Evaluation of a rheumatology patient prioritization triage system

Abstract

Purpose

Patient triage systems have been used to prioritize referred patients to facilitate timely treatment of acutely ill patients, but there is limited data to support the effectiveness of these systems as implemented in the clinic setting. Therefore, the purpose of this study was to evaluate the accuracy of a specialty provider triage system.

Data sources

A prospective study design was conducted (N = 103) to compare the pre- and postappointment provider-assigned, prioritization system acuity scores. The intraclass correlation coefficient (ICC), paired t-test, and the Bland–Altman plotting method were used to summarize and analyze the data.

Conclusions

The ICC between the pre- and postappointment acuity scores was 0.50 (p < .001) with no significant difference between the average means (t = −1.17; p = .24). The Bland–Altman plot suggests scores were typically within the limits of agreement. Our findings suggest the specialty provider triage system was effective at accurately classifying rheumatologic patient acuity in this sample.

Implications for practice

When resources are limited and delayed evaluations and treatments result in negative health outcomes, the use of triage systems is likely an effective strategy to reduce the impact of limited provider availability relative to patient census.

Introduction

Rheumatology is a medical specialty focusing on the treatment of a vast number of pain, autoimmune, degenerative, and inflammatory conditions (e.g., rheumatoid arthritis, osteoarthritis, fibromyalgia, systemic lupus erythematosus). Many of these rheumatologic conditions are often classified as arthritic. In the United States, there were 4946 rheumatologists charged with treating the population of persons with rheumatic conditions, which is estimated to be over 48 million (Centers for Disease Control and Prevention, 2009; Helmick et al., 2008; Lawrence et al., 2008). With an expected increase in the prevalence of arthritic conditions, a shortage of 2609 rheumatologists by 2025 is projected (Deal et al., 2007; Kirwan, 1997).

This shortage of rheumatologists relative to the number of persons seeking care results in significant wait times for rheumatology clinic appointments. An estimated 1–17 months elapse between a rheumatology patient’s first symptoms to his or her rheumatology appointment (Hernández-García et al., 2000; Kumar et al., 2007). Timely diagnosis and treatment of rheumatologic conditions is imperative for optimal treatment and delays in care result in worsened health outcomes including permanent damage to organ systems increasing morbidity, mortality, and rates of disability (Oliver & St. Clair, 2008; Yelin, Such, Criswell, & Epstein, 1998).

To address the disproportionate need for rheumatology care relative to the supply of rheumatology healthcare providers, many rheumatology practices use patient-triage systems. These systems identify referred, acutely ill (urgent need for an appointment) rheumatology patients, and prioritize their care in order to provide time-appropriate treatment (Sathi, Whitehead, & Grennan, 2003).

A number of specific triage systems are used by rheumatology and other specialty practices where the need for services outweighs the number of available providers. Triage systems can be grouped into those triaged by a specialty provider (Graydon & Thompson, 2008; Harrington & Walsh, 2001; Sathi et al., 2003) or by the referring provider (Slade et al., 2008). Typically those systems in which new referrals are triaged by a specialty provider involve review of referral information by a provider trained in rheumatology who then assigns a score based on the patient’s rheumatologic acuity (Graydon & Thompson, 2008; Harrington & Walsh, 2001; Sathi et al., 2003). Appointments are scheduled based on the acuity score considering appointment availability (Harrington & Walsh, 2001; Slade et al., 2008). These systems also identify patients who were inappropriately referred to rheumatology or the outpatient setting and redirect them to appropriate services (Harrington & Walsh, 2001). Systems in which referrals were triaged by the referring provider for outpatient specialty services involve referring providers assigning patients a score based on urgency for outpatient services (Mariotti, Meggio, de Pretis, & Gentilini, 2008). The patients are then given appointment times with maximum waiting intervals based on the score (Mariotti et al., 2008). Both systems aim to identify acutely ill patients and reduce wait times for those patients that need to be seen urgently.

There is conflicting evidence on the accuracy of patient triage systems at correctly classifying acutely ill patients. Evaluation of prioritization systems in which patients were triaged by a rheumatology provider suggests good agreement between pre- and postconsultation score (κ= .71); yet, the efficacy of these triage systems is hinged on accurate information from the referring provider. Evaluations of referring provider systems have suggested poor to moderate agreement (κ= .44–.47, p = .001) between the referring and consulting provider (Mariotti et al., 2008; Slade et al., 2008).

In summary, the inequity between the high number of persons with rheumatic conditions and the few specialty providers available to provide this population with specialty care has resulted in the use of patient prioritization triage systems, which aid in identifying and providing timely treatment to high-acuity rheumatologic patients. The dearth of and conflicting research findings in regards to these systems has resulted in a lack of a standardized, evidence-based triage system; therefore, further research is needed to improve quality in patient care. Therefore, we conducted a prospective study to compare the pre- and postappointment provider-assigned, triage-system acuity scores assigned using a patient prioritization triage system at a rheumatology clinic in a large university medical center.

Description of the current triage system

The established patient prioritization triage system evaluated had been in place for several years at the practice and consisted of a board-certified rheumatologist reviewing documents sent by a referring provider and then assigning an acuity score, thereby determining rheumatology appointment immediacy. Acuity score levels ranged from 1 to 4, with lower scores indicating more urgency: (1) urgent (e.g., systemic vasculitis); (2) emergency (e.g., active inflammatory arthritis); (3) next available (e.g., stable inflammatory arthritis); and (4) lowest priority (occupational musculoskeletal condition). Additional health information was often requested from the referring provider (i.e., laboratory results, office notes) and then re-reviewed. The acuity score was entered into a secured, Health Insurance Portability and Accountability Act compliant World Wide Web-based database (Sharepoint). This Web-based database contained key information provided by referring providers (including date of referral, laboratory results, key assessment findings, and pertinent medical history). The database also contained information entered by the rheumatology department, including acuity score, presumptive diagnosis, and date of appointment. Appointments were made by a scheduling center based on the preappointment acuity score.

Methods

Approval was granted from the university Medical Institutional Review Board. A written informed consent was given to all of the providers explaining the study and to request their participation. A power analysis was carried out prior to data collection. With a sample size of at least 100 and an alpha level of .05, the power of the paired t-test to detect a significant difference between the two assessments was estimated to be at least 84%, assuming the ratio of the average difference in score relative to the SD would be at least 0.3. This was a relatively conservative assumption to make, because in this context a small effect size would be a ratio of 0.2, while a medium effect size would be 0.5 (Cohen, 1988).

Sample and setting

All persons newly referred to the university rheumatology clinic from an outside facility aged 18 or above between May 2014 and September 2014 who were provided a preappointment acuity score were eligible for inclusion in the study. The sample comprised 103 adults who met inclusion criteria.

Protocol

Preappointment acuity scores were obtained from a Sharepoint system used by the clinic prior to the initiation of the study. To evaluate the accuracy of the score, providers were asked to score patient acuity using the triage scoring system following the initial consultation. All information was deidentified by clinic staff who assigned patients a participant number prior to the patient being seen for consultation and entered data into an Excel spreadsheet that was provided to the research team. This information included date of referral, age, gender, county of residence, scheduled rheumatology provider, appointment date, and presumptive diagnosis. Data collection sheets (with the deidentified patient number assigned by clinic staff) were given to providers after their initial consultation with patients; these sheets asked the provider to assign the patient a postappointment acuity score and document a presumptive diagnosis based on the information gathered during the consultation. This deidentified information was then entered in the Excel spreadsheet with the corresponding preappointment acuity scores and demographic information collected prior to the appointment by clinic staff.

Analysis

Descriptive statistics, including means and SDs or frequency distributions, were used to summarize demographic and clinical characteristics of study participants. To evaluate the efficacy of the triage system, the Bland–Altman method was implemented and included an intraclass correlation coefficient (ICC), a paired t-test, and Bland–Altman plot (Altman & Bland, 1983; Martin Bland & Altman, 1986). ICC was used as the test of association to determine the correlation between pre- and postacuity scores. The paired t-test was used to determine whether the average difference between the pre- and the postappointment acuity score was significantly different from zero (Rayens, Svavarsdottir, & Burkart, 2010). The Bland–Altman plot was used as a graphical way of depicting agreement between the acuity scores by assessing differences between the “gold-standard” or most accurate rating (postappointment acuity score) and triage system (preappointment acuity score; Rayens et al., 2010). Data analysis was performed using Statistical Package for the Social Sciences (SPSS) 21 software; an alpha level of .05 was used to determine statistical significance.

Results

Summaries of patient demographic and clinical characteristics are displayed in Table 1. The mean age of participants was 46.1 years (SD = 13.3; range = 18–85). Consistent with rheumatologic disease in the United States, the majority of the sample was female (81%) (Klippel, Stone, & White, 2008). Patients waited an average of 36.5 days (SD = 15.8) from the time of outside provider referral until their initial clinic consultation. The predominant working rheumatology diagnoses included osteoarthritis (34.9%), fibromyalgia syndrome (24.3%), rheumatoid arthritis (16.5%), and systemic lupus erythematosus (3.8%).

Table 1.

Demographic and clinical characteristics of study participants (N = 103)

	Mean (SD); range or n (%)
Age (years)	46.1 (13.3); 18–85
Gender
Male	20 (19.4%)
Female	83 (80.6%)
Time from referral to appointment (days)	36.5 (15.8); 3–105
Predominant working diagnoses
Osteoarthritis	36 (34.9%)
Fibromyalgia syndrome	25 (24.3%)
Rheumatoid arthritis	17 (16.5%)
Systemic lupus erythematosus	4 (3.8%)
Preappointment acuity score	3.3 (0.7); 2–4
Postappointment acuity score	3.4 (0.9); 0–4

The ICC for the acuity scores was 0.50 (p < .001), which suggests moderate agreement according to the guidelines established by Landis and Koch (1977). There was not a significant difference between the pre- and postappointment acuity scores according to the paired t-test (t = −1.17; p = .24). The Bland–Altman plot (Figure 1) suggests that the pre- and postappointment triage scores were within the limits of agreement; only 8 of the 103 differences were outside of the limit of bias (±2 SD). Three patients were given a higher acuity score and five patients were given a lower acuity score following their clinical evaluation (relative to their triage acuity score prior to being seen in the clinic). evaluation (relative to their triage acuity score prior to being seen in the clinic).

Figure 1.

Bland–Altman plot.

Discussion

The specialty provider prioritization triage system evaluated in this study was effective in identifying acutely ill rheumatologic patients as evidenced by the moderate agreement found between the pre- and postappointment acuity scores (ICC = 0.50). Paired t-test results found no significant difference between the means, and only 7.8% of the scores fell outside of limits of agreement as shown on the Bland–Altman plot. This is consistent with other research aimed at evaluating the accuracy of similar triage systems (Sathi et al., 2003). Sathi et al. (2003) found that good agreement (κ= .71) existed between scores based on the referring information and those after consultation. Importantly, most scores that were different between the pre- and postassessments were because of a lower priority assigned at postassessment, suggesting that acutely ill persons received timely care.

Despite the triage systems being used and inappropriate referrals being redirected to appropriate services, we found that patients still waited on average 37 days before being seen by a rheumatology healthcare provider. This is important considering that up to 41% of patient requests for consultation with a rheumatology provider are inappropriate; thus, wait times would have likely been significantly longer had this system not been in place (Harrington & Walsh, 2001). Because accuracy in referring information is imperative in order to receive an appropriate acuity score (i.e., the lack of a presumptive diagnosis, symptom duration, and involved joints can lead to acutely ill patients receiving lower acuity scores), future studies should aim to develop quality improvement strategies to improve the accuracy and efficiency in which health information is provided from referring providers to subspecialties (Graydon & Thompson, 2008).

The time required to implement the prioritization triage systems is an additional limitation to widespread use. Future studies should be directed at making system improvements to advance feasibility by streamlining the process of assessing and recording this information.

When designing this study, we felt it was important to evaluate the prioritization triage system as it was implemented in the rheumatology clinic setting. However, some limitations were present because of this approach to conducting the study. In this clinical setting, the pre-and postappointment triage acuity scores were often assigned by different providers. This may have contributed to variability despite the use of a standardized scoring system. Additionally, even though well-established criteria existed for determining scores, there is some variability based on a provider’s knowledge and experience and thus scores could be affected by the provider assigning the acuity score. Because of the time lag in appointment wait times, it is possible that patient acuity changed over the time that elapsed between the initial acuity score to when the patient was seen in the rheumatology clinic. For this reason, the comparison of pre- and postassessment acuity is conservative because some discrepancy may be as a result of patient status changes rather than errors in acuity measurement.

Conclusion

This study indicates that there is agreement between the pre- and postappointment acuity scores. This supports the effectiveness of provider-assigned patient prioritization triage systems at accurately identifying acutely ill patients referred to a rheumatology specialty care clinic. Because timely access to care will continue to be a challenge for providers as the population ages and imbalances between supply and demand worsen, improving referral processes will be imperative to facilitate quality patient care outcomes. Our findings suggest that prioritization triage systems are one mechanism that can positively affect these quality outcomes.