OBJECTIVE. The objective of this study was to identify risk factors for developing chronic idiopathic thrombocytopenic purpura in a cohort of pediatric patients with idiopathic thrombocytopenic purpura.
METHODS. We conducted a retrospective cohort analysis of 259 children who were diagnosed with idiopathic thrombocytopenic purpura between 1991 and 2000 at 1 of 8 managed care organizations that comprise the Vaccine Safety Datalink. We reviewed the charts of 595 potential patients with idiopathic thrombocytopenic purpura from the 8 Vaccine Safety Datalink sites and excluded patients with known causes of thrombocytopenia. Chronic idiopathic thrombocytopenic purpura was defined as having thrombocytopenia for 6 months beyond the initial diagnosis. The risk for developing chronic idiopathic thrombocytopenic purpura was assessed using simple and multivariable analyses.
RESULTS. Of the 259 cases of idiopathic thrombocytopenic purpura, 197 (76%) were acute, 60 (23%) were chronic, and 2 (1%) could not be determined. Among the acute cases, the mean duration of illness was 22 days. There was 1 serious bleeding outcome in the cohort. In multivariable regression analysis, the patients with chronic illness were older, less likely to present with mucosal bleeding, less likely to have had an acute illness before diagnosis, and more likely to present with a platelet count >20000/μL than children with acute idiopathic thrombocytopenic purpura. In particular, children whose illness was diagnosed at ≥10 years of age and who had platelet counts ≥20000/μL had an approximate fivefold risk for progressing to chronic disease when compared with children who presented at ≤2 years of age with platelet counts <20000/μL.
CONCLUSIONS. Although idiopathic thrombocytopenic purpura tends to be a benign and self-limited condition, acute and chronic idiopathic thrombocytopenic purpura seem to be distinct illnesses defined by age, platelet count, bleeding symptoms, and the presence of acute illness before diagnosis. Physicians should be aware of these differences when advising their patients and families.
Childhood idiopathic (or immune) thrombocytopenic purpura (ITP) is a hemorrhagic disorder characterized by low platelet counts, spontaneous bruising, purpuric or petechial rash, and mucosal bleeding. It occurs at a rate of ∼2 to 5 cases per 100000 person-years in otherwise healthy children younger than 15 years.1–6 Between 70% and 80% of the cases are acute and spontaneously resolve within 6 months of presentation. The remaining 20% to 30% of the cases are classified as chronic ITP, which is formally defined as having low platelet counts for >6 months after diagnosis. The most serious complication of ITP is intracranial hemorrhage, which occurs in <0.5% of the cases.7–9
Studies on the natural history of childhood ITP contain descriptive data on presenting symptoms, diagnosis patterns (eg, the use of bone marrow examinations), patient follow-up, demographic characteristics, and response to treatment.7,8,10–15 The study populations were ascertained from single hematology-oncology centers in the United States and Australia, national health registries in Europe and the 5 Nordic countries, and an international registry with 38 participating countries. The methods for ascertaining the ITP cases ranged from detailed medical chart review to questionnaires completed by participating physicians. When comparing acute with chronic cases, these studies showed that children with chronic ITP are more likely to be older and present with higher platelet counts. They demonstrated that chronic ITP tends to occur with equal frequency in boys and girls, although 1 study found significantly more chronic cases among girls aged 6 to 14 years.12 Two of the studies also suggested an interaction between age and presenting platelet count: 1 showed a widening discrepancy in presenting platelet counts between acute and chronic cases as age at diagnosis increases, and the other demonstrated an increasing proportion of chronic cases as both presenting platelet counts and age increase. Although these results suggest that both age and presenting platelet count predict chronic ITP, the analyses were limited to bivariate comparisons that were unadjusted for the other predictors. They therefore did not examine how strongly these characteristics were associated with chronic ITP in an analysis that considered multiple risk factors, such as gender, recent acute illness, medication exposure, recent measles-mumps-rubella (MMR) vaccination (a known cause of immune thrombocytopenia),16–18 and the presence of mucosal bleeding at diagnosis.
To address these gaps in the research, we conducted a multisite, population-based cohort study of medical chart–verified cases of childhood ITP. The objective of this study was to identify risk factors that significantly predict the likelihood that an initial diagnosis of ITP progresses to chronic disease.
ITP cases were identified within the 8 managed care organizations (MCOs) that comprise the Vaccine Safety Datalink (VSD) project, a Centers for Disease Control and Prevention–funded project that uses large administrative databases to conduct epidemiologic studies of vaccine safety.19 All of the MCOs in the VSD either own their own hospitals or contract with pediatric hospitals to hospitalize children. In all instances, the hospital course and discharge summaries are part of the medical chart that was reviewed for children with ITP as part of this study. There are ∼5.5 million children younger than 18 years annually in the VSD cohort. Cases for this study were collected from the years 1991 through 2000.
ITP cases were ascertained in a 3-step process. Potential cases were identified from the VSD automated databases, which contain medical encounter data from the outpatient, inpatient, emergency department, and laboratory settings of the 8 participating MCOs. Seven of the 8 MCOs provided data for children up to 18 years of age; the eighth provided data only for the 12- to 23-month age group. In step 1, automated databases were screened to identify children with either 2 platelet counts ≤50000/μL in a 6-week period or 1 platelet count ≤50000/μL and an associated International Classification of Diseases, Ninth Revision (ICD-9) diagnosis code of 287.0 to 287.9 within 6 weeks. In step 2, potential cases from step 1 were excluded when they had an ICD-9 code for a condition that is known to cause thrombocytopenia, such as aplastic anemia (ICD-9 284.x), defibrination syndrome (ICD-9 286.6), acquired hemolytic anemia (ICD-9 283.x), or malignant neoplasm (ICD-9 140.x–165.x, 170.x–176.x, 179.x–208.x). In step 3, the patients’ medical charts from all MCO sites were reviewed by a pediatrician at 1 VSD site to confirm case status. The reviewer was blinded to MMR vaccination status.
Using the medical charts, ITP cases were defined as having at least 1 platelet count ≤50000/μL with normal red and white blood cell indices and the presence of clinical signs and symptoms of spontaneous bleeding. Cases were excluded during chart review when the patients were exposed to platelet-depleting medication (phenytoin, valproic acid, sulfonamide antibiotics) or when they had a viral illness that is known to cause thrombocytopenia (eg, varicella, Epstein-Barr virus) within 6 weeks before diagnosis. Asymptomatic cases (ie, no signs of bruising or bleeding) that were serendipitously found, such as a low platelet count found on a complete blood count screening for anemia or for workup of febrile illness, were also excluded. The ITP resolution date was defined as the date of the first platelet count >100000/μL and no evidence of a drop in platelet count in the subsequent weeks. A case of ITP was considered chronic when the thrombocytopenia lasted for ≥6 months or when the medical chart described the illness as chronic ITP ≥6 months after the initial diagnosis.
Univariate and bivariate statistics were used to describe the case population and to assess differences in gender, disease state (acute versus chronic), and MMR vaccination status. Exploratory data analyses showed that a parametric survival regression with a log normal distribution was appropriate for the analysis of the timing of ITP resolution.20 The parametric survival regression models were used to evaluate the effects of the following variables on development of chronic ITP: gender, age at diagnosis, medication exposure, MMR vaccination exposure, acute illness within 6 weeks of ITP diagnosis, presenting symptoms, and presenting platelet count. The follow-up period for each case began at the date of ITP diagnosis, and cases were followed either until the date of ITP resolution or until thrombocytopenia persisted beyond 6 months, whichever came first. Age at diagnosis was transformed into a categorical variable on the basis of the 25th, 50th, and 75th percentiles of its distribution. The age groupings were as follows: <2 years, 2 to 4 years, 5 to 10 years, and >10 years. Presenting platelet count was dichotomized to ≥20000/μL versus <20000/μL because the effect of the variable was not linear. We chose this specific cut point because 20000/μL represented the 75th percentile of the distribution, as well as the level below which serious bleeding episodes tend to occur and when treatment is recommended.8,21,22 Furthermore, chronic ITP is generally associated with presenting platelet counts >20000/μL.7,8,11,12
Using the electronic VSD databases, 595 potential ITP cases were identified between January 1, 1991, and December 31, 2000. Within this group, 259 newly diagnosed cases of ITP were confirmed after chart review, representing an incidence rate of ∼2.7 cases per 100000 years of person-time follow-up. Of these, 197 (76%) were acute and 60 (23%) were chronic ITP cases (2 cases were unknown). In 6 (10%) of the chronic cases, the clinical data for the date of ITP onset, such as complete blood count results, treatments received, recent acute illnesses, and recent medications, could not be extracted from the medical charts. In 12 (6%) of the acute cases, the ITP resolution date could not be confirmed, and the presenting platelet count could not be ascertained in 1 (0.5%) acute case. Certain analyses, therefore, were conducted on subpopulations of 252 (97%) patients with complete medical chart data for the ITP onset date and on 185 (94%) of the patients with acute ITP and confirmed resolution dates.
A total of 336 potential cases were excluded by chart review. Reasons for exclusion were laboratory error, lupus, drug-induced thrombocytopenia, thrombocytopenia as a result of pregnancy, serendipitous case finding, Epstein-Barr virus infection, febrile illness, malaria, neonatal thrombocytopenia, leukemia, varicella, sepsis, AIDS, and incomplete medical chart. Because of the conservative inclusion criteria of platelet counts ≤50000/μL, there were only 13 (2%) serendipitous findings of thrombocytopenia among the 595 charts reviewed.
The gender distribution of the 259 cases was approximately equal (Table 1), with a mean age at diagnosis of 6.2 years (95% confidence interval [CI]: 5.6–6.9). On average, girl presented at approximately the same age as boys. The cases were diagnosed evenly throughout the year (P = .14), with a nonsignificant peak incidence in February (n = 32; 12%) and a low incidence in August (n = 13; 5%).
Among the 252 cases with complete data for the ITP onset date, the mean presenting platelet count was 13600/μL (95% CI: 11700–15500), and the median value was 8000/μL. Seventy-nine percent of the patients presented with platelet counts <20000/μL. These patients tended to have the most severe clinical symptoms, such as mucosal bleeding, when compared with patients who presented with platelet counts >20000/μL.
A total of 180 (70%) of the cases had an acute illness or reported being ill within 6 weeks of ITP onset. These illnesses included otitis media, upper respiratory infection, febrile illness, gastroenteritis, and streptococcal pharyngitis.
Eighty percent of the patients received at least 1 medication for their ITP. Of those who received medication, 67% were treated with intravenous immunoglobulin, 42% were treated with corticosteroids, and 15% received both medications.
Overall, the mean duration of illness from presentation to resolution in the acute cases was ∼22 days (95% CI: 18–25). Seventy-seven percent of the 197 acute ITP cases resolved within 1 month of presentation, and the remaining 23% resolved between 1 and 4 months after presentation.
The medical chart review revealed 1 serious hemorrhagic outcome after the initial diagnoses of ITP (rate: 0.39%; 95% CI: 0.01%–2.13%). A 16-month-old boy was hospitalized for 19 days as a result of recurrent epistaxis and hematochezia that required multiple platelet and red blood cell transfusions. The patient subsequently recovered but had chronic ITP. In the entire case population, no intracranial hemorrhages, fatalities, or subsequent diagnoses of chronic and/or serious illnesses that are known to cause thrombocytopenia were identified.
Acute Versus Chronic ITP
When compared with the patients with acute illness, the patients with chronic illness tended to be older and female and present with higher platelet counts (Table 1). At presentation, the patients with chronic and acute illness had mean ages of 10.0 years (95% CI: 8.5–11.4) and 5.1 years (95% CI: 4.5–5.8), respectively (P < .001). As shown in Fig 1, the proportion of chronic illness increases as age at presentation increases. Approximately half of the patients with acute illness were male, whereas the patients with chronic illness tended to be female (58% female vs 42% male; P = .15). The patients with chronic illness presented with a mean platelet count of 16400/μL (95% CI: 12100–20600), whereas the patients with acute illness cases had mean presenting platelet count of 12900/μL (95% CI: 10700–15000), but the difference was not statistically significant (P = .14). Mucosal bleeding was also more common in the patients with acute illness (50%) than in the patients with chronic illness (23%; P = .002). The patients with acute illness, however, were not more likely to receive treatment or a bone marrow examination than the patients with chronic illness. Treatment (as defined in Table 1) was administered to 83% and 79% (P = .55) of the patients with chronic and acute ITP, respectively; bone marrow examinations were conducted in 25% and 18% (P = .25) of these 2 respective groups.
In a multivariable survival regression analysis, older age at diagnosis, higher presenting platelet count, lack of mucosal bleeding at presentation, and lack of a previous acute illness were significant predictors of development of chronic ITP (Table 2). Patients who had received a diagnosis of an acute illness 6 weeks before their ITP diagnosis were less likely to develop chronic ITP than patients without a history of a previous acute illness (relative risk [RR]: 0.44; 95% CI: 0.26–0.80). Those who presented with mucosal bleeding were also less likely to progress to chronic ITP when compared with patients without documented mucosal bleeding at diagnosis (RR: 0.50; 95% CI: 0.31–0.74). In the multivariable analysis, gender, current medications, and recent MMR vaccination were not associated with an increased risk for development of chronic ITP.
Children who presented at ages ≥10 years were at higher risk for developing chronic ITP than those who presented at ages ≤2 years (RR: 3.60; 95% CI: 1.90–6.82). There was, however, a statistical interaction between age and presenting platelet count on the risk for progression to chronic ITP. When compared with children aged ≤2 years, the risk among patients aged ≥10 years increased as their presenting platelet counts increased (Table 2). Using our initial cut point of 20000/μL, patients who were older than 10 years and had an initial platelet count >20000/μL had an approximately fivefold increased risk when compared with patients who were aged ≤2 years and had lower presenting platelet counts (RR: 4.87; 95% CI: 1.52–15.64). When the cut point was increased to 30000/μL and 35000/μL, the risks among patients who were ≥10 years increased to 11- and 25-fold, respectively.
In this large, US-based, multisite cohort study of the risk for development of chronic ITP in children, we found that most patients presented with cutaneous bleeding, platelet counts <20000/μL, and a recent acute illness within 6 weeks of diagnosis. A majority of the children also had an acute form of ITP that resolved within 1 month of diagnosis, and 1 child had a serious bleeding outcome.
Our multivariable analyses show that acute illness before ITP diagnosis and the presence of mucosal bleeding symptoms at diagnosis were inversely related to the risk for development of chronic ITP. Because mucosal bleeding is more likely to bring a child to diagnosis, our results support previous findings that insidious onset of symptoms was associated with chronic illness.10,14 Our results also demonstrate that both age and presenting platelet count modified the risk for progression to chronic illness. Patients who received the diagnosis after 10 years of age and had platelet counts >20000/μL were at particularly high risk for developing chronic ITP, and the risk continued to increase with progressively higher platelet counts. Our data, therefore, suggest that acute and chronic ITP may be distinct illnesses and that the relationship between age and presenting platelet count could have important clinical implications. On the basis of this information, pediatricians may offer different advice to families of children who have ITP and are older and have higher platelet counts and be more cautious about the resolution of their child's symptoms, especially if the child does not have mucosal bleeding or a history of a recent acute illness.
Although it has been suggested that girls are at increased risk for developing chronic ITP,12,18 our multivariable analyses showed that girls were not more likely than boys to have ITP for >6 months. Similarly, neither recent MMR vaccination nor current medications were associated with an increased risk for development of chronic illness.
Children who developed chronic ITP were no more likely to have had a bone marrow examination or to have received treatment than children with acute ITP. Furthermore, there was only hemorrhagic complication in the entire cohort. Thus, in our analysis, treatment could not be shown to offer protection from complications of ITP or to prevent progression to chronic ITP. Given the lack of evidence for treatment effectiveness12,13,23–25 and the debate about whether any treatment is indicated in this condition,6,11,15,17,26–32 our findings suggest that cautious observation of children with ITP does not put them at any increased risk for developing chronic disease.
To our knowledge, this study presents the first multivariable analysis to examine the risk for chronic ITP in a large, multisite cohort of childhood cases from the United States. One potential limitation of the study, however, is that the case population may underrepresent the true population of childhood ITP. Thrombocytopenia is formally defined as having a platelet count <150000/μL, yet our case population was limited to children with at least 1 platelet count <50000/μL. We used this conservative criterion because spontaneous bleeding seldom occurs with platelet counts >50000/μL,33 and cases with platelet counts between 50000 and 150000/μL are not likely to come to medical attention; therefore, extending the criteria to include patients with platelet counts >50000/μL could lead to biased results, because this population of patients may not represent the true population of patients with asymptomatic ITP. Limiting the population to patients with lower platelet counts helped to ensure that the study population accurately represented the true population of patients with clinically relevant, symptomatic ITP.
Our conservative case ascertainment criteria may also explain why our estimated incidence rate of 2.7 cases per 100000 person-years is at the lower end of the published range of 2 to 5 cases per 100000 person-years.1–6 We believe that this modest estimate can largely be attributed to our study's exclusion of mild cases with platelet counts >50000/μL. We identified 1 serious bleeding complication in the cohort, which corresponds with much of the published data demonstrating that <1% of patients with childhood ITP develop severe bleeding complications.7–9 If we had included patients who presented with platelets between 50000/μL and 150000/μL, then we believe that our estimated incidence rate would have been closer to the high end of the published range.
Another potential limitation is that the follow-up data from our chart review did not extend beyond 6 months after the initial diagnoses of ITP. We limited the follow-up to 6 months because it represents the time after which persistent thrombocytopenia is defined as chronic. A recent study, however, showed that ∼26% of children with persistent thrombocytopenia recovered from ITP between 7 and 12 months after diagnosis.34 This suggests that the current definition of chronic ITP may not be appropriate. We believe that future studies are needed to help to define a more precise timeline that differentiates chronic from acute ITP.
ITP is an uncommon condition that tends to resolve spontaneously within 6 months of presentation. Although usually self-limited, ITP can be a worrisome diagnosis for families and clinicians. Serious hemorrhagic complications of ITP seldom occur, regardless of whether symptoms persist beyond 6 months. Our results suggest that acute and chronic ITP are separate conditions associated with clinical differences in bleeding symptoms, previous acute illness, age, and presenting platelet count. In practice, physicians may need to consider these characteristics when advising their patients and families about potential resolution of ITP.
This work was supported by the Vaccine Safety Datalink, funded by the Centers for Disease Control and Prevention (contract 200-2002-00732).
We acknowledge Candido Chacon for database management and chart tool design for this study.
- Accepted July 20, 2007.
- Address correspondence to Jason Glanz, PhD, Institute for Health Research, Kaiser Permanente Colorado, PO Box 378066, Denver, CO 80237-8066. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
- Chu YW, Korb J, Sakamoto KM. Idiopathic thrombocytopenic purpura. Pediatr Rev.2000;21 (3):95– 104
- Walker RW, Walker W. Idiopathic thrombocytopenia, initial illness and long term follow up. Arch Dis Child.1984;59 (4):316– 322
- ↵Lilleyman JS. Intracranial haemorrhage in idiopathic thrombocytopenic purpura. Paediatric Haematology Forum of the British Society for Haematology. Arch Dis Child.1994;71 (3):251– 253
- ↵Watts RG. Idiopathic thrombocytopenic purpura: a 10-year natural history at the Childrens Hospital of Alabama. Clin Pediatr (Phila).2004;43 (8):691– 702
- ↵Robb LG, Tiedman K. Idiopathic thrombocytopenic purpura: predictors of chronic disease. Arch Dis Child.1990;65 (5):502– 506
- ↵Reid, MM. Chronic idiopathic thrombocytopenic purpura: incidence, treatment, and outcome. Arch Dis Child.1995;72 (2):125– 128
- ↵Miller E, Waight P, Farrington P, Andrews N, Stowe J, Taylor B. Idiopathic thrombocytopenic purpura and MMR vaccine. Arch Dis Child.2001;84 (3):227– 229
- ↵France EK, Glanz J, Xu S, et al. The risk of idiopathic thrombocytopenia following measles-mumps-rubella immunization in children. Pediatrics.2008;121 (3). Available at: www.pediatrics.org/cgi/content/full/121/3/e41
- ↵Chen RT, Glasser JW, Rhodes PH, et al. Vaccine Safety Datalink project: a new tool for improving vaccine safety monitoring in the United States. Pediatrics.1997;99 (6):765– 773
- ↵Allison P. Survival Analysis Using the SAS System. Cary, NC: SAS Institute;1996
- ↵George JN, Woolf SH, Raskob GE, et al. Idiopathic thrombocytopenic purpura: a practice guideline developed by explicit methods for the American Society of Hematology. Blood.1996;88 (1):3– 40
- ↵Bolton-Maggs P, Tarantino MD, Buchanan GR, Bussel JB, George JN; American Society of Pediatric Hematology/Oncology. The child with immune thrombocytopenic purpura: is pharmacotherapy or watchful waiting the best initial management? A panel discussion from the 2002 meeting of the American Society of Pediatric Hematology/Oncology. J Pediatr Hematol Oncol.2004;26 (2):146– 151
- Bolton-Maggs PH. Idiopathic thrombocytopenic purpura. Arch Dis Child.2000;83 (3):220– 222
- ↵Buchanan GR. Thrombocytopenia during childhood: what a pediatrician needs to know. Pediatr Rev.2005;26 (11):401– 409
- Copyright © 2008 by the American Academy of Pediatrics