Of all UM participants, 76% are female; therefore, for this initial investigation, we chose to focus on comparing the health of women in UM with that of a larger female population. For comparison, we adopted a substantial portion of questions from the Australian Longitudinal Study on Women’s Health (ALSWH). The ALSWH survey was inaugurated in 1996 and includes data from approximately 75,000 women [13-16].

We designed this study to answer the following questions. (1) Do women who participate in UM appear to have overall health substantially different from that of the general female population, as represented by ALSWH respondents? (2) In which specific aspects of health do UM participants differ, and in which aspects are there no differences? (3) Is UM participants’ health correlated with age, education, or their country of residence?

UM is a complementary medicine health care organization founded in 1999 near Lismore, in Eastern Australia. UM has approximately 200 male and 500 female regular attendees at its workshops and conferences. We drew the study sample from these regular participants.

Information on the UM website [17] describes the organization as follows:

This survey is the first scientific investigation to look at the health of the UM participants.

UM has developed several treatment modalities. One such modality, called Esoteric Connective Tissue Therapy, is being studied in a randomized controlled trial on chronic low back pain [18]. These treatment modalities were used by 99.7% (367/368) of UM respondents in the 12 months prior to taking the survey.

Like other institutions that support self-empowerment of women, such as the family court, women’s shelters, and social services, UM has been the target of criticism in the form of “a vigorous, determined campaign” [19] by a small number of detractors. In contrast, at least 34 registered medical professionals are among the regular visitors to UM events. The curious coexistence of harsh critics, medical professional advocates, and anecdotal evidence of substantial benefit could make UM an interesting object of study.

The full UM survey comprised 43 questions from the ALSWH, as well as a newly developed menstrual attitudes questionnaire; the latter is not discussed in this paper. We chose the ALSWH items to maximize comparability of the 2 groups. For comparisons, we used the electronic ALSWH data books [20], which give frequencies, mean scores, and, in some cases, standard deviations.

The ALSWH website [21] states that the ALSWH is

An additional 17,000 participants were added after the definition was written.

Our study population of interest comprised 500 UM-participating women who, although they were consumers of complementary medical services, had profiles that differed in important ways from those of typical complementary medicine adherents [22]: UM participants, as recorded in this survey, did not have poorer health or a higher use of registered medical professionals (nor was their use substantially lower) than the ALSWH respondents. One difference is that UM women were more likely to be middle-aged.

The portion of the data collection that was relevant to this study was a quantitative, cross-sectional online survey of women’s health. We did not collect any directly identifying data; however, we did collect indirectly identifying data such as partial medical history and age in years. Due to privacy considerations, portions of the data will not be available for inclusion in a public repository. We present age as a range, and we exclude the medical history data.

We recruited participants via 2 overlapping mailing lists of 650 and 350 UM members, and by distributing flyers at several UM-sponsored events in 2015.

Ethical approval was given by the University of Queensland School of Public Health Research Ethics Committee on June 23, 2015 (CS23062015). The first item on the survey explained the purpose of the survey and asked participants to either grant their consent or decline to do so. The study is registered with the Australian New Zealand Clinical Trials Registry (ACTRN12617000972325). Multimedia Appendix 1 shows a Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.

Deidentified data will be made available from the corresponding author on reasonable request. Data will not include potential identifiers, as outlined by Hrynaszkiewicz et al [23]. Specifically, this means that age will be categorized into intervals, and a list of medical procedures undergone and major illnesses will be excluded. Regrettably, the data cannot be shared in a public repository, as with such a small group, it would be relatively easy to identify individuals even with the above measures taken.

The survey was completed online in July to September 2015, using the Survs survey platform (Enough Pepper Lda). In the case of duplicates, we removed 1 of them.

We conducted informal focus groups with UM-participating women to help us identify the ALSWH items that seemed most relevant for comparison in this preliminary study, and not to exceed an average of 75 minutes in response time. Demographic items related to age, BMI, menstrual status, education, and lifestyle questions about smoking, alcohol use, number of general practitioner visits, and number of UM event visits were included. In addition to the 36-Item Short Form Survey (SF-36) [24], other standardized scales we incorporated were the Center for Epidemiologic Studies Depression Scale (CES-D) [25,26], Perceived Control Scale [27,28], and an ALSWH-developed multi-item summed score for perceived stress [14,29]. A further 6 items solicited information about sleep quality, 47 past diagnoses of medical issues, and frequency of 24 physical symptoms in the previous 12 months, with response options being “often,” “sometimes,” “rarely,” and “never.”

Of the ALSWH questions, 27 were asked twice, once with reference to the present, and once with reference to the time of the respondent’s first attendance at a UM event.

We calculated Cohen d using the Stata command esizei (version 14.2; StataCorp LLC). The ALSWH standard deviations are reported to only 1 significant digit, so we assumed the maximum possible standard deviation (eg, 0.1 became 0.149 and 0.0 became 0.049, reducing Cohen d), from which we used the pooled standard deviation [30] with Welch’s approximation.

A superior approach for calculating P values and effect sizes would be to compare the UM respondents with their closest ALSWH counterparts, taking account of age, education, and BMI. When the full ALSWH data are available, we will take this approach.

This survey produced 449 responses, of which 373 (83.1%) were complete. The respondents answered from 17 countries (Australia, n=273; United Kingdom, n=97; Germany, n=26; the Netherlands, n=11; other European countries, n=18; United States, n=11; rest of the world, n=13). Of the respondents, 13 did not consent, 3 were male, 20 did not give their menses status, 5 were less than 18 years of age, and 1 completed the survey twice, leaving 407 valid and 338 completed responses (Figure 1).

Table 1 and Table 2 show demographic and survey administration data. The UM group was more highly educated than the ALSWH cohort. The average UM group survey respondent’s age was 48 (range 18-86) years; age was normally distributed (Shapiro-Wilk P=.42). The proportion of smokers among UM women was 1.6% (4/240), compared with 13.9% among ALSWH women. The rate of alcohol use among UM women was 1.8% (6/338) versus 86.3% for ALSWH women.

The UM respondents’ average BMI of 21.0 kg/m² was substantially lower than the ALSWH weighted BMI of 26.1 kg/m² (see Table 3). A regression analysis showed that the BMI for the UM respondents was related to neither education (P=.63) nor age (P=.56), in contrast to data from the ALSWH and the Australian Bureau of Statistics showing obesity rates increasing to ages 55-64 (men) and 65-74 (women) [31]. A fractional polynomial graph (a visual means for determining whether a relationship is linear or, for example, U-shaped) of age and BMI shows an almost straight horizontal line; that is, there is no apparent relationship (Figure 2).

Table 3 shows the lower BMI scores for UM respondents. UM responses scored higher (better) on the 4 common psychological health scales that measure stress, the level of perceived control, depression, and the mental and physical health scales of the SF-36.

Table 4 compares the proportion of UM respondents who reported ever having had a particular health issue (including the time before they became associated with UM) with weighted ALSWH responses. The UM respondents had substantially higher rates than the ALSWH group of abnormal Pap tests and diagnoses of low iron, and substantially lower rates of diabetes, hypertension, and thrombosis.

P values use Pearson chi-square for the hypothesis that the rows and columns in the 2-way listing of UM respondents and ALSWH respondents in Table 4 are independent.

UM respondents had fewer sleep issues than did ALSWH respondents, with 68.9% (228/331) of the former and 27.9% of the latter reporting no issues at all. For lying awake most of the night, the percentages were 2.4% (UM, 8/331) versus 14.2% (ALSWH); taking a long time to get to sleep (38/331, 11.5% vs 28.8%), being kept awake by worrying (28/331, 8.5% vs 18.9%), and sleeping badly at night (44/331, 13.3% vs 35.2%) (Multimedia Appendix 2).

Multimedia Appendix 3 shows the symptoms for which the UM respondents were noticeably different (in all cases noticeably fewer incidences) from their ALSWH counterparts. The lower rates of back pain, allergies, breathing difficulties, depression, panic attacks, headaches, migraines, hot flashes, and night sweats could be of interest (Figure 3).

We detected no difference of any substance for an effect of education, including BMI, Stress Scale, Perceived Control Scale, CES-D, and SF-36. Multimedia Appendix 4 shows this in graphical form.

UM survey responses scored more highly than ALSWH responses on composite physical and mental health (see Figure 4). For ALSWH respondents, physical scores peaked at age 18 to 23 years and mental scores peaked at age 73 to 78 years. UM survey respondents’ scores were similar to peak ALSWH scores; that is, UM respondents had the higher physical scores of the very young and the higher mental scores of the very old.

Observed effect sizes (Cohen d) ranged from 0.6 to 11.9 (Table 3). These values are with one exception higher than the 0.8 considered to denote a large effect [33,34].

A cross-sectional survey is not able to establish causality, so it is possible that participants for whom UM events held appeal were already healthier. However, there are indications in the data that this was not the case. The UM survey asked respondents to answer 27 of the 43 ALSWH questions as per their memory of how they were feeling at the time of their first UM event and to answer the same questions in relation to the present. Multimedia Appendix 2 shows BMI changes. Retrospective reports are always subject to the vagaries of memory and have low reliability, but until a prospective longitudinal study is undertaken, these data are the best available. The data indicate the UM respondents were of average or below-average physical health and below-average mental health at the time of their first UM event. Examples are the proportion who often had back pain in the previous 12 months: 25.8% (UM past, 87/337) versus 3.3% (UM present, 11/336) versus 17.8% (ALSWH); those who had depression: 18.3% (61/333) versus 0.6% (2/332) versus 6.6%; and those who had allergies: 26.0% (86/331) versus 0.9% (3/332) versus 14.9%.

The results show that the UM participants, who on average were 48 years old and had been associated with UM for 8 years, had a lower BMI than the general population; experienced substantially less frequent back pain, lower stress, and depression scores; and scored higher on general mental and physical health, vitality, and perceived level of control. UM participants also had noticeably lower lifetime diagnoses of hypertension, used less hormone replacement therapy, had fewer sleep issues, and had notably fewer instances of allergies, sinusitis, anxiety, breathing difficulties, panic attacks, headaches, migraines, hot flashes, and night sweats. Major differences in lifestyle choices were apparent between the UM participants and the ALSWH population, with very low rates of both alcohol use and smoking in the UM group.

There are other groups that demonstrate aspects of above-average health. What may make it worthwhile to investigate the UM group further is the breadth of health indicators for which their scores are better than those of the general population.

If it is the case that UM participants are healthier than their population cohorts, and that they were not already healthier when they began participating in UM, this raises the question of whether UM practices or interventions could be responsible for this difference and whether there is any plausible mechanism that could explain UM’s effects, if any.

The organization’s founder, Serge Benhayon, has described UM as encompassing the concept:

One limitation of the study is that only summary ALSWH data were available for comparison. This precluded a case-control analysis, wherein UM respondents would be individually matched to ALSWH respondents. A second limitation is that the UM sample was self-selected and much smaller than the ALSWH group, although 449 respondents constitutes a large majority of the estimated 500 to 600 eligible UM event visitors (women in 2015 who had participated in at least one UM event). While this study is not able to make causal inferences about UM, it might have served to inform identification of promising variables to include in future studies. A third limitation is the possibility of response bias or memory bias among UM respondents. A fourth limitation is that 17% of UM respondents dropped out during the survey; however, the dropout population did not differ significantly in their health outcomes from the rest of the respondents.

A future study using regression analysis to uncover the association of lifestyle choices and demographics with physical and mental health may be of value. As mentioned previously, a true longitudinal, controlled study of UM participants would be worthwhile, as it would give a more accurate picture of participants’ initial health levels, the type and level of specific UM services they access, and the trajectory of their health changes. Such investigation of dose-response relationships and causal factors could inform development of lifestyle modification programs that individuals could use in concert with medicine.

UM-participating women appear to be notable in that they exhibit better than average health and are not members as a result of any competitive selection, unlike senior Whitehall public servants, who are selected on a competitive basis; nor are UM-participating women part of a population with a limiting cultural or religious tradition. UM women also come from a wide range of backgrounds, ages, and countries. Identifying a group of women who are not part of the obesity epidemic and measuring their other health indicators may be of use for further research. Therefore, further research on this group of women could be worthwhile.