Article identification process
An initial scope of the PROM literature suggested that relevant papers would be found in two literature clusters; the treatment satisfaction literature and papers using or discussing transition items/external anchor questions/global judgements to validate a PROM measure. A search of the databases described above identified 53 articles relating to transition questions or related questions (i.e. global evaluative judgements). We excluded 50 papers as they contained the word ‘transition’ but did not contain a measure or questions of a transition type. From the reference lists of the 3 remaining papers 22 additional papers were produced, of these 3 were excluded for the same reasons as above. Of the remaining 22 papers 10 additional papers were identified and included, resulting in 32 papers that were included in the transition item review.
Ten key papers concerning patient satisfaction with the outcomes of treatment were identified from a review of the above databases. Scrutiny of their reference lists produced 74 papers of possible relevance, of these 37 were excluded because they dealt more broadly with the patient satisfaction. We therefore included 47 papers in this strand of the review. A total of 79 papers from both search streams were therefore the subject of the combined review.
Approaches to elicit patient reports on the outcomes of treatment
Our literature review confirmed that the two most commonly used approaches to elicit patient reports on the outcomes of treatments were measures of patient satisfaction and measures which employed global ratings of improvement. Other approaches exist which, while not strictly within the scope of this review deserve highlighting. One such approach is to measure the outcomes of treatment against patient specific valued goals, an example of which is the Patient Generated Index (PGI) [
4]. Such measures allow for an assessment of the benefit of treatment against patient specific expectations, often employing a more complicated format with sets of linked transition questions. Such measures may be particularly good at detecting clinically significant change over time, but their complexity may render them less readily useful for some patient populations. This review therefore focused the two most commonly used approaches.
Measures of Patient Satisfaction with the outcomes of treatment are used frequently, and often as a consequence of the importance of collecting clinical trial data with an interest in treatment outcomes important to patients. A second approach is to use a Global Rating of Improvement to assess the benefits of a treatment received; this often takes the form of an overall evaluative question. Health Transition Question(s) (HTQs) are a type of global evaluation question which directly ask patients to assess whether their health or functioning has stayed the same, improved or worsened when compared with a previous (often pre intervention) time point. The latter two approaches reflect a steer away from the construct of satisfaction in recognition of problems associated with this approach e.g. positive skews in data. The following sections of this paper present a selection of findings from the literature review that serve as typical examples of these approaches. Table
1 details a selection of measures that are good examples of well validated and robust measures used to elicit patient reports of the outcomes of treatment. The latter sections of the review outline the advantages and disadvantages of using such approaches and potential strategies to consider when using these methods.
Table 1
A selection of well designed measures to elicit patient report of the outcomes of treatment
Satisfaction with/assessment of the outcomes of treatment
| Diabetes Treatment Satisfaction Questionnaire (DTSQ). [ 5] | An 8 item measure of patient satisfaction with diabetes treatment. | Developed by qualitative work to ensure comprehensive and authentic issues were covered. Assessed psychometrically and analysed in relation to covariates. |
| Revised version of the above. | Detects greater responsiveness to improvements than the original DTSQ. |
Oxford Elbow Score (OES) [ 7] | A 12-item PRO developed to assess the outcomes of elbow surgery. | Shown to be valid, reliable and sensitive to change after rigorous testing. |
Questionnaire on the perceptions of patients about shoulder surgery. [ 8] | A 12-item PRO for patients having shoulder operations. | A short, practical, reliable, valid outcome measure that is sensitive to clinically important changes. |
Questionnaire on the perceptions of patients about total hip replacement. [ 9] | A 12-item PRO for patients having total hip replacement (THR). | As above. |
Questionnaire on the perceptions of patients about total knee replacement. [ 10] | A 12-item questionnaire for patients having a total knee replacement (TKR). | As above. |
Measures containing transition items/global ratings of change
| The Evaluation Ranking Scale (ERS) [ 11] | The ERS asks patients to rank and then rate six dimensions or characteristics of the services they have received. | Compared with a global measure of satisfaction the ERS was more specific, more discriminating, and resulted in lower satisfaction scores [ 11]. |
Patient Judgements of Hospital Quality (PJHQ) [ 12] | Designed to assess the health change associated with hospital stay/treatment over 11 scales. | This measure was subject to extensive and rigorous devolvement and testing that included patient reported open-ended responses about the quality of hospital care, and interviews with hospital administrators, physicians and nurses [ 12]. |
Patient Global Impression of Change Scale (PGIC) [ 13] | Measures patient evaluations of their health change in relation to treatment. | Captures what patients consider to be important changes in pain ratings [ 14] and cancer specific quality of life scores [ 15, 16]. Also a potential correlate of clinical opinion [ 17]. Used in trials of chronic pain [ 17, 18] and recommended as a core outcome measure of global improvement [ 19]. |
The Functional Status Index (FSI) [ 20] | A patient specific measure of change in maximal physical, mental, and emotional function with a transition component that measures change from patient specific norms. | As part of the development it was compared with the Sickness Impact Profile (SIP) [ 21] and performed well, showing sensitivity to change over time [ 22]. |
The Health Transition Index (HTI) [ 23] | Patient rated change in health between two time periods using a 5 point ordinal scale (1 = much better than a year ago; 2 = somewhat better than a year ago; 3 = about the same; 4 = somewhat worse than a year ago; and 5 = much worse than a year ago) | HTI was used as an external anchor to assess the responsiveness of the SF36 [ 23], the HAQ [ 24] and a disease specific health status measure AIMS2 [ 25] in psoriatic arthritis [ 26] and detected as much change as clinical examination [ 26]. |
Short Form 36 (SF36) [ 23] | The SF-36 is a health survey with 36 questions. It yields an 8-scale profile of functional health and well-being scores as well as psychometrically-based physical and mental health. The HTQs have five response categories from “much better” to “much worse”. | The HTQ was assessed among a large general practice sample and correlated well with change measured prospectively [ 27]. The discriminative properties of the HTQs were demonstrated in a similar large population study against prospective change [ 28]. This study was able to successfully distinguish groups whose health had improved compared to those whose health deteriorated. |
Measures of treatment satisfaction
Within the treatment satisfaction literature we identified three methods commonly used; a single global evaluation question i.e.
“How satisfied are you with your current treatment?” a set of separate measures for each aspect of treatment received, and a composite measure comprising of a global item and a set of separate items [
29]. Treatment satisfaction measures also usually include a Likert scale with some also including a visual analogue scale (VAS).
Patient satisfaction with outcomes of treatment has been measured in a vast array of health conditions and procedures. In particular, the fields of diabetes care [
5,
6,
30], orthopaedic surgery [
8‐
10,
31], renal treatment [
32] and asthma [
33] have been fruitful areas for this research, and stand out for applying methodological rigour in the development of measures to assess this (see Table
1). The acceptability of treatments has also been measured in relation to behavioural treatments for children with conduct disorders [
34] and mental health treatments [
35].
Global ratings of improvements/change and health transition questions
As a consequence of some of the problems associated with the data on satisfaction (i.e. positive data skews and undifferentiated data sets), researchers have developed other approaches to obtain this information. The main alternative is to gather patient ratings of improvement or assessments of health change in response to treatment i.e. “Overall, how would you compare your health with the way it was before your surgery, is it much better now, a little better now, about the same, a little worse or much worse?”.
One such approach is to collect a Global Rating of Change (GRC) related to treatment. This approach provides an opportunity for patients to combine all of the components of their experience (e.g. pain relief, improvements in functioning) into one overall evaluative measure of the treatment they receive [
19]. These approaches have also been used to investigate participants’ judgments of the clinical importance of change in other outcome measures [
14,
36]. These questions are commonly used as an external anchor by which to assess the responsiveness of measure to patient rated meaningful health change (See the FACT (Functional Assessment of Cancer Therapy questionnaire and Health Related Quality of life in a study of oncology patients [
15])). Global ratings of change have also been used to elucidate clinically important changes in scores in quality of life of instruments in chronic heart and lung disease [
37,
38], in asthma [
38] and cancer treatment [
39]. Indeed, Jaeschke et al. (1989) concluded that in the absence of a gold standard measure external global ratings represent a credible alternative for establishing the meaning of change in a new measure [
37].
An example of a measure specifically designed for this purpose is the Patient Global Impression of Change Scale (PGIC) [
13]. This rating scale measures patient evaluations of their health change in relation to treatment from “very much improved” to “very much worse” using a visual analogue scale, and has two variants; one for use by the clinician and one for use by the patient. The PGIC has been used in trials of chronic pain [
17,
18] and recommended as a core outcome measure of global improvement [
19].
Central to the importance of measuring PROs are methods that assess the changes in health related status over time. Health Transition Questions (HTQs) do this by directly asking patients to assess whether they consider their health or functioning to have stayed the same, improved or worsened compared with a previous (often pre intervention) time point i.e. “The last time we talked, you said that (physical activity from the baseline index or previous transition) was the most physically strenuous thing you could do? In terms of physical activity now are you: much better, slightly better, the same, slightly worse or much worse?”.
HTQs are employed in a number of ways in health status measurement, and with the exception of studies that have assessed a HTQ within a measure [
27,
28,
40‐
42], and the exception of several notable papers [
43‐
45] there has been little discussion of the widespread use of these questions. This is perhaps a reflection of their perceived usefulness in the absence of a gold standard, much like global questions, they are often used as an external measure or benchmark by which to compare the responsiveness of an existing measure [
15,
37,
46‐
51], or during development of a new measure [
7‐
10,
22,
52,
53]. Studies conducted for the latter purposes thus perform indirect assessments of HTQs in the process of using them as the external benchmark. Studies that directly assess a HTQ within a scale often do this by measuring change assessed prospectively, i.e. by calculating change scores in samples known to have experienced a clinically important change compared to those that have not experienced such a change [
28]. Patient and clinician ratings of patient health change were collected to assess the responsiveness of the Sickness Impact Profile (SIP) and the American Rheumatism Association (ARA) functional scale [
44]. This study found that only changes in SIP physical dimension and patient self-rating showed significant correlation with clinically estimated changes, and that transition items registered changes in clinical status that were not detected by the functional scales [
54]. The benefit of using HTQs to probe patient evaluation has been acknowledged by several authors [
41,
53], mainly as a result of their validity and practicality. Further support for HTQs is provided by studies that assess health status measures against HTQs and standard clinical measurement [
24,
36,
41,
42,
48,
49].