Abstract
Objective:
Thrombocytopenia (TCP), defined as platelet counts <150,000/µL, is a common complication of severe chronic liver disease (CLD). This retrospective study estimated the prevalence of thrombocytopenia in a large population of CLD patients and compared medical resource utilization and medical care costs by TCP status.
Methods:
A retrospective analysis was conducted on a longitudinal administrative claims database from a large US commercial health plan. Patients assigned CLD diagnosis codes from January 1, 2000–December 31, 2003 were identified; annual ambulatory visits, ER visits, inpatient stays, and general and CLD-related medical care costs for patients with vs without TCP (identified using diagnosis codes and platelet count data if available) were compared.
Results:
Of 56,445 patients with an ICD-9-CM diagnosis for CLD, 1289 (2.3%) had a diagnosis for TCP. CLD patients with vs without a TCP diagnosis had >2.5-times the annual number of liver disease-related ambulatory visits (3.6 vs 1.4; odds ratio [OR] = 2.6, p < 0.01); were 13-times more likely to have a liver-related inpatient stay (OR = 13.0, p < 0.01); were nearly 4-times more likely to have a liver-related ER visit (OR = 3.9, p < 0.01); had 3.5-fold greater mean annual overall medical care costs ($43,560 vs $12,270, p < 0.01); and had 7-fold greater annual liver disease-related medical care costs ($9940 vs $1420, p < 0.01). Similar results were seen for patients with platelet count data indicating TCP.
Limitations:
CLD and TCP are not always diagnosed, nor is diagnosis uniform or standardized; administrative claims data are subject to coding errors, and individuals covered are not necessarily representative of the general US population. The number of CLD patients in this study with TCP (n = 1289) is small relative to that expected in the general US population.
Conclusions:
In this analysis, CLD patients with TCP used significantly more medical resources and incurred significantly higher medical care costs than those without TCP.
Introduction
Chronic liver disease (CLD) includes many conditions with varying etiologies including infectious hepatitis, hepatic fibrosis, cirrhosis, and liver cancer. Recent epidemiologic data indicate that liver disease from all causes is the 12th leading cause of mortality in the US, accounting for more than 29,000 deaths annuallyCitation1. Thrombocytopenia is especially common in patients with advanced or end-stage liver disease, occurring in 64–76% of patients with cirrhosisCitation2,Citation3 and in up to 45% of patients with hepatitis C virus infectionCitation4. The presence of thrombocytopenia is most commonly associated with the presence of hepatic fibrosis, cirrhosis, and portal hypertensionCitation2. Inadequate thrombopoietin production, chronic viral hepatitis, platelet sequestration caused by splenic pooling or splenomegaly, and iatrogenic interventions such as interferon (IFN)-based antiviral therapy and cancer chemotherapy can all contribute to the development of thrombocytopeniaCitation2,Citation5–11. Thrombocytopenia is generally an indicator of advanced liver disease and warrants comprehensive diagnostic evaluation.
The presence of thrombocytopenia can increase the potential for bleeding, which results in an elevated morbidity and mortality risk in patients with liver disease. For example, thrombocytopenia is a well-characterized and significant clinical problem for CLD patients who have esophageal varices due to cirrhosis. Bleeding esophageal varices can severely reduce liver function in thrombocytopenic patients; for patients who are on waiting lists for orthotopic liver transplantation, such liver deterioration may render them unsuitable for transplantCitation3. Severe thrombocytopenia is also of clinical concern for CLD patients who require chemotherapy for solid tumors or hematologic malignancies; such therapies induce both myelosuppression and, less frequently, hepatic toxicities, and pre-existing thrombocytopenia may exclude such patients from needed treatmentCitation3. Thrombocytopenic patients often require multiple platelet transfusions, which can be associated with complications such as systemic infections, iron overload, and platelet refractorinessCitation12,Citation13.
Thrombocytopenia can also be a significant clinical issue for CLD patients who are undergoing surgical and diagnostic procedures including endoscopy, liver biopsy, abdominal or thoracic surgery, dental procedures, invasive cardiovascular diagnostic procedures/treatments, and orthopedic surgeryCitation14–18. For example, in cirrhotic patients undergoing abdominal surgery, bleeding accounts for 60% of all causes of death, and patients with prolonged prothrombin time (PT) have 7–14-times the mortality rate of patients with normal PTCitation14. Because of the risk of severe bleeding in patients with platelet counts under 50,000/µL, invasive procedures may be avoided or delayed in the patients with platelet counts below this threshold, regardless of surgical specialtyCitation19. Percutaneous liver biopsy, a principal method for diagnosing patients with suspected liver disease, is generally not recommended for patients with platelet counts <60,000/µL unless platelet transfusions are given in advanceCitation15. Similarly, a recent international consensus report discouraged major neurosurgery in patients with platelet counts <100,000/µLCitation19.
The objective of this retrospective study was to use diagnosis codes to estimate the prevalence of thrombocytopenia and related complications in a large clinical population of CLD patients and to compare medical resource utilization and medical care costs by thrombocytopenia status.
Patients and methods
Study design and data source
A retrospective analysis was conducted on a longitudinal administrative claims database including insurance eligibility, pharmacy and medical claims and laboratory results data from a large integrated US commercial health plan covering 23 million individuals from all geographic regions of the country. This study analyzed claims submitted by healthcare providers directly to the plan from January 1, 2000 through December 31, 2003 on behalf of 12.7 million patients. Approximately 40% of these patients had laboratory data. The authors conducted a similar retrospective analysis of the same database and outcomes in HCV patients by thrombocytopenia statusCitation20.
Claims data were coded according to insurance industry standards and collected for all types of care for physician offices, emergency rooms (ER), inpatients and outpatients in hospitals, surgery centers, and others. Claims for ambulatory services submitted by individual providers used the Centers for Medicare & Medicaid Services (CMMS) equivalent of the Health Care [financing administration] (HCFA)-1500 format. Inpatient or same day surgery claims for hospital facility services used the UB-82 or UB-92 format. Each facility service record may contain multiple diagnoses and procedures, coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Outpatient or clinic services and provider records are coded with the Current Procedural Terminology (CPT-4) schema or HCFA Common Procedure Coding System (HCPCS) codes. The cost of medical services used in this study corresponds to the transaction amount paid for the service to the provider by the health plan and patient (enrollees’ copays and coinsurance).
Pharmacy claims data include drug name, dosage form, drug strength, fill date, days of supply, average wholesale price, and de-identified patient and prescriber data, allowing for longitudinal tracking of medication fills and refills.
Study population
Adult patients included in the analysis were enrolled continuously for at least 1 year before the index date (date of the initial claim bearing a CLD diagnosis) and for 1 year following the index date (follow-up period). Patients with CLD were identified by ICD-9-CM diagnosis codes 571 (CLD and cirrhosis), 572 (liver abscess and sequelae of CLD), and 573 (other disorders of liver), between January 1, 2001 and December 31, 2003. Thrombocytopenia was identified by ICD-9-CM diagnosis codes 287.3–287.5 between January 1, 2001 and December 31, 2004. Patients diagnosed with liver cancer (155.x, 197.7, V10.07), who received a liver transplant (996.8, V42.7) or platelet transfusion (99.05), or who were diagnosed with various bleeding events also were identified, using diagnostic codes noted in . All patients diagnosed with CLD are included in the analysis cohort regardless of disease etiology. To confirm the diagnosis of thrombocytopenia, sensitivity analyses were performed on a sub-set of patients for whom complete laboratory data (e.g., platelet counts) were available. These analyses enabled correlations between platelet counts and major bleeding events (identified using the ICD-9-CM diagnosis codes and procedures and the CPT-4/HCPCS procedures listed in ). The study population and the total number of patients in each group are shown in .
Figure 1. Patient selection scheme from a population of adults diagnosed with chronic liver disease.
Table 1. Diagnosis codes used in this study.
Medical resource utilization was determined by comparing the annual ambulatory or outpatient visits, ER visits, and inpatient stays for CLD patients with evidence of thrombocytopenia vs those patients without a thrombocytopenia diagnosis or such evidence. Costs included in this analysis are the allowable or transaction costs (i.e., monies paid to the provider by the insurer or managed care plan), and by the patient (out-of-pocket responsibility). Costs are expressed in 2003 US dollars.
Analyses
The prevalence of thrombocytopenia in CLD patients was estimated by dividing the number of patients with diagnosed and coded thrombocytopenia during the 3-year study period by the total number of patients diagnosed with CLD. Demographics were described for CLD patients, including age, gender, and the number of patients with certain comorbid conditions or therapeutic interventions (i.e., anemia, neutropenia, liver cancer, IFN therapy, liver transplant). For evaluable patients with available laboratory results data, platelet counts were described in increments of 50,000/µL, and the number of bleeding events or platelet transfusions during the follow-up period was determined. includes a list of bleeding events considered in this analysis.
Negative binomial regression models were used in estimating or predicting incremental differences in medical resource utilization between CLD patients with or without diagnosed thrombocytopenia, where medical resources included annual ambulatory (outpatient) visits, ER visits, and inpatient stays. All analyses controlled for age, gender, health plan region, physician specialty, comorbid conditions, medications in baseline period, baseline inpatient visits, baseline ER visits, and baseline ambulatory visits. Annual overall medical care costs, liver disease-related medical care costs, and medical resource utilization during the study period were compared using multivariate modeling for CLD patients with and without diagnosed and coded thrombocytopenia.
As a confirmatory analysis, medical resource utilization and medical care costs were compared in the sub-set of patients with evaluable laboratory results data. CLD patients with platelet counts ≤100,000/µL during the study window (n = 453) were matched 1:4 with CLD patients with platelets >100,000/µL (n = 1812). Logistic regression analysis was performed on the medical resource utilization. Creation of such comparison cohorts increases the likelihood of tractable regression models and ensures robust estimates of effects. For the purposes of this analysis, patients with a platelet count ≤100,000/µL were considered to have clinically relevant thrombocytopenia, although the laboratory standards consider the lower range of normal to be <150,000/µL. Descriptive statistics are provided in this analysis; correlation coefficients are reported as measures of association. All analyses used Statistical Analysis System software® (Cary, NC).
Results
Demographic and health characteristics
The final study population consisted of 56,445 patients with CLD () as defined by the ICD-9-CM diagnostic codes 571.xx–573.xx in . Of these, 1289 (2.3%) had a diagnosis code for thrombocytopenia. Annual prevalence of thrombocytopenia among patients with CLD ranged from 3.3–4.1% over 3 years of the study.
Demographic and selected health characteristics are shown in , for the total CLD population and for the sub-population of matched cohorts (1:4) of patients with thrombocytopenia by platelet count (≤100,000/µL) and without thrombocytopenia by platelet count (>100,000/µL). For the total CLD population, the mean age of patients with vs without diagnosed thrombocytopenia was similar (51 years vs 48 years, respectively). There was a higher percentage of men among the CLD patients with a thrombocytopenia diagnosis (62.6%) compared with the percentage of male CLD patients without a thrombocytopenia diagnosis (49.4%). Compared with CLD patients without a thrombocytopenia diagnosis, there was a higher percentage of CLD patients with a thrombocytopenia diagnosis who had anemia (54.2% vs 18.5%, respectively) or neutropenia (20.8% vs 1.7%, respectively). Although the number of individuals was relatively low, the percentages of CLD patients with liver cancer (5.7% vs 1.5%, respectively) and liver transplants (2.1% vs < 1%, respectively) were higher in patients diagnosed with thrombocytopenia compared with patients without diagnosed thrombocytopenia. Additionally, a higher proportion of CLD patients with a thrombocytopenia diagnosis received IFN therapy during the study period compared with CLD patients without a thrombocytopenia diagnosis (5.9% vs 2.0%, respectively).
Table 2. Baseline demographic and selected health characteristics of all chronic liver disease patients in a managed care database and matched patients with or without thrombocytopenia.
Laboratory results data including platelet counts were available for 35.7% (n = 20,151) of all patients with CLD. In this sub-set, the mean platelet count during the study period was 167,110/µL (standard deviation [SD] = 110,059/µL) in CLD patients diagnosed with a thrombocytopenia diagnosis at some point during the study period, compared with 253,213/µL (SD = 69,571/µL) in CLD patients without a thrombocytopenia diagnosis. Patients diagnosed with thrombocytopenia underwent a greater number of platelet count assessments (mean = 3.68; SD = 4.78) compared to those without a thrombocytopenia diagnosis (mean = 2.21; SD = 2.47). The distribution of index platelet counts by thrombocytopenia diagnosis status is shown in . Based on this analysis, no thrombocytopenia diagnosis code was reported for a small percentage of patients (1.2%) with laboratory results evidence of platelet counts <100,000/µL.
Figure 2. Distribution of index platelet counts in chronic liver disease patients with laboratory results data (n = 20,151) by thrombocytopenia diagnosis.
Bleeding events and transfusions
The number and percentage of bleeding events and platelet transfusions by thrombocytopenia diagnosis status during the study period are shown in . Of the CLD patients with a thrombocytopenia diagnosis code, 27.8% had a bleeding event compared with 10.0% of the CLD patients without a thrombocytopenia diagnosis code. There was a strong statistical correlation between the mean platelet count and bleeding events in CLD patients with a thrombocytopenia diagnosis code (p < 0.007), but not in CLD patients without a thrombocytopenia diagnosis code (p = 0.614). A higher percentage of CLD patients with a thrombocytopenia diagnosis code received platelet transfusions (8.1%, 104/1289) compared with CLD patients without a thrombocytopenia diagnosis code (<1%, 79/55,156). Among patients with CLD receiving platelet transfusions, an average of two transfusions were given.
Table 3. Number of chronic liver disease patients with a bleeding event or platelet transfusion by thrombocytopenia diagnosis during the 12-month follow-up period.
Medical resource utilization
A negative binomial regression analysis showed that CLD patients with thrombocytopenia diagnoses had 2.5-times more liver disease-related ambulatory visits compared with CLD patients without thrombocytopenia diagnoses (95% confidence interval [CI] = 2.4–2.8; p < 0.01). Similarly, logistic regression estimates suggest that CLD patients with thrombocytopenia diagnoses were 4-times more likely to have liver disease-related ER visits (odds ratio [OR] = 3.9; 95% CI = 3.0–5.0; p < 0.01) and almost 13-times more likely to have liver disease-related inpatient hospital stays (OR = 12.9; 95% CI = 10.7–15.6; p < 0.01) than patients without thrombocytopenia (Appendix 1). However, fewer than 2% of CLD patients had liver-related inpatient stays and ER visits.
Logistic regression analysis was performed for matched cohorts of CLD patients with evidence of thrombocytopenia according to platelet count (≤100,000/µL) and without thrombocytopenia according to platelet count (>100,000/µL) (). Mean annual ambulatory visits were 3.2-fold greater (5.8 visits per year) in CLD patients with at least one platelet count ≤100,000/µL compared with CLD patients without such evidence of low platelet counts (1.8 visits per year; p < 0.01; ). CLD patients with evidence of low platelet counts were 6-times more likely to have ER visits (p < 0.01) and 14.5-times more likely to be hospitalized (p < 0.01) than CLD patients without evidence of low platelets. There were significantly more ambulatory visits, ER visits, and inpatient hospitalizations among CLD patients with thrombocytopenia than in CLD patients without thrombocytopenia, regardless of whether thrombocytopenia was indicated by diagnosis code or by platelet counts (p < 0.05) (i.e., observed in both the original analysis population and the confirmatory cohorts with lab results data).
Figure 3. Estimated mean annual ambulatory visits, emergency room visits, and inpatient stays in patients with chronic liver disease.
Medical care costs in patients with CLD
Mean annual medical care costs in patients with CLD are shown in . Overall and liver disease-related costs were significantly higher in CLD patients with thrombocytopenia, whether thrombocytopenia was indicated by diagnosis code or by platelet counts. Compared with CLD patients without thrombocytopenia, CLD patients with a thrombocytopenia diagnosis code had 3.5-fold greater overall medical care costs ($43,562 vs $12,271; p < 0.01) and 7-fold greater liver disease-related costs ($9938 vs $1417; p < 0.01). Similarly, patients with platelet counts ≤100,000/µL had overall medical care costs that were 3-fold greater ($40,959 vs $13,462; p < 0.01) and liver disease-related costs that were 8.8-fold greater ($14,224 vs $1610; p < 0.01) compared with CLD patients with platelet counts >100,000/µL.
Figure 4. Mean annual liver disease-related and overall medical care costs in patients diagnosed with chronic liver disease.
Impact of comorbid conditions on medical resource utilization and cost
In models controlling for the influence on medical resource utilization and costs of comorbid conditions reported by those with CLD, none of the comorbid conditions analyzed had a significant (p ≤ 0.01) independent effect on liver-related ambulatory visits, ER visits, or inpatient stays (Appendix 1). In contrast, modeling results suggested that heart disease, non-traumatic joint disorders, anemia, and gastrointestinal (GI) disorders (excluding upper GI disorders) substantively affect liver-related costs (Appendix 2). Liver disease-related costs among CLD patients who reported these comorbid conditions were ∼80% of liver disease costs for patients without such comorbid conditions.
Discussion
In this retrospective analysis of a medical claims database, thrombocytopenia diagnosis codes were present in only a small percentage (2.3%) of patients with CLD. CLD patients with thrombocytopenia by diagnosis code or platelet counts used significantly more medical resources and incurred significantly higher medical care costs during the follow-up period than those without evidence of thrombocytopenia. CLD patients with thrombocytopenia had a significantly greater number of ambulatory visits, ER visits, and inpatient hospitalizations compared with CLD patients without thrombocytopenia (p < 0.05 for all analyses) (). Overall and liver disease-related costs were both significantly higher in CLD patients with thrombocytopenia vs those patients without thrombocytopenia (). The significantly higher medical resource utilization and medical care costs were present in patients regardless of whether thrombocytopenia was indicated by diagnosis code or by platelet counts.
The mechanism underlying the relative contribution of thrombocytopenia to increased medical resource utilization is unclear in this analysis, except in those patients admitted to the hospital for bleeding events indicated by the diagnosis codes noted in . Thrombocytopenia is more common in patients with cirrhosis and correlates with the extent of structural change and severity of liver disease. As patients progress from compensated cirrhosis, in which the liver still performs its core functions, to decompensated cirrhosis, when the liver is extensively scarred and unable to function properly, the potential for life-threatening complications including bleeding increases—a clinical situation worsened by low platelet counts. Thrombocytopenia is often actively managed to mitigate bleeding potential, but many therapies used to treat thrombocytopenia, notably platelet transfusions, increase the costs of managing patients with CLD, increase the likelihood of transfusion reactions and complications, and do not have a lasting effect on platelet counts. The results of this retrospective analysis corroborate prior studies indicating that thrombocytopenia is a significant, independent predictor of increased mortality, medical resource utilization, and medical care costsCitation12,Citation21, and support the value of cost-effective treatments that avoid the potential complications of therapies such as platelet transfusions.
Patients with CLD often develop chronic medical conditions other than liver disease, and these comorbid conditions might additionally impact medical resource use and cost. Controlling for the impact of 12 common comorbidities in estimating the influence of thrombocytopenia on liver-related costs suggested that comorbid heart disease, non-traumatic joint disorders, anemia, and GI disorders (excluding upper GI disorders) might reduce liver-related costs (spend) compared to CLD patients without such comorbidities. In contrast, neither these nor other comorbidities had a significant impact on medical resource utilization nor overall healthcare costs. These results might suggest that proportionately less was spent on liver-related medical care in the presence of heart disease, joint disorders, anemia, or chronic GI conditions among CLD patients than in cases where such comorbid conditions are absent. Comorbid conditions could either reduce the ability of patients with CLD to receive liver-related care or increase the spend on comorbid conditions as a proportion of overall spend, thus reducing the amount spent on liver-related care. These observations warrant further analysis.
The authors conducted a companion study analyzing the same database by thrombocytopenia status for demographics, medical resource utilization, and medical care costs for patients with chronic hepatitis C viral infectionCitation20. The HCV patient population (n = 7905) was numerically smaller than the more inclusive CLD population (n = 56,445) described in this report but had similar outcomes. Thrombocytopenic HCV patients had a higher percentage of claims for bleeding events and platelet transfusions, and a significantly greater mean number of ambulatory, ER, and inpatient visits than those without thrombocytopenia. These findings collectively indicate that thrombocytopenia, whether the result of hepatitis or other primary hepatopathies, creates therapeutic challenges and adds significantly to demands on healthcare resources and associated costs. Although thrombocytopenia occurs in only a small percentage of CLD patients (< 3% in our study population), a pro-active approach to thrombocytopenia diagnosis, cost-effective treatment, and monitoring of at-risk patients is justified by the clinical benefits and potential treatment cost reductions that result.
The annual prevalence of thrombocytopenia by diagnosis code in patients with CLD ranged from 3.3–4.1% in this study. This rate is much lower than that reported in some prior studies of patients with cirrhosisCitation2,Citation3 or patients with HCV infection who had not been diagnosed with CLDCitation22. Thrombocytopenia may not have been coded in some patients in whom it occurred or was suspected if it was not the primary reason for the medical care visit or hospital stay. Thrombocytopenia is not always recorded on administrative insurance claims as a pertinent reason for medical care or clinic visits. Sometimes there is no formal laboratory testing for thrombocytopenia; if there is testing, sometimes platelet counts are not sufficiently low as to impact clinical care management, in which case thrombocytopenia would not be diagnosed or noted. In our analysis, lower platelet counts were associated with an increased likelihood of bleeding events and an increased likelihood of receiving platelet transfusions (). Platelet counts ≤100,000/µL were associated with significantly greater medical resource utilization (). Thus, it may be useful to identify and sub-categorize patients with thrombocytopenia according to median platelet counts, or duration of time with low platelets, in order to provide optimal medical care and anticipate medical resource utilization within the healthcare system. Another advantage to capturing platelet counts data is that platelets are a useful proxy for severity of liver disease, which claims data do not routinely include (i.e., biopsy results or results of structural tests of cirrhosis and fibrosis).
Platelet transfusions are the standard of care for the clinical management of severe thrombocytopenia. In clinical practice, the platelet count for diagnosing thrombocytopenia (∼ 50,000 to < 150,000/µL) or for administering platelet transfusions varies depending on the patient’s current medical condition and history. A platelet count of <50,000/µL is considered indicative of clinically relevant thrombocytopeniaCitation23, with less acute thrombocytopenia diagnosed as platelet counts <150,000/µL,Citation2 the lower limit of normal in healthy individuals. Bleeding risk in patients undergoing liver biopsy has been shown to increase significantly when platelet counts are ≤60,000/µLCitation24. A platelet count of ≤10,000/µL has been proposed as a suitable trigger for prophylactic platelet transfusion in the absence of a hemostatic challenge, although a 20,000/µL threshold is used by many institutionsCitation25,Citation26. The data from our analyses indicate that a platelet count of ≤100,000/µL may be a useful prognostic indicator of increased medical resource utilization ().
From a clinical perspective, platelet counts should be monitored in patients with liver disease. From a medical cost standpoint, platelets <100,000/µL appear to be predictive of higher costs and may be an indicator for a pro-active management strategy.
Platelet transfusions have been shown to be a significant predictor of increased length of hospital stay and cost per case in cirrhotic patients and in patients undergoing certain forms of chemotherapyCitation21,Citation27. Patients undergoing platelet transfusions are subject to complications such as infections and allergic or other reactions in 5–30% of casesCitation25,Citation28,Citation29, and refractoriness in a high percentage of casesCitation13,Citation25. Additionally, the relatively high cost and the difficulty of donor recruitment make prophylactic platelet transfusions less likely to be used in patients with low but adequate platelet countsCitation25. Alternative treatments for thrombocytopenia are needed that are safe, efficacious, and cost effectiveCitation28. Novel, non-invasive treatments that effectively increase platelet counts to a safe level could result in substantially reduced medical resource utilization in thrombocytopenic CLD patients by decreasing hospitalization and reducing inpatient stays. Limiting the use of transfusions could also reduce demand on the blood supply. The thrombopoietin receptor (TPO-R) agonists eltrombopag and romiplostim have been approved in the US for the treatment of chronic immune thrombocytopenia, and the novel oral TPO-R agonist, E5501, is currently in development. TPO-R agonists are also being evaluated in patients with liver diseaseCitation30,Citation31. Phase 2 studies are currently recruiting participants (NCT00914927, NCT01153919). Recently, data were reported from a phase 3 study of eltrombopag treatment for CLD patients with thrombocytopenia undergoing an elective invasive procedure (ELEVATE)Citation32. In this study, platelet counts were increased and platelet transfusions and bleeding events were decreased for patients treated with eltrombopag compared to those treated with placebo. However, the study was terminated early due to an increased incidence of thrombotic events (TEs) in the eltrombopag arm (six patients, 4.1%) compared to the placebo arm (two patients, 1.4%) (OR = 2.827, 95% CI = 0.695–11.501). All TEs except one (acute myocardial infarction in the placebo group) occurred in the portal vein system. Portal vein thrombosis has also been reported with romiplostim treatment of CLDCitation33. TEs are a theoretical risk of TPO-R agonist therapy; however, this is the first report of a randomized study observing an increased risk of TEs using one of these agents. Studies are needed to determine the risks and benefits of TPO-R agonists in CLD.
Several aspects of this retrospective analysis contributed to its strengths and research value. This analysis consisted of a large population of more than 56,000 patients with CLD distributed throughout the US and evaluated over a 3-year period. Patients with irrelevant or confounding conditions were identified and excluded from the analysis using ICD-9 diagnosis codes. The result was a targeted, non-pediatric population of patients with CLD that could be evaluated for differences in medical resource utilization based on the presence of thrombocytopenia without the influence of other confounding comorbidities. Administrative claims data have the advantages of providing a readily available, relatively inexpensive, anonymous source of diagnostic information that can be analyzed for medical resource utilization and medical costs. Retrospective claims data have been shown to have a higher degree of accuracy and congruence with medical records and the patients’ actual disease condition compared with self-reported conditions in patient surveysCitation34,Citation35. In addition, although the data reported are from 2001–2003, relative differences rather than absolute differences between groups were analyzed, so an analysis from 2010 is likely to be similar.
Limitations
Although administrative claims are a powerful source of data for research applications, certain inherent limitations should be recognized. CLD and thrombocytopenia are not always diagnosed, nor is diagnosis uniform or standardized. Some patients with low platelets may not have been recorded as such, and, in these patients, thrombocytopenia may have impacted clinical management with this impact consequently not recognized in the claims database. Some individuals with CLD may not have sought treatment and would have been excluded from the analytic frame. Individuals with bleeding may not submit a medical claim if the bleeding does not require medical intervention. In addition, administrative claims data are collected for purposes of payment, not for verification of quality of care or treatment pattern research, and are subject to coding errors. Diagnosis codes are not always sufficiently specific to capture the severity of the condition, progression of disease, or its duration. Plan enrollment is not always of sufficient length to model disease progression from onset through treatment and resolution. Thus, the medical care profiled here represents a limited longitudinal view. The individuals covered by a particular health plan or insurer may not be representative of the general US population and may differ in their demographic and health characteristics, especially since such a database represents only those patients actively seeking medical care. The prevalence of CLD in this database is 2.3%, which is much lower than the estimated prevalence of CLD in the general population (10%)Citation36. Medical care is also influenced by medical benefits available from the insurance provider; as such, the population of CLD patients in this database may look different from that of another commercial database, or a non-commercial database such as one from Medicare, Medicaid, or the Veterans Administration.
Conclusions
CLD patients with evidence of thrombocytopenia had significantly higher numbers of liver-related ambulatory or outpatient visits, ER visits, inpatient hospitalizations, and significantly greater overall and liver disease-related medical care costs during the study period, compared with CLD patients without thrombocytopenia. The thrombocytopenic group was twice as likely to have liver disease-related ambulatory visits, 4-times more likely to have a liver disease-related ER visit, 13-times more likely to have a liver disease-related inpatient stay, and had 7-fold greater annual liver disease-related medical care costs. Platelet counts <100,000/µL are a predictor of much higher medical care utilization and escalating costs. Novel treatments that safely and effectively treat thrombocytopenia may improve quality of care for patients with CLD and result in decreased utilization of medical resources, including platelet transfusions.
Transparency
Declaration of funding
Funding for this study was provided by GlaxoSmithKline (GSK). Funding for this study was restricted.
Declaration of financial/other relationships
Two authors (KG, DT) are employees of GSK. The lead author (FP) reported receipt of research grants from GSK.
Acknowledgments
All listed authors meet the criteria for authorship set forth by the International Committee for Medical Journal Editors, were actively involved in the preparation, writing, and revision of the manuscript, and approved of the final manuscript. Mark Dana and Ted Everson, PhD, of AOI Communications, L.P., contributed to writing the manuscript, and Mehul Delal, PhD, a former employee of GSK, assisted in the analyses. Kimberly Marino of GSK provided critical review of this manuscript. Some of the data from this study were presented at Digestive Disease Week in Washington, DC, in May 2007.
References
- Xu J, Kochanek KD, Murphy SL, et al. Deaths: final data for 2007. Natl Vital Stat Rep 2010;58:1-135
- Bashour FN, Teran JC, Mullen KD. Prevalence of peripheral blood cytopenias (hypersplenism) in patients with nonalcoholic chronic liver disease. Am J Gastroenterol 2000;95:2936-9
- Giannini EG. Review article: thrombocytopenia in chronic liver disease and pharmacologic treatment options. Aliment Pharmacol Ther 2006;23:1055-65
- Louie KS, Micallef JM, Pimenta JM, et al. Prevalence of thrombocytopenia among patients with chronic hepatitis C: a systematic review. J Viral Hepat 2011;18:1-7
- Adinolfi LE, Giordano MG, Andreana A, et al. Hepatic fibrosis plays a central role in the pathogenesis of thrombocytopenia in patients with chronic viral hepatitis. Br J Haematol 2001;113:590-5
- Dienstag JL, McHutchison JG. American Gastroenterological Association technical review on the management of hepatitis C. Gastroenterology 2006;130:231-64
- Rios R, Sangro B, Herrero I, et al. The role of thrombopoietin in the thrombocytopenia of patients with liver cirrhosis. Am J Gastroenterol 2005;100:1311-6
- Nudo CG, Wong P, Hilzenrat N, et al. Elderly patients are at greater risk of cytopenia during antiviral therapy for hepatitis C. Can J Gastroenterol 2006;20:589-92
- Ong JP, Younossi ZM. Managing the hematologic side effects of antiviral therapy for chronic hepatitis C: anemia, neutropenia, and thrombocytopenia. Cleve Clin J Med 2004;71(3 Suppl):S17-21
- Seipelt G, Hofmann WK, Martin H, et al. Comparison of toxicity and outcome in patients with acute myeloid leukemia treated with high-dose cytosine arabinoside consolidation after induction with a regimen containing idarubicin or daunorubicin. Ann Hematol 1998;76:145-51
- Zimmermann T, Böcher WO, Biesterfeld S, et al. Efficacy of an escalating dose regimen of pegylated interferon alpha-2a plus ribavirin in the early phase of HCV reinfection after liver transplantation. Transpl Int 2007;20:583-90
- Brown RS, Jr. A pharmacoeconomic analysis of thrombocytopenia in chronic liver disease. Aliment Pharmacol Ther 2007;26:41-8
- Murphy MF, Waters AH. Clinical aspects of platelet transfusions. Blood Coagul Fibrinolysis 1991;2:389-96
- Amitrano L, Guardascione MA, Brancaccio V, et al. Coagulation disorders in liver disease. Semin Liver Dis 2002;22:83-96
- Friedman LS. Controversies in liver biopsy: who, where, when, how, why? Curr Gastroenterol Rep 2004;6:30-6
- McVay PA, Toy PT. Lack of increased bleeding after liver biopsy in patients with mild hemostatic abnormalities. Am J Clin Pathol 1990;94:747-53
- Cobb WS, Heniford BT, Burns JM, et al. Cirrhosis is not a contraindication to laparoscopic surgery. Surg Endosc 2005;19:418-23
- Inabnet WB, Deziel DJ. Laparoscopic liver biopsy in patients with coagulopathy, portal hypertension, and ascites. Am Surg 1995;61:603-6
- Provan D, Stasi R, Newland AC, et al. International consensus report on the investigation and management of primary immune thrombocytopenia. Blood 2010;115:168-86
- Poordad F, Theodore D, Sullivan J, et al. Medical resource utilisation and healthcare costs in patients with chronic hepatitis C viral infection and thrombocytopenia. J Med Econ 2011;14:194-206
- Aggarwal A, Ong JP, Younossi ZM, et al. Predictors of mortality and resource utilization in cirrhotic patients admitted to the medical ICU. Chest 2001;119:1489-97
- Nagamine T, Ohtuka T, Takehara K, et al. Thrombocytopenia associated with hepatitis C viral infection. J Hepatol 1996;24:135-40
- Afdhal NH, Esteban R. Introduction: thrombocytopenia in chronic liver disease – treatment implications and novel approaches. Aliment Pharmacol Ther 2007;26(1 Suppl):1-4
- Sharma P, McDonald GB, Banaji M. The risk of bleeding after percutaneous liver biopsy: relation to platelet count. J Clin Gastroenterol 1982;4:451-3
- McCullough J. Current issues with platelet transfusion in patients with cancer. Semin Hematol 2000;37:3-10
- Rinder HM, Arbini AA, Snyder EL. Optimal dosing and triggers for prophylactic use of platelet transfusions. Curr Opin Hematol 1999;6:437-41
- Elting LS, Cantor SB, Martin CG, et al. Cost of chemotherapy-induced thrombocytopenia among patients with lymphoma or solid tumors. Cancer 2003;97:1541-50
- Brown RS, Jr. Review article: a pharmacoeconomic analysis of thrombocytopenia in chronic liver disease. Aliment Pharmacol Ther 2007;26 (1 Suppl):41-8
- Perrotta PL, Snyder EL. Non-infectious complications of transfusion therapy. Blood Rev 2001;15:69-83
- Dusheiko G. Thrombopoietin agonists for the treatment of thrombocytopenia in liver disease and hepatitis C. Clin Liver Dis 2009;13:487-501
- Tillmann HL, McHutchison JG. Use of thrombopoietic agents for the thrombocytopenia of liver disease. Semin Hematol 2010;47:266-73
- Afdhal N, Giannini E, Tayyab GN, et al. Eltrombopag in chronic liver disease patients with thrombocytopenia undergoing an elective invasive procedure: results from ELEVATE, a randomised clinical trial. J Hepatol 2010;52:Abstract 1185
- Dultz G, Kronenberger B, Azizi A, et al. Portal vein thrombosis as complication of romiplostim treatment in a cirrhotic patient with hepatitis C-associated immune thrombocytopenic purpura. J Hepatol 2011;55:229-32
- Tyree PT, Lind BK, Lafferty WE. Challenges of using medical insurance claims data for utilization analysis. Am J Med Qual 2006;21:269-75
- Raina P, Torrance-Rynard V, Wong M, et al. Agreement between self-reported and routinely collected health-care utilization data among seniors. Health Serv Res 2002;37:751-74
- American Liver Foundation. An introduction to liver care. American Liver Foundation, 2009. http://www.liverfoundation.org/downloads/alf_download_29.pdf. Accessed April 14, 2011
Table A1. Regression results for MRU in chronic liver disease.
Table A2. Regression results for liver-related and total healthcare costs in chronic liver disease.