Introduction

The World Health Organization (WHO) estimates that over 400 million people have diabetes worldwide, and this is projected to increase to reach 592 million by 2035 [1]. Poorly managed diabetes can lead to serious and possibly fatal complications such as cardiovascular disease, renal failure, nerve damage and blindness [2,3,4].

Diabetes mellitus (DM) is a chronic metabolic disorder in which blood glucose levels are higher than normal for a long period of time. These high blood glucose levels are attributed to abnormal disturbances of insulin production and/or function [5]. Diabetes is caused by either lack of insulin production by the pancreas (type 1 diabetes, T1D), when the amount of insulin produced by the pancreas is insufficient to carry out all blood glucose regulation processes, or by decreased insulin sensitivity by the body cells (type 2 diabetes, T2D). Diabetes can also be caused by a combination of low insulin production as well as low insulin sensitivity or be due to hormonal dysregulation in pregnancy [5].

Self-care practices relevant to self-management of diabetes include adherence to prescribed treatment and clinical management plans, adopting a healthy lifestyle and having a balanced diet [6]. In addition, many patients also use complementary and alternative medicine (CAM) [7]. The WHO defines CAM as a ‘broad set of health care practices that are not part of that country’s own tradition or conventional medicine and are not fully integrated into the dominant health-care system’ [8].

CAM use is known to be prevalent in patients with diabetes as a supplement to their existing orthodox diabetes treatments, as a replacement, or for reasons that might not be directly related to diabetes such as using CAM for energy and general wellbeing [7]. Various factors may influence CAM use by patients with diabetes. A study of 3978 U.S. adults suggested that CAM use by patients who were diabetic for more than 10 years or patients who had a functional limitation caused by diabetes were more likely to use CAM compared to patients with less severe diabetes [9]. In addition, the study reported that 77% of patients who used CAM for the treatment of diabetes used CAM as a supplement to conventional treatment, while 23% used CAM as a replacement [9].

CAM users often perceive CAM to be an effective means of lowering blood glucose levels and treating side effects of prescribed diabetic medications [10,11,12,13,14,15]. However, adverse outcomes of CAM use have also been reported. For example, CAM can affect the management of diabetes by either direct herb–drug interaction with the use of herbal remedies or indirectly by affecting medication adherence when using herbal or any other CAM types [6].

There is a lack of an up-to-date systematic review that investigates the prevalence of CAM use by patients with diabetes. Patient sources of health-related information have changed immensely in the past decade [16]. In particular, increasing availability and use of web-based information sources, including social media and online health information in recent years, may encourage and inhibit CAM use in long term health conditions [17]. An up-to date systematic review on the prevalence of CAM use by patients with diabetes will help healthcare professionals to consider patient use of CAM when counselling patients, supporting adherence and identifying the risks of interactions and adverse effects when CAMs are used in conjunction with prescribed treatments.

The aim of this study was to systematically review the global prevalence of CAM use amongst adults with diabetes. Specific objectives were to identify the types of CAM that are used by the population with diabetes and to identify differences in CAM use amongst different populations with diabetes, including types of diabetes, demographic characteristics, duration of diabetes and presence or absence of diabetic complications.

Methods

This systematic review was informed by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and checklist [18]. A protocol was developed as per the PRISMA protocol guideline (protocol ID CRD42019125036).

Data sources and searches

Cochrane Library, MEDLINE, Embase, CINAHL, AMED, Web of Science, Google Scholar and PROSPERO databases were searched for the past 10 years covering 2009 to June 2019. Open Grey was searched for grey literature. Search terms and an example search strategy are listed in Supplementary Table 1. The review was restricted to studies published in English. Studies that recruited participants who are adult patients with diabetes, 18 years of age and older and reported partially or exclusively the prevalence and use of CAM amongst patients with diabetes were included. Studies which either focused on CAM use in conjunction with conventional treatments or as a replacement were considered.

Study selection

Screening and selection were performed independently by two review authors (AA, VP) and were carried out in three phases. Titles and abstracts were screened for inclusion of possible relevant studies followed by assessment of full texts for eligibility. Reference lists of included studies were screened. If a title was considered relevant; the study was manually searched and the abstract examined.

Data extraction and quality assessment

Data on study characteristics, prevalence of CAM use as well as types of CAM used by patients with diabetes were extracted. Two review authors (AA, VP) independently assessed the quality of included studies using the critical appraisal tool from the Joanna Briggs Institute (JBI) checklist [19]. Studies were classified into high, moderate and low quality based on the results of the JBI checklist (Supplementary Table 1). The quality assessment in included studies was focused on three fields: clarity of inclusion criteria and study setting and sampling, appropriateness of approaches to data collection and analysis, and outcome measurement (i.e. use of CAM). Included studies were judged to be of ‘high quality’ if quality criteria were satisfied by at least 7 items, ‘moderate quality’ for scores of 3–6 and ‘low quality’ for scores ≤2 [20]. All studies were included regardless of their quality.

Data synthesis and analysis

A quantitative synthesis of aggregate level data on prevalence was performed. Study specific results were reported as percentage prevalence with exact 95% confidence intervals (95%CI). When sufficient data were available for within-study comparisons of prevalence between dichotomous groups, e.g. sex, then relative prevalence ratios (PRs) together with 95% confidence intervals were calculated. Meta-analyses of proportions and PRs were performed using a random effects model fit using the method of Der Simonian & Laird [21]. Heterogeneity was assessed using the I2 statistic, the between study standard deviation and calculation of 95% prediction intervals (95%PrI) for the prevalence in a new study [22, 23]. Data are presented in forest plots which include pooled estimates where appropriate. All analyses were performed using STATA version 15.

Results

A total of 2623 unique titles were screened of which 38 articles met the inclusion criteria (Fig. 1). After applying quality assessment, studies fell into these categories (8 high quality studies, 30 moderate quality studies and no low quality studies). Details of critical appraisal results are available in Supplementary Table 2.

Fig. 1
figure 1

Prisma flow diagram

Study characteristics

Included studies originated from 25 different countries. Participants were mostly recruited from diabetes clinics and healthcare centres (Table 1). Fifteen of the studies enrolled participants with either T1D or T2D, and 23 studies only included patients with T2D. Out of the included 38 studies, 37 were cross-sectional surveys and one analysed data from another cohort study (Table 1).

Table 1 Study characteristics

Types of CAM

Sixteen studies focused exclusively on herbal and nutritional supplement use by patients with diabetes (Table 1). The remaining 22 studies discussed other CAM types. Fourteen of those 22 studies that investigated other CAM types also reported the use of herbal and nutritional supplements as a form of CAM [12,13,14, 25, 32, 34, 36, 38, 39, 41, 49, 52,53,54]. A total of 35 different CAM types were reported in at least one study. CAM types used by patients with diabetes and mentioned in the most studies were acupuncture (n = 6 studies), Mind–body therapies (n = 6 studies) religious and spiritual healing (n = 5 studies) and homoeopathy (n = 4 studies) (Table 2).

Table 2 List of complementary and alternative medicine types as cited by included studies

Within the 31 studies which reported the use of herbal and nutritional supplements by patients with diabetes, a total of 223 different herbal and nutritional supplements were reported (Supplementary Table 3). The five herbs that were mentioned in the most studies were, cinnamon (Cinnamomum verum) and fenugreek (Trigonella foenum-graecum) each reported in 18 different studies, garlic (Allium sativum) reported in 17 studies, aloe vera (Aloe Vera) reported in 14 studies and black seed (Nigella sativa) reported in 12 studies.

Prevalence of CAM use

The highest prevalence of CAM (all types) use was reported at 89% by two studies, one each from India and Jordan followed by studies in Tanzania (78%), Sri Lanka (76%) and Iran (75%) (Table 3) [12, 35, 39, 45, 46]. The lowest prevalence of CAM use was 17% as reported by a study conducted in Jordan [15]. A study in Australia reported a prevalence of 8%, but the study gathered data from patients about their visits to CAM practitioners only and did not include data on CAM use in general by patients with diabetes [44]. Other studies reporting the lowest prevalence of CAM use included studies in Libya (29%), Saudi Arabia (26%), USA (26%), Israel (23%) and Jordan (17%) [15, 26, 29, 43, 52]. Pooled prevalence of CAM use was 51% (95%CI 43%, 59%). However, heterogeneity was very high (I2 = 99%) with the predictive interval ranging from 8% to 93%. (Fig. 2).

Table 3 Included studies’ overall and subgroups prevalence of CAM use
Fig. 2
figure 2

Forest plot showing pooled prevalence of complementary and alternative medicine in diabetes

Subgroup analysis

Study level factors

Meta-analysis was conducted for results stratified at the study level by continent. I2 was 97.5% and predictive intervals were found to be wide. The highest prevalence rates of 76% were observed in Europe (PrI inestimable), followed by Africa 55%, (95%PrI 0.17, 0.90) from seven studies. The lowest prevalence rates were observed in North America 45%, (95%PrI 0.04, 0.92) from five studies (Fig. 3).

Fig. 3
figure 3

Meta-analysis of study level factors in relation to CAM use (prevalence proportion by continent). CAM, complementary and alternative medicine

Patient level factors

Subgroup analyses were conducted across ethnicity (reported in eight studies). All meta-analyses at subgroup level also showed high levels of heterogeneity. Results were as follows: for the ethnicity subgroup, no predictive interval could be estimated other than PrI for the group of ‘other ethnicities’ prevalence ratio 0.57 (95%CI 0.39–0.75); the estimated predictive intervals ranged between 0.00 and 1.00, I2 = 64.05%) (Fig. 4).

Fig. 4
figure 4

Ethnicity subgroup forest plot showing prevalence of CAM use. CAM, Complementary and alternative medicine

For analysis stratified by binary subgroups within-study comparative data were extractable for sex (31 studies), type of diabetes (7 studies), duration of diabetes (15 studies) and presence or absence of diabetic complications (10 studies). Within study pooled estimates PRs for patients with no diabetic complications versus patients with diabetic complications gave a prevalence ratio (PR) 0.81 (95%CI 0.66, 0.99), (95%PrI 0.39–1.67) (I2 = 89%) (Supplementary Fig. 1). For patients who had had diabetes for more than 5 years versus less than 5 years pooled PR was 1.71 (95%CI 1.04, 1.32), (95%PrI 0.73, 1.88) (I2 of 83%) (Supplementary Fig. 2). For male versus female participants, pooled PR was 0.86 (95%CI 0.81, 0.91), (95%PrI 0.64, 1.16) (I2 of 72%) (Supplementary Fig. 3). Pooled PR for patients with T2D versus T1D patients was 1.00 (95%CI 0.83, 1.20), 95%(PrI 0.56, 1.77) (I2 = 75%) (Supplementary Fig. 4).

Additional outcomes

CAM as a complementary or alternative treatment

Eight of the 38 included studies assessed whether CAM was used as an additional treatment or as an alternative treatment to conventional medicines. Prevalence of CAM use as an additional treatment to prescribed medicine was 78% (95%CI 56%, 94%) with 95% PrI (4%, 1.00%) (I2 = 98%), and the percentage of patients who used CAM as an alternative to their prescribed medicine was 21% (95%CI 12%, 31%) with 95% PrI (0.%, 63%) (I2 = 89%) (Supplementary Figs. 5, 6).

Patients’ disclosure of CAM use to healthcare professionals

The percentage of patients who do not disclose their CAM use to healthcare professionals was 67% (95%CI 58%, 76%) with 95% PrI (25%, 97%) (I2 = 98%) (Fig. 5).

Fig. 5
figure 5

Patient who do not disclose the use of CAM to the healthcare professionals. CAM, complementary and alternative medicine

Discussion

This study provides up-to-date data on the global prevalence of CAM use by patients with diabetes as reported in the peer reviewed research literature. The last literature review on CAM use of patients with diabetes was published in 2007 [7] which reviewed studies conducted in nine countries and reported prevalence ranging from 17% to 73%. A similarly wide variation in prevalence rate of 8–89% was observed in our updated review that included studies from 25 countries.

According to the included studies, CAM use is common amongst patients with diabetes for the purpose of diabetes management. Most of the studies showed that the participants they recruited used CAM as an additional approach to conventional treatment, while in other studies, the reason for CAM use (additional or alternative) was not specified. Only seven studies reported that some patients with diabetes used CAM as the sole means of managing their diabetes. Most of the included studies were conducted in healthcare settings. Therefore, patients who do not use conventional treatments for diabetes may not have been included. The prevalence of patients with diabetes who use CAM in the general population with diabetes is hence likely to be higher than the estimates provided by the included studies.

The meta-analysis of the prevalence data demonstrated extreme variation in prevalence of CAM use amongst patients with diabetes across studies.

Strengths and limitations

This systematic review was a protocol driven review with a pre-specified aim, objectives and methodology. A range of relevant databases were used which covered the prevalence of CAM used by patients with diabetes globally. Data collection methods varied amongst studies. Some studies used structured questionnaires while other studies used semi-structured questionnaires. In addition there was a wide variation in the nature of the study settings. The content of the questionnaires used to collect the data on prevalence and nature of CAMs are likely to influence patient response. Therefore, the included studies may have underreported the nature and extent of CAM use by study participants. In addition, our study only included studies published in the English language.

Implications for practice and research

This systematic review shows that CAM use amongst patients with diabetes is prevalent in many populations. This review suggests that healthcare professionals should consider use of CAM by patients with diabetes when advising them about using their prescribed treatments and monitor their medication adherence while using any forms of CAM. They should also be aware of patients’ use of herbal supplements as some forms of herbal medicine can lead to herb–drug interactions [57]. For example, the most frequently mentioned herbal supplement used by patients with diabetes was cinnamon. It is reported that cinnamon has a potentiating effect on diabetic drugs increasing the risk of hypoglycaemia [58]. For example, cinnamon shows an inhibitory effect on the CYP3A4 enzyme in rabbits which potentiates the effect of pioglitazone if combined with cinnamon, leading to a hypoglycaemic effect [59]. Aloe vera, which was the most frequently reported CAM by the studies included in our review, has been linked to potential interaction with 45 different drugs, including diabetic drugs such as glimepiride [60]. Concomitant use of Aloe vera and glimepiride can produce hypoglycaemic effects as Aloe vera has an inhibiting effect on ATP sensitive potassium channels in pancreatic β cells leading to additional release of insulin [61].

Understanding CAM use patterns and considering any possible interactions between them and other potential medications could help healthcare professionals to appropriately minimize drug-related problems or herb–drug interactions. It could help them to encourage their patient to discuss their CAM use and offer the opportunity to provide better advice for patients with diabetes.

The observed prevalence of CAM, and the many varieties of CAM that are used by patients with diabetes, call for revision of diabetes management guidelines. The National Institute for Health and Care Excellence (NICE) guideline on management of diabetes does not explicitly advise healthcare professionals to discuss patient use of herbal medicines or CAM in their consultation [6]. Guidelines should enable healthcare professionals to counsel patients with diabetes, their families and carers, who should all be educated about the safe use of CAM in conjunction with prescribed medicines.

Owing to the variable and often high prevalence of CAM use amongst patients with diabetes worldwide, research that generates evidence-based information about CAM is needed. This includes effectiveness and safety profiles of commonly used CAMs, including herbal medicines as identified in this systematic review.

This systematic review has identified that on average up to two-thirds of patients who use CAM do not disclose this to their healthcare professionals. Use of CAM such as herbal medicines could be incorporated as part of the comprehensive medication review services offered at community pharmacies and primary care [62].

Future studies need to consider the perspectives of patients with diabetes who do not visit conventional healthcare facilities for the management of diabetes to provide a better estimate of prevalence rates. In addition, there is a need to gather evidence on the factors that affect use and non-use of CAM by patients with diabetes.

Conclusion

A wide variation in prevalence rate of CAM use in diabetes (8–89%) was observed, and pooled prevalence of CAM use was 51%. Our findings show that CAM use by patients with diabetes is common. Healthcare professionals should be aware of the use of CAM by patients with diabetes to ensure treatment optimization and medication adherence. Future studies should incorporate patient and healthcare professionals’ perspectives of CAM use in diabetes, evaluate patient outcomes through the use of healthcare databases and carefully designed prospective studies, and identify opportunities to promote rational use of CAM through evidence-based guidelines and patient-centred approaches.