Construct Validity and Item Response Theory Analysis of the PROMIS-29 v2.0 in Recipients of Lumbar Spine Surgery

The PROPr can be measured using different PROMIS domain item sets

BACKGROUND

The Patient-Reported Outcomes Measurement Information System (PROMIS) Preference Score (PROPr) is estimated from descriptive health assessments within the PROMIS framework. The underlying item response theory (IRT) allows researchers to measure PROMIS health domains with any subset of items that are calibrated to this domain. Consequently, this should also be true for the PROPr. We aimed to test this assumption using both an empirical and a simulation approach.

METHODS

Empirically, we estimated 3 PROMIS Pain inference (PI) scores from 3 different item subsets in a sample of n=199 cancer patients: 4 PROMIS-29 items (estimate: θ4), the 2 original PROPr items (θ2), and 10 different items (θ10). We calculated mean differences and agreement between θ4, and θ2 and θ10, respectively, and between their resulting PROPr4, PROPr2, PROPr10, using intraclass correlation coefficients (ICC) and Bland-Altman (B-A) plots with 95 %-Limits of Agreement (LoA). For the simulation, we used the IRT-model to calculate all item responses of the entire 7 PROPr domain item banks from the empirically observed PROMIS-29+cognition θ. From these simulated item banks, we chose the 2 original PROPr items per domain to calculate PROPrsim and compared it to PROPr4 again using ICC and B-A plots.

RESULTS

θ4 vs θ10 showed smaller bias (-0.012, 95 %-LoA -0.88;0.85) than θ4 vs θ2 (0.025, 95 %-LoA -0.95;1.00. ICC>0.85 (p<0.001) in both θ-comparisons. PROPr4 vs PROPr10 showed lower bias (0.0012, 95 %-LoA -0.039;0.042) than PROPr4 vs PROPr2 (-0.0029, 95 %-LoA -0.049;0.044). ICC>0.98 (p<0.0001) on both PROPr-comparisons. Mean PROPrsim was larger than mean PROPr4 (0.0228, 95 %-LoA -0.1103; 0.1558) and ICC was 0.95 (95 %CI 0.93; 0.97).

CONCLUSION

Different item subsets can be used to estimate the PROMIS PI for calculation of the PROPr. Reduction to 2 items per domain rather than 4 does not significantly change the PROPr estimate on average. Agreements differ across the spectrum and in individual comparisons.

Preoperative patient-reported physical health-related quality of life predicts short-term postoperative outcomes in brain tumor patients

BACKGROUND

Patient-reported outcome measures (PROMs) are increasingly used to assess patients' perioperative health. The PROM Information System 29 (PROMIS-29) is a well-validated global health assessment instrument for patient physical health, though its utility in cranial neurosurgery is unclear.

OBJECTIVE

To investigate the utility of preoperative PROMIS-29 physical health (PH) summary scores in predicting postoperative outcomes in brain tumor patients.

METHODS

Adult brain tumor patients undergoing resection at a single institution (January 2018-December 2021) were identified and prospectively received PROMIS-29 surveys during pre-operative visits. PH summary scores were constructed and optimum prediction thresholds for length of stay (LOS), discharge disposition (DD), and 30-day readmission were approximated by finding the Youden index of the associated receiver operating characteristic curves. Bivariate analyses were used to study the distribution of low (z-score≤-1) versus high (z-score>-1) PH scores according to baseline characteristics. Logistic regression models quantified the association between preoperative PH summary scores and post-operative outcomes.

RESULTS

A total of 157 brain tumor patients were identified (mean age 55.4±15.4 years; 58.0% female; mean PH score 45.5+10.5). Outcomes included prolonged LOS (24.8%), non-routine discharge disposition (37.6%), and 30-day readmission (19.1%). On bivariate analysis, patients with low PH scores were significantly more likely to be diagnosed with a high-grade tumor (69.6% vs 38.85%, p=0.010) and less likely to have elective surgery (34.8% vs 70.9%, p=0.002). Low PH score was associated with prolonged LOS (26.1% vs 22%, p<0.001), nonroutine discharge (73.9% vs 31.3%, p<0.001) and 30-day readmission (43.5% vs 14.9%, p=0.003). In multivariate analysis, low PH scores predicted greater LOS (odds ratio [OR]=6.09, p=0.003), nonroutine discharge (OR=4.25, p=0.020), and 30-day readmission (OR=3.93, p=0.020).

CONCLUSION

The PROMIS-29 PH summary score predicts short-term postoperative outcomes in brain tumor patients and may be incorporated into prospective clinical workflows.

The Relationship Between Preoperative Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference and Pain Intensity Scores and Early Postoperative Pain and Opioid Consumption After Lumbar Fusion

Background The Patient-Reported Outcomes Measurement Information System (PROMIS) pain interference and pain intensity measures quantify separate dimensions of pain from the patient's perspective. This study aimed to assess differences in these outcomes and to evaluate whether baseline PROMIS pain scores could be used as a leading indicator of increased pain and opioid consumption during early recovery after lumbar fusion. Methods A retrospective review of 199 consecutive patients undergoing posterolateral fusion (PLF) at a single institution was performed. All patients underwent one to three level lumbar PLF and preoperatively completed the PROMIS pain intensity and PROMIS pain interference measures. Multivariate linear regression was used to assess the relationship between preoperative PROMIS scores and postoperative pain numeric rating scale (NRS) and oral morphine milligram equivalents (OMME) by day after controlling for age, sex, and body mass index (BMI). Results In comparison to patients with the lowest preoperative pain intensity scores, those with the highest scores required significantly more OMME on postoperative day (POD) zero and one (both p<0.05) and had higher pain NRS on POD one (p=0.02). Patients with the highest pain interference scores reported higher pain NRS on POD zero (p=0.02) but required similar OMME at all time points. After controlling for age, sex, and BMI, each one-point increase in preoperative PROMIS pain interference scores was associated with increased OMME on POD zero (β=0.29, p=0.04) and POD one (β=0.64, p=0.03). Conclusions Patients with high pain intensity reported higher levels of pain and required more opioids during the first 24 hours postoperatively, while those with high pain interference reported higher levels of pain on the day of surgery but utilized similar amounts of opioids. After risk adjustment, increased baseline PROMIS pain interference scores - but not pain intensity - were associated with increased opioid use. These results suggest that both measures should be considered when identifying patients at risk for increased pain and opioid consumption after PLF.

How Well Do Seven Self-Report Measures Represent Underlying Back Pain Impact?

BACKGROUND

The extent to which different measures of back pain impact represent an underlying common factor has implications for decisions about which one to use in studies of pain management and estimating one score from others.

AIMS

To determine if different self-report back pain impact measures represent an underlying pain latent variable and estimate associations with it.

METHOD

Seven pain impact measures completed by Amazon Mechanical Turk adults are used to estimate internal consistency reliability and associations: Oswestry Disability Index (ODI), Roland-Morris Disability Questionnaire (RMDQ), short form of the Örebro Musculoskeletal Pain Questionnaire (OMPQ), Subgroups for Targeted Treatment (STarT) Back Tool, the Graded Chronic Pain Scale (GCPS) disability score, PEG (Pain intensity, interference with Enjoyment of life, interference with General activity), and Impact Stratification Score (ISS).

RESULTS

The sample of 1,874 adults with back pain had an average age of 41 and 52% were female. Sixteen percent were Hispanic, 7% non-Hispanic Black, 5% non-Hispanic Asian, and 71% non-Hispanic White. Internal consistency reliability estimates ranged from 0.710 (OMPQ) to 0.923 (GCPS). Correlations among the measures ranged from 0.609 (RMDQ with OMPQ) to 0.812 (PEG with GCPS). Standardized factor loadings on the pain latent variable ranged from 0.782 (RMDQ) to 0.870 (ISS).

CONCLUSIONS

Scores of each measure can be estimated from the others for use in research.

Specific and shared mechanisms associated with treatment for chronic neck pain: study protocol for the SS-MECH trial

BACKGROUND

Treatment mechanisms involve the steps or processes through which an intervention unfolds and produces change in an outcome variable. Treatment mechanisms can be specific to the intervention provided (i.e. pain modulation) or shared with other treatments (i.e. reduced fear of movement). Whether specific and shared treatment mechanisms are different across interventions and whether they lead to the outcomes seen in trials is largely unknown. The management of individuals with chronic neck pain routinely include manual therapy (MT) and resistance exercise (RE), as both approaches are included in clinical practice guidelines and both yield similar outcomes.

OBJECTIVES

Our study plans to answer two research questions: 1) what are the specific mechanisms associated with MT versus interventions (and are these different), and 2) what are the shared mechanisms associated with these interventions, and do specific or shared mechanisms mediate clinical outcomes?

METHODS

This study will involve a 2-group parallel (1:1) single-blinded randomized trial to compare the specific and potential shared treatment mechanisms between these two approaches. We will enroll individuals with a history of chronic neck pain and evaluate whether specific or shared mechanisms mediate clinical outcomes.

RESULTS

We hypothesize that MT and RE approaches will both exhibit different specific treatment mechanisms, and that both approaches will exhibit shared treatment mechanisms, which will notably influence outcomes at both discharge and 6-months.

CONCLUSIONS

This study is important because it will help identify what specific or shared treatment mechanisms are associated with different interventions and, how different treatment mechanisms influence clinical outcomes.

Psychometric properties and reference values of the Patient-Reported Outcomes Measurement Information System (PROMIS) sleep item banks in the Dutch general population

Psychometric properties of the v1.0 Patient-Reported Outcomes Measurement Information System (PROMIS®) sleep disturbance (27 items) and sleep-related impairment (SRI; 16 items) item banks, short forms derived from the item bank, and simulated computerised adaptive test (CAT), were assessed in a representative sample of 1,006 adults from the Dutch general population. For sleep disturbance all items fitted the item response theory model. Four items showed differential item functioning (i.e., lack of measurement invariance) for age and two for language but the impact on scores (expressed as T-scores) was small. Reliable scores (r > 0.90) were found for 92.2%-96.3% of respondents with the full bank, short forms with six and eight items, and CAT, but for only 25.6% with the four-item short form. For SRI two items did not fit the item response theory model. Four items showed differential item functioning for language but the impact on T-scores was small. Reliable scores were found for 82.1% with the full bank, for 47.8%-69.5% with short forms and CAT. T-scores of 49.7 and 49.3 represent the average score of the Dutch general population for sleep disturbance and SRI, respectively. In conclusion, sufficient structural validity, reliability, and cross-cultural validity was found for the full banks but short forms of four items are not reliable enough for clinical practice. For SRI we recommend the full item bank if this is the primary outcome.

Self-reported health measures in burn survivors undergoing burn surgery following acute hospitalization: A burn model system national database investigation

INTRODUCTION

Health Related Quality of Life (HRQoL) surveys such as PROMIS-29 may facilitate shared decision-making regarding surgery after burn injury. We aimed to examine whether scar revision and contracture release surgery after index hospitalization was associated with differences in HRQoL.

METHODS

Patient and PROMIS-29 Profile v2.0 data were extracted from the Burn Model System (BMS) at 6-, 12-, and 24-months after burn. PROMIS-29 measures 7 health-related domains. Linear regression was performed to identify associations between independent burn patient variables (e.g. scar-related surgery) and PROMIS-29 scores. Socio-demographic and injury variables were analyzed using logistic regression to determine the likelihood of undergoing burn-related surgery.

RESULTS

Of 727 participants, 201 (27.6%) underwent ≥ 1 scar/contracture operation within 24 months of injury. Number of operations at index hospital admission and range of motion (ROM) deficit at discharge were correlated with an increased likelihood of undergoing subsequent scar/contracture surgery (p < 0.05). Participants undergoing scar/contracture surgery and those that were Medicaid insured reported significantly worse HRQoL for PROMIS domains: anxiety, depression, and fatigue (p < 0.05).

CONCLUSIONS

After adjusting for burn severity and available confounders, participants who underwent scar-related burn reconstructive surgery after index hospitalization reported overall worse Health-Related Quality of Life (HRQoL) in multiple domains.