, luckily, basable on within-patient ('paired') comparisons]. - More questions? /JH

Dr. Jørgen Hilden, thanks again sir.

As in cohort study, we have two groups, first exposed group include subjects with and without outcome, second non-exposed group which also include subjects with and without outcome.

When I see my proposed cohort study, it only include septic patients with and without mortality; where objective is comparison of two scales that predict mortality.

What you suggest if I go with "observational study" design?

Dear Muhammad, You are a victim of the "design tyrranny" that crept into epidemiology and biostatistics around 1980. I am referring to the idea that every study represents a textbook design and preferably must have that design's name or label in its title. That is not how the world is. - No, no, a well-designed study is a well-designed study whether or not you can find a matching study, and a label for the design, in textbooks. If you cannot find the match, it is the textbook writer's fault, not yours!

If you need a label after all, then 'consecutive clinical sample' is fine. That this is a subclass of observational studies is admittedly true because no intervention is involved, but it is misleadingly unspecific to use 'observational' as your design label.

Good luck! Jørgen

Dr. Jørgen Hilden,

I am grateful to you for giving your worthy time and suggestions.