Health care utilization and costs associated with functional status in patients with psoriatic arthritis

Abstract

BACKGROUND:

The Health Assessment Questionnaire Disability Index (HAQ-DI) has been validated and widely used in psoriatic arthritis (PsA) clinical trials for the assessment of patient functional status. Significant improvements in the HAQ-DI have been reported in response to therapeutic interventions; however, few US studies have evaluated the economic impact of functional disability in patients with PsA.

OBJECTIVE:

To evaluate the association of functional status with health care resource utilization (HCRU) and total health care costs in US patients diagnosed with PsA.

METHODS:

This retrospective study included adult patients with PsA enrolled in FORWARD between July 2009 and June 2019 who completed 1 or more HAQ-DI questionnaires between January 2010 and December 2019. Patient demographics, clinical characteristics, and patient-reported outcomes were collected from the most recent questionnaire. HCRU and total health care costs (2019 US dollars) for all hospitalizations, emergency department (ED) visits, outpatient visits, diagnostic tests, and procedures were assessed for the 6 months prior to survey completion. Negative binomial regression models (HCRU outcomes) and generalized linear models with γ distribution and log-link function (cost outcomes) were used to assess the relationship between HAQ-DI and HCRU and cost outcomes, respectively.

RESULTS:

A total of 828 patients with PsA who completed HAQ-DI questionnaires were included. The mean (SD) age was 58.5 (13.5) years, 72.3% were female, and 92.3% were White. The mean (SD) disease duration was 17.5 (12.4) years, and the mean (SD) HAQ-DI score at the time of the patients’ most recent questionnaire was 0.9 (0.7). More severe functional disability, measured by higher HAQ-DI score, was significantly associated with increased risk (incident rate ratio [95% CI]) of hospitalizations (1.68 [1.11-2.55]), ED visits (2.09 [1.47-2.96]), outpatient visits (1.14 [1.05-1.24]), and diagnostic tests (1.42 [1.16-1.74]). There was also a significant positive association between greater HAQ-DI score and increased total annualized health care costs (incremental amount [95% CI], 1.13 [1.03-1.23]) and medical costs (1.38 [1.13-1.69]), but there was no significant association found with pharmacy costs. Total adjusted average patient medical costs increased with increasing HAQ-DI score.

CONCLUSIONS:

Among patients with PsA enrolled in FORWARD, more functional disability—as measured by higher HAQ-DI scores—was associated with greater HCRU and increased total health care costs. These results suggest that improving functional status in patients with PsA may reduce economic burden for health care payers and systems.

Plain language summary

Psoriatic arthritis (PsA) can cause pain and stiffness that may affect what patients do every day. Patients can answer certain questions as part of the Health Assessment Questionnaire Disability Index (HAQ-DI) to show how PsA can affect how they are able to do common things. Our study shows that patients with a worse HAQ-DI score have more hospitalizations, emergency department (ED) visits, doctor visits, and tests and spend more money to care for their disease.

Implications for managed care pharmacy

The physical burden of PsA is widely assessed through the HAQ-DI. Higher HAQ-DI scores were associated with increased hospitalizations, ED visits, outpatient visits, diagnostic testing, and health care and medical expenditures in this study of US patients with PsA. These results suggest that improved functional status in patients with PsA, possibly through therapeutic interventions, may reduce the resource and economic burden for health care payers and systems.

Psoriatic arthritis (PsA) is a heterogeneous chronic inflammatory disease that affects several aspects of the musculoskeletal system.^1,2 Patients with PsA present with diverse clinical manifestations that can include peripheral arthritis, axial disease, dactylitis, enthesitis, and skin and nail involvement.^1,2 PsA is estimated to affect 0.25% of the total US population.³ The typical age of PsA onset is 30 to 50 years, and patients can experience a high lifetime impact of disease burden.⁴ Without adequate intervention, PsA can cause progressive and permanent joint damage.⁵

The substantial burden of physical disability, pain, and cutaneous involvement experienced by patients with PsA has been shown to significantly impact patient quality of life and physical function.^6,7 The Health Assessment Questionnaire Disability Index (HAQ-DI) has been validated and is widely used in PsA clinical trials and real-world settings for the assessment of patient functional status.^8,9 Several studies in PsA have shown significant improvements in HAQ-DI scores with biologic therapies.^9-14 HAQ-DI scores have also been significantly associated with clinical disease severity and are a predictor of clinical response in patients with PsA.^15,16

In addition to reduced quality of life and physical function, the clinical manifestations related to PsA also correspond with a considerable economic burden and increased health care resource utilization (HCRU) that is required for disease treatment and management.¹⁷ Patients with PsA have been shown to have significantly higher total health care costs and increased HCRU compared with controls,^18,19 and worse physical disability has been associated with significant economic burden.²⁰ However, most studies evaluating the impact that physical impairment has on medical costs and HCRU have been performed outside of the United States^20-23; there are limited data on the economic impact of functional disability in US patients with PsA.^19,24-26 The purpose of this study was to assess the relationship between HAQ-DI scores and HCRU and total health care costs in patients with PsA from a national US registry.

Methods

DATA SOURCE AND PATIENT POPULATION

FORWARD, the National Databank for Rheumatic Diseases, is a longitudinal observational registry that collects patient-reported data via questionnaires administered every 6 months.^27,28 Patients with patient-reported, physician-diagnosed disease conditions are recruited from rheumatology clinics throughout the United States. Questionnaires are administered online, by mail, or by telephone. Data collected include patient demographics, comorbidities, symptoms, functional status, quality of life, medication use, and HCRU.

This retrospective cohort study included adult patients with a diagnosis of PsA enrolled in FORWARD between July 2009 and June 2019 who completed 1 or more questionnaires, including the HAQ-DI (Supplementary Figure 1 ^{(370.6KB, pdf)} , available in online article). Patients with a primary diagnosis of rheumatoid arthritis, reactive arthritis, or an autoimmune disorder other than PsA were excluded.

ASSESSMENTS AND OUTCOMES

Data were collected from the most recent questionnaire and included (1) patient demographics (age, sex, race, education level, insurance type, marital status, and geographical location), (2) clinical characteristics (disease duration, smoking status, rheumatic disease comorbidity index,²⁹ and comorbid conditions), (3) treatment profile (disease-modifying antirheumatic drugs [DMARDs], biologics, prednisone, nonsteroidal anti-inflammatory drugs, and opioids), (4) patient-reported outcome (PRO) measures of health and functional status, including the most commonly used PRO in PsA, HAQ DI scores (20 questions in 8 categories; scale 0-3 each category; a higher score represents worse functional ability),^8,30 patient-reported heel pain (none, mild, moderate, or severe), patient-reported back pain (none, mild, moderate, or severe), and Patient-Reported Outcomes Measurement Information System 29-item questionnaire (PROMIS-29; composed of categories measuring physical function, anxiety, depression, fatigue, pain interference, sleep disturbance, and ability to participate in social roles and activities; scale 1-5 each category; a higher score is worse for all categories, with the exception of physical function and ability to participate in social roles and activities), (5) work productivity measures (questionnaire responses for current working status (no vs yes), % work time missed, and % impairment while working), and (6) patient-reported HCRU for hospitalizations (based on Diagnosis Related Group codes), emergency department (ED) visits, outpatient visits, procedures, and diagnostic tests. Missing data may have occurred owing to the varying time points PROs were added to the FORWARD registry; missing data were not imputed, and when possible, the last value was carried forward. The FORWARD registry uses a series of data quality control checks to guarantee data of high standard, and complete details are previously published.²⁸

HCRU and medical costs (2019 US dollars) were evaluated for the 6 months prior to survey completion and annualized by multiplying by 2 in order to facilitate interpretation using the assumption of resource utilization at a uniform rate. Information on patient-reported HCRU collected from questionnaires was then used to estimate medical costs based on 2019 Medicare physician fee schedule for medical services and ProspectoRx for pharmacy costs. Costs comprised medical costs (costs for all hospitalizations, ED visits, outpatient visits, diagnostic costs, and procedures [eg, X-ray, computed tomography scan, or magnetic resonance imaging]), pharmacy costs (costs associated with physician-prescribed medications, including conventional and biologic disease-modifying antirheumatic drugs, general arthritis drugs, opioids, and other costs [medications for the following categories: general, gastrointestinal, proton pump inhibitors, H2 inhibitors, osteoporosis, biphosphates, heart, antilipemics, anticoagulants, antituberculars, antipsychotropics, sleep, and anticonvulsants]), and total health care costs (medical costs and pharmacy costs).

The primary objective was to evaluate the association of functional status as measured by the HAQ-DI with annualized HCRU and health care costs. Secondary objectives were to assess HCRU and total health care costs by HAQ-DI score increments using predictive models and evaluate the association between additional PRO measures and HCRU and health care costs.

DATA ANALYSIS

Data on patient demographics, clinical characteristics, treatment profiles, PRO measures, and work productivity were summarized descriptively.

For analyses of associations between functional disability and outcomes of interest, HAQ-DI score was evaluated in 0.5-unit increments for descriptive analysis but analyzed as a continuous variable in the models. The association between HAQ-DI score and annualized HCRU outcomes was assessed using negative binomial regression models adjusted for confounders. Selection of confounders was based on those determined to have a statistically significant (P < 0.05) relationship with the outcome of interest in univariate models. Additionally, the relationship between HAQ-DI score and annualized health care costs was assessed using a generalized linear model with γ distribution and log-link function adjusted for confounders.

HAQ-DI scores are specific to patients with PsA, and some researchers may be interested in cross-disease quality of life measures, so we have included analysis of the associations between PROMIS-29 categories and HCRU and costs. These associations were determined through the transformation of PROMIS-29 raw scores to standardized T-score metrics with a mean (SD) of 50 (10). PROMIS-29 raw score conversion tables used to produce the final T-scores have previously been described.³¹

Multivariable regression models were conducted to estimate the relationship between HCRU and cost outcomes (total health care cost, total medical costs, and pharmacy cost) and functional disability. A sensitivity analysis was computed similar to the multivariable primary analyses, with each model adjusted for race, income, education level, and geographic region. P values < 0.05 were considered statistically significant. All analyses were performed using STATA software version 14.2 (StataCorp, LLC).

Results

STUDY POPULATION AND PATIENT CHARACTERISTICS

A total of 828 patients with PsA who completed 1 or more HAQ-DI questionnaires between 2009 and 2019 were included in this analysis. Overall, 72.3% of patients were female and 92.3% were White (Table 1). The mean (SD) age and disease duration were 58.5 (13.5) and 17.5 (12.4) years, respectively. The mean (SD) HAQ-DI score at the time of the patients’ most recent questionnaire was 0.9 (0.7). More than one-third of patients (36.4%) had a rheumatic disease comorbidity index score of greater than or equal to 3, and the most common comorbidities experienced by patients in this study population were psoriasis (67.5%), hypertension (36.6%), and dyslipidemia (33.5%). Additionally, 59.4% of patients reported that they experienced moderate or severe back pain, and nearly one-half of patients (46.9%) experienced any heel pain. Only 42.7% of patients were employed, and these patients reported a mean (SD) of 8.9% (51.5%) missed work time related to PsA.

TABLE 1.

Demographics, Clinical Characteristics, and Disease Activity Among Patients With PsA Who Completed 1 or More HAQ-DI Questionnaires

Characteristica	Patients with PsA (N = 828)
Age, mean (SD), years	58.5 (13.5)
Sex, n (%)
Female	599 (72.3)
Male	229 (27.7)
White race, n (%) [n]	697 (92.3) [755]
Education (highest year of education), n (%)	n = 753
College graduate or master’s or doctorate degree	572 (76.0)
Insurance status, n (%)	n = 828
Private	263 (31.8)
Medicare	362 (43.7)
Medicaid	68 (8.2)
Other	120 (14.5)
None	15 (1.8)
Urban geographic location, n (%)	638 (79.2) [806]
Year most recent questionnaire completed, n (%)
2010	102 (12.3)
2011	90 (10.9)
2012	86 (10.4)
2013	71 (8.6)
2014	51 (6.2)
2015	49 (5.9)
2016	52 (6.3)
2017	51 (6.2)
2018	110 (13.3)
2019	166 (20.1)
Physician-confirmed diagnosis, n (%)b	455 (55.0)
Disease duration, mean (SD), years [n]	17.5 (12.4) [759]
Rheumatic disease comorbidity index (0-9), n (%)	n = 828
0	136 (16.4)
1	215 (26.0)
2	176 (21.3)
≥ 3	301 (36.4)
Comorbid condition, n (%) [n]c
Psoriasis	553 (67.5) [819]
Hypertension	303 (36.6) [828]
Dyslipidemia	274 (33.5) [819]
Depression	202 (24.4) [828]
Gastrointestinal disorders	142 (17.2) [828]
Cardiovascular disease	140 (16.9) [828]
Diabetes mellitus	105 (12.7) [828]
Smoking status, n (%)	n = 828
Ever smoker	318 (38.4)
PsA treatment history (current and past use), n (%) [n]d
DMARDs	413 (49.9) [801]
Biologics	376 (45.4) [801]
Prednisone	93 (11.2) [735]
NSAIDs	298 (36.0) [801]
Opioids	218 (26.3) [828]
PsA disease activity and quality of life, mean (SD) [n]
HAQ-DI	0.9 (0.7) [828]
PROMIS-29, mean (SD) [n]
Physical function	41.8 (9.2) [407]
Anxiety	51.6 (9.3) [38]
Depression	51.3 (10.1) [37]
Fatigue	54.7 (12.1) [406]
Pain interference	59.1 (9.8) [274]
Sleep disturbance	53.0 (9.3) [404]
Satisfaction with social roles	47.7 (10.3) [407]
Heel pain, n (%)	n = 490
Mild	123 (25.1)
Moderate/severe	107 (21.8)
Back pain, n (%)	n = 662
Mild	178 (26.9)
Moderate/severe	393 (59.4)
Work productivity
Current working status (n = 801), n (%)
Unemployed	27 (3.4)
Employed	342 (42.7)
Retired	241 (30.1)
Disabled	126 (15.7)
Other (eg, going to school, retired, housekeeping)	65 (8.1)
Missed work time, mean (SD), %	8.9 (51.5) [321]

^a [n] represents the number of patients with data available.

^b FORWARD is a self-reported longitudinal study. Diagnosis is validated by physician when possible.

^c Comorbidities reported by more than 10% of the study cohort.

^d Treatment history included use at the time of survey completion and use within the 6 months prior.

DMARD = disease-modifying antirheumatic drug; HAQ-DI = Health Assessment Questionnaire Disability Index; NSAID = nonsteroidal anti-inflammatory drug; PROMIS-29 = Patient-Reported Outcomes Measurement Information System 29-item questionnaire; PsA = psoriatic arthritis.

HCRU AND HEALTH CARE COSTS FOR PATIENTS WITH PsA

Adjusted mean (SD) annualized HCRU was primarily attributed to the number of outpatient visits (19.86 [14.26]) and diagnostic tests (5.07 [6.76]) for patients with PsA included in this study (Table 2). Overall, mean (SD) annualized total health care cost was $46,320 ($39,075), which was driven by pharmacy costs ($39,997 [$36,961]). Additionally, annualized mean (SD) total medical cost related to PsA disease management was $6,323 ($12,699), and expenditures due to hospitalizations ($3,262 [$11,906]) and outpatient visits ($2,234 [$1,651]) were the highest among medical costs.

TABLE 2.

Average Annualized HCRU and Costs Among Patients With PsA (N = 828)

	Mean (SD)
HCRU, no. of uses
Hospitalization	0.30 (0.94)
ED visit	0.51 (1.27)
Outpatient visit	19.86 (14.26)
Diagnostic tests	5.07 (6.76)
Procedures	0.41 (1.02)
Medical costs, USD
Total direct health care costs	46,320 (39,075)
Medical costsa	6,323 (12,699)
Hospitalization	3,262 (11,906)
ED visit	190 (476)
Outpatient visit	2,234 (1,651)
Diagnostic test	473 (638)
Procedure	164 (514)
Pharmacy costsb	39,997 (36,961)

^a Includes hospital admissions, ED visits, outpatient visits, diagnostic tests, and procedures.

^b Pharmacy costs include costs associated with physician-prescribed medications, including conventional and biologic disease-modifying antirheumatic drugs, general arthritis treatments, opioids, and other costs.

ED = emergency department; HAQ-DI = Health Assessment Questionnaire Disability Index; HCRU = health care resource utilization; PsA = psoriatic arthritis; USD= U S dollar.

DECREASED PHYSICAL FUNCTION IS ASSOCIATED WITH INCREASED HCRU

A direct relationship between physical function, as assessed by the HAQ-DI, and HCRU was found among patients with PsA included in this study. A 1.0-unit increase in HAQ-DI score, which represents increased physical disability, was significantly associated with increased risk (incidence rate ratio [IRR] [95% CI]) number of hospitalizations (1.68 [1.11-2.55]; P < 0.05), ED visits (2.09 [1.47-2.96]; P < 0.01), outpatient visits (1.14 [1.05-1.24]; P < 0.01), and diagnostic tests (1.42 [1.16-1.74]; P < 0.01) (Figure 1).

FIGURE 1.

The Relationship Between HAQ-DI and HCRU in Patients With PsA^a

Furthermore, a sensitivity analysis, which adjusted models for race, income, education, and geographical area, in addition to the significant confounders selected in the univariate models, further demonstrated the direct relationship between HAQ-DI score and HCRU. In this sensitivity analysis, a significantly increased risk (IRR [95% CI]) of hospitalizations (1.73 [1.08-2.79]; P < 0.05), ED visits (2.07 [1.37-3.11]; P < 0.01), outpatient visits (1.16 [1.07-1.27]; P < 0.01), and diagnostic testing (1.50 [1.21-1.85]; P < 0.01) was also associated with increased HAQ-DI score (Supplementary Figure 2 ^{(370.6KB, pdf)} ).

Additionally, there was a significant inverse relationship between the PROMIS-29 categories that assess improved physical function and satisfaction with ability to perform social roles and HCRU for patients with PsA included in this study; a higher score for each of these 2 categories represented a better outcome (Supplementary Figure 3 ^{(370.6KB, pdf)} ). A higher score for physical function was associated with a minor but significant decreased risk of hospitalizations, ED visits, outpatient visits, and diagnostic testing; similarly, a higher score for satisfaction with ability to perform social roles was associated with a small but significant reduced risk of ED visits, outpatient visits, and diagnostic testing. Furthermore, a significant direct relationship between increased pain interference assessment in the PROMIS-29 questionnaire and ED visits, outpatient visits, and diagnostic testing was determined. Similarly, patient-reported moderate/severe low back pain was significantly associated with a 27% and 37% increased risk (IRR [95% CI]) of outpatient visits (1.27 [1.14-1.42]; P < 0.01) and diagnostic testing (1.37 [1.04-1.79]; P < 0.05) compared with none/mild low back pain as reference.

ASSOCIATION OF HAQ-DI SCORE WITH MEDICAL COSTS

There was a significant direct relationship (incremental amount [IA] [95% CI]) between increased HAQ-DI score and increased total health care costs and medical costs, as evidenced by a 1.13 (1.03-1.23; P < 0.01) and 1.38 (1.13-1.69; P < 0.01) times increase in costs, respectively, for each 1.0-unit increase in HAQ-DI score (Table 3). In the sensitivity analysis, adjusted annualized total health care costs and medical costs increased 13% and 41%, respectively, for each 1.0-unit increase in HAQ-DI score among patients with PsA who completed the HAQ-DI questionnaire (IA [95% CI], 1.13 [1.02-1.25]; P < 0.05 and 1.41 [1.14-1.74]; P < 0.01, respectively) (Table 3). There was no significant association between pharmacy costs and HAQ-DI score (IA [95% CI], 1.06 [0.96-1.18]) in the primary analysis; however, pharmacy costs significantly increased with increasing HAQ-DI score in the sensitivity analysis (1.16 [1.04-1.30]; P < 0.01). Additionally, according to model predictions, the average annualized patient medical costs increased across worsening HAQ-DI score for patients with PsA included in this study (Figure 2).

TABLE 3.

Adjusted Association Between Health Care Costs and PRO Measures in Patients With PsA Who Completed the HAQ-DI

Medical costs, USD	Primary analysisa (N = 828)	Sensitivity analysisb (N = 828)
Total health care
Base value	15,804	15,647
Incremental amountc (95% CI)	1.13 (1.03-1.23)d P = 0.007	1.13 (1.02-1.25)e P = 0.017
Total medical f
Base value	4,238	2,299
Incremental amountc (95% CI)	1.38 (1.13-1.69)d P = 0.002	1.41 (1.14-1.74)d P = 0.002
Pharmacy
Base value	9,730	8,769
Incremental amountc (95% CI)	1.06 (0.96-1.18) P = 0.227	1.09 (0.96-1.22) P = 0.164

^a Adjusted for covariates significant at 5% in bivariate analyses: geographic region, rheumatic disease comorbidity index, and opioid use.

^b Model adjusted for race, income, education, and geographical area.

^c Multiplicative factor that represents the incremental dollar amount for change from corresponding base value per 1-unit increase in HAQ-DI score.

^d P < 0.01.

^e P < 0.05.

^f Includes costs for hospitalizations, emergency department visits, outpatient visits, procedures, and diagnostic tests.

HAQ-DI = Health Assessment Questionnaire Disability Index; PRO = patient-reported outcome; PsA = psoriatic arthritis; USD = US dollar.

FIGURE 2.

Average Annualized Patient Medical Costs Across HAQ-DI Scores in Patients With PsA^a

Moreover, there was a significant inverse relationship between improvements in physical function and satisfaction with social roles of the PROMIS-29 questionnaire and minor reductions in total health care costs and medical costs with physical function and pharmacy costs and satisfaction with social roles (Supplementary Table 1 ^{(370.6KB, pdf)} ).

Discussion

This study demonstrates the relationship between functional disability, as measured by HAQ-DI score, and HCRU and health care costs in patients with PsA in a real-world setting in the United States. In this study population, a significant association was found between increasing HAQ-DI score, which represents decreased physical function, and increased risk of HCRU and health care costs. Overall, the main drivers of the relationship between HAQ-DI score and HCRU were hospitalizations, ED visits, outpatient visits, and diagnostic testing, and total health care expenditures were predominately attributed to pharmacy costs related to PsA. Our findings demonstrate the high HCRU and economic burden related to impaired physical function seen with PsA; however, there are few studies with similar data available, especially in the United States.

One retrospective observational study using US claims data for 10,832 patients with PsA reported similar trends in HCRU and costs over a 12-month period.²⁵ Overall mean (SD) HCRU frequency was lower than that seen in our study, yet office visits (10.7 [7.0]) were the main driver of HCRU, which is consistent with our results. Additionally, similar medical costs and trends in resource use were reported, with the highest mean (SD) costs attributed to hospitalizations ($3,064 [$17,444]) and office visits ($1,458 [$2,097]), and pharmacy costs made up more than one-half of the medical plus prescription costs, as in our study cohort. A systemic literature review of European studies that assessed the cost burden associated with PsA reported the estimated annual cost per patient to be $10,924 to $17,050 for all patients and an even higher cost ($57,000) for patients with severe disease; costs were calculated in USD through conversion to international dollars using purchasing power parity exchange rates for 2015.²² Among studies included in the literature review, the range for direct costs for patients with PsA were determined to be $3,693 to $8,871, which is similar to the mean (SD) medical costs estimated in our study cohort ($6,323 [$12,699]). Similar patterns in the primary drivers of HCRU and medical costs found in our study are comparable to those previously reported, although patients in our study population had higher resource use and increased health care costs.

The high HCRU and health care costs observed in our study cohort may be impacted by the population characteristics included, such as age, comorbidities, and treatment history. The population included in this study had a relatively higher mean (SD) age of 58.5 (13.5) years compared with similar studies, more than one-third of patients (36.4%) had a rheumatic disease comorbidity index (0-9) score of greater than or equal to 3, and nearly one-half of patients had a treatment history with DMARDs (49.9%) or biologics (45.4%).

The increased risk of developing comorbidities is established in PsA, and several comorbid conditions have been associated with the disease.^32,33 A large observational study that assessed claims data for patients with PsA found that patients had significantly higher medication use and PsA-associated comorbidities compared with matched controls without PsA.¹⁹ Overall, patients with PsA had a significantly higher comorbidity burden related to the disease, as evidenced by greater HCRU for inpatient admissions, length of hospital stay, ED visits, and outpatient services, as well as higher annual direct medical costs and all-cause pharmacy costs (all P < 0.0001) than matched controls without PsA. Additionally, a retrospective study that included claims data for 2,560 patients with PsA who initiated a biologic within 1 year of the start of the study reported similar trends in HCRU and health care costs as seen with our study cohort.²⁶ Among patients who newly initiated a biologic, mean (SD) total health care costs ($27,021 [$42,199]) and pharmacy costs ($11,509 [$20,640]) were lower than those seen in our study, which could be because more than one-half of patients in that study remained on their index biologic, potentially reducing the need for increased HCRU and costs related to treatment changes for that population. However, as seen with our study population, medical costs were primarily attributed to high outpatient costs ($11,154 [$19,280]), and mean (SD) number of outpatient visits (28.60 [22.21]) was the main driver of HCRU. Higher medical costs and HCRU were found among patients who discontinued or switched biologics, which are both known trends in PsA,^34-36 likely due to the recommended stepwise treatment strategies and inadequate response to therapies.^37,38 Furthermore, in the study by Greenberg et al²⁵ discussed above, increased comorbidities, as measured by Elixhauser Comorbidity Index score, and baseline biologic use were determined to be significant predictors (coefficient [95% CI]) of all-cause direct costs with 1.95 (1.78-2.13; P < 0.001) and 3.12 (2.98-3.25; P < 0.001) times increase in direct costs, respectively. Future studies are needed to evaluate associations between patient characteristics, such as age, comorbidity burden, prescription medication use, and prior exposure to DMARDs, which may impact HCRU and medical costs in regard to physical function.

Each of the patient characteristics that may have contributed to the increased HCRU and health care costs may also affect patient quality of life and impact physical function, which could explain, in part, the associations observed in our study; however, our findings are in line with other comparable real-world studies. A similar survey-based study of a cohort of 101 patients with PsA in the United Kingdom found that patients with more impaired functional disease status, as evidenced by higher HAQ-DI scores, had higher total health care costs, demonstrated by a significant positive association between total costs and HAQ-DI score.²⁰ Additionally, a study that included data for 2,822 patients enrolled in registries in the United Kingdom found a significant direct association between increased HAQ-DI score and total health care costs (β coefficient [SE], 2.05 [0.006]; P < 0.001) and reported that prescription costs accounted for more than one-third (34%) of total costs.²³ Similar findings were seen in our sensitivity analysis, in which pharmacy costs significantly increased with worsening HAQ-DI score, when adjusted for race, income, education, and geographical region of patients. The findings from our study provide further evidence of the direct relationship between impaired physical function and HCRU and costs, as measured by HAQ-DI score.

LIMITATIONS

Our study provides important information on the relationship between HCRU and health care costs among patients with PsA, yet this study does have inherent limitations due to the nature of the real-world observational study design. Because participation in the FORWARD is voluntary, participants may play a more proactive role in disease management that may not be representative of the general population of patients with PsA. Furthermore, most patients in this cohort were White, earned a college or graduate degree, and were from urban regions, which does not represent diverse patient populations and may impact the generalizability of these findings. The long study period presents another limitation because the therapeutic landscape for PsA has changed substantially over the course of the study; however, to include a robust study sample, we retained the full cohort over the 10-year study period, which would reduce by approximately one-half if limited to those patients who completed a questionnaire in the last 5 years to account for the initiation of notable approvals for PsA treatments. Additionally, data for HCRU and other questionnaire components are patient reported and may be subject to recall bias, and estimates for medical costs may be imprecise if costs inaccurately related to comorbid conditions skewed the values. Moreover, HCRU and cost data were captured over a 6-month period and annualized by multiplying by 2, which assumes HCRU and costs over a 12-month period were consistent with those of the 6-month observation period. This study is also questionnaire based, so the time sequence of HCRU and costs outcomes and functional scores cannot be clearly identified; thus, the results should be interpreted within this limitation. However, IRRs are used because variables are reported as number per given time period and analyzed similar to prior publications.³⁹ Additional studies need to be conducted over longer time periods and with assessments for the impact of patient characteristics on HCRU and medical costs to substantiate the findings from this study.

Conclusions

More severe disease activity and worse physical disability—as measured by higher HAQ-DI scores—were associated with greater HCRU and increased health care costs for patients with PsA in this study population. There was a significant direct relationship between increased HAQ-DI score and increased HCRU for hospitalizations, ED visits, outpatient visits, and diagnostic testing observed in our study. Additionally, for every 1.0-unit increase in HAQ-DI score, which represents decreased physical function, total health care and medical costs significantly increased. These results suggest that interventions that improve functional status may reduce economic burden for patients and health care systems throughout the course of disease.