The use of patient-reported outcome measures to classify Type 1 & 2 SLE activity

Abstract

Objective.

We developed a model that categorizes SLE activity into two dimensions: Type 1 SLE consists of inflammatory activity, including arthritis, nephritis, and rashes; Type 2 SLE includes fatigue, myalgia, mood disturbance, and cognitive dysfunction. Patient-reported outcome (PRO) measures have received attention as a way to capture symptomatology of SLE. The objective of this study was to explore the use of existing PRO measures to classify Type 1 & 2 SLE activity.

Methods.

SLE patients completed three questionnaires: Systemic Lupus Activity Questionnaire (SLAQ), Polysymptomatic Distress Scale (PSD), and Patient Health Questionnaire (PHQ-2). SLEDAI and physician global assessments (PGA; 0–3) for Type 1 and Type 2 activity were also recorded. High Type 1 SLE activity was defined as cSLEDAI ≥4 (scored without labs), SLEDAI ≥6, active nephritis, or Type 1 PGA ≥1.0. High Type 2 SLE activity was defined as Type 2 PGA ≥1.0. Patients with both high Type 1 and 2 activity were defined as Mixed SLE, and patients with low Type 1 and 2 activity were defined as Minimal SLE. Data were reduced with a factor analysis. Using a reduced set of 13 variables, multinomial logistic regression models estimated the probability of Minimal, Type 1, Type 2, and Mixed SLE classification.

Results.

The study included 208 patients with SLE. The model accurately predicted the clinician-based Type 1 & 2 SLE classification in 63% of patients; 73% of patients had their Type 1 SLE activity accurately predicted and 83% had their Type 2 SLE activity accurately predicted. Performance varied by group: 87% of Minimal patients were correctly predicted to be in the Minimal SLE group, yet only about one-third of patients in the Type 1 group were correctly predicted to be in the Type 1 group.

Conclusions.

Our findings indicate Type 2 SLE activity can be identified by patient-reported data. The use of PROs was not as accurate at predicting Type 1 activity. These findings highlight the challenges of using PROs to categorize and classify SLE symptoms since some manifestations of Type 1 activity (e.g., nephritis) may be essentially clinically silent while other Type 1 manifestations may cause severe symptoms.

Introduction

Systemic lupus erythematosus (SLE) is a systemic autoimmune disease associated with a wide range of signs and symptoms that vary among patients. While patients with SLE may experience symptoms with a clear-cut relationship to inflammation (e.g., arthritis and rash), they also commonly experience symptoms whose relationship to inflammation is less certain. These symptoms, which include fatigue, myalgias and widespread pain, are among the most frequently reported complaints of patients; furthermore, patients often consider these symptoms as signs of active lupus. As a group, these symptoms can significantly impact patients’ quality of life especially as they do not usually respond to anti-inflammatory or immunosuppressive medications.^1–8

The frequency and severity of symptoms in SLE that are not clearly related to inflammation represent an important challenge in patient evaluation and management. While physicians may ascribe these symptoms to concomitant depression or fibromyalgia, patients identify them as manifestations of their SLE. This situation led our group to develop the Type 1 & 2 SLE Model to categorize symptomatology of SLE into a more unified framework that incorporates the perspective of both patients and providers.⁹ In this conceptual model, Type 1 SLE symptoms, which are considered by physicians to represent active lupus, include synovitis, rashes, serositis and CNS lupus; objective laboratory findings (e.g., proteinuria) are also indicators of Type 1 SLE. Type 2 SLE symptoms include fatigue, myalgia, mood disturbance, widespread pain and cognitive dysfunction. Importantly, in this model, both Type 1 and Type 2 symptoms are manifestations of SLE despite differences in the potential role of inflammation in their etiology.

Patient-reported outcome (PRO) measures have received attention as a way to capture more completely symptomatology of SLE and thereby better assess disease activity and its relationship to disease outcome. The Duke Lupus Clinic has been successfully using PROs in the clinical setting, however there is still not widespread use of PROs in many rheumatology settings, however, since current measures of disease activity rely on observable and measurable findings, despite the severity of symptoms that can be experienced by patients with SLE. Another factor affecting the use of PROs may relate to the discordance between physician-reported lupus activity (i.e., Type 1 SLE) and patient-reported activity (i.e., Type 2 SLE). According to over 1,000 patients with SLE, a patient-defined flare includes extreme fatigue, aching joints, muscle weakness or pain, and forgetfulness.¹⁰ Of these symptoms, only joint pain would be counted as active lupus by a physician; the other symptoms, which are likely manifestations of Type 2 SLE activity, would usually not be judged as evidence of inflammatory disease activity.

As shown in other studies, integration of PRO measures into routine clinical care can improve patient-clinician communication and facilitate more meaningful and focused conversations.^11–16 In our work on SLE activity, we have used PROs to define Type 2 activity,¹⁷ but currently, no single PRO measure has been designed specifically to allow categorization of patients according to the Type 1 & 2 SLE Model. A single measure indicating the extent of Type 1 & 2 SLE activity could allow for better assessment of disease activity. We were interested in the possibility that existing PRO measures could be combined in an instrument to distinguish between Type 1 & Type 2 SLE activity. Such a PRO measure could also improve therapy by better targeting of interventions to symptoms that most affect patients while limiting potential over-treatment with immunosuppressives. The objective of this study was therefore to explore the use of existing PRO measures to create a single, cohesive measure to identify Type 1 & Type 2 SLE activity without physician input.

Methods

Patient Population.

All patients were ≥18 years old and met 1997 ACR or 2012 SLICC criteria for SLE.^{18, 19} Patients were enrolled in the Duke Lupus Registry (DLR), a prospective registry of adult patients with SLE who receive rheumatology care from six treating rheumatologists in the outpatient Duke University Clinic. All patients signed informed consent to participate in the registry (Duke Health IRB Pro00008875). This was a cross-sectional analysis of patients’ first visit in the registry between 2018 and 2019 (Duke Health IRB Pro00102300).

Data Collection.

At each clinic visit, patients with SLE completed questionnaires, and their treating rheumatologist completed disease activity measures, including the SLE Disease Activity Index (SLEDAI), Physician’s Global Assessment of Disease Activity (PGA) for Type 1 activity, and a PGA for Type 2 activity.^20–22 The PGAs allowed rheumatologists to rate the severity of Type 1 & 2 SLE activity separately on scales ranging from 0 (no activity) to 3 (severe activity). All data were stored in a secure REDCap database.

Patient-Reported Outcomes Measures

The Polysymptomatic Distress Scale (PSD) includes many components of Type 2 SLE and consists of two subscales: widespread pain index (WPI) and symptom severity score (SSS).^23–26 For the WPI, patients indicated if they had experienced pain in any of 19 areas of the body in the past 1 month. The SSS asked patients to report the extent of their fatigue, cognitive symptoms, and waking unrefreshed over the past 1 month with response options of (0) no symptoms, (1) slight, mild or intermittent problems, (2) moderate problems, often present at a moderate level, and (3) severe, continuous, life-disturbing problems. Additionally, patients indicated if they had experienced a headache, pain or cramps in the lower abdomen, or depression in the last 6 months. The total SSS ranged from 0 to 12. The total PSD score, combining the WPI and SSS, ranged from 0 to 31.

The Systemic Lupus Activity Questionnaire (SLAQ) is a 24-symptom patient-reported measure of SLE activity that moderately correlates with the physician-reported Systemic Lupus Activity Measure.^{27, 28} The SLAQ asks patients to report if they have had experienced lupus symptom in the past 1 month and indicate the severity of each symptoms (mild, moderate, or severe). The SLAQ covers a range of symptoms including traditional Type 1 symptoms (e.g., swelling in joints, pain or stiffness in joints, rash on cheeks), as well as Type 2 symptoms (e.g., fatigue, forgetfulness, and depression). In our clinic, we include 7 additional symptoms for patients to report experiencing in the past month: anxiety, dry eyes or dry mouth, edema, elevated urine protein, foamy urine, hypertension, and pain or burning with urination.

The SLAQ also asks patients to report if they had a flare in the past 1 month and the severity of the flare (mild, moderate or severe). Finally, patients rate their lupus disease activity during the last 1 month based on their most active day on a scale from 0 (no activity) to 10 (most activity).

The Patient Health Questionnaire (PHQ-2) is a PRO that asks how often over the past 2 weeks patients had been bothered by (a) little interest or pleasure during things and (b) feeling down, depressed, or hopeless.²⁹ Response options included (0) not at all, (1) several days, (2) more than half the days, and (3) nearly every day. Total scores range from 0 to 6, with a score of ≥3 indicating that a major depressive disorder was likely.

Clinician-based classification of Type 1 & 2 SLE activity.

For this study, patients were classified based on their Type 1 and Type 2 SLE activity. High Type 1 SLE activity was defined as clinical SLEDAI ≥4 (scored without labs), SLEDAI ≥6, active nephritis, or Type 1 PGA ≥1.0.¹⁷ High Type 2 SLE activity was defined as Type 2 PGA ≥1.0. The four potential classification groups were:

• Minimal: Low Type 1 Activity, Low Type 2 Activity

• Type 1: High Type 1 Activity, Low Type 2 Activity

• Type 2: Low Type 1 Activity, High Type 2 Activity

• Mixed: High Type 1 Activity, High Type 2 Activity.

Analysis

Factor Analysis for Data Reduction.

In total, 78 patient-reported and clinician-reported variables were available for inclusion in the analyses. A factor analysis was conducted to reduce the number of variables for inclusion in the model. All components of the SLAQ, PSD, PHQ-2, SLEDAI, Type 1 PGA, and Type 2 PGA were included. The factor analysis used an oblique rotation to enhance interpretability among multiple factors. The number of factors retained was based on review of the scree plot (presenting amount of variance explained by each consecutive factor) and clinical interpretability.^{30, 31} We removed and combined PRO indicators based on:

1. indicators with dual loadings >0.3 were candidates for deletion,

2. clinical utility, with symptoms that were rare or non-specific to SLE considered for deletion, and

3. highly correlated items (>.70), which were reviewed by clinicians and indicators were removed if they were determined by clinicians to be duplicative.

Multinomial logistic regression models.

Using the reduced set of variables identified from the factor analysis as independent variables and clinician classification of patients’ SLE type as the dependent variable, we estimated the probability of Minimal, Type 1, Type 2, and Mixed SLE classification with a multinomial logistic regression model. We compared the model’s predicted group membership with the clinician’s classified SLE group. The average predicted probability was calculated for each group, and the proportion of patients with correctly classified Type 1 & Type 2 SLE activity was determined. The hit rate, the chance hit rate, and Huberty’s I index were calculated to evaluate the model’s ability to predict group classification over chance. Huberty’s I index of >0.35 indicates a large effect.³² Differences in means for patient-reported symptoms by Type 1 & 2 SLE Classification groups were examined by ANOVA and differences in proportions were tested by Fisher’s exact test. All analyses were conducted in SAS 9.4 (Copyright © 2016 by SAS Institute Inc., Cary, NC, USA.) and Mplus 8.2 (Muthén, L.K. and Muthén, B.O. (1998–2017). Mplus User’s Guide. Eighth Edition. Los Angeles, CA: Muthén & Muthén).

Results

The study included 208 patients with SLE. Most were female (93%) and Black (59%), with an average age of 43.6 years (range: 21–78) (Table 1). Almost half had a history of lupus nephritis; 12% had active nephritis at the study visit. Based on the study definition for Type 1 & 2 SLE Classifications by clinicians, 41% were in the Minimal group at the study visit, 20% in the Type 1 group, 16% in the Type 2 group, and 22% in the Mixed group. Patients in the Type 2 group were older (average age 50 years compared to 45 years in the Minimal group, 38 years in the Type 1 group, and 41.5 years in the Mixed group), but otherwise the groups had similar demographics.

Table 1.

Cohort characteristics.

	Overall	Minimal SLE	Type 1 SLE	Type 2 SLE	Mixed SLE	p-value
	n=208	n=86	n=42	n=34	n=46
Age, mean (SD)	43.6 (14.0)	44.7 (14.6)	38.1 (12.1)	50.4 (13.3)	41.5 (13.0)	0.0009
Duration of SLE, mean (SD)	13.8 (8.8)	13.1 (9.1)	12.9 (9.3)	16.9 (9.5)	13.8 (6.6)	0.2
Female, n (%)	193 (93%)	77 (90%)	38 (90%)	34 (100%)	44 (96%)	0.2
Race, n (%)						0.3
Asian	3 (1%)	1 (1%)	0 (0%)	1 (3%)	1 (2%)
Black	122 (59%)	49 (57%)	31 (74%)	17 (50%)	25 (54%)
Other	12 (6%)	4 (5%)	3 (7%)	1 (3%)	4 (9%)
White	71 (34%)	32 (37%)	8 (19%)	15 (44%)	16 (35%)
Hispanic (n=206), n (%)	10 (5%)	5 (6%)	3 (7%)	0 (0%)	2 (4%)	0.5
History of lupus nephritis, n (%)	87 (42%)	44 (51%)	21 (50%)	7 (21%)	15 (33%)	0.006
Clinical Features
Active nephritis (n=200), n (%)	23 (12%)	0 (0%)	13 (33%)	0 (0%)	10 (24%)	<0.0001
SLEDAI (n=191), mean (SD)	3.1 (3.4)	1.3 (1.5)	6.2 (3.4)	0.9 (1.3)	5.5 (3.4)	<0.0001
Type 1 PGA (n=207), mean (SD)	0.6 (0.6)	0.1 (0.2)	1.1 (0.5)	0.3 (0.2)	1.2 (0.4)	<0.0001
Type 2 PGA, mean (SD)	0.7 (0.8)	0.2 (0.2)	0.3 (0.2)	1.6 (0.5)	1.5 (0.5)	<0.0001
Medications, n (%)
Prednisone	99 (48%)	36 (42%)	29 (69%)	9 (26%)	25 (54%)	0.001
Hydroxychloroquine	184 (88%)	76 (88%)	37 (88%)	27 (79%)	44 (96%)	0.2
Mycophenolate	63 (30%)	26 (30%)	18 (43%)	7 (21%)	12 (26%)	0.2
Azathioprine	40 (19%)	20 (23%)	8 (19%)	4 (12%)	8 (17%)	0.6
Methotrexate	36 (17%)	11 (13%)	9 (21%)	5 (15%)	11 (24%)	0.3
Belimumab	16 (8%)	3 (3%)	4 (10%)	2 (6%)	7 (15%)	0.1

Factor Analysis.

The Scree plot provided evidence of 4 to 6 relevant factors. Based on the clinical review of factor loadings for each of these solutions, the 4-factor solution was the most interpretable (Suppl Table 1). Factor 1 consisted of primarily Type 2 symptoms, including muscle weakness and pain, fatigue, cognitive dysfunction, and widespread pain. Factor 2 consisted of physical symptoms, including patient-reported rashes, seizures, and swollen glands. Factor 3 was primarily active Type 1 symptoms with SLEDAI components of pericarditis, vasculitis, pyuria, hematuria, mucosal ulcers, pleurisy, and proteinuria. Factor 4 included laboratory-based indicators of inflammation, including thrombocytopenia, low C3/C4, anti-dsDNA+, and active lupus nephritis.

Data Reduction.

We first removed SLEDAI components, as well as Type 1 PGA and Type 2 PGA, as potential variables since they are part of the definition for Type 1 & 2 SLE activity. Looking across factors, we removed and combined PRO measures based on a constellation of attributes:

1. The ability of the indicator to differentiate across factors (loading >0.3 on one factor and <0.3 on other factors). Good differentiators included patient-reported flare in the past 3 months, muscle weakness, muscle pain, swelling in joints, and pain or stiffness in joints.

2. The clinical utility of the variable, including specificity for SLE or frequency, with some symptoms too rare to be worth asking all patients about their occurrence. Non-specific symptoms removed during this step included anxiety, unusual headache, weight loss, other skin rash, swollen glands, dry eyes or dry mouth, Raynaud’s, dark blue or purple spots you could feel on your skin, persistent numbness/tingling in arms or legs, hypertension, edema, pain or burning with urination, stomach or belly pain, and pain or cramps in lower abdomen. Conditions that were dropped due to their rarity included stroke (self-reported in 4 patients) and seizures (self-reported in 2 patients). Two variables for active lupus nephritis, elevated urine protein and foamy urine, were excluded because the standard for this outcome is lab-based and not patient-reported.

3. The strong correlation between variables indicating an overlap of concepts (r > 0.7). Some variables with strong correlations were combined, such as muscle weakness and muscle pain; swelling in joints and pain or stiffness in joints; shortness of breath and chest pain with a deep breath; left upper, left lower, right upper, and right lower arm pain; left upper, left lower, right upper, and right lower leg pain; upper and lower back pain. The variables “feeling depressed” from the SLAQ, “depression” from the PSD, and the PHQ-2 were correlated, and the PHQ-2 was retained due to being a validated measure of major depressive disorder.

A total of 13 variables were included for data analysis (Table 2).

Table 2.

Variables included in the analysis following data reduction.

Instrument	Variable	Definition
SLAQ	Lupus flare in the past 3 months	None, Mild, Moderate or Severe
	Muscle weakness or muscle pain	Highest value of the two variables: None, Mild, Moderate or Severe
	Rash on cheeks (shaped like a butterfly), rash or feeling sick after going out in the sun, bald patches on scalp or clumps of hair on pillow, Sores in mouth or nose	Highest value of the four variables: None, Mild, Moderate or Severe
	Fatigue	None, Mild, Moderate or Severe
	Fever (>101° F, 38.5° C) taken by thermometer	None, Mild, Moderate or Severe
	Shortness of breath and chest pain	None, Mild, Moderate or Severe
	Forgetfulness	None, Mild, Moderate or Severe
PHQ-2	Depression	PHQ-2 score of ≥3 indicated depression
Polysymptomatic Distress Scale: Widespread Pain Index	Leg pain	Pain in the left upper, left lower, right upper, or right lower leg over the past week
	Arm pain	Pain in the left upper, left lower, right upper, or right lower arm over the past week
	Back pain	Pain in the upper or lower back over the past week
	Widespread Pain Index	Total number areas of pain (0–19) over the past week

Patient-reported symptoms across Type 1 & 2 SLE Groups.

Patient-reported symptoms on the SLAQ, PSD, and PHQ-2 were significantly different across Type 1 & 2 SLE groups (p<0.001 for all symptoms, Table 3). Patients in the Minimal SLE group were largely asymptomatic, with almost half of patients in the Minimal SLE group reporting that symptoms on the SLAQ that were either not present or mild. Patients in the Type 1 SLE group reported on average mild symptoms, with mild muscle weakness or pain, swelling or stiffness in joints, rash or other mucocutaneous features, fatigue, and forgetfulness. Patients in the Type 2 and Mixed SLE groups had similar patterns of moderate to severe symptoms of muscle weakness or pain, swelling or stiffness in joints, and fatigue. These patients had on average six to seven areas of widespread pain, with the majority reporting leg or back pain. One-third of patients in the Mixed SLE group met the PHQ-2 criteria for depression.

Table 3.

Patient-reported symptoms by clinician-based classification of Type 1 & 2 SLE activity.

	Overall	Minimal	Type 1	Type 2	Mixed
	n=208	n=86	n=42	n=34	n=46
	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	p-value
SLAQ (0–3)a
Flare	1.0 (1.0)	0.4 (0.7)	1.2 (0.8)	1.0 (1.0)	1.7 (1.0)	<0.0001
Muscle weakness or pain	1.2 (1.0)	0.7 (0.8)	1.0 (0.9)	1.6 (1.0)	1.8 (1.0)	<0.0001
Swelling in joints or stiffness	1.4 (1.0)	0.9 (0.8)	1.3 (0.7)	1.8 (0.9)	2.1 (0.8)	<0.0001
Rash on cheeks, rash after going out in sun, alopecia, ulcers	1.0 (0.9)	0.5 (0.7)	0.9 (0.8)	1.3 (0.9)	1.6 (1.0)	<0.0001
Fatigue	1.5 (1.0)	1.0 (0.8)	1.3 (0.8)	2.1 (0.8)	2.3 (1.0)	<0.0001
Fever	0.2 (0.6)	0.1 (0.3)	0.2 (0.5)	0.2 (0.4)	0.6 (1.0)	<0.0001
Shortness of breath or chest pain	0.7 (0.9)	0.5 (0.7)	0.7 (0.7)	0.9 (0.9)	1.1 (1.1)	0.0006
Forgetfulness	0.8 (0.9)	0.4 (0.7)	0.7 (0.8)	1.3 (0.8)	1.4 (0.9)	<0.0001
Polysymptomatic Distress Scale
Widespread Pain Index (0–19)	3.7 (3.8)	1.9 (2.0)	2.6 (2.5)	5.6 (4.0)	6.8 (4.6)	<0.0001
Leg pain, n (%)	81 (39%)	18 (21%)	16 (38%)	20 (59%)	27 (59%)	<0.0001
Arm pain, n (%)	59 (28%)	12 (14%)	13 (31%)	14 (41%)	20 (43%)	0.0005
Back pain, n (%)	97 (47%)	28 (33%)	13 (31%)	21 (62%)	35 (76%)	<0.0001
PHQ-2
Depression, n (%)	35 (17%)	5 (6%)	4 (10%)	8 (24%)	18 (39%)	<0.0001

^a0 = none, 1 = mild, 2 = moderate, 3 = severe

Multinomial logistic regression.

Results from the multinomial logistic regression model reflected the descriptive results with significant differences in patient-reported symptoms across groups (Suppl Table 2). Based on predicted Type 1 & 2 SLE Classifications from the PRO variables in the model, 53% were in the Minimal SLE group at the study visit, 14% in the Type 1 group, 12% in the Type 2 group, and 21% in the Mixed group. The model accurately predicted the clinician-based Type 1 & 2 SLE classification in 63% of patients, with a chance hit rate of 29% and Huberty’s I index of 0.48, indicating a large effect of the model to accurately classify patients over chance.

Overall, 73% of patients had their Type 1 SLE activity accurately predicted and 83% had their Type 2 SLE activity accurately predicted. Performance of the model varied by group. Among patients who were assigned to the Minimal SLE group, 87% were correctly predicted to be in the Minimal SLE group, and 57% of patients assigned to the Mixed SLE group were correctly predicted to be in the Mixed SLE group. However, slightly more than one-third of patients in the Type 1 and Type 2 groups were correctly predicted to be in their respective groups (Figure 1). Around one-third of patients in the Type 2 group were inaccurately predicted to have active Type 1 symptoms. Overall, among patients who had active lupus nephritis (n=23), only 48% were predicted to have active Type 1 SLE activity.

Figure 1.

Distribution of multinomial logistic regression model predicted Type 1 & 2 SLE classification for each clinician-based classification of Type 1 & 2 SLE activity.

Almost half of the patients in the Type 1 group were misclassified by the model as being in the Minimal group, indicating low Type 1 activity. Among these 18 misclassified patients, the average SLEDAI score was 5.7 (SD: 2.8), although 28% had a clinical SLEDAI of 0 (range: 0–4). One-third of misclassified patients had active lupus nephritis without significant physical symptoms reported on the SLAQ or PSD. Additionally, about one-third had inflammatory arthritis scored on the SLEDAI, but only reported mild to moderate joint pain and swelling.

Discussion

This study builds upon our prior work and provides new insights into the use of PROs to categorize symptomatology in SLE, recognizing that SLE is clinically heterogeneous and that the etiology of symptoms is diverse. Our findings indicate that utilization of a subset of patient-reported indicators from the SLAQ, PSD and PHQ-2, can accurately predict the Type 1 & 2 SLE Classification for 63% of patients. As our data indicated, Type 2 SLE activity is more easily identified by patient-reported data and can be accurately predicted in 83% of patients by PROs. The use of PROs was not as accurate at predicting Type 1 activity, however, and was only able to correctly identify half of patients in the Type 1 SLE group. These findings highlight the challenges of using PROs to categorize and classify symptoms in the Type 1 & 2 SLE Model since some manifestations of Type 1 activity (e.g., nephritis) may be essentially clinically silent while other Type 1 manifestations may cause severe symptoms.

This study confirms the experience of many rheumatologists in practice: severe patient-reported symptoms do not invariably indicate active inflammation and, furthermore, often do no correlate with physician indices of disease activity.^{17, 33, 34} These findings are consistent with our previous studies which showed that patients, who according to their rheumatologist had Type 2 SLE activity without Type 1 activity, frequently reported a recent disease flare as well as moderate-severe symptoms of dry eyes, oral/nasal ulcers, shortness of breath, and stiff and swollen joints.¹⁷ Previous studies have shown discordance in how patients and their rheumatologists score disease activity, with many patients reporting higher levels of activity than their rheumatologists report.^35–39 Our findings are consistent with that experience.

The discordance in the reporting of disease activity by patients and physicians likely relates to the differences in their respective definition and assessment of active disease. Patients often consider disease activity in terms of their psychological and physical well-being, defining flares as extreme fatigue, aching joints, muscle weakness or pain, and forgetfulness.^{10, 40} In contrast, rheumatologists may base their assessment of disease activity on clinical and laboratory signs of inflammation that can be clearly attributed to SLE.⁴⁰ In our clinic, patient and physician assessments of disease activity are positively correlated in patients without Type 2 symptoms, but not in patients with Type 2 symptoms. This finding suggests that symptoms of chronic pain, fatigue, and cognitive dysfunction may be drivers of the discordance.¹⁷

In the current study, patients in the Type 2 and Mixed SLE groups reported similar symptomatology in terms of its severity, making it difficult to use patient-reported measures to distinguish between the two groups. It appears that Type 2 and Mixed SLE activity must be differentiated through laboratory or physician-reported abnormalities in addition to PROs, which reliably identify Type 2 SLE activity. Of note, many patients with Type 1 SLE activity, primarily lupus nephritis, do not experience physical symptoms, thereby preventing distinction from Minimal SLE patients through a PRO measure.

Our study has limitations. These limitations include its cross-sectional nature and its inclusion of only English-speaking patients from a single academic center in the Southeastern United States. Additionally, the clinician-based classification of Type 1 & 2 SLE activity has not yet been validated and is based on definitions used in our Lupus Clinic to make decisions on therapy or management. The PROs are currently used in the Duke Lupus Clinic for clinical care, with the Type 2 PGA based on symptoms reported by patients and scaled accordingly by the treating rheumatologist. Rheumatologists are not blinded to the PROs, which may influence the correlation between the PROs and Type 2 PGA. Also, the number of patients in the Type 2 and Mixed SLE groups was relatively small, which may have limited our ability to accurately predict groups with the available instruments.

In summary, this study found that, while PRO measures can be used to capture symptoms experienced by patients that reflect Type 2 SLE activity, the assessment of active Type 1 SLE, particularly active lupus nephritis, remains challenging using PROs alone. Additionally, the PRO measures were unable to adequately distinguish between patients with only Type 2 SLE activity and patients with Mixed SLE activity. While PROs remain a valuable adjunct to identify the symptoms that most impact a patient’s quality of life, the PROs in the SLAQ and PSD were unable to distinguish between Type 2 symptoms alone compared to Type 2 symptoms associated with Type 1 symptoms, limiting their use in directing immunosuppression-focused treatments.