![[ figure 1 image ]](/images/0041f1.gif)
IDIOPATHIC thrombocytopenic purpura (ITP, also known as primary immune thrombocytopenic purpura) is a hematologic disorder for which appropriate diagnostic and treatment strategies are uncertain. In 1994, the American Society of Hematology (ASH) established a panel to produce explicitly developed practice guidelines for the diagnosis and management of ITP. "Explicitly developed," evidence-based practice guidelines, which are being issued increasingly by medical specialty societies, combine a critical appraisal of scientific evidence with practice recommendations that state clearly to what extent the guidelines are based either on published scientific evidence or opinion (eg, clinical experience).1-4 More details about the clinical practice guideline movement are provided elsewhere.5-7
Diagnosis
The diagnosis of ITP is based principally on the history, physical examination, complete blood count, and examination of the peripheral smear, which should exclude other causes of thrombocytopenia. Further diagnostic studies (see Table 7) are generally not indicated in the routine work-up of patients with suspected ITP, assuming that the history, physical examination, and blood counts are compatible with the diagnosis of ITP and do not include atypical findings that are uncommon in ITP or suggest other etiologies. Patients with risk factors for human immunodeficiency virus (HIV) infection should be tested for HIV antibody, and an abdominal computed tomographic (CT) scan or ultrasound examination is appropriate in patients with suspected splenomegaly on initial physical examination. Bone marrow aspiration should be performed to establish the diagnosis in patients with persistent thrombocytopenia (lasting more than 6 to 12 months) and in those unresponsive to intravenous Ig (IVIg), but it should not be performed to establish the diagnosis before initiating IVIg therapy. Additional testing is also generally unnecessary, and sometimes inappropriate, when performed on a routine basis to establish the diagnosis before splenectomy or to evaluate patients who have not responded to glucocorticoid therapy, IVIg, and splenectomy (see Table 7).
Treatment
Children with platelet counts >30,000 should not be hospitalized and do not routinely require treatment if they are asymptomatic or have only minor purpura; they should not be given glucocorticoids, IVIg, or anti-Rh(D) as routine initial treatment. Children with platelet counts <20,000 and significant mucous membrane bleeding and those with counts <10,000 and minor purpura should be treated with specific regimens of IVIg or glucocorticoids (see text). Patients with severe, life-threatening bleeding should be hospitalized and receive conventional critical care measures, along with treatment for ITP: appropriate regimens include high-dose parenteral glucocorticoid therapy, IVIg, and platelet transfusions.
Splenectomy is clearly appropriate or inappropriate in specific clinical situations (see text). If an elective splenectomy is planned, appropriate preoperative therapy includes prophylactic IVIg therapy for patients with platelet counts <30,000, and IVIg, parental glucocorticoids, and anti-Rh(D) for patients with platelet counts <10,000. Inappropriate preoperative prophylaxis includes IVIg, oral glucocorticoid therapy, or anti-Rh(D) when platelet counts exceed 50,000, parenteral glucocorticoid therapy when platelet counts exceed 30,000, and platelet transfusions when platelet counts exceed 20,000.
When ITP symptoms persist after primary treatment (glucocorticoid, IVIg) and splenectomy, further treatment is indicated in children with platelet counts <30,000 who have active bleeding. Panel members suggested many treatments as reasonable options but did not reach consensus on any single regimen, reflecting the lack of evidence that any single treatment is more effective than another.
Diagnosis
The diagnosis of ITP is based principally on the history, physical examination, complete blood count, and examination of the peripheral smear, which should exclude other causes of thrombocytopenia. Further diagnostic studies (see Table 7) are generally not indicated in the routine work-up of patients with suspected ITP, assuming that the history, physical examination, and blood counts are compatible with the diagnosis of ITP and do not include atypical findings that are uncommon in ITP or suggest other etiologies. Patients with risk factors for HIV infection should be tested for HIV antibody. Bone marrow aspiration is appropriate to establish the diagnosis in patients over age 60 and in patients considering splenectomy. Additional testing is also generally unnecessary, and sometimes inappropriate, when performed on a routine basis to establish the diagnosis before splenectomy or to evaluate patients who have not responded to glucocorticoid therapy and splenectomy (see Table 7). Preoperative thyroid function testing is appropriate to rule out occult hyperthyroidism or hypothyroidism before elective splenectomy.
Treatment
Patients with platelet counts >20,000 should not be hospitalized if they are either asymptomatic or have only minor purpura. Patients with counts >50,000 do not routinely require treatment; they should not be given glucocorticoids or IVIg as routine initial treatment. IVIg is also inappropriate as initial treatment in patients with counts >30,000 who are asymptomatic or have only minor purpura. However, treatment is indicated in patients with platelet counts <20,000 to 30,000, and those with counts <50,000 and significant mucous membrane bleeding (or risk factors for bleeding, such as hypertension, peptic ulcer disease, or a vigorous lifestyle). Initial therapy with glucocorticoids (eg, prednisone) is appropriate in such patients. Hospitalization is appropriate for patients with platelet counts <20,000 who have significant mucous membrane bleeding. Patients with severe, life-threatening bleeding should also be hospitalized and should receive conventional critical care measures, along with treatment for ITP: appropriate regimens include high-dose parenteral glucocorticoid therapy, IVIg, and platelet transfusions.
Splenectomy is clearly appropriate or inappropriate in specific clinical situations (see text). It should not be performed as initial therapy in patients who have no bleeding, minor purpura, or even mucous membrane bleeding. In a patient who has had bleeding symptoms (eg, epistaxis, menorrhagia), splenectomy is often appropriate if platelet counts remain below 30,000 after 4 to 6 weeks of medical treatment. If an elective splenectomy is planned, appropriate preoperative therapy includes prophylactic IVIg or oral glucocorticoid therapy for patients with platelet counts <20,000. Inappropriate preoperative prophylaxis includes IVIg, oral or parenteral glucocorticoid therapy, and anti-Rh(D) when platelet counts exceed 50,000, and platelet transfusions when platelet counts exceed 10,000.
When ITP symptoms persist after primary treatment (glucocorticoid) and splenectomy, further therapy is recommended in patients with platelet counts <30,000 who have active bleeding. The most commonly recommended first-choice treatment options include IVIg, glucocorticoids, accessory splenectomy, and no additional treatment, but other agents may also be appropriate (see text). Women with ITP who are of childbearing age and have counts <10,000 after splenectomy and other treatments should be discouraged from becoming pregnant.
Diagnosis
The diagnosis of ITP during pregnancy generally does not require special laboratory testing (see Table 7). The patient's blood pressure should be measured to rule out preeclampsia as an alternative diagnosis; liver function testing is also appropriate. Patients with risk factors for HIV infection should be tested for HIV antibody.
Treatment
Recommendations for pregnant women are different from other adults in some situations. Pregnant women with ITP and platelet counts >50,000 do not routinely require treatment and should not receive glucocorticoids or IVIg as routine initial therapy. Women with counts of 30,000 to 50,000 in the first or second trimester also should not receive routine initial treatment. Treatment is required for women with platelet counts <10,000, and for those with platelet counts of 10,000 to 30,000 who are in their second or third trimester or are bleeding. IVIg is appropriate initial treatment for women with platelet counts <10,000 in the third trimester, and for those with counts of 10,000 to 30,000 who are bleeding. In pregnant women who have failed glucocorticoid and IVIg therapy, splenectomy is appropriate in the second trimester in women with platelet counts <10,000 who are bleeding. Splenectomy should not be performed in asymptomatic pregnant women with platelet counts >10,000.
As labor and delivery approach, women with ITP do not require testing for maternal platelet antibodies. Percutaneous umbilical vein blood sampling (PUBS) or fetal scalp vein sampling to measure the fetal platelet count and predict the risk of neonatal bleeding are not necessarily required. PUBS and fetal scalp vein sampling are unnecessary in pregnant women without known ITP even with platelet counts as low as 40,000 at term. Women with ITP should be delivered by cesarean section in selected circumstances (see text). In general, assuming the fetal platelet count (and the platelet count of previous babies) is unknown, cesarean section is not indicated when the maternal platelet count is >50,000. If the fetal platelet count is known, cesarean section is appropriate if the fetal count is <20,000. A maternal platelet count of >50,000 is considered sufficient to prevent complications from excessive maternal bleeding at vaginal delivery or cesarian section. Prophylactic platelet transfusions before delivery are appropriate in women with counts <10,000 who (1) have a planned cesarean section or (2) have epistaxis or other mucous membrane bleeding and are expected to deliver vaginally, but are unnecessary in women with platelet counts >30,000 and no bleeding symptoms.
Diagnosis
The neonatal platelet count should generally be measured for 3 to 4 days after birth. Brain imaging (eg, ultrasound) should be performed if the platelet count at birth is <20,000; brain imaging is also appropriate if the count is 20,000 to 50,000, even in the absence of neurologic abnormalities.
Treatment
In newborns without evidence of intracranial hemorrhage (ICH), treatment with IVIg is appropriate if the infant's platelet count is <20,000. Newborns with platelet counts of 20,000 to 50,000 do not necessarily require IVIg treatment. Newborns with counts >50,000 should not be treated with IVIg or glucocorticoids. Newborns with imaging evidence of ICH should be treated with combined glucocorticoid and IVIg therapy if the platelet count is <20,000; they should not be treated with glucocorticoids alone. Women with ITP should not be discouraged from breast feeding.
The ASH selected ITP because of the frequency with which it is encountered by hematologists and because of uncertainty regarding the relative effectiveness and safety of current diagnostic tests and treatments. Although there are no reliable epidemiologic data on the incidence of ITP, estimates are that 10 to 125 per 1,000,000 persons (children and adults) develop ITP each year.8 The goal of the panel was to issue explicitly developed recommendations, based as much as possible on published, scientific evidence, regarding the diagnosis and treatment of patients with known or suspected ITP.
The 15-member panel included 13 hematologists selected to represent the ASH membership. The hematologists included both university-affiliated physicians with research interests in ITP and private practitioners. Panel members represented both pediatric and adult medicine perspectives. The panel also included two members with expertise in clinical epidemiology and practice guideline methodology.
The panel defined ITP as isolated thrombocytopenia with no clinically apparent associated conditions or other causes of thrombocytopenia (eg, HIV infection, systemic lupus erythematosus, lymphoproliferative disorders, myelodysplasia, agammaglobulinemia or hypogammaglobulinemia, drug-induced thrombocytopenia, alloimmune thrombocytopenia, congenital/hereditary nonimmune thrombocytopenia). No specific criteria establish the diagnosis of ITP; the diagnosis relies on the exclusion of other causes of thrombocytopenia. For purposes of this review, the panel excluded from consideration patients with clinically apparent coexisting conditions that can cause immune thrombocytopenia (eg, systemic lupus erythematosus). Patients with isolated abnormalities on serologic tests (eg, antinuclear or antiphospholipid antibodies) but without a clinically evident disorder such as systemic lupus erythematosus were not excluded because positive serologic tests are frequently encountered in patients with typical ITP.9,10 However, the panel recognized that patients with thrombocytopenia and an associated autoimmune disease may have an illness comparable to ITP.
A computerized search of the MEDLINE database, performed in April 1994, sought English-language articles published between 1966 and 1994. Search terms (Medical Subject Headings) included: "THROMBOCYTOPENIA," "PLATELET COUNT," "AUTOIMMUNE THROMBOCYTOPENIC PURPURA," "COMPLETE BLOOD COUNT," "BONE MARROW EXAMINATION," "RETICULOCYTE COUNT," "ANTINUCLEAR ANTIBODY TEST," "IGG," "DIAGNOSIS (SH)," and "THERAPY (SH)." The database was also searched on the text word "ITP." The computerized search retrieved 581 articles. This initial reference list underwent substantial expansion after being supplemented with relevant articles from the files of panel members, publications from 1989 through 1995 retrieved with alternate search software ("Reference Update"), and cross-checking against the bibliographies of retrieved articles to identify additional publications (especially those published before 1966). Case reports, case series of less than five patients, review articles, and letters-to-the-editor without primary data were excluded from review. Statements in this report about the number of studies that have examined the efficacy of specific treatments and statements that "no published evidence is available" do not include case reports and other categories of inadmissible evidence.
Each article was evaluated independently by two panel members (J.N.G., G.E.R.) to assess scientific validity and verify results. Scientific validity was assessed using published guidelines.11-15 Literature on the clinical course of ITP was evaluated for the presence of an inception cohort of consecutive patients, an explicit referral pattern, complete follow-up, and use of objective outcome criteria. The term "inception cohort" refers to a group of patients identified at an early and uniform point in the course of their disease so that patients who die or completely recover are included with patients in whom the disease persists. Most of the ITP literature reviewed in this report pertains to therapy. The strength of the evidence for individual therapeutic approaches was assessed using the "level of evidence" criteria outlined in Table 1.13,14 Evidence tables in the Results section only present data from level I and level II studies.
| Level of Evidence | Study Design |
|---|---|
| I Strongest | Randomized trials with low false-positive and false-negative errors. |
| II | Randomized trials with high false-positive and false-negative errors. |
| III | Nonrandomized studies with concurrent control group. |
| IV | Nonrandomized studies with historical control group. |
| V Weakest | Case series without a control group. |
Data from refs 11, 12, and 14.
Most of the literature on the treatment of ITP consists of case series without a control group (level V). For those therapies for which only level V evidence is available, or for which no evidence is available, and for issues on diagnosis that have not been addressed by clinical studies, the opinion of the panel was assessed. Survey instruments were used to assess quantitatively the opinion and strength of consensus of the panel, and these data provide the basis for statements about opinion in the text and tables. The survey instruments were designed at panel meetings in which members were asked to identify the key diagnostic and treatment practices for which opinion would be assessed. The appropriateness of these practices was intentionally not discussed at the meeting to avoid influencing the responses by the opinions of more assertive panel members. A 41-page questionnaire addressing these practices was mailed to panel members in 1994 to be completed independently, without discussion with one another. The questionnaire, which included separate pediatric and adult sections, asked respondents to measure the necessity and appropriateness of diagnosis or treatment in over 1,300 clinical scenarios. In these surveys, "Necessary" was defined as a test or treatment that should be performed; "Appropriate" was defined as a test or treatment that may or may not be necessary, but performing it is not wrong; "Unnecessary" was defined as a test or treatment that need not be performed, but is not necessarily inappropriate; "Inappropriate" was defined as a test or treatment that should not be performed. Questions relating to adult patients were completed by 11 panel members, and questions relating to pediatric patients were completed by six respondents. A second, 25-page questionnaire was circulated in early 1995 to examine opinions regarding pregnancy and newborn care and to clarify opinions regarding issues identified in the 1994 survey. The 1995 survey examined over 600 issues and was completed by 13 panel members.
Using a modified RAND scoring system,16,17 the questionnaire asked panelists to quantify the strength of their opinion on a 1 to 9 scale; "9" represented strong agreement with the appropriateness/necessity of the practice and "1" represented strong disagreement. The mean response for each question provided an overall assessment of the panel's opinion regarding the necessity and appropriateness of specific practices. Panel votes are presented in this report only when there was agreement among the panel regarding the necessity or appropriateness of an intervention (mean panel score of 7.0 to 9.0) or agreement that the intervention is unnecessary or inappropriate (mean panel score of 1.0 to 3.0).
The strength of the panel's inter-observer agreement about the appropriateness/necessity of tests or treatments was graded using the standard deviations (SDs) for responses to each question (Table 2). Panel responses were classified as category A ("Complete or Almost Complete Unanimity"), for example, if the variance in panel member responses to a specific question was more than two SDs below the mean variance. Thus, a score of "1.5, A" signified strong agreement among the panel that the intervention is unnecessary/inappropriate, with most panelists assigning scores close to 1.0. A score of 7.5, D meant that, on average, the panel considered the intervention necessary/appropriate, but that wide variation in the responses of individual panel members was noted. These scores were arbitrarily considered as representing a consensus if the mean score was 3 or less or 7 or more.
| Score | Definition |
|---|---|
| Appropriateness of Necessity Scores* | |
| 1.0-3.0 | "Inappropriate" or "unnecessary" (depending on question). |
| 3.01-6.99 | Uncertain appropriateness or necessity. |
| 7.0-9.0 | "Appropriate" or "necessary" (depending on question). |
Consensus Codes[dagger] | |
| A | "Complete or almost complete unanimity" (panel variance more than 2 SD below the mean variance). |
| B | "Strong agreement" (panel variance 1 to 2 SD below the mean variance). |
| C | "Moderate agreement" (panel variance less than 1 SD below the mean variance). |
| D | "Moderate disagreement" (panel variance less than 1 SD above the mean variance). |
| E | "Strong disagreement" (panel variance greater than 1 SD above the mean variance). |
*Represents mean panel score for response to questions asking for ranking of appropriateness/necessity on a scale of "1" to "9," with "1" representing most "inappropriate/unnecessary" and "9" representing most "appropriate/necessary." Separate scores were obtained for appropriateness and necessity by asking separate, individually worded questions. "Necessary" = test should be performed, "Appropriate" = test may or may not be necessary, but performing it is not wrong, "Unnecessary" = test need not be performed (but is not necessarily inappropriate), "In appropriate" = test should not be performed.
[dagger]Strength of agreement among the panel members about appropriateness/necessity, ie, the variance of responses around the mean panel score.
FIGURE 1. Panel responses to the question, "Is it necessary/appropriate to order a bone marrow aspiration/biopsy to establish the diagnosis of ITP in all adult patients at presentation?" This question assumed that the history, physical examination, and initial blood counts with examination of the blood smear are compatible with the diagnosis of ITP and do not include atypical findings that are uncommon in ITP or suggest other etiologies. Data points ([bullet]) represent the responses of individual panel members (N = 11) on a scale of 1-9, with 1 reflecting complete disagreement and 9 reflecting complete agreement. Horizontal lines ([line]) represent the mean panel score. Letter codes, which describe the variance, are defined in Table 4.
![[ figure 2 image ]](/images/0041f2.gif)
FIGURE 2. Panel responses to a question regarding the appropriateness of offering no specific initial treatment for children presenting with ITP and symptoms of only minor purpura. Data points ([bullet]) represent the responses of individual panel members (N = 6) on a scale of 1-9, with 1 reflecting complete disagreement and 9 reflecting complete agreement. Horizontal lines ([line]) represent the mean panel score. Letter codes, which describe the variance, are defined in Table 4.
In almost all aspects of ITP level I evidence is lacking, and there are few level II, III, or IV studies to allow firm, evidence-based recommendations. In general, only level V evidence, or no studies, were available for making recommendations. Therefore, the panel issued recommendations based on opinion, indicating the mean panel score and variance to permit readers to judge the strength of the consensus. Although the sample sizes of voting members were small and some confidence intervals for panel votes were wide, the results can help readers assess the strength of opinion behind specific recommendations. The basis of recommendations is explicitly labeled in the text so that reader can appreciate which recommendations are based on evidence and which are based on opinion. The inherent weakness of opinion-based recommendations is acknowledged; these recommendations should not form the basis for definitive decisions on health care policy. Indications for which the panel could not reach consensus (scores of 3.1 to 6.9) are generally not listed in the text; thus, recommendations frequently address only the "extremes" of inappropriate and appropriate practice and do not comment on intermediate clinical scenarios that may be common. The fact that the panel did not reach consensus regarding these indications does not necessarily signal the appropriateness or inappropriateness of clinicians' decisions to administer tests or treatments in these settings.
This practice guideline describes a range of approaches to the diagnosis and management of ITP. Its recommendations are not intended to serve as inflexible rules, and they are not inclusive of all proper methods of care or other methods of care that may achieve similar results. Adherence to the guideline will not ensure a successful outcome in every case. The ultimate judgment regarding the care of a particular patient should be made by the physician in light of the clinical data and circumstances presented by the patient and the diagnostic and treatment options available.
Before the final panel meeting, the report was independently reviewed by eight private practice and university-based hematologists with expertise in adult and/or pediatric ITP (Drs Neil Abramson, Jacksonville, FL; Barbara Alving, Washington, DC; Diana Beardsley, New Haven, CT; Jack Levin, San Francisco, CA; Joan Parkhurst, Oklahoma City, OK; Graham Pineo, Calgary, Alberta, Canada; Gary Ratkin, St Louis, MO; Samuel Silver, Ann Arbor, MI).
Clinical Course
A critical issue in caring for children with ITP is determining which patients require treatment, either at the time of diagnosis or in the management of chronic disease. To make informed management decisions, prognostic information is needed to predict (1) how platelet counts will respond, with or without therapy, (2) likely health outcomes without treatment, and (3) whether early response to intervention reduces the incidence of adverse outcomes.
Evidence. There have been no large prospective studies which assembled an inception cohort of children with ITP and followed the clinical course of untreated patients to document the incidence of clinically important bleeding and mortality. Data on the clinical course of untreated ITP in children come from two types of evidence: (1) case series in which selected children with ITP were not treated and were followed to document the incidence of spontaneous remission, clinically important bleeding, and mortality, and (2) data from untreated control groups in relatively small, brief randomized clinical trials evaluating the effectiveness of alternative treatments. The case-series data are summarized in Table 3.18-29
The best data on untreated disease come from two series in which about 75% of patients were not treated initially.22,25 Most patients had platelet counts <50,000 at presentation, and in one of the reports22 most had platelet counts <20,000. Of the 221 untreated children, 2 (0.9%) had fatal bleeding associated with the acute presentation, and 191 (87%) had a complete remission from ITP. The platelet count normalized in 2 to 8 weeks, with one half to two thirds of the patients recovering within 4 weeks.22,25
There are limitations to the inferences that can be drawn from these data. First, 25% of the inception cohort in each series were selected for treatment. If clinicians treated patients with the most serious clinical presentations, then the clinical course in the remaining patients may underestimate the frequency of important bleeding and mortality, and may overestimate the rate of spontaneous remission. However, even if it is assumed that patients selected for treatment would not have had a spontaneous remission, then the "least frequency" estimate of the probability of complete spontaneous remission is 191 of 298 (64%). Second, 7% to 14% of patients were lost to follow-up, some of whom may have suffered a relapse of ITP, with bleeding complications or death. In the remaining series in Table 3, the children selected to be followed without treatment represented only 10% to 56% of the inception cohort; patients with more severe clinical presentations were generally treated. Thus, the untreated patients in Table 3 may represent a select population with mild to moderate symptoms who may have a more favorable prognosis than the average child with ITP.
| Hemorrhagic Complications[double dagger] | Patients With Persistant Thrombocytopenia§ | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Author | Location | Years | Patients (no.) | Patients Responding With No Therapy* | Patients in Remission at 6 mo[dagger] | ICH | Fatal ICH | Other Deaths | Patients in Remission At Last Follow-up[dagger] | No. | Spontaneous Recovery | Deaths From Hemorrhage |
| Komrower and Watson18 | UK | 1948-1953 | 43 | 18/24 | 25|| | 4 | 3 | 3¶ | 31 | 9 | 2 | 0 |
| Choi and McClure19 | Canada | 1950-1964 | 239 | 20/25 | 105/161 | 1 | 1 | 0 | 120/128 | 18 | 5 | 0 |
| Walker and Walker20 | UK | 1950-1980 | 177 | 51/63 | 138 | 1 | 0 | 1# | 162 | 15 | 5 | 0 |
| Ramos et al21 | US | 1952-1977 | 150 | -- | 135 | 0 | 0 | 0 | 143 | 13 | 6 | 0 |
| Lusher and Zuelzer22 | US | 1956-1964 | 146 | 101/109 | 129/142** | 1 | 1 | 0 | 135/142 | -- | -- | -- |
| Simons et al23 | US | 1956-1973 | 84 | 19/20 | 50 | 1 | 2 | 0 | 58 | -- | -- | -- |
| Benham and Taft14 | Australia | 1958-1966 | 132 | 13/15 | 97** | 2 | 0 | 0 | 116 | 26 | 10 | 0 |
| Lamm and Lovric25 | Australia | -- | 152 | 91/112 | 116 | 1 | 1 | 0 | 147 | 29 | 23 | 1 |
| den Ottolander et al26 | Netherlands | -- | 77 | 23/35 | 38/75 | 1 | 1 | 0 | 60 | -- | -- | -- |
| Hoyle et al17 | UK | 1962-1982 | 136 | 35/41 | 97/132 | 1 | 1 | 0 | 117/132 | 16 | 9 | 0 |
| Zaki et al28 | Kuwait | 1981-1986 | 60 | 18/23 | 41 | 0 | 0 | 0 | 44 | 16 | 2 | 0 |
| Robb and Tiedeman29 | Australia | 1968-1987 | 297 | -- | 236/289 | 3 | 3 | 0 | 163 | 37 | 4 | 2 |
| Summary | 1,693 | 389/467 | 1,207/1,597 | 16 | 13 | 4 | 1,396/1,574 | 179 | 66 (37%) | 3 | ||
| (83%) | (76%) | (0.9%) | (0.7%) | (0.2%) | (89%) | (2%) | ||||||
Abbreviation: ICH, intracranial hemorrhage.
In these series, the upper age limit ranged from 12 to 16 years old.
*The numerator is the number of patients with a complete response, typically defined as a normal platelet count without relapse; the denominator is the total number of patients managed without specific initial therapy. The response rate for untreated patients is greater than the overall response at 6 mo because of selection of patients with good prognostic features for no treatment.
[dagger]A different denominator from the original number of patients indicates that some patients were not followed long enough to be included in the estimate.
[double dagger]Note that 6 of the total 17 deaths occurred in the first (and smallest) study of patients before 1953 Komrower and Watson.18 Omitting this study, the frequency of fatal intracranial hemorrhage is 9/1,650 (0.5%) and the overall mortality is 10/1,650 (0.6%). Eight of the fatal ICH occurred acutely, within 5 wk of diagnosis; the other 5 occurred between 1 to 2 years after diagnosis. All 4 of the other hemorrhagic deaths occurred acutely within 5 wk.
§Persistant thrombocytopenia was defined as >6 mo after diagnosis, except for Lusher and Zuelzer22 and Benham and Taft,34 who defined persistant thrombocytopenia as >12 mo after diagnosis.
||Includes 7 patients who had splenectomy <6 mo after diagnosis. No patients were treated with glucocorticoids in this study.
¶Described only as "uncontrollable bleeding."
#Death due to hemorrhage (not ICH) and presumed sepsis 2 wk after diagnosis.
**The distinction of acute v chronic ITP is determined at 12 mo rather than 6 mo in these two reports, and these are the data for remission at 12 mo.
Only limited observational data are available regarding the complications of intracranial hemorrhage. In a review of 14 children with intracranial hemorrhage, Woerner et al33 reported that 4 died and 2 others may have had neurologic sequelae. Of the 30 children with intracranial hemorrhage described in this report33 and the references in Table 3, 12 (40%) occurred within the first 12 days after diagnosis, including 2 patients with a history of head trauma. The intracranial hemorrhages in the other 18 patients occurred between 1 month and 5 years after diagnosis, typically after glucocorticoids and splenectomy failed to induce a remission. At least 24 of these 30 patients were reported before 1981, when IVIg therapy was initially described.34
Unlike ITP in adults, persistent thrombocytopenia is uncommon in children. In the 12 case series in Table 3, 10 defined chronic disease as 6 months of thrombocytopenia and 2 studies defined it as 12 months.22,24 In the 12 series, ITP resolved in 1,207 (76%) of the 1,597 children who were followed for these time periods. Features of the presenting illness that were associated with an increased risk of chronic persistent thrombocytopenia included a history of purpura for more than 2 to 4 weeks before diagnosis,23-25,29 female sex,21-23,28 age over 10 years,23,28 and a higher platelet count at presentation.22 The fate of children with chronic ITP is uncertain, although about one third appear to have spontaneous remissions several months to many years after diagnosis.35,36
Diagnosis
Few clinical studies have evaluated the sensitivity and specificity of the diagnostic tests used for children with suspected ITP, because in the absence of a "gold-standard" test for ITP the diagnosis is based only on the presence of thrombocytopenia with no other apparent cause. Other etiologies are uncommon: in a study of 127 consecutive children with suspected ITP who had bone marrow aspirations, other causes of thrombocytopenia were identified in only 5 (4%) children, all of whom had atypical presenting features.37 Therefore, in the absence of additional scientific evidence on the accuracy or effectiveness of diagnostic tests for ITP, the panel's recommendations regarding the history, physical examination, laboratory tests, and special procedures are based entirely on opinion.
Directed history and physical examination. By definition, the diagnosis of ITP cannot be made without a compatible history and physical examination that excludes other causes of thrombocytopenia. The most likely alternate causes vary with the age of the child. For example, many case series exclude infants less than 4 to 6 months old in part because neonatal alloimmune or autoimmune thrombocytopenia cannot be ruled out at this age. The most important elements of the history and physical examination identified by the panel are presented in Table 4. The maternal and birth history are especially relevant when evaluating infants. The presence of congenital anomalies in the patient or family members may be a clue for congenital thrombocytopenia, an important consideration in children with persistent thrombocytopenia.38
| History Bleeding symptoms Type of bleeding Severity of bleeding Duration of bleeding Hemostasis with prior invasive procedures Systemic symptoms, especially of recent (within 6 wk) viral illness or exposure to viruses such as varicella, or recurrent infections suggesting immunodeficiency; symptoms of an autoimmune disorder Recent live virus immunization Medications, including heparin, quinidine/quinine, and sulfonamides, which may cause thrombocytopenia, and aspirin, which may exacerbate bleeding Risk factors for HIV infection, including maternal HIV status Family history of thrombocytopenia or hematologic disorder In an infant <6 mo old, include perinatal and maternal history Comorbid conditions, which may increase the risk of bleeding Lifestyle, including vigorous and potentially traumatic activities Physical examination Bleeding signs Type of bleeding (including retinal hemorrhages) Severity of bleeding Liver, spleen, and lymph nodes Evidence for infection Presence of dysmorphic features suggestive of congenital disorder, including skeletal anomalies, auditory acuity Specific Congenital Syndromes to Exclude Fanconi syndrome Thrombocytopenia-absent radius Wiskott-Aldrich syndrome Alport syndrome (and its variants) Bernard-Soulier syndrome May-Hegglin anomaly Gray platelet syndrome |
Complete blood count with examination of the peripheral blood smear. A complete blood count and an examination of the peripheral blood smear are essential in ITP. The principal features of the examination of the blood smear that were identified by the panel are the same for children and adults (Table 5). Although most patients with ITP present with platelet-related bleeding, the condition may be first detected by the incidental discovery of thrombocytopenia on routine blood counts. Because ITP is defined by a low platelet count without another apparent cause, the clinician must know the normal values for the laboratory. Aside from thrombocytopenia, the blood counts of patients with ITP should be normal or otherwise readily explained by a coincident disorder (eg, thalassemia minor). The presence of platelet clumps suggests pseudothrombocytopenia (see Adult section, below). Anemia, if present, may be caused by bleeding or iron deficiency resulting from chronic thrombocytopenia, but this is uncommon in children. White blood cell morphology should be normal, although some children with ITP may have atypical lymphocytes or eosinophilia.22,25,40
Other laboratory data. Recommendations regarding other laboratory tests were derived from opinion by a questionnaire completed by six panel members (see text above). The recommendations assume that the history, physical examination, and initial blood counts and smear are compatible with the diagnosis of ITP and do not include atypical findings that are uncommon in ITP or suggest other disease etiologies. For example, a direct antiglobulin test, which the panel did not recommend for patients with a typical presentation of ITP, may be appropriate if the peripheral smear shows red blood cell polychromatophilia with poikilocytosis and spherocytes. Indications for which the panel did not reach consensus (score of 3.1-6.9) are not listed in the text but are summarized in Table 6.
| Consistent with the diagnosis of ITP 1. Thrombocytopenia. Platelets are normal in size or may appear larger than normal, but consistently giant platelets (approaching the size of red blood cells) should be absent. 2. Normal red blood cell morphology. 3. Normal white blood morphology. Not consistent with the diagnosis of ITP 1. Predominant giant platelets. 2. Red blood cell poikilocytosis, schistocytes, polychromatophilia (unless response to bleeding), macrocytes, nucleated red blood cells. 3. Leukocytosis or leukopenia, with immature or abnormal cells (although atypical lymphocytes and eosinophilia may occur in children with ITP). |
| Tests of Uncertain Appropriateness (meean panel scores 3.01-6.99) | ||
|---|---|---|
| Indications | Children | Adults |
| To establish the diagnosis in all patients at presentation | ANA (a), direct antiglobulin (a), HIV (a), bone marrow (a), platelet antigen-specific antibody, mean platelet volume, reticulocyte count | ANA, direct antiglobulin, lupus anticoagulant/APLA (a), chemistry profile, coagulation studies, chest x-ray (a), HIV, bone marrow, mean platelet volume, reticulocyte count (a), thyroid function, urinalysis (a) |
| To establish the diagnosis before splenectomy | ANA, direct antiglobulin, lupus anticoagulant/APLA, abdominal CT/ultrasound (a), serum immunoglobulins (a), platelet antigen-specific antibody | ANA, direct antiglobulin, lupus anticoagulant/APLA, serum complement, abdominal CT/ultrasound, bone marrow (n), chest x-ray, platelet antigen-specific antibody, platelet survival, thyroid function |
| To establish the diagnosis in patients who fail to respond to primary treatment (eg, glucocorticoid) and splenectomy | ANA, direct antiglobulin, lupus anticoagulant/APLA (a), abdominal CT/ultrasound (a), serum immunoglobulins, platelet-associated IgG, platelet antigen-specific antibody, platelet survival, thyroid function | ANA, direct antiglobulin, lupus anticoagulant/APLA, serum complement, abdominal CT/ultrasound, chest x-ray, platelet-associated IgG, platelet antigen-specific antibody, platelet survival, thyroid function |
| Other tests of uncertain appropriateness: ANA, to establish the diagnosis in pregnant and nonpregnant women; lupus anticoagulant/APLA, to establish the diagnosis in women at presentation (a) and pregnant women; abdominal CT/ultrasound, for suspected splenomegaly on physical examination in children (n) and adults; HIV, in adult patients with no risk factors for HIV infection; thyroid function, to rule out thyroid disease in all patients at presentation (a) and before elective splenectomy (n). | ||
Tests that the panel considered unnecessary/inappropriate for routine evaluation of all patients (mean scores, 1.0-3.0) are listed in Table 7. Tests which the panel considered appropriate/unnecessary (mean scores, 7.0-9.0) are described in the text. Listed here are the specific clinical scenarios for which the panel assigned a mean panel score of 3.01-6.99, not reaching consensus on whether the test is appropriate/necessary. (a) = appropriateness uncertain, but testing is not necessary, (n) = necessity uncertain, but testing is appropriate.
| Unnecessary, But May be Appropriate (Mean Panel Score for Necessity, Consensus Code) | Unnecessary and Inappropriate (Mean Panel Score for Appropriateness, Consensus Code) |
|---|---|
| Children | |
| Platelet antigen-specific antibody (1.3, B) Mean platelet volume (1.8, B) Bone marrow (2.0, B) HIV test (2.0, B) Antinuclear antibody (2.0, C) Direct antiglobulin test (2.5, C) | Platelet survival study (1.0, A) Chest x-ray (1.0, A) Abdominal CT or ultrasound (1.0, A) Coagulation studies (1.2, A) Serum complement level (1.7, C) Lupus anticoagulant/APLA (2.0, C) Bleeding time (2.0, C) Platelet-associated IgG (2.2, C) Thyroid function tests (2.3, D) Serum chemistry profile* (2.7, D) Urinalysis (2.8, D) Serum immunoglobulin level (3.0, D) |
| Adults | |
| Lupus anticoagulant/APLA (1.8, B) Platelet antigen-specific antibody (1.7, C) Direct antiglobulin test (2.1, B) Chest x-ray (2.1, C) Mean platelet volume (2.4, D) Reticulocyte count (2.6, D) Urinalysis (2.6, C) Thyroid function tests (2.9, D) | Bleeding time (1.7, C) Platelet survival study (2.4, C) Serum complement (2.6, D) Abdominal CT/ultrasound (2.6, D) Platelet-associated IgG assay (3.0, D) |
| Pregnant women | |
| Platelet antibody (1.4, B) Serum fibrin D-dimer (2.4, D) PT/PTT (2.6, C) Lupus anticoagulant/APLA (2.9, D) Uric acid (2.9, D) | None |
Tests of uncertain appropriateness/necessity are listed in Table 6.
*Including LDH, BUN, creatinine, and liver function tests.
(1) To establish the diagnosis before splenectomy: Tests that the panel considered unnecessary for this purpose included platelet antigen-specific antibody assay (2.0, C), abdominal CT scan or ultrasound (2.0, C), and serum Ig level (3.0, D). Tests that the panel considered unnecessary and inappropriate included (scores are for appropriateness): serum complement level (1.8, C), chest x-ray (2.5, C), thyroid function studies (2.5, D), platelet survival study (2.5, D), and platelet-associated IgG assay (2.7, C).
(2) To establish the diagnosis in patients who have failed to respond to glucocorticoid therapy, IVIg, and splenectomy: Tests that the panel considered unnecessary for this purpose included platelet-associated IgG assay (1.2, A), platelet antigen-specific antibody assay (2.0, C), abdominal CT scan or ultrasound (2.8, C), platelet survival study (2.8, D), lupus anticoagulant or antiphospholipid antibody (3.0, C), and thyroid function testing (3.0, D). Tests that the panel considered unnecessary and inappropriate included chest x-ray (2.2, C) and serum complement level (2.5, D) (scores are for appropriateness).
Treatment
Essentially all evidence regarding the efficacy of treatment of ITP is indirect, inferred by measuring a surrogate outcome, platelet count, rather than a health outcome such as bleeding or mortality. The panel accepted the platelet count as a useful surrogate outcome, because numerous studies of thrombocytopenia show a correlation between platelet counts and clinically important bleeding.41-44 The limitations of this assumption are highlighted by several factors. First, the association between platelet count and clinically important bleeding has been demonstrated principally in patients with thrombocytopenia with conditions other than ITP. Second, the platelet count may not reflect beneficial or potential harmful effects of treatment that are independent of an effect on platelets.
Even an effect on the platelet count is difficult to validate convincingly based on currently available data, because evidence of treatment efficacy consists largely of reports from uncontrolled case series (level V evidence, the weakest category, Table 1). Without an internal control group for comparison, such studies are unable to clarify whether the favorable results were due to the treatment under study or would have occurred even without treatment (or with another treatment). Although the potential adverse effects of certain treatments for ITP are known, a valid framework for the systematic comparison of benefits and harms is lacking, making it difficult to determine when a treatment results in more harm than good. Given these gaps in the evidence, treatment recommendations in this report rely largely on opinion.
Hospitalization
Evidence. There have been no studies to evaluate the effectiveness of hospitalizing children with ITP.
Recommendations. In the absence of evidence, the opinion of the panel was that hospitalization is appropriate for a child with severe, life-threatening bleeding, regardless of the platelet count (9.0, A), and for a child with a platelet count of <20,000 and mucous membrane bleeding that may require clinical intervention (8.2, C). Hospitalization is inappropriate for a child with a platelet count of 20,000 to 30,000 who is asymptomatic (2.8, D) or for a child with a platelet count >30,000 who is either asymptomatic or has only minor purpura (1.0 to 1.5, B) (Table 8). Indications for hospitalization under intermediate conditions are less clear. Hospitalization may also be appropriate for children with platelet counts <20,000 who may be inaccessible or noncompliant (8.2, B) or whose parents request hospitalization (7.0 to 7.4, B).
Emergency Treatment
Evidence. Although there are no published data on the efficacy of different treatments for the management of children with urgent, life-threatening bleeding, evidence regarding the morbidity and mortality associated with severe hemorrhage from thrombocytopenia is extensive.18,33,45
Recommendations. The opinion of the panel was that the serious consequences of severe, life-threatening bleeding justify the use of several regimens. Assuming that conventional critical care measures are already underway, there was strong agreement (9.0, A) among panel members that appropriate interventions include platelet transfusions, high-dose parenteral glucocorticoid (eg, 30 mg/kg methylprednisolone daily for 3 days), and IVIg, either alone or in combination with glucocorticoids. See more detailed discussion of these treatments below.
Observation (No Specific Initial Treatment)
Evidence. Evidence about the outcomes of not treating ITP is derived from studies of the clinical course of untreated cases (see "Clinical Course"). Two level I studies31,32 and many level V studies suggest that 30% to 70% of children recover from severe thrombocytopenia, achieving platelet counts of 50,000 to 100,000 within 3 weeks without specific treatment. Level I evidence indicates that platelet count recovery is more rapid with either IVIg or glucocorticoid therapy than with no specific treatment30-32,46-48 (see Table 9), but it remains uncertain if this effect on platelet count influences morbidity or mortality. Moreover, the data come from children with severe thrombocytopenia at presentation; no comparable studies have been performed on children with less severe thrombocytopenia. Although it may seem intuitive that less severe thrombocytopenia would provide an even weaker indication for intervention, there is some evidence that children with higher platelet counts may have a greater risk of chronic, persistent thrombocytopenia.22,49 However, there is no evidence that the risk of developing chronic ITP is lowered by treatment.
Recommendations. Current evidence is inadequate to recommend which groups of children with ITP can be safely managed without therapy. The opinion of the panel was that it was appropriate to withhold specific treatment for asymptomatic children with platelet counts of 20,000 to 30,000 (7.0, C), and more strongly for children with platelet counts >30,000 who are asymptomatic or who have only minor purpura (8.3 to 9.0, A-C) (Table 8, Fig 2). The panel acknowledges that some pediatric hematologists who were not represented on the panel do not recommend specific treatment for children presenting with severe thrombocytopenia (platelet counts <20,000); these hematologists believe that careful observation is sufficient and preferable. The panel believed that withholding specific treatment was inappropriate for children with a platelet count <50,000 who present with significant mucous membrane bleeding (1.0, A for platelet count <30,000; 2.0, B for platelet count of 30,000 to 50,000). Not treating children with severe life-threatening bleeding was considered inappropriate (1.0, A) at any platelet count. Although the panel considered it appropriate (7.7 to 8.7, B-C) to withhold treatment at the parents' request for children with platelet counts >30,000, it was considered inappropriate (2.8, D) to do so if the platelet count was <10,000.
Glucocorticoid Therapy
Evidence. Level I and II studies of the efficacy of glucocorticoids are summarized in Table 9. Randomized clinical trials (level I and II) have shown that glucocorticoids increase the platelet count more quickly than when no specific treatment is administered. For example, the median time to achieve a platelet count of >50,000 was 4 days with prednisone treatment (4 mg/kg/d for 7 days, then tapered) versus 16 days in untreated children.32 The efficacy of glucocorticoids has only been demonstrated in terms of platelet recovery time and not in terms of morbidity or mortality. All relevant data come from children with acute ITP of recent onset. There have been no randomized controlled studies of glucocorticoid treatment in children with chronic thrombocytopenia.
Three general cateories of regimens for glucocorticoids have been evaluated: (1) 1 to 2 mg/kg/d or 60 mg/m2/d of oral prednisone for approximately 21 days27,30,31,46,47,50-52 (level I, II, and V evidence); (2) 4 mg/kg/d of oral prednisone for 7 days then tapered32,48 (level I evidence); and (3) 10 to 30 mg/kg/d of oral or IV methylprednisolone for several days52-57 (level II, III, V evidence). Because ITP in children is typically self-limited, the duration of treatment was limited in many studies to 21 days. Initial reports used 2 mg/kg/d, comparable to the adult dose, but more recent studies have used 4 mg/kg/d, which is well-tolerated because the duration of treatment is short. In recent studies,32,48 the dose of 4 mg/kg/d was continued for only 7 days and the dose was then tapered and discontinued on day 21. Several studies using very high doses (10 to 50 mg/kg/d of methylprednisolone for 3 to 7 days) suggest that platelet count recovery is as rapid as that seen with IVIg,52,55-57 but similar findings have also been reported with a dose of 4 mg/kg/d for the first 7 days.32,48
The potential adverse effects of glucocorticoid therapy include all of the signs and symptoms of hypercortisolism in Cushing syndrome, including facial swelling, weight gain, hyperglycemia, hypertension, cataracts, and behavioral abnormalities.58 The toxicities of glucocorticoids are dose and duration dependent. Glucocorticoid therapy may increase the risk of growth retardation in children.59
Recommendations. There is level I evidence that children with acute ITP and severe thrombocytopenia experience more rapid recovery of platelets if given glucocorticoids, but it is unknown if this influences morbidity or mortality. There is also inadequate evidence of the efficacy of glucocorticoids in other patient categories (less severe thrombocytopenia, chronic ITP) to develop definitive recommendations based on the data. The opinion of the panel was that in patients with platelet counts <50,000 it is appropriate (7.0 to 8.4, B-D) to treat severe, life-threatening bleeding initially with high-dose oral (eg, prednisone, 4 to 8 mg/kg/d) or parenteral (eg, methylprednisolone, 30 mg/kg/d) glucocorticoid. High doses of oral glucocorticoid are also appropriate as initial therapy for children with mucous membrane bleeding and platelet counts <20,000 (7.6, C) and for those with minor purpura and platelet counts <10,000 (7.0, D). The panel considered glucocorticoids inappropriate (1.0 to 2.2, A-C) as initial therapy for children with platelet counts >30,000 and no symptoms or only minor purpura (Table 8). Treatment for the sole purpose of determining responsiveness or confirming the diagnosis was considered inappropriate for high-dose parenteral glucocorticoids in patients with platelet counts >10,000, for conventional-dose oral glucocorticoids in patients with platelet counts >20,000, and for high-dose oral glucocorticoids in patients with platelet counts >30,000 (1.0 to 2.8, A-D). When oral glucocorticoids are used, level I studies suggest that the regimens of 1.5 or 2 mg/kg/d for 14 to 21 days,30,46,47 60 mg/m2/d for 21 days,31 or 4 mg/kg/d for 7 days, followed by a tapering dose until day 21,32,48 are more effective than no treatment. These regimens have not been compared with each other, and some may be more effective than others in rapidly reaching a platelet count that may reduce the risk of serious hemorrhage.
| Treatment Options | |||
|---|---|---|---|
| Platelet Count <20,000 | Appropriate (mean panel scores 7-9) | Appropriateness Uncertain (mean panel scores, 3.1-6.9) | Inappropriate (mean panel scores, 1-3) |
| Asymptomatic | No treatment*, hospitalization, conventional-dose oral glucocorticoid,[dagger] high-dose oral glucocorticoid,[double dagger] high-dose parenteral glucocorticoid,§ IVIg (1 g/kg × 1 d), IVIg (total dose of 2 g/kg given over 2-5 d), anti-D|| | ||
| Minor purpura | IVIg (1 g/kg × 1 d), (7.2, D)[dagger] High-dose oral glucocorticoid, (7.0, D)¶ | Hospitalization, conventional-dose oral glucocorticoid, high-dose parenteral glucocorticoid, IVIg (total dose of 2 g/kg given over 2-5 d), anti-D | No treatment (2.5 D)Ÿ |
| Mucous membrane bleeding that may require clinical intervention | IVIg (1 g/kg × 1 d) (8.3, B) Hospitalization, (8.2, C) IVIg (total dose of 2 g/kg given over 2-5 d), (7.8, B) High-dose oral glucocorticoid, (7.6, C) | Conventional-dose oral glucocorticoid, high-dose parenteral glucocorticoid, anti-D | No treatment (1.0 A) |
| Severe, life threatening bleeding | Hospitalization (9.0, A) IVIg (1 g/kg × 1 d) (8.8, A-B) High-dose parenteral glucocorticoid (8.0-8.4, B-C) IVIg (total dose of 2 g/kg given over 2-5 d), (7.8, C) High-dose oral glucocorticoid (7.0-7.4, C-D) | Conventional-dose oral glucocorticoid therapy, anti-D | No treatment (1.0, A) |
| Platelet Count 20-30 × 103 | Appropriate (mean panel scores 7-9) | Appropriateness Uncertain (mean panel scores, 3.1-6.9) | Inappropriate (mean panel scores, 1-3) |
| Asymptomatic | No treatment (7.0, C) | Conventional-dose oral glucocorticoid, high-dose oral glucocorticoid, IVIg (1 g/kg × 1 d), IVIg (total dose of 2 g/kg given over 2-5 d), anti-D | High-dose parenteral glucocorticoid, (2.6, C) Hospitalization (2.8, D) |
| Minor purpura | No treatment, hospitalization, conventional-dose oral glucocorticoid, high-dose oral glucocorticoid, IVIg (1 g/kg × 1 d), IVIg (total dose of 2 g/kg given over 2-5 d), anti-D | High-dose parenteral glucocorticoid, (2.6, C) | |
| Mucous membrane bleeding that may require clinical intervention | Hospitalization, conventional-dose oral glucocorticoid, high-dose oral glucocorticoid, high-dose parenteral glucocorticoid, IVIg (1 g/kg × 1 d, IVIg (total dose of 2 g/kg given over 2-5 d), anti-D | No treatment (1.0, A) | |
| Severe, life-threatening bleeding | Hospitalization (9.0 A) IVIg (1 g/kg × 1 d) (8.5, B) IVIg (total dose of 2 g/kg given over 2-5 d) (8.2, C) High-dose parenteral glucocorticoid (7.6, C) High-dose oral glucocorticoid (7.4, C) | Conventional-dose oral glucocorticoid, anti-D | No treatment (1.0, A) |
| Platelet count 30-50 × 104 | Appropriate (mean panel scores 7-9) | Appropriateness Uncertain (mean panel scores, 3.1-6.9) | Inappropriate (mean panel scores, 1-3) |
| Platelet Count 30-50 × 104 | Appropriate (mean panel scores 7-9) | Appropriateness Uncertain (mean panel scores, 3.1-6.9) | Inappropriate (mean panel scores, 1-3) |
| Asymptomatic | No treatment (9.0, A) | IVIg (total dose of 2 g/kg given over 2-5 d), (1.0, A) IVIg (1 g/kg × 1 d) (1.2, A) Anti-D (1.2, A) High-dose parenteral glucocorticoid (1.2, B) Hospitalization (1.5, B) High-dose oral glucocorticoid (2.0, C) Conventional-dose oral glucocorticoid (2.0, C) | |
| Minor purpura | No treatment (8.3, C) | IVIg (total dose of 2 g/kg given over 2-5 d) (1.0, A) IVIg (1 g/kg × d) (1.2, A) Anti-D (1.2, A) High-dose parenteral glucocorticoid (1.2, B) Hospitalization (1.5, B) High-dose oral glucocorticoid (2.2, C) Conventional-dose oral glucocorticoid (2.2, C) | |
| Mucous membrane bleeding that may require clinical intervention | Hospitalization, conventional-dose oral glucocorticoid, high-dose oral glucocorticoid, IVIg (1 g/kg × 1 d), IVIg (total dose of 2 g/kg given over 2-5 d) | No treatment (2.0, B) High-dose parenteral glucocorticoid (2.8, D) Anti-D (3.0, D) | |
| Severe, life-threatening bleeding | Hospitalization (9.0, A) IVIg (1 g/kg × 1 d) (8.0, C) High-dose oral glucocorticoid (7.4, C) IVIg (total dose of 2 g/kg given over 2-5 d) (7.3, D) High-dose parenteral glucocorticoid (7.0, D) | Conventional-dose oral glucocorticoid | No treatment (1.0, A) Anti-D (3.0, D) |
"Appropriate" and "Not appropriate" = mean panel score of 7.0-9.0 or 1.0-3.0, respectively. "Appropriate" = treatment may or may not be necessary, but performing it is not wrong, "Inappropriate" = treatment should not be performed. Mean panel score is graded on a scale of "1" to "9" with "1" representing low appropriateness and "9" representing high appropriateness. Letter codes following panel scores reflect strength of agreement, the panel consensus (defined by standard deviation) around the mean panel score. "A" = complete or virtual unanimity, "B" = strong agreement, "C" = moderate agreement, "D" = moderate disagreement, "E" = strong disagreement (see Table 4).
*"No treatment" implies careful observation. In patients with major risk factors for bleeding (eg, elevated blood pressure, ulcer disease, vigorous lifestyle), not treating is considered inappropriate in all patients if the platelet count is 20-30 × 105 (2.3 C), 10-20 × 105 (1.3, B) or <10 × 105 (1.0, A). Not treating patients less than 3 years of age is also considered inappropriate if the platelet count is 10-20 × 105 (1.6, B) or less than 10 × 105 (2.4, B).
[dagger]Eg, 1-2 mg/kg/d of prednisone.
[double dagger]Eg, 4-8 mg/kg/d of prednisone.
§Eq, 30 mg/kg/d of methylprednisolone.
||Anti-D given intravenously.
¶These recommendations were made only for patients with platelet counts <10,000.
IVIg
Evidence. Clinical trials of IVIg therapy for ITP are summarized in Table 9. One level I study has shown that initial IVIg treatment of children with acute ITP increases the platelet count more rapidly than no specific treatment and than glucocorticoid therapy.32 Five level V studies34,60-63 suggest that IVIg will increase the platelet count substantially in a majority of patients, although some do not respond. Less than 10% of patients with chronic ITP have sustained, normal platelet counts without further treatment; in others thrombocytopenia recurs in several weeks to several months. No controlled data clarify whether these occasional prolonged responses without further treatment are different from those that would be observed in untreated children. Repeated treatments with IVIg may sustain platelet counts at a level of >20,000 to 30,000 and be useful to avoid splenectomy. For both acute and chronic ITP, there is no evidence that treatment with IVIg diminishes mortality or morbidity.
The first reported IVIg regimen was 0.4 g/kg daily for 5 consecutive days. Subsequent studies suggested that 1 g/kg for 1 day64 or 0.4 g/kg/d for 2 days50 may be sufficient in most responding patients. Recently, a randomized trial showed that a single dose of 0.8 g/kg achieves the same results as the former regimen with less cost and possibly fewer side effects.48
| Study Population | Outcome | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Authors | N | Age | Follow-up | Randomized Treatment Arms | Outcome Measure | Platelet Count | Bleeding Symptoms | Adverse Effects | Deaths | Evidence Level |
| McWilliams and Maurer46 | 27 | 6 yr (mean) | NR | Prednisone (2 mg/kg/d × 21 d) No treatment | Median time to platelet count of 150K | 21 d 60 d (P <= .03) | NR* NR | NR* NR | 0 0 | I |
| Sartorius31 | 93 | 6 mo-16 yr | >6 mo | Prednisone (60 mg/m2/d × 21 d, then tapered) Placebo | Proportion with platelet count >30K and >100K, and with negative Rumpel-Leede test | Prednisone > placebo (P < .01) | Prednisone < placebo (by Rumpel-Leede test (P < .01) | NR | 0 0 | I |
| Buchanan and Holtkamp30 | 27 | <11 yr | 28 d | Prednisone (2 mg/kg/d × 14 d, then taper to d 21) Placebo | Platelet count, bleeding time, clinical bleeding score at d 0-28 | Prednisone > placebo (P < .05) only at d 7 | Prednisone < placebo (P < .05) only at d 7 (bleeding time and clinical score) | Increased appetite, weight gain | 0 0 | I |
| Imbach et al50 | 94 | <16 yr | 1 yr | Prednisone (60 mg/m2/d × 21 d, follow-up protocol for poor response/remissions) | Proportion with platelet count >100K | 77% | NR | 77% with weight gain or acne | 0 | II |
| IVIg (0.4 g/kg/d × 5 d, follow-up protocol for poor response/remissions) | 83% (no difference) | NR | 22% with headache, fever, vomiting, vertigo | 1 died on day 6, excluded from analysis | ||||||
| Mazzucconi et al47 | 61 | 2-12 yr | >6 mo | Prednisone (0.5 mg/kg/d × 1 mo or until platelet normalization) | Proportion with platelet count >150K | 62% | NR | NR | NR | I |
| Prednisone (1.5 mg/kg/d × 1 mo or until platelet normalization) | 81% (P < .05) | NR | NR | NR | ||||||
| Belluci et al51 | 160 | <15 yr | >12 mo | Prednisone (0.25 mg/kg/d × 3 wk) Prednisone (1 mg/kg/d × 3 wk) | Proportion with platelet count >100K for >3 mo | 71% 77% (no difference) | NR NR | NR NR | 0 0 | II |
| Khalifa et al56 | 30 | 2 mo-15 yr | >6 mo | Methylprednisolone (IV, 10 mg/kg/d × 5 d) Prednisone (2 mg/kg/d × 4 wk) IVIg (0.4 g/kg/d × 5d) | Mean platelet count on days 1-14 | Methylprednisolone = IVIg > prednisone (P < .001) | NR NR NR | NR NR NR | 0 0 0 | I |
| Ozsoyle et al57 | 20 | 2 mo-11 yr | >6 mo | Methylprednisolone (po. 30 mg/kg/d × 3 d, then 20 mg/kg/d × 4 d) IVIg (0.4 g/kg/d × 5 d) | Proportion with platelet count >150K in 3 d, 6 mo | 60%, 90% (no difference) 60%, 75% | None NR | NR NR | 0 0 | II |
| Blanchette et al32 | 53 | 7 mo-14 yr | 180 d | Prednisone (4 mg/kg/d × 7 d, then tapered to d 21) | Median time to platelet count >20K, >50K | 2 d, 4 d | NR | Weight gain, behavioral change | 0 | I |
| IVIg (1 g/kg/d × 2 d) No therapy | 1 d, 2 d (P < .01 v prednisone) 4 d, 16 d (P < 0.1 v either treatment) | NR | 75% with nausea, vomiting, headache, fever | 0 | ||||||
| Blanchette et al48 | 146 | 6 mo-18 yr | 6-32 mo | Prednisone (4 mg/kg/d × 7 d then tapered to d 21)) | Median time to platelet count >20k, >50K | 2 d, 3 d | NR | None | 0 | I |
| IVIg (1 g/kg/d × 2 d) IVIg (0.8 g/kg once) | 2 d, 2 d 1 d, 2 d (P < .05 v prednisone) | NR NR | 16-18% with fever, nausea, vomiting, headache | 0 0 | ||||||
| Anti-D (25 µg/kg/d × 2 d) | 2 d, 2.5 d (P < .05 v both IVIg regimens) | NR | 24% with hemoglobin <10 g/dL | 0 | ||||||
| Albayrak et al55 | 57 | 2 mo-17 yr | 6 mo | Methylprednisolone (po, 30 mg/kg/d × 7 d) | Mean platelet count on days 0-30 | Mean >100K by d 4. No difference among groups | NR | Increased appetite and Cushingoid appearance | 0 | II |
| Methylprednisolone (po, 50 mg/kg/d × 7 d) | 1 ICH | Increased appetite and Cushingoid appearance | 0 | |||||||
| IVIg (0.5 g/kg/d × 5 d) | NR | 1 pt with aseptic meningitis; 2 with headache, vomiting | 0 | |||||||
*NR, Not reported.
Recommendations. There is level I evidence that children with acute, previously untreated ITP experience more rapid recovery of platelets with IVIg than with glucocorticoids or no specific therapy, but it is unclear whether this enhancement of platelet recovery influences bleeding or mortality or if there are circumstances in which the disadvantages of IVIg might outweigh its benefits. There is inadequate evidence regarding the efficacy of IVIg in other patient categories to develop definitive recommendations based on data. The opinion of the panel was that, regardless of the platelet count, it is appropriate (7.3 to 8.8, A-D) to treat severe, life-threatening bleeding initially with IVIg. IVIg was also considered appropriate as initial therapy for children with platelet counts <10,000 and minor purpura (1 g/kg for 1 day, 7.2, D) and for children with platelet counts <20,000 and mucous membrane bleeding (7.8 to 8.3, B). In all categories, a dose of 1 g/kg administered on 1 day received higher panel ratings (7.2 to 8.8, A-D) than a total dose of 2.0 g administered over 2 to 5 days (6.4 to 8.2, B-D). IVIg was considered appropriate initial treatment in children with platelet counts below 20,000 in whom inaccessibility or noncompliance is a concern (7.6 to 8.7, B-C). The panel considered IVIg inappropriate (1.0 to 1.2, A) in children with platelet counts >30,000 who are asymptomatic or have only minor purpura (Table 8).
Anti-Rh(D)
Evidence. One level I trial48 (Table 9) compared anti-Rh(D) to IVIg and glucocorticoid as initial therapy in patients with acute ITP and platelet counts <20,000 at presentation. The time required to increase platelet counts to >20,000 and >50,000 was slightly longer with anti-Rh(D) than with glucocorticoid or IVIg therapy. There are no level I or II data comparing anti-Rh(D) treatment to no treatment, nor is there evidence regarding the effectiveness of anti-Rh(D) in reducing mortality or morbidity from bleeding. Four level V studies72-75 suggest that anti-Rh(D) may increase the platelet count in about 80% of children with acute and chronic ITP, and that repeated treatments may postpone the need for splenectomy, but the responses are generally transient, lasting a median time of 5 weeks.
The only clinically important adverse effect of anti-Rh(D) appears to be alloimmune hemolysis. All Rh(D)+ patients develop a positive direct antiglobulin test after treatment, accompanied by a transient (1 to 2 weeks) decrease in hemoglobin concentration of about 0.5 to 2 g/dL. Although in two studies 4% to 24% of patients had a hemoglobin concentration of <10 g/dL after 7 to 14 days,48,74 red blood cell transfusion was not required.
Recommendations. There is level I evidence indicating that anti-Rh(D) increases the platelet count less rapidly than IVIg or glucocorticoids in children with acute, severe thrombocytopenia (platelet count <20,000). Based on opinion, the panel considered initial treatment with anti-Rh(D) inappropriate (1.0 to 3.0, A-D) for children presenting with platelet counts >30,000 (Table 8). The use of anti-Rh(D) in chronic ITP was not addressed in the panel survey.
Splenectomy
Evidence. Compared to adults, children with ITP are less likely to undergo splenectomy. Sixteen case series (level V evidence)18-29,76-79 describe outcomes from splenectomy over the past 40 years. In most instances, splenectomy was performed in children in whom thrombocytopenia had persisted for more than 1 year and who had clinically important bleeding. In some case series, children underwent splenectomy earlier in the course of their illness because of uncontrollable hemorrhage that was unresponsive to glucocorticoid therapy. Splenectomy is less frequent in more recent case series.36 These data consistently show that most children (72% of the 271 children undergoing elective splenectomy in the 16 case series) achieve a complete remission from ITP after splenectomy. An effect of splenectomy on morbidity or mortality has not been shown directly. There are few data on accessory splenectomy in children; it is discussed under Adult Treatment below.
The potential adverse effects of splenectomy include the operative and postoperative complications of bleeding and infection. An important concern for late morbidity and mortality after splenectomy is the long-term risk of fatal bacterial infection, particularly in children less than 5 years old, in whom the risk may be 1 death per 300 to 1,000 patient-years.80-82 However, most of these observations involved splenectomy for other diseases and predated the current practice of presplenectomy immunization and the administration of postsplenectomy prophylactic penicillin. Prophylactic penicillin has been shown to reduce the risk of infection in children with sickle cell anemia,83 and this observation may be generalizable to other asplenic children.
Recommendations. Although all available evidence is level V, the consistency of observations, the frequency of complete responses to splenectomy, and similar observations in larger samples of adult patients with chronic ITP suggest that splenectomy is an effective therapy. However, there are inadequate data to make evidence-based recommendations on the appropriate indications and timing for splenectomy, on when the harms of splenectomy might outweigh its potential benefits, or on appropriate preoperative management. Many of the case series predated the use of IVIg and anti-Rh(D) therapy, which can provide intermittent support for children with recurrent, symptomatic thrombocytopenia and thereby postpone or avoid the need for splenectomy. The occurrence of spontaneous complete remissions in some children with chronic ITP may also lessen the need for this procedure.
The panel reached consensus on only selected indications for splenectomy, such as persistence of disease 12 months after diagnosis with bleeding symptoms and a platelet count of <10,000 (7.5 to 9.0, A-C for ages 3 to 12 years) or of 10,000 to 30,000 with bleeding symptoms (7.6 to 7.9, B for ages 8 and 12), but it considered only certain scenarios. These scenarios assume that primary treatment (glucocorticoid, IVIg, and/or anti-D) was only transiently successful and that there are no medical contraindications to the surgery. The panel had strong disagreement (5.0, E) about the appropriateness of emergency splenectomy in the case of urgent, life-threatening bleeding in which conventional critical care measures are already underway.
If an elective splenectomy is planned, preoperative prophylaxis that the panel considered appropriate to reduce the risk of intraoperative and postoperative bleeding included (1) IVIg (8.8, A), parental glucocorticoid (7.2, D), and anti-D (7.2, D) therapy for platelet counts <10,000 and (2) IVIg therapy for platelet counts of 10,000 to 20,000 (8.5, B) or 20,000 to 30,000 (7.7, B). Indications that were considered inappropriate for preoperative prophylaxis included IVIg for platelet counts >50,000 (2.6, D), platelet transfusion for platelet counts of 20,000 to 30,000 (2.3, D) or >30,000 (1.0, A), anti-D for platelet counts >50,000 (1.0, A), oral glucocorticoid therapy for platelet counts >50,000 (2.6, D), and parenteral glucocorticoid therapy for platelet counts of 30,000 to 50,000 (2.2, C) or >50,000 (1.0, A).
The panel endorsed the recommendations of the Advisory Committee on Immunization Practices that, at least 2 weeks before elective splenectomy, children should be immunized with Hemophilus influenzae type b vaccine and, if over 2 years of age, with polyvalent pneumococcal vaccine and quadrivalent meningococcal polysaccharide vaccine.84
Other Treatments
Evidence. Only four level V case series have evaluated other treatment modalities (plasma infusion, azathioprine, danazol, and interferon) for ITP in children.85-88 The modalities are described in the subsequent section on treatment of adults.
Recommendations. There is insufficient evidence to make recommendations about alternative treatment modalities when ITP symptoms persist after primary treatment and splenectomy, or to assess when the benefits of such treatments outweigh their potential harms. Furthermore, the data on the clinical course of ITP in children do not clarify whether further treatment is even necessary under these circumstances. Based on opinion, the panel did not recommend further treatment of children with platelet counts >30,000 who have failed to respond to splenectomy and have no bleeding symptoms (2.0, B for platelet count of 30,000-50,000; 1.0, A for platelet count >50,000). Further treatment was recommended (9.0, A) for children with platelet counts <30,000 who have active bleeding. The panel considered many treatments (and no treatment) to be reasonable options, reflecting the lack of evidence that any single treatment is better than another.
Clinical Course
An understanding of the clinical course of ITP in adults is essential to make informed management decisions, to know which patients require treatment either at the time of diagnosis or in the management of chronic disease, and to estimate morbidity and mortality, with and without treatment.
Evidence. ITP in adults is typically a chronic disease. However, the clinical course of untreated disease is uncertain, because, in contrast to children, patients with symptomatic thrombocytopenia are generally treated initially with glucocorticoids. Despite this bias, which would tend to underestimate the severity of untreated disease, the data suggest that the course of ITP is more serious in adults than in children, with an estimated rate of fatal hemorrhage of 5%, due mainly to intracranial hemorrhage (Table 10). Most data on fatal hemorrhages were collected in previous decades, when platelet transfusions and IVIg were unavailable and supportive care for critical complications was less effective. Thus, current mortality rates may be less than 5%. At equivalent platelet counts, hemorrhagic complications may be more common in older patients.89,90 There are no long-term follow-up data on outcomes in adults with incidentally discovered asymptomatic thrombocytopenia. In addition, the relative incidence of symptomatic versus incidentally discovered thrombocytopenia is unknown.
| Hemorrhagic Complications[dagger] | Patients With Persistant Thrombocytopenia§ | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Author | Location | Years | Patients (no.) | Patients With Complete Remission on No Therapy* | ICH | Other | Other Deaths | Patients in Complete Remission* at Last Follow-up[double dagger] | No. | Spontaneous Recovery | Deaths From Hemorrhage |
| Watson-Williams et al91 | Scotland | 1928-1957 | 78 | 12/26|| | 4 | 1 | 0 | 46/62 (74)¶ | -- | -- | -- |
| Carpenter et al92 | US | 1945-1959 | 46# | 0/12 | 2 | 0 | 0 | 20/35 (57) | -- | -- | -- |
| Thompson et al93 | US | 1945-1970 | 66 | 3 | -- | -- | 4 | 37/62 (60) | -- | -- | -- |
| Meyers94 | US | 1950-1961 | 71 | 0 | 2 | 0 | 2 | 56/57 (84) | 15 | 0 | 2 |
| Jiji et al95 | US | 1951-1972 | 91 | 0 | -- | -- | 5 | 53/86 (64) | -- | -- | -- |
| Picozzi et al96 | US | 1959-1969 | 38 | 0 | 0 | 0 | 0 | 25/36 (69) | 16 | 8 | 0 |
| Ikkala et al97 | Finland | 1966-1973 | 41 | 1 | -- | 2 | 1 | 27/38 (71) | 14 | 0 | 0 |
| DiFino et al98 | US | 1971-1979 | 62 | 0 | 0 | 3 | 3 | 34/51 (67) | 18 | 0 | 3 |
| Jacobs et al99/ | South Africa | 1971-1981 | 148 | 1 | 1 | 1 | 0 | 78/146 (53) | 18 | 0 | 0 |
| den Ottolander et al26 | Netherlands | -- | 69 | 1 | -- | -- | -- | 39/69 (57) | -- | -- | -- |
| Pizzuto and Ambriz100 | Central and South America# | -- | 934 | 9 | 27** | -- | 19** | 577/887 (65) | 384 | 14 | 20 |
| Cortelazzo et al90 | Italy | 1982-1989 | 117 | --[dagger][dagger] | -- | -- | 1 | 33/67 (49) | -- | -- | -- |
| Summary | 1,761 | 27 | 36 | 7 | 35 | 1,027/1,606 (64%) | 465 | 22 (5%) | 25 (5%) | ||
Abbreviation: ICH, intracranial hemorrhage.
*Complete remission is defined as a normal platelet count on no therapy continuing to the time of the last observation. In almost all patients there was no opportunity to observe a spontaneous remission because steroids were begun at the time of diagnosis. In the first two series, a substantial number of patients were untreated when, before 1950, splenectomy was the only effective modality.
[dagger]Acute hemorrhagic deaths are arbitrarily defined as occurring within 6 mo of diagnosis. Other hemorrhagic deaths occurred after 6 mo, or the time was not specified--and it was unclear even if all of these were due to hemorrhage from ITP.
[double dagger]The number of patients is less than the original series by deaths and patients lost to follow-up. This estimate is largely dependent on the duration of follow-up, which was variable.
§Patients who failed to achieve a complete response to glucocorticoid, splenectomy, and subsequent therapy. In contrast to children, persistence is defined as lack of response to treatment rather than by an arbitrary time.
||This group contained 3 children less than 12 years old, and it was not stated if they were among the patients whose ITP resolved.
¶Numbers in parentheses are percentages.
#This study was a collaborative effort of 10 institutions.
**This report stated that 27 of the total of 46 hemorrhagic deaths were due to ICH, 9 due to gastrointestinal or pulmonary bleeding, and 10 due to "massive" purpura, but these etiologies were not distinguished according to time from diagnosis.
[dagger][dagger]49 patients with platelet counts over 30,000/µl and no bleeding symptoms were not treated and apparently had no major hemorrhagic complications. Whether any complete remissions occurred is not stated. These patients are not included in the estimate of patients in complete remission at last follow-up.
Diagnosis
History and physical examination. The history and physical examination are aimed at detecting alternative causes of thrombocytopenia. The most important elements of the history and physical examination identified by the panel are presented in Table 11. The primary objective of the history is to assess the type of bleeding and to distinguish pl