1
$\begingroup$

I am doing a logistic regression . My outcome is a categorical (yes/ no) pain after surgery. The predictors i wish to model for includes the type of anaesthesia , among other predictors. The problem is the types i wish to include are general anaesthetic, plain spinal anaesthetic, spinal anaesthetic with morphine and finally spinal anaesthetic with diamorphine. Now, the 3 spinal anaesthetic categories are not mutually exclusive unlike general anaesthetic versus spinal anaesthetic alone. Is this appropriate for logistic regression? Or should i be using spare predictors ; i.e. GA vs Spinal and then spinal vs spinal morphine vs spinal diamorphine.. Any advice (in plain language please) would be gratefully received.

$\endgroup$
1
  • 1
    $\begingroup$ Could you please explain how these are not mutually exclusive? That is tantamount to saying that a patient could have received two or more of these types of anesthesia. Is that really the case? If so, then the solution may depend on what comparisons you plan to make. What are they? $\endgroup$ – whuber Jan 28 '15 at 14:49
2
$\begingroup$

One option is to make the categories:

General anesthetic

  1. Spinal anesthetic with neither morphine nor diamorphine
  2. SA with morphine only
  3. SA with diamorphine only
  4. SA with both

Then they are mutually exclusive.

$\endgroup$
2
  • $\begingroup$ So would i be right in considering the 4 groups : General anaesthetic , Spinal alone, Spinal with morphine and spinal with diamorphine ? The spinal anaesthetics (have similarities in terms of basic drug used; i.e. local anaesthetic) ; none/ morphine or diamorphine are adjuvants... $\endgroup$ – neesax2003 Jan 28 '15 at 15:25
  • $\begingroup$ That shouldn't be a problem $\endgroup$ – Peter Flom Jan 28 '15 at 17:01
1
$\begingroup$

What do you mean by they are not mutually exclusive? Can a patient be assigned in more than one anaesthesia category at the same time? No. A patient is assigned to the general anaesthesia, plain spinal anaesthesia, spinal anaesthesia with morphine or spinal anaesthesia with diamorphine category. One who is assigned to the plain spinal anaesthesia category cannot at the same time be assigned in the spinal anaesthesia with morphine category. This means each category is distinct. And your intent is to determine if the outcome differs by type of anaesthesia. So all you need to do is to choose one of the anaesthesia categories as your reference category and to run the logistic regression. Then the result of the analysis for each of the remaining three categories would be interpreted relative to the reference category. (However, if your objective is to just compare the outcome between patients receiving general anaesthesia and those receiving any type of spinal anaesthesia, then you can merge the three spinal anaesthesia categories and you can compare 'general anaesthesia vs spinal anaesthesia'.)

You also can read this and this about binary logistic regression.

$\endgroup$
7
  • $\begingroup$ Thank you for the very helpful comment. I wasn't sure about that myself and hence the question. So.....spinal anaesthetic is a standard anaesthetic technique (uses local anaesthetic) and you can add nothing to it ....or an adjuvant such as morphine or diamorphine . So, i was wondering if there was quite an overlap between the groups (to consider them mutually exclusive).....as opposed to a general anaesthetic and a spinal anaesthetic (alone). $\endgroup$ – neesax2003 Jan 28 '15 at 15:22
  • 1
    $\begingroup$ If you are not sure whether each patient received only one of the mentioned anaesthetics or a combination of them, it is necessary to go back to the raw data (if possible) to clarify the issue. Or, have correspondence with the agency that provided you with the data and get clarifications. Based on the clarifications you provide I will edit my answer. $\endgroup$ – Ayalew A. Jan 28 '15 at 15:37
  • $\begingroup$ What I understood from your post was that plain spinal anaesthesia refers to the standard anaesthesia technique (eg., lidocaine) with nothing added to it. Could it be different? $\endgroup$ – Ayalew A. Jan 28 '15 at 15:49
  • $\begingroup$ Besides, can the two adjuvants be given together? If not, then you have four mutually exclusive categories: general anaesthesia, standard spinal anaesthesia (plain), plain + morphine and plain + diamorphine. The fact that the basic drug is the same doesn't matter. Each category is distinct. Plain is different from plain + morphine, etc. $\endgroup$ – Ayalew A. Jan 28 '15 at 16:06
  • $\begingroup$ Thanks Ayalew! There's no confusion about the anaesthetic used. You are right , spinal anaesthetic (plain) refers to bupivacaine alone, whereas the other groups used bupivacaine + morphine or bupivacaine + diamorphine. The fourth group being General anaesthetic . So in my view, there is more in common in the 3 spinal related groups (no additive, morphine or diamorphine) than the fourth (GA) , so with that in mind, are they different enough to be treated as separate (mutually exclusive groups)? I have actually done the analysis with the 4 groups as separate with no significant contribution to $\endgroup$ – neesax2003 Jan 28 '15 at 16:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.