An analysis of sickness absence in chronically ill patients receiving Complementary and Alternative Medicine: A longterm prospective intermittent study
© Moebus et al; licensee BioMed Central Ltd. 2006
Received: 16 March 2005
Accepted: 12 February 2006
Published: 12 February 2006
The popularity of complementary and alternative medicine (CAM) has led to a growing amount of research in this area. All the same little is known about the effects of these special treatments in every-day practice of primary care, delivered by general practitioners within the health insurance system. From 1994 to 2000 more than 20 German Company health insurances initiated the first model project on CAM according to the German social law. Aim of this contribution is to investigate the effectiveness of multi-modal CAM on chronic diseases within primary health care.
A long-term prospective intermittent study was conducted including 44 CAM practitioners and 1221 self-selected chronically ill patients (64% women) of whom 441 were employed. Main outcome measure is sick-leave, controlled for secular trends and regression-to-the mean and self-perceived health status.
Sick-leave per year of 441 patients at work increased from 22 (SD ± 45.2) to 31 (± 61.0) days within three years prior to intervention, and decreased to 24 (± 55.6) in the second year of treatment, sustaining at this level in the following two years. Detailed statistical analysis show that this development exceeds secular trends and the regression-toward-the-mean effect. Sick-leave reduction was corroborated by data on self-reported improvement of patients' health status.
Results of this longterm observational study show a reduction of sick leave in chronically ill patients after a complex multimodal CAM intervention. However, as this is an uncontrolled observational study efficacy of any specific CAM treatment can not be proven. The results might indicate an general effectiveness of CAM in primary care, worthwhile further investigations. Future studies should identify the most suitable patients for CAM practices, the most appropriate and safe treatments, provide information on the magnitude of the effects to facilitate subsequent definitive randomised controlled studies that will help to position complementary and alternative medicine in health care.
The dilemma is becoming ever more obvious: although mainstream medicine, science and health policy refuse Complementary and Alternative Medicine (CAM) as being scientific and efficient, the popularity of the use of CAM is at a premium and growing [1–5]. This conflicting situation is, above all, fostered by the still prevailing shortcomings of good scientific evidence for efficacy and effectiveness of CAM procedures, as well as by the increasing chronic diseases within the population – those disease patterns for which mainstream medicine itself acknowledges having no satisfactory solutions.
Against this background, 22 German company health insurance funds (BKK) in the Rhine-Ruhr area initiated a project according to the German social law lasting from 1994 to 2000. Only within this project, health insurances were allowed to pay for CAM therapies.
The scientific evaluation of the project had to be carried out on account of legal requirements. Therefore, an observational study with quality control steps according to standard operation procedures (SOP) has been performed, committing sickness absence as a main study outcome. Sickness absence is widely accepted as an objectively and integrated measure of morbidity in the working population [6–10], though it is still a seldom used outcome measure in epidemiological/clinical studies. Even more seldom are studies which directly link absenteeism as one outcome variable and the individual health status .
The purpose of the study is to provide profound information about the effectiveness of CAM in chronically ill patients. The purpose, however, is not to prove that a specific CAM treatment is effective for a specific disease. Rather, it is meant to investigate possible longterm effects in everyday practice in primary care. In this contribution, we examine the overall effectiveness of CAM interventions on trends in sick-leave, together with patients' health related quality of life. Particularly, we investigate the extent of the regression-towards-the-mean effect, an often ignored ubiquitous statistical phenomenon in pre-/post-treatment measurements [12–14].
We conducted a longterm prospective cohort study with intraindividual pre/post comparisons using patients' questionnaires, documentations of the participating physicians, and health insurance data.
Specifically approved Complementary and Alternative Medicine Procedures*
holistic anamnese (up to 1 hour)
electric acupuncture (except: according to Voll)
electric acupuncture (not by Voll)
oxygen therapies except: oxyon-, hyperbaric oxygen- and oxygen multistep-therapy
Data was collected from different sources: Case report forms for documenting therapy processes and data of routine health insurance records (sickness absence, in hospital, medication). Patients' self-administered questionnaires comprising health status, on the basis of the questionnaire of the German National Health Survey (NHS). The NHS was part of the German Cardiovascular Prevention Study , that comprises three nationwide cross sectional surveys in the former West Germany between 1981–1996. The NHS is a major source of the German health reporting system and is provided as public use files, frequently used for scientific research.
Sick-leave days (SLD) and duration of sickness leave per case (DSL) served as indicators for morbidity. The cumulative sick-leave incidence (CSLI, the proportion of patients with any sick-leave in a given observation year) served as ill health indicator for the whole study group. Standardised rate ratios (SRR) for SLD were obtained by dividing sick leave days by the mean values in strata defined by age classes and gender in the reference population. As reference population all members of the German Company Health Insurance Funds were chosen. Sick leave data of the reference population are documented in the annual health reports .
The SLD distribution shows strong positive skewness with 40% of the patients having no sick-leave in a given observation year. Applying a log-transformation reveals an excess of patients with no sick leave in a given observation year. Therefore, we treat the number of patients without sick leave – or its complement, the CSLI per year – as a distinct parameter for sick-leave in the entire study group. The CSLI and measures of SLD and SRR distributions characterise the time dependence of sick leave in the study population. The quartiles of those measures are presented with their 95% confidence intervals (95% CI).
Changes in the CSLI are subjected to a χ2-test-of-fit to a uniform distribution over the observation years. Repeated measures multivariate analysis of variance based on ranks was used to explore changes in SLD. The repeated measures design confines analysis to dependent samples, that is, to well defined subcohorts with non-missing sick-leave data in every observation year compared.
The kind of study design – observational long term study including retrospective data, a pre/post comparison, no explicit control group and highly selected patients – suggests an effect of regression-toward-the-mean (RTM) in our data [12, 17, 18], a statistical phenomenon, first defined by Galton . RTM originates from selection by value and poor correlation of variables, and is of particular importance for repeated measurements as in longitudinal studies. In short: RTM is associated with Pearsons' correlation r between e.g. the measured values of a quantity at one point in time and the values of the same quantity at the next point in time. The effect, observable in subgroups selected by the value of the quantity (for example, the upper 25% of the population at the first point in time) is strongest for zero correlation and vanishes at r = 1. Unawareness of this phenomenon may lead to misinterpretation of data.
In order to check up on RTM we (1) controlled for common societal trends standardising sick-leave data to the population of all German company health insurances (SRR). Assuming that sick-leave five to four years before inclusion to the study is independent from study participation, we (2) estimated the effect of RTM from the fifth to the fourth year in our cohort: the strongly skewed distributed SRR is represented by the classes (SRR = 0, 0<SRR< = 1, SRR>1), correlation is then given by a transition matrix between adjacent points in time (years). We observed that in both remote years five and four years before inclusion, mean SRR in the model experiment was stationary at 0.86 for 321 patients with sick-leave data reported in both these years. The respective 95%CIs are 0.69–1.03 five years and 0.7–1.02 four years before inclusion, indicating comparable sick leave levels of the patients to the level in the reference population five and four years before enrolment. A mean SRR slightly lower unity may be accountable for by demographic characteristics of patients volunteering for CAM treatment. (3) For the present argument, the observed steady state of mean SRR is decisive. We therefore may estimate the matrix describing the process of regression-toward-the-mean in the reference population from the participants' data in the indicated consecutive remote years. Multiplying this estimate for the transition matrix with the vector of class percentages in a given year yields the class percentages in the next year as expected under regression-toward-the-mean. To compare classified SRR distributions between years we employed Cochran-Mantel-Haenszel (CMH) statistics.
Note, that all significance levels associated with the above tests are understood to be strictly descriptive. Analyses were performed with SAS software package.
Baseline characteristics of the study population
Whole study population (n = 1221)
Study population at work (n = 441)
Age (years), mean (± SD)
Duration of disease (years), mean (± SD)
Duration of treatment (years), mean (± SD)
Health status*, mean (± SD)
3.2 (0.86) (response: 93.9 %)
3.2 (0.82) (response: 97.5 %)
2.7 (0.91) (response: 74.3 %)
2.6 (0.91) (response: 74.6%)
2.6 (0.94) (response: 51.1 %)
2.6 (0.88) (response: 54.7%)
Standardising our data according to age and sex, study participants represent a higher social status than the NHS population. The greatest differences appeared, as expected, in reference to the health status: patients engaged more in sports, cared more for their health and were to a higher degree convinced that one can influence ones own health status. The current health state was considered to be much poorer, and distress such as backache, pain in neck and shoulders, feeling of weakness, and weariness were more often represented among study participants compared to the German population.
The mean CAM treatment lasted 2.0 years (mean intervention time). Patients were treated on average with three different therapy modalities, with acupuncture, face time, homoeopathy, oxygen therapy, and neuraltherapy as the most often used. Surprisingly, phytotherapy seemed to have only a small meaning, although of the drug groups practitioners prescribed most often, phytotherapeutics ranked on the fourth place with 7%, besides homoeopathics (15%), vitamins/minerals (9%) and bacteria lysates (8%).
Sick leave days per patient and year, stratified by sex (unstandardized)
Sick leave days/patient/year
5 (4; 9)
6 (4; 9)
5 (3; 10)
7 (4; 9)
8 (5; 12)
4 (1; 9)
9 (6; 12)
8 (5; 12)
10 (6; 14)
9 (7; 11)
8 (5; 11)
11 (7; 16)
5 (3; 7)
5 (3; 8)
5 (0; 8)
5 (2; 7)
5 (1; 8)
5 (0; 10)
4 (0; 6)
5 (0; 9)
1 (0; 5)
Standardized Rate Ratio (SRR) of the sick leave quota*
Standardized Rate Ratio
Proportion of patients with improved versus unchanged or impaired selfreported health status after treatment
Health status after treatment
Characteristics of our study cohort are typical for user of CAM: On average, they are more likely middle aged, female, multimorbid, higher educated, of poorer subjective health, with chronic pain, non-life-threatening health problems and are high utilisers of the health care system [4, 19–22]. The latter item is especially interesting since our outcome parameter, absenteeism, is an internationally established indicator for work related morbidity , and was also used to study effects of acupuncture in randomised trials . Furthermore absenteeism imposes considerable direct and indirect costs on the employer and health insurances. Accordingly, the revealed sustained reduction of absenteeism, exceeding regression-toward-the-mean effects and societal trends, might indicate at least an indirect economical effect of the multifaceted CAM intervention, particularly with regard to the reduction of long spells. Together with the improvement of the health status this is a plausible argument for a beneficial treatment for this kind of patients. Up to which degree the observed effects might have an overall economical benefit has yet to be proven.
The observed pattern of sick leave with higher rates of absenteeism in men is different to results of many other epidemiological studies reporting consistently higher levels of absenteeism in women [e.g. [7, 24–26]]. This could be due to the patient-based cohort group here, with especially highly selected chronically ill men. Since it is known that women prefer CAM therapies, one can assume, that only those male study members entered the study, showing the highest perceived morbidity. Self-perceived well-being on the other hand is in some way associated with taking sick leave .
In Germany, almost all insurance companies initiated projects according to the German social law [28, 29]. Although none of them included this wide spectrum of CAM therapies – homoeopathy and acupuncture are mostly involved – those examining sick-leave found similar results . Contrasting to our study, they did not control for regression-toward-the-mean and secular trend. However, the evaluation of these projects will give comprehensive insights into the use of CAM in primary care patients. In the period from 1995–2007 the insurance funds in Germany will altogether have spent approximately € 510 million for CAM treatments and their scientific evaluations (€ 12 mio) .
Within this project study physicians were allowed to practice a variety of CAM procedures without financial restriction. On a first glance, this might be quite unfavourable for evaluation purposes. Otherwise, for the first time, we are able to achieve insights of general practices of CAM practitioners in primary care.
Concerning study limitations (1) a major bias might derive from a spontaneous regression of disease, but this seems unlikely given the long-term chronic condition of the patients and the very long observation time. (2) Study subjects were not aware of the fact that the principal outcome would be absenteeism, so it is unlikely to assume that participants were less absent from work because of their knowledge about the main outcome. (3) The observed temporal course of sick-leave could rely only on regression-toward-the-mean, especially with this highly selected group of patients. But again, the long retrospective and prospective time span observed made it possible to control for regression-toward-the-mean, demonstrating an effect that exceeds this phenomenon. (4) A potential bias might arise from different sample sizes at different observations years. However, neither a last observation carried forward nor a complete case analysis resulted in major modifications of our results. (5) It must be taken into account that the feature of being unfit to work is influenced by several confounding factors [31, 32]. A remarkable influence on the number of SLDs is the severity of the chronic disease, as well as different kinds of working conditions and workload. Absenteeism is a complex process, and although a poor health status is related to higher rates of absenteeism , there are also different coping behaviours that influence taking sick-leave . This phenomenon might explain the relatively high proportion of chronically ill patients without any sick-leave in one given year. Nevertheless, the pattern of our sick-leave data, combined with self-reported health status is consistent with constant effects of the CAM intervention on sick-leave in our cohort. (6) At least, possible co-interventions might have influenced the observed changes. During the observation period patients could have received additional conventional treatments. When asking patients for further therapies, we found that more than 60% received additional treatments, but in most instances these referred to treatments of dentists, gynaecologists, ophthalmologists or orthopaedists, seldom to treatments of radiologists, neurologists, ENTs. Less than 5% of the patients reported co-intervention from other general practitioners. So the impact of co-intervention seemed to be low at a first glance. Nevertheless, when analysing the prescription patterns of the study practitioners, we could find some (modern) conventional pharmaceutical therapies, e.g. Aarane, a bronchodilatator for asthma . Further analysis are planned to describe the diverse strategies of CAM practitioners regarding treatment patterns, number of different CAM specialities and conventional therapies used, treatment durations, costs in association with patients satisfaction and health improvement.
Finally, some comments on the relevance of our results have to be made relating to the observational study design. With the special situation of evaluating a project according to the German social law, we had to regard some conditions that differ from clinical studies. First, establishing a reliable control group was not feasible since patients interested in participating were highly in favour of being treated with this variety of CAM interventions. It is known from other studies that patients with a strong preference for a particular treatment will refuse randomisation [36, 37]. What is more, it had to be expected that our chronically ill patients would refuse those conventional therapies which they had already experienced without success years before. Second, logistical reasons excluded a random invitation. Third, no comparable population with definitely no access to CAM was available due to data protection reasons and lack of interest of conventional practitioners to cooperate. The absence of a suitable control group could partly be compensated by the long observation time and by the use of routine health data for comparison, but nevertheless limits the generalisability of our conclusions.
Profound observational studies are pragmatic approaches to questions concerning the practices used for particular diseases, the numbers and types of patients who use them, and how well patients respond to treatment . Therefore, the evaluation of this intervention project reveals comprehensive insights of multi-modal CAM use in general practice. The value and meaning of results of well-designed observational studies have been extensively discussed [39–45] and we agree with Rosenbaum , who states that results from observation studies strengthen the evidence, but do not prove the treatment caused its ostensible effects. We are well aware of the fact that this study by no way could adress the question of efficacy (efficacy here refers to an intervention that has been shown to be superior to placebo in randomised controlled trials) of any of the CAM treatments applied. It is meant to show possible long-term effects in association with the special intervention in real-life practice. The definition of a relevant endpoint, the confirmation of a study protocol with a clear study question and the provision for quality assessment as prerequisites of a careful observational study , make this study an approach in which an investigation of a whole system has been undertaken in its proper context.
Extensive research in CAM is needed, that will help to define the place of these interventions in health care. This is called for, seeing as the continuing public demand for CAM, combined with still missing sound scientific research, will affect and challenge health care delivery.
From a public health perspective we definitely need to answer questions such as: Which procedures of CAM are safe and effective and for which group of patients? How does CAM meets health care needs, and is CAM associated with high usage of all type of health care? The answering of these questions has to consider social, cultural, political, and economic contexts to maximize the contribution of CAM to the health care system globally. Therefore we need the whole scope of scientific methods available – experimental and randomized controlled trials to outcome studies – in order to be able to make informed and rational decisions within the health care system. What is urgently needed to cope with these challenging tasks is a comprehensive and targeted funding for CAM research, which is almost completely missing in most countries. Altogether this study provides another small building block for the long-term objective of a pragmatic integration of mainstream medicine and CAM that better meets the needs and wishes of all patients – not only in Germany.
complementary and alternative medicine
cumulative sick-leave incidence
duration of sickness leave per case
ear nose and throat
German National Health Survey
standardised rate ratio
We sincerely thank the following individuals for their commitment and helpful work: Herbert Hirche for his scientific contribution, Claudia Ose and Anja Marr for their excellent research assistance, and Roswitha Beneda for her reliable longtime data management. We thank the personnel of all participating Company Health Insurances (BKKs) for their help in providing routine health records. We wish to thank the Zentrum zur Dokumentation für Naturheilverfahren, Essen, especially Dr. Schlebusch and Marianne Kolvenbach for the coordination of the technical part of the study, and special thanks to the participating practitioners for their engagement. At last we are very grateful to Lorraine Frisina for her carefully proof-reading of the manuscript.
- Fisher P, Ward A: Complementary Medicine in Europe. BMJ. 1994, 309: 107-111.View ArticlePubMedPubMed CentralGoogle Scholar
- MacLennan AH, Wilson DH, Tayloy AW: Prevalence and cost of use of alternative medicine in Australia. Lancet. 1996, 347: 569-573. 10.1016/S0140-6736(96)91271-4.View ArticlePubMedGoogle Scholar
- Wotton CW, Sparber A: Surveys of complementary and alternative medicine: Part I. General trends and demographic groups. J Altern Compl Med. 2001, 7: 195-208. 10.1089/107555301750164307.View ArticleGoogle Scholar
- Thomas KJ, Nicholl JP, Coleman P: Use and expenditure on complementary medicine in England: a population based survey. Comp Ther Med. 2001, 9: 2-11. 10.1054/ctim.2000.0407.View ArticleGoogle Scholar
- Harris P, Rees R: The prevalence of complementary and alternative medicine use among the general population: a systematic review of the literature. Compl Ther Med. 2000, 8: 88-96. 10.1054/ctim.2000.0353.View ArticleGoogle Scholar
- Marmot M, Feeney A, Shipley M, North F, Syme SL: Sickness absence as a measure of health status and functioning: from the UK Whitehall II study. J Epidemiol Community Health. 1995, 49: 124-130.View ArticlePubMedPubMed CentralGoogle Scholar
- Tellnes G, Bjerkedal T: Epidemiology of Sickness Certification. Scand J Soc Med. 1989, 17: 245-251.PubMedGoogle Scholar
- Hemingway H, Shipley M, Stansfeld S, Marmot M: Sickness absence from back pain, psychosocial work characteristics and employment grade among office workers. Scand J Work Environ Health. 1997, 23: 121-129.View ArticlePubMedGoogle Scholar
- Allebeck P, Mastekaasa A: Risk factors for sick leave – general studies. Scand J Public Health. 2004, 63: 49-108. 10.1080/14034950410021853.View ArticleGoogle Scholar
- Bödeker W: Associations between workload and diseases rarely occurring in sickness absence data. J Occup Environ Med. 2001, 43: 1081-1088.View ArticleGoogle Scholar
- Alexanderson K: Sickness absence: a review of performed studies with focused on levels of exposures and theories utilized. Scand J Soc Med. 1998, 26: 241-249.PubMedGoogle Scholar
- Campbell DT, Kenny DA: A primer on regression artifacts. 1999, The Guilford Press: New YorkGoogle Scholar
- Bland JM, Altman DG: Statistic notes: regression towards the mean. BMJ. 1994, 308: 1499-View ArticlePubMedPubMed CentralGoogle Scholar
- Streiner DL: Regression toward the mean: its etiology, diagnosis, and treatment. Can J Psychiatry. 2001, 46: 72-76.PubMedGoogle Scholar
- GCP Study Group: The German Cardiovascular Prevention Study (GCP): Design and methods. Eur H J. 1988, 9: 1058-1066.Google Scholar
- Federal Association of Company Health Insurance Funds: Health Report 1994. (in German) BKK Bundesverband: Essen
- Yudkin PL, Stratton IM: How to deal with regression to the mean in intervention studies. Lancet. 1996, 347: 241-243. 10.1016/S0140-6736(96)90410-9.View ArticlePubMedGoogle Scholar
- Chen S, Cox C, Cui L: A more flexible regression-to-the-mean model with possible stratification. Biometrics. 1998, 54: 939-947.View ArticlePubMedGoogle Scholar
- Astin JA: Why patients use alternative medicine: Results of a national study. JAMA. 1998, 279: 1548-1553. 10.1001/jama.279.19.1548.View ArticlePubMedGoogle Scholar
- Bausell RB, Lee WL, Berman BM: Demographic and Health-Related Correlates of Visits to Complementary and Alternative Medical Providers. Med Care. 2001, 39: 190-196. 10.1097/00005650-200102000-00009.View ArticlePubMedGoogle Scholar
- Druss BG, Rosenheck RA: Association between use of unconventional therapies and conventional medical services. JAMA. 1999, 282: 651-656. 10.1001/jama.282.7.651.View ArticlePubMedGoogle Scholar
- Leung JM, Dzankic S, Manku K, Yuan S: The prevalence and predictors of the use of alternative medicine in presurgical patients in five California hospitals. Anesth Analg. 2001, 93: 1062-1068. 10.1097/00000539-200110000-00053.View ArticlePubMedGoogle Scholar
- Vickers AJ, Rees RW, Zollman CE, Rob McCarney, Smith CM, Ellis N, Fisher P, Van Haselen R: Acupuncture for chronic headache in primary care: large, pragmatic, randomised trial. BMJ. 2004, 328: 744-747. 10.1136/bmj.38029.421863.EB.View ArticlePubMedPubMed CentralGoogle Scholar
- Feeney A, North F, Head J, Canner R, Marmot M: Socioeconomic and sex differentials in reason for sickness absence from the Whitehall II Study. Occup Environ Med. 1998, 55: 91-98.View ArticlePubMedPubMed CentralGoogle Scholar
- Kivimäki M, Vahtera J, Thomson L, Griffiths A, Cox T, Pentti J: Psychosocial factors predicting employee sickness absence during economic decline. J App Psychol. 1997, 82: 858-872. 10.1037/0021-9010.82.6.858.View ArticleGoogle Scholar
- Kristensens TS: Sickness absence and work strain among Danish slaughterhosue workers: an analysis of absence from work regarded as coping behaviour. Soc Sci Med. 1991, 32: 15-27. 10.1016/0277-9536(91)90122-S.View ArticleGoogle Scholar
- Hörnquist JO, Hansson B, Zar M: Well-being and future sick-leave. Multivariate analysis with regard to preceding sick-leave. Eur J Pub Health. 1997, 7: 284-290. 10.1093/eurpub/7.3.284.View ArticleGoogle Scholar
- Melchart D, Linde K, Streng A, Reitmayr S, Hoppe A, Brinkhaus B, Becker-Witt C, Wagenpfeil S, Pfaffenrath V, Hammes M, Willich SN, Weidenhammer W: Acupuncture randomised trials (ART) in patients with migraine or tension-type headache – Design and protocols. Forsch Komplementärmed Klass Naturheilkd. 2003, 10: 179-184. 10.1159/000073473.View ArticlePubMedGoogle Scholar
- Walach H, Güthlin C: Effects of homeopathy and acupuncture in general practice – Intermediate results of a longitudinal observational study as seen in work-absenteeism. Forsch Komplementärmed Klass Naturheilkd. 2002, 7: 36-10.1159/000046861.Google Scholar
- Marstedt G, Moebus S: Gesundheitsberichterstattung des Bundes Heft 9 – Inanspruchnahme alternativer Methoden in der Medizin. (Federal Health Report No. 9: Health care utilization and prevalence of complementary and alternative medicine in Germany). 2002, Verlag Robert Koch Institut: Berlin, [http://www.rki.de]Google Scholar
- Breaugh JA: Predicting absenteeism from prior absenteeism and work attitudes. J Applied Psychology. 1981, 66: 555-560. 10.1037//0021-9010.66.5.555.View ArticleGoogle Scholar
- North F, Syme SL, Feeney A, Head J, Shipley MJ, Marmot MG: Explaining socioeconomic differences in sickness absence: the Whitehall II study. BMJ. 1993, 306: 361-366.View ArticlePubMedPubMed CentralGoogle Scholar
- Rael EGS, Stansfeld SA., Shipley M, Head J, Feeney A, Marmot M: Sickness absence in the Whitehall II study, London: the role of social support and material problems. J Epidemiol Community Health . 1995, 49: 474-481.View ArticlePubMedPubMed CentralGoogle Scholar
- Kristensens TS: Sickness absence and work strain among Danish slaughterhouse workers: an analysis of absence from work regarded as coping behaviour. Soc Sci Med. 1991, 32: 15-27. 10.1016/0277-9536(91)90122-S.View ArticleGoogle Scholar
- Moebus S, Lehmann N, Hoffmann B, Jöckel KH: Prescription patterns of practitioners treating chronically ill patients with complementary and alternative medicine (CAM) – Results of a longterm outcome study. ISTAHC Annu Meet. 2002, 18: 215-Google Scholar
- Richardson MA, Post-White J, Singletary SE, Justice B: Recruitment for complementary/alternative medicine trials: who participates after breast cancer. Ann Behav Med. 1998, 20: 190-198.View ArticlePubMedGoogle Scholar
- Ellis PM: Attitudes towards and participation in randomised clinical trials in oncology: a review of the literature. Ann Oncol. 2000, 11: 939-945. 10.1023/A:1008342222205.View ArticlePubMedGoogle Scholar
- Nahin RL, Straus SE: Research into complementary and alternative medicine: problems and potential. BMJ. 2001, 322: 161-164. 10.1136/bmj.322.7279.161.View ArticlePubMedPubMed CentralGoogle Scholar
- White AR, Ernst E: The case for uncontrolled clinical trials: a starting point for the evidence base for CAM. Complementar Ther Med. 2001, 9: 11-115.View ArticleGoogle Scholar
- Cummings P, Weiss NS: Case series and exposure series: the role of studies without controls in providing information about the etiology of injury or disease. Injury Prevention. 1998, 4: 54-57.View ArticlePubMedPubMed CentralGoogle Scholar
- Kunz R, Oxman AD: The unpredictability paradox: Review of empirical comparisons of randomised and non-randomised clinical trials. BMJ. 1998, 317: 1185-1190.View ArticlePubMedPubMed CentralGoogle Scholar
- Concato J, Shah N, Horwitz RI: Randomised, controlled trials, observational studies, and the hierarchy of research designs. N Engl J Med. 2000, 342: 1887-1892. 10.1056/NEJM200006223422507.View ArticlePubMedPubMed CentralGoogle Scholar
- Britton A, McKee M, Black N, McPerson K, Sanderson C, Bain C: Choosing between randomised and non-randomised studies: A systematic review. Health Technol Assess. 1998, 13: 1-136.Google Scholar
- Benson K, Hartz AJ: A comparison of observational studies and randomised, controlled trials. N Engl J Med. 2000, 342: 1878-1886. 10.1056/NEJM200006223422506.View ArticlePubMedGoogle Scholar
- Lüdtke R, Weber U, Fischer I, Friese KH, Moeller H: An example on the value of non-randomisation in clinical trials in complementary medicine. Forsch Komplementärmed Klass Naturheilk. 2003, 9: 105-109. 10.1159/000057272.View ArticleGoogle Scholar
- Rosenbaum PR: Choice as an alternative to control in observational studies. Stat Sci. 1999, 14: 259-304. 10.1214/ss/1009212410.View ArticleGoogle Scholar
- Moses L: Measuring effects without randomised trials? Options, problems, challenges. Med Care. 1995, 33: AS8-AS14.PubMedGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2458/6/28/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.