- Journal List
- HHS Author Manuscripts
- PMC6642029
As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice
Br J Haematol. Author manuscript; available in PMC 2020 Aug 1.
Published in final edited form as:
Br J Haematol. 2019 Aug; 186(3): 490–498.
Published online 2019 May 26. doi:10.1111/bjh.15932
PMCID: PMC6642029
NIHMSID: NIHMS1023391
PMID: 31131442
Vivek A. Upadhyay,1,7 Benjamin P. Geisler,1,7 Lova Sun,1,7 Lynne Uhl,2,7 Richard M. Kaufman,3,7 Christopher Stowell,4,7 Robert S. Makar,4,7,* and Pavan K. Bendapudi4,5,6,7,*
Author information Copyright and License information PMC Disclaimer
The publisher's final edited version of this article is available free at Br J Haematol
Associated Data
- Supplementary Materials
Abstract
The PLASMIC score is a recently described clinical scoring algorithm that rapidly assesses the probability of severe ADAMTS13 (a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13) deficiency among patients presenting with thrombotic microangiopathy. Using a large multi-institutional cohort, we explored whether an approach utilizing the PLASMIC score to risk-stratify patients with suspected immune thrombotic thrombocytopenic purpura (iTTP) could lead to significant cost savings. Our consortium consists of institutions with an unrestricted approach to ADAMTS13 testing (Group A) and those that require pre-approval by the transfusion medicine service (Group B). Institutions in Group A tested more patients than those in Group B (P <0.001) but did not identify more cases of iTTP (P=0.29) or have lower iTTP-related mortality (P=0.84). Decision tree cost analysis showed that applying a PLASMIC score-based strategy to screen patients for ADAMTS13 testing in Group A would have reduced costs by approximately 27% over the 12-year period of our study compared to the current approach. Savings were primarily driven by a reduction in unnecessary therapeutic plasma exchanges, but lower utilization of ADAMTS13 testing and subspecialty consultations also contributed. Our data indicate that using the PLASMIC score to guide ADAMTS13 testing and the management of patients with suspected iTTP could be associated with significant cost savings.
Keywords: Thrombotic thrombocytopenic purpura, cost savings/effectiveness, TTP, ADAMTS13, PLASMIC
Introduction
Immune thrombotic thrombocytopenic purpura (iTTP) is a rare but lethal subtype of thrombotic microangiopathy (TMA) in which patients experience a severe acquired deficiency in the von Willebrand factor cleaving protease, ADAMTS13 (a disintegrin and metalloproteinase with a thrombospondin type 1 motif, member 13). Therapeutic plasma exchange (TPE) has been demonstrated to dramatically reduce iTTP-related mortality (Rock, et al 1991) but it is costly and associated with a significant risk of complications, particularly severe allergic reactions and hypocalcaemia (Hovinga and Lämmle 2006). At most centres, the ADAMTS13 test is not performed in-house, and results are not available for real-time clinical decision-making (Franchini and Mannucci 2008, Sadler 2008). In response to this need, the PLASMIC scoring algorithm (Table I), was developed and validated as a clinical assessment tool to reliably determine the pre-test probability of severe ADAMTS13 deficiency in patients presenting with thrombocytopenia and TMA (Bendapudi, et al 2017a, Bendapudi, et al 2017b, Jajosky, et al 2017, Li, et al 2018). The score utilizes presenting laboratory values and components of the clinical history to stratify TMA patients into those at low (score 0–4), intermediate (score 5) and high risk (score 6–7) for severe ADAMTS13 deficiency in order to identify the subgroup of individuals most likely to have iTTP and benefit from TPE.
Table I.
The PLASMIC Score.
| Component | Score |
|---|---|
| Platelet count < 30 × 109/l | 1 |
| Combined lysis parameter (any one of the following: reticulocyte count > 2.5%; indirect bilirubin > 34.2 μmol/l; absent haptoglobin) | 1 |
| International Normalised Ratio < 1.5 | 1 |
| Creatinine < 176.8 μmol/l | 1 |
| Mean corpuscular volume < 90 fl | 1 |
| No active cancer or cancer therapy within last year | 1 |
| No history of bone marrow or solid organ transplantation | 1 |
| Total Possible Score | 7 |
Open in a separate window
Score 0–4: Low risk of severe ADAMTS13 deficiency (0–4%)
Score 5: Intermediate risk of severe ADAMTS13 deficiency (5–25%)
Score 6–7: High risk of severe ADAMTS13 deficiency (62–82%)
The development of standardized clinical approaches to TMA that maximize efficacy while minimizing the costs and complications associated with unnecessary TPE represents an unmet need in our field. Practice patterns vary widely, with most centres relying solely on clinician judgment to determine which patients receive ADAMTS13 testing and undergo TPE. We hypothesized that a diagnostic and management algorithm based on the PLASMIC score would minimize unnecessary testing and TPE procedures and yield significant cost savings by facilitating the early identification of patients with iTTP. Using this approach (Bendapudi, et al 2017b), management of patients with TMA would be stratified according to the PLASMIC score. To study this concept, we utilized data from the Harvard TMA Research Collaborative registry to evaluate the impact of existing resource management strategies within our consortium and to compare a PLASMIC score-based algorithm to a standard of care, in which ADAMTS13 testing and TPE are driven primarily by clinician assessment.
Methods
Identification of Suspected iTTP Patients in the Harvard TMA Research Collaborative Registry
Our registry consists of patients seen at Beth Israel Deaconess Medical Center, Brigham and Women’s Hospital and Massachusetts General Hospital, between 8 January 2004 and 5 December 2015. Details of the registry have been described previously (Bendapudi, et al 2015). Briefly, ADAMTS13 test results were recorded, along with 34 demographic, clinical and laboratory parameters for each consecutive adult patient who presented with TMA and suspected iTTP during the study period and had ADAMTS13 testing performed. Admissions data for all three institutions were obtained via the Massachusetts Center for Health Information and Analysis (Center for Health Information and Analysis 2018). This work has been approved by the institutional review boards of all participating institutions.
Clinical Diagnostic Categories
Patients were identified as having iTTP or assigned one of 10 other clinical diagnoses based on predefined criteria and without regard to ADAMTS13 activity level except in situations where it was necessary to distinguish iTTP from haemolytic-uraemic syndrome. During this process, relevant clinical and laboratory data from the patient’s electronic medical record proximate to the date of the ADAMTS13 test were reviewed by at least two of the study authors and the diagnostic category was assigned by consensus.
Analysis of Clinical Management
Among patients receiving TPE, the number of sessions and the number of plasma units used were tallied until cessation of TPE treatment. Hospital length of stay was calculated using the index admission and discharge dates regardless of the date of ADAMTS13 testing. Haematology consultation was determined via the presence of a haematology consultation note in the medical record. Likewise, blood bank involvement was defined as either the presence of a consultation note from the transfusion medicine service and/or TPE documentation via a procedure note.
Patient Eligibility
Our study evaluated ADAMTS13 testing patterns over time as well as diagnoses and resource utilization for those patients suspected of having iTTP. We charted all ADAMTS13 assays sent during the study period (2004–2015) across the consortium. We then eliminated those patients for whom active iTTP was either not under consideration (such as those with outpatient assays), patients with missing data, and all paediatric patients, resulting in an eligible patient population for analysis. These eligible patients were divided by consortium group, A or B, where Group A included those patients from consortium hospitals with unrestricted testing practices and Group B included those patients from consortium hospitals with restrictive testing practices. These two groups were subsequently analysed for testing patterns, costs and diagnostic accuracy. Finally, PLASMIC scores were calculated for those eligible patients who presented with thrombocytopenia (<150 × 109 platelets/l) and microangiopathic haemolytic anaemia (defined as the presence of schistocytes on the peripheral blood smear).
Cost Analysis
Economic evaluations are usually performed in two different ways: either alongside a clinical trial or in model-based extrapolation (Drummond, et al 2015). A decision tree, a common and relatively simple form of model-based extrapolation (Barton, et al 2004), was created for the two strategies using the TreeAge Pro 2017, R1 software package (TreeAge Software, Inc., Williamstown, MA). By accounting for the costs associated with each management decision made during the care of patients with suspected iTTP, our model decision tree (Figures S1A-S1C) calculated, as aggregates and by component, the costs for each of the two testing strategies studied. We compared the unrestricted testing approach used by Group A and a hypothetical PLASMIC score-driven strategy in the same cohort of patients. Due to the lack of information on patients in Group B who did not receive approval to proceed with ADAMTS13 testing, we were only able to perform a decision tree-based analysis for Group A and a hypothetical cohort of patients from Group A who were risk-stratified by the PLASMIC algorithm as described below.
Key input parameters included prevalence of iTTP, distribution of PLASMIC scores, and the cost of resources that would be most directly affected by implementation of a PLASMIC score-driven strategy, namely ADAMTS13 assays, consultations from the haematology and transfusion medicine services, and use of plasma and TPE (Table SI). Costs are based upon allowable Medicare charges as a proxy for true costs ( Kuntz et al 2017). In the unrestricted testing strategy used in Group A, the rate of patients who received a given treatment or service was based on historical percentages within Group A. These percentages were directly measured for haematology and blood bank consultations. TPE rates were calculated by taking the mean numbers of TPE sessions per patient in the cohorts with and without severe ADAMTS13 deficiency. For both consultation and TPE treatments, we assumed these trends would continue without significant deviation. For the PLASMIC-based strategy, all high-risk patients received haematology and blood bank consultations, and for the remaining intermediate and low-risk groups, we calculated the historical rate of consultations according to PLASMIC score and assumed a similar rate of consultation. All high-risk patients would receive TPE at the rate seen historically in this category (83.5%), while intermediate- and low-risk patients would not receive up-front TPE. In addition to the TPE calculations, we also assumed that for intermediate risk patients, empiric plasma therapy (20–30 ml/kg x 4 days) would be given to 30% of patients, based on historical patterns observed in our cohort. Using this approach, we calculated the average costs for these specific resources for each strategy during the study period spanning 2004–2015.
Statistical Methods
To evaluate the number of TPE sessions administered per patient between the two groups, the Shapiro-Wilk test of normality was applied, and the Mann-Whitney U test was used to compare median values. Comparisons of proportions between the study groups or different ADAMTS13 testing strategies were made using the chi square or Fisher’s exact test. We performed statistical calculations in MedCalc version 16.8.4 (MedCalc Software, Ostend, Belgium) and Microsoft Excel version 16.16.5 (Redmond, WA).
Results
Baseline Characteristics of the Study Groups
A total of 647 patients were enrolled in the Harvard TMA Collaborative Registry between 8 January 2004 and 5 December 2015 (Figure 1). Of these, 135 met exclusion criteria and were removed from the analysis. The remaining eligible population (N=512) was then divided into two groups based on testing site. Group A (N = 412) consisted of those patients who were evaluated at institutions with an unrestricted ADAMTS13 testing strategy at which there were no limitations on ADAMTS13 testing by clinicians. Group B (N=100) consisted of patients who were evaluated at an institution with a restrictive ADAMTS13 testing strategy in which requests for ADAMTS13 testing were reviewed and subject to approval by the transfusion medicine service. Most baseline characteristics of the two groups were similar (Table II). However, Groups A and B differed in terms of the proportion of patients who received TPE (114/412 (27.7%) in Group A vs. 66/100 (66.7%) in Group B, P <0.0001) as well as the number of sessions administered (median [interquartile range] of 0 [0–3] in Group A vs. 5 [1–10] in Group B, P <0.0001). Within the Group A and Group B populations, the PLASMIC score was calculated for the subset of patients who presented with thrombocytopenia and evidence of TMA (N = 264 for Group A and N = 86 for Group B).
Figure 1.
Study Design iTTP: immune thrombotic thrombocytopenic purpura; TMA: thrombotic microangiopathy; TPE: therapeutic plasma exchange.
Table II.
Demographic, clinical, and laboratory features of the Group A and B patient cohorts.
| Group A (N=412) | Group B (N=100) | |
|---|---|---|
| Demographic Features | ||
| Age, years | 54 (39–66) N=412 | 51 (38–64) N=100 |
| Sex, female (%) | 233 (56.6) N=412 | 59 (59.0) N=100 |
| Ethnicity, Caucasian (%) | 298 (75.3) N=396 | 72 (75.8) N=95 |
| Clinical Data | ||
| Cancer treatment within 1 year (%) | 111 (26.9) N=412 | 22 (22.0) N=100 |
| Prior transplant (%) | 72 (17.5) N=412 | 12 (12.0) N=100 |
| Fever (%) | 143 (34.9) N=410 | 25 (25.0) N=100 |
| Neurological symptoms (%) | 177 (43.4) N=408 | 37 (37.0) N=100 |
| Received TPE (%) | 114 (27.7) N=412 | 66 (66.0) N=100 |
| Number of TPE Procedures | 0 (0–3) N=410 | 5 (1–10) N=85 |
| 90-day Survival (%) | 231 (64.2) N=360 | 63 (73.3) N=86 |
| Laboratory Data | ||
| Severe ADAMTS13 deficiency (%)* | 54 (13.1) N=412 | 27 (27.0) N=100 |
| Inhibitor present (%)† | 47 (11.4) N=412 | 25 (25.0) N=100 |
| Platelet count (x109/l) | 46 (22–92) N=412 | 33 (20–70) N=100 |
| Haematocrit (%) | 26.9 (23.2–30.7) N=412 | 26.3 (22.9–29.9) N=100 |
| Schistocytes present (%) | 276 (68.1) N=405 | 87 (90.6) N=96 |
| Creatinine (μmol/l) | 159.1 (97.2–300.6) N=412 | 132.6 (97.2–318.2) N=100 |
| Lactate dehydrogenase (u/l) | 668 (375–1175) N=397 | 862 (586–1439) N=99 |
| International normalised ratio | 1.2 (1.1–1.5) N=412 | 1.1 (1.1–1.2) N=100 |
| Bilirubin (μmol/l) | 22.2 (11.9–41.0) N=411 | 32.5 (17.1–51.3) N=100 |
| Blood group O (%) | 196 (48.6) N=403 | 48 (49.0) N=98 |
Open in a separate window
Relevant clinical features and laboratory findings proximate to the time of ADAMTS13 testing are reported. Continuous variables are reported as median (interquartile range) and statistical comparison made using the Mann-Whitney test. Categorical variables are reported as number (%) and statistical comparison is made using the Fisher Exact test.
*Severe deficiency of ADAMTS13 is defined as an activity level ≤10%.
†A positive inhibitor is defined as a titre >0.4 Bethesda Units.
TPE: therapeutic plasma exchange.
Trends in ADAMTS13 Testing and iTTP Diagnoses
We first compared trends in ADAMTS13 testing, iTTP treatment and mortality outcomes between Groups A and B to determine if restricting ADAMTS13 testing could be done without compromising patient care in a real-world setting. The number of ADAMTS13 assays sent in Group A per 100,000 inpatient admissions increased more than nine-fold during the study period, from 8.15 per 100,000 admissions in 2004 to 69 per 100,000 admissions in 2015, P <0.0001 (Figure 2A). During the same time-frame, the number of patients found to have severe ADAMTS13 deficiency in Group A remained steady (4.98 cases per 100,000 admissions in 2004–2009 vs. 3.91 cases per 100,000 admissions between 2010–2015, P=0.72, Figure 2B). Additionally, the number of services ordering ADAMTS13 assays and the number of assays performed on an outpatient basis increased significantly in Group A between 2004 and 2015 (Figure S2). By comparison, Group B did not experience an increase in ADAMTS13 testing (23.42 per 100,000 admissions in 2004 versus 22.91 per 100,000 admissions in 2015, P=0.93, Figure 2A) or in the number of patients determined to have severe ADAMTS13 deficiency (6.31 cases per 100,000 admissions per year from 2004–2009 compared to 5.09 cases per 100,000 admissions per year from 2010–2015, P=0.72, Figure 2B).
Figure 2.
Comparison of ADAMTS13 testing, PLASMIC scores, and iTTP diagnoses for patients with suspected iTTP in Groups A (N=412) and B (N=100). A: ADAMTS13 testing per 100,000 admissions by year. B: number of immune thrombotic thrombocytopenic purpura (iTTP) diagnoses per 100,000 admissions within the two groups. C: PLASMIC scores during the study period for eligible patients with microangiopathic haemolytic anaemia in Group A D: PLASMIC scores during the study period for eligible patients with microangiopathic haemolytic anaemia in Group B. *Incomplete data available for 2010 TMA: thrombotic microangiopathy.
Across the entire study period, Group A institutions sent significantly more ADAMTS13 tests per 100,000 admissions (34.28 vs. 21.30 in Group B, P <0.0001, Table III) despite having a similar number of patients with severe ADAMTS13 deficiency compared to Group B institutions (4.49 v. 5.75 per 100,000 admissions, P=0.29). In terms of iTTP management, a significantly higher proportion of patients in Group B received TPE (14.07 per 100,000 admissions vs. 9.48 per 100,000 admissions in Group A, P=0.01). However, there was no difference in iTTP-related mortality between the two groups (0.17 vs. 0.21 deaths per 100,000 admissions, P=0.84).
Table III.
iTTP-related events in Group A and Group B during the study period, per 100,000 admissions.
| Event (per 100,000 admissions) | Group A (n = 412) | Group B (n = 100) | P* |
|---|---|---|---|
| ADAMTS13 Test Sent | 34.28 | 21.30 | <0.0001 |
| Severe ADAMTS13 Deficiency (≤10) | 4.49 | 5.75 | 0.2924 |
| Patients Receiving TPE | 9.48 | 14.07 | 0.0103 |
| iTTP Mortality (90 Day) | 0.17 | 0.21 | 0.8394 |
Open in a separate window
*Statistical comparison was made using the chi-square test.
iTTP: immune thrombotic thrombocytopenic purpura; TPE: therapeutic plasma exchange.
To better understand the risk profile of patients receiving ADAMTS13 testing in each group, we stratified patients with TMA by PLASMIC score. In Group A, individuals with low risk PLASMIC scores (score 0–4) accounted for the majority of ADAMTS13 testing (146/264, 55% of all assays during study period, Figure 2C). By contrast, Group B had a smaller average proportion of patients with low-risk PLASMIC scores who received ADAMTS13 testing (27/86, 31%, P <0.05 for comparison to Group A)(Figure 2D). Taken together, these data indicate that the increasing number of ADAMTS13 assays sent by Group A institutions was not associated with a corresponding increase in the number of patients diagnosed with iTTP or a difference in iTTP-related mortality and that these tests were largely performed on patients who had a low pre-test probability for severe ADAMTS13 deficiency.
Modelling a PLASMIC Score-Based Approach to iTTP Diagnosis and Management
We sought to explore the potential cost savings associated with the use of a diagnostic and treatment algorithm that incorporates the PLASMIC score into the management of patients with suspected iTTP (Figure 3). Under a PLASMIC-score driven approach, ADAMTS13 activity testing and downstream iTTP-related consultations and treatment would be deferred for patients with a low-risk PLASMIC score (score 0–4) and efforts would be focused instead on identifying alternative causes of TMA. Patients at intermediate risk (score 5) would be considered for plasma infusion (30 ml/kg on day 1 followed by 15 ml/kg on days 2–4) at the discretion of the treating physician while awaiting the results of ADAMTS13 testing, and those at high risk would be considered for empiric TPE until the results of ADAMTS13 testing were known. Compared to the status quo, we found that a PLASMIC score-based algorithm would have significantly decreased the proportion of patients without severe ADAMTS13 deficiency who underwent ADAMTS13 testing (86.9% vs 54.2%, P <0.0001, Table IV) or treatment with TPE (55.3% v. 29.2%, P <0.001) during the study period.
Figure 3.
Proposed PLASMIC score-based algorithm for iTTP workup and management. iTTP: immune thrombotic thrombocytopenic purpura;MAHA: Microangiopathic haemolytic anaemia; Plt: platelet count; TPE: therapeutic plasma exchange.
Table IV.
Effect of a PLASMIC-score based management algorithm on resource utilization in Group A.
| Cases assayed for ADAMTS13 (n) | ADAMTS13 Activity (cases, n) ‡ | Cases Receiving TPE (n) | TPE Treatment by ADAMTS13 Activity (cases, n) ‡‡ | |||
|---|---|---|---|---|---|---|
| ≤10% | >10% | ≤10% | >10% | |||
| Group A* | 412 | 54 | 358 | 114 | 51 | 63 |
| Group A + PLASMIC† | 118 | 54 | 64 | 72 | 51 | 21 |
Open in a separate window
*Consortium data, 2004–2015.
†Projected by decision tree cost analysis for the period between 2004–2015.
‡P <0.0001for Group A vs Group A + PLASMIC by Fisher’s exact test.
‡‡P <0.001for Group A vs Group A + PLASMIC by Fisher’s exact test.
TPE: therapeutic plasma exchange.
We next determined the iTTP-related costs for the unrestricted testing strategy currently used in Group A and compared these to the costs of the proposed PLASMIC-score-driven approach over the entire study period. Cost savings analysis demonstrated a 27% reduction in overall cost with the PLASMIC score-based strategy compared to the unrestricted testing approach currently utilized by Group A institutions ($1,845,348 vs. $2,530,092, Table V). Therefore, our model suggests that a total of $684,744 in iTTP-related costs would have been saved within Group A if a PLASMIC score-based strategy had been in place during the approximately 12-year study period. The reduction in costs was driven primarily by a decrease in TPE treatments, which accounts for 87.7% of total savings (Figure S3). Additionally, the PLASMIC score-based strategy would have created savings by reducing the number of haematology consultations (8.1% of total savings), transfusion medicine consultations (0.5% of total savings) and ADAMTS13 tests (3.7% of total savings). The number of patients with an intermediate-risk PLASMIC score receiving plasma resulted in a small added cost, which reduced total savings by $26,368 (3.7%).
Table V.
Effect of a PLASMIC-score based management algorithm on costs in Group A.*
| Assays | Haematology Consultation | Blood Bank Consultation | Plasma | TPE | Total Costs | Relative Savings | Absolute Savings | |
|---|---|---|---|---|---|---|---|---|
| Group A | 36,668 | 71,688 | 28,016 | - | 2,393,720 | 2,530,092 | - | - |
| Group A + PLASMIC† | 10,712 | 47,380 | 24,308 | 26,368 | 1,736,580 | 1,845,348 | 27% | 684,744 |
Open in a separate window
*Reflects the costs of care for all patients in Group A (N=412) under the two scenarios. All figures are in $US
†Projected by decision tree cost analysis for the period between 2004–2015.
TPE: therapeutic plasma exchange.
Discussion
This study evaluated the hypothesis that an algorithmic approach to the workup of patients with suspected iTTP could lead to significant cost savings. First, we compared data from institutions with (Group A) and without (Group B) an unrestricted ADAMTS13 testing policy and found that the restrictive policy used by Group B led to a significant reduction in patients evaluated and treated for suspected iTTP without affecting the rate of iTTP diagnoses or iTTP-related mortality. Importantly, all institutions in our consortium are large tertiary care academic medical centres with similar patient populations and catchment areas. Second, we compared the use of a diagnostic and management strategy based on the PLASMIC score against current practice at institutions with an unrestricted testing policy (Group A). Taken together, our findings suggest that incorporation of the PLASMIC score into the upfront management of patients with suspected iTTP could lead to significant cost savings by focusing clinical attention and resources on the patients most likely to have severe ADAMTS13 deficiency.
We found a striking increase in the number of patients receiving ADAMTS13 testing between 2004–2015 in a setting where testing is unregulated. Importantly, this observation was driven largely by increased testing in patients with a low pre-test probability of having severe ADAMTS13 deficiency and did not result in a higher number of true iTTP cases identified; rather, an unrestricted testing strategy was associated with increased health care utilization, particularly with regard to TPE and expert consultations. By contrast, the approach taken in Group B, which involves manually screening requests for ADAMTS13 testing and deferring testing in patients felt to be at low risk for iTTP, appeared to effectively control iTTP-related resource utilization while capturing a similar number of true iTTP cases compared to Group A. The higher rate of TPE observed in Group B probably partly reflects the fact that a higher proportion of Group B patients receiving ADAMTS13 testing were categorised as intermediate- or high-risk by the PLASMIC score. Given that the number of ADAMTS13 assays performed Group B reflect the implementation of a pre-approval process, we were unable to establish equivalent starting populations for decision tree analysis and therefore did not explore the implications on costs or outcomes associated with a PLASMIC score-based approach within Group B.
Our study expands on recent work by Kim et al (2017), examining the cost-effectiveness of utilizing the PLASMIC score in combination with in-house or send-out ADAMTS13 activity testing to manage cases of suspected iTTP. Consistent with those results and independent of whether in-house ADAMTS13 testing is preferable to send-out assays, we found that including the PLASMIC score in clinical decision making has the potential to achieve cost savings over the current unregulated approach in use at many institutions. The PLASMIC score-based approach achieves savings primarily by eliminating costs associated with the diagnosis and management of low-risk patients. Furthermore, the sustained trend towards increased testing suggests that our model may underestimate future savings that could be possible with a PLASMIC score-based approach. We also did not study other potential cost savings stemming from the accurate triage of patients with suspected iTTP, such as expenses associated with patient transfers between hospitals for administration of TPE or savings stemming from a reduction in TPE-related complications, such as allergic reactions, citrate toxicity, infection and bleeding. In contrast to the work of Kim et al (2017), we did not assess the use of an in-house ADAMTS13 assay in combination with our PLASMIC score-based approach and were able to show that a score-based approach alone could produce savings in settings where in-house ADAMTS13 testing is not available.
A key concern regarding use of the PLASMIC score as a gate-keeper for ADAMTS13 testing is that this approach could lead to missed cases of iTTP. While no algorithm is perfect, score-based decision support has proven successful in the diagnosis of other lethal haematological conditions, including venous thromboembolism (Ginsberg 1996, Wells, et al 2003, Wells, et al 1995) and heparin-induced thrombocytopenia (Lo, et al 2006). Although we did not examine patient outcomes, several factors argue against missed cases being a significant concern. First, the PLASMIC score is intended to supplement clinical judgment, not to replace it. Under the score-based approach, patients with low risk scores who are felt to have presentations consistent with iTTP would still receive ADAMTS13 testing. Second, the sensitivity of the PLASMIC score is almost 100% for true cases of iTTP both within the Harvard dataset and external cohorts (Bendapudi, et al 2017a, Jajosky, et al 2017, Li, et al 2018), making misclassification of iTTP patients unlikely. Third, the PLASMIC score has been shown to out-perform consensus clinical assessment by physicians alone in the diagnosis of iTTP (Bendapudi, et al 2017a). Fourth, increased ADAMTS13 testing has not been shown to capture additional cases of true iTTP, as demonstrated by the comparison between Groups A and B.
Our work is not without important limitations. First, our model is based on several fixed assumptions derived from historical trends within our consortium and may not be externally valid in the future or in other settings. Second, we were not able to evaluate the PLASMIC score-based strategy in Group B because the patient population in this cohort has undergone manual pre-screening, making decision tree analysis unfeasible. Nevertheless, because approximately 32% of patients receiving ADAMTS13 testing in Group B during the study period had a low-risk PLASMIC score, there is reason to believe that adding a score-based approach to the manual pre-screening strategy could result in further savings. Finally, we did not formally assess cost-effectiveness. However, in light of our data suggesting that a score-based approach is unlikely to lead to missed cases of iTTP together with the demonstrated cost savings, a PLASMIC score-based strategy would probably be the economically dominant one.
In summary, we investigated ADAMTS13 testing at three large academic medical centres and found that sites with an unrestricted testing strategy sent a significantly greater number of samples for ADAMTS13 assays without detecting additional cases of iTTP or reducing iTTP-related mortality. Replacing an unrestricted testing approach with a PLASMIC score-based algorithm for the diagnosis of patients with suspected iTTP could lead to significant cost savings. Given these findings, centres should consider stewardship policies around ADAMTS13 testing that incorporate a PLASMIC score-based strategy to risk-stratify suspected cases of iTTP.
Supplementary Material
Supp info
Click here to view.(243K, docx)
Acknowledgements:
This work was generously supported by the Luick Family Fund of the Massachusetts General Hospital. P.K.B. was supported by a career development grant from the National Heart, Lung, and Blood Institute (1K08 HL136840–01).
Footnotes
Conflicts of Interest: The authors declare no competing interests.
References
- Barton P, Bryan S & Robinson S (2004) Modelling in the economic evaluation of health care: selecting the appropriate approach. 9, 110–118. [PubMed] [Google Scholar]
- Bendapudi PK, Li A, Hamdan A, Uhl L, Kaufman R, Stowell C, Dzik W & Makar RS (2015) Impact of severe ADAMTS13 deficiency on clinical presentation and outcomes in patients with thrombotic microangiopathies: the experience of the Harvard TMA Research Collaborative. British journal of haematology, 171, 836–844. [PubMed] [Google Scholar]
- Bendapudi PK, Hurwitz S, Fry A, Marques MB, Waldo SW, Li A, Sun L, Upadhyay V, Hamdan A & Brunner AM, Viswanathan S, Kaufman RM, Uhl L, Stowell CP, Dzik WH & Makar RS, (2017a) Derivation and external validation of the PLASMIC score for rapid assessment of adults with thrombotic microangiopathies: a cohort study. The Lancet Haematology, 4, e157–e164. [PubMed] [Google Scholar]
- Bendapudi PK, Upadhyay V, Sun L, Marques MB & Makar RS (2017b) Clinical Scoring Systems in Thrombotic Microangiopathies. Semin Thromb Hemost, 43, 540–548. [PubMed] [Google Scholar]
- Center for Health Information and Analysis. (2018) Massachusetts Hospital Profiles: Data through Fiscal Year 2016. Available at: http://www.chiamass.gov/assets/docs/r/hospital-profiles/2016/Massachusetts-Hospitals-Profiles-Compendium-2016.pdf.
- Drummond MF, Sculpher MJ, Claxton K, Stoddart GL & Torrance GW (2015) Methods for the economic evaluation of health care programmes. Oxford university press. [Google Scholar]
- Franchini M & Mannucci PM (2008) Advantages and limits of ADAMTS13 testing in thrombotic thrombocytopenic purpura. Blood Transfusion, 6, 127. [PMC free article] [PubMed] [Google Scholar]
- Ginsberg JS (1996) Management of venous thromboembolism. N Engl J Med, 335, 1816–1828. [PubMed] [Google Scholar]
- Hovinga K & Lämmle B (2006) Complications of plasma exchange in patients treated for clinically suspected thrombotic thrombocytopenic purpura-hemolytic uremic syndrome. Transfusion, 46, 154–156. [PubMed] [Google Scholar]
- Jajosky R, Floyd M, Thompson T & Shikle J (2017) Validation of the PLASMIC score at a University Medical Center. Transfusion and Apheresis Science, 56, 591–594. [PubMed] [Google Scholar]
- Karen MKuntz LBR, Owens Douglas K., Sanders Gillian D., Trikalinos Thomas A., and Solomon Joshua A. (2017) Decision Models in Cost-Effectiveness Analysis In: Cost-Effectiveness in Health and Medicine (ed. by Neumann Peter J., G TG, Sanders Gillian D., Russell Louise B., Siegel Joanna E.). Oxford University Press. [Google Scholar]
- Kim CH, Simmons SC, Williams LA III, Staley EM, Zheng X & Pham HP (2017) ADAMTS13 test and/or PLASMIC clinical score in management of acquired thrombotic thrombocytopenic purpura: a cost‐effective analysis. Transfusion, 57, 2609–2618. [PMC free article] [PubMed] [Google Scholar]
- Kuntz KM, Russell LB, Owens DK, Sanders GD, Trikalinos TA, & Solomon JA (2017) Decision Models in Cost-Effectiveness Analysis In: Cost-Effectiveness in Health and Medicine (ed. by Neumann PJ, Sanders GD, Russell LB, Siegel JE & Ganiates TG). Oxford University Press, Uk. [Google Scholar]
- Li A, Khalighi PR, Wu Q & Garcia DA (2018) External validation of the PLASMIC score: a clinical prediction tool for thrombotic thrombocytopenic purpura diagnosis and treatment. J Thromb Haemost, 16, 164–169. [PMC free article] [PubMed] [Google Scholar]
- Lo GK, Juhl D, Warkentin TE, Sigouin CS, Eichler P & Greinacher A (2006) Evaluation of pretest clinical score (4 T’s) for the diagnosis of heparin-induced thrombocytopenia in two clinical settings. J Thromb Haemost, 4, 759–765. [PubMed] [Google Scholar]
- Rock GA, Shumak KH, Buskard NA, Blanchette VS, Kelton JG, Nair RC & Spasoff RA (1991) Comparison of plasma exchange with plasma infusion in the treatment of thrombotic thrombocytopenic purpura. New England Journal of Medicine, 325, 393–397. [PubMed] [Google Scholar]
- Sadler JE (2008) Von Willebrand factor, ADAMTS13, and thrombotic thrombocytopenic purpura. Blood, 112, 11–18. [PMC free article] [PubMed] [Google Scholar]
- Wells PS, Hirsh J, Anderson DR, Lensing AW, Foster G, Kearon C, Weitz J, D’Ovidio R, Cogo A & Prandoni P (1995) Accuracy of clinical assessment of deep-vein thrombosis. Lancet, 345, 1326–1330. [PubMed] [Google Scholar]
- Wells PS, Anderson DR, Rodger M, Forgie M, Kearon C, Dreyer J, Kovacs G, Mitchell M, Lewandowski B & Kovacs MJ (2003) Evaluation of D-dimer in the diagnosis of suspected deep-vein thrombosis. N Engl J Med, 349, 1227–1235. [PubMed] [Google Scholar]
