I think there are several important issues here:
i) the 'one size fits all' approach may not be appropriate:
There are likely to be differences in risks and benefits of antibiotic treatent between the 'well-looking but thin' child who is either MAM or SAM, and complicated, sick children in terms of the degree of immune dysfunction, subclinical infection and reductive adaptation. I agree with Prof Golden that the true answer about uncomplicated 'walk-in' SAM is about rates of readmission and death. However, we do not yet have these data.
Antibiotics are not recommended for uncomplicated moderate acute malnutrition (MAM), but it seems unlikely that a child with MUAC 11.6cm is much different to one with MUAC 11.4cm if they do not have a clinical complaint.
There is an addiional question regarding HIV infected children who are already taking cotrimoxazole prophylaxis... do they need something more during uncomplicated SAM?
There is good evidence from other fields that short course antibiotics are effective in preventing serious infection in vulnerable children. Cotrimmoxazole prevented more than 90% of pneumonias in children with measles in Guinea Bissau: see
http://www.ncbi.nlm.nih.gov/sites/ppmc/articles/PMC1702442/
Malnutrition is the commonest cause of secondary immune deficiency worldwide. Epidemiological evidence of increased susceptibility to death from infectious disease (even in mild or moderate malnutrition) is overwhelming. The mechanisms are less clear. A key aspect is likely to be impaired barriers to infection at the skin and gut/respiratory mucosa. In terms of systemic immune deficiency, there are relatively few consistent findings. T cells responses such as delay type hypersensitivity (eg tuberculin test) do appear consistently impaired, but other mechanisms either have not been studied or have been studied at a time of active infection, making interpretation difficult.
In Kilifi, Kenya, we have used a demographhic surveillance system (DSS) to track survival following inpatient rehabiliation and treatment of complicated SAM. We have found a very high mortality from common infectious diseases for 6 months after treatment, which remains above that of the avergae population even at one year. Although this my reflect social disadvantage, we believe it is good evidence that these children have taken a severe 'hit' to immunity, which takes time to recover, probably longer than anthropometric recovery. These children were genrally admitted with very complicated SAM: septicaemia, shock and severe infections at presentation. However, subsequent mortality, even for MAM who were admitted becuase they were sick are well above baseline.
ii) These treatments must be tested in well designed clinical trials:
The questions are of enormous public health importance, but there are real risks of cost, logistics and antimicrobial resistance. For this reason, I support the trial that is going on in Malawi. If this eventually points to the need a large international trial, then that should be done. This is normal for major public health questions such as malaria treatment (eg recent Aquamat trial), malaria vaccine (current RTSS trial) or pneumococcal vaccines to need large, international and well funded trials. The same is true of nutritional interventions, especially in relation to cost-effectiveness. Evidence from such trials is the key to knowing what to do in practice, but also the key to harnessing donor and govenment support for implementation.
After seeing our post-discharge mortality data, we are currently conducting a randomised trial along the Kenyan coast of daily cotrimoxazole (septrin) prophylaxis among HIV-negative children with complicated SAM. They will receive the study drug (cotrimoxazole or placebo) at stabilisation and continue for 6 months, with follow up to 12 months. All other usual treatment and nutritional rehabilitation is given. The primary endpoint is mortality, secondary endpoints are readmission (including full microbiological work up), toxicity and antimicrobial resistance of carriage (nose and rectal) and disease-causing bacteria.
http://clinicaltrials.gov/ct2/show/NCT00934492
iii) There are two sides to the question of antimicrobial resistance:
a) To what extent does resistance reduce the efficacy of the antibiotic treatment? Again, there are likely to be differences between antibiotics. One key piece of data is that in the HIV trials, cotrimoxazole prophylaxis was highly effective, even in areas of high levels of cotrimoxazole resistance (see:
http://www.ncbi.nlm.nih.gov/pubmed/15555666 ). Thus, the inhibitory concentration of drug to prevent overt infection may be cosiderably lower than that needed to treat an established infection. Similarly, low doses of penicillins are effective in preventing infections in other type of immune deficiency (eg splenectomy)
b) To what extent does the widespread (necessary or unecessary) use of an antibiotic cause it? This differs between antibiotics and usage patterns. Ceftrixone, for example, is extremely good at rapidly causing resistance, both becuase of it's mode of action on bacteria and because it is recirculated through the bile system and reabsorbed in the gut exposing resident gut bacteria over it's long half life. Resistance genes can easily jump from resident to pathogenic bacteria in the gut.
In summary, these are questions of great public health importance that need to be tackled with adequately powered (large) multi centre trials with proper follow up, detection of death and readmission and the presence and important of antimicrobial resistance. As a community, we shoudl first be willing to acknowledge the areas where there are not good data from clinical trials to guide our practice, and secondly not shy away from doing the trials that are needed, as they are the only way these questions can be answered.