Clinical Ethics in Anesthesiology. A Case-Based Textbook
4. Research and publication
32. The ethics of research on pain and other symptoms for which effective treatments already exist
Monica Escher and Samia Hurst
Dr. Smith is an anesthesia resident on the hospital “pain service,” who has been asked to consult on a patient who has been admitted for palliative care for terminal colon cancer. The patient has had breakthrough pain on maximal oral therapy. He feels nauseous and limits his oral intake. Subcutaneous (SQ) administration of morphine is initiated, and brings partial relief. The patient requests epidural analgesia. The attending on the pain service is conducting a placebo-controlled crossover research study evaluating the efficacy of administration of SQ recombinant human hyaluronidase to improve the absorption of SQ morphine in patients with terminal illness. Each study subject will receive morphine via a different method on each of 3 days: intravenous morphine, SQ morphine plus placebo, and SQ morphine plus hyaluronidase. As part of her role on the pain service, Dr. Smith is supposed to help recruit clinical research subjects, but she is bothered by the fact that this patient is terminally ill, is in considerable pain, and is requesting a specific therapy that is not on the research protocol. Is it ethical for her to try to recruit him for this study?
Research on the management of pain and other symptoms is crucial to provide better care for acute, postoperative, and chronic symptoms, as well as better symptom management at the end of life. However, participation in clinical research places subjects at risk of harm for the benefit of others. This tension is intrinsic to all clinical research, and underlies the need for protection of human subjects. A number of national and international regulatory documents aim to protect participants in research while also taking into account the interests of future patients by allowing research to be conducted1 All refer to similar basic principles, which have been synthesized thus: social value, scientific validity, fair subject selection, favorable risk/benefit ratio, independent review, informed consent, and respect for enrolled participants. All of the requirements are equally necessary, and each is relevant to research on the management of pain and other symptoms, which requires the same ethical protections as any other form of research with human subjects. Some aspects, however, do present difficulties more specific to the context of symptom management studies. These include difficulties related to scientific methodology, fair subject selection, obtaining a favorable risk/benefit ratio, and informed consent.
Although individual patients may benefit indirectly from research participation, the intention of research is to generate valid data to inform the care of future patients. The research may involve delaying pain relief in participants. The investigator is responsible for minimizing risks and avoiding unnecessary harm for the patients. He/she is also accountable for the scientific validity and clinical relevance of the experimental question. Methodological issues in the design of symptom management trials, the collection of adverse effects, and the reporting of data, have ethical importance. Failure to give proper attention to these issues can make the results of a trial difficult to interpret and prevent comparisons. This limits usefulness in clinical practice, causing research to fall short of its purported goal and fail to fulfill its commitments towards patients and society. Risks to human subjects must be justified in part by the social benefit of research. Therefore any risk, however small, in a study that cannot answer its research question is excessive.
Randomized, double-blind trials have become standard in acute as well as in chronic pain trials. The inclusion of patients with moderate to severe pain at baseline may be crucial for sensitivity, i.e., the ability to detect the analgesic effect of the tested drug. Primary outcomes are usually a difference in pain intensity and a measure of pain relief. Analgesic consumption after a surgical procedure has also been used as a surrogate endpoint because it can prove more convenient to measure. Data other than pain intensity and pain relief are also useful in characterizing patients’ responses. In acute pain trials onset and duration of analgesia, the time to rescue analgesia and the number of patients requiring a rescue dose at various time points all provide valuable information. Observation periods must be long enough to gather relevant data. As chronic pain impacts different aspects of the patient’s life, decreasing health-related quality of life, chronic pain trials must address issues considered important by patients, such as physical and emotional functioning, sleep, relations to family, and social activities. Whenever available, a standard set of outcome measures should be used to assess these dimensions. Studies should also assess what magnitude of change brings a meaningful difference for the patient, as opposed to finding a statistical difference only. Defining a “meaningful difference” is therefore necessary. Such an approach enables researchers to determine which patients benefit from treatment and to perform responder analyses.
Overall patient satisfaction depends on treatment efficacy in alleviating pain, and on treatment adverse effects. Single-dose studies do not adequately characterize drug toxicity because most side effects are dose dependent and will occur only after repeated dosing. Careful monitoring of adverse effects is warranted in all pain trials to assess the balance between risks and benefits in the specific clinical context. It also helps interpret the results in trials where analgesic consumption is an outcome, as sedation, confusion, or nausea may all lead to reduced opioid use without achievement of adequate pain relief. Whatever the method, the type, frequency, and severity of adverse events should be recorded for placebo and active treatment, and the causal relationship between adverse events and the study drug should be evaluated.
For results to be readily understandable, data must be reported in a way that relates to clinical practice. Data on pain intensity and pain relief do not have a normal distribution but are highly skewed, in acute pain as well as in chronic pain trials. Giving mean values does not reflect clinical reality as some patients may benefit considerably from the intervention, while others do not experience any improvement. Analysis of individual patient data has been proposed as a more relevant way of reporting results.2 In addition to absolute changes in pain intensity, percent change provides valuable information, and correlates with patient global satisfaction.3
Fair subject recruitment
Potential subjects should be selected based on the potential for generalizability of study results, and to maximize the risk/benefit ratio for enrolled subjects. They should not be recruited on the basis of misleading expectations, lesser ability to defend themselves, or because they are persons whose risk is somehow discounted.
The problem of misleading expectations
A major expectation of informed consent in clinical research is to help patients understand the difference between the goals of the research and those of medical care – the research goal is to acquire generalizable knowledge that will help improve the care of future patients, rather than providing the best care to individual patients. Chronic pain patients have often received treatments that did not satisfy them, and may welcome access to a novel treatment they believe will work better. This makes them particularly prone to “therapeutic misconception.” Their hope for personal benefit can also lead them to minimize or overlook inconveniences and risks. Most chronic pain trials involve fixed-schedule drug titration, prohibit or limit the use of other pain medication, and impose constraints on nonpharmacological and alternative treatments. Moreover, chronic pain patients often experience impaired physical, emotional and social functioning, all of which have been associated with the therapeutic misconception.4
Therapeutic misconception is also ethically problematic because it represents misplaced trust, which can endanger the physician–patient relationship.5 Participants may question the researcher’s competence and blame him if their expectations are not fulfilled. This risk may be especially high with patients suffering from chronic pain, as it is difficult to treat, with even the best treatments available providing only partial relief. Moreover, patients may experience the side effects of medication without benefiting from its therapeutic effect. Patients’ distrust towards physician investigators can extend and impact on their relationship with their own physician, thus possibly jeopardizing routine clinical care.
Is research at the end of life a special case?
The exclusion of terminally ill patients from research would not be easy to defend. Research is useful to improve therapies at the end of life and thus help the very people whom such a ban would attempt to protect. Terminally ill patients do enroll in studies, and research has shown that, when given hypothetical scenarios, a majority of palliative care patients want to participate in research6 with motives including the hope of personal benefit, altruism towards patients going through the same ordeals7 and the wish to maintain hope.8 Moreover, terminal disease does not by itself invalidate the capacity for informed consent. Were it so, we would not routinely respect the choices expressed by the terminally ill in their wills.
The conduct of research with terminally ill patients is, however, often described as requiring special care due to risks of coercion or exploitation. Coercion is best described as a credible and strong threat exerted by a person that limits the options in a negative way available to another person. Having limited options through no fault of anyone’s does not constitute coercion. Indeed, the possibility of enrolling in a clinical trial actually gives terminally ill patients an additional option.
Exploitation is the unfair distribution of the benefits and burdens of a transaction. Research participation always carries burdens. Although it can carry the prospect both of direct benefits to subjects and of benefits to their community from the social value of research, both are always uncertain in a given study. Are terminally ill patients at greater risk during research, and/or are they less likely to benefit from research participation than they should? Assessing risks and benefits of participation in research for terminally ill patients is a complicated question. Effects deemed sufficient by researchers – and which underlie the choice of research question and much of the methodology – may or may not reflect what terminally ill patients would consider a clinically desirable effect.9 A patient’s evaluation of both risks and benefits may shift during terminal illness, and that shift will no doubt vary between individuals. At minimum, the risks incurred by terminally ill patients should never be discounted, and the risk/benefit assessment must take into account the circumstances of terminal disease.
A favorable risk/benefit ratio
The use of a placebo control when a recognized effective treatment already exists is controversial. On the one hand, giving a placebo to subjects randomized to the control group deprives them of an effective treatment, leading to recommendations against placebo controls in such standards such as the Declaration of Helsinki.10 On the other hand, using a recognized treatment as a control can lead to methodological difficulties. A positive noninferiority trial can mean three very different things: both treatments are equally effective, both are equally ineffective, or the trial was underpowered to detect the difference. This tension was recognized in the 2008 Helsinki declaration, which states that placebo controls can be used “where for compelling and scientifically sound methodological reasons the use of placebo is necessary to determine the efficacy or safety of an intervention and the patients who receive placebo or no treatment will not be subject to any risk of serious or irreversible harm.” It then adds that “Extreme care must be taken to avoid abuse of this option.”
How do we tell the difference between those situations where a placebo control is acceptable, and those where it is not? Emanuel and Miller propose that this be based on scientific rationale and a risk criterion.11 Placebos should be excluded where this would deprive patients of “effective, life-saving, or at least life-prolonging treatment.” While it can be justifiable to conduct a placebo-controlled trial in cases where there is some potential harm to patients, the methodological rationale must be compelling. Such rationale may exist if the placebo response rate is high, the condition rare, variable or with frequent spontaneous remissions, or if existing therapies are only partly effective or have serious side effects.
A scientific rationale is necessary, but not sufficient to make a placebo-controlled trial ethically justified. A placebo control should not increase the subjects’ risk of death or irreversible morbidity, or disability. Nor should it substantially increase their risk of other serious harms or severe discomfort. Judging which trials are acceptable on these criteria is a delicate balance and should be specifically addressed during the ethics review. Such trials must exclude participants who are at increased risk, limit the period of placebo administration as much as possible, monitor subjects carefully, provide rescue medication, and outline specific criteria for withdrawal from the study. The use of placebo, as well as its risks and rationale, must be made clear to potential subjects during informed consent.
The question of placebo use is also relevant to research on invasive interventions like surgical treatments, sometimes raising the question of placebo surgery.12 Sham surgery is controversial; requires placing control subjects at risk while additionally depriving them of effective treatment. Trials including sham interventions are, however, not always riskier than other accepted clinical trials, and can be important to protect future patients from the harms of ineffective interventions.
In practice, they should be assessed with at least the same criteria used for placebo-controlled drug trials. Particular care should be given to exploring alternatives, minimizing risks, and assessing the risk/benefit ratio. Informed consent should be particularly careful, as potential subjects may be more likely to confuse research with clinical care in the case of an invasive procedure. Some sham intervention trials may be justified, provided they have no valid safer alternative and respect the criteria outlined above.
Consenting to research on symptom management
Informed consent is particularly important in research, because it allow subjects to make an informed and voluntary choice to participate – or refuse to participate– in a project where they will take risks for the benefit of others.
Adequate informed consent requires that potential subjects understand the relevant aspects of their choice, are capable of decision-making and are free to accept or refuse participation. Assessment of each of these elements requires a degree of judgment. In the case of research on the management of pain and other symptoms, some of these elements can be more difficult to attain. Similarly, concerns could arise in emergency situations, or in situations where consent might be solicited during acute pain, or immediately following acute administration of anesthetic agents.
Although none of these situations necessarily represents an obstacle to informed consent, or to enrollment in research, they do require special precautions. Two aspects are particularly relevant. The first one is the risk that decision-making capacity could be affected. Since chronic pain, and chronic administration of stable doses of analgesic medication do not by themselves affect decision-making capacity, this question is most relevant in the presence of acute pain or the acute administration of analgesia. However, studies have shown that most patients perceive that they retain decision-making capacity in these circumstances.13
The second aspect is that, in emergency situations, there is limited time for potential subjects to think through a decision to participate in research. There are several potential solutions. First, the urgency might be reduced by the study design. If, for example, the emergency is due to severe pain rather than the need for immediate surgery, a protocol might allow treatment to proceed as usual at first, with enrollment and randomization only after the emergency has resolved. If this is not possible, care should be taken to separate those aspects of the trial that must be initiated during the emergency and those which do not. Consent should be sought at a later time for nonurgent aspects of the trial. Finally, neutral supervision of enrollment can be utilized: a clinician not involved in the trial is given veto power on the enrollment of a patient if her considered judgment is that participation is against the patient’s interests. Whichever strategy is chosen, it should be described in detail in the protocol submitted for ethics review.
Several issues contribute to Dr. Smith’s discomfort in our case scenario. (1) The patient is in pain, which she has an obligation to try to relieve. (2) The patient is terminally ill, and therefore may be vulnerable to undue pressure if he is under the therapeutic misconception that the study will necessarily provide him with better pain relief. (3) Dr. Smith understands that for portions, if not all, of the study, the patient may experience deterioration in his pain control, depending on the efficacy of SQ and IV morphine compared to his oral medications, and depending on the efficacy of the study drug, hyaluronidase. She is torn between her desire to research pain in terminally ill patients to benefit others, and her obligation to offer this patient the best and most individualized care, which could be epidural analgesia.
The solutions to some of these issues are relatively straightforward. If the patient appears competent to make medical decisions, he can certainly be informed of the risks of poorer pain control, as well as the potential benefit of better pain control for himself as well as other patients through participation in the study. Ultimately, he will have to determine if the clinical trial and its potential risks and benefits are acceptable to him. As with many terminally ill patients, the opportunity to benefit others by participating in research may have special meaning for him, and may also provide him with hope during his palliative care. While he is a “vulnerable” patient, automatically excluding him from being a clinical research subject solely on the basis of his diagnosis is arguably unethical, as long as coercion is avoided, and careful attention is paid to correcting therapeutic misconceptions he may have. Care should be taken if he agrees to be a subject to clearly outline the plans for “rescue” therapy if any, and allow for voluntary withdrawal from the study if/whenever the patient changes his mind.
However, for Dr. Smith, being both a researcher and a primary treating physician for a subject of clinical research carries several risks. The patient may be more likely to view the research study as medical care, and be unable to overcome therapeutic misconception because of Dr. Smith’s dual role in his care, for example. If he is denied epidural analgesia in favor of participation in a research study, and his pain is poorly controlled, he may not only come to mistrust Dr. Smith, but transfer that mistrust to other physicians involved in his clinical care. While the study itself may be ethical, and recruitment of terminally ill patients acceptable, Dr. Smith should consider either withdrawing from the medical team caring for the patient in favor of doing the clinical research, or ask another physician to obtain consent from the patient so that she can participate solely in his clinical care and avoid the “dual” role of researcher and clinician.
• Subjects who participate in clinical research take risks for the benefit of future patients, and may not realize personal benefits themselves. It is therefore an ethical obligation of clinical researchers to pay close attention to study design to fulfil the promise of research to patients and to society, and to justify the risks to which the subjects of research are subjected.
• Researchers always have the obligation to obtain informed consent from research subjects. This can require special precautions in emergency situations, or in situations where consent might be solicited during acute pain, or immediately following acute administration of anaesthetic agents.
• Patient subjects often confuse the intention of research trials and have the misleading expectation that the research is intended to benefit them.
• Even with explicit patient consent, the researcher has the obligation to minimize risk and discomfort to the patient.
• Particular attention to subject comfort and safety is important in placebo-controlled trials.
• Research involving symptomatic relief at the end of life is ethical, but requires special considerations. Patient assessment of benefits and burdens may shift during the trial and require reconsideration.
1* The National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (1979); ICH Steering Committee (1996); CIOMS (2002); World Medical Association (2008); Department of Health and Human Services, (March 1983).
2 Moore, R. A., Edwards, J. E., and McQuay, H. J. (2005). Acute pain: individual patient meta-analysis shows the impact of different ways of analysing and presenting results. Pain, 116(3), 322–31.
3 Farrar, J. T., Young, J. P. Jr., LaMoreaux, L., et al. (2001). Clinical importance of changes in chronic pain intensity measured on an 11-point numerical pain rating scale. Pain, 94(2), 149–58.
4* Appelbaum, P. S., Lidz, C. W., and Grisso, T. (2004). Therapeutic misconception in clinical research: frequency and risk factors. IRB, 26(2), 1–8.
5* de Melo-Martin, I. and Ho, A. (2008). Beyond informed consent: the therapeutic misconception and trust. J Med Ethics, 34(3), 202–5.
6* Pautex, S., Herrmann, F. R., and Zulian, G. B. (2005). Is research really problematic in palliative care? A pilot study. J Pain Symptom Manage, 30(2), 109–11.
7* Todd, A. M., Laird, B. J., Boyle, D., et al. (2009). A systematic review examining the literature on attitudes of patients with advanced cancer toward research. J Pain Symptom Manage, 37(6), 1078–85.
8* Agrawal, M., Grady, C., Fairlough, D. L., et al. (2006). Patients’ decision-making process regarding participation in phase I oncology research. J Clin Oncol, 24(27), 4479–84.
9* Horrobin, D. F. (2003). Are large clinical trials in rapidly lethal diseases usually unethical? Lancet, 361(9358), 695–7.
10 World Medical Association (2008). Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Subjects. Retrieved February 24, 2009, from http://www.wma.net/e/policy/b3.htm(accesed October 3, 2006).
11* Emanuel, E. J. and Miller, F. G. (2001). The ethics of placebo-controlled trials – a middle ground. N Engl J Med, 345(12), 915–19.
12* Horng, S. and Miller, F. G. (2002). Is placebo surgery unethical? N Engl J Med, 347(2), 137–9.
13* Kay, R. and Siriwardena, A. K. (2001). The process of informed consent for urgent abdominal surgery. J Med Ethics, 27(3), 157–61.
CIOMS (2002). Guideline 4: Individual Informed Consent. International Ethical Guidelines for Biomedical Research Involving Human Subjects. Geneva: CIOMS.
Emanuel, E. J., Grady, C., Crouch, R. A., et al. (2008). The Oxford Textbook of Clinical Research Ethics. Oxford, New York: Oxford University Press.
Emanuel, E. J., Wendler, D., and Grady, C. (2000). What makes clinical research ethical? JAMA, 283(20), 2701–11.
Faden, R. and Beauchamp, T. (1986). A History and Theory of Informed Consent. Oxford, New York: Oxford University Press.
Grisso, T. and Appelbaum, P. S. (1998). Assessing Competence to Consent to Treatment. A Guide for Physicians and Other Health Care Professionals. New York: Oxford, Oxford University Press.
Hawkins, J. S. and Emanuel, E. J. (2005). Clarifying confusions about coercion. Hastings Cent Rep 35(5), 16–19.
International Association for the Study of Pain. Ethical Guidelines for Pain Research in Humans. http://www.iasp-pain.org.