Secular trends in the incidence and prevalence of rheumatoid arthritis within members of an integrated health care delivery system

Assessing the potential of GPT-4 to perpetuate racial and gender biases in health care: a model evaluation study

BACKGROUND

Large language models (LLMs) such as GPT-4 hold great promise as transformative tools in health care, ranging from automating administrative tasks to augmenting clinical decision making. However, these models also pose a danger of perpetuating biases and delivering incorrect medical diagnoses, which can have a direct, harmful impact on medical care. We aimed to assess whether GPT-4 encodes racial and gender biases that impact its use in health care.

METHODS

Using the Azure OpenAI application interface, this model evaluation study tested whether GPT-4 encodes racial and gender biases and examined the impact of such biases on four potential applications of LLMs in the clinical domain-namely, medical education, diagnostic reasoning, clinical plan generation, and subjective patient assessment. We conducted experiments with prompts designed to resemble typical use of GPT-4 within clinical and medical education applications. We used clinical vignettes from NEJM Healer and from published research on implicit bias in health care. GPT-4 estimates of the demographic distribution of medical conditions were compared with true US prevalence estimates. Differential diagnosis and treatment planning were evaluated across demographic groups using standard statistical tests for significance between groups.

FINDINGS

We found that GPT-4 did not appropriately model the demographic diversity of medical conditions, consistently producing clinical vignettes that stereotype demographic presentations. The differential diagnoses created by GPT-4 for standardised clinical vignettes were more likely to include diagnoses that stereotype certain races, ethnicities, and genders. Assessment and plans created by the model showed significant association between demographic attributes and recommendations for more expensive procedures as well as differences in patient perception.

INTERPRETATION

Our findings highlight the urgent need for comprehensive and transparent bias assessments of LLM tools such as GPT-4 for intended use cases before they are integrated into clinical care. We discuss the potential sources of these biases and potential mitigation strategies before clinical implementation.

FUNDING

Priscilla Chan and Mark Zuckerberg.

Association between patient-reported outcomes and impairments in work and activity in patients with rheumatoid arthritis in clinical remission: A retrospective analysis using the IORRA database

OBJECTIVES

To explore the patient-reported outcomes (PROs) associated with work productivity and activity impairment (WPAI) in patients with rheumatoid arthritis (RA) in clinical remission.

METHODS

We enrolled patients with RA ≥18 years and with a simplified disease activity index ≤3.3 from the Institute of Rheumatology, Rheumatoid Arthritis data set collected in October 2017. The pain-visual analogue scale, patients' global assessment visual analogue scale (VAS), Japanese version of the Healthcare Assessment Questionnaire (J-HAQ) Disability Index, and duration of morning joint stiffness were selected as the PROs. To evaluate work productivity and activity, the WPAI for RA instrument (WPAI-RA) was used. To assess the contribution of each PRO to the WPAI-RA score, an analysis of variance model was constructed.

RESULTS

The mean age of the 2614 patients was 62.4 years; 85.1% were female. Median values of the WPAI-RA score were 1.1% for absenteeism, 6.5% for presenteeism, 7.4% for work impairment, and 10.2% for activity impairment. Morning joint stiffness contributed the most to absenteeism (18.0%), while pain-VAS contributed the most to presenteeism (57.4%), work productivity loss (51.1%), and daily activity impairment (53.7%). J-HAQ was the second most contributing factor to presenteeism (17.4%), work productivity loss (16.3%), and daily activity impairment (26.0%).

CONCLUSIONS

The pain-VAS and J-HAQ highly contributed to WPAI in patients with RA in clinical remission.

Increased circulating sclerostin levels in rheumatoid arthritis patients: an updated meta-analysis

BACKGROUND

Sclerostin, a regulator of bone metabolism and vascular calcification involved in regulating the Wnt/β-catenin signaling pathway, has been shown to be involved in the pathogenesis of rheumatoid arthritis (RA). However, current results regarding the circulating sclerostin level of RA patients are debatable. This study aimed to evaluate the circulating level of sclerostin in RA patients and briefly summarize its role.

METHOD

PubMed, EMBASE, and the Cochrane Library databases were systematically searched till May 27, 2021, for eligible articles. Useful data from all qualified papers were systematically extracted and analyzed using Stata 12.0 software (Stata Corp LP, College Station, TX, USA).

RESULTS

Overall, 13 qualifying studies including 1030 cases and 561 normal controls were analyzed in this updated meta-analysis. Forest plot of this meta-analysis showed that RA patients had higher circulating sclerostin levels (P < 0.001, standardized mean difference [SMD] = 0.916, 95% CI: 0.235-1.597) compared to normal controls. Subgroup analyses implied that age, region, and assay method were associated with sclerostin level in RA patients.

CONCLUSION

RA patients have higher circulating sclerostin levels, and these was influenced by age, region, and assay method.

Representation of adults with rheumatoid arthritis in US-based randomized controlled exercise trials: a systematic review with meta-analysis

The purpose of this study was to conduct a systematic review with meta-analysis to determine the representation of adults with rheumatoid arthritis (RA) according to sex, race, and ethnicity in exercise randomized controlled trials (RCTs) conducted in the USA. Exercise RCTs ≥ 4 weeks conducted in the USA in adults ≥ 18 years with RA were eligible. Studies were retrieved by searching six electronic databases, cross-referencing and searching a clinical trials registry. Dual, independent, study selection and data abstraction were conducted. The primary outcomes were the proportion of participants in each study according to sex as well as race/ethnicity. Results were pooled meta-analytically using the inverse-variance heterogeneity (IVhet) model after applying the double-arcsine transformation. Of the 1030 unique articles screened, five RCTs representing 353 participants with RA were included. The pooled participant prevalence was 83% (95% CI = 73 to 92%) for women and 17% (95% CI = 8 to 27%) for men, suggesting an over-representation of women and an under-representation of men by approximately 7.4% based on current prevalence US estimates for each. Qualitative examination for race and ethnicity demonstrated an under-representation of racial/ethnic minority groups. There is a lack of representation of men with RA in US-based randomized controlled exercise intervention studies. Additional US-based randomized controlled exercise trials, including greater inclusion and reporting of the racial/ethnic composition of participants, are also needed. Key Points • This systematic review with meta-analysis of US studies found an under-representation of men in randomized controlled trials examining the effects of exercise in those with rheumatoid arthritis (RA). • Qualitative examination according to race/ethnicity found both a lack of reporting as well as under-representation of selected racial/ethnic minorities in US-based randomized controlled exercise studies among adults with RA. • This study highlights the need for additional US-based randomized controlled trials of exercise in adults with RA that better represent the RA population in the USA.

Incidence, sociodemographic factors and treatment penetration of rheumatoid arthritis and psoriatic arthritis in Norway

OBJECTIVES

To evaluate nationwide incidence, sociodemographic associations and treatment penetration of rheumatoid arthritis (RA) and psoriatic arthritis (PsA) in Norway.

METHODS

The study combined data from nationwide registries on the total Norwegian adult population (age ≥ 18). From the Norwegian Patient Registry, incident RA and PsA cases during 2011-2015 were identified with records of first and second healthcare episodes listing RA/PsA diagnostic codes, and ≥ 1 episode in an internal medicine or rheumatology unit with RA/PsA code during the two-year period after the first episode. Dispensed DMARD prescriptions were obtained from the Norwegian Prescription Database. Persons with dispensed DMARD prescriptions or biologic DMARDs given in hospitals > 12 months before the index date were excluded.

RESULTS

Incidence of RA/PsA in Norway was 42/26 per 100,000 person-years (55/28 among women and 28/23 among men). RA peak incidence was observed at ages 70-79 in both sexes, whereas the peak incidence of PsA occurred at ages 50-59. Age- and sex-standardized incidences of RA and PsA were lower among persons with higher education levels. Within a year from the index date, 82.4/57.4% of RA/PsA patients used synthetic DMARDs while 9.4/9.5% used biologic DMARDs.

CONCLUSIONS

Register-based incidence estimates for RA and PsA in Norway are similar to other Nordic countries, but slightly higher than in previous Norwegian studies. Furthermore, we found that higher socioeconomic status was associated with lower incidence of both RA and PsA. Although conventional synthetic DMARDs were less often used in early PsA than RA, frequency of biologic DMARD prescriptions was comparable.

The efficacy of low dose short-term prednisone therapy for remission induction in newly diagnosed rheumatoid arthritis patients

BACKGROUND

The ACR/EULAR recommendations endorse the use of glucocorticoids (GCs) for rheumatoid arthritis (RA) patients' flares and as a bridge to a DMARD. However, the recommendation of low dose short-term monotherapy with (GCs) remains open to the discretion of the clinician. The aim of this study was to assess whether a short-term use of low dose prednisone monotherapy was effective in inducing remission in newly diagnosed RA patients.

METHODS

A retrospective analysis of patients newly diagnosed with RA at a Community Health Center in North Dakota was performed based on the ACR/EULAR RA classification criteria. Demographic and clinical data were abstracted from patients' medical charts. Patients treated with (< 10 mg/day) of prednisone up to 6 months were included. Response to prednisone was analyzed according to pre- and post-treatment DAS28-ESR score and EULAR response criteria.

RESULTS

Data on 201 patients were analyzed. The mean prednisone dose was 8 mg/day (range: 5-10; SD = 1.2) and the mean treatment duration was 42.2 days (12-177; 16.9). Disease severity significantly improved from baseline to follow-up for: tender joint count (8.6 ± 4.8 vs. 1.5 ± 3.3; P < 0.001), swollen joint count (6.2 ± 5.0 vs. 1.4 ± 3.0; P < 0.001), and visual analog pain score (4.8 ± 2.6 vs. 2.1 ± 2.5; P < 0.001). DAS28-ESR disease severity significantly improved from baseline to follow-up: (5.1 ± 1.2 vs. 2.7 ± 1.3; P < 0.001). Per EULAR response criteria, 69.7% of patients showed good response to treatment and 20.4% showed moderate response. 54.2% of patients reached remission.

CONCLUSION

Short-term use of low dose prednisone monotherapy induced disease remission and improved clinical severity of RA in the majority of newly diagnosed patients.

The prevalence of rheumatoid arthritis in Spain

Rheumatoid arthritis (RA) prevalence is believed to be around 1% worldwide, although it varies considerably among different populations. The aim of EPISER2016 study was to estimate the prevalence of RA in the general adult population in Spain. We designed a population-based cross-sectional study. A national survey was conducted between November 2016 and October 2017 involving a probabilistic sample from the general population aged 20 years or older. Subjects were randomly selected for phone screening using a computer-assisted telephone interviewer system. Positive RA screening results were evaluated by a rheumatologist. Cases fulfilled the 1987 ACR and/or the 2010 ACR/EULAR criteria; previous diagnosis established by a rheumatologist and clearly identified in medical records were also accepted regardless of the criteria used. Prevalence estimates with 95% CI were calculated taking into account the design of the sample (weighting based on age, sex, and geographic origin using as a reference the distribution of the population in Spain). 4916 subjects participated in the study and 39 RA cases were confirmed. RA estimated prevalence was 0.82% (95% CI 0.59–1.15). Mean age of RA cases was 60.48 (14.85) years, they were more frequently women (61.5%), from urban areas (74.4%), non-smokers (43.6%), and with a high body mass index (53.8% with overweight). Extrapolating to the population in Spain (approximately 37 million are ≥ 20 years old), it was estimated that there were between 220,000 and 430,000 people aged 20 years or older with RA. No undiagnosed cases were detected, which could be related to the establishment of early arthritis clinics around the country, increasing the rates of diagnosis during early phases of RA.

Prevalence of patients with rheumatoid arthritis and age-stratified trends in clinical characteristics and treatment, based on the National Database of Health Insurance Claims and Specific Health Checkups of Japan

AIM

To estimate the prevalence and age-stratified treatment trends and clinical characteristics of rheumatoid arthritis (RA) in Japan.

METHOD

Using 7 RA definitions, the prevalence of RA in those aged ≥16 years was estimated using the National Database of Health Insurance Claims and Specific Health Checkups of Japan in the fiscal year 2017. We analyzed age-stratified trends in characteristics and treatments.

RESULTS

Of 1 116 122 patients aged ≥16 years with at least 1 RA-related International Classification of Diseases-10 code, 825.7 thousand patients (women, 76.3%) were assessed as having RA with an estimated prevalence of 0.65%. The highest age-stratified prevalence was 1.63% in patients aged 70-79 years. Overall, 60.8% and 7.0% of patients with RA were aged ≥65 years and ≥85 years, respectively. Methotrexate use was most frequent in patients aged 50-59 years (73.0%) and least frequent in patients aged ≥85 years (38.2%). Biologic disease-modifying antirheumatic drugs use was 50.9% in patients aged 16-19 years and decreased to 13.7% in those aged ≥85 years. Preference for the use of tumor necrosis factor inhibitors versus abatacept decreased from 24.0:1 to 1.7:1 in patients aged 16-19 years and ≥85 years, respectively. The prevalence of cardiovascular disease was 3.5% in patients aged 60-69 years and 12.1% in those aged ≥85 years. Overall RA-related orthopedic surgeries were most prevalent in patients aged 70-79 years.

CONCLUSION

The estimated prevalence of patients with RA in Japan was 0.65%. Age-stratified treatment trends and clinical characteristics have been described in a super-aged society for the first time.