1573
recording displacement of the tympanic membrane during stapedial reflex contraction elicited by a loud sound. High perilymphatic pressure displaces the resting position of the stapes footplate laterally and leads to a more inward-going tympanic displacement on stapedial contraction. A transducer probe attached to a headset is placed in the patient’s external auditory meatus and computer-based instrumentation allows the measurement of small movements of the tympanic membrane; 1000 Hz stimuli of increasing sound
EDITORALS
used to induce controlled stapes contraction. Patency of the cochlear aqueduct is checked by observing a change in mean tympanic membrane movement with change in posture from pressure level
Ear-wave to ventricular shunts Some readers may be surprised that intracranial pressure can be monitored via the ear. However, Marchbanks and his colleagues have developed a technique that could prove useful in hydrocephalic children with ventricular shunts.1 Ventricular shunts are fundamentally simple pieces of equipment consisting of plastic tubing, a one-way valve, and a reservoir or pumping device. It is sometimes very difficult to know whether a shunt is working or not, especially when children present with vague or intermittent symptoms-headaches, nausea, drowsiness, or blurred vision. Is the shunt blocked? Pumping of the shunt reservoir is an unreliable method of assessment2 and may even induce blockage in children with slit-like ventricles. When a shunt is blocked there may be virtually no change in ventricular size on computed tomography or magnetic resonance imaging. Injection of isotope into the shunt can confirm patency but gives no information about intracranial pressure, and this technique carries a small risk of introducing infection. Sequential transcranial doppler studies sometimes help but do not provide an absolute measure of pressure. Other techniques such as telethermography and telemetric pressure sensing are not of proven value. Thus, there is a need for a reliable non-invasive means of assessing shunt function. In 1989, Reid, Marchbanks, and colleagues3 reported a pilot study of their audiological technique for monitoring intracranial pressure. The basis of the method is the fact that intracranial pressure is transmitted via the cochlear aqueduct to the cochlear perilymph when the aqueduct is patent. Perilymphatic pressure can be assessed indirectly by
are
sitting to lying. The pilot study3 showed a relation between tympanic membrane displacement and mean intracranial pressure, defined either clinically or by direct measurement in 58 patients aged 5-77 years with various underlying conditions. The most consistent results were obtained in young patients with hydrocephalus and benign intracranial hypertension. In a subsequent study, Moss et all used the technique for long-term assessment of intracranial pressure in 43 hydrocephalic children aged 4-17 years with shunts; the patients were tested over a period of 18 months when they were clinically well. 11 (26%) of these children were later admitted with symptoms suggestive of acute shunt blockage. The tympanic intracranial pressure assessment correlated with clinical and/or operative findings in 10 cases. In 1 patient the tympanic method suggested raised pressure but the clinical assessment was that the shunt was patent (ie, there was one false-positive result). The advantage of the tympanic displacement technique is that it is non-invasive and quick. The procedure takes about 10 min and the results are available immediately. Moss et aP report that in 1 patient management was greatly helped by the twenty-two serial assessments that were done. There are disadvantages. A skilled audiologist is required to carry out the test and to interpret the result. The technique is not suitable for continuous intracranial pressure monitoring or for detecting waves of pressureand it does not measure absolute intracranial pressure. The procedure cannot be used in children with an absent stapedial reflex as a result of brain stem or middle ear dysfunction or in patients on ventilators because acoustic reflexes are absent whether or not muscle relaxants or sedation are used.3Restlessness will likewise make measurements difficult, and this could be important in younger children who might well object to the insertion of transducer probes into their ears. None of the other non-invasive techniques for assessment of shunt function is ideal, although
imaging can help to monitor flow in shunt tubing or ventricles. The tympanic membrane displacement technique is unlikely to be the complete answer, but in many dynamic magnetic
resonance
1574
children it may well provide a useful serial intracranial pressure abnormalities.
guide
to
-
SM, Marchbanks RJ, Burge DM. Long-term assessment of intracranial pressure using the tympanic membrane displacement measurement technique. Eur J Pediatr Surg 1991; 1 (suppl): 25-26. 2. Piatt JH. Physical examination of patients with cerebrospinal fluid shunts: is there useful information in pumping the shunt? Pediatrics 1. Moss
1992; 89: 470-73. 3. Reid A, Marchbanks RJ, Bateman DE, Martin AM, Brightwell AP, Pickard JD. Mean intracranial pressure monitoring by a non-invasive
audiological technique: a pilot study. J Neurol Neurosurg Psychiatry 1989; 52: 610-12.
Zelen
protocols
randomised trial, informed consent of the patient or guardian is usually obtained before allocation is made to one or other treatment. The subject has the right to be told the relevant risks and alternatives and to withhold consent, whatever the investigator believes are the benefits of participation.! To some clinicians this presents a difficulty in that the doctor may be perceived by the patient not to know the best choice, but recruitment is not ethical unless the experimental treatment is believed to be at least as effective as conventional therapy.Patients do not know at the time of consent which treatment they will receive and there is a chance that some will withdraw later if not given the treatment they then prefer. If clinicians are half-hearted in recruiting, or patients are resistant, the credibility of the trial suffers. In 1979, Zelen proposed alternative trial designs with randomisation before consent.3,4 In one form (double consent), all eligible patients are randomised to one of two possible treatments and are then asked for informed consent. If they agree, they are given their allotted treatment. If they do not, they are given the comparator or some suitable alternative. Clinicians may be more willing to approach suitable patients because they can discuss, at the outset, the merits of the specific treatment patients will receive in the trial. For the patient, the decision is simply whether or not to have the allotted treatment. Analysis is by intention to treat, irrespective of the treatment actually given, to avoid bias.4If there are few refusals, the reduction in the statistical power of the trial is offset by increased recruitment. However, if the overall refusal rate is 15%, twice the number of patients will have to be recruited.5 Double-blind trials are impossible with In
a
this approach. In another design (single consent), Zelen proposed that only patients allocated to the test treatment be asked for consent.3The comparator would be the best standard treatment, which any patient refusing consent would also receive. The notion that some patients would be studied without the investigators’ seeking consent provoked considerable controversy.5 The US Code of Federal Regulations and the EC Guidelines on Good Clinical Practice imply that, in any research on human beings, the subjects (or their guardian) should be informed of the risks and benefits and give consent.4,6 Patients in the control group
would still be part of a study, possibly requiring additional investigations, and their records would be reviewed, albeit for purely statistical purposes. In a review of trials in which these alternative designs had been adopted, Zelen found them of benefit in placebo-controlled studies and in comparisons of operations for breast cancer, an area
designs are unpopular.Other applications were less successful, and in one investigation the design was abandoned when the rate where randomised
of refusals reached 30%.4,5 Zelen noted two studies of extracorporeal membrane oxygenation in newborn babies with pulmonary hypertension in which consent was not sought from parents of babies allocated standard therapy; it was judged unethical to discuss a possibly life-saving procedure with the parents and then withhold it.7,8 In a randomised trial of arthroscopy of the knee vs medical management, recruitment with a conventional design yielded only 2 of 28 eligible patients in a year.9 Patients were then asked if they would accept arthroscopy before the randomisation envelope was opened. In the ensuing discussion with the patient, the assigned treatment was given prominence before consent was requested. Recruitment increased six-fold but half of the eligible patients declined before randomisation. Rifampicin has been shown in some cases to increase the bactericidal activity of antibiotic combinations including aminoglycosides against Pseudomonas spp in vitro and in laboratory anirnals.10 Korvick et al11 used the single-consent design in a trial of treatment of Pseudomonas bacteraemia with or without rifampicin in addition to a beta-lactam and an aminoglycoside. Patients allocated to the test group and already receiving standard therapy were asked for consent to the addition of rifampicin. 6 of 58 patients who were asked for consent refused and were given standard treatment but were analysed with the test group. Bacteriological failure was significantly less common in the group allocated to rifampicin (1/58 vs 9/63, p 0-02), irrespective of the severity of illness. The study design was said to be justified because recruitment was simplified and because no additional procedures were carried out on patients as the result of participation in the trial. Truly informed consent is seldom obtained, patients often failing to remember much of what they have been toldP However, in the Zelen designs, although the patient can refuse the allotted treatment, he or she cannot withdraw from the study and may not be asked for consent at all. The Zelen protocol increases recruitment when physicians are unwilling to participate in a conventional randomised study and few patients prefer one treatment over another. Opportunities for placebo-controlled trials are very limited. Assessments of new agents usually require investigations in addition to those made routinely, which could exclude the single-consent format. If recruitment in a randomised trial of antibiotics is slow =
recording displacement of the tympanic membrane during stapedial reflex contraction elicited by a loud sound. High perilymphatic pressure displaces the resting position of the stapes footplate laterally and leads to a more inward-going tympanic displacement on stapedial contraction. A transducer probe attached to a headset is placed in the patient’s external auditory meatus and computer-based instrumentation allows the measurement of small movements of the tympanic membrane; 1000 Hz stimuli of increasing sound
EDITORALS
used to induce controlled stapes contraction. Patency of the cochlear aqueduct is checked by observing a change in mean tympanic membrane movement with change in posture from pressure level
Ear-wave to ventricular shunts Some readers may be surprised that intracranial pressure can be monitored via the ear. However, Marchbanks and his colleagues have developed a technique that could prove useful in hydrocephalic children with ventricular shunts.1 Ventricular shunts are fundamentally simple pieces of equipment consisting of plastic tubing, a one-way valve, and a reservoir or pumping device. It is sometimes very difficult to know whether a shunt is working or not, especially when children present with vague or intermittent symptoms-headaches, nausea, drowsiness, or blurred vision. Is the shunt blocked? Pumping of the shunt reservoir is an unreliable method of assessment2 and may even induce blockage in children with slit-like ventricles. When a shunt is blocked there may be virtually no change in ventricular size on computed tomography or magnetic resonance imaging. Injection of isotope into the shunt can confirm patency but gives no information about intracranial pressure, and this technique carries a small risk of introducing infection. Sequential transcranial doppler studies sometimes help but do not provide an absolute measure of pressure. Other techniques such as telethermography and telemetric pressure sensing are not of proven value. Thus, there is a need for a reliable non-invasive means of assessing shunt function. In 1989, Reid, Marchbanks, and colleagues3 reported a pilot study of their audiological technique for monitoring intracranial pressure. The basis of the method is the fact that intracranial pressure is transmitted via the cochlear aqueduct to the cochlear perilymph when the aqueduct is patent. Perilymphatic pressure can be assessed indirectly by
are
sitting to lying. The pilot study3 showed a relation between tympanic membrane displacement and mean intracranial pressure, defined either clinically or by direct measurement in 58 patients aged 5-77 years with various underlying conditions. The most consistent results were obtained in young patients with hydrocephalus and benign intracranial hypertension. In a subsequent study, Moss et all used the technique for long-term assessment of intracranial pressure in 43 hydrocephalic children aged 4-17 years with shunts; the patients were tested over a period of 18 months when they were clinically well. 11 (26%) of these children were later admitted with symptoms suggestive of acute shunt blockage. The tympanic intracranial pressure assessment correlated with clinical and/or operative findings in 10 cases. In 1 patient the tympanic method suggested raised pressure but the clinical assessment was that the shunt was patent (ie, there was one false-positive result). The advantage of the tympanic displacement technique is that it is non-invasive and quick. The procedure takes about 10 min and the results are available immediately. Moss et aP report that in 1 patient management was greatly helped by the twenty-two serial assessments that were done. There are disadvantages. A skilled audiologist is required to carry out the test and to interpret the result. The technique is not suitable for continuous intracranial pressure monitoring or for detecting waves of pressureand it does not measure absolute intracranial pressure. The procedure cannot be used in children with an absent stapedial reflex as a result of brain stem or middle ear dysfunction or in patients on ventilators because acoustic reflexes are absent whether or not muscle relaxants or sedation are used.3Restlessness will likewise make measurements difficult, and this could be important in younger children who might well object to the insertion of transducer probes into their ears. None of the other non-invasive techniques for assessment of shunt function is ideal, although
imaging can help to monitor flow in shunt tubing or ventricles. The tympanic membrane displacement technique is unlikely to be the complete answer, but in many dynamic magnetic
resonance
1574
children it may well provide a useful serial intracranial pressure abnormalities.
guide
to
-
SM, Marchbanks RJ, Burge DM. Long-term assessment of intracranial pressure using the tympanic membrane displacement measurement technique. Eur J Pediatr Surg 1991; 1 (suppl): 25-26. 2. Piatt JH. Physical examination of patients with cerebrospinal fluid shunts: is there useful information in pumping the shunt? Pediatrics 1. Moss
1992; 89: 470-73. 3. Reid A, Marchbanks RJ, Bateman DE, Martin AM, Brightwell AP, Pickard JD. Mean intracranial pressure monitoring by a non-invasive
audiological technique: a pilot study. J Neurol Neurosurg Psychiatry 1989; 52: 610-12.
Zelen
protocols
randomised trial, informed consent of the patient or guardian is usually obtained before allocation is made to one or other treatment. The subject has the right to be told the relevant risks and alternatives and to withhold consent, whatever the investigator believes are the benefits of participation.! To some clinicians this presents a difficulty in that the doctor may be perceived by the patient not to know the best choice, but recruitment is not ethical unless the experimental treatment is believed to be at least as effective as conventional therapy.Patients do not know at the time of consent which treatment they will receive and there is a chance that some will withdraw later if not given the treatment they then prefer. If clinicians are half-hearted in recruiting, or patients are resistant, the credibility of the trial suffers. In 1979, Zelen proposed alternative trial designs with randomisation before consent.3,4 In one form (double consent), all eligible patients are randomised to one of two possible treatments and are then asked for informed consent. If they agree, they are given their allotted treatment. If they do not, they are given the comparator or some suitable alternative. Clinicians may be more willing to approach suitable patients because they can discuss, at the outset, the merits of the specific treatment patients will receive in the trial. For the patient, the decision is simply whether or not to have the allotted treatment. Analysis is by intention to treat, irrespective of the treatment actually given, to avoid bias.4If there are few refusals, the reduction in the statistical power of the trial is offset by increased recruitment. However, if the overall refusal rate is 15%, twice the number of patients will have to be recruited.5 Double-blind trials are impossible with In
a
this approach. In another design (single consent), Zelen proposed that only patients allocated to the test treatment be asked for consent.3The comparator would be the best standard treatment, which any patient refusing consent would also receive. The notion that some patients would be studied without the investigators’ seeking consent provoked considerable controversy.5 The US Code of Federal Regulations and the EC Guidelines on Good Clinical Practice imply that, in any research on human beings, the subjects (or their guardian) should be informed of the risks and benefits and give consent.4,6 Patients in the control group
would still be part of a study, possibly requiring additional investigations, and their records would be reviewed, albeit for purely statistical purposes. In a review of trials in which these alternative designs had been adopted, Zelen found them of benefit in placebo-controlled studies and in comparisons of operations for breast cancer, an area
designs are unpopular.Other applications were less successful, and in one investigation the design was abandoned when the rate where randomised
of refusals reached 30%.4,5 Zelen noted two studies of extracorporeal membrane oxygenation in newborn babies with pulmonary hypertension in which consent was not sought from parents of babies allocated standard therapy; it was judged unethical to discuss a possibly life-saving procedure with the parents and then withhold it.7,8 In a randomised trial of arthroscopy of the knee vs medical management, recruitment with a conventional design yielded only 2 of 28 eligible patients in a year.9 Patients were then asked if they would accept arthroscopy before the randomisation envelope was opened. In the ensuing discussion with the patient, the assigned treatment was given prominence before consent was requested. Recruitment increased six-fold but half of the eligible patients declined before randomisation. Rifampicin has been shown in some cases to increase the bactericidal activity of antibiotic combinations including aminoglycosides against Pseudomonas spp in vitro and in laboratory anirnals.10 Korvick et al11 used the single-consent design in a trial of treatment of Pseudomonas bacteraemia with or without rifampicin in addition to a beta-lactam and an aminoglycoside. Patients allocated to the test group and already receiving standard therapy were asked for consent to the addition of rifampicin. 6 of 58 patients who were asked for consent refused and were given standard treatment but were analysed with the test group. Bacteriological failure was significantly less common in the group allocated to rifampicin (1/58 vs 9/63, p 0-02), irrespective of the severity of illness. The study design was said to be justified because recruitment was simplified and because no additional procedures were carried out on patients as the result of participation in the trial. Truly informed consent is seldom obtained, patients often failing to remember much of what they have been toldP However, in the Zelen designs, although the patient can refuse the allotted treatment, he or she cannot withdraw from the study and may not be asked for consent at all. The Zelen protocol increases recruitment when physicians are unwilling to participate in a conventional randomised study and few patients prefer one treatment over another. Opportunities for placebo-controlled trials are very limited. Assessments of new agents usually require investigations in addition to those made routinely, which could exclude the single-consent format. If recruitment in a randomised trial of antibiotics is slow =
