Clinically significant newly presenting autoimmune thrombocytopenic purpura in adults: a prospective study of a population-based cohort of 245 patients


P. R. A. Taylor, Department of Haematology, Royal Victoria Infirmary, Queen Victoria Road, Newcastle upon Tyne NE1 4LP, UK. E-mail:


Summary. The true incidence and prognosis of autoimmune thrombocytopenic purpura (ITP) in adults is unknown. We present the results of a prospective study in a population-based cohort of newly presenting adults (≥ 16 years) with ITP and platelet count of < 50 × 109/l, which took place between 1 January 1993 and 31 December 1999 in the former Northern Health Region in the UK (population 3·08 million). A total of 245 cases were confirmed by bone marrow examination with a median follow-up of 60 months (range 6–78 months). There were 134 females/111 males (1·2:1). Overall incidence was 1·6 per 105 per annum. Absolute incidence was similar for both sexes, with highest age-specific incidence in those aged > 60 years. Thirty patients (12%) presented with frank bleeding, and 28% were asymptomatic. Forty-five patients (18%) received no treatment, and 135 (55%) received first-line treatment only. Thirty patients (12%) underwent splenectomy. There were four deaths (1·6%) from bleeding and/or the complications of therapy in this cohort, but only one was in the acute phase of the illness. The majority of patients (155 out of 245) achieved remission (platelet count > 100 × 109/l), with a further 59 (24%) in partial remission with no symptoms (platelet count 30–100 × 109/l). This population-based study suggests that the traditional view of adult ITP as being a predominantly chronic disease that preferentially affects females needs to be modified.

Autoimmune thrombocytopenic purpura (ITP) was first described in 1735 by Werlhof (Editorial, 1963), who described a young woman with sudden onset of spontaneous petechiae, ecchymoses and mucous membrane haemorrhage. In the American Society for Haematology guidelines (George et al, 1996), ITP is defined as ‘isolated thrombocytopenia with no clinically apparent associated conditions or other causes of thrombocytopenia’. It is thus a condition that, to a large extent, is a diagnosis of exclusion of, for example, human immunodeficiency virus infection, systemic lupus erythematosus, lymphoproliferative disorders, myelodysplasia, agammaglobulinaemia, drug-induced thrombocytopenia, alloimmune thrombocytopenia, congenital/hereditary non-immune thrombocytopenia, relying on the presence of thrombocytopenia and an otherwise normal full blood count, with no clinically apparent associated conditions or drug therapy that may cause thrombocytopenia (George et al, 1998).

Although ITP in adults is not considered to be rare, the true incidence of the disorder is unknown. Although several demographic studies have been carried out in children, current estimates of the incidence among adults are taken from studies with patients diagnosed between 1936 and 1985 (Jacobs & Wood, 1986; Cortelazzo et al, 1991; Kaufman et al, 1993) or are extrapolated from the data on children. These studies report an excess of females, with a peak age of incidence around 40 years (Doan & Bouroncle, 1960).

Clinical observation and experience indicate a spectrum of manifestations ranging from trivial bruising to catastrophic haemorrhage. The incidence of fatal haemorrhage is as high as 10·4% according to some studies (Stasi et al, 1995; George et al, 1996; Frederiksen & Schmidt, 1999). The literature suggests that older patients appear to have more severe bleeding manifestations (Cortelazzo et al, 1991). It has been reported that ≈ 43% of patients achieve a remission with a platelet count of > 100 × 109/l, thus implying that the majority of patients have a continuous or relapsing chronic disease state (George et al, 1995).

Clinical experience and anecdote supported by some published data led us to believe that the true demographic features of this disease differ from the published literature. This led the members of the Northern Region Haematology Group to initiate an audit. Prospective collection of data from adult patients with ITP within the Northern Health Region of England was commenced on 1 January 1993. The intention of the audit was to provide a current review of demographics, clinical manifestations and response to treatment in ITP. It is a prospective population-based cohort study allowing assessment, within this defined population, of the effect of changes in management introduced over the last half-century and thus provides a more contemporaneous review of ITP than that found in published texts (which mostly reflects practice before 1970).

Patients and methods

All patients with ITP aged 16 years and over presenting between 1 January 1993 and 31 December 1999 within the Northern Health Region of England (population 3·08 million) were registered prospectively by the haematologist who made the diagnosis.

Eligibility criteria.  For the purposes of this audit, the diagnosis of ITP required exclusion of other haematological pathology by bone marrow aspirate and biopsy. The date of diagnosis was considered to be the date of the marrow biopsy. Patients with platelet counts ≥ 50 × 109/l were excluded from the audit as they were considered to be clinically less problematic. Patients with gestational and therapy-related thrombocytopenia and all those with a proven cause for thrombocytopenia, e.g. drug related, were also excluded. Infection-associated cases were not excluded from the cohort. Eligible cases therefore had presumed autoimmune but otherwise idiopathic thrombocytopenia.

Clinical details.  Full presenting details were obtained from the hospital notes of all registered new cases of ITP. Information collected included the presenting full blood count, age at diagnosis, date of presentation and initial therapy instituted. Additional information on medications, co-existing medical problems with particular reference to preceding viral symptoms, presence of systemic lupus erythromatosus or other autoimmune disorder and co-morbidity with malignant disease was also collected.

Follow-up was to 30 June 2000 with median follow-up of 60 months (range 6–78 months). At first follow-up, details of presenting symptoms and signs, with particular reference to haemorrhagic symptoms and response to initial treatment, were recorded. The requirement for subsequent therapy, with a record of the response, and whether splenectomy was performed was also made. Whether the patients undergoing splenectomy had been registered on the Regional Asplenia Register (Spickett et al, 1999) was also audited.

For the purposes of this study, the current information on symptoms, details of subsequent treatment, date of last clinic attendance and most recent platelet count were collected at the final follow-up (date last seen nearest to 30 June 2000).

Treatment.  Clinicians instituted the appropriate therapy according to their clinical judgement and not according to a set protocol.

Remission status.  A normalization of the platelet count to ≥ 100 × 109/l was deemed complete remission (CR). Remission status was taken as the count on the date the patient was last seen in the clinic and/or platelet count was recorded (6–78 months disease duration). Partial remission was defined as platelet count above the count at presentation but < 100 × 109/l, with or without haemorrhagic symptoms or signs. However, if the platelet count failed to attain 30 × 109/l, this was considered not a remission.

Chronic ITP was defined as persistent thrombocytopenia (< 50 × 109/l) for more than 6 months after diagnosis.

Mortality.  The date and cause of death were documented if applicable.

Population statistics.  Data were collated, managed and analysed using Microsoft excel software. Age-specific population information was obtained from the UK 1991 Census from the Office of Population Censuses & Surveys (1995).


Study eligibility

Of the 343 patients registered, 98 were excluded from the study. There were therefore 245 evaluable patients.


The principle reasons for exclusion from the study are detailed below. The most common reasons were lack of bone marrow examination (n = 40) and drug-related thrombocytopenia (n = 28). Other reasons for exclusion were age < 16 years (n = 2), platelet count ≥ 50 × 109/l (n = 2), relapse of previous ITP (n = 3), gestational ITP (n = 2), systemic lupus erythematosus (n = 8), no records available (n = 7) and other causes (n = 6).

Quinine was the commonest drug-related cause of thrombocytopenia (n = 13/28; 46·4%). The platelet count at diagnosis in these patients was within the range 1–27 × 109/l, with a median count of 4 × 109/l. There were four males and nine females, with a median age of 72 years (range 45–82 years); three of the 13 patients were asymptomatic. All responded to withdrawal of quinine and/or first-line treatment, with normalization of the platelet count.

In those patients taking drugs other than quinine (including gold, thiazide diuretics, antiepileptics, antischizophrenic drugs and cotrimoxazole), the median age was 50 years (range 17–76 years), with three males and 12 females in this category. In this cohort, the platelet count at diagnosis was in the range 2–32 × 109/l, with a median of 10 × 109/l. Five of the 15 patients in this group were asymptomatic. Six of the 14 patients who were treated in this group failed to respond to first-line therapy, with three proceeding to splenectomy.

Age and gender distribution

The age range of the patients presenting with confirmed ITP was 16–91 years, with a median age at diagnosis of 56 years of age. There was a female to male ratio of 1·2:1 (134 females to 111 males).

Annual incidence

The overall incidence of ITP in patients presenting with platelet count < 50 × 109/l was 1·6/105/year. Figure 1 shows the incidence rates according to age and gender. The incidence was approximately equal for gender except for patients aged 45–59 years. The highest age-specific incidence was in those aged > 60 years.

Figure 1.

Age/gender-specific incidence per 100 000 population of newly presenting clinically significant autoimmune thrombocytopenia. Age/gender-specific incidence was greatest in patients aged > 60 years. There was no gender difference, except for the 45–59 age group.

Associated disorders

Within our cohort of 245 cases, 13 patients had a previous or concomitant history of solid tumours, and three patients had a previous history of haematological malignancies; bone marrow examination revealed no evidence of infiltration at the time of diagnosis of ITP in any of these patients. The three patients with previous haematological malignancy [one acute myeloid leukaemia, one Non-Hodgkin's lymphoma (NHL) and one Hodgkin lymphoma] attained CR from their ITP and have maintained this without relapse of either their primary disorder or their ITP.

Five patients had co-existing inflammatory bowel disease, but only one was receiving treatment for this complaint at the time of diagnosis with ITP. One patient presented with ITP at age 32 years, with a previous history of autoimmune haemolytic anaemia at 14 years old.

‘Post-infection’ thrombocytopenia

Twenty-two (16 females and six males) of the 245 patients had a history suggesting preceding infection, two of whom had proven infectious mononucleosis. A further patient was IgM positive for both cytomegalovirus and human parvovirus B19 at presentation. In all other cases, the clinical history was the only evidence of infectious illness. Thirteen of 22 patients presented with platelet count < 10 × 109/l, although only one had frank bleeding.

Two patients achieved spontaneous remission. Seven of 19 patients who received steroids as initial treatment had no response to therapy and proceeded to treatment with immunoglobulin; three patients eventually underwent splenectomy.

Nineteen of 22 patients remain in continuous CR 6–76 months after diagnosis. Two patients are in partial remission with no symptoms (PRNS) requiring no therapy, with neither patient having received immunoglobulin or undergone splenectomy.

Seasonal variation

No evidence of seasonal variation of presentation was demonstrated in this study cohort (data not shown).

Haemorrhagic symptoms at presentation

Presenting symptoms by platelet count are summarized in Table I.

Table I.  Presenting symptoms by presenting platelet count and gender.
Platelet count (× 109/l)Patient genderHaemorrhagePurpuraAsymptomatic
  1. Men were more likely than women to present with frank bleeding, although this did not reach statistical significance.

(n = 114) (21%)(10%)(70%)(62%)(9%)(28%)
(n = 51) (5%)(16%)(71%)(64%)(24%)(20%)
(n = 26) (40%)(10%)(20%)(52%)(40%)(38%)
(n = 54) (4%)(4%)(34%)(52%)(62%)(44%)
Total 245111M:134F(14%)(10%)(58%)(59%)(27%)(31%)

With lower platelet counts, most patients were symptomatic with frank bleeding as well as purpura; however, once the platelet count reached > 30 × 109/l, frank bleeding became less problematic, and purpura predominated, although the majority of patients were asymptomatic. There was evidence of a difference in presenting haemorrhagic symptoms with age and gender (see Table II). Ten per cent of females presented with frank bleeding compared with 14% of males, but the difference between females and males did not achieve statistical significance (P = 0·3, chi-squared test).

Table II.  Frank bleeding at presentation (%) by age and gender.
Age range (years)MaleFemale
  1. Females showed an increased incidence of haemorrhage only in the 16–29 age group.

(n = 53)(4%)(15%)
(n = 43)(14%)(14%)
(n = 36)(19%)(0%)
(n = 67)(22%)(14%)
(n = 46)(13%)(6%)
Total (all ages)16/11114/134
(n = 245)(14%)(10%)

There was one acute haemorrhagic death in the study patients. Full details are given in the section on Deaths.

Spontaneous remissions

There were five patients who underwent spontaneous remission from their ITP; one of particular note attained spontaneous remission after Herpes zoster infection 5 years after diagnosis.

Response to therapy

Full details of response to therapy are summarized in Tables III and IV.

Table III.  Number of treatment modalities given according to presenting platelet count.
First line
Third line
or more
Total proceeding
to splenectomy
  1. The majority of patients responded to first-line therapy. Only 12% of the patient cohort underwent splenectomy during the course of their disease.

< 10 × 109/l27726911
(n = 114)(2%)(68%)(23%)(8%)(10%)
10–19 × 109/l62611813
(n = 51)(12%)(51%)(22%)(16%)(25%)
20–29 × 109/l713604
(n = 26)(27%)(50%)(23%)(0%)(15%)
≥  30 × 109/l3119312
(n = 54)(57%)(35%)(6%)(2%)(4%)
Total 24546135461830
Table IV.  Clinical outcome by presenting platelet count.
Platelet countCRPRNSPR M/NRLost to follow-upHaemorrhagic death
  1. CR, complete remission; PRNS, partial remission with no symptoms; PR, partial remission; M/NR, minimal/no response.

  2. Platelet count at diagnosis did not predict outcome.

< 10 × 109/l89153411
(n = 114)(78%)(13%)(2%)(3%)  
10–19 × 109/l31141212
(n = 51)(60%)(27%)(2%)(4%)  
20–29 × 109/l1870100
(n = 26)(69%)(27%)(0%)(4%)  
≥  30 × 109/l172311000
(n = 54)(33%)(45%)(2%)(5%)  
Total (245)1555951723

Of the 114 patients who presented with a platelet count < 10 × 109/l, the majority (78%) eventually achieved CR, with 67% having first-line treatment only and 10% undergoing splenectomy. The prognosis was excellent; < 20% of cases developed chronic ITP that required ongoing treatment.

Fifty-one patients presented with a platelet count between 10 and 19 × 109/l, with 60% achieving CR. Fifty-one per cent of patients required first-line therapy only, 25% proceeded to splenectomy, while 33% were considered to have chronic ITP.

As the platelet count at presentation rose, fewer patients required therapy. Of the 26 patients with a presenting platelet count of 20–29 × 109/l, 69% achieved CR and 50% of patients required first-line therapy only. The majority of the 54 patients (57%) presenting with a platelet count > 30 × 109/l did not receive therapy. Of those patients who required therapy, 33% achieved CR. Most of this cohort continued to have platelet counts between 30 and 100 × 109/l. There were no haemorrhagic deaths recorded.


Thirty patients (12%) proceeded to splenectomy. Treatment before splenectomy and post-operative response is detailed in Table V. The three patients with platelet counts > 30 × 109/l at presentation subsequently became symptomatic as a result of falling platelet counts, and therefore had a splenectomy. Seven of the 30 patients did not achieve CR but are currently asymptomatic. One patient (patient 18) died in the post-operative period on d 15 as a consequence of sepsis.

Table V.  Clinical details of patients who underwent splenectomy.
count (× 109/l)
  1. S, steroids; IgG, immunoglobulin; N, none; AZA, azathiaprine; SPL, splenectomy; DAN, danazol; IFN, interferon; CR, complete remission; PRNS, partial remission with no symptoms; MI, myocardial infarction.

  2. No patients underwent splenectomy as first-line treatment. The majority of patients attained complete remission after splenectomy.

127F4SIgG + SPL  CR
335M13SSPL  CR
425M6S + IgGSPL  CR
516F22NSPL  CR
742F20SSPL  CR
838F20SSPL  CR
946M37NSPL  CR
1079F24NSPL  CR
1765M48SS + IgGSPL Died (MI)
1880F10SSSPL Died (Post-op sepsis)

All patients who underwent splenectomy were registered on the Regional Asplenia Register (Spickett et al, 1999). Follow-up to date has identified no late post-splenectomy infective deaths in this study cohort of patients.


Twenty-seven patients died within the study period. The details are analysed according to platelet count at diagnosis (see Table VI). Three deaths were considered to be directly attributable to ITP, but only one of these three died at presentation.

Table VI.  Cause of death by presenting platelet count.
Platelet count at presentation (× 109/l)Cause of death
  1. Only four patients died of bleeding and/or therapy that was directly attributable to their ITP.

< 10Bleeding
Carcinoma of bronchus
Carcinoma of colon (n = 2)
Refractory anaemia with excess blasts
Myocardial infarction (n = 2)
Non-Hodgkin's lymphoma
10–19 ‘Old age’
Bleeding (n = 2)
Carcinoma of ovary
Chronic lymphocytic leukaemia
Cerebral vascular accident (NB. platelet count normal)
Post-splenectomy complications
20–29Post-operative bowel resection (ulcerative colitis)
Carcinoma of breast
≥  30 ‘Old age’
Carcinoma of bronchus
Carcinoma of breast (n = 2)
Myocardial infarction
Chronic obstructive airways disease (n = 2)
Ruptured aortic aneurysm

One patient died in the acute phase of the disease, on d 5. This patient was a 19-year-old-male, who had concomitant acute autoimmune haemolytic anaemia, and died of gastrointestinal haemorrhage while receiving steroid therapy. A further two patients died later in the course of their disease, one of a gastrointestinal bleed having developed low-grade NHL and another while on warfarin. One patient died as a consequence of sepsis on the 15th day after splenectomy.


The authors of the American Society for Haematology practice guidelines for ITP identified priorities for future research following their evidence-based literature review (George et al, 1996). We have attempted to address aspects of three of their priorities in the current study:

  • 1The need for a prospective study of the clinical course of untreated ITP in patients presenting with mild or moderate thrombocytopenia and no clinically important bleeding, including long-term follow-up and the outcomes of bleeding and mortality.
  • 2The need to obtain data on the clinical course of chronic refractory ITP, especially in those untreated patients without clinically important bleeding.
  • 3The need to assess the prognostic relation of the platelet count to bleeding and mortality.

In relation to point 1, we have considered only patients who have moderate thrombocytopenia.

It is likely that our study will be the last to contain bone marrow confirmation of diagnosis as it has been recommended (George et al, 1996) that performing a bone marrow biopsy is unnecessary in patients under the age of 60 years with otherwise normal blood films.

ITP is generally reported to be commoner in women than in men, and this is supported by numerous reports in the literature (Watson-Williams et al, 1958; Doan & Bouroncle, 1960; Thompson et al, 1972; DiFino et al, 1980; Pizzuto & Ambriz, 1984; Jacobs & Wood, 1986; Cortelazzo et al, 1991; Kaufman et al, 1993; George et al, 1995; Portielje et al, 2001; Godeau et al, 2002) (see Table VII). The present study, which is population based, found that, although the absolute majority of patients presenting with ITP are women, when the population at risk is taken into account, the age/gender-specific incidence is similar (see Fig 1). The actual incidence shows a biphasic distribution with most patients presenting during the seventh and eighth decades (Fig 1). This contrasts with the incidence of ITP in adults extrapolated from the incidence in children, which has been estimated to be 6·6 per 100 000 (George et al, 1995). However, our study was restricted to patients with platelet counts of ≤ 50 × 109/l. A higher threshold would have produced a greater incidence than 1·6 per 100 000 per year as found in the present study.

Table VII.  Summary of previous studies of ITP.
StudyNo. of patientsMedian age (range)Ratio M:FPlatelet counts (× 109/l) included in studyBM confirmationAt study entry % with chronic ITPBleeding at presentation-line therapyResponse to first-line treatment% having splenectomy(%) Death rate
  • *

    Randomized trial.

  • NA, not available; N/A, not applicable.

  • The majority of studies described patients who were diagnosed before the era of readily available automated blood counts. There has been only one recent randomized study of treatment in this disorder.

Watson-Williams et al (1958)58NA1:2·6< 100N21%NA75%390·1
Doan & Bouroncle (1960)271(0–80)1:2·1< 50YNA33%NA621·4
Thompson et al (1972)66(14–80)1:2·1< 100YNANA49% CR 26% PR540·1
Jiji et al (1973)54NANA ‘Significantreduction’Y100%NA24%880·1
DiFino et al (1980)62NA1:2< 30Y71%63%43%600·1
Picozzi et al (1980)38(0–70)NANANA100%5%3% CR 39% PR68%0·1
Pizzuto & Ambriz (1984)934(20–65)1:3·3NANANANA65%43%5·0
Jacobs & Wood (1986)14838 (13–38)1:3·210–180YNA1%19%69%NA
Cortelazzo et al (1991)11743 (16–84)1:2·9< 100NA100%7%NA28%0·1
Kaufman et al (1993)255491:1·8< 30YNANANANANA
Portielje et al (2001)15238 (15–86)1:1·7< 100Y0%82%59%57%2·6%
Godeau et al (2002)*12238 (24–59)1:1·9< 20Y0%Excluded from study40%N/AN/A
Present study24556 (16–91)1:1·2< 50Y0%12%68%12%1·6%

The median age at presentation in the five studies in which this information is available (Jacobs & Wood, 1986; Cortelazzo et al, 1991; Kaufman et al, 1993; Portielje et al, 2001; Godeau et al, 2002) was 38–49 years, whereas in the current study, it is 56 years. It is possible that a degree of selection in previous studies (e.g. reports from tertiary referral centres) may account for this difference in incidence. The median age of the general population has increased significantly over the last three decades, and this may have further influenced the age incidence patterns. Also, many of our patients (27·8%) were asymptomatic and only diagnosed incidentally after a routine full blood count. These patients would probably not have been identified before the automated full blood count analysis introduced over the last three decades, as platelet counts would not have been performed routinely.

Another characteristic of our patient cohort that differs from previously published observations is the pattern of incidence of haemorrhagic manifestations. As shown in Table I, this unselected cohort showed a non-statistically significant trend for males to be more likely to present with frank bleeding, whereas in females, the presenting symptoms were more likely to be purpura.

In the present study, the death rate from causes directly related to ITP or its treatment was 1·6% with only one patient dying at presentation. This would be in keeping with the experience of others (Watson-Williams et al, 1958; Doan & Bouroncle, 1960; Thompson et al, 1972; JiJi et al, 1973; DiFino et al, 1980; Picozzi et al, 1980; Pizzuto & Ambriz, 1984; Cortelazzo et al, 1991; Portielje et al, 2001) where the reported incidence was 0–5%.

Although not strictly ITP, but a drug-related thrombocytopenia, it was interesting to note that quinine was the commonest drug identified as causing thrombocytopenia (46·4%). In two of these patients, this diagnosis was made only when ‘relapse’ occurred and a more detailed history was taken; the consequence of patients being unaware that tonic water may contain the drug quinine. We feel that quinine should be specifically enquired about when obtaining a drug ingestion history in patients with ITP, particularly as these are older patients presenting with a low platelet count who are mostly symptomatic. Withdrawal of the drug resulted in a good response.

The most frequently used form of first-line treatment for our cohort was steroid therapy. Although not completely comparable, the response rate was similar to that reported in the most recent studies (Portielje et al, 2001; Godeau et al, 2002). The total number of patients progressing to splenectomy was 12% (see Table III). This is lower than in the literature (28–88%, see Table VII) (Watson-Williams et al, 1958; Doan & Bouroncle, 1960; Thompson et al, 1972; JiJi et al, 1973; DiFino et al, 1980; Picozzi et al, 1980; Pizzuto & Ambriz, 1984; Jacobs & Wood, 1986; Cortelazzo et al, 1991; Portielje et al, 2001) and may reflect a more selected group of patients in other studies.

In this study cohort, presenting platelet count was not a predictor of likelihood of progression to splenectomy. Table III shows that the majority of patients became clinically stable or achieved remission after treatment with steroids. A small number required further effective second-line treatment. The prognosis was excellent: only 81 (33%) had chronic ITP (with a platelet count < 50 × 109/l). The majority of these patients were asymptomatic, not requiring therapy (and 19 never received any treatment at all for their ITP).

In conclusion, this population-based study of clinically significant ITP showed that, contrary to previously published studies, the age/gender-specific incidence was approximately equal for males and females, and the maximum age-specific incidence was in the eighth decade. This is in contradiction to the generally accepted epidemiological data found in textbooks (George et al, 1995), and may reflect the changing age profile of the population, the referral patterns of patients in previous studies and the introduction of automated cell counters. The platelet count at presentation was not found to be a prognostic indicator for splenectomy; interestingly, it was the cohort with a presenting platelet count of 10–19 × 109/l who were more likely to proceed to splenectomy. Patients with a platelet count > 30 × 109/l were mostly asymptomatic and required no treatment. Response to first-line therapy was good, with only 32·2% of patients proceeding to further therapy. Generally, our treatment approach relied less on splenectomy as a first- or second-line therapy compared with other studies. The death rate for ITP was low, with three out of 245 patients dying of bleeding, and only one of the three was in the initial phase of the disease. It is interesting to note that only a minority of patients required further intervention after steroid therapy, with relatively few progressing to chronic ITP.


The authors wish to thank Dr M. M. Reid for his critical reading of this manuscript, and Jane Lord for secretarial help. Our thanks also to the members of the Northern Region Haematology Group without whose co-operation this study would have been impossible. Dr P. Taylor is funded by a grant from the Northern and Yorkshire R&D Department.

Northern Region Haematology Group

Dr M. Abela, Dr T. Biss, Dr N. M. Browning, Dr P. J. Carey, Dr J. Cavet, Dr J. Chandler, Dr C. Chapman, Dr M. Collin, Dr P. Condie, Dr L. Crossman, Dr M. Dewar, Dr H. Dignum, Dr M. J. Galloway, Dr D. K. Goff, Dr P. J. Hamilton, Dr J. Hanley, Dr A. Hendrick, Dr A. Iqbal, Dr G. H. Jackson, Dr F. M. Keenan, Dr P. Kesteven, Dr A. L. Lennard, Dr S. Marshall, Dr Z. T. Maung, Dr P. J. Mounter, Dr I. Neilly, Dr A. Nicolle, Dr H. G. O'Brien, Dr S. G. O'Brien, Dr M. M. Reid, Dr R. Sharples, Dr D. Stainsby, Dr G. L. Stark, Dr G. P. Summerfield, Dr K. Talks, Dr H. Tinegate, Dr C. Tiplady, Dr M. Velangi, Dr J. Wallis, Dr A. W. Wells, Dr N. West, Dr P. J. Williamson, Dr A. Wood, Dr A. Youart.