It is not clear to me here why high satisfaction would be viewed as reflecting insensitivity or inaccuracy? What I suspect is you have a ceiling effect that you are working with given the predominantly high scores. You may find this document helpful.

http://srmo.sagepub.com/view/the-sage-encyclopedia-of-social-science-research-methods/n102.xml

Indeed, reported high satisfaction might be accurate. Let me better explain my search: I suspect that many patients reporting to be satisfied might actually have experienced different things and I would like to be able to differentiate them.

For example: 2 patients might have a negative experience but for one the original problem did not fall within the patients perception of the doctors duty, so this patient would still report to be satisfied, while for an other patient this problem did fall in the doctors' duty and this patient might report to be dissatisfied IF he feels it was the doctors "fault" but still report to be satisfied IF he feels it was not the doctors fault. Together with satisfied patients who had a positive experience I than have 3 satisfied reports for 3 different experiences.

In other words, are there instruments that are more sensible than others to differentiate between different "satisfied" patients?

an article I read on this is: Williams B, Coyle J & Healy D. 1998. The meaning of patient satisfaction: an explanation of high reported levels. Soc. Sci. Med. 47, 1351-1359

If I understood correctly in this case you are looking to parcel out the "satisfaction." If so then I would start with what are the "areas" that this satisfaction can be parceled into and generate the item set. You can set the items up using a system similar to what is used in the Student Satisfaction Inventory. Expectations and satisfaction for the two rating perhaps. Perhaps satisfaction and physician influence as the two ratings. It depends on what you wish to measure. See teh format at:

http://survey.noellevitz.com/index.cfm?sAction=survey&z=1399987578

Thank you for this idea, it's very interesting. It would mean using a "customized" instrument, which than would not be validated nor would it be known to be reliable. This is difficult for scientific comparison. But I am very attracted to your suggestion of parcelling "satisfaction"....Does anyone know if something likewise has been done for physician-patient interaction?

The Student Satisfaction Inventory setup seems a good idea too! Thank you for your input.

If you construct it using the literature, you are building in content related validity. I am not sure which reliability you are working to look for, but with survey data there are multiple options. You can look at the reliability across person per item - treating persons as items in the traditional sense. You could also, more conventionally, look at the items per section (scale) for reliability. Reliability covers consistency - consistent understanding across raters, consistent responses across items per scale, Consistent responding across time.

You might want to have a look at pages 125-127 of the following document:

http://www.improvingchroniccare.org/downloads/health_literacy_universal_precautions_toolkit.pdf

I find this really useful.

Dear Mirjam,

Interesting question- I think the main thing to note is that patient experience is a multi-dimensional construct encompassing a number of elements of care. However, it can be broadly divided into two domains:

i) patient reports: patients’ accounts of what happened (e.g. did you receive information about your treatment?); and

ii) patient evaluation: patients’ ratings of their experience (e.g. how would you rate the information you received about your treatment?).

As you would expect, satisfaction would fall under the second one, and it is just one type of patient evaluation(there are others). In general, patient reports are better at distinguishing experience (as opposed to satisfaction, which can include expectation), what you are trying to look at (especially if the question is asking about a specific aspect of care). [For a more detail discussion of this and refs, I have attached a very recent paper of ours].

Therefore, in the UK health system, the National General Practice Patient Survey mostly focuses on report items. For an example of this, see Q23 on pg 4, this might be of use to you, and it is validated:

http://www.gp-patient.co.uk/download/Questionnaires/Y5Q4%20GP%20Patient%20Survey%20questionnaire.pdf On average, what we find is that this question does show substantial variation according ethnicity, deprivation, health status, and a number of other socio-demographics factors.

Hope this helps!

thank you all for your helpful comments! I will read the articles with great interest.

Dear Mirjam,

You might read this paper, it explains some theoretical approaches in measuring the patient evaluations. As the authors suggest, I think it is also interesting to consider the importance that patients give to each aspect of care/communication process.

http://onlinelibrary.wiley.com/doi/10.1046/j.1369-6513.1998.00004.x/abstract;jsessionid=85406FD6B7437381A3928A6772154F83.f03t02?systemMessage=Wiley+Online+Library+will+be+disrupted+Saturday%2C+7+June+from+10%3A00-15%3A00+BST+%2805%3A00-10%3A00+EDT%29+for+essential+maintenance&userIsAuthenticated=false&deniedAccessCustomisedMessage=

A really interesting problem, and some great comments. My two cents, as a service designer, is some questions (I don't have ready answers, but note others cite some really good papers):

1. The "Halo" Effect: understanding the context for measurement. Patients naturally feel grateful and relieved when cared for, so have a bias to be polite/ positive and forgive issues they would not forgive in a bank (for example). This means "satisfaction" can have a hidden normative bias and can systematically over-estimate the quality of experience/ service. Is there a more "contextually valid" (for the patient) measure?

2. The "Satisfaction" Effect: understanding the semantic gap. "Satisfaction" is a vague, abstract concept (it denotes "enough") and is a passive measure. So if I don't expect much, I can be satisfied with what I get. And when I say I'm "quite satisfied", you (the researcher) are probably quite satisfied too, though in reality you're not actually learning anything substantial. There is a significant gap between the dynamics of a service experience and the abstractness of the measure. Is this the best "meaning/ word/ concept" to measure the dynamics of patient experiences?

3. The "Utility" Effect: understanding the bias against improvement. When someone says they are "very satisfied" this can suggest to us researchers that all is as good as it could be, and we can relax. Of course, it just means the communication (for example) was above the patient's knowledge of what the best might be, or what else might be possible. And in fact, there will still be many opportunities for improvement and innovation the patient (and/ or the team) is unaware of. Is this the best tool to measure opportunities for improvement and innovation?

i think questionnaire will serve you the best