Psychometric performance of the OECD Patient-Reported Indicator Surveys Patient Questionnaire (PaRIS-PQ) among adults with chronic conditions in Norway.
Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation · 2026
Abstract originale
PurposeThe OECD Patient-Reported Indicator Surveys (PaRIS) initiative aims to provide standardised, internationally comparable patient-reported indicators for people living with chronic conditions receiving primary or generalist ambulatory care. This study evaluated the psychometric performance of the multi-item scales in the Norwegian PaRIS Patient Questionnaire (PaRIS-PQ).MethodsData were drawn from the 2023 Norwegian PaRIS Main Survey. Respondents aged ≥ 45 years with at least one self-reported chronic condition were included. Measures were PROMIS Global Physical Health, PROMIS Global Mental Health, and WHO-5 Well-Being Index as patient-reported outcome measures (PROMs); the Person-Centred Coordinated Care Experience Questionnaire (P3CEQ) as a patient-reported experience measure (PREM); and the Porter-Novelli Scale (PNS) and self-management confidence as capability-related measures. Analyses assessed structural validity, internal consistency reliability, measurement invariance, item functioning, and construct validity.ResultsOf 8,683 respondents, 6,691 reported at least one chronic condition. PROM scales performed best, with acceptable internal consistency (α = 0.80-0.89) and strongest confirmatory support for PROMIS Global Physical Health (CFI = 0.992, RMSEA = 0.080). The PREM (P3CEQ) showed weaker performance, with suboptimal fit for the total scale (CFI = 0.818, RMSEA = 0.132), limited scalar invariance, and low internal consistency for the care coordination subscale (α = 0.61). Capability-related measures showed mixed performance. Metric invariance was supported across all tested models, whereas scalar invariance was less consistent. Construct validity findings were consistent with expectations.ConclusionPROM scales showed the most consistent psychometric support. PREM and capability-related measures were more heterogeneous and require cautious, scale-specific interpretation, especially in subgroup comparisons and when used as latent measures. Evaluation and refinement of weaker experience- and capability-related measures is warranted.
Parole chiave
Leggi il testo integrale originale
Testo estratto dalla pubblicazione: sono esclusi immagini, media, supplementi e parti soggette a diritti distinti.
Psychometric performance of the OECD Patient-Reported Indicator Surveys Patient Questionnaire (PaRIS-PQ) among adults with chronic conditions in Norway
Abstract
Purpose
The OECD Patient-Reported Indicator Surveys (PaRIS) initiative aims to provide standardised, internationally comparable patient-reported indicators for people living with chronic conditions receiving primary or generalist ambulatory care. This study evaluated the psychometric performance of the multi-item scales in the Norwegian PaRIS Patient Questionnaire (PaRIS-PQ).
Methods
Data were drawn from the 2023 Norwegian PaRIS Main Survey. Respondents aged ≥ 45 years with at least one self-reported chronic condition were included. Measures were PROMIS Global Physical Health, PROMIS Global Mental Health, and WHO-5 Well-Being Index as patient-reported outcome measures (PROMs); the Person-Centred Coordinated Care Experience Questionnaire (P3CEQ) as a patient-reported experience measure (PREM); and the Porter–Novelli Scale (PNS) and self-management confidence as capability-related measures. Analyses assessed structural validity, internal consistency reliability, measurement invariance, item functioning, and construct validity.
Results
Of 8,683 respondents, 6,691 reported at least one chronic condition. PROM scales performed best, with acceptable internal consistency (α = 0.80–0.89) and strongest confirmatory support for PROMIS Global Physical Health (CFI = 0.992, RMSEA = 0.080). The PREM (P3CEQ) showed weaker performance, with suboptimal fit for the total scale (CFI = 0.818, RMSEA = 0.132), limited scalar invariance, and low internal consistency for the care coordination subscale (α = 0.61). Capability-related measures showed mixed performance. Metric invariance was supported across all tested models, whereas scalar invariance was less consistent. Construct validity findings were consistent with expectations.
Conclusion
PROM scales showed the most consistent psychometric support. PREM and capability-related measures were more heterogeneous and require cautious, scale-specific interpretation, especially in subgroup comparisons and when used as latent measures. Evaluation and refinement of weaker experience- and capability-related measures is warranted.
Supplementary Information
The online version contains supplementary material available at https://doi.org/10.1007/s11136-026-04429-z.
Keywords: Psychometrics, Primary Health Care, Chronic Disease, Patient-reported experience measures, Patient-reported outcome measures, Surveys and Questionnaires
Background
Health systems worldwide face increasing pressures from ageing populations, rising multimorbidity and the growing number of people living with chronic conditions [1]. Chronic conditions substantially affect functioning, quality of life, and account for a major share of health-care utilisation and expenditures across countries [2]. Many countries are therefore redesigning health systems towards more people-centred and integrated care, with primary care playing a pivotal role in prevention, long-term management, coordination, continuity and self-management support [3]. Despite this role, comprehensive and comparable information on how people with chronic conditions experience their care, and how well services support their functioning, well-being and ability to manage their health, remains limited [4].
Patient-reported outcome measures (PROMs) and patient-reported experience measures (PREMs) complement clinical and administrative data by capturing people’s functioning, well-being, capabilities and experiences of care [5]. However, many existing measures focus on single diseases, specific interventions or isolated encounters and may fail to capture central elements of person-centred and coordinated care for individuals receiving services across multiple providers [2, 4, 6]. This limits the coverage and comparability of patient-reported indicators within and across health systems.
To address these gaps, the Organisation for Economic Co-operation and Development (OECD) launched the Patient-Reported Indicator Surveys (PaRIS), which aim to strengthen international healthcare performance measurement by focusing on outcomes and care experiences among people living with chronic conditions [2]. PaRIS comprises a patient questionnaire and a primary care practice questionnaire designed to support cross-country comparison of patient-reported indicators among adults aged 45 years and older receiving primary or generalist ambulatory care in light of practice characteristics [6–8]. Patient-level information from the PaRIS Patient Questionnaire (PaRIS-PQ) [9] is complemented by organisational and provider-reported data from the PaRIS Primary Care Practice Questionnaire (PaRIS-PCPQ) [8]. The PaRIS conceptual framework links patient-reported outcomes and experiences to characteristics of primary care delivery, health system design, policy and context, enabling analyses of how practice- and system-level factors are associated with patients’ outcomes and experiences across countries and population subgroups [6].
The PaRIS-PQ integrates established scales with targeted items covering remaining subdomains of the PaRIS framework [9]. The OECD flagship report summarises core domains through the “PaRIS-10” key indicators: physical, mental and social functioning, well-being, general health, confidence to self-manage, experienced coordination, person-centred care, experienced quality and trust [2]. The usefulness of these indicators depends on the psychometric robustness of the underlying multi-item scales from which several are constructed.
International development and Field Trial evaluation of the PaRIS-PQ have provided initial support for the questionnaire, including acceptable patient-level reliability for most measures [9, 10]. However, more restrictive forms of cross-country measurement invariance were not fully supported, and the Field Trial findings informed revisions to the questionnaire before implementation of the Main Survey [9, 10]. The Field Trial therefore provided an important foundation but did not establish the measurement properties of the final questionnaire within each national survey population.
Norway participated in both the Field Trial and the 2023 Main Survey. In Norway, the national general practitioner (GP) registry was used to establish a provider-linked, registry-informed sampling design among people with chronic conditions. National findings were comparatively favourable for patients’ well-being and trust in the health system, but weaker for confidence in managing their own health and experienced coordination [11]. Previous analyses of Norwegian data also showed potential selection effects related to GP participation [12].
Even when measures are developed within an international survey programme, psychometric evaluation in new national and linguistic contexts remains important because measures translated or transferred across settings may not retain equivalent dimensionality, reliability, or measurement invariance [13]. The measurement properties of the multi-item scales in the Norwegian PaRIS-PQ have not yet been evaluated using Main Survey data. Such evaluation is therefore important to support interpretation of Norwegian PaRIS results and comparisons across population subgroups.
Based on the Norwegian PaRIS Main Survey data, the aim of this study was to evaluate the psychometric performance of the multi-item scales included in the Norwegian PaRIS-PQ among persons aged 45 years and older with one or more chronic conditions. Psychometric evaluation included structural validity, internal consistency reliability, construct validity, measurement invariance across key sociodemographic groups, and item-level performance. The findings may inform scale-specific interpretation of Norwegian PaRIS results, subgroup comparisons and further refinement of PaRIS measures.
Methods
Data collection and sample
The present study was a psychometric evaluation using cross-sectional data from the 2023 Norwegian PaRIS Main Survey, conducted as part of the OECD PaRIS [2, 9]. PaRIS is designed to assess health outcomes and care experiences of primary care users, with a focus on people with chronic conditions. PaRIS includes respondents aged 45 years and older who had at least one contact with their primary care practice in the past six months. Given its design feature, the present psychometric analyses were restricted to respondents reporting at least one chronic condition.
Data collection followed the PaRIS study protocol [7]. In Norway, the national GP Registry, which covers all regular GPs and almost the entire population, was used for the first stage of the provider-linked sampling procedure [11, 14]. A random sample of 250 GPs was invited to participate in PaRIS and complete the PaRIS-PCPQ. Of these, 121 responded; three were excluded because they had no or only one eligible patient or had stopped practising, leaving 118 participating GPs. GP questionnaire data were not included in the present psychometric analyses; GP participation served to define the sampling frame for the patient sample. In the second stage, eligible patients were identified using the Norwegian Control and Payment of Health Reimbursements Database (KUHR) [15] as those listed with a participating GP, aged 45 years or older, and with at least one consultation during the six months preceding sample selection.
Patient data collection took place in May–June 2023. Included patients (n = 19,637) were invited using a secure survey link and unique access code, primarily through Helsenorge [16]. Up to three SMS reminders were sent to digital non-respondents. Patients not registered for digital communication (12.3%) received a postal invitation and one postal reminder, both including a Norwegian paper questionnaire. Additional registry data from KUHR, the GP Register and Statistics Norway were linked to the survey data.
Questionnaire and multi-item scales
The development of the PaRIS-PQ has been described elsewhere [9]. In brief, candidate measures were mapped to the PaRIS conceptual framework, appraised using the Evaluating the Measurement of Patient-Reported Outcomes (EMPRO) tool [17], refined through a modified Delphi process, and selected to ensure conceptual coverage while limiting respondent burden [6, 7, 9].
Existing Norwegian translations were incorporated for the Patient-Reported Outcomes Measurement Information System (PROMIS) Global Health item set, the WHO-5 Well-Being Index (WHO-5), and several original items from the Person-Centred Coordinated Care Experience Questionnaire (P3CEQ). The remaining content evaluated in the present study was translated or adapted through the centralised PaRIS translation process, using the Translation, Review, Adjudication, Pretesting and Documentation (TRAPD) approach to support conceptual equivalence across countries while allowing adaptation to the national context [18]. The assembled Norwegian PaRIS-PQ was cognitively tested in 20 interviews prior to field implementation. Because of the questionnaire length, cognitive testing used three separate interview protocols and did not assess the full patient questionnaire within a single interview. In the Main Survey, the PaRIS-PQ was available in the two official written standards of Norwegian, as well as English [11].
The present psychometric evaluation included the multi-item scales in the Norwegian PaRIS-PQ, summarised in Supplementary Table S1. The full 10-item PROMIS Global Health item set was administered. It comprises two four-item summary scales, PROMIS Global Physical Health (PROMIS GPH) and PROMIS Global Mental Health (PROMIS GMH), originally scored by averaging the four constituent items within each scale, while the two remaining items, general health and satisfaction with social roles, are recommended for separate reporting [19]. As the present study evaluated multi-item scales, only PROMIS GPH and PROMIS GMH were included in the psychometric analyses. The WHO-5 [20], the PaRIS version of the P3CEQ [21], and the 10-item Porter–Novelli Scale (PNS) [22] were administered in their entirety. For confidence in self-management, PaRIS included four items derived from the Medicare Current Beneficiary Survey (MCBS) [23] and one item from the eHealth Literacy Scale (eHEALS) [24]; the full eHEALS was not administered. Although originating from different measures, these items were evaluated together in the PaRIS Field Trial as a measure of self-management confidence within the broader health and healthcare capabilities domain [9, 10]. The MCBS items assess confidence in performing specific self-management tasks, whereas the eHEALS item assesses confidence in using online health information for health-related decision-making [23, 24]. Given their different source measures and content, the five-item MCBS + eHEALS specification was evaluated exploratorily alongside a four-item MCBS-only specification.
Single-item PaRIS measures were not designed as psychometric scales and were therefore not evaluated. The PROMIS GPH, PROMIS GMH and WHO-5 were treated as PROMs; the P3CEQ as a PREM; and the PNS and self-management confidence item set as measures within the health and healthcare capabilities domain [6, 9, 10].
Statistical analysis
In line with the OECD PaRIS approach [9], the psychometric analyses evaluated reflective measurement specifications for the multi-item scales, item sets and empirically identified factors. Reflective models were prespecified for PROMIS GPH, PROMIS GMH and WHO-5, and for the proposed P3CEQ subscales [19–21]. The P3CEQ total scale was examined as a candidate reflective specification based on PaRIS Field Trial findings, whereas PNS factors and both self-management confidence specifications were evaluated exploratorily following empirical examination of dimensionality [9, 10]. In line with current COSMIN guidance, factor-analytic and item response theory (IRT) findings were interpreted as evidence of structural validity, while internal consistency and measurement invariance were interpreted within the same reflective measurement framework [25].
Structural validity was evaluated using confirmatory factor analysis (CFA) for scales with well-established hypothesised structures, and exploratory factor analysis (EFA) followed by CFA for scales with less certain dimensionality. EFA and CFA were conducted using the same dataset, without splitting the sample into separate exploratory and confirmatory subsets. PROMIS GPH and PROMIS GMH were evaluated primarily using CFA, consistent with the original validation study [19], and WHO-5 was likewise evaluated as a unidimensional scale [20]. For these prespecified PROM scales, EFA was used only to descriptively check whether item correlations supported the expected structures, whereas for P3CEQ, PNS and the self-management confidence item sets, EFA was used to examine the number and interpretation of factors before CFA models were specified.
The P3CEQ was originally conceptualised as comprising two related subdomains, Person-Centredness (P3CEQ-PC) and Care Coordination (P3CEQ-CC) [21]. However, PaRIS Field Trial analyses indicated high intercorrelations between the subdomains and a single dominant factor [9]. Therefore, both the total scale and proposed subscales were examined. For the PNS, Field Trial analyses identified two dimensions, active engagement and reliance on health professionals [9, 10]. EFA preceded confirmatory testing in the Norwegian sample. For self-management confidence, PaRIS Field Trial analyses operationalised a broader construct combining P3CEQ10, MCBS items and eHEALS [9]; here, P3CEQ10 was retained within the P3CEQ analyses, while the MCBS + eHEALS and MCBS-only specifications were examined separately.
EFA used principal axis factoring on Pearson correlation matrices with pairwise complete observations. Factorability was assessed using the Kaiser–Meyer–Olkin (KMO) statistic, with values ≥ 0.70 considered desirable, and Bartlett’s test of sphericity was required to be statistically significant [26–28]. Factor retention was guided by eigenvalues, scree plots and parallel analysis [28]. Items with primary loadings below 0.40 or cross-loadings above 0.30 were interpreted as showing limited factorial specificity [29]. Internal consistency reliability was assessed using Cronbach’s alpha (α), with values ≥ 0.70 considered acceptable [30].
CFA was used to test the prespecified or EFA-informed measurement models. Item responses were specified as ordered categorical and models were estimated using weighted least squares with mean and variance adjustment (WLSMV). Model fit was evaluated using the mean- and variance-adjusted χ² statistic, Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardised Root Mean Square Residual (SRMR). Because χ² is sensitive to large samples, model fit was assessed primarily using approximate fit indices. CFI and TLI ≥ 0.95, RMSEA < 0.07, and SRMR < 0.05 were used as reference criteria of good fit [31, 32]. For models with few degrees of freedom, RMSEA may be overly sensitive; fit indices were therefore interpreted jointly and considered alongside standardised factor loadings [33, 34].
Measurement invariance was evaluated for the multi-item scales and subscales retained for subgroup comparisons after the initial structural validity assessment. Multigroup CFA was conducted across gender, age group and educational level using the same WLSMV specification. For the ordered-categorical indicators, invariance was tested sequentially using configural, threshold, metric and scalar models, with increasingly restrictive equality constraints across groups. Model fit was assessed using the fit indices and criteria described above. Invariance was considered supported when decreases in CFI were ≤ 0.010 and increases in RMSEA were ≤ 0.015 relative to the less constrained model [35–37]. Metric invariance was interpreted as supporting comparisons of associations across groups, whereas scalar invariance was considered important for comparing latent means and interpreting observed subgroup differences.
Item-level performance was examined using IRT. Factor-analytic and IRT approaches were used to evaluate complementary aspects of psychometric performance: factor analyses assessed dimensionality and structural validity, while IRT assessed item-level functioning and measurement precision. For scales with EFA/CFA evidence of a dominant underlying dimension, graded response models (GRM) for ordered categorical items were estimated using marginal maximum likelihood; where structural support was mixed, IRT was applied exploratorily [38]. The GRM was selected because it accommodates ordered categorical responses and permits discrimination to vary across items. Discrimination (a), thresholds (b), global fit using the limited-information C2 statistic with RMSEA, Standardised Root Mean Square Residual (SRMSR) and CFI, and item fit using S-X² and item-level RMSEA were assessed [39–41]. Discrimination parameters were interpreted descriptively as very low (0.01–0.34), low (0.35–0.64), moderate (0.65–1.34), high (1.35–1.69), or very high (≥ 1.70); these categories were treated as approximate guidelines rather than strict quality criteria [42]. Local item dependence was examined using Yen’s Q3 residual correlations. Adjusted Q3 values were calculated by subtracting the mean off-diagonal Q3 from each item-pair residual correlation [43, 44]. Adjusted Q3 values above 0.20, based on unrounded estimates, were used as a heuristic screening value for possible local item dependence rather than as an absolute statistical criterion; detailed results are provided in Supplementary Table S2. Test information curves, item information curves, and category probability curves were inspected to evaluate measurement precision, information distribution, latent trait coverage, and response category functioning. An overview of the test information curves for all IRT models is provided in Supplementary Figure S1, and model-specific test information, item information, and category probability curves are provided in Supplementary Figures S2–S11.
Construct validity was assessed by testing a priori hypotheses regarding the magnitude of correlations among theoretically related and distinct scale scores, consistent with COSMIN guidance [25], the PaRIS conceptual framework and previous Field Trial analyses [6, 9, 10]. Pearson correlations were calculated between theoretically related outcome, experience, and capability-related domains, with Spearman correlations used as sensitivity analyses. PROMs were expected to correlate moderately to strongly with each other. PROM–PREM correlations were expected to be weak to moderate, PROM–capability correlations to be weak, and PREM–capability correlations to be weak to moderate.
All analyses were conducted in R version 4.2.3, using psych for EFA and internal consistency analyses, lavaan and semTools for CFA and measurement invariance testing, and mirt for IRT analyses.
Results
Sample
Of 8,683 respondents (response rate 44.2%), 8,536 had valid data on chronic conditions; of these, 6,691 (78.4%) reported at least one chronic condition and were included in the analyses (Table 1).
Table 1Respondent characteristics of the analytical sample (≥1 chronic condition)
n %
Gender (n= 6,691)
Women 3,667 54.8
Men 3,024 45.2
Age group (years) (n= 6,691)
45–54 1,280 19.1
55–69 2,929 43.8
70–79 1,685 25.2
≥80 797 11.9
Number of chronic conditions (n= 6,691)
1 2,969 44.4
2 2,088 31.2
≥3 1,634 24.4
Education level (n= 6,653)*
Low 1,079 16.2
Medium 3,133 47.1
High 2,441 36.7
* Education level was classified according to Statistics Norway: low = primary and lower secondary education, including no completed education; medium = upper secondary and post-secondary vocational education; high = university or college education
Factor analysis and internal consistency
EFA and reliability results for the P3CEQ, PNS and self-management confidence specifications are shown in Tables 2, 3, 4 and CFA results for all scales are shown in Table 5. For PROMIS GPH, PROMIS GMH and WHO-5, descriptive EFAs were compatible with the expected structures, and internal consistency was acceptable (α = 0.80–0.89) (results not shown). CFA results were strongest for PROMIS GPH, with CFI, TLI and SRMR meeting the reference criteria, while RMSEA was slightly above the criterion. PROMIS GMH and WHO-5 showed broadly supportive but mixed fit, with CFI and SRMR meeting the criteria but RMSEA and TLI not. Standardised factor loadings were moderate to strong across the three PROM models (0.66–0.90).
Table 2Exploratory factor analysis and reliability statistics for the P3CEQ total scale, P3CEQ-PC subscale, and P3CEQ-CC subscale
Person-Centred Coordinated Care Experience Questionnaire P3CEQ total scaleCronbach’s α = 0.728 P3CEQ-PC subscaleCronbach’s α = 0.745 P3CEQ-CC subscaleCronbach’s α = 0.606
Factor loading Corrected item–total correlation Cronbach’s α if item deleted Factor loading Corrected item–total correlation Cronbach’s α if item deleted Factor loading Corrected item–total correlation Cronbach’s α if item deleted
P3CEQ 1 Discussion of what matters most 0.550 0.488 0.695 0.515 0.445 0.718
P3CEQ 2 Involvement in care decisions 0.655 0.504 0.698 0.675 0.554 0.704
P3CEQ 3 Being treated as a whole person 0.659 0.505 0.698 0.683 0.552 0.704
P3CEQ 4 Having to repeat information 0.441 0.320 0.716 0.468 0.375 0.730
P3CEQ 5 Care organised to meet patient needs 0.714 0.571 0.690 0.719 0.588 0.698 0.548 0.407 0.560
P3CEQ 6 Single professional coordinating care 0.352 0.335 0.728 0.444 0.347 0.571
P3CEQ 7 Care plan addressing patient needs 0.331 0.330 0.724 0.449 0.361 0.555
P3CEQ 8 Support for self-management 0.520 0.442 0.697 0.516 0.466 0.718 0.530 0.352 0.557
P3CEQ 9 Timely and useful self-management information 0.481 0.455 0.694 0.439 0.395 0.734 0.577 0.432 0.517
P3CEQ 10 Confidence in managing own health 0.331 0.247 0.725 0.354 0.305 0.742
Abbreviations: EFA exploratory factor analysis, P3CEQ Person-Centred Coordinated Care Experience Questionnaire, P3CEQ-PC P3CEQ Person-Centredness subscale, P3CEQ-CC P3CEQ Care Coordination subscale
Total analytic sample = 6,691. EFAs were based on pairwise complete observations; harmonic N was 6,198 for the P3CEQ total scale, 6,122 for the P3CEQ-PC subscale, and 6,436 for the P3CEQ-CC subscale
Table 3Exploratory factor analysis and reliability statistics for the PNS
Porter-Novelli Scale PNS Factor 1: Active engagementCronbach’s α = 0.738 PNS Factor 2: Health information comprehensionCronbach’s α = 0.795 PNS Factor 3: Reliance and cooperation with health professionalsCronbach’s α = 0.581
Factor loadings Corrected item–total correlation Cronbach’s α if item deleted Factor loadings Corrected item–total correlation Cronbach’s α if item deleted Factor loadings Corrected item–total correlation Cronbach’s α if item deleted
PNS 1 Reliance on professionals for health information 0.573 0.373 0.502
PNS 2 Perceived complexity of health issues 0.821 0.660 0.643
PNS 3 Active disease prevention 0.473 0.438 0.733
PNS 4 Reliance on professionals for health decisions 0.448 0.296 0.569
PNS 5 Importance of health information 0.725 0.565 0.657
PNS 6 Health information for oneself and family 0.831 0.606 0.633
PNS 7 Difficulty understanding written health information 0.829 0.660 0.678
PNS 8 Working with professionals to manage health 0.681 0.486 0.417
PNS 9 Discussing relevant information with professionals 0.341 0.315 0.545
PNS 10 Understanding personal health risks 0.598 0.526 0.685
Abbreviations: EFA exploratory factor analysis, PNS Porter–Novelli Scale
Total analytic sample = 6,691. EFA was based on pairwise complete observations; harmonic N = 6,634
Table 4Exploratory factor analysis and reliability statistics for MCBS items combined with the eHEALS item and for the MCBS items alone
Patients’ confidence in self-management MCBS + eHEALSCronbach’s α = 0.772 MCBSCronbach’s α = 0.788
Factor loadings Corrected item–total correlation Cronbach’s α if item deleted Factor loadings Correcteditem–total correlation Cronbach’s α if item deleted
MCBS 1 Confidence in following home-care instructions 0.758 0.619 0.705 0.787 0.649 0.709
MCBS 2 Confidence in following lifestyle-change instructions 0.719 0.588 0.716 0.743 0.615 0.726
MCBS 3 Confidence in recognising need for medical care 0.704 0.612 0.707 0.684 0.608 0.729
MCBS 4 Confidence in recognising medication side effects 0.613 0.546 0.731 0.578 0.516 0.777
eHEALS Confidence in using online health information for decisions 0.407 0.369 0.788
Abbreviations: EFA exploratory factor analysis, MCBS Medicare Current Beneficiary Survey, eHEALS eHealth Literacy Scale
Total analytic sample = 6,691. EFAs were based on pairwise complete observations; harmonic N was 5,801 for the MCBS + eHEALS specification and 6,264 for the MCBS-only specification
Table 5Confirmatory factor analysis for all scales and subscales
Model Chi-square (χ²) df p CFI TLI RMSEA SRMR Standardised factor loadings
PROMIS GPH (n = 6688) 57.69 2 < 0.001 0.992 0.977 0.080 0.015
PROMIS GPH 1 Physical health 0.818
PROMIS GPH 2 Everyday physical functioning 0.823
PROMIS GPH 3 Average fatigue 0.752
PROMIS GPH 4 Average pain 0.663
PROMIS GMH (n = 6689) 293.01 2 < 0.001 0.973 0.919 0.167 0.028
PROMIS GMH 1 Quality of life 0.785
PROMIS GMH 2 Mental health 0.867
PROMIS GMH 3 Satisfaction with social activities and relationships 0.822
PROMIS GMH 4 Emotional problems 0.719
WHO-5 (n = 6686) 829.95 5 < 0.001 0.960 0.920 0.167 0.030
WHO-5 1 Feeling cheerful and in good spirits 0.903
WHO-5 2 Feeling calm and relaxed 0.863
WHO-5 3 Feeling active and vigorous 0.813
WHO-5 4 Interest in daily life 0.806
WHO-5 5 Feeling refreshed on waking 0.765
P3CEQ total scale (n = 6691) 2015.57 35 < 0.001 0.818 0.766 0.132 0.078
P3CEQ 1 Discussion of what matters most 0.623
P3CEQ 2 Involvement in care decisions 0.753
P3CEQ 3 Being treated as a whole person 0.781
P3CEQ 4 Having to repeat information 0.500
P3CEQ 5 Care organised to meet patient needs 0.807
P3CEQ 6 Single professional coordinating care 0.451
P3CEQ 7 Care plan addressing patient needs 0.458
P3CEQ 8 Support for self-management 0.572
P3CEQ 9 Timely and useful self-management information 0.555
P3CEQ 10 Confidence in managing own health 0.375
P3CEQ-PC subscale (n = 6690) 1075.21 20 < 0.001 0.895 0.853 0.120 0.058
P3CEQ 1 Discussion of what matters most 0.598
P3CEQ 2 Involvement in care decisions 0.768
P3CEQ 3 Being treated as a whole person 0.794
P3CEQ 4 Having to repeat information 0.524
P3CEQ 5 Care organised to meet patient needs 0.804
P3CEQ 8 Support for self-management 0.577
P3CEQ 9 Timely and useful self-management information 0.519
P3CEQ 10 Confidence in managing own health 0.400
P3CEQ-CC subscale (n = 6680) 263.16 5 < 0.001 0.891 0.783 0.147 0.057
P3CEQ 5 Care organised to meet patient needs 0.665
P3CEQ 6 Single professional coordinating care 0.540
P3CEQ 7 Care plan addressing patient needs 0.583
P3CEQ 8 Support for self-management 0.595
P3CEQ 9 Timely and useful self-management information 0.630
PNS three-factor model (n = 6684) 4088.07 32 < 0.001 0.891 0.847 0.107 0.085
PNS 1 Reliance on professionals for health information 0.466
PNS 2 Perceived complexity of health issues 0.863
PNS 3 Active disease prevention 0.590
PNS 4 Reliance on professionals for health decisions 0.207
PNS 5 Importance of health information 0.789
PNS 6 Health information for oneself and family 0.810
PNS 7 Difficulty understanding written health information 0.846
PNS 8 Working with professionals to manage health 0.839
PNS 9 Discussing relevant information with professionals 0.618
PNS 10 Understanding personal health risks 0.746
MCBS + eHEALS (n = 6680) 1073.52 5 < 0.001 0.884 0.768 0.214 0.069
MCBS 1 Confidence in following home-care instructions 0.868
MCBS 2 Confidence in following lifestyle-change instructions 0.823
MCBS 3 Confidence in recognising need for medical care 0.722
MCBS 4 Confidence in recognising medication side effects 0.652
eHEALS Confidence in using online health information for decisions 0.441
MCBS (n = 6678) 839.64 2 < 0.001 0.904 0.713 0.290 0.067
MCBS 1 Confidence in following home-care instructions 0.874
MCBS 2 Confidence in following lifestyle-change instructions 0.828
MCBS 3 Confidence in recognising need for medical care 0.716
MCBS 4 Confidence in recognising medication side effects 0.636
Abbreviations: χ² chi-square statistic, df degrees of freedom, CFI Comparative Fit Index, TLI Tucker–Lewis Index, RMSEA Root Mean Square Error of Approximation, SRMR Standardised Root Mean Square Residual, PROMIS GPH Patient-Reported Outcomes Measurement Information System Global Physical Health, PROMIS GMH Patient-Reported Outcomes Measurement Information System Global Mental Health, WHO-5 WHO-5 Well-Being Index, P3CEQ Person-Centred Coordinated Care Experience Questionnaire, P3CEQ-PC P3CEQ Person-Centredness subscale, P3CEQ-CC P3CEQ Care Coordination subscale, PNS Porter–Novelli Scale, MCBS Medicare Current Beneficiary Survey, eHEALS eHealth Literacy Scale, WLSMV weighted least squares with mean and variance adjustment
Models were estimated using WLSMV with indicators treated as ordered categorical; the mean- and variance-adjusted χ² statistic and robust CFI, TLI and RMSEA are reported
For the P3CEQ total scale, P3CEQ-PC subscale and P3CEQ-CC subscale, EFA factorability was acceptable to good (KMO = 0.72–0.84; Bartlett’s tests all p < 0.001). Eigenvalues suggested two factors for the P3CEQ total scale, but the two-factor solution lacked conceptual clarity and showed substantial cross-loadings. Consistent with the PaRIS Field Trial [9], a one-factor solution was retained, explaining 27% of the variance (α = 0.73), with three items loading below 0.40. For the P3CEQ-PC subscale, evidence for a second factor was limited, and the subscale was treated as unidimensional (31% variance explained; α = 0.75). For the P3CEQ-CC subscale, EFA supported a one-factor solution, but internal consistency was below criterion (26% variance explained; α = 0.61). Corresponding CFA results showed suboptimal fit for the P3CEQ total scale and both subscales; none of the models met the prespecified global fit criteria.
The PNS items showed acceptable factorability (KMO = 0.75; Bartlett’s p < 0.001), and EFA supported a three-factor structure explaining 44% of the variance: active engagement, health information comprehension, and reliance and cooperation with health professionals. Internal consistency was acceptable for PNS Factors 1 and 2 (α = 0.74 and 0.80), but below criterion for PNS Factor 3 (α = 0.58). In CFA, the correlated three-factor model did not meet the prespecified global fit criteria, although standardised factor loadings suggested broadly interpretable dimensions.
The self-management confidence specifications showed acceptable EFA factorability (KMO = 0.76 and 0.73; Bartlett’s tests both p < 0.001), with one-factor solutions explaining 43% of the variance for MCBS + eHEALS and 49% for MCBS-only. Internal consistency was acceptable for both specifications (α = 0.77 and 0.79). In CFA, neither model provided clear confirmatory support for robust unidimensional measurement; MCBS item loadings were moderate to strong in both models, while the eHEALS item showed the weakest loading in the five-item model.
Measurement invariance
Measurement invariance results are shown in Table 6. Threshold and metric invariance criteria were met across all tested models and grouping variables, although absolute model fit was weaker for several P3CEQ, PNS Factor 3 and self-management confidence models.
Table 6Fit indices for measurement invariance models for all scales across gender, age group, and educational level
Chi-square (χ²) df CFI RMSEA SRMR ΔCFI ΔRMSEA
PROMIS GPH
Gender (n = 6688)
Configural 55.93 4 0.997 0.062 0.015 – –
Threshold 68.67 12 0.997 0.038 0.015 0.000 −0.025
Metric 83.57 15 0.997 0.037 0.018 −0.001 −0.001
Scalar 129.22 18 0.995 0.043 0.018 −0.002 0.006
Age groups (n = 6688)
Configural 62.27 8 0.998 0.064 0.015 – –
Threshold 100.89 32 0.997 0.036 0.015 −0.001 −0.028
Metric 138.74 41 0.996 0.038 0.018 −0.001 0.002
Scalar 475.71 50 0.981 0.071 0.023 −0.015 0.034
Educational level (n = 6650)
Configural 72.91 6 0.996 0.071 0.018 – –
Threshold 90.49 22 0.996 0.037 0.018 0.000 −0.033
Metric 111.30 28 0.996 0.037 0.020 −0.001 −0.001
Scalar 140.75 34 0.994 0.038 0.021 −0.001 0.001
PROMIS GMH
Gender (n = 6689)
Configural 303.22 4 0.991 0.150 0.029 – –
Threshold 243.31 12 0.993 0.076 0.029 0.002 −0.074
Metric 236.24 15 0.993 0.066 0.030 0.000 −0.010
Scalar 353.59 18 0.990 0.075 0.029 −0.003 0.008
Age groups (n = 6689)
Configural 291.57 8 0.992 0.146 0.028 – –
Threshold 284.39 32 0.992 0.069 0.028 0.001 −0.077
Metric 279.40 41 0.993 0.059 0.029 0.000 −0.010
Scalar 571.02 50 0.985 0.079 0.029 −0.008 0.020
Educational level (n = 6651)
Configural 295.91 6 0.991 0.148 0.029 – –
Threshold 285.18 22 0.992 0.073 0.029 0.001 −0.074
Metric 270.25 28 0.993 0.062 0.030 0.001 −0.011
Scalar 405.21 34 0.989 0.070 0.030 −0.004 0.008
WHO-5
Gender (n = 6686)
Configural 850.29 10 0.989 0.159 0.031 – –
Threshold 635.21 25 0.992 0.085 0.031 0.003 −0.073
Metric 634.73 29 0.992 0.079 0.032 0.000 −0.006
Scalar 687.50 33 0.991 0.077 0.032 −0.001 −0.002
Age groups (n = 6686)
Configural 870.56 20 0.989 0.160 0.033 – –
Threshold 751.08 62 0.991 0.082 0.033 0.002 −0.078
Metric 675.19 74 0.992 0.070 0.033 0.001 −0.012
Scalar 1085.45 86 0.987 0.083 0.033 −0.005 0.014
Educational level (n = 6648)
Configural 799.49 15 0.990 0.154 0.031 – –
Threshold 635.41 45 0.992 0.077 0.031 0.003 −0.077
Metric 591.12 53 0.993 0.068 0.031 0.001 −0.009
Scalar 720.01 61 0.991 0.070 0.031 −0.002 0.002
P3CEQ total scale
Gender (n = 6691)
Configural 2048.82 70 0.917 0.092 0.079 – –
Threshold 2116.65 78 0.915 0.088 0.079 −0.002 −0.004
Metric 1710.90 87 0.932 0.075 0.079 0.017 −0.014
Scalar 1957.08 94 0.922 0.077 0.080 −0.010 0.002
Age groups (n = 6691)
Configural 2070.70 140 0.920 0.091 0.080 – –
Threshold 2160.64 164 0.918 0.085 0.080 −0.003 −0.005
Metric 1778.84 191 0.935 0.071 0.081 0.017 −0.015
Scalar 2304.70 212 0.914 0.077 0.082 −0.021 0.006
Educational level (n = 6653)
Configural 2034.41 105 0.919 0.091 0.079 – –
Threshold 2134.21 121 0.915 0.087 0.079 −0.004 −0.004
Metric 1660.61 139 0.936 0.070 0.079 0.021 −0.016
Scalar 2150.01 153 0.916 0.077 0.081 −0.020 0.006
P3CEQ-PC subscale
Gender (n = 6690)
Configural 1112.22 40 0.948 0.090 0.059 – –
Threshold 1167.44 48 0.946 0.084 0.059 −0.002 −0.006
Metric 955.53 55 0.957 0.070 0.059 0.011 −0.014
Scalar 1039.36 62 0.953 0.069 0.060 −0.004 −0.001
Age groups (n = 6690)
Configural 1076.32 80 0.953 0.086 0.059 – –
Threshold 1144.42 104 0.951 0.077 0.059 −0.002 −0.009
Metric 991.06 125 0.959 0.064 0.060 0.008 −0.013
Scalar 1327.02 146 0.944 0.070 0.061 −0.015 0.005
Educational level (n = 6652)
Configural 1068.75 60 0.951 0.087 0.058 – –
Threshold 1151.73 76 0.947 0.080 0.058 −0.003 −0.007
Metric 915.90 90 0.960 0.064 0.059 0.012 −0.016
Scalar 1041.78 104 0.954 0.064 0.059 −0.005 −0.001
P3CEQ-CC subscale
Gender (n = 6680)
Configural 262.76 10 0.951 0.087 0.057 – –
Threshold 288.75 13 0.947 0.080 0.057 −0.004 −0.007
Metric 244.85 17 0.956 0.063 0.057 0.009 −0.016
Scalar 255.50 19 0.955 0.061 0.057 −0.002 −0.002
Age groups (n = 6680)
Configural 269.99 20 0.953 0.087 0.058 – –
Threshold 308.37 29 0.947 0.076 0.058 −0.006 −0.011
Metric 265.02 41 0.958 0.057 0.059 0.010 −0.019
Scalar 441.82 47 0.925 0.071 0.060 −0.032 0.014
Educational level (n = 6642)
Configural 260.25 15 0.953 0.086 0.057 – –
Threshold 305.77 21 0.946 0.078 0.057 −0.008 −0.008
Metric 249.25 29 0.958 0.059 0.057 0.012 −0.020
Scalar 454.90 33 0.920 0.076 0.063 −0.038 0.017
PNS Factor 1: Active engagement
Gender (n = 6680)
Configural 140.63 4 0.993 0.101 0.025 – –
Threshold 116.67 12 0.994 0.051 0.025 0.002 −0.050
Metric 100.69 15 0.995 0.041 0.025 0.001 −0.010
Scalar 106.19 18 0.995 0.038 0.025 0.000 −0.003
Age groups (n = 6680)
Configural 166.06 8 0.992 0.109 0.025 – –
Threshold 156.78 32 0.993 0.048 0.025 0.002 −0.060
Metric 143.61 41 0.995 0.039 0.026 0.001 −0.010
Scalar 173.74 50 0.993 0.039 0.025 −0.001 0.000
Educational level (n = 6642)
Configural 150.07 6 0.991 0.104 0.026 – –
Threshold 167.69 22 0.991 0.055 0.026 0.000 −0.049
Metric 154.49 28 0.992 0.045 0.027 0.001 −0.010
Scalar 165.65 34 0.992 0.042 0.027 0.000 −0.003
PNS Factor 3: Reliance and cooperation with health professionals
Gender (n = 6683)
Configural 229.88 4 0.960 0.130 0.041 – –
Threshold 265.57 12 0.955 0.080 0.041 −0.005 −0.050
Metric 225.36 15 0.963 0.065 0.042 0.008 −0.015
Scalar 465.14 18 0.921 0.086 0.043 −0.042 0.021
Age groups (n = 6683)
Configural 243.49 8 0.959 0.133 0.043 – –
Threshold 305.47 32 0.952 0.072 0.043 −0.007 −0.061
Metric 272.07 41 0.960 0.058 0.045 0.007 −0.013
Scalar 655.10 50 0.895 0.085 0.045 −0.065 0.027
Educational level (n = 6645)
Configural 222.95 6 0.962 0.128 0.041 – –
Threshold 280.03 22 0.955 0.073 0.041 −0.007 −0.055
Metric 270.43 28 0.957 0.063 0.044 0.003 −0.010
Scalar 730.26 34 0.878 0.096 0.044 −0.080 0.034
MCBS + eHEALS
Gender (n = 6680)
Configural 1067.22 10 0.959 0.178 0.069 – –
Threshold 1081.94 15 0.959 0.146 0.069 0.000 −0.032
Metric 849.80 19 0.968 0.114 0.069 0.009 −0.032
Scalar 810.59 23 0.970 0.101 0.069 0.002 −0.013
Age groups (n = 6680)
Configural 1093.97 20 0.959 0.179 0.069 – –
Threshold 1113.08 35 0.959 0.136 0.069 0.000 −0.044
Metric 866.52 47 0.969 0.102 0.070 0.010 −0.034
Scalar 880.43 59 0.969 0.091 0.070 0.000 −0.011
Educational level (n = 6642)
Configural 1078.10 15 0.958 0.179 0.070 – –
Threshold 1102.40 25 0.957 0.140 0.070 −0.001 −0.039
Metric 858.60 33 0.967 0.106 0.071 0.010 −0.033
Scalar 922.09 41 0.965 0.099 0.072 −0.002 −0.008
MCBS
Gender (n = 6678)
Configural 835.08 4 0.967 0.249 0.067 – –
Threshold 842.62 8 0.966 0.177 0.067 0.000 −0.073
Metric 707.94 11 0.972 0.138 0.067 0.006 −0.039
Scalar 678.49 14 0.973 0.119 0.067 0.001 −0.019
Age groups (n = 6678)
Configural 871.48 8 0.965 0.254 0.067 – –
Threshold 873.88 20 0.966 0.160 0.067 0.000 −0.094
Metric 735.98 29 0.972 0.121 0.068 0.006 −0.039
Scalar 747.05 38 0.972 0.106 0.068 0.000 −0.015
Educational level (n = 6640)
Configural 822.19 6 0.966 0.248 0.067 – –
Threshold 837.83 14 0.966 0.163 0.067 0.000 −0.085
Metric 709.48 20 0.971 0.125 0.067 0.006 −0.038
Scalar 755.34 26 0.970 0.113 0.067 −0.002 −0.012
Abbreviations: χ² chi-square statistic, df degrees of freedom, CFI Comparative Fit Index, RMSEA Root Mean Square Error of Approximation, SRMR Standardised Root Mean Square Residual, PROMIS GPH Patient-Reported Outcomes Measurement Information System Global Physical Health, PROMIS GMH Patient-Reported Outcomes Measurement Information System Global Mental Health, WHO-5 WHO-5 Well-Being Index, P3CEQ Person-Centred Coordinated Care Experience Questionnaire, P3CEQ-PC P3CEQ Person-Centredness subscale, P3CEQ-CC P3CEQ Care Coordination subscale, PNS Porter–Novelli Scale, MCBS Medicare Current Beneficiary Survey, eHEALS eHealth Literacy Scale, WLSMV weighted least squares with mean and variance adjustment
Models were estimated using WLSMV with indicators treated as ordered categorical
ΔCFI and ΔRMSEA were calculated as the fit index for the more constrained model minus that for the preceding model. Invariance was supported when ΔCFI ≥ −0.010 and ΔRMSEA ≤ 0.015
For the WHO-5 age-group analysis, the two lowest response categories of item 2 were collapsed because the lowest category was unobserved in one age group
Scalar invariance was supported for PROMIS GPH across gender and educational level, but not age group. For PROMIS GMH, scalar invariance was supported across gender and educational level, but not age group. WHO-5 met scalar criteria across all three grouping variables.
Scalar invariance was supported for the P3CEQ total scale across gender but not age group or educational level. For P3CEQ-PC, scalar criteria were met across gender and educational level but not age group, whereas P3CEQ-CC met scalar criteria across gender only.
PNS Factor 1 met scalar criteria across all grouping variables, whereas PNS Factor 3 did not meet scalar criteria across any grouping variable. PNS Factor 2 was not tested because its two-item structure was considered insufficient for conventional multigroup CFA.
Both self-management confidence specifications met scalar criteria across all three grouping variables.
Item response theory analyses
IRT results are shown in Table 7. An overview of the test information curves is provided in Supplementary Figure S1, with model-specific IRT curves presented in Supplementary Figures S2–S11. The test information curves showed differences in the magnitude and distribution of measurement information across the models, with maximum precision occurring at different regions of the respective latent traits. Item-level S-X² tests indicated statistically significant misfit for most items across models, including items within the otherwise stronger PROM models (Table 7).
Table 7Graded response model parameter estimates for scales included in the item response theory analyses
a b1 b2 b3 b4 b5 S-X² p
PROMIS GPH (n = 6588)
PROMIS GPH 1 Physical health 2.508 -1.548 -0.358 1.206 2.541 41.234 0.001
PROMIS GPH 2 Everyday physical functioning 2.544 -2.817 -1.745 -1.025 -0.257 58.465 <0.001
PROMIS GPH 3 Average fatigue 2.069 -3.594 -1.926 -0.288 0.838 70.279 <0.001
PROMIS GPH 4 Average pain 1.584 -4.431 -1.667 -0.141 1.579 69.571 <0.001
PROMIS GMH (n = 6603)
PROMIS GMH 1 Quality of life 2.277 -2.070 -0.886 0.544 2.078 77.591 <0.001
PROMIS GMH 2 Mental health 3.291 -2.108 -1.098 0.058 1.266 185.302 <0.001
PROMIS GMH 3 Satisfaction with social activities and relationships 2.595 -2.013 -1.048 0.278 1.637 59.923 <0.001
PROMIS GMH 4 Emotional problems 1.828 -3.506 -1.626 -0.165 1.298 117.878 <0.001
WHO-5 (n = 6598)
WHO-5 1 Feeling cheerful and in good spirits 3.685 -2.751 -1.592 -0.966 -0.177 1.575 132.242 <0.001
WHO-5 2 Feeling calm and relaxed 3.127 -2.705 -1.679 -0.960 -0.181 1.480 315.969 <0.001
WHO-5 3 Feeling active and vigorous 2.480 -1.987 -1.175 -0.437 0.316 1.854 171.497 <0.001
WHO-5 4 Interest in daily life 2.522 -3.080 -1.587 -0.908 -0.003 1.713 155.752 <0.001
WHO-5 5 Feeling refreshed on waking 2.084 -1.949 -1.157 -0.431 0.268 1.917 106.710 <0.001
P3CEQ total scale (n = 5009)
P3CEQ 1 Discussion of what matters most 1.383 -3.201 -1.251 0.586 145.974 <0.001
P3CEQ 2 Involvement in care decisions 2.053 -2.911 -1.852 -0.190 78.637 0.001
P3CEQ 3 Being treated as a whole person 2.316 -2.609 -1.730 -0.528 59.040 0.042
P3CEQ 4 Having to repeat information 0.986 -3.002 -1.506 0.682 105.068 <0.001
P3CEQ 5 Care organised to meet patient needs 2.524 -2.557 -1.695 -0.239 64.025 0.009
P3CEQ 6 Single professional coordinating care 0.781 0.212 NA NA 38.120 0.009
P3CEQ 7 Care plan addressing patient needs 0.819 1.101 NA NA 29.401 0.044
P3CEQ 8 Support for self-management 1.179 -1.399 -0.482 0.533 346.183 <0.001
P3CEQ 9 Timely and useful self-management information 1.049 -1.095 -0.139 1.434 611.625 <0.001
P3CEQ 10 Confidence in managing own health 0.727 -3.675 0.208 2.965 90.525 0.001
P3CEQ-PC subscale (n = 5053)
P3CEQ 1 Discussion of what matters most 1.296 -3.347 -1.308 0.598 122.316 <0.001
P3CEQ 2 Involvement in care decisions 2.141 -2.855 -1.825 -0.196 56.439 0.044
P3CEQ 3 Being treated as a whole person 2.440 -2.555 -1.702 -0.528 44.762 0.209
P3CEQ 4 Having to repeat information 1.064 -2.834 -1.428 0.645 106.230 <0.001
P3CEQ 5 Care organised to meet patient needs 2.502 -2.567 -1.708 -0.252 66.353 0.003
P3CEQ 8 Support for self-management 1.186 -1.401 -0.495 0.521 356.267 <0.001
P3CEQ 9 Timely and useful self-management information 0.960 -1.169 -0.153 1.520 706.029 <0.001
P3CEQ 10 Confidence in managing own health 0.784 -3.446 0.194 2.785 58.982 0.079
P3CEQ-CC subscale (n = 6088)
P3CEQ 5 Care organised to meet patient needs 1.608 -3.140 -2.049 -0.282 190.030 <0.001
P3CEQ 6 Single professional coordinating care 1.039 0.172 99.587 <0.001
P3CEQ 7 Care plan addressing patient needs 1.199 0.842 32.096 <0.001
P3CEQ 8 Support for self-management 1.253 -1.310 -0.411 0.542 542.344 <0.001
P3CEQ 9 Timely and useful self-management information 1.350 -0.904 -0.081 1.239 394.568 <0.001
PNS Factor 1: Active engagement (n = 6599)
PNS 3 Active disease prevention 1.329 -3.770 -3.061 -1.169 1.364 255.068 <.001
PNS 5 Importance of health information 2.505 -2.666 -2.425 -1.304 0.509 179.916 <.001
PNS 6 Health information for oneself and family 3.024 -2.613 -2.272 -1.142 0.595 72.793 <.001
PNS 10 Understanding personal health risks 2.007 -3.229 -2.883 -1.767 0.898 145.488 <.001
PNS Factor 3: Reliance and cooperation with health professionals (n = 6587)
PNS 1 Reliance on professionals for health information 1.153 -3.036 -2.028 -0.557 1.480 297.021 <0.001
PNS 4 Reliance on professionals for health decisions 0.683 -3.037 -0.807 1.401 4.598 276.627 <0.001
PNS 8 Working with professionals to manage health 2.978 -2.221 -1.558 -0.551 1.021 146.486 <0.001
PNS 9 Discussing relevant information with professionals 1.131 -3.414 -1.847 0.176 2.620 193.629 <0.001
MCBS + eHEALS (n = 4802)
MCBS 1 Confidence in following home-care instructions 3.270 -0.612 0.587 2.151 160.069 <0.001
MCBS 2 Confidence in following lifestyle-change instructions 2.667 -1.117 0.271 1.802 156.682 <0.001
MCBS 3 Confidence in recognising need for medical care 1.787 -1.121 0.390 2.264 81.247 <0.001
MCBS 4 Confidence in recognising medication side effects 1.357 -1.943 -0.252 1.624 79.738 <0.001
eHEALS Confidence in using online health information for decisions 0.779 -4.349 -1.893 0.668 250.835 <0.001
MCBS (n = 5920)
MCBS 1 Confidence in following home-care instructions 3.513 -0.605 0.578 2.122 192.931 <.001
MCBS 2 Confidence in following lifestyle-change instructions 2.749 -1.110 0.267 1.788 190.924 <.001
MCBS 3 Confidence in recognising need for medical care 1.708 -1.146 0.398 2.321 106.556 <.001
MCBS 4 Confidence in recognising medication side effects 1.283 -2.011 -0.263 1.682 193.320 <.001
Abbreviations: PROMIS GPH Patient-Reported Outcomes Measurement Information System Global Physical Health, PROMIS GMH Patient-Reported Outcomes Measurement Information System Global Mental Health, WHO-5 WHO-5 Well-Being Index, P3CEQ Person-Centred Coordinated Care Experience Questionnaire, P3CEQ-PC P3CEQ Person-Centredness subscale, P3CEQ-CC P3CEQ Care Coordination subscale, PNS Porter–Novelli Scale, MCBS Medicare Current Beneficiary Survey, eHEALS eHealth Literacy Scale, NA not applicable
a = discrimination parameter; b1–b5 = category thresholds; S-X² = item-level fit statistic; p < 0.05 indicates lack of fit
PROMIS GPH showed the strongest IRT support, with good global fit and high to very high discrimination. PROMIS GMH and WHO-5 showed mixed but broadly acceptable fit, with elevated RMSEA but good CFI and SRMSR values, ordered thresholds, and very high discrimination. Across the PROMIS and WHO-5 models, information was concentrated at lower to moderate trait levels, with reduced precision at higher levels (Supplementary Figures S1–S4).
The P3CEQ models showed mixed IRT performance. The total scale had suboptimal global fit and variable discrimination, with P3CEQ6, P3CEQ7 and P3CEQ10 contributing least information. P3CEQ-PC provided more information than P3CEQ-CC, while both subscales showed reduced precision at higher trait levels (Supplementary Figures S5–S7).
For the multidimensional PNS, Factors 1 and 3 were modelled separately; PNS Factor 2 was not modelled because it comprised only two items. PNS Factor 1 showed stronger IRT support than PNS Factor 3, with better fit and more consistent discrimination. Within PNS Factor 3, PNS8 contributed substantially more information than the other items (Supplementary Figures S8–S9).
The self-management confidence specifications showed the weakest IRT support. Both the MCBS + eHEALS and MCBS-only models had suboptimal global fit, with high RMSEA and lower CFI values than the PROM models, and variable discrimination. Their information was concentrated primarily around the central to moderately higher regions of the respective latent traits. The eHEALS item showed moderate discrimination and contributed least information in the five-item model, while MCBS4 showed moderate discrimination and contributed least information in the MCBS-only model (Supplementary Figures S10–S11).
Local item dependence was generally limited but was identified for selected item pairs, with the clearest pattern occurring in the self-management confidence models (Supplementary Table S2).
Construct validity
The pattern of inter-scale correlations was broadly consistent with expectations (Table 8). The PROMs were strongly intercorrelated, whereas PROM–PREM associations were weak to moderate. The P3CEQ total scale and P3CEQ-PC and P3CEQ-CC subscales were highly intercorrelated. Associations involving capability-related measures were generally weaker: PROM–PNS correlations were weak, whereas P3CEQ–PNS correlations were weak to moderate. The two self-management confidence specifications were almost perfectly correlated (r = 0.97). Correlations with the reverse-coded self-management confidence scores were negative and weak to moderate for both PROMs and PREM, indicating that higher PROM and PREM scores were associated with greater self-management confidence.
Table 8Pearson correlations between scale scores
PROMIS GPH PROMIS GMH WHO-5 P3CEQ total scale P3CEQ-PC subscale P3CEQ-CC subscale PNS Factor 1 PNS Factor 2 PNS Factor 3 MCBS MCBS + eHEALS
PROMIS GPH 1.000 0.588 0.690 0.196 0.293 0.084 0.133 0.257 -0.035 -0.274 -0.262
PROMIS GMH 1.000 0.777 0.260 0.342 0.140 0.135 0.242 0.027 -0.330 -0.320
WHO-5 1.000 0.292 0.360 0.185 0.111 0.174 0.072 -0.292 -0.286
P3CEQ total scale 1.000 0.920 0.920 0.208 0.216 0.450 -0.402 -0.381
P3CEQ-PC subscale 1.000 0.744 0.235 0.294 0.421 -0.477 -0.453
P3CEQ-CC subscale 1.000 0.138 0.102 0.423 -0.240 -0.222
PNS Factor 1 1.000 0.284 0.252 -0.339 -0.331
PNS Factor 2 1.000 -0.049 -0.436 -0.435
PNS Factor 3 1.000 -0.190 -0.174
MCBS 1.000 0.966
MCBS + eHEALS 1.000
Abbreviations: PROMIS GPH Patient-Reported Outcomes Measurement Information System Global Physical Health, PROMIS GMH Patient-Reported Outcomes Measurement Information System Global Mental Health, WHO-5 WHO-5 Well-Being Index, P3CEQ Person-Centred Coordinated Care Experience Questionnaire, P3CEQ-PC P3CEQ Person-Centredness subscale, P3CEQ-CC P3CEQ Care Coordination subscale, PNS Porter–Novelli Scale; MCBS, Medicare Current Beneficiary Survey, eHEALS eHealth Literacy Scale, PROM patient-reported outcome measure, PREM patient-reported experience measure
For the MCBS and MCBS + eHEALS scores, higher values indicate lower self-management confidence; thus, negative correlations with PROM and PREM scores indicate that higher PROM and PREM scores are associated with greater self-management confidence
Discussion
This study evaluated the psychometric performance of the multi-item scales included in the Norwegian PaRIS-PQ in a large national sample of people with chronic conditions receiving primary care. Overall, psychometric performance varied across scales. The PROMs showed the strongest and most consistent support, whereas the PREM and capability-oriented measures showed more complex measurement properties. The findings support scale-specific interpretation, especially for subgroup comparisons and latent scale use.
The PROMIS physical and mental health scales and the WHO-5 performed best overall. PROMIS GPH showed strong confirmatory support, with CFI, TLI and SRMR meeting the reference criteria and RMSEA slightly exceeding the criterion, while PROMIS GMH and WHO-5 showed strong factor loadings, acceptable internal consistency and broadly supportive, although mixed, CFA results. The elevated RMSEA values, particularly for PROMIS GMH and WHO-5, should therefore be interpreted alongside the stronger CFI, SRMR, factor loading and internal consistency results. Overall, this pattern is consistent with prior evidence supporting the two-summary-score structure of PROMIS Global Health and the unidimensional structure of WHO-5, and supports their use for assessing physical health, mental health and well-being within the PaRIS conceptual framework and PaRIS-10 indicator set [2, 6, 10, 19, 20].
The P3CEQ showed weaker and less stable measurement properties. The total scale showed suboptimal CFA fit, and scalar invariance was only partly supported, with support across gender but not age group or educational level. Subscale results were also uneven, with weaker internal consistency for P3CEQ-CC. This is consistent with the original UK validation, which noted overlap between person-centredness and coordination items and challenges related to response formats [21], and with the PaRIS Field Trial, which pointed more strongly to a dominant general factor than to a clearly separable two-factor structure (9, 10). The Norwegian findings suggest that neither the total scale nor the current subscale specification is fully robust, supporting further refinement of item overlap, response formats and scoring.
The PNS showed a more differentiated and uneven structure. The international Field Trial reported two factors reflecting active engagement and reliance on health professionals [9, 10], whereas the Norwegian EFA supported three factors. PNS Factor 1, reflecting active engagement, showed the strongest psychometric support and stable cross-group measurement properties. PNS Factor 3 performed less well, with lower reliability, no scalar invariance and weaker IRT fit, and its stronger correlation with P3CEQ suggests that it may partly capture patients’ evaluations of their relationship with, or support from, health professionals. PNS Factor 2, reflecting health information comprehension, comprised only two items, which limited further psychometric testing, but the same two PNS-derived items have been operationalised as a health literacy measure in recent Norwegian PaRIS analyses and were associated with PREMs and PROMs [14]. The PNS should therefore be interpreted as multidimensional and partly exploratory rather than as a single robust reflective scale, consistent with the original Porter–Novelli Scale as a brief screening tool and with the Field Trial recommendation for further work on cross-country variation in health information and decision-making preferences [10, 22].
The self-management confidence item sets also showed mixed support. The PaRIS Field Trial evaluated a broader self-management confidence construct comprising P3CEQ10, four MCBS items and one eHEALS-derived item [9, 10], whereas the present study retained P3CEQ10 within the P3CEQ analyses and evaluated the MCBS + eHEALS and MCBS-only specifications separately. Both specifications showed exploratory support, but CFA and IRT provided limited support for robust unidimensional latent measurement. Removing eHEALS did not improve model fit, while both specifications met scalar invariance criteria across gender, age group and educational level. Local item dependence was also most evident for the self-management confidence specifications, particularly between MCBS3 and MCBS4, which may inflate reliability and information estimates, further supporting cautious interpretation of these item sets.
The weaker performance of some experience- and capability-related measures should be interpreted in relation to the reflective specifications evaluated. This is particularly relevant for the PNS and self-management confidence item sets, which were not originally developed as conventional unidimensional scales in their present PaRIS configurations [9, 10, 22]. Weaker fit therefore indicates limited support for the reflective model tested rather than poor relevance of the individual items. Conversely, poor reflective-model fit does not by itself establish a formative measurement model, which would require separate conceptual justification and empirical evaluation [45].
Item-level fit provided an additional qualification. S-X² indicated statistically significant misfit for most items, including within the otherwise stronger PROM models. Given the large analytical sample, the chi-square-based S-X² tests may also have detected relatively small departures from the GRM as statistically significant. These findings were therefore interpreted in conjunction with global model fit, discrimination, information and local dependence rather than in isolation. The widespread item-level misfit indicates that further item-level evaluation is warranted across the PaRIS-PQ scales, including measures with otherwise supportive scale-level performance.
Measurement invariance and construct validity have practical implications for PaRIS-PQ scale use. Metric invariance was supported for all tested models, supporting analyses of associations across demographic groups. Scalar invariance was less consistent across measures. PROMIS GPH and PROMIS GMH met scalar criteria across gender and educational level but not age group. WHO-5, PNS Factor 1 and both self-management confidence specifications met scalar criteria across all grouping variables. Scalar support was more variable for the P3CEQ models, whereas PNS Factor 3 did not meet scalar criteria across any grouping variable. Correlations supported construct validity and related but distinct outcome, experience and capability domains, with strong PROM intercorrelations and weaker PROM–PREM, PROM–capability and PREM–capability associations, consistent with the Field Trial [9, 10].
Cross-country comparability requires separate consideration. Differences in healthcare-system organisation are explicitly part of the contextual variation that PaRIS is designed to examine; nested survey design and the associated conceptual framework link patient-reported outcomes and experiences to practice- and system-level characteristics [6, 7]. The use of common measures and harmonised translation and survey procedures is intended to support cross-country comparison [2, 7, 9] but does not in itself establish measurement invariance [13]. In the Field Trial, multigroup CFA of the PROMs supported configural invariance, whereas metric models showed greater misfit and scalar invariance was not supported [9, 10]. The present study, based on data from Norway only, cannot assess cross-country measurement invariance; this requires analyses using data from multiple countries, which is an important avenue for further research for the OECD PaRIS initiative. Cross-country comparisons of scale scores should be interpreted based on the results from the analysis of the PaRIS database at the international level.
Taken together, the Norwegian findings support scale-specific rather than uniform use of the PaRIS-PQ. They support the PaRIS strategy of combining established scales with additional items while showing that measurement properties vary across domains and require context-specific evaluation [9, 10, 13]. For current use, the PROMIS GPH, PROMIS GMH and WHO-5 have the strongest overall support for monitoring and research, whereas the P3CEQ, PNS factors beyond active engagement, and the exploratory self-management confidence item sets warrant more cautious, purpose-specific interpretation. In particular, subgroup mean comparisons should be interpreted cautiously where scalar invariance was not supported, and latent-scale analyses are less well supported where dimensionality or model fit was mixed. The psychometric evaluation also identifies opportunities to refine item selection, response formats and scale construction in the experience and capability domains. Further analyses using cross-country Main Survey data should examine item-level functioning and measurement invariance to determine whether the observed limitations are country-specific or more general and to guide future refinement and use of PaRIS measures. Future PaRIS cycles should assess scale performance alongside representativeness and provider participation, given Norwegian evidence of GP participation-related selection [12].
Strengths and limitations
This study provides the first comprehensive psychometric evaluation of the multi-item scales in the Norwegian PaRIS-PQ using Main Survey data. It covers scales underpinning key PaRIS indicators and applies complementary methods, including EFA, CFA, internal consistency assessment, measurement invariance testing and IRT analyses. The large national sample strengthened precision and enabled subgroup analyses, while the provider-linked design was closely aligned with the intended PaRIS implementation model. The study also benefited from the Norwegian translation and cognitive testing process, which supported linguistic and conceptual adaptation before field implementation.
Some limitations should also be noted. First, the evaluation was restricted to multi-item scales; important single-item PaRIS indicators, including trust, overall perceived quality and general health, were outside the scope of the analyses. Second, several scales, subscales and factors contained few items, resulting in low degrees of freedom in some CFA models and constraining both CFA and IRT modelling. CFA fit indices, particularly RMSEA in models with few degrees of freedom, should therefore be interpreted alongside factor loadings, reliability, other psychometric evidence, and the complementary IRT findings. Third, cognitive testing in Norway used three separate protocols rather than the full patient questionnaire and therefore could not fully assess overall cognitive burden. Finally, as the study was conducted in one national context, cross-country measurement invariance could not be assessed.
Conclusions
The Norwegian PaRIS-PQ showed varied psychometric performance across its multi-item scales. PROMIS GPH, PROMIS GMH and WHO-5 demonstrated the strongest and most consistent support and appear suitable for monitoring and research. The P3CEQ, the PNS beyond the active engagement factor, and the self-management confidence item sets showed more mixed evidence and require cautious, scale-specific interpretation, especially in subgroup comparisons where scalar invariance was not supported and when treated as reflective latent measures. The findings support further refinement of experience and capability-related measures with weaker psychometric support, particularly those related to person-centredness, coordination and self-management support. Evidence from other countries will be important for assessing generalisability within the PaRIS initiative.
Supplementary Information
Acknowledgements
We thank patients and providers participating in the Norwegian PaRIS surveys, and the national reference group for contributing to all phases of PaRIS in Norway. We also thank the rest of the Norwegian PaRIS team, OECD’s PaRIS team and the PaRIS consortium for their effort and collaboration on the PaRIS surveys.
Abbreviations
PROMPatient-reported outcome measure
PREMPatient-reported experience measure
OECDOrganisation for Economic Co-operation and Development
PaRISPatient-Reported Indicator Surveys
PaRIS-PQPaRIS Patient Questionnaire
PaRIS-PCPQPaRIS Primary Care Practice Questionnaire
PaRIS-10PaRIS ten key indicators
GPGeneral practitioner
KUHRNorwegian Control and Payment of Health Reimbursements Database
EMPROEvaluating the Measurement of Patient-Reported Outcomes
TRAPDTranslation, Review, Adjudication, Pretesting and Documentation
PROMISPatient-Reported Outcomes Measurement Information System
PROMIS GPHPatient-Reported Outcomes Measurement Information System Global Physical Health
PROMIS GMHPatient-Reported Outcomes Measurement Information System Global Mental Health
WHO-5WHO-5 Well-Being Index
P3CEQPerson-Centred Coordinated Care Experience Questionnaire
PNSPorter–Novelli Scale
MCBSMedicare Current Beneficiary Survey
eHEALSeHealth Literacy Scale
CFAConfirmatory factor analysis
EFAExploratory factor analysis
P3CEQ-PCP3CEQ Person-Centredness subscale
P3CEQ-CCP3CEQ Care Coordination subscale
KMOKaiser–Meyer–Olkin
WLSMVWeighted least squares with mean and variance adjustment
CFIComparative Fit Index
TLITucker–Lewis Index
RMSEARoot Mean Square Error of Approximation
SRMRStandardised Root Mean Square Residual
IRTItem response theory
GRMGraded response model
SRMSRStandardised Root Mean Square Residual
DPIAData Protection Impact Assessment
Authors’ contributions
J.M.V. initiated the study. H.H.I. led the psychometric evaluation, performed all statistical analyses, interpreted the results, and drafted the manuscript. O.B. contributed throughout the Norwegian PaRIS study. L.H.E.-D. contributed to the Norwegian questionnaire adaptation and cognitive testing. J.M.V., I.P. and C.K. contributed expertise on the PaRIS conceptual framework, questionnaire development and interpretation of the psychometric findings. All authors critically reviewed and approved the final manuscript.
Funding
Open access funding provided by Norwegian Institute of Public Health (FHI). The authors received no specific funding for this work. The Ministry of Health and Care Services financed the Norwegian part of the international costs of participating in the PaRIS surveys.
Data availability
Anonymous data are available from the corresponding author on reasonable request. Public Use Files from the PaRIS patient survey, including anonymised data from Norway, are available through the OECD PaRIS Public Use Files website.
Declarations
Ethics approval and consent to participate
The study was part of a national quality assurance project (“OECD’s Patient-reported Indicator survey in Norway 2021–25”, project number: 3103). Research Director Atle Fretheim of the Norwegian Institute of Public Health approved the Data Protection Impact Assessment (DPIA) for the quality assurance project. According to the joint body of the Norwegian Regional Committees for Medical and Health Research Ethics, research approval is not required for quality assurance projects.
Informed consent was obtained for all participants. Returning the completed questionnaire constituted patient consent, which is the standard procedure in all patient-experience surveys conducted by the Norwegian Institute of Public Health. Patients were informed that participation was voluntary. The study was performed in accordance with relevant guidelines and regulations including the Declaration of Helsinki.
Consent to publish
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
1.OECD. (2019). Realising the full potential of primary health care. OECD Publishing.
2.OECD. (2025). Does Healthcare Deliver? Results from the Patient-Reported Indicator Surveys (PaRIS). OECD Publishing.
3.OECD. (2020). Realising the Potential of Primary Health Care. OECD Publishing.
4.Kendir, C., Gonzalez de la Fuente, A., Kringos, D., van den Berg, M., Valderas, J. M., & Klazinga, N. (2026). Patient-reported outcome and experience measures in primary care: a scoping review of systematic collection and use. Bmc Health Services Research, 26(1), 278.
5.Valderas, J. M., & Alonso, J. (2008). Patient reported outcome measures: a model-based classification system for research and clinical practice. Quality Of Life Research, 17(9), 1125–1135.
6.Valderas, J. M., Porter, I., Martin-Delgado, J., Rijken, M., de Jong, J., Groene, O., et al. (2025). Development of the Patient-Reported Indicator Surveys (PaRIS) conceptual framework to monitor and improve the performance of primary care for people living with chronic conditions. Bmj Quality & Safety, 34(8), 529–536.
7.de Boer, D., van den Berg, M., Ballester, M., Bloemeke, J., Boerma, W., de Bienassis, K., et al. (2022). Assessing the outcomes and experiences of care from the perspective of people living with chronic conditions, to support countries in developing people-centred policies and practices: study protocol of the International Survey of People Living with Chronic Conditions (PaRIS survey). British Medical Journal Open, 12(9), e061424.
8.Bloemeke-Cammin, J., Groene, O., Ballester, M., Guanais, F., Groenewegen, P., Kendir, C., et al. (2024). International cross-cultural development and field testing of the primary care practice questionnaire for the PaRIS survey (PaRIS-PCPQ). BMC Prim Care, 25(1), 168.
9.Valderas, J. M., Porter, I., Evans, J. P., Heijmans, M., Rijken, M., Groene, O. (2026). International survey of people living with chronic conditions: development and evaluation of the PaRIS Patient Questionnaire (PaRIS-PQ) in 18 countries. Bmj Quality & Safety, 35(10), 687–697. 10.1136/bmjqs-2025-018548
10.van den Berg, M., Kendir, C., Castelblanco, D., Larrain, N., Guanais, F., Groene, O., et al. (2024). PaRIS Field Trial Report: Technical report on the international PaRIS survey of people living with chronic conditions (Vol. ,166OECD Health Working Papers, No. 166). OECD Publishing.
11.Sjetne, I. S., Norman, R. M., & Bjertnæs, Ø. A. (2025). Patient-Reported Indicator Surveys (PaRIS): Norsk resultatrapport. Oslo: Folkehelseinstituttet; Report No.: ISBN 978-82-8406-515-1.
12.Bjertnaes, O., Skudal, K. E., van den Berg, M. J., Porter, I., Holmboe, O., Norman, R. M., et al. (2024). International survey of people living with chronic conditions (PaRIS survey): effects of general practitioner non-participation on the representativeness of the Norwegian patient data. Bmc Health Services Research, 24(1), 1257.
13.Zhao, Y., Summers, R., Gathara, D., & English, M. (2024). Conducting cross-cultural, multi-lingual or multi-country scale development and validation in health care research: A 10-step framework based on a scoping review. J Glob Health, 14, 04151.
14.Bjertnaes, O., Rijken, M., Porter, I., Norman, R. M., Sjetne, I. S., Iversen, H. H. (2026). Identifying patient and provider determinants of primary care experiences and outcomes for persons with chronic conditions: a multilevel analysis of a nation-wide survey in Norway. Fam Pract,43(3).
15.Norwegian Directorate of Health Norwegian Control and Payment of Health Reimbursements Database (KUHR) [Internet]. Oslo: Norwegian Directorate of Health; [Available from: https://helsedata.no/en/forvaltere/norwegian-directorate-of-health/norwegian-control-and-payment-of-health-reimbursements-database-kuhr/
16.Norsk helsenett. About Helsenorge [Internet]. Oslo: The Norwegian Directorate of Health (2026). [Available from: https://www.helsenorge.no/en/about-helsenorge/
17.Valderas, J. M., Ferrer, M., Mendívil, J., Garin, O., Rajmil, L., Herdman, M., et al. (2008). Development of EMPRO: A Tool for the Standardized Assessment of Patient-Reported Outcome Measures. Value in Health, 11(4), 700–708.
18.University of Michigan SRC. (2011). Guidelines for Best Practice in Cross-cultural Surveys: Full Guidelines: Survey Research Center. Institute for Social Research, University of Michigan.
19.Hays, R. D., Bjorner, J. B., Revicki, D. A., Spritzer, K. L., & Cella, D. (2009). Development of physical and mental health summary scores from the patient-reported outcomes measurement information system (PROMIS) global items. Quality Of Life Research, 18(7), 873–880.
20.Topp, C. W., Østergaard, S. D., Søndergaard, S., & Bech, P. (2015). The WHO-5 Well-Being Index: a systematic review of the literature. Psychotherapy And Psychosomatics, 84(3), 167–176.
21.Lloyd, H., Fosh, B., Whalley, B., Byng, R., & Close, J. (2019). Validation of the person-centred coordinated care experience questionnaire (P3CEQ). International Journal For Quality In Health Care, 31(7), 506–512.
22.Maibach, E. W., Weber, D., Massett, H., Hancock, G. R., & Price, S. (2006). Understanding consumers’ health information preferences: development and validation of a brief screening instrument. J Health Commun, 11(8), 717–736.
23.Parker, J. L., Regan, J. F., & Petroski, J. (2014). Beneficiary activation in the Medicare population. Medicare Medicaid Res Rev, 4(4)–E14.
24.Norman, C. D., & Skinner, H. A. (2006). eHEALS: The eHealth Literacy Scale. Journal Of Medical Internet Research, 8(4), e27.
25.de Arruda, G. T., Terwee, C. B., Elsman, E. B. M., Avila, M. A., Gagnier, J. J., & Mokkink, L. B. (2025). Explanation & Elaboration document of the COSMIN Reporting Guideline 2.0 for studies on measurement properties of patient-reported outcome measures. Quality Of Life Research, 34(7), 1891–1899.
26.Kaiser, H. F. (1974). An Index of Factorial Simplicity. Psychometrika, 39(1), 31–36.
27.Bartlett, M. S. (1950). Tests of Significance in Factor Analysis. British Journal of Statistical Psychology, 3(2), 77–85.
28.Watkins, M. W. (2018). Exploratory Factor Analysis: A Guide to Best Practice. Journal of Black Psychology, 44(3), 219–246.
29.Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric Theory. McGraw-Hill Companies, Incorporated.
30.Streiner, D. L., Norman, G. R., & Cairney, J. (2014). Health Measurement Scales: Oxford University Press; 01 Jan 2015.
31.Hu, L. T., & Bentler, P. M. (1999). Cutoff Criteria for Fit Indexes in Covariance Structure Analysis: Conventional Criteria Versus New Alternatives. Structural Equation Modeling-a. Multidisciplinary Journal, 6(1), 1–55.
32.Hooper, D., Coughlan, J., & Mullen, M. (2008). Structural Equation Modeling: Guidelines for Determining Model Fit. Electron J Bus Res Methods,6(1), 53–60.
33.Kenny, D. A., Kaniskan, B., & McCoach, D. B. (2015). The Performance of RMSEA in Models With Small Degrees of Freedom. Sociological Methods & Research. ;44(2015)(3):486–507.
34.Chen, F., Curran, P. J., Bollen, K. A., Kirby, J., & Paxton, P. (2008). An Empirical Evaluation of the Use of Fixed Cutoff Points in RMSEA Test Statistic in Structural Equation Models. Sociol Methods Res, 36(4), 462–494.
35.Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling-a Multidisciplinary Journal, 9(2), 233–255.
36.Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling-a Multidisciplinary Journal, 14(3), 464–504.
37.Putnick, D. L., & Bornstein, M. H. (2016). Measurement Invariance Conventions and Reporting: The State of the Art and Future Directions for Psychological Research. Developmental Review, 41, 71–90.
38.Samejima, F. (1969). Estimation of Latent Ability Using a Response Pattern of Graded Scores. Psychometrika, 34(S1), 1–97.
39.Nguyen, T. H., Han, H. R., Kim, M. T., & Chan, K. S. (2014). An introduction to item response theory for patient-reported outcome measurement. Patient, 7(1), 23–35.
40.Kang, T., & Chen, T. T. (2011). Performance of the generalized S-X item fit index for the graded response model. Asia Pacific Education Review, 12(1), 89–96.
41.Maydeu-Olivares, A. (2013). Goodness-of-Fit Assessment of Item Response Theory Models. Measurement: Interdisciplinary Research & Perspective, 11(3), 71–101.
42.Baker, F., & Kim, S-H. The Basics of Item Response Theory Using R2017.
43.Yen, W. M. (1984). Effects of Local Item Dependence on the Fit and Equating Performance of the Three-Parameter Logistic Model. Applied Psychological Measurement, 8(2), 125–145.
44.Christensen, K. B., Makransky, G., & Horton, M. (2017). Critical Values for Yen’s Q(3): Identification of Local Dependence in the Rasch Model Using Residual Correlations. Applied Psychological Measurement, 41(3), 178–194.
45.Bollen, K., & Lennox, R. (1991). Conventional Wisdom on Measurement: A Structural Equation Perspective. Psychological Bulletin, 110, 305–314.
Associated Data
Supplementary Materials
Data Availability Statement
Anonymous data are available from the corresponding author on reasonable request. Public Use Files from the PaRIS patient survey, including anonymised data from Norway, are available through the OECD PaRIS Public Use Files website.
Come citare questo studio
Iversen HH, Bjertnaes O, Ellingsen-Dalskau LH, Porter I, Kendir C, Valderas JM. (2026). Psychometric performance of the OECD Patient-Reported Indicator Surveys Patient Questionnaire (PaRIS-PQ) among adults with chronic conditions in Norway.. Quality of life research : an international journal of quality of life aspects of treatment, care and rehabilitation. DOI: 10.1007/s11136-026-04429-z
Consulta la pubblicazione originale ↗