Dear Paul,
In terms of criteria for where RCT is not required, I would like to refer
to a paper by Nick Black which has tackled the issue for surgical
procedures:
Black N (1999). Evidence-based surgery: a passing fad? World J Surg 23,
789-793.
He suggests there are four main reasons: "experimantation may be
unnecessary, inappropriate, impossible, or inadequate", and offers examples
for those:
1.Unnecessary: when the effect is so dramatic that unkown confounding
factors could be ignored, like immobalisation of fractured bones. (I guess
insulin for IDDM falls in this category)
2. Inappropriate: a) where the outcome is rare
b) outcome happens far in the future and long follow-ups
are required. e.g. loosening of artificial hip joints
c)when randomisation could reduced effectiveness. For
example randomising patients for open surgery vs laparoscopic one; as
surgeons may be expert in one and not the others.
3. Impossible: a) clinicians refusal to participate.
b) ethical considerations e.g. cardiac transplantation vs
medical management
c) legal obstacles ; an example from US has been referred to.
d) difficulty of confronting research community
4. Inadequate: low external validity of many trials in surgery
As you could see it is a list of reasons why RCT may not be needed, also
why RCTs may not be able to answer the question. Both would probably lead
to the same conclusion, i.e. it is not needed.
I believe there are ways to overcome some of the situations that Black has
described, for example 2c one. The argument in 2c is very simialr to what
has been claimed on irrelevance of RCTs for assessing homeopathy.
Nevertheless, I found the paper very thoughtful and interesting, and I
think it tackles most of the possible scenarios.
Regards
Arash
--
Dr Arash Rashidian
Dept of Health Sciences
University of York
On Nov 13 2003, Paul Glasziou wrote:
> Dear All, Many thanks for your suggestions about what treatments don't
> need randomised trials. The most suggested was insulin for juvenile
> diabetes, but a number were suggested twice and many were already on Ben
> Djulbegovic's list (see below). There were only two (but very helpful)
> suggestions on criteria. So here is the full list - I don't guarantee all
> these are dramatic enough to qualify. In the next stage we plan to look
> at which treatments fit which of several possible criteria. In the
> meantime, if you have other suggestions, please let me know. Thanks to
> all for the excellent contributions,
>
> Paul Glasziou
> Department of Primary Health Care &
> Director, Centre for Evidence-Based Practice, Oxford
> ph: 44-1865-227055 www.cebm.net
> ---------------------
> 1. Dramatic therapies
>
> Endocrine
> * insulin in diabetic coma (2 people) insulin in diabetic
> keto-acidosis - survival
> * thyroid extract for myoedema - cure
> thyrostatics for thyrotoxicosis - cure
> * steroids in the treatment of adrenal cortical hypofunction?
>
> Hematological
> Hemotherapy to treat Hemophilia.
> * Blood transfusion in acute severe blood loss
>
> Nutritional
> * vitamin B12 in pernicious anemia; liver extract for PA and
> subacute combined - cure of PA
> citrus fruits to treat scurvy
>
> Infectious
> streptomycin for TB meningitis - return of consciousness and
> survival
> * penicillin in pneumonia (in bacterial endocarditis)
> gancyclovir for CMV retinitis in AIDs - complete arrest of the
> disease.
>
> Cardiovascular
> Defibrillation for ventricular fibrillation. People lived!
> automatic external defibrillators in cardiac arrest
>
> Musculoskeletal
> total hip and knee replacements
> Closed reduction of fracture of long bones with deformity?
>
> Neurological
> cholinesterase inhibitors for myastheia gravis - immediate return
> of muscle power
>
> Surgical
> appendicectomy in perforated appendicitis
> sewing up wounds,
> ether anesthetic
> maintaining sterile conditions for surgery
>
> Cancer
> * Imatinib in treatment of CML (chronic myeloid leukemia)
> * Combination chemotherapy with cisdiamminedichlorplatinum,
> vinblastine and bleomycine in disseminated testicular cancer.
> Other
> some organ transplantation (e..g liver transplantation in
> fulminant acute hepatitis)
> some antidotes in poisoning
> recompression for decompression illness.
>
> * From Ben Djulbegovic's list at:
> http://www.hsc.usf.edu/~bdjulbeg/oncology/practice-change.htm
>
> 2. suggest WHY they are so convincing that a trial is unnecessary
>
> Nino Cartabellotta suggested: We compiled a list of criteria for
> accepting in clinical practice health interventions without RCTs evidence
> (the criteria should be all satisfied): 1. bad outcome of disease if
> untreated (high/very high control event rate) 2. drammatic benefit of
> treatment (high relative risk reduction) 3. accettable side effects of
> treatment (high/very high number need to harm) 4. no alternative
> treatment 5. convincing physiopatological basis
>
> And Douglas Badenoch pointed out:
> The CEBM's Levels of evidence (www.cebm.net/levels_of_evidence.asp)
> includes a Level 1c for "All or none" case series in Therapy, Prognosis,
> Differential Diagnosis and Economic Analysis.
>
> This has been defined as "Met when all patients died before the
> [intervention] became available, but some now survive on it; or when some
> patients died before the [intervention] became available, but none now
> die on it."
>
> The questions would be, to my mind
> 1 How many is enough for such a case-series?
> 2 How well are the PICO elements of the research defined, measured
> and implemented in the study?
> 3 Do the PICO elements of the study relate to the PICO elements of
> your clinical question?
>
|