Further prospective randomised controlled trial is necessary

Editor—Paula McDonald and colleagues have shown that a pneumococcal campaign based in general practice is feasible and offers the potential to increase substantially the proportion of patients at risk of pneumococcal infections who are vaccinated.1 Unfortunately, there remains a lack of robust evidence, particularly outcome data, to support pneumococcal vaccination in all the at risk groups identified in their paper. For example, one of the studies quoted by McDonald and colleagues fails to show any evidence of a protective effect of vaccination in patients with alcohol dependence or cirrhosis, sickle cell disease, chronic renal failure, or Hodgkin's disease.2 These conditions, however, are included in most lists of target groups for pneumococcal vaccination, including McDonald and colleagues'.

Butler et al's work also shows that the vaccine's efficacy in the conditions with a higher prevalence, such as congestive cardiac failure, may be as low as 17% or as high as 88% when confidence intervals are considered.2 Small sample sizes make conclusions difficult to draw.

Other work has relied on the development of pneumococcal antibodies rather than reductions in morbidity and mortality. We have collated details of 40 trials and 15 other papers on this topic. The general trends show that the groups at greatest risk from pneumococcal infection are those with the most impaired immune response. This potentially limits the usefulness of pneumococcal vaccination programmes.

Fine et al have published a meta-analysis of randomised controlled trials of pneumococcal vaccination, which McDonald and colleagues did not cite.3 They concluded that “evidence from randomized controlled trials fails to demonstrate vaccine efficacy for pneumococcal infection-related or other medical outcomes in the heterogeneous group of subjects currently labelled as high risk.”

A protocol for a systematic review of the literature in relation to pneumococcal vaccination has been posted in the Cochrane Database of Systematic Reviews.4 A further prospective randomised controlled trial is necessary. Such a trial will need to look at outcome measures reflecting reductions in infection, morbidity, and mortality, with subgroup analysis by underlying illness and age, to provide a definitive answer to whether McDonald and colleagues' primary care based model would save lives or waste large amounts of healthcare resources.

Expert advice is unclear

Editor—Paula McDonald and colleagues' study of pneumococcal vaccination in general practice shows what might be achieved by a targeted vaccination campaign.1 The authors state that they based their guidelines pragmatically on the best available advice. Although, like many other general practitioners, I am in favour of clinical guidelines,2 I have some concerns about the validity and practical application of recent expert advice on pneumococcal vaccination.

The Department of Health recommends vaccination for chronic lung disease, but whether this includes chronic asthma is not clear.3 In the same guidelines general practitioners are encouraged to give pneumococcal vaccine when immunising patients against influenza, as some practices in McDonald and colleagues' study would have preferred; patients with asthma are considered to be a target group for immunisation against influenza.3. In the United States, where pneumococcal vaccination is considered to be cost effective, it is recommended for asthmatic patients.4 I find it difficult to see why a patient with severe chronic asthma should be excluded from vaccination while a patient with mild chronic bronchitis is included. The authors did not reference the expert advice on which they based the exclusion of patients with asthma from pneumococcal vaccination in this study.

Another practical difficulty is that of reimmunisation. The Department of Health advises against reimmunisation except for patients with asplenia, hyposplenism, or the nephrotic syndrome,3 whereas a recent review suggested monitoring antibody concentrations three to four weeks after vaccination and then annually.5 If reimmunisation was necessary then there would be resource implications for practices and perhaps also a reluctance among patients to reattend if, as this study suggests, the rate of side effects from the vaccination was over 50%.

Poor uptake of vaccine may be due to poor knowledge or attitudes among doctors as well as patients. I accept that knowledge may be constantly changing, but effective vaccination programmes in primary care will be impeded by unclear expert advice.

Datasheet for vaccine contradicts authors' recommendations

Editor—Paula McDonald and colleagues include patients with immunodeficiency and immunosuppression as target groups for pneumococcal vaccination,1 in agreement with the Department of Health's guidelines.2 The current datasheet for the 23-valent vaccine (Pneumovax II; Pasteur Mérieux), however, states that the vaccine is contraindicated for patients less than 10 days before or during immunosuppressive treatment.3 As a further contraindication, the datasheet mentions patients with Hodgkin's disease who have received extensive chemotherapy or nodal irradiation, or both, whereas McDonald and colleagues specifically include patients with Hodgkin's disease as a target group.

Recommending a vaccine for groups in which it is contraindicated by the manufacturer puts prescribers in a difficult position; which advice do we follow?

Authors' reply

Editor—Neal Maskrey and Martin Parkinson suggest that there is not enough evidence to support the Department of Health's policy of vaccinating high risk groups. The meta-analysis of randomised controlled trials that they cite, however, provides little evidence to support or disprove their view1: the randomised controlled trials that included patients relevant to the Department of Health's guidelines found only four confirmed pneumococcal infections in total. Randomised controlled trials are not helpful in evaluating pneumococcal vaccine because no reliable outcome measure for pneumococcal pneumonia exists. This makes results difficult to interpret unless there is a clear difference between groups, as in the early trials in a population with a high incidence of the disease. Randomised controlled trials based on confirmed infections (bacteraemia and meningitis) require unfeasibly large numbers. The trial that Maskrey and Parkinson want to do would require 117 000 people in each risk group to show 50% efficacy against confirmed pneumococcal infections—for example, all HIV positive patients diagnosed in Britain for the next 39 years. Such a trial would be unlikely to get ethical approval, as the vaccine is already licensed for these groups.

Case-control studies and indirect cohort analysis are more powerful than randomised controlled trials in conditions with a low incidence and high morbidity. Several such studies have found evidence of efficacy of the vaccine against confirmed pneumococcal infections, including in some common at risk groups. Limited recruitment of patients has precluded definitive estimates of efficacy in some disorders, but there is a trend towards reduced efficacy in immunosuppressed patients. Pneumococcal infections, however, are serious, and costly to treat. Vaccination is cost effective in high risk conditions even if the vaccine has low efficacy and protection is short lived.2

We need a better test for pneumococcal infections. Meanwhile, an estimated 60–90 patients per health authority will die of pneumococcal infections annually. We can ignore this, or we can base our decisions on the best evidence available. In our view, it is reasonable to vaccinate groups known to be at high risk from pneumococcal infections if the efficacy of the vaccine is unknown but antibody response has been shown.

Recent evidence supports an age related policy for vaccination.3 4 This would simplify the situation, which A N Siriwardena suggests is necessary, but younger patients at high risk would still need vaccinating. We listed our advisers at the end of our paper.

Finally, we have asked the manufacturer to review its datasheet regarding Hodgkin's disease. Our protocol (not published) advised on timing of vaccination, as do the Department of Health's guidelines (widely available).

Small audit showed that only 14% of patients were offered pneumococcal vaccine

Editor—Paula McDonald and colleagues' paper on the implementation of a pneumococcal vaccination campaign highlighted not only the effectiveness of such a campaign but also the pressing need for such campaigns.1 Although the Department of Health recommends the use of pneumococcal vaccine in several high risk groups,2 the authors found that only 17% of general practitioners offered the vaccine before the campaign.

We have found a similar poor degree of compliance with these recommendations in a group of patients with chronic obstructive pulmonary disease in East Yorkshire. Patients with this disease comprise one of the high risk groups for which pneumococcal vaccination is recommended. In a random sample of 40 patients admitted to hospital in 1995 with an acute exacerbation of chronic obstructive pulmonary disease, records were available for 28. Only four of the 28 had been offered pneumococcal vaccine by their general practitioner at any time before that admission. All these patients had symptomatically severe chronic obstructive pulmonary disease, as shown by both their admission to hospital and their average of 7.2 primary care consultations for their chest condition in the year before their admission.

The low number of patients who were offered pneumococcal vaccine contrasted greatly with the finding that 22 had been offered influenza vaccine in the year before admission. Other work has also shown relatively good rates of acceptance of influenza vaccination in high risk groups.3 Although the sample size in our study was small, there was a striking difference between the immunisation rate with pneumococcal vaccine and that with influenza vaccine. Differences in the potential for remuneration might explain part of the difference between these rates.

McDonald and colleagues' paper suggests that further education could result in a similar proportion of high risk people receiving pneumococcal vaccine as receive influenza vaccine. We are therefore repeating our study to find out if the rate of pneumococcal immunisation has increased in the two years since the original audit. Further education and specific campaigns are needed in both primary and secondary care to increase compliance with the Department of Health's guidelines.