STATISTICS IN MEDICINE, VOL. 9,45-51 (1990)
SICKLE CELL ANAEMIA TRIAL MARILYN HUGHES GASTON National Institutes of Health, National Heart, Lung and Blood Institute, Division of Blood Diseases and Resources, Sickle Cell Disease Branch, 7550 Wisconsin Avenue, Room 504, Bethesda, Maryland 20892, U.S.A.
AND
JOEL VERTER National Institutes of Health, National Heart, Lung and Blood Institute, Division of Epidemiology and Clinical Applications, Biostatistics Research Branch, 7550 Wisconsin Avenue, Room 211, Bethesda, Maryland 20892, U.S.A
SUMMARY The major cause of death in sickle cell anaemia is from infection, especially infection caused by Streptococcus pneumoniae. Meningitis, pneumonia and septicaemia caused by this organism are the primary types of infection leading to death. Children under three years of age are at highest risk. We have known for over twenty years that approximately 30 per cent of the infants born with sickle cell anaemia will become infected in the first three years of life and one-third can be expected to die from the infection. These data were the reason that we conducted the Prophylactic Penicillin Study (PROPS),a trial to investigate the effectiveness of oral prophylactic penicillin in preventing severe infection due to S. pneumoniae. This investigation was a very efficient, cost effective study because of its timeliness and its conduct within the framework of an ongoing study. Moreover, the question being answered was simple and focused with upto-date data that permitted accurate estimates of sample size and incidence.
BACKGROUND We have known for over twenty years that babies born with sickle cell anaemia have an increased susceptibility to severe infections, specifically infection due to Streptococcus pneurnoniae. Infection, especially septicaemia and meningitis, caused by this organism, is the primary cause of death in We have also known that this illness. Children under three years of age are at highest approximately 30 per cent of these infants will become infected in the first three years of life and the ~ - ~illness is often fulminant, progressing from case fatality rate may be as high as 35 per ~ e n t . This the onset of fever to death in less than 12 hours.’** This was known when the Cooperative Study of Sickle Cell Disease (CSSCD) was initiated in 1978. This study, a nationwide, multicentre collaborative investigation, was designed to investigate the natural history of sickle cell disease from birth to death.’ A major objective of that study was to evaluate infection and mortality in children under five years old. Because of the availability of pneumococcal vaccines and improved programmes of care with early diagnosis of sickle cell anaemia, we assumed we would observe a clear reduction in the incidence and mortality from these infections. Much to everyone’s surprise, however, the rate of pneumococcal infection in
0277-6715/90/010045-07$05.00 0 1990 by John Wiley & Sons, Ltd.
46
M. H. GASTON AND J. VERTER
children under five had not changed in the past twenty years, but most importantly, the 30 per cent case fatality rate was alarmingly high." Therefore, we initiated the Prophylactic Penicillin Study (PROPS). METHODS
Organization The Sickle Cell Disease Branch, Division of Blood Diseases and Resources of the National Heart, Lung, and Blood Institute (NHLBI), established the Prophylactic Penicillin Study group in 1983. The group included 12 clinical centres participating in the Cooperative Study of Sickle Cell Disease and I1 other institutions outside the larger study. The CSSCD centres received no additional funding and the 1 1 outside institutions received no funding at all. There were central laboratories: the National Hemoglobinopathy Laboratory, Centers for Disease Control (CDC), provided the centralized diagnostic evaluation of haemoglobin within a larger programme receiving funding already. The Laboratory of Clinical Investigations, NIAID, NIH, measured antibody titres to S . pneurnoniae and urine penicillin levels, and serotyped isolated pneumococcal organisms. This laboratory was the sole expense of the study to the NHLBI. The trial was monitored by two NIH staff persons, one in the programme office of the Sickle Cell Disease Branch, and the other in the Biostatistics Research Branch who comprised the PROPS Data Coordinating Center. The existing policy board of the CSSCD served as the Data and Safety Monitoring Board for PROPS. DESIGN The PROPS was a multicentre, randomized, double-blind, placebo-controlled trial conducted after obtaining informed consent. PROPS, which utilized a common protocol,' was designed to test whether the administration of oral penicillin twice a day would reduce the incidence of documented severe bacterial infection in children with sickle cell anaemia under three years of age. The primary endpoint was a documented severe infection due to S. pneurnoniae, that is, meningitis, pneumonia, septicaemia. The secondary endpoint was a documented severe infection due to an organism other than S . pneumoniae, especially those due to Haemophilus injuenzae, to which these children are also susceptible. A documented infection was defined as an infection from which the organism was isolated and serotyped. Death was not an endpoint since the study did not have the power to detect a difference; however, its occurrence was monitored. STATISTICAL ANALYSIS AND SAMPLE SIZE On the basis of the CSSCD data which showed a one-year incidence of 10 per cent of pneumococcal septicaemia, it was estimated that the same incidence could be expected in the placebo group. Additional assumptions included in the sample size estimate were an 80 per cent reduction in the infection rate, a one-sided type I error rate of 0.05, and a power of 0.80 (type I1 error of 0.20). These assumptions resulted in a sample-size requirement of 108 patients in each study group (treatment and placebo), who had to be followed for a minimum of one year. Comparisons of baseline factors between groups were done with chi-square tests or t-tests, as appropriate. The Fisher exact test was used to compare the proportions of documented severe
47
SICKLE CELL ANAEMIA TRIAL PROPS
PRIMARY ENDPOINT
'I
"1
/
10
0
Placebo
4
a
12
16
20
MONTHS SINCE RANDOMIZATION N 105 0.0 Penicillin 110 0.0 placebo
102 0.01 108 0.02
91 0.02 96 0.06
76 0.02 78 0.09
64 0.02 61 0.12
Figure 1. Cumulative rate for all patients of S. pneumoniae Septicemia. Numbers of patients at risk and septicemia rates in the penicillin and placebo groups at 4-month intervals appear below the curves (one tailed P value =0.003)
infection in the two study groups. We used a one-tailed test of significance to evaluate this comparison, and this was appropriate given our design assumptions. Children were recruited from center's existing populations of children with sickle cell anaemia under three years of age and also from newly diagnosed babies at birth. They were randomized into an oral penicillin group receiving 125 mg tablets of penicillin V potassium twice a day, or a placebo group which received a 50 mg tablet of vitamin C. Both parents and centre staff were blinded. A common protocol was used. Patients were seen on entry into the trial and at subsequent visits arranged according to their routine schedule of well baby visits, that is, every three months. The only exception to the schedule was if an infection occurred.
RESULTS The trial was terminated eight months early after the occurrence of 15 episodes of pneumococcal septicaemia, 13 in the placebo group and 2 in the penicillin group. The observed 84 per cent reduction in the incidence of infection in the group treated with penicillin yielded a nominal onesided P value of 0.0025. The observed one-year incidence of pneumococcal septicaemia in the placebo group was 0.09. The penicillin group had an incidence of 0.02 (Figure 1). Although death was not a primary study end point, it must be noted that three deaths occurred in the placebo group and none in the penicillin group. An important finding was that four children in the placebo group had a fulminant course that in three cases proceeded from the onset of fever to death in less than nine hours. The fourth child also had a fulminant course and survived the disseminated intravascular coagulation to which the others succumbed; however, he had an associated cerebrovascular accident and remains alive, but with severe neurologic impairments.
48
M. H. GASTON AND J. VERTER
COST The total cost over the two year period for the central laboratory at NIAID to perform serotyping of organisms, pneumococcal antibodies and urine penicillin levels, was $75,000 dollars or approximately $350 per patient.
DISCUSSION This randomized, double-blind, multicentre trial demonstrated the effectiveness of prophylaxis with oral penicillin in significantly decreasing the incidence of pneumococcal septicaemia in children with sickle cell anaemia. The risk of septicaemia from S . pneurnoniae was decreased by 84 per cent (13 of 110 patients versus 2 of 105), and no deaths occurred in the group that received penicillin. Other important clinical implications include the fact that this infection must be prevented to be successful in decreasing mortality, because it can be so overwhelming that once a fulminant course is in progress, attempts to treat it are met with failure. Additionally, since this overwhelming infection can occur as early as three months of age, this trial demonstrates the need to diagnose babies at birth and initiate prophylactic penicillin by that age. The actual cost of $75,000 for the central laboratory underlines this trial as efficient and cost effective. Even if we include the donated personnel time from the clinical centres, the in-house costs to the Institute for programme administration, the mailing costs and the costs of the Data and Safety Monitoring Board, the total cost is approximately $350,000. The obvious reasons for our effective, low-cost study are that it was small with a simple, well focused question to be answered; it was conducted within an on-going investigation with one-half of the clinical centres in place. Trained, knowledgeable staff were already functioning effectively upon initiation of the study and all staff contributed to the design and implementation. In addition, most of the patients had already been identified so that a major effort for recruitment was not necessary. Moreover, the processes for entry into the study were already in place and working, that is, procedures for blood collection, form completion and mailing of forms and blood specimens and quality control. The efficiency was also enhanced by the fact that the statistical assumptions, design, and sample size were based on actual, and up-to-date data from the CSSCD which permitted accurate estimates of incidences and sample size estimates. The inclusion of the protocol into the usual routine well baby care being administered by all the paediatric centres added greatly to the efficiency and cost saving measures. Almost all of the data collected was during regularly scheduled well baby visits; additional data was kept to a minimum at entry, exit, at a certain age, and if an infection occurred. This permitted a streamlining of data collection, patient visits, laboratory requirements etc. Lastly, the co-operation and commitment of the investigators and the staff, and the willingness to participate without funding was related to the timeliness of the trial. The finding by the CSSCD that no change had occurred in the incidence and mortality over a twenty year period, despite the availability of pneumococcal vaccines, improved programmes of care, and better trained health care workers regarding sickle cell anaemia, underlined the necessity for an immediate trial to document the effectiveness of penicillin administered daily. All of these factors played major roles in the success of this inexpensive trial which is having an immeasurable public health impact across the country on newborn screening programmes in general and the survival of children with this disease in particular.
SICKLE CELL ANAEMIA TRIAL
49
REFERENCES 1. Robinson, M. G., and Watson, R. J. ‘Pneumococcal meningitis in sickle cell anemia’, New England Journal of Medicine, 274, 1006-1008 (1966). 2. Seeler, R. A., Metzger, W. and Mufson, M. A. ‘Diplococcus pneumoniae infecions in children with sickle cell anemia’, American Journal of Diseases of Children, 123, 8-10 (1972). 3. Barrett-Connor, E. ‘Bacterial infection and sickle cell anemia: an analysis of 250 infections in 166 patients and a review of the literature, Medicine (Baltimore), 50, 97-112 (1971). 4. Powars, D. R. ‘Natural history of sickle cell disease-the first ten years’, Seminars in Hematology, 21, 267-285 (1975). 5. Overturf, G. D., Powars, D. and Baraff, L. J. ‘Bacterial meningitis and senticemia in sickle cell disease’, American Journal of Diseases of Children, 131, 784-787 (1977). 6. Powars, D., Overturf, G., Weiss, J., Lee, S. and Chan, L. ‘Pneumococcal septicemia in children with sickle cell anemia: changing trend of survival’, Journal of the American Medical Association, 245, 1839-1842 (1981). 7. Kabins, S. A. and Lerner, C. ‘Fulminant pneumococcemia and sickle cell anemia’, Journal of the American Medical Association, 211, 4677471 (1970). 8. Seeler, R. A. ‘Deaths in children with sickle cell anemia: a clinical analysis of 19 fatal instances in Chicago’, Clinical Pediatrics, 11, 634-637 (1972). 9. Gaston, M. and Rosse, W. F. ‘The cooperative study of sickle cell disease: review of study design and objectives’, American Journal of Pediatric Hematology-Oncology, 4, 197-201 (1982). 10. Gaston, M. H. Falletta, J., Verter, J., Chilcote, R. and Wethers, D. ‘Infection in children with sickle cell disease’, Pediatric Research, 16, 204 (1982). 11. Gaston, M. H., Verter, J. I., Woods, G., et al. ‘Prophylaxis with oral penicillin in children with sickle cell anemia: a randomized trial’, New England Journd of Medicine, 314, 1593-1599 (1986).
DISCUSSION
Dr. Peduzzi: How did you arrive at an 80 per cent reduction in the infection rate? That is an extremely high estimate. Did you have any evidence for arriving at that or did you base the reduction upon the number of patients that you could get? You did indeed observe an 84 per cent reduction, so your assumption was on target.
Dr. Gaston: We just knew that penicillin should work. It is almost amazing that for 20 years we had the data on the problems with the pneumococcus and this disease, and nobody had ever done this trial before. Dr. Peduzzi: I have been involved in about half a dozen trial designs and every physician group that I have dealt with have always felt that their product would work but never to the level of 80 per cent.
Dr. Gaston: Infectious disease people have taught us how sensitive the pneumococcal organism is to penicillin and so we just shot for the moon. Dr. Hennekens: Whenever a clinician comes to me and is concerned about looking for large effects with an agent like streptokinase or r-TPA, I say, ‘We are not dealing with penicillin’. Dr. Friedewald: You had two classes of clinics - old and new. Did you see any difference in the performances of the two? Dr. Gaston: We did. Even though we trained the second group, it took them a while to get started. One of them had to establish a newborn screening programme. The new clinics were slower in their entry of patients and in returning forms. Overall, however, the quality of their data was similar to the old clinics.
50
M. H. GASTON AND J. VERTER
Dr. Friedewald: I know adherence was an issue in this trial. Could you summarize your experience? That would give some insight into the true effectiveness of penicillin. Dr. Gaston: That was the worst part of the trial and the most frustrating for us. We tried to monitor adherence by counting pills and by looking at the urine penicillin levels. Unfortunately, a third of the urine data was missing. We intended to get urine assays at routine visits and certainly if an infection occurred. Between visits we gave parents filter paper to put urine on and mail to the lab so we could check it. Hardly any of these urine collections were done. There was none done at the time of infection because you have this drastically ill baby in the emergency room and nobody has time to collect a urine in a crisis situation. We had a compelling result so that if compliance is as bad as we expect from other studies, perhaps taking pills every other day, or on the weekend is just enough penicillin to provide adequate levels. I think the fact that we gave it orally is the key. Perhaps giving penicillin every now and then does keep a relatively constant level. The fact that the organism is so clearly susceptible means that even if the compliance is erratic, the penicillin does not give the pneumococcus a chance to colonize and therefore we abort the infection. Possibly, the parent might start giving whatever pill was available at the first sign of infection. If the pill happens to be penicillin, that would mean starting penicillin very early in the couse of the illness, that might have been enough to make a difference. Dr. Tielsch: Suppose you hadn’t already built this incredible infrastructure, but your gut feeling was that the effect was going to be 80 per cent. Would a clinical trial be necessary or could you have not just got a couple of paediatricians together and just tried the treatment. Dr. Gaston: I don’t think we would have. We certainly have had several centres in the past trying to answer the question. I’m surprised nobody asked why we had to add centres in addition to those in the natural history study (CSSCD).When all the centres in the natural history study (CSSCD)saw the data, half ran out to put their children on penicillin and said that they were not participating in any kind of trial as it was unethical. The other half said these data pointed to the necessity of a trial to make sure the apparent effect was real. Therefore, a trial was necessary to change medical practice. Dr. Yusuf: Marilyn, yours was truly a remarkable trial. If you were not at NHLBI but you had exactly this idea and submitted a grant, I would suspect that the peer reviewers would shoot such an application to pieces: How could you expect an 80 per cent benefit? How can one person manage all the data? How can you get away with so little data? How can you persuade people to participate for free? Do you think you could have persuaded people to fund this trial? Dr. Gaston: No, I think we were very lucky that we were able to do the study. We took this on as part of our very busy schedules, the co-operation was excellent, the question to be answered was so important, and the required pieces were in place. Dr. Byar: How long do you recommend using prophylactic penicillin and how does that relate directly to the results of the study? Dr. Gaston: We don’t know when to stop it. As a matter of fact, we think that question is a very important one to answer. There are investigators who are not going to stop penicillin, even though the data from the CSSCD clearly show that the infection rate drops to about 2 per cent per year after three years of age and to 1 per cent after five years of age. Once a baby develops overwhelming pneumococcal infection nobody wants to stop penicillin. There is another issue that is even more critical. There are no data about the effects of starting penicillin at the age of two months. What is this doing to their ability to develop their own natural immunity to the
SICKLE CELL ANAEMIA TRIAL
51
pneumococcus? If you keep babies on penicillin for five years when the incidence goes markedly down, might you be impairing their ability to fight the pneumococcus when you stop the penicillin? Anecdotal data support the idea that when penicillin is stopped the incidence of infections goes up. So in another trial we are randomizing at age five into those continuing penicillin and those put on a placebo. Dr. Tognoni: Since you are so concerned about compliance, did you consider using benzathine penicillin instead of oral penicillin?
Dr. Gaston: We decided to use oral penicillin because we didn’t feel that in the real world we could give these tiny babies shots every three weeks for very long. It is difficult to find enough muscle mass repeatedly in small infants. We were also afraid parents would not continue to bring babies in to get these horrible shots every month and compliance would still be a problem. Dr. Slichter: You mentioned that the three children who died in the control group all had pneumococcal vaccine. Why is the pneumococcal vaccine not effective? Was the serotyping of the pneumococcal such that the pneumococcal vaccine did not cover the actual infection that the children had? Dr. Gaston: We now know that first of all the vaccine is ineffective in children under two. In addition, certain serotypes in the vaccine are poorly immunogenic and therefore infection results with the serotypes despite the vaccine. Dr. Ferris: A number of discussions have pointed out that sometimes clinics will participate in clinical trials at little or no cost. It is certainly gratifying to see that sometimes this works out successfully. Lest everyone wonder why we should pay for clinics if we can enroll them for free, let me describe some experiences we have had with clinics participating at no cost. We have more difficulty dealing with clinics that are participating for free, even though we try to limit the amount of data they collect. In studying eyes, we don’t have a simple endpoint like mortality that is rather easy to collect. We have had considerable difficulty assuring that the quality of the data, particularly visual acuity results, are adequate when we aren’t paying for a clinic co-ordinator or a visual function technician. Although having cIinics co-operate without compensation may work in some specific circumstances, that model certainly isn’t generalizable to all trials. Dr. Kathryn Davis: To follow up on Dr. Ferris’s comment, I have participated in two streptokinase trials with different funding mechanisms. The Western Washington Intracoronary Streptokinase Trial (WWIST I) was funded by this thousand-ponts-of-light modality while the intravenous trial (WWIST 11) was funded by NHLBI. The unfunded trial worked very well, but most hospitals are not willing to do this repeatedly. Although they may be enthusiastic once, I don’t think they would continue t o participate time after time in unfunded trials.
SICKLE CELL ANAEMIA TRIAL MARILYN HUGHES GASTON National Institutes of Health, National Heart, Lung and Blood Institute, Division of Blood Diseases and Resources, Sickle Cell Disease Branch, 7550 Wisconsin Avenue, Room 504, Bethesda, Maryland 20892, U.S.A.
AND
JOEL VERTER National Institutes of Health, National Heart, Lung and Blood Institute, Division of Epidemiology and Clinical Applications, Biostatistics Research Branch, 7550 Wisconsin Avenue, Room 211, Bethesda, Maryland 20892, U.S.A
SUMMARY The major cause of death in sickle cell anaemia is from infection, especially infection caused by Streptococcus pneumoniae. Meningitis, pneumonia and septicaemia caused by this organism are the primary types of infection leading to death. Children under three years of age are at highest risk. We have known for over twenty years that approximately 30 per cent of the infants born with sickle cell anaemia will become infected in the first three years of life and one-third can be expected to die from the infection. These data were the reason that we conducted the Prophylactic Penicillin Study (PROPS),a trial to investigate the effectiveness of oral prophylactic penicillin in preventing severe infection due to S. pneumoniae. This investigation was a very efficient, cost effective study because of its timeliness and its conduct within the framework of an ongoing study. Moreover, the question being answered was simple and focused with upto-date data that permitted accurate estimates of sample size and incidence.
BACKGROUND We have known for over twenty years that babies born with sickle cell anaemia have an increased susceptibility to severe infections, specifically infection due to Streptococcus pneurnoniae. Infection, especially septicaemia and meningitis, caused by this organism, is the primary cause of death in We have also known that this illness. Children under three years of age are at highest approximately 30 per cent of these infants will become infected in the first three years of life and the ~ - ~illness is often fulminant, progressing from case fatality rate may be as high as 35 per ~ e n t . This the onset of fever to death in less than 12 hours.’** This was known when the Cooperative Study of Sickle Cell Disease (CSSCD) was initiated in 1978. This study, a nationwide, multicentre collaborative investigation, was designed to investigate the natural history of sickle cell disease from birth to death.’ A major objective of that study was to evaluate infection and mortality in children under five years old. Because of the availability of pneumococcal vaccines and improved programmes of care with early diagnosis of sickle cell anaemia, we assumed we would observe a clear reduction in the incidence and mortality from these infections. Much to everyone’s surprise, however, the rate of pneumococcal infection in
0277-6715/90/010045-07$05.00 0 1990 by John Wiley & Sons, Ltd.
46
M. H. GASTON AND J. VERTER
children under five had not changed in the past twenty years, but most importantly, the 30 per cent case fatality rate was alarmingly high." Therefore, we initiated the Prophylactic Penicillin Study (PROPS). METHODS
Organization The Sickle Cell Disease Branch, Division of Blood Diseases and Resources of the National Heart, Lung, and Blood Institute (NHLBI), established the Prophylactic Penicillin Study group in 1983. The group included 12 clinical centres participating in the Cooperative Study of Sickle Cell Disease and I1 other institutions outside the larger study. The CSSCD centres received no additional funding and the 1 1 outside institutions received no funding at all. There were central laboratories: the National Hemoglobinopathy Laboratory, Centers for Disease Control (CDC), provided the centralized diagnostic evaluation of haemoglobin within a larger programme receiving funding already. The Laboratory of Clinical Investigations, NIAID, NIH, measured antibody titres to S . pneurnoniae and urine penicillin levels, and serotyped isolated pneumococcal organisms. This laboratory was the sole expense of the study to the NHLBI. The trial was monitored by two NIH staff persons, one in the programme office of the Sickle Cell Disease Branch, and the other in the Biostatistics Research Branch who comprised the PROPS Data Coordinating Center. The existing policy board of the CSSCD served as the Data and Safety Monitoring Board for PROPS. DESIGN The PROPS was a multicentre, randomized, double-blind, placebo-controlled trial conducted after obtaining informed consent. PROPS, which utilized a common protocol,' was designed to test whether the administration of oral penicillin twice a day would reduce the incidence of documented severe bacterial infection in children with sickle cell anaemia under three years of age. The primary endpoint was a documented severe infection due to S. pneurnoniae, that is, meningitis, pneumonia, septicaemia. The secondary endpoint was a documented severe infection due to an organism other than S . pneumoniae, especially those due to Haemophilus injuenzae, to which these children are also susceptible. A documented infection was defined as an infection from which the organism was isolated and serotyped. Death was not an endpoint since the study did not have the power to detect a difference; however, its occurrence was monitored. STATISTICAL ANALYSIS AND SAMPLE SIZE On the basis of the CSSCD data which showed a one-year incidence of 10 per cent of pneumococcal septicaemia, it was estimated that the same incidence could be expected in the placebo group. Additional assumptions included in the sample size estimate were an 80 per cent reduction in the infection rate, a one-sided type I error rate of 0.05, and a power of 0.80 (type I1 error of 0.20). These assumptions resulted in a sample-size requirement of 108 patients in each study group (treatment and placebo), who had to be followed for a minimum of one year. Comparisons of baseline factors between groups were done with chi-square tests or t-tests, as appropriate. The Fisher exact test was used to compare the proportions of documented severe
47
SICKLE CELL ANAEMIA TRIAL PROPS
PRIMARY ENDPOINT
'I
"1
/
10
0
Placebo
4
a
12
16
20
MONTHS SINCE RANDOMIZATION N 105 0.0 Penicillin 110 0.0 placebo
102 0.01 108 0.02
91 0.02 96 0.06
76 0.02 78 0.09
64 0.02 61 0.12
Figure 1. Cumulative rate for all patients of S. pneumoniae Septicemia. Numbers of patients at risk and septicemia rates in the penicillin and placebo groups at 4-month intervals appear below the curves (one tailed P value =0.003)
infection in the two study groups. We used a one-tailed test of significance to evaluate this comparison, and this was appropriate given our design assumptions. Children were recruited from center's existing populations of children with sickle cell anaemia under three years of age and also from newly diagnosed babies at birth. They were randomized into an oral penicillin group receiving 125 mg tablets of penicillin V potassium twice a day, or a placebo group which received a 50 mg tablet of vitamin C. Both parents and centre staff were blinded. A common protocol was used. Patients were seen on entry into the trial and at subsequent visits arranged according to their routine schedule of well baby visits, that is, every three months. The only exception to the schedule was if an infection occurred.
RESULTS The trial was terminated eight months early after the occurrence of 15 episodes of pneumococcal septicaemia, 13 in the placebo group and 2 in the penicillin group. The observed 84 per cent reduction in the incidence of infection in the group treated with penicillin yielded a nominal onesided P value of 0.0025. The observed one-year incidence of pneumococcal septicaemia in the placebo group was 0.09. The penicillin group had an incidence of 0.02 (Figure 1). Although death was not a primary study end point, it must be noted that three deaths occurred in the placebo group and none in the penicillin group. An important finding was that four children in the placebo group had a fulminant course that in three cases proceeded from the onset of fever to death in less than nine hours. The fourth child also had a fulminant course and survived the disseminated intravascular coagulation to which the others succumbed; however, he had an associated cerebrovascular accident and remains alive, but with severe neurologic impairments.
48
M. H. GASTON AND J. VERTER
COST The total cost over the two year period for the central laboratory at NIAID to perform serotyping of organisms, pneumococcal antibodies and urine penicillin levels, was $75,000 dollars or approximately $350 per patient.
DISCUSSION This randomized, double-blind, multicentre trial demonstrated the effectiveness of prophylaxis with oral penicillin in significantly decreasing the incidence of pneumococcal septicaemia in children with sickle cell anaemia. The risk of septicaemia from S . pneurnoniae was decreased by 84 per cent (13 of 110 patients versus 2 of 105), and no deaths occurred in the group that received penicillin. Other important clinical implications include the fact that this infection must be prevented to be successful in decreasing mortality, because it can be so overwhelming that once a fulminant course is in progress, attempts to treat it are met with failure. Additionally, since this overwhelming infection can occur as early as three months of age, this trial demonstrates the need to diagnose babies at birth and initiate prophylactic penicillin by that age. The actual cost of $75,000 for the central laboratory underlines this trial as efficient and cost effective. Even if we include the donated personnel time from the clinical centres, the in-house costs to the Institute for programme administration, the mailing costs and the costs of the Data and Safety Monitoring Board, the total cost is approximately $350,000. The obvious reasons for our effective, low-cost study are that it was small with a simple, well focused question to be answered; it was conducted within an on-going investigation with one-half of the clinical centres in place. Trained, knowledgeable staff were already functioning effectively upon initiation of the study and all staff contributed to the design and implementation. In addition, most of the patients had already been identified so that a major effort for recruitment was not necessary. Moreover, the processes for entry into the study were already in place and working, that is, procedures for blood collection, form completion and mailing of forms and blood specimens and quality control. The efficiency was also enhanced by the fact that the statistical assumptions, design, and sample size were based on actual, and up-to-date data from the CSSCD which permitted accurate estimates of incidences and sample size estimates. The inclusion of the protocol into the usual routine well baby care being administered by all the paediatric centres added greatly to the efficiency and cost saving measures. Almost all of the data collected was during regularly scheduled well baby visits; additional data was kept to a minimum at entry, exit, at a certain age, and if an infection occurred. This permitted a streamlining of data collection, patient visits, laboratory requirements etc. Lastly, the co-operation and commitment of the investigators and the staff, and the willingness to participate without funding was related to the timeliness of the trial. The finding by the CSSCD that no change had occurred in the incidence and mortality over a twenty year period, despite the availability of pneumococcal vaccines, improved programmes of care, and better trained health care workers regarding sickle cell anaemia, underlined the necessity for an immediate trial to document the effectiveness of penicillin administered daily. All of these factors played major roles in the success of this inexpensive trial which is having an immeasurable public health impact across the country on newborn screening programmes in general and the survival of children with this disease in particular.
SICKLE CELL ANAEMIA TRIAL
49
REFERENCES 1. Robinson, M. G., and Watson, R. J. ‘Pneumococcal meningitis in sickle cell anemia’, New England Journal of Medicine, 274, 1006-1008 (1966). 2. Seeler, R. A., Metzger, W. and Mufson, M. A. ‘Diplococcus pneumoniae infecions in children with sickle cell anemia’, American Journal of Diseases of Children, 123, 8-10 (1972). 3. Barrett-Connor, E. ‘Bacterial infection and sickle cell anemia: an analysis of 250 infections in 166 patients and a review of the literature, Medicine (Baltimore), 50, 97-112 (1971). 4. Powars, D. R. ‘Natural history of sickle cell disease-the first ten years’, Seminars in Hematology, 21, 267-285 (1975). 5. Overturf, G. D., Powars, D. and Baraff, L. J. ‘Bacterial meningitis and senticemia in sickle cell disease’, American Journal of Diseases of Children, 131, 784-787 (1977). 6. Powars, D., Overturf, G., Weiss, J., Lee, S. and Chan, L. ‘Pneumococcal septicemia in children with sickle cell anemia: changing trend of survival’, Journal of the American Medical Association, 245, 1839-1842 (1981). 7. Kabins, S. A. and Lerner, C. ‘Fulminant pneumococcemia and sickle cell anemia’, Journal of the American Medical Association, 211, 4677471 (1970). 8. Seeler, R. A. ‘Deaths in children with sickle cell anemia: a clinical analysis of 19 fatal instances in Chicago’, Clinical Pediatrics, 11, 634-637 (1972). 9. Gaston, M. and Rosse, W. F. ‘The cooperative study of sickle cell disease: review of study design and objectives’, American Journal of Pediatric Hematology-Oncology, 4, 197-201 (1982). 10. Gaston, M. H. Falletta, J., Verter, J., Chilcote, R. and Wethers, D. ‘Infection in children with sickle cell disease’, Pediatric Research, 16, 204 (1982). 11. Gaston, M. H., Verter, J. I., Woods, G., et al. ‘Prophylaxis with oral penicillin in children with sickle cell anemia: a randomized trial’, New England Journd of Medicine, 314, 1593-1599 (1986).
DISCUSSION
Dr. Peduzzi: How did you arrive at an 80 per cent reduction in the infection rate? That is an extremely high estimate. Did you have any evidence for arriving at that or did you base the reduction upon the number of patients that you could get? You did indeed observe an 84 per cent reduction, so your assumption was on target.
Dr. Gaston: We just knew that penicillin should work. It is almost amazing that for 20 years we had the data on the problems with the pneumococcus and this disease, and nobody had ever done this trial before. Dr. Peduzzi: I have been involved in about half a dozen trial designs and every physician group that I have dealt with have always felt that their product would work but never to the level of 80 per cent.
Dr. Gaston: Infectious disease people have taught us how sensitive the pneumococcal organism is to penicillin and so we just shot for the moon. Dr. Hennekens: Whenever a clinician comes to me and is concerned about looking for large effects with an agent like streptokinase or r-TPA, I say, ‘We are not dealing with penicillin’. Dr. Friedewald: You had two classes of clinics - old and new. Did you see any difference in the performances of the two? Dr. Gaston: We did. Even though we trained the second group, it took them a while to get started. One of them had to establish a newborn screening programme. The new clinics were slower in their entry of patients and in returning forms. Overall, however, the quality of their data was similar to the old clinics.
50
M. H. GASTON AND J. VERTER
Dr. Friedewald: I know adherence was an issue in this trial. Could you summarize your experience? That would give some insight into the true effectiveness of penicillin. Dr. Gaston: That was the worst part of the trial and the most frustrating for us. We tried to monitor adherence by counting pills and by looking at the urine penicillin levels. Unfortunately, a third of the urine data was missing. We intended to get urine assays at routine visits and certainly if an infection occurred. Between visits we gave parents filter paper to put urine on and mail to the lab so we could check it. Hardly any of these urine collections were done. There was none done at the time of infection because you have this drastically ill baby in the emergency room and nobody has time to collect a urine in a crisis situation. We had a compelling result so that if compliance is as bad as we expect from other studies, perhaps taking pills every other day, or on the weekend is just enough penicillin to provide adequate levels. I think the fact that we gave it orally is the key. Perhaps giving penicillin every now and then does keep a relatively constant level. The fact that the organism is so clearly susceptible means that even if the compliance is erratic, the penicillin does not give the pneumococcus a chance to colonize and therefore we abort the infection. Possibly, the parent might start giving whatever pill was available at the first sign of infection. If the pill happens to be penicillin, that would mean starting penicillin very early in the couse of the illness, that might have been enough to make a difference. Dr. Tielsch: Suppose you hadn’t already built this incredible infrastructure, but your gut feeling was that the effect was going to be 80 per cent. Would a clinical trial be necessary or could you have not just got a couple of paediatricians together and just tried the treatment. Dr. Gaston: I don’t think we would have. We certainly have had several centres in the past trying to answer the question. I’m surprised nobody asked why we had to add centres in addition to those in the natural history study (CSSCD).When all the centres in the natural history study (CSSCD)saw the data, half ran out to put their children on penicillin and said that they were not participating in any kind of trial as it was unethical. The other half said these data pointed to the necessity of a trial to make sure the apparent effect was real. Therefore, a trial was necessary to change medical practice. Dr. Yusuf: Marilyn, yours was truly a remarkable trial. If you were not at NHLBI but you had exactly this idea and submitted a grant, I would suspect that the peer reviewers would shoot such an application to pieces: How could you expect an 80 per cent benefit? How can one person manage all the data? How can you get away with so little data? How can you persuade people to participate for free? Do you think you could have persuaded people to fund this trial? Dr. Gaston: No, I think we were very lucky that we were able to do the study. We took this on as part of our very busy schedules, the co-operation was excellent, the question to be answered was so important, and the required pieces were in place. Dr. Byar: How long do you recommend using prophylactic penicillin and how does that relate directly to the results of the study? Dr. Gaston: We don’t know when to stop it. As a matter of fact, we think that question is a very important one to answer. There are investigators who are not going to stop penicillin, even though the data from the CSSCD clearly show that the infection rate drops to about 2 per cent per year after three years of age and to 1 per cent after five years of age. Once a baby develops overwhelming pneumococcal infection nobody wants to stop penicillin. There is another issue that is even more critical. There are no data about the effects of starting penicillin at the age of two months. What is this doing to their ability to develop their own natural immunity to the
SICKLE CELL ANAEMIA TRIAL
51
pneumococcus? If you keep babies on penicillin for five years when the incidence goes markedly down, might you be impairing their ability to fight the pneumococcus when you stop the penicillin? Anecdotal data support the idea that when penicillin is stopped the incidence of infections goes up. So in another trial we are randomizing at age five into those continuing penicillin and those put on a placebo. Dr. Tognoni: Since you are so concerned about compliance, did you consider using benzathine penicillin instead of oral penicillin?
Dr. Gaston: We decided to use oral penicillin because we didn’t feel that in the real world we could give these tiny babies shots every three weeks for very long. It is difficult to find enough muscle mass repeatedly in small infants. We were also afraid parents would not continue to bring babies in to get these horrible shots every month and compliance would still be a problem. Dr. Slichter: You mentioned that the three children who died in the control group all had pneumococcal vaccine. Why is the pneumococcal vaccine not effective? Was the serotyping of the pneumococcal such that the pneumococcal vaccine did not cover the actual infection that the children had? Dr. Gaston: We now know that first of all the vaccine is ineffective in children under two. In addition, certain serotypes in the vaccine are poorly immunogenic and therefore infection results with the serotypes despite the vaccine. Dr. Ferris: A number of discussions have pointed out that sometimes clinics will participate in clinical trials at little or no cost. It is certainly gratifying to see that sometimes this works out successfully. Lest everyone wonder why we should pay for clinics if we can enroll them for free, let me describe some experiences we have had with clinics participating at no cost. We have more difficulty dealing with clinics that are participating for free, even though we try to limit the amount of data they collect. In studying eyes, we don’t have a simple endpoint like mortality that is rather easy to collect. We have had considerable difficulty assuring that the quality of the data, particularly visual acuity results, are adequate when we aren’t paying for a clinic co-ordinator or a visual function technician. Although having cIinics co-operate without compensation may work in some specific circumstances, that model certainly isn’t generalizable to all trials. Dr. Kathryn Davis: To follow up on Dr. Ferris’s comment, I have participated in two streptokinase trials with different funding mechanisms. The Western Washington Intracoronary Streptokinase Trial (WWIST I) was funded by this thousand-ponts-of-light modality while the intravenous trial (WWIST 11) was funded by NHLBI. The unfunded trial worked very well, but most hospitals are not willing to do this repeatedly. Although they may be enthusiastic once, I don’t think they would continue t o participate time after time in unfunded trials.
