Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Drug-induced agranulocytosis


Agranulocytosis (abnormally low levels of some white cells in the blood) can occur when drugs cause injury to the bone marrow. It is a serious condition, and about 1 in 10 cases results in death. Drug-induced agranulocytosis is rare, perhaps affecting a few people per million of population per year.

Because it is rare, agranulocytosis rates are difficult to measure. It is difficult to know whether certain drugs cause agranulocytosis, and what the risk is. A new and rather good study from Barcelona helps [1].


The population of Barcelona (about four million people) was observed from 1980 to 2001 for cases of agranulocytosis reported by 17 haematology units. All potential cases were found through weekly calls to the units, using defined definitions for agranulocytosis based on laboratory results. These included blood count results, and, usually, a bone marrow aspirate.

Exclusions included children under two years, patients receiving treatments (like chemotherapy or immunosuppressants) known to interfere with bone marrow function, or with conditions (like leukaemia or AIDS) known to involve impaired bone marrow function. Patients also had to be able to participate in interviews relating to drug use.

For each case four controls were selected randomly, and personnel unaware of their status, using a structured questionnaire, interviewed cases and controls. Details of drug use within the previous six months were obtained. Drug exposure was defined as use in the week before an index day, defined as the day when the first symptom of agranulocytosis occurred.


There were 79 million person years of observation, during which there were 396 cases of agranulocytosis, with 273 admitted to hospital because of agranulocytosis (community cases) and 123 cases in which agranulocytosis developed during an admission. The overall incidence was 5 cases per million per year, with 3.5 per million per year for community cases. Incidence was similar in men and women, but increased with age (Figure 1), and more than half the community cases were older than 64 years.

Figure 1: Agranulocytosis incidence by sex and age

Overall fatality in the four weeks after diagnosis was 9.1%, and 7% for community cases. Fatality was also age-related, and was much higher in those older than 64 years (Figure 2).

Figure 2: Case fatality rate for agranulocytosis by age, over 20 years

Drugs significantly associated with agranulocytosis in 177 community cases are shown in Table 1, together with the attributable incidence. These drugs accounted for about two-thirds of cases. For some of these there were few case and control patients receiving the drugs, with 10 or fewer for calcium dobesilate, spironolactone, and carbamazepine. No drug had an attributable incidence of more than one case per million population per year.

Table 1: Drug-related agranulocytosis for those with a statistically significant association, giving numbers of patients using the drug in cases and controls, odds ratios, and attributable incidence

Cases 177/Control 586
Odds ratio
(95% CI)
Attributable incidence
(per million per year)
Ticloplidine hydrochloride
103 (13 to 840)
Calcium dobesilate
78 (4.5 to 1300)
Antithyroid drugs
53 (5.8 to 480)
26 (8.4 to 79)
20 (2.3 to 180)
11 (1.2 to 100)
8.0 (2.1 to 31)
ß-lactam antibiotics
4.7 (1.7 to 13)
Diclofenac sodium
3.9 (1.0 to 15)


Agranulocytosis was defined according to sensible criteria, comprehensibly identified and followed up, in a large defined population with a universal free health care service of high quality, for over 20 years. Few cases were likely to be missed. These are great strengths of the study. Weaknesses were that for some drugs there were very few events, and some of the drugs significantly associated with agranulocytosis are not commonly used around the world, and some not now in Spain.

The study helps in several ways. It highlights those drugs that are likely to be associated with increased risk of agranulocytosis, helping to make sure cases are not missed. It tells us to keep a special look at older people on these particular drugs. And it also informs us of a lack of association with agranulocytosis for the many drugs used in a typical population.

Many questions remain unanswered. A Lancet editorial [2] has examined issues around agranulocytosis with dipyrone. Part of the problem is low numbers of events, and part is apparently varying rates associated with drugs in different parts of the world. There is still much to be learned.


  1. L Ibáñez et al. Population-based drug-induced agranulocytosis. Archives of Internal Medicine 2005 165: 869-874.
  2. JE Edwards, HJ McQuay. Dipyrone and agranulocytosis: what is the risk? Lancet 2002 360: 1438.

previous or next story