- Email: [email protected]
Scientific case research in complementary and alternative medicine—A review
Complementary Therapies in Medicine (2013) 21, 388—395
Available online at www.sciencedirect.com
journal homepage: www.elsevierhealth.com/journals/ctim
Scientific case research in complementary and alternative medicine—–A review Michael Teut a,∗, Klaus Linde b a
Institute for Social Medicine, Epidemiology and Health Economics, Charité University Medical Center, Luisenstr. 57, 10117 Berlin, Germany b Institute of General Practice, Technische Universität München, Munich, Germany Available online 23 May 2013
KEYWORDS Review; Single case research; Complementary and alternative medicine; Case reports; Case studies
Summary The description of individual cases is probably the most important didactic tool for teaching in medicine, especially in complementary and alternative medicine (CAM). However, only very rarely is the information provided in traditional single case reports sufficient to answer scientific questions. These reports typically are demonstrations of the solution to a clinical problem. To contribute to scientific discussion single-case research must ask critical questions whose answers are open. Two fundamental questions are: (1) Is my observation reliable? (2) Which factors, other than my treatment, may explain the observed outcome? In this review we will give an introduction to single-case research, as well as present and explain single-case designs as a tool for research and discuss their relevance and applicability for clinical practice in CAM. This review deals exclusively with single-case research on treatment effects and covers observational single-case studies, progressive and repetitive experimental single case designs, n-of-1-RCT, multiple baseline design, best case series and meta-analysis of single-case studies. © 2013 Elsevier Ltd. All rights reserved.
Contents Introduction.............................................................................................................. Quantitative and qualitative approaches to single-case research ......................................................... Observational single-case studies and case reports ....................................................................... Experimental single-case designs......................................................................................... Progressive single-case designs ...................................................................................... AB-/ABC-designs.............................................................................................. Repetitive single case designs ....................................................................................... ABA-/ABAB-designs ........................................................................................... N-of-1-RCT ...................................................................................................
∗
Corresponding author. Tel.: +49 30 450529234; fax: +49 30 450529902. E-mail address: [email protected] (M. Teut).
0965-2299/$ — see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ctim.2013.04.006
389 389 390 390 390 390 391 391 391
Scientific case research in complementary and alternative medicine
Designs involving several or multiple single-case studies or case reports ................................................. Best case series ..................................................................................................... Single case-studies embedded in studies of groups of patients....................................................... Multiple baseline design ............................................................................................. Meta-analysis of single case studies ................................................................................. Cognition-based medicine ........................................................................................... Discussion ................................................................................................................ Conflict of interests ...................................................................................................... Acknowledgements....................................................................................................... References ...............................................................................................................
Introduction Why are single-case studies needed? The most important didactic tool for teaching in medicine is probably the description of individual cases. Listening to a patient’s narrative of his/her disease, including their personal pattern of signs and symptoms and what happened once they received treatment is remarkable. Proponents of CAM treatments typically refer to reports of individual cases to highlight the benefits of their method. But why is it that, in general, case reports play a minor role in providing evidence to scientifically prove that a treatment is effective? Most case reports in CAM today are retrospective accounts of an impressive case as perceived and documented by a therapist. They serve more as a narrative to demonstrate a problem or the solution to a problem in clinical practice and can be considered very important for teaching sessions, as a didactic tool, to help students remember key information. They should be exemplary and encapsulate all the relevant clinical issues. A well told story of a case will stick in one’s memory and the crucial points can be called upon if a similar patient appears for treatment in future clinical practice. Sometimes observations in single cases can even generate new hypotheses or identify new treatment options. However, most case reports cannot be considered systematic research. Instead of ‘‘demonstrating’’ something, empirical research tries to answer open questions. The methods used must be transparent and stand a critical evaluation. For example, imagine a report by a therapist who is convinced to have a fabulous cure for a patient with advanced cancer. Even if one is only slightly sceptical, numerous questions will immediately arise: Did the patient truly suffer from this specific type of advanced cancer? Was the patient truly cured? If so, did he/she receive other treatments which might have contributed to the cure? Could it have been a spontaneous remission? Only very rarely is the information provided in traditional single case reports sufficient to answer such questions. To be considered research, critical questions must be addressed in a case report or a single-case study. In our view, the two most fundamental questions are: (1) Is my observation reliable? (2) Which factors other than my treatment may explain the observed outcome? Surprisingly, the number of systematic single-case studies published in the field of CAM is very small. This is a pity, in our opinion, because such research could contribute
389
392 392 392 392 393 393 393 394 394 394
considerably to the critical evaluation of CAM therapies and CAM research may benefit from improved understanding of benefits, limitations and utility of single-cases. Furthermore, although the workload associated with high-quality single-case studies must not be underestimated, they are probably the most feasible research tool for the busy practitioner. In this review we want to give an introduction to single-case research, to present and explain single-case designs as a tool for research and discuss their relevance and applicability for the clinical practice. This article exclusively addresses single-case research on treatment effects. The aim of all such research is to provide evidence for causalities during the therapeutic process. We do not cover etiological, prognostic, diagnostic or procedural research. Throughout this article we will use the term ‘‘case report’’ if we refer to retrospective analyses of single patients and the term ‘‘single-case study’’ if there is at least some prospective element to the study.
Quantitative and qualitative approaches to single-case research While single-case research is applied in diverse fields, such as medicine, education, rehabilitation, sport and athletic performance, it has particularly strong roots in psychology. Several textbooks summarizing methodological approaches are available (e.g. Refs. 1—3). These textbooks focus mainly on experimental single-case studies (see below) with graphical and statistical analyses. In such studies the influence of an ‘‘independent variable’’ (e.g. the treatment vs. a control condition) on a predefined outcome (e.g. blood pressure) is investigated. The outcome is measured repeatedly (e.g. daily over 12 weeks). The findings are then displayed using graphs. Specific statistical analyses are applied to test whether changes in outcomes exceed those expected by chance.4 This approach to single-case research is strongly quantitatively orientated and the principles in drawing inferences regarding causality do not differ greatly from research performed with groups of individuals. A second important field of research are qualitative case studies. This type of research is often carried out by medical anthropologists or social scientists. Especially when a researcher is investigating a field of research where little is known, a qualitative case study or case series might be a good approach to enter the field, to collect data and develop scientific hypotheses and theories. A qualitative method
390 may also be used to identify questions and develop instruments for quantitative research.5 Applied to the search for therapeutic effects qualitative research or a mixed strategy of qualitative and quantitative research methods (‘‘mixed methods approach’’) is especially valuable to explore the subjective and complex experiences and the theoretical concepts of patients and therapists.6
Observational single-case studies and case reports Observational single-case studies are scientifically less conclusive than experimental single case studies. However, as they do not interfere or only slightly interfere with routine clinical practice, they are easier to perform. Furthermore, while in many countries one should always obtain written informed consent from the patient, one does not need approval from an ethical review board. If you have very good documentation, it will be possible, at least in some situations, to retrospectively do a rigorous case analysis. From a methodological point of view, it is much better to analyze a case prospectively. Before actually starting the treatment, you need to decide with the patient what you will analyze during the course of the disease and do this accurately and systematically. It is highly recommended that you clearly formulate objectives you aim to reach with the treatment, to verify the clinical diagnosis according to the highest standard and to carefully document how the course of the disease was — depending on the condition — in the last hours, days or months. One should use validated instruments to measure symptoms, signs, quality of life etc. before starting the treatment and predefine when these are to be measured. In addition, you should carefully qualitatively document all important observations before, during and after treatment as needed. Take care to ensure that several perspectives (patient, therapist, possibly also relatives) are reflected in your documentation. Detail and consistency increase the conclusiveness more than anything else. You are free to treat as you would treat outside of the study, but ensure you document carefully. An example of a prospective case study is provided in Box 1. The main drawback of prospective case studies is that you have to decide in advance whether you will study a particular patient. It might be that this patient turns out to be non-compliant or that the course of disease in this patient is worse than in the patients you usually treat. The likelihood of having selected a patient which is not a ‘‘best’’ case is high. Semi-prospective case studies could be a compromise, balancing the advantages of retrospective case studies (focus on particularly interesting cases) and prospective case studies (systematic documentation). If early in the treatment process you become aware that a specific case might be of particular interest, you should carefully document what has happened so far and then start the prospective documentation. There are some cases in the literature where semiprospective methods seem to have been used as the documentation is quite systematic. However, publications of such single-case studies rarely describe their methods in detail, so it is difficult to know whether the approach
M. Teut, K. Linde
Box 1 An example of a prospective observational single-case study Teut and Warning8 prospectively observed an 87-yearold woman with severe joint pain due to osteoarthritis of the knee who presented with side effects from a potent pain medication (fentanyl). Together with the patient, the following outcome measures were chosen: pain intensity (using a numerical rating scale), walking distance and activities of daily living (measured with the Barthel Index). The patient was treated (in an in-patient setting) with leeches, herbal medicine, physiotherapy and three single doses of metamizol. Under therapy the patient experienced a clear reduction in pain (from 8 to 3 points on the numerical rating scale). She regained walking ability (increase in walking distance from 0 to 70 m) and showed improvements in the Barthel Index (from 45 to 65). Using the criteria of figural correspondence, several of these suggest a causal treatment effect: there was a fast and marked improvement within a few days post treatment, although the complaints had been stable for about three months. The timing of improvement was predicted based on the evidence available. Comment: This is a carefully documented case-study which suggests a causal relationship between improvement and the treatment applied. However, it also shows how difficult it is in routine care to prove causality beyond reasonable doubt.
was retrospective (based on systematic high quality routine documentation) or semi-prospective.7
Experimental single-case designs Experimental single-case studies typically share two characteristics: first, they consist of several predefined treatment and/or observation phases which are often designated as phases A, B, C, etc. (see Fig. 1). We propose to differentiate progressive and repetitive approaches (see below). Second, the analysis focuses on a quantitative or graphical comparison of the different phases often using specific statistical methods (single case statistics). The sequence or timing of the treatment or observation periods can be non-randomized or randomized.
Progressive single-case designs AB-/ABC-designs Most basic experimental single-case studies follow an AB or an ABC design (see Fig. 1, upper part), where A is a preassessment phase (or treatment 1), B the treatment phase (or treatment 2), and C a follow-up phase (or treatment 3). Box 2 summarizes an example. We call the approach progressive because single phases follow each other without being repeated. There can be more complex progressive approaches in which, for example, there is a second intervention phase (B ) and a second follow-up phase (C ), to investigate whether effects are long lasting. Progressive approaches can be applied in almost any chronic condition
Scientific case research in complementary and alternative medicine Intensity of complaints
a)
A
B
b)
A
B
391
This to some extent increases the validity of such studies, particularly if assessments are made by an independent observer who does not know when the treatment began, or if a placebo intervention was used during the baseline phase. If several such studies are combined to a multiplebaseline study (see below) the conclusiveness increases considerably.
C
Repetitive single case designs
c)
WO
A
WO
A
B
WO
B
B
WO
A
WO
Time/Phase
Figure 1 Basic designs of experimental single-case studies: (a) ABC design; (b) ABAB design; (c) variable phases with washout (WO) periods in between. Taken from Witt and Linde26 with permission from Elsevier.
Box 2 Example of a single-case study using an ABC design Lee et al.10 observed a 35-year-old man with advanced lung cancer (stage IV, metastases in the stomach and bone) treated with radiotherapy and chemotherapy six months prior and currently receiving only morphine for pain. At the study entry, his general condition was poor (Karnofsky Performance Scale 40, constantly high pain levels) requiring special care and assistance over the previous month. After six days of pre-treatment observation, the patient received eight sessions Korean external Qigong therapy on alternate days over 16 days, followed by a two-week post-treatment observation phase. A variety of symptoms were assessed repeatedly using a visual analog scale. The results were presented in graphs. Comment: The study shows convincingly that the patient improved. After the initiation of the Qigong treatment, symptoms improved rapidly, consistently and distinctively. Improvements seem to persist in the short follow-up phase. A main weakness is the short pre-treatment observation period with only two measurements, which leaves some uncertainty as to whether the symptoms at the study entry were particularly bad. This would increase the likelihood of regression to the mean or spontaneous improvement (although the authors report that the patient felt bad at least for the previous month).
without interfering too much with routine practice. The obvious limitation of such an approach is that it is not possible to assess whether the effects are reproducible due to the lack of alternation of the different phases. It is possible to introduce a random element into such a study by randomizing, for example, how long the pre-treatment phase lasts and when the treatment starts.9
ABA-/ABAB-designs A more complex design is the ABA-design, which is called reversal design. An experimental variable is repeatedly introduced and withdrawn. If a patient, for example, suffers from a chronic condition and symptoms re-appear at the same intensity shortly after disrupting a treatment, repetitive approaches are possible. If, for example, a patient with chronic pain receives the short acting analgesic X for three days, he/she might have less pain during that period but after stopping the medication the pain returns. He/she could then try analgesic Y in the same way. If this is repeated once again in a study this is described as an ABAB design (see Fig. 1b). Depending on the number and sequence of treatments, a variety of designs are possible (ABBA, ABBABAAB, etc.). Wash-out phases must be included if treatments have longer-lasting effects (see Fig. 1c). Several statistical methods are available to analyze repetitive single-case studies.1—3 The advantage of more complex designs is that confounders are less likely to occur repeatedly. Designs that finish with a treatment phase, such as ABAB overcome the ethical objection of comparing with the baseline phase. More complex multiple sequential withdrawal designs can be used to analyze the separate effects of different components of complex settings. N-of-1-RCT The conclusiveness of a repetitive study can be greatly enhanced if the multiple treatment phases are randomized. Such a study can then be designed as a n-of-1 trial. The number-of-1-randomized-controlled-trial-design (n-of1-RCT) is based on the single case experimental design but applies the principles of conventional RCT in extension.11 Multiple, randomized crossovers of phase pairs are applied to the individual patient in crossover order. This design can be used when a rapid onset of the intervention can be shown, which ceases soon after the treatment phase and a therapeutic effect is not maintained without reapplying the intervention. A minimum of three phase pairs is considered to provide sufficient data for a conclusion. This design is often performed as double-blind and placebocontrolled to assess the best individual therapeutic evidence in cases with conflicting medical therapies, or to individually prove the therapeutic benefit of a therapeutic option.12 A statistical analysis is performed to assess the individual treatment effect. It is also possible to evaluate case series,13 by focusing on the best therapy for an individual patient. For the evaluation of the best individual treatment option the n-of-1-RCT method was considered by Guayatt et al. to be even superior to evidence gained from systematic reviews or meta-analyses.14 The n-of-1-RCT is also
392
Box 3 An example of a randomized n-of-1 trial Estrada and Young16 performed a double-blind multiple cross-over, placebo-controlled trial in a 61-year-old African-American male who had experienced side effects with several antihypertensive medications and was not willing to try any others. The patient received three treatment pairs (each single phase for three weeks) of a garlic-smelling placebo or a garlic preparation in randomized order. To provide a washout period, data from the first week after each cross-over were not included in the statistical analysis. While the patient was taking garlic, the mean systolic blood pressure was 146.9 mm compared to 148.9 under placebo (p < 0.05), the mean diastolic blood pressure was 87.2 vs. 89.6 (p < 0.05). The patient identified the capsule he was taking in 75% of the weeks. After reviewing the results of the study the patient insisted on receiving only garlic for the management of his hypertension in spite of the small effect. Comment: While unblinding might have introduced some bias, the study suggests that there was a small blood pressure lowering effect with the garlic treatment.
a very interesting option for medical CAM research.15 The approval of an ethical review board may be required if a nonconventional or experimental therapy is tested for scientific reasons. In Box 3 one of the few examples of n-of-1 trials in CAM is summarized.
Designs involving several or multiple single-case studies or case reports Best case series A best case series is a documentation of a number of cases which have gone particularly well. If such a best case series indicates surprisingly positive outcomes in a number of individuals, it can stimulate further systematic research. For example, the Office of Cancer Complementary and Alternative Medicine (OCCAM) of the US National Cancer Institute launched a best case series program in oncology (http://www.cancer.gov/cam/bestcase intro.html). Providers of CAM therapies are invited to submit their best cases. To be considered conclusive, reports must meet stringent criteria. In the case of the OCCAM program these are: (1) well-documented, definitive diagnosis of cancer; (2) an adequate documentation of disease response; (3) absence of confounders (no concurrent treatments with known therapeutic potential); (4) an adequately documented treatment history. An example from such a best case series is summarized in Box 4. Obviously, best case series can be performed for other conditions using adapted criteria.
M. Teut, K. Linde
Box 4 An example of a best case series Banerji et al.17 published a report on a best case series, evaluated in the program of the Office of Cancer Complementary and Alternative Medicine (OCCAM) of the US National Cancer Institute. Lung and esophageal carcinoma patients were treated with homeopathic remedies in an Indian centre according to a specific (Banerji) protocol until there was complete regression of the tumors. Case records including pathology and radiology reports from 14 patients were submitted for review by the US NCI BCS Program. Four of these cases had independent confirmation of the diagnosis and radiographic responses and were accepted as sufficient information for the NCI to initiate further investigation. The results of the review were deemed to be sufficient to warrant an NCI-initiated prospective research follow-up in the form of an observational study.
Single case-studies embedded in studies of groups of patients It is possible to integrate single-case studies into observational or experimental studies in larger groups of patients. For example, Hamre et al.18 performed a prospective observational study of anthroposophic art therapy in 161 patients. Within the study the intensity of complaints, quality of life, depressive symptoms, use of other treatments, satisfaction with the treatment and a variety of other outcome measures were documented repeatedly up to 24 months after inclusion. Sixteen patients were selected for additional single case analysis for which additional interviews were performed. Using figural correspondence, the authors tried to assess the likelihood of a causality of observed treatment effects. Qualitative interviews are an excellent tool to further understand subjective experiences of patients and therapists and complement quantitative data. Walach et al.19 conducted a 1-year observational study of 18 patients with chronic headaches, who had previously participated in a placebo-controlled randomized trial of classical homeopathy. Patients filled in diaries which were analyzed using single-case statistical methods. Single-case studies embedded in studies of groups of patients can serve to collect additional qualitative information to improve the understanding of processes, to provide more detailed and individualized evaluation and possibly to strengthen the conclusiveness.
Multiple baseline design If the therapeutic intervention is thought to have a sustained therapeutic effect on the patient, re-intervening with the baseline phase will not be useful to control the treatment phase. In these cases, the AB-design is the option to study therapeutic effects. A multiple baseline design across participants is an extension of the classical AB-design and aims to systematically compare the therapeutic intervention between several patients in an AB-setting. The basic idea is that the introduction of the treatment phase is progressively
Scientific case research in complementary and alternative medicine Intensity of complaints
393
Cognition-based medicine Cognition-based medicine (CBM) takes quite a different approach to investigate causal inferences and is primarily applied for observational single-case studies and case reports. CBM was developed by Helmut Kiene et al. based on Karl Duncker’s Gestalt theory. For more details we refer to the paper on Cognition Based Medicine in this special issue.
Discussion
Time
Figure 2 Multiple baseline design: several patients are observed over a longer period of time with repeated measurements. The timing of the switch from baseline to treatment (fat lines) can be randomized. Taken from Witt and Linde26 with permission from Elsevier.
delayed in successive participants. In the first phase (baseline) they do not receive the treatment investigated. The duration of the baseline phase and the timing of the switch to the treatment phase are variable. If possible, the timing of the switch is randomized and remains unknown to both the patient and the treatment provider until the treatment is commenced. If a treatment effect can be demonstrated during the introductory treatment phase, the effect will most likely be due to the intervention (Fig. 2). It is unlikely that a confounding effect could repeatedly coincide with the introduction of the therapeutic intervention. Confounding variables would therefore be seen as a change in all baselines. If in such a study, a placebo can be used or if outcome measurements can be performed by a blinded independent observer, the validity of a multiple baseline study can be greatly enhanced. If a consistent response pattern with marked improvements after initiating treatment emerges, such designs can provide quite impressive preliminary evidence. Therefore, they are sometimes used as pilot studies for larger observational or experimental studies.20,21 In psychological behavioral research, where multiple baseline designs (MBD) are mainly applied, they have been used not only across participants, but also across settings and behaviors.
Meta-analysis of single case studies If multiple single-case studies on the same condition using a similar methodology become available, these can be summarized in a systematic review or a meta-analysis. To the best of our knowledge, this has not yet been done in CAM due to the lack of a sufficient number of similar single case studies. However, what has been done already is a series of randomized n-of-1 trials designed prospectively in a manner to allow for combined analysis.22
Single-case study designs have not yet been integrated systematically or widely into CAM research although they deserve much more attention. The need to develop modern methods of case collection, documentation and evaluation has already been recognized but only rarely has it been systematically applied.23 We believe that this is a pity because such research could contribute considerably to a critical evaluation of CAM therapies. Furthermore, although the workload associated with high-quality single-case studies must not be underestimated single case studies are probably the most feasible research tool for the busy practitioner. Can the effectiveness of a CAM therapy be proven by single cases? Yes and no. In principle, it is possible to show a causal treatment effect in a single patient with a rigorous single-case study. However, showing that a treatment is generally effective in a given condition is not possible (unless there are consistent results from a large number of similar case studies). Single-case studies are great instruments to improve the ability for (self-)critical observation. They can be used to generate hypotheses, to carefully observe and to analyze clinical processes. They can be used in quality assessments and serve as teaching material. It is unlikely that case studies will provide evidence that will convince skeptics, but open clinicians may listen carefully. If a large number of high quality single-case studies would become available combined analysis may provide further insights. In summary, it would be unrealistic to expect that singlecase studies could revolutionize CAM research — but they may well contribute important pieces in an evidence puzzle. Single case study designs have certain limitations which restrict their use in practice:
- The more complex a therapeutic intervention is, the more difficult the analysis becomes. - Introducing a combination of therapeutic interventions at the same time complicates the analysis. - A continuous documentation of interventions and outcome parameters is essential but a time consuming task. - Assessment of potential bias factors can become detective’s work. - Prospective or semi-prospective scientific case research requires planning, which may delay the start of treatment. - Experimental therapies may require the approval of an ethical board.
394 The ideal conditions for most single-case studies are: - Stable baseline, usually corresponding with chronic diseases. - If the disease is self resolving or spontaneous remissions occur frequently, the single case designs might not be applicable. - Introducing only one new or repeated therapeutic element at a time. - No carry-over effects between treatment phases. - Continuous assessment of outcome parameters (high number of measuring points). - Motivated patients with good compliance. - Clinical setting with enough time for research. Guidelines like CONSORT statement (http://www. consort-statement.org/) have helped to improve the quality of clinical trials. The implementation or adaptation of guidelines on documentation, evaluation and reviewing of case reports might also be very helpful for CAM case research.24 An important issue to discuss is the ethics. The approval of an ethical review board is considered necessary if a non-conventional or experimental therapy is tested for scientific reasons, but not in cases where routine therapies are applied for the individual benefit of the patient.25 Ethical review boards are often not familiar with CAM and singlecase research and might see the case in a different light than a CAM researcher or practitioner: the CAM therapy involved might not be at all understood as routine therapy and nearly automatically be classified as ‘‘experimental’’. In cases where medical treatments are involved, authorities might even classify a single-case study as a variation of a phase 2 drug trial with all the legal and organizational consequences. Today many medical journals invite the submission of case reports, but opportunities to study single case research methodology are still rare. We believe that single-case study designs can be a very useful tool for CAM researchers: for fundamental clinical research, as a pilot study and to develop hypotheses for bigger and more expensive clinical trials. They may well bridge the gap between CAM practitioners and researchers, as expertise in both fields is needed to conduct a high quality single-case study. Especially for general practitioners, single-case studies might turn out to be a very interesting field of research.
Conflict of interests There are no conflicts of interest. There was no funding for this research.
Acknowledgements Parts of this article are taken from a book chapter by Klaus Linde in the book Witt C.M., Linde K.: Clinical Research in Complementary and Integrative Medicine — A Practical Training book. München: Elsevier, 2011, ISBN: 978-0-7020-3476, and from Teut M: Scientific case studies in homeopathy. New Directions in homeopathy research,
M. Teut, K. Linde edited by Claudia Witt and Henning Albrecht, KVC Publishers, Essen 2009. We thank both publishers for giving their permission to use our previous work.
References 1. Petermann F, Müller JM. Clinical psychology and single-case evidence — a practical approach to treatment, planning and evaluation. Chichester: Wiley; 2001. 2. Julius H, Schlosser RW, Goetze H. Kontrollierte Einzelfallstudien. Göttingen: Hogrefe; 2000. 3. Franklin RD, Allison DB, Gorman BS, editors. Design and analysis of single-case research. Mahwah: Lawrence Erlbaum Associates; 1997. 4. Parker RI, Brossart DF. Evaluating single case research data: a comparison of seven statistical methods. Behavior Therapy 2003;34:189—211. 5. Thompson CB, Walker L. Basics of research (part 12): qualitative research. Air Medical Journal 1998;17(2):65—70. 6. Broom A. Using qualitative interviews in CAM research: a guide to study design, data collection and data analysis. Complementary Therapies in Medicine 2005;13:65—73. 7. Wode K, Schneider T, Lundberg I, Kienle GS. Mistletoe treatment in cancer-related fatigue: a case report. Cases Journal 2009;2:77. 8. Teut M, Warning A. Blutegel, Phytotherapie und Physiotherapie bei Gonarthrose — eine geriatrische Fallstudie. Forsch Komplementärmed 2008;15:269—72. 9. Schulte D, Walach H. F.M. Alexander technique in the treatment of stuttering — a randomized single-case intervention study with ambulatory monitoring. Psychotherapy and Psychosomatics 2006;75:190—1. 10. Lee MS, Yang SH, Lee KK, Moon SR. Effects of Qi therapy (external Qigong) on symptoms of advanced cancer: a single case study. European Journal of Cancer 2005;14:457—62. 11. Guyatt G, Sackett DL, Adachi J, Roberts R, Chong J, Rosenbloom D, et al. A clinician’s guide for conducting randomised trials in individual patients. Canadian Medical Association Journal 1988;139(6):497—503. 12. Nicles CJ, Clavarino AM, Del Mar CB. Using n-of-1 trials as clinical tool to improve prescribing. British Journal of General Practice 2005;55:175—80. 13. Nathan PC, Tomlison G, Dupuis LL, Greenberg ML, Ota S, Bartels U, et al. A pilot study of ondansetron plus metopimazine vs. ondansetron monotherapy in children receiving highly emetogenic chemotherapy: a Bayesian randomized serial n-of-1-trials design. Support Care Cancer 2006;14:268—76. 14. Guayatt G, Haynes B, Jaeschke R, Cook D, Greenalgh T, Mead M, et al. Introduction: the philosophy of evidence based medicine. In: Guayatt G, Rennie D, editors. User’s guide to the medical literature: a manual for evidence based clinical practice. Chicago: American Medical Association Press; 2002. p. 3—11. 15. Johnston BC, Mills E. n-of-1 randomized controlled trials: an opportunity for complementary and alternative medicine evaluation. Journal of Alternative and Complementary Medicine 2004;10(6):979—84. 16. Estrada CA, Young MJ. Patient preferences for novel therapy: an n-of-1 trial of garlic in the treatment of hypertension. Journal of General Internal Medicine 1993;6:619—21. 17. Banerji P, Campbell DR, Banerji P. Cancer patients treated with the Banerji protocols using homoeopathic medicine: a best case series program of the National Cancer Institute USA. Oncology Reports 2008;20:69—74. 18. Hamre HJ, Glockmann A, Kiene H. Wirksamkeitsbeurteilung der Anthroposophischen Kunsttherapie: Einzelfallstudie
Scientific case research in complementary and alternative medicine eingebettet in eine prospektive Kohortenstudie. Merkurstab 2004;57:194—203. 19. Walach H, Lowes T, Mussbach D, Schamell U, Springer W, Stritzl G, et al. The long-term effects of homeopathic treatment of chronic headaches: one year follow-up and single case time series analysis. British Homoeopathic Journal 2001;90: 63—72. 20. Kazdin A, Kopel S. On resolving ambiguities of the multiplebaseline design: problems and recommendations. Behavior Therapy 1975;6:601—8. 21. Watson PJ, Workman EA. The non-concurrent multiple baseline across-individuals design: an extension of the traditional multiple baseline design. Journal of Behavior Therapy and Experimental Psychiatry 1981;12(3): 257—9.
395
22. Coxeter PD, Schluter PJ, Eastwood HL, Nikles CJ, Glasziou P. Valerian does not appear to reduce symptoms for patients with chronic insomnia in general practice using a series of randomised n-of-1 trials. Complementary Therapies in Medicine 2003;11:215—22. 23. Baas C. The pitfalls of clinical case research: lessons from the Delphi Project. Homeopathy 2004;93(1):21—6. 24. Hung SK, Hillier S, Ernst E. Case reports of adverse effects of herbal medicinal products (HMPs): a quality assessment. Phytomedicine 2011;18:335—43. 25. Irweg L, Glasziou P, March L. Ethics of n-of-1 trials. Lancet 1995;345:469. 26. Witt CM, Linde K. Clinical research in complementary and integrative medicine. A practical training book. München: Elesvier; 2011.
Available online at www.sciencedirect.com
journal homepage: www.elsevierhealth.com/journals/ctim
Scientific case research in complementary and alternative medicine—–A review Michael Teut a,∗, Klaus Linde b a
Institute for Social Medicine, Epidemiology and Health Economics, Charité University Medical Center, Luisenstr. 57, 10117 Berlin, Germany b Institute of General Practice, Technische Universität München, Munich, Germany Available online 23 May 2013
KEYWORDS Review; Single case research; Complementary and alternative medicine; Case reports; Case studies
Summary The description of individual cases is probably the most important didactic tool for teaching in medicine, especially in complementary and alternative medicine (CAM). However, only very rarely is the information provided in traditional single case reports sufficient to answer scientific questions. These reports typically are demonstrations of the solution to a clinical problem. To contribute to scientific discussion single-case research must ask critical questions whose answers are open. Two fundamental questions are: (1) Is my observation reliable? (2) Which factors, other than my treatment, may explain the observed outcome? In this review we will give an introduction to single-case research, as well as present and explain single-case designs as a tool for research and discuss their relevance and applicability for clinical practice in CAM. This review deals exclusively with single-case research on treatment effects and covers observational single-case studies, progressive and repetitive experimental single case designs, n-of-1-RCT, multiple baseline design, best case series and meta-analysis of single-case studies. © 2013 Elsevier Ltd. All rights reserved.
Contents Introduction.............................................................................................................. Quantitative and qualitative approaches to single-case research ......................................................... Observational single-case studies and case reports ....................................................................... Experimental single-case designs......................................................................................... Progressive single-case designs ...................................................................................... AB-/ABC-designs.............................................................................................. Repetitive single case designs ....................................................................................... ABA-/ABAB-designs ........................................................................................... N-of-1-RCT ...................................................................................................
∗
Corresponding author. Tel.: +49 30 450529234; fax: +49 30 450529902. E-mail address: [email protected] (M. Teut).
0965-2299/$ — see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ctim.2013.04.006
389 389 390 390 390 390 391 391 391
Scientific case research in complementary and alternative medicine
Designs involving several or multiple single-case studies or case reports ................................................. Best case series ..................................................................................................... Single case-studies embedded in studies of groups of patients....................................................... Multiple baseline design ............................................................................................. Meta-analysis of single case studies ................................................................................. Cognition-based medicine ........................................................................................... Discussion ................................................................................................................ Conflict of interests ...................................................................................................... Acknowledgements....................................................................................................... References ...............................................................................................................
Introduction Why are single-case studies needed? The most important didactic tool for teaching in medicine is probably the description of individual cases. Listening to a patient’s narrative of his/her disease, including their personal pattern of signs and symptoms and what happened once they received treatment is remarkable. Proponents of CAM treatments typically refer to reports of individual cases to highlight the benefits of their method. But why is it that, in general, case reports play a minor role in providing evidence to scientifically prove that a treatment is effective? Most case reports in CAM today are retrospective accounts of an impressive case as perceived and documented by a therapist. They serve more as a narrative to demonstrate a problem or the solution to a problem in clinical practice and can be considered very important for teaching sessions, as a didactic tool, to help students remember key information. They should be exemplary and encapsulate all the relevant clinical issues. A well told story of a case will stick in one’s memory and the crucial points can be called upon if a similar patient appears for treatment in future clinical practice. Sometimes observations in single cases can even generate new hypotheses or identify new treatment options. However, most case reports cannot be considered systematic research. Instead of ‘‘demonstrating’’ something, empirical research tries to answer open questions. The methods used must be transparent and stand a critical evaluation. For example, imagine a report by a therapist who is convinced to have a fabulous cure for a patient with advanced cancer. Even if one is only slightly sceptical, numerous questions will immediately arise: Did the patient truly suffer from this specific type of advanced cancer? Was the patient truly cured? If so, did he/she receive other treatments which might have contributed to the cure? Could it have been a spontaneous remission? Only very rarely is the information provided in traditional single case reports sufficient to answer such questions. To be considered research, critical questions must be addressed in a case report or a single-case study. In our view, the two most fundamental questions are: (1) Is my observation reliable? (2) Which factors other than my treatment may explain the observed outcome? Surprisingly, the number of systematic single-case studies published in the field of CAM is very small. This is a pity, in our opinion, because such research could contribute
389
392 392 392 392 393 393 393 394 394 394
considerably to the critical evaluation of CAM therapies and CAM research may benefit from improved understanding of benefits, limitations and utility of single-cases. Furthermore, although the workload associated with high-quality single-case studies must not be underestimated, they are probably the most feasible research tool for the busy practitioner. In this review we want to give an introduction to single-case research, to present and explain single-case designs as a tool for research and discuss their relevance and applicability for the clinical practice. This article exclusively addresses single-case research on treatment effects. The aim of all such research is to provide evidence for causalities during the therapeutic process. We do not cover etiological, prognostic, diagnostic or procedural research. Throughout this article we will use the term ‘‘case report’’ if we refer to retrospective analyses of single patients and the term ‘‘single-case study’’ if there is at least some prospective element to the study.
Quantitative and qualitative approaches to single-case research While single-case research is applied in diverse fields, such as medicine, education, rehabilitation, sport and athletic performance, it has particularly strong roots in psychology. Several textbooks summarizing methodological approaches are available (e.g. Refs. 1—3). These textbooks focus mainly on experimental single-case studies (see below) with graphical and statistical analyses. In such studies the influence of an ‘‘independent variable’’ (e.g. the treatment vs. a control condition) on a predefined outcome (e.g. blood pressure) is investigated. The outcome is measured repeatedly (e.g. daily over 12 weeks). The findings are then displayed using graphs. Specific statistical analyses are applied to test whether changes in outcomes exceed those expected by chance.4 This approach to single-case research is strongly quantitatively orientated and the principles in drawing inferences regarding causality do not differ greatly from research performed with groups of individuals. A second important field of research are qualitative case studies. This type of research is often carried out by medical anthropologists or social scientists. Especially when a researcher is investigating a field of research where little is known, a qualitative case study or case series might be a good approach to enter the field, to collect data and develop scientific hypotheses and theories. A qualitative method
390 may also be used to identify questions and develop instruments for quantitative research.5 Applied to the search for therapeutic effects qualitative research or a mixed strategy of qualitative and quantitative research methods (‘‘mixed methods approach’’) is especially valuable to explore the subjective and complex experiences and the theoretical concepts of patients and therapists.6
Observational single-case studies and case reports Observational single-case studies are scientifically less conclusive than experimental single case studies. However, as they do not interfere or only slightly interfere with routine clinical practice, they are easier to perform. Furthermore, while in many countries one should always obtain written informed consent from the patient, one does not need approval from an ethical review board. If you have very good documentation, it will be possible, at least in some situations, to retrospectively do a rigorous case analysis. From a methodological point of view, it is much better to analyze a case prospectively. Before actually starting the treatment, you need to decide with the patient what you will analyze during the course of the disease and do this accurately and systematically. It is highly recommended that you clearly formulate objectives you aim to reach with the treatment, to verify the clinical diagnosis according to the highest standard and to carefully document how the course of the disease was — depending on the condition — in the last hours, days or months. One should use validated instruments to measure symptoms, signs, quality of life etc. before starting the treatment and predefine when these are to be measured. In addition, you should carefully qualitatively document all important observations before, during and after treatment as needed. Take care to ensure that several perspectives (patient, therapist, possibly also relatives) are reflected in your documentation. Detail and consistency increase the conclusiveness more than anything else. You are free to treat as you would treat outside of the study, but ensure you document carefully. An example of a prospective case study is provided in Box 1. The main drawback of prospective case studies is that you have to decide in advance whether you will study a particular patient. It might be that this patient turns out to be non-compliant or that the course of disease in this patient is worse than in the patients you usually treat. The likelihood of having selected a patient which is not a ‘‘best’’ case is high. Semi-prospective case studies could be a compromise, balancing the advantages of retrospective case studies (focus on particularly interesting cases) and prospective case studies (systematic documentation). If early in the treatment process you become aware that a specific case might be of particular interest, you should carefully document what has happened so far and then start the prospective documentation. There are some cases in the literature where semiprospective methods seem to have been used as the documentation is quite systematic. However, publications of such single-case studies rarely describe their methods in detail, so it is difficult to know whether the approach
M. Teut, K. Linde
Box 1 An example of a prospective observational single-case study Teut and Warning8 prospectively observed an 87-yearold woman with severe joint pain due to osteoarthritis of the knee who presented with side effects from a potent pain medication (fentanyl). Together with the patient, the following outcome measures were chosen: pain intensity (using a numerical rating scale), walking distance and activities of daily living (measured with the Barthel Index). The patient was treated (in an in-patient setting) with leeches, herbal medicine, physiotherapy and three single doses of metamizol. Under therapy the patient experienced a clear reduction in pain (from 8 to 3 points on the numerical rating scale). She regained walking ability (increase in walking distance from 0 to 70 m) and showed improvements in the Barthel Index (from 45 to 65). Using the criteria of figural correspondence, several of these suggest a causal treatment effect: there was a fast and marked improvement within a few days post treatment, although the complaints had been stable for about three months. The timing of improvement was predicted based on the evidence available. Comment: This is a carefully documented case-study which suggests a causal relationship between improvement and the treatment applied. However, it also shows how difficult it is in routine care to prove causality beyond reasonable doubt.
was retrospective (based on systematic high quality routine documentation) or semi-prospective.7
Experimental single-case designs Experimental single-case studies typically share two characteristics: first, they consist of several predefined treatment and/or observation phases which are often designated as phases A, B, C, etc. (see Fig. 1). We propose to differentiate progressive and repetitive approaches (see below). Second, the analysis focuses on a quantitative or graphical comparison of the different phases often using specific statistical methods (single case statistics). The sequence or timing of the treatment or observation periods can be non-randomized or randomized.
Progressive single-case designs AB-/ABC-designs Most basic experimental single-case studies follow an AB or an ABC design (see Fig. 1, upper part), where A is a preassessment phase (or treatment 1), B the treatment phase (or treatment 2), and C a follow-up phase (or treatment 3). Box 2 summarizes an example. We call the approach progressive because single phases follow each other without being repeated. There can be more complex progressive approaches in which, for example, there is a second intervention phase (B ) and a second follow-up phase (C ), to investigate whether effects are long lasting. Progressive approaches can be applied in almost any chronic condition
Scientific case research in complementary and alternative medicine Intensity of complaints
a)
A
B
b)
A
B
391
This to some extent increases the validity of such studies, particularly if assessments are made by an independent observer who does not know when the treatment began, or if a placebo intervention was used during the baseline phase. If several such studies are combined to a multiplebaseline study (see below) the conclusiveness increases considerably.
C
Repetitive single case designs
c)
WO
A
WO
A
B
WO
B
B
WO
A
WO
Time/Phase
Figure 1 Basic designs of experimental single-case studies: (a) ABC design; (b) ABAB design; (c) variable phases with washout (WO) periods in between. Taken from Witt and Linde26 with permission from Elsevier.
Box 2 Example of a single-case study using an ABC design Lee et al.10 observed a 35-year-old man with advanced lung cancer (stage IV, metastases in the stomach and bone) treated with radiotherapy and chemotherapy six months prior and currently receiving only morphine for pain. At the study entry, his general condition was poor (Karnofsky Performance Scale 40, constantly high pain levels) requiring special care and assistance over the previous month. After six days of pre-treatment observation, the patient received eight sessions Korean external Qigong therapy on alternate days over 16 days, followed by a two-week post-treatment observation phase. A variety of symptoms were assessed repeatedly using a visual analog scale. The results were presented in graphs. Comment: The study shows convincingly that the patient improved. After the initiation of the Qigong treatment, symptoms improved rapidly, consistently and distinctively. Improvements seem to persist in the short follow-up phase. A main weakness is the short pre-treatment observation period with only two measurements, which leaves some uncertainty as to whether the symptoms at the study entry were particularly bad. This would increase the likelihood of regression to the mean or spontaneous improvement (although the authors report that the patient felt bad at least for the previous month).
without interfering too much with routine practice. The obvious limitation of such an approach is that it is not possible to assess whether the effects are reproducible due to the lack of alternation of the different phases. It is possible to introduce a random element into such a study by randomizing, for example, how long the pre-treatment phase lasts and when the treatment starts.9
ABA-/ABAB-designs A more complex design is the ABA-design, which is called reversal design. An experimental variable is repeatedly introduced and withdrawn. If a patient, for example, suffers from a chronic condition and symptoms re-appear at the same intensity shortly after disrupting a treatment, repetitive approaches are possible. If, for example, a patient with chronic pain receives the short acting analgesic X for three days, he/she might have less pain during that period but after stopping the medication the pain returns. He/she could then try analgesic Y in the same way. If this is repeated once again in a study this is described as an ABAB design (see Fig. 1b). Depending on the number and sequence of treatments, a variety of designs are possible (ABBA, ABBABAAB, etc.). Wash-out phases must be included if treatments have longer-lasting effects (see Fig. 1c). Several statistical methods are available to analyze repetitive single-case studies.1—3 The advantage of more complex designs is that confounders are less likely to occur repeatedly. Designs that finish with a treatment phase, such as ABAB overcome the ethical objection of comparing with the baseline phase. More complex multiple sequential withdrawal designs can be used to analyze the separate effects of different components of complex settings. N-of-1-RCT The conclusiveness of a repetitive study can be greatly enhanced if the multiple treatment phases are randomized. Such a study can then be designed as a n-of-1 trial. The number-of-1-randomized-controlled-trial-design (n-of1-RCT) is based on the single case experimental design but applies the principles of conventional RCT in extension.11 Multiple, randomized crossovers of phase pairs are applied to the individual patient in crossover order. This design can be used when a rapid onset of the intervention can be shown, which ceases soon after the treatment phase and a therapeutic effect is not maintained without reapplying the intervention. A minimum of three phase pairs is considered to provide sufficient data for a conclusion. This design is often performed as double-blind and placebocontrolled to assess the best individual therapeutic evidence in cases with conflicting medical therapies, or to individually prove the therapeutic benefit of a therapeutic option.12 A statistical analysis is performed to assess the individual treatment effect. It is also possible to evaluate case series,13 by focusing on the best therapy for an individual patient. For the evaluation of the best individual treatment option the n-of-1-RCT method was considered by Guayatt et al. to be even superior to evidence gained from systematic reviews or meta-analyses.14 The n-of-1-RCT is also
392
Box 3 An example of a randomized n-of-1 trial Estrada and Young16 performed a double-blind multiple cross-over, placebo-controlled trial in a 61-year-old African-American male who had experienced side effects with several antihypertensive medications and was not willing to try any others. The patient received three treatment pairs (each single phase for three weeks) of a garlic-smelling placebo or a garlic preparation in randomized order. To provide a washout period, data from the first week after each cross-over were not included in the statistical analysis. While the patient was taking garlic, the mean systolic blood pressure was 146.9 mm compared to 148.9 under placebo (p < 0.05), the mean diastolic blood pressure was 87.2 vs. 89.6 (p < 0.05). The patient identified the capsule he was taking in 75% of the weeks. After reviewing the results of the study the patient insisted on receiving only garlic for the management of his hypertension in spite of the small effect. Comment: While unblinding might have introduced some bias, the study suggests that there was a small blood pressure lowering effect with the garlic treatment.
a very interesting option for medical CAM research.15 The approval of an ethical review board may be required if a nonconventional or experimental therapy is tested for scientific reasons. In Box 3 one of the few examples of n-of-1 trials in CAM is summarized.
Designs involving several or multiple single-case studies or case reports Best case series A best case series is a documentation of a number of cases which have gone particularly well. If such a best case series indicates surprisingly positive outcomes in a number of individuals, it can stimulate further systematic research. For example, the Office of Cancer Complementary and Alternative Medicine (OCCAM) of the US National Cancer Institute launched a best case series program in oncology (http://www.cancer.gov/cam/bestcase intro.html). Providers of CAM therapies are invited to submit their best cases. To be considered conclusive, reports must meet stringent criteria. In the case of the OCCAM program these are: (1) well-documented, definitive diagnosis of cancer; (2) an adequate documentation of disease response; (3) absence of confounders (no concurrent treatments with known therapeutic potential); (4) an adequately documented treatment history. An example from such a best case series is summarized in Box 4. Obviously, best case series can be performed for other conditions using adapted criteria.
M. Teut, K. Linde
Box 4 An example of a best case series Banerji et al.17 published a report on a best case series, evaluated in the program of the Office of Cancer Complementary and Alternative Medicine (OCCAM) of the US National Cancer Institute. Lung and esophageal carcinoma patients were treated with homeopathic remedies in an Indian centre according to a specific (Banerji) protocol until there was complete regression of the tumors. Case records including pathology and radiology reports from 14 patients were submitted for review by the US NCI BCS Program. Four of these cases had independent confirmation of the diagnosis and radiographic responses and were accepted as sufficient information for the NCI to initiate further investigation. The results of the review were deemed to be sufficient to warrant an NCI-initiated prospective research follow-up in the form of an observational study.
Single case-studies embedded in studies of groups of patients It is possible to integrate single-case studies into observational or experimental studies in larger groups of patients. For example, Hamre et al.18 performed a prospective observational study of anthroposophic art therapy in 161 patients. Within the study the intensity of complaints, quality of life, depressive symptoms, use of other treatments, satisfaction with the treatment and a variety of other outcome measures were documented repeatedly up to 24 months after inclusion. Sixteen patients were selected for additional single case analysis for which additional interviews were performed. Using figural correspondence, the authors tried to assess the likelihood of a causality of observed treatment effects. Qualitative interviews are an excellent tool to further understand subjective experiences of patients and therapists and complement quantitative data. Walach et al.19 conducted a 1-year observational study of 18 patients with chronic headaches, who had previously participated in a placebo-controlled randomized trial of classical homeopathy. Patients filled in diaries which were analyzed using single-case statistical methods. Single-case studies embedded in studies of groups of patients can serve to collect additional qualitative information to improve the understanding of processes, to provide more detailed and individualized evaluation and possibly to strengthen the conclusiveness.
Multiple baseline design If the therapeutic intervention is thought to have a sustained therapeutic effect on the patient, re-intervening with the baseline phase will not be useful to control the treatment phase. In these cases, the AB-design is the option to study therapeutic effects. A multiple baseline design across participants is an extension of the classical AB-design and aims to systematically compare the therapeutic intervention between several patients in an AB-setting. The basic idea is that the introduction of the treatment phase is progressively
Scientific case research in complementary and alternative medicine Intensity of complaints
393
Cognition-based medicine Cognition-based medicine (CBM) takes quite a different approach to investigate causal inferences and is primarily applied for observational single-case studies and case reports. CBM was developed by Helmut Kiene et al. based on Karl Duncker’s Gestalt theory. For more details we refer to the paper on Cognition Based Medicine in this special issue.
Discussion
Time
Figure 2 Multiple baseline design: several patients are observed over a longer period of time with repeated measurements. The timing of the switch from baseline to treatment (fat lines) can be randomized. Taken from Witt and Linde26 with permission from Elsevier.
delayed in successive participants. In the first phase (baseline) they do not receive the treatment investigated. The duration of the baseline phase and the timing of the switch to the treatment phase are variable. If possible, the timing of the switch is randomized and remains unknown to both the patient and the treatment provider until the treatment is commenced. If a treatment effect can be demonstrated during the introductory treatment phase, the effect will most likely be due to the intervention (Fig. 2). It is unlikely that a confounding effect could repeatedly coincide with the introduction of the therapeutic intervention. Confounding variables would therefore be seen as a change in all baselines. If in such a study, a placebo can be used or if outcome measurements can be performed by a blinded independent observer, the validity of a multiple baseline study can be greatly enhanced. If a consistent response pattern with marked improvements after initiating treatment emerges, such designs can provide quite impressive preliminary evidence. Therefore, they are sometimes used as pilot studies for larger observational or experimental studies.20,21 In psychological behavioral research, where multiple baseline designs (MBD) are mainly applied, they have been used not only across participants, but also across settings and behaviors.
Meta-analysis of single case studies If multiple single-case studies on the same condition using a similar methodology become available, these can be summarized in a systematic review or a meta-analysis. To the best of our knowledge, this has not yet been done in CAM due to the lack of a sufficient number of similar single case studies. However, what has been done already is a series of randomized n-of-1 trials designed prospectively in a manner to allow for combined analysis.22
Single-case study designs have not yet been integrated systematically or widely into CAM research although they deserve much more attention. The need to develop modern methods of case collection, documentation and evaluation has already been recognized but only rarely has it been systematically applied.23 We believe that this is a pity because such research could contribute considerably to a critical evaluation of CAM therapies. Furthermore, although the workload associated with high-quality single-case studies must not be underestimated single case studies are probably the most feasible research tool for the busy practitioner. Can the effectiveness of a CAM therapy be proven by single cases? Yes and no. In principle, it is possible to show a causal treatment effect in a single patient with a rigorous single-case study. However, showing that a treatment is generally effective in a given condition is not possible (unless there are consistent results from a large number of similar case studies). Single-case studies are great instruments to improve the ability for (self-)critical observation. They can be used to generate hypotheses, to carefully observe and to analyze clinical processes. They can be used in quality assessments and serve as teaching material. It is unlikely that case studies will provide evidence that will convince skeptics, but open clinicians may listen carefully. If a large number of high quality single-case studies would become available combined analysis may provide further insights. In summary, it would be unrealistic to expect that singlecase studies could revolutionize CAM research — but they may well contribute important pieces in an evidence puzzle. Single case study designs have certain limitations which restrict their use in practice:
- The more complex a therapeutic intervention is, the more difficult the analysis becomes. - Introducing a combination of therapeutic interventions at the same time complicates the analysis. - A continuous documentation of interventions and outcome parameters is essential but a time consuming task. - Assessment of potential bias factors can become detective’s work. - Prospective or semi-prospective scientific case research requires planning, which may delay the start of treatment. - Experimental therapies may require the approval of an ethical board.
394 The ideal conditions for most single-case studies are: - Stable baseline, usually corresponding with chronic diseases. - If the disease is self resolving or spontaneous remissions occur frequently, the single case designs might not be applicable. - Introducing only one new or repeated therapeutic element at a time. - No carry-over effects between treatment phases. - Continuous assessment of outcome parameters (high number of measuring points). - Motivated patients with good compliance. - Clinical setting with enough time for research. Guidelines like CONSORT statement (http://www. consort-statement.org/) have helped to improve the quality of clinical trials. The implementation or adaptation of guidelines on documentation, evaluation and reviewing of case reports might also be very helpful for CAM case research.24 An important issue to discuss is the ethics. The approval of an ethical review board is considered necessary if a non-conventional or experimental therapy is tested for scientific reasons, but not in cases where routine therapies are applied for the individual benefit of the patient.25 Ethical review boards are often not familiar with CAM and singlecase research and might see the case in a different light than a CAM researcher or practitioner: the CAM therapy involved might not be at all understood as routine therapy and nearly automatically be classified as ‘‘experimental’’. In cases where medical treatments are involved, authorities might even classify a single-case study as a variation of a phase 2 drug trial with all the legal and organizational consequences. Today many medical journals invite the submission of case reports, but opportunities to study single case research methodology are still rare. We believe that single-case study designs can be a very useful tool for CAM researchers: for fundamental clinical research, as a pilot study and to develop hypotheses for bigger and more expensive clinical trials. They may well bridge the gap between CAM practitioners and researchers, as expertise in both fields is needed to conduct a high quality single-case study. Especially for general practitioners, single-case studies might turn out to be a very interesting field of research.
Conflict of interests There are no conflicts of interest. There was no funding for this research.
Acknowledgements Parts of this article are taken from a book chapter by Klaus Linde in the book Witt C.M., Linde K.: Clinical Research in Complementary and Integrative Medicine — A Practical Training book. München: Elsevier, 2011, ISBN: 978-0-7020-3476, and from Teut M: Scientific case studies in homeopathy. New Directions in homeopathy research,
M. Teut, K. Linde edited by Claudia Witt and Henning Albrecht, KVC Publishers, Essen 2009. We thank both publishers for giving their permission to use our previous work.
References 1. Petermann F, Müller JM. Clinical psychology and single-case evidence — a practical approach to treatment, planning and evaluation. Chichester: Wiley; 2001. 2. Julius H, Schlosser RW, Goetze H. Kontrollierte Einzelfallstudien. Göttingen: Hogrefe; 2000. 3. Franklin RD, Allison DB, Gorman BS, editors. Design and analysis of single-case research. Mahwah: Lawrence Erlbaum Associates; 1997. 4. Parker RI, Brossart DF. Evaluating single case research data: a comparison of seven statistical methods. Behavior Therapy 2003;34:189—211. 5. Thompson CB, Walker L. Basics of research (part 12): qualitative research. Air Medical Journal 1998;17(2):65—70. 6. Broom A. Using qualitative interviews in CAM research: a guide to study design, data collection and data analysis. Complementary Therapies in Medicine 2005;13:65—73. 7. Wode K, Schneider T, Lundberg I, Kienle GS. Mistletoe treatment in cancer-related fatigue: a case report. Cases Journal 2009;2:77. 8. Teut M, Warning A. Blutegel, Phytotherapie und Physiotherapie bei Gonarthrose — eine geriatrische Fallstudie. Forsch Komplementärmed 2008;15:269—72. 9. Schulte D, Walach H. F.M. Alexander technique in the treatment of stuttering — a randomized single-case intervention study with ambulatory monitoring. Psychotherapy and Psychosomatics 2006;75:190—1. 10. Lee MS, Yang SH, Lee KK, Moon SR. Effects of Qi therapy (external Qigong) on symptoms of advanced cancer: a single case study. European Journal of Cancer 2005;14:457—62. 11. Guyatt G, Sackett DL, Adachi J, Roberts R, Chong J, Rosenbloom D, et al. A clinician’s guide for conducting randomised trials in individual patients. Canadian Medical Association Journal 1988;139(6):497—503. 12. Nicles CJ, Clavarino AM, Del Mar CB. Using n-of-1 trials as clinical tool to improve prescribing. British Journal of General Practice 2005;55:175—80. 13. Nathan PC, Tomlison G, Dupuis LL, Greenberg ML, Ota S, Bartels U, et al. A pilot study of ondansetron plus metopimazine vs. ondansetron monotherapy in children receiving highly emetogenic chemotherapy: a Bayesian randomized serial n-of-1-trials design. Support Care Cancer 2006;14:268—76. 14. Guayatt G, Haynes B, Jaeschke R, Cook D, Greenalgh T, Mead M, et al. Introduction: the philosophy of evidence based medicine. In: Guayatt G, Rennie D, editors. User’s guide to the medical literature: a manual for evidence based clinical practice. Chicago: American Medical Association Press; 2002. p. 3—11. 15. Johnston BC, Mills E. n-of-1 randomized controlled trials: an opportunity for complementary and alternative medicine evaluation. Journal of Alternative and Complementary Medicine 2004;10(6):979—84. 16. Estrada CA, Young MJ. Patient preferences for novel therapy: an n-of-1 trial of garlic in the treatment of hypertension. Journal of General Internal Medicine 1993;6:619—21. 17. Banerji P, Campbell DR, Banerji P. Cancer patients treated with the Banerji protocols using homoeopathic medicine: a best case series program of the National Cancer Institute USA. Oncology Reports 2008;20:69—74. 18. Hamre HJ, Glockmann A, Kiene H. Wirksamkeitsbeurteilung der Anthroposophischen Kunsttherapie: Einzelfallstudie
Scientific case research in complementary and alternative medicine eingebettet in eine prospektive Kohortenstudie. Merkurstab 2004;57:194—203. 19. Walach H, Lowes T, Mussbach D, Schamell U, Springer W, Stritzl G, et al. The long-term effects of homeopathic treatment of chronic headaches: one year follow-up and single case time series analysis. British Homoeopathic Journal 2001;90: 63—72. 20. Kazdin A, Kopel S. On resolving ambiguities of the multiplebaseline design: problems and recommendations. Behavior Therapy 1975;6:601—8. 21. Watson PJ, Workman EA. The non-concurrent multiple baseline across-individuals design: an extension of the traditional multiple baseline design. Journal of Behavior Therapy and Experimental Psychiatry 1981;12(3): 257—9.
395
22. Coxeter PD, Schluter PJ, Eastwood HL, Nikles CJ, Glasziou P. Valerian does not appear to reduce symptoms for patients with chronic insomnia in general practice using a series of randomised n-of-1 trials. Complementary Therapies in Medicine 2003;11:215—22. 23. Baas C. The pitfalls of clinical case research: lessons from the Delphi Project. Homeopathy 2004;93(1):21—6. 24. Hung SK, Hillier S, Ernst E. Case reports of adverse effects of herbal medicinal products (HMPs): a quality assessment. Phytomedicine 2011;18:335—43. 25. Irweg L, Glasziou P, March L. Ethics of n-of-1 trials. Lancet 1995;345:469. 26. Witt CM, Linde K. Clinical research in complementary and integrative medicine. A practical training book. München: Elesvier; 2011.