Statistical Methods in Acupuncture Research


A Review

by Danica Pietrzak


Randomized control trials (RCT) were first created by Sir Austin Bradford in 1923. He had derived these mathematical models that would “describe and calibrate the complex responses of the human body to therapeutic interventions” (Meldrum et. al, 2000) [1]. These were noble attempts to capture idiosyncrasies across diverse human physiologies. By creating a method that directly compares a group of treated people to a group of untreated people, the individual and unique reactions of each person can be generalized into a symptomatology viable for people outside the treatment group. But, with the RCT, the environment of both treated and untreated groups must be carefully controlled so ideally the only difference between them is the treatment itself. When this control is achieved in a trial, RCTs perform like a dream, and truly live up to their “gold standard” image in research.

However, while RCTs may work well in particular scenarios, they have become a blanket solution in research. There is not one test that works for everything, and yet RCTs have become a nationwide standard practice, and continued support from the scientific community only furthers this problem. Because RCTs are the “standard”, trials based on this method, even poorly constructed ones, will receive more funding and interest, pushing other methods and research aside (Grossman et. al, 2005) [2]. This is a large concern as the type of research done is directly affected by the method used, not by its promise or necessity as a research area. This means that when the study’s success is altered by the choice of research method, a bias in all of research is established before the trial can even begin. This is particularly obvious with research that does not have controlled inputs. While a different research method may be more effective at producing results, RCTs are so heavily emphasized that different methods are not funded, emphasized, and published.

This concern, along with many others, is particularly prominent in the field of acupuncture research. While acupuncture has standard practices, the use of these practices varies across practitioners and standards. This creates a largely varied methodology that clashes with the standardization necessary for practicable RCTs. Another additional concern is the ineffective reporting done even with a practicable RCT trial (Mukaino et al. 2005, Wang et al. 2001, Karst et al. 2007, Spence et al. 2004) [3]. However, these struggles between the research material and methodology is not exclusive to RCTs.

In fact, observational studies also have certain design flaws. Introduced as somewhat of an alternative to RCTs, observational studies commonly step in when time, cost, or morality prevent an RCT from being effective. Observational studies are far more prone to overt and hidden biases than their cousin, the RCT, meaning a resounding verdict is more difficult to come by with his trial method (Rosenbaum et al. 2005) [4]. But observational studies will still give verdicts where RCTs are not functional. If the research area is something such as smoking, and it is suspected to be harmful to participants, then an RCT cannot simply make half of a group start or continue smoking for the sake of the trial. An observational study, instead, can monitor those making the active choice to smoke, and use their data to better understand the effects of smoking. A similar issue arises with long-term effects, especially mortality rates. Here, there is the fact that research should not, however slowly, kill its participants for the data. Another component of long-term effects lies in the name: long-term. RCTs are generally expensive and a monster to plan. Stretching a trial out years on end or including a large number of participants would practically hemorrhage money from the research group. An observational study, again, can function on longer time frames and have larger numbers of people without such dire financial consequences, meaning it is far better suited to these conditions. Observational studies, therefore, can handle the increased study size and time an RCT cannot (Frakt et al. 2015) [5].

Another study, the regression point displacement design, is a quasi-experimental design that is commonly employed when control groups are not practically or ethically feasible in situations where the treatment may be harmful. This method compares the expected results of a study with the actual results. However, this method tends to be more practical for smaller collections of data, meaning it is not as useful for large or long studies (Clark et al. 2021), which are essential components of research that can be covered via an RCT or an Interrupted Time Series [6]. This, unfortunately means that while worth mention, this method is not viable for a broad area of research such as acupuncture. 

With a different quasi-experimental design in mind, the interrupted time series shows more promise in the field of acupuncture. This application takes pre implementation trends and compares them with the post implementation data for treatments that are applied en masse, and therefore difficult to create a viable control group for. This means that the total effect of acupuncture can be monitored. However, this design is not without its flaws. Inappropriately conducted data analysis, lack of accounting for potential biases and a lack of general clarity for adequate sample size per time point (Ewusie et al. 2020) [7]. With this in mind, there is an available control group for acupuncture treatments, and the actual application of this method would more than likely require more time to appropriately devise.

It is quite idealistic to hope there would be one design to fit a large number of research areas. However, there are some methods that are more practiced and published than others, particularly in the field of acupuncture, specifically the RCT and observational study. As seen above, these studies are not without their own unique challenges, which make the recently developing methodologies that much more interesting, now and hopefully even more so in the future.

Annotated Bibliography

 Meldrum, Marcia L. "A brief history of the randomized controlled trial: From oranges and lemons to the gold standard." Hematology/oncology clinics of North America 14, no. 4 (2000): 745-760.


Grossman, Jason, and Fiona J. Mackenzie. "The randomized controlled trial: gold standard, or merely standard?." Perspectives in biology and medicine 48, no. 4 (2005): 516-534.

Mukaino Y, Park J, White A, Ernst E. The effectiveness of acupuncture for depression. Acupunct Med 2005;23:70–76 


Karst M, Winterhalter M, Munte S, et al. Auricular acupuncture for dental anxiety: A randomised controlled trial. Anaesth Analg 2008;104:295–300


Wang, Shu-Ming, Carol Peloquin, and Zeev N. Kain. "The use of auricular acupuncture to reduce preoperative anxiety." Anesthesia & Analgesia 93.5 (2001): 1178-1180.


Spence, D. Warren, et al. "Acupuncture increases nocturnal melatonin secretion and reduces insomnia and anxiety: a preliminary report." The Journal of neuropsychiatry and clinical neurosciences 16.1 (2004): 19-28.

Rosenbaum, Paul R. "Observational study." Encyclopedia of statistics in behavioral science (2005)


Frakt, Austin B. "An observational study goes where randomized clinical trials have not." Jama 313, no. 11 (2015): 1091-1092



Clark, David, Emmeline Edwards, Peter Murray, and Helene Langevin. "Implementation Science Methodologies for Complementary and Integrative Health Research." The Journal of Alternative and Complementary Medicine 27, no. S1 (2021): S-7.


Ewusie, Joycelyne E., Charlene Soobiah, Erik Blondal, Joseph Beyene, Lehana Thabane, and Jemila S. Hamid. "Methods, applications and challenges in the analysis of interrupted time series data: A scoping review." Journal of Multidisciplinary Healthcare 13 (2020): 411.

  • Danica's Notes: Review of interrupted time series, discusses strengths and weaknesses, and discusses its applications