Estimating the Risk for Sepsis after Splenectomy in Hereditary Spherocytosis

Methods

Twenty-nine families, unrelated as far as is known, were studied over periods as long as 45 years. Interviews were done in 1993-1994 to gather information on 432 family members. Splenectomy had been done in 226 of 342 family members who were known to be affected with hereditary spherocytosis. Many affected patients who had been diagnosed as a result of a systematic study of families had elected not to have a splenectomy. The apparently high proportion of affected family members (342 of 432) is due to selective interviewing done on the basis of preexisting data [7, 8]. Participants were asked about hospitalization for serious infection. For the 33 persons who died after having had a splenectomy, family members were asked about the nature of the deaths or hospital records were reviewed. Dates of birth, splenectomy, and last contact or death were collected and used to estimate the risk for fatal sepsis after splenectomy for persons with hereditary spherocytosis. The age at which splenectomy had been done and the years of survival after splenectomy were calculated.

In calculating the rate of fatal infections, it was assumed that the death rate (hazard rate) was constant. Ninety-five percent CIs were calculated as described by Gardner and Altman [9].

Results

The rate of fatal infection for the 226 persons who had had a splenectomy was 0.73 per 1000 years (95% CI, 0.015 to 1.5). Four fatal fulminant infections occurred in four different families during 5461 years of life after splenectomy: A 5-year-old girl in family 203 died of sepsis 2 years after splenectomy; a 48-year-old woman in family 170 died of sepsis 30 years after splenectomy; a 31-year-old man in family 192 died a few hours into an acute febrile illness 18 years after splenectomy; and a 47-year-old man in family 193 died a few hours into an acute febrile illness 23 years after splenectomy. Splenectomy had been done in these four patients before 1967. Pneumococcal vaccine first became available some years later. None of the four was receiving prophylactic penicillin.

Table 1 contains the following information about families in my study: the number of persons who had had splenectomy; the number of years at risk after splenectomy; and the mean, median, and range of years after splenectomy.

One of the 35 children who had had a splenectomy before the age of 6 years died of sepsis (family 203). The 35 children have lived a total of 894 years after surgery; the mortality rate for children is thus 1.12 per 1000 years of life after splenectomy. The mortality rate for the 191 persons who were older than 6 years at splenectomy was 0.66 per 1000 years of life after splenectomy. The relative risk for young children compared with others was 1.70, but the 95% CI for this ratio is so wide (0.18 to 16) that no significance should be given to this estimate of relative risk.

Discussion

In addition to the four fatal infections, two near-fatal episodes of meningitis occurred in women aged 19 and 18 years who had had a splenectomy 8 and 11 years earlier, respectively (families 166 and 181). Because two of the persons who died were known to me before this survey, only two instances of fatal sepsis were discovered as a result of the survey. One of the fatal infections had been previously reported [8]. Although a rate of fatal sepsis of 0.73 per 1000 years of life after splenectomy may seem to represent a small risk, it is many times greater than that in the general population; such a rate in a city with a population of 200 000 would result in 146 deaths per year due to fulminant sepsis. I have not found an estimate of the rate of fatal sepsis in a population of healthy persons with spleens.

Because three of the four deaths due to sepsis occurred many years after splenectomy was done, these data do not support the belief that most overwhelming infections occur in the first few years after surgery. The process used to gather the data makes it unlikely that any fatal fulminant infections in asplenic persons were missed.

Because of genetic diversity, it is likely that the risk for sepsis after splenectomy varies among families. Family 193 had one fatal infection in 65 years at risk, and family 192 had one fatal infection in 602 years at risk. The ratio of these rates is 9.2. Because 29 families had a small number of events in a large number of years, it is not appropriate to conclude that family 193 is at greater risk than are other families. In fact, the 95% CI for the ratio is 0.58 to 145.

Because all four deaths occurred in patients who had had a splenectomy before pneumococcal vaccine was available, it is highly probable that the risk for sepsis in persons who have had a splenectomy in the last 15 years is less than that seen here. A study from Denmark found that the incidence of pneumococcal infection after splenectomy in children was dramatically lower after the introduction of pneumococcal vaccine and the promotion of early penicillin therapy for febrile children who have had a splenectomy [10].