Comparing patient‑reported outcome and experience measures in international health surveys

3. Results across selected comparable indicators

Further comparative analysis was conducted on a selected set of indicators from PaRIS, EHIS, the IHP survey and PVS. The comparative analysis was conducted for a selected set of indicators. All surveys were standardised to a common demographic reference using a post-stratification approach in which the PaRIS population served as the reference population, strict harmonisation was applied among service users with chronic conditions and results were compared across countries.

To ensure stricter cross‑country comparisons, further analysis was conducted harmonising populations, indicators, and demographic structures across PaRIS, EHIS, the IHP survey and PVS. This analysis compared selected indicators for people aged 45+ who live with chronic conditions and had a recent healthcare contact across PaRIS, EHIS, the IHP survey and PVS (Box 3).

Box 3. Comparing a selected set of indicators through further harmonisation, standardisation and weighting

• Four indicators were examined: self-rated health (PROM), experienced quality of care (PREM), confidence to self-manage, and hospitalisations. To enhance comparability across surveys, response categories were harmonised by recoding the responses.

• Analyses were limited to countries available in both PaRIS and the comparison survey. The PaRIS-EHIS comparison included 13 countries (Belgium, Czechia, France, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Romania, Slovenia, Spain and Norway). The PaRIS-PVS comparison included four countries (Greece, Italy, Romania and the United States). The initial PaRIS-IHP set included six countries (Australia, Canada, France, the Netherlands, Switzerland and the United States). Indicator coverage also differed within the IHP survey: quality of care and self-rated health were available for Canada, Switzerland and the United States, but self-management confidence was only asked in Switzerland.

• Age scope differed across comparisons. For PaRIS-IHP, analyses were restricted to respondents aged 65 and over, reflecting the older adult focus of that comparison. For PaRIS-EHIS and PaRIS-PVS, analyses included all respondents aged 45 and over, as both EHIS and PVS provided sufficient sample sizes across this broader age range.

• To improve comparability across surveys, all estimates were standardised to a common age sex reference distribution based on the PaRIS design targets for respondents aged 45 and over. Targets for ages 45‑54, 55‑64, 65‑74 and 75+ were renormalised for the restricted age group and combined with original sex targets (55% female; 45% male) to form a four‑cell reference structure. Post-stratification weights were then generated for each survey and country by aligning observed age-sex proportions to these reference targets, with weights normalised to preserve effective sample sizes.

3.1. Self-rated general health

Self‑rated health status is measured in PaRIS, EHIS, the IHP survey and PVS using closely related questions, though with some differences in response scales and category ordering. All four surveys ask respondents to assess their health “in general,” but EHIS uses a scale ranging from “very good” to “very bad,” while PaRIS, the IHP survey and PVS use variations of an “excellent” to “poor” scale, with PVS presenting responses in reverse order. Taking into account these differences in response options and labelling, results were aligned into harmonised outcome categories for comparative analysis (Table 3). The differences in scale construction and threshold placement, particularly the treatment of “fair” health in EHIS, hinder cross‑survey comparisons. Nevertheless, the exercise is useful to shed light on the effect of methodological differences in the self-rated general health assessment.

Table 3. Harmonisation of response categories for self-rated general health

	PaRIS	EHIS	the IHP survey	PVS
Question	In general, would you say your health is...	How is your health in general?	In general, how would you describe your own health?	In general, would you say your health is?
Response categories (in original order)	1. Excellent 2. Very good 3. Good 4. Fair 5.Poor	1. Very good 2. Good 3. Fair 4. Bad 5. Very bad	1. Excellent 2. Very good 3. Good 4. Fair 5.Poor	1. Poor 2. Fair 3. Good 4. Very good 5. Excellent
Recoding (and reversing order in PVS)	Positive <- excellent OR very good OR good; compared to fair OR poor	Positive <- very good OR good; compared to fair OR bad OR very bad	Positive <- excellent OR very good OR good; compared to fair OR poor	Positive <- excellent OR very good OR good; compared to fair OR poor
Final reported outcome	Percentage of people reporting positive self-rated health

Figure 3 shows that cross-survey country gradients remain broadly similar in most countries, but differences in absolute levels become much larger once the comparison is restricted to service users with chronic conditions and standardised to the PaRIS standard population.

In the overall comparison (Panel A), PaRIS and EHIS show broadly consistent rankings but with systematic level differences that vary by country. The differences between PaRIS and EHIS range from +13 p.p. in Portugal and Slovenia where more respondents in PaRIS report positive self-rated health to ‑16 p.p. in Italy and ‑11 p.p. in Spain where fewer respondents in PaRIS report positive self-rated health. In the standardised and harmonised comparison (Panel B), EHIS is lower than PaRIS in 11 of the 13 overlapping countries, with gaps ranging from +30 p.p. in Romania and +24 in Slovenia where more respondents in PaRIS report positive self-rated health to +2 points in Norway and ‑5 points in Italy where fewer respondents in PaRIS report positive self-rated health.

In the comparison between PaRIS and the IHP survey (Panel A), more respondents in PaRIS generally report positive self‑rated health than the IHP survey, though the magnitude and direction of differences vary by country. Across the six overlapping countries in the crude comparison, differences range from +11 p.p. in France, +6 points in Canada, and +3 points in Switzerland, where more respondents in PaRIS report positive self‑rated health, to –1 point in the Netherlands, where fewer respondents in PaRIS report positive self‑rated health. Australia shows alignment between the two sources. In the standardised and harmonised comparison among service users with chronic conditions (Panel B), PaRIS reports higher levels of positive self‑rated health than the IHP survey in all six overlapping countries, with gaps widening substantially compared to the crude results. Differences range from +17 p.p. in France and +12 points in Canada to smaller gaps of +6 points in Australia and +5 points in the United States.

Figure 3. Self-rated general health varies on average 6.2 p.p. across surveys

Note. The data are sorted from largest to smallest based on the PaRIS results.

In the comparison between PaRIS and PVS (Panel A), country rankings are broadly aligned but level differences are substantial and consistently favour PVS in most overlapping countries. Among the four countries with data in both sources, differences range from +2 p.p. in Greece, where PaRIS reports slightly higher positive self‑rated health than PVS, to –18 p.p. in Italy, where PaRIS reports lower levels than PVS. Smaller gaps are observed in the United States (–3 p.p.) and Romania (–4 p.p.). In the standardised and harmonised comparison among service users with chronic conditions (Panel B), the pattern reverses with more respondents in PaRIS reporting positive self-rated health than PVS in all four overlapping countries, with gaps ranging from +23 p.p. in Greece and +14 in the United States to +3 points in Italy and +5 points in Romania.

PaRIS is used as the reference survey for the restriction and harmonisation exercise, and by construction its results change only minimally when restricted to the standardised population with chronic conditions. Among the remaining surveys, EHIS is the most affected by restriction and harmonisation. The adjustments lead to large and systematic downward shifts relative to PaRIS, often reversing the direction of differences observed in the crude comparison and substantially widening gaps in many countries. This indicates a sensitivity of EHIS estimates to population composition and survey design, particularly its coverage of the general population compared with service users with chronic conditions.

Another factor that deserves attention is the difference in the wording of the response categories between PaRIS and EHIS. PaRIS uses the North American style scale (Excellent/ Very good /Good /Fair/ Poor), whereas EHIS uses the European-style scale (Very good/ Good/ Fair/ Bad/ Very bad). Although both scales contain five categories and are conventionally dichotomised at the same cut-off when constructing a “positive self-rated health” indicator, the labels are not psychometrically equivalent. The PaRIS scale is positively skewed in its anchors (four of five categories are non-negative), while the EHIS scale is symmetric and with explicit negative anchors at the lower end. Evidence from survey methodology shows that such differences shift the psychological midpoint of the scale, alter the distribution of responses, and complicate direct cross-survey comparisons of self-rated health (Jürges, Avendano and Mackenbach, 2008[28]).

PVS is also affected, though generally less than EHIS. The IHP survey is moderately affected by restriction and harmonisation. Differences between PaRIS and the IHP survey are already present in the crude comparison and tend to widen after standardisation, but without systematic patterns. In the crude comparison, PVS frequently reports higher levels of positive self‑rated health than PaRIS, but after restriction and harmonisation this pattern reverses, with more service users with chronic conditions in PaRIS reporting positive self-rated health.

3.2. Experienced quality of care

Experienced quality of care is measured across PaRIS, the IHP survey and PVS using conceptually similar but not identical questions. PaRIS asks respondents to rate the medical care that they have received in the past 12 months from their primary care centre, while the IHP survey captures overall satisfaction with healthcare received over the past 12 months, and PVS focusses on the quality of care received from a specific healthcare facility over the same period. Response scales also differ slightly, with PaRIS and PVS using quality‑based ratings (from “excellent” to “poor”) and the IHP survey using satisfaction‑based categories (from “very satisfied” to “very dissatisfied”). Table 4 presents the harmonisation of response categories regarding the experienced quality of care outcome. While this recoding enables high‑level comparability across surveys, differences in question focus, single consultation versus care over a year, and differences in response framing (quality versus satisfaction) should be considered when interpreting results.

Table 4. Harmonisation of response categories for experienced quality of care

	PaRIS	the IHP survey	PVS
Question	Overall, how do you rate the medical care that you have received in the past 12 months from your primary care centre?	Overall, how satisfied are you with the quality of healthcare you have received during the past 12 months?	Overall, how would you rate the quality of healthcare you received in the past 12 months from this healthcare facility?
Response categories (in original order)	1. Excellent 2. Very good 3. Good 4. Fair 5. Poor 6. Have not received medical care in the last 12 months 7.Not sure	1. Very satisfied 2. Somewhat satisfied 3. Neither satisfied nor dissatisfied 4. Somewhat dissatisfied 5. Very dissatisfied	1. Excellent 2. Very good 3. Good 4. Fair 5. Poor
Recoding	Positive <- excellent OR very good OR good; compared to fair OR poor; NA <- Have not received medical care in the last 12 months OR not sure	Positive <- very satisfied OR somewhat satisfied OR neither satisfied nor dissatisfied; compared to somewhat dissatisfied OR very dissatisfied	Positive <- excellent OR very good OR good; compared to fair OR poor
Final reported outcome	Percentage of people reporting good quality of care

Figure 4 shows that cross‑survey country gradients for experienced quality of care remain broadly similar across PaRIS, the IHP survey and PVS, but, similar to self‑rated health, absolute levels diverge more strongly once the comparison is restricted to service users with chronic conditions and standardised to the PaRIS reference population.

In the overall comparison (Panel A), PaRIS generally reports higher levels of experienced quality of care than the IHP survey, although the size and direction of differences vary by country. Among the five overlapping countries, differences range from +17 p.p. in Canada, +7 points in France, where more respondents in PaRIS report positive experiences of care, to –6 points in Australia and –1 point in the United States, where PaRIS reports lower levels. Comparisons with PVS in the crude data among four overlapping countries show higher levels of positive results for PVS than PaRIS, with gaps ranging from –14 p.p. in Romania and –7 points in the United States, to –2 points in Italy, and –5 points in Greece.

In the standardised and harmonised comparison among service users with chronic conditions (Panel B), PaRIS systematically reports higher experienced quality of care than the IHP survey in all overlapping countries. Differences range from +16 p.p. in Canada and +7 points in Australia to +4 points in Switzerland, France and the United States. Comparisons with PVS show that PaRIS and PVS are closely aligned in the United States and Italy (no difference), while PaRIS reports lower levels in Greece (‑15 points) and in Romania (–13 points).

Overall, the results indicate that experienced quality of care is more sensitive to restriction and harmonisation than crude comparisons imply, particularly for the IHP survey, where standardisation and harmonisation reveal systematically higher levels in PaRIS, while differences with PVS remain country‑specific.

Figure 4. Experienced quality of care

Note. The data are sorted from largest to smallest based on the PaRIS results.

3.3. Confidence to self-manage

Confidence to manage one’s own health is assessed in both PaRIS and PVS using broadly similar but not identical questions. PaRIS asks respondents about their confidence to manage their own health and well-being, while PVS frames the question in terms of personal responsibility for managing overall health. Response categories differ substantially across surveys in wording and scale, but both distinguish varying levels of confidence. For comparative purposes, responses were grouped into a harmonised positive outcome, while noting that differences in question framing and response options may influence interpretation (Table 5).

Table 5. Harmonisation of response categories for confidence to self-manage

	PaRIS	PVS
Question	How confident are you that you can manage your own health and well-being?	When you think about your health, how confident are you that you are the person who is responsible for managing your overall health?
Response categories (in original order)	1. Very confident 2. Confident 3. Somewhat confident 4. Not confident at all	1. Very confident 2. Somewhat confident 3. Not too confident 4. Not at all confident
Recoding	Confident <-Very confident OR confident; compared to Somewhat confident OR not confident at all	Confident <-Very confident OR somewhat confident; compared to Not too confident OR not at all confident
Final reported outcome	Percentage of people with positive outcomes

Figure 5 indicates that cross‑survey country gradients for confidence to self‑manage health are broadly consistent between PaRIS and PVS, but absolute levels differ markedly, reflecting both population coverage and important differences in question framing. While PaRIS asks respondents about their confidence in managing their own health and well-being, PVS focusses more narrowly on whether individuals see themselves as the person responsible for managing their overall health, a formulation that is likely to elicit higher affirmative responses.

Figure 5. More respondents in PVS reported feeling responsible to manage compared to the number of PaRIS respondents having confidence to manage

Note. The data are sorted from largest to smallest based on the PaRIS results.

In the overall comparison (Panel A), PVS reports substantially higher levels of confidence than PaRIS in all four overlapping countries. Differences range from ‑23 p.p. in the United States (73% in PaRIS and 96% in PVS) to very large gaps of ‑43 points in Greece, ‑43 points in Romania, and ‑62 points in Italy, where PaRIS results are much lower. Despite these large level differences, the relative ordering of countries is broadly similar across the two surveys, with the United States reporting the highest levels and Italy the lowest, suggesting consistent cross‑country gradients.

In the standardised and harmonised comparison among service users with chronic conditions (Panel B), results change only marginally compared to the crude comparison. PaRIS estimates remain largely stable or decline slightly (for example from 40% to 35% in Greece), while PVS levels remain very high across all countries. As a result, gaps persist and remain substantial, ranging from ‑22 p.p. in the United States to ‑46 points in Greece, ‑46 points in Romania, and ‑59 points in Italy.

These large and persistent differences are consistent with differences in question framing and conceptual focus. PaRIS asks respondents about their confidence in managing their health and well-being, which may implicitly anchor responses in the context of chronic conditions and ongoing care needs. By contrast, PVS explicitly frames self‑management in terms of personal responsibility for overall health, which may prompt a broader, more normative assessment of agency and control. As a result, the two surveys may be capturing related but distinct constructs, with PVS eliciting generally higher affirmative responses. This highlights the limits of comparability for measures of self‑management, even when response categories are harmonised and populations are standardised.

3.4. Hospitalisation

Hospitalisation in the past 12 months is measured in PaRIS, EHIS and PVS using questions focussed on overnight stays, though with minor differences in wording and response formats. PaRIS records the number of overnight hospitalisations, EHIS asks directly about inpatient hospital stays of one night or longer, and PVS refers to overnight stays in a healthcare facility. While response categories vary, from binary yes/no options, to counts of hospitalisations and a “not sure” option in PaRIS, all three surveys were aligned to a harmonised indicator distinguishing between any hospitalisation and none (Table 6). These results are therefore broadly comparable, although small differences in question framing and response options should be considered when interpreting results and cross‑survey differences.

Table 6. Harmonisation of response categories for hospitalisation

	PaRIS	EHIS	PVS
Question	In the last 12 months, have you been in a hospital for one night or longer?	In the past 12 months have you been in hospitals as an inpatient, that is overnight or longer?	In the past 12 months did you stay overnight at a healthcare facility as a patient?
Response categories (in original order)	1. No 2. Yes, once 3. Yes, twice 4. Yes, 3 or more times 5. Not sure	1. Yes 2. No	1. No 2. Yes
Recoding	Hospitalisation: Yes <- Yes, once OR yes, twice OR yes, 3 or more times; Hospitalisation: No <- No; NA <- Not sure	Hospitalisation: Yes <- Yes; Hospitalisation: No <- No	Hospitalisation: Yes <- Yes; Hospitalisation: No <- No
Final reported outcome	Percentage of people reporting being hospitalised in the past 12 months

Figure 6 shows that cross‑survey country gradients for hospitalisation in the past 12 months are generally consistent across PaRIS, EHIS and PVS, but PaRIS reports higher absolute levels, with differences widening after restriction and standardisation. These patterns reflect both population coverage, PaRIS focusses on service users, many with chronic conditions, and survey framing, as EHIS and PVS capture hospitalisation in the general population.

Figure 6. In general, PaRIS respondents reported higher hospitalisation in the last 12 months compared to other surveys, the gap generally narrowed down after strict comparison

Note. The data are sorted from largest to smallest based on the PaRIS results.

In the overall comparison (Panel A), PaRIS consistently reports higher hospitalisation rates than EHIS in all overlapping countries. Differences are particularly large in Romania (+25 p.p.) followed by Italy (+13 points) and go down to +4 points in Slovenia and Iceland. Comparisons with PVS also show higher hospitalisation in PaRIS in most overlapping countries, with gaps of +15 points in Romania, and +12 points in Italy to +7 points in Greece and the United States, indicating greater reported hospital use among PaRIS respondents in the crude data.

In the standardised and harmonised comparison among service users with chronic conditions (Panel B), hospitalisation levels increase across all surveys, but PaRIS remains higher than EHIS in nearly all countries, except Iceland where EHIS respondents are 1 p.p. higher than PaRIS. Differences vary from +18 p.p. in Romania and +11 points in Italy to +1 point in Slovenia and the Netherlands. Comparisons with PVS show a similar pattern where more respondents in PaRIS report hospitalisation in Romania (+10 p.p.), Italy (+6 points), Greece (+5 points) and the United States (+3 points).

Overall, the results indicate that hospitalisation is sensitive to restriction and harmonisation, with PaRIS consistently capturing higher use of inpatient care than EHIS and PVS. Even after restricting the comparison to service users with chronic conditions and standardising to the PaRIS population, hospitalisation in most countries remains higher in PaRIS than in EHIS. This likely reflects both a higher-risk population and the fact that PaRIS respondents are already connected to primary care and an identified provider. In this context, hospital use may be more frequent not only because needs are greater, but also because primary care is more likely to detect deterioration, initiate referrals, and connect patients to tests, specialist care or treatments that are still often delivered in hospitals (Lyall et al., 2022[29]). Hospital admission may also itself increase subsequent primary care use, since discharge commonly requires closer follow-up, medication review and monitoring to prevent complications that may lead to readmission (Moneme et al., 2023[30]). In settings with structural unmet need, contact with primary care may therefore uncover previously unaddressed conditions and generate further hospital-based care. Romania may illustrate this pattern. Once Romania is excluded, however, cross-country dispersion becomes very similar across PaRIS and EHIS, suggesting that the main difference is not greater system sensitivity in PaRIS, but a population that is both more clinically complex and more connected to care pathways through which hospital need is identified and managed.