만성피로

Background: Tiredness is one of the most frequent complaints in primary care. Although often self-limiting and frequently associated with psychosocial stress, patients but also their physicians are often uncertain regarding a serious cause and appropriate diagnostic work-up. We conducted a systematic review and meta-analysis of studies reporting on differential diagnosis of fatigue in primary care.

Methods: MEDLINE, EMBASE and conference abstracts were searched for primary care based studies of patients presenting with tiredness. Twenty-six studies were included. We report on anaemia, malignancy, serious organic disease, depression and the chronic fatigue syndrome (CFS) as causes of tiredness as presenting complaint.

Results: We found considerable heterogeneity of estimates which was reduced by limiting our analysis to high quality studies. Prevalences were as follows-anaemia: 2.8 % (CI (confidence interval) 1.6–4.8 %); malignancy: 0.6 % (CI 0.3–1.3 %); serious somatic disease: 4.3 % (CI 2.7–6.7 %); depression 18.5 % (CI 16.2–21.0 %). Pooling was not appropriate for CFS. In studies with control groups of patients without the symptom of tiredness, prevalence of somatic disease was identical to those complaining of tiredness. Depression, however, was more frequent among those with tiredness.

Conclusions: Serious somatic disease is rare in patients complaining of tiredness. Since prevalence is similar in patients without tiredness, the association may not be causal. Extensive investigations are only warranted in case of specific findings from the history or clinical examination. Instead, attention should focus on depression and psychosocial problems.

Electronic supplementary material: The online version of this article (doi:10.1186/s12875-016-0545-5) contains supplementary material, which is available to authorized users.

Search result and study selection: The electronic literature search was carried out in October 2010 and produced 3733 publications. 83 of them met our inclusion criteria. A few studies resulted in more than one publication so that we found 81 relevant studies overall. Forty-six studies dealt with the aetiology of tiredness. We excluded 20 studies in which patients were systematically asked about tiredness, the remaining 26 studies investigating tiredness as a reason for encounter or secondary complaint form the sample of our review. Further details of the selection process are given in a flow-chart (Fig. 1).Fig. 1Flow of search

Included studies: Most studies were conducted in Europe (n = 12). The number of included patients ranged from 22 to 10,279. Detailed information about the included studies is given in Table 1 and Additional file 2.Table 1Characteristics of the included studiesFirst author, yearCountrySample sizeSymptom definitionAnaemiaMalignancySevere somatic diseaseDepressionCFSFriedlander, 1962 USA71“Unbearable fatigue”Morrell, 1972 GB58New symptom that wasn’t brought up at a doctor’s visit during the last 12 months, main reason for encounterXMorrison, 1980 USA176Cause unknownXXXJerrett, 1981 GB300Tiredness as a central problem (“tiredness”, “need of tonic”, “run down”, etc.)XXXXSugarman, 1984 USA118Cause unknownXXXKnottnerus, 1986 NL174New SymptomXNelson, 1987 USA71At least 1 month, main reason for encounterKroenke, 1989 USA823-year incidence of the symptom of 1000 randomly selected patient filesValdini, 1989 USA22Since at least 1 year, cause unknownXXXKirk, 1990 USA71Since at least 1 month, important problemXCathebras, 1992 CDN93Through direct questioning (“Why did you come here today?”) or if the doctor indicated that tiredness was a reason for encounterXXXElnicki, 1992 USA52Since > 1 month, cause unknown, main reason for encounterXXXXGerber, 1992 USA88Tiredness as one of the chief complaintsXRidsdale, 1993 GB220“being knackered, lethargic, run down or tired all the time”, since at least 2 weeks, main reason for encounterXXXFuhrer, 1994 F287Since at least 2 weeksXHall, 1994 GB197Diagnosis of tiredness in the patient’s fileXMaeno, 2002 J157“General fatigue”, main reason for encounterXAndrea, 2003 NL322“Fatigue-related visite”XDarbishire, 2003 GB141Since at least 6 months, cause unknownXXGialamas, 2003 AUS342“Tiredness”, “fatigue”, “weariness”, “weakness”, “lethargy” or “malaise” in patient’s recordXXKenter, 2003 NL10.297Episodes of care with tiredness as the reason for encounterXXVital Durand, 2004 F120Since at least 6 months, cause unknownBelanger, 2005 CDN36“Fatigue” (no precise definition)Kenter, 2007 NL385Diagnosis “tiredness”XKoch, 2009 NL296New symptom, cause unknownXXNijrolder, 2009 NL571New symptom (no encounter due to tiredness in the last 6 months), main reason for encounterXXXXX

Quality of included studies: Nineteen of the included studies had a prospective design, 5 were retrospective analyses of clinical records and 2 were based on registries. In 15 studies consecutive recruitment was mentioned explicitly. The inclusion and exclusion criteria were well-defined in most of the studies (n = 24). Only 11 studies were conducted in more than one health care facility. The number of drop-outs and the reasons for withdrawal were specified in 12 studies. In 20 studies the diagnostic categories were well-defined. In 21 studies the reference standards to detect underlying illnesses were appropriate. Participating patients were subjected to the same reference standard in 12 studies. Exactly the same diagnostic investigations were applied to the patients in 8 studies. Six of the 26 studies had a control group of non-fatigued patients. Details are shown in Table 2.Table 2Methodical quality of the included studies+ YesWas the recruitment of patients prospective?Where all patients with the symptom included in the study?Are inclusion and exclusion criteria precisely defined?Was the study design multicentered?Were the number of drop-outs and the reasons for withdrawal specified?Were the diagnostic categories well-defined?Was the reference standard that was used appropriate?Was every patient subjected to the reference standard?Were the same diagnostic procedures applied to all patients?Was there an appropriate comparison group?− No? Unclear+/−Criteria fulfilled for part of the aetiologiesAndrea, 2003 +++−−?−−−+Belanger, 2005 ++++?++++−Cathebras, 1992 +++−−?+++/−+Darbishire, 2003 +?++?+++/−+/−−Elnicki, 1992 +++−−++++/−−Friedlander, 1962 +?−−−+−−−−Fuhrer, 1994 ++++?+++++Gerber, 1992 +++−+++++−Gialamas, 2003 −−++++++/−+/−−Hall, 1994 −−−−−++c??−Jerrett, 1981 +++−−++/−+/−+/−−Kenter, 2003 −++++++−−+Kenter, 2007 ++++?+++++Kirk, 1990 +++++?+++−Knottnerus, 1986 +?+−++++++Koch, 2009 +++++++++−Kroenke, 1989 −?+−+?+−−−Maeno, 2002 +++−+++++−Morrell, 1972 +++−+?+/−−−−Morrison, 1980 −?+−+++−−−Nelson, 1987 +?++−?−−−?Nijrolder, 2009 +?+++++−−−Ridsdale, 1993 +?++−+++/−+/−−Sugarman, 1984 −++−++++/−+/−−Valdini, 1989 +?+−−++++/−−Vital Durand, 2004 +++?−+++−−

Tiredness in primary care: underlying diseases: We present results for the following etiological categories: anaemia, malignancies, serious somatic diseases, depression and CFS (Fig. 2).Fig. 2Quantitative analysis of prevalence rates of anaemia, malignancies, severe somatic diseases and depression among patients presenting with fatigue in primary care of selected studies

Thirteen studies assessed the frequency of anaemia in the respective study population. The prevalence rates ranged from 0 to 6.9 %; see Fig. 3. Three studies used precisely defined diagnostic criteria and an appropriate diagnostic work-up. The pooled estimate within this group was 2.8 % (95 % CI: 1.6–4.8) (see Fig. 3).Fig. 3Forest plot, frequencies of anaemia and quantitative analysis

Knottnerus et al. reported the prevalence of anaemia in a control group and, interestingly, found no significant difference between tired patients and those consulting for other reasons. The test used a type 1 error of level alpha of 0.05 and had a power (1-ß) of 0.09 to find a difference of 0.5 mmol/l in the mean haemoglobin level.

Three studies reported the prevalence of malignancies among patients complaining of tiredness. The prevalence rate of malignancies in our quantitative synthesis was 0.6 % (CI 0.3–1.3 %) (see Fig. 2). Jerrett, who diagnosed a carcinoma of the lung in one patient, points out that the patient’s main complaints were cough and dyspnea. Tiredness was only an additional complaint.

Eleven studies analysed the frequency of serious somatic diseases (see Fig. 4). In this category we included anaemia, malignancies and other serious illnesses associated with tiredness for which treatments are available. We conducted a meta-analysis with 3 studies that had prospective study designs and used defined diagnostic standards. The prevalence rate of serious somatic diseases was 4.3 % (CI 2.7–6.7 %) (see Fig. 3).Fig. 4Forest plots, frequencies of severe somatic diseases and quantitative analysis

Diabetes, anaemia and hypothyroidism were most frequent. In the study of Cathebras et al. serious somatic illnesses did not occur more frequently among the fatigued than among the non-fatigued study population.

Depression was investigated in 14 studies (see Fig. 5). Six studies used standardized instruments (HADS (Hospital Anxiety and Depression Scale), SCL-90 (Symptom Check List), CES-D (Center for Epidemiological Studies Depression Scale) ). These latter studies found that up to 68.2 % of the fatigued study population scored positive. However, because of somatic and mental comorbidities these estimates may be too high. In the study of Cathebras et al. all recruited patients underwent a structured interview (DIS (Diagnostic Interview Schedule). Here about 17 % (CI 10.2–26.4 %) of the patients presenting with tiredness were classified as being depressive. Most studies relied on GPs diagnosing depression according to their routine.Fig. 5Forest plots, frequencies of depressions and quantitative analysis

We performed a quantitative synthesis of 6 of these studies. Current depression was identified in 18.5 % (CI 16.2–21.0 %) of the patients (see Fig. 2). Three studies, which were not included in the meta-analysis for methodological reasons, found a significantly higher prevalence of depression in fatigued compared to non-fatigued patients.

Three studies presented estimates of the frequency of CFS among fatigued primary care attenders. Two reported low rates of 1.9 % (CI 0.0–10.3 %) and 0.7 % (CI 0.2–1.8 %) (see Fig. 2). Darbishire et al. who used the CDC (Centers for Disease Control and Prevention) definition reported a high rate of 31.2 % (CI 23.7–39.5 %). They only included patients who had suffered from tiredness for at least 6 months and had not received a diagnosis associated with tiredness. Hence two important diagnostic criteria of CFS were met by all included patients, which explains the high prevalence. Given the small number of studies we did not pool results.

Background: Chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME) has been emerging as a significant health issue worldwide. This study aimed to systemically assess the prevalence of CFS/ME in various aspects of analyses for precise assessment.

Methods: We systematically searched prevalence of CFS/ME from public databases from 1980 to December 2018. Data were extracted according to 7 categories for analysis: study participants, gender and age of the participants, case definition, diagnostic method, publication year, and country of the study conducted. Prevalence data were collected and counted individually for studies adopted various case definitions. We analyzed and estimated prevalence rates in various angles: average prevalence, pooled prevalence and meta-analysis of all studies.

Results: A total of 1291 articles were initially identified, and 45 articles (46 studies, 56 prevalence data) were selected for this study. Total 1085,976 participants were enrolled from community-based survey (540,901) and primary care sites (545,075). The total average prevalence was 1.40 ± 1.57%, pooled prevalence 0.39%, and meta-analysis 0.68% [95% CI 0.48–0.97]. The prevalence rates were varied by enrolled participants (gender, study participants, and population group), case definitions and diagnostic methods. For example, in the meta-analysis; women (1.36% [95% CI 0.48–0.97]) vs. men (0.86% [95% CI 0.48–0.97]), community-based samples (0.76% [95% CI 0.53–1.10]) vs. primary care sites (0.63% [95% CI 0.37–1.10]), adults ≥ 18 years (0.65% [95% CI 0.43–0.99]) vs. children and adolescents < 18 years (0.55% [95% CI 0.22–1.35]), CDC-1994 (0.89% [95% CI 0.60–1.33]) vs. Holmes (0.17% [95% CI 0.06–0.49]), and interviews (1.14% [95% CI 0.76–1.72]) vs. physician diagnosis (0.09% [95% CI 0.05–0.13]), respectively.

Conclusions: This study comprehensively estimated the prevalence of CFS/ME; 0.89% according to the most commonly used case definition CDC-1994, with women approximately 1.5 to 2 folds higher than men in all categories. However, we observed the prevalence rates are widely varied particularly by case definitions and diagnostic methods. An objective diagnostic tool is urgently required for rigorous assessment of the prevalence of CFS/ME.

Characteristics of the included studies: From four major public databases, a total of 1291 articles were initially identified, and 45 articles finally met the inclusion criteria for this study. The first study of selected articles was reported in 1990, and the latest one was in 2018. Forty-six studies (one article contained two studies) were conducted in 13 countries (41 prospectively and 5 retrospectively). Four studies applied multiple case definitions (2 studies applied 4 case definitions, 2 studies applied 3 case definitions), providing a total of 56 prevalence data (Fig. 1). The total number of participants was 1085,976 from 30 community-based studies (33 data, n = 540,901) and 16 primary care studies (23 data, n = 545,075). Thirty-four studies (44 data) focused on the general population (n = 956,526, age ≥ 18 years), 7 focused on children and adolescents (n = 117,307, < 18 years), and 5 focused on a particular population (n = 12,143; nurses, 2; company employees, 1; livestock employees, 1; Gulf War veterans, 1). The average number of participants in each study was 23,608 ± 48,092 (± SD). Of the 46 studies, 21 studies (24 data) reported gender-related information (males, n = 62,070; females, n = 68,772) (Table 1).Table 1Characteristics of studies on the prevalence of CFS/ME (± SD)GroupCommunityPrimary careTotalNumber of articles included (%)30 (67)15 (33)45 (100)Number of studies includeda(%)30 (65)16 (35)46 (100)Prospective25 (54)16 (35)41 (89)Retrospective4 (9)1 (2)5 (11)Number of prevalence datab (%)33 (59)23 (41)56 (100)Total number of participants540,901545,0751085,976Mean N. of participants18,030 ± 38,09434,067 ± 61,32523,608 ± 48,092Total number of known sexc120,76510,077130,842 Male58,752331862,070 Female62,013675968,772 M:F ratio48:5225:7545:55Mean age of participantsd41.1 ± 11.338.0 ± 6.540.0 ± 9.9 Male41.1 ± 11.338.0 ± 6.839.9 ± 10.0 Female41.2 ± 11.438.1 ± 6.140.0 ± 9.8N. of (studies)/prevalence data by subgroup (N. of participants) Publication year 1990–20008(177,201)15(318,391)(18)/23(495,592) 2001–201019(86,512)4(9895)(20)/23(96,407) 2011–20186(277,188)4(216,789)(8)/10(493,977) Population group General population (≥ 18 years)24(489,961)20(466,565)(34)/44(956,526) Children/adolescents (< 18 years)6(43,671)1(73,636)(7)/7(117,307) Specific populatione3(7269)2(4874)(5)/5(12,143) Case definition (8 case definitions)23(339,192)11(298,739)34(637,931) CDC-1994 4(20,037)4(27,454)8(47,491) Holmes 2(116,520)2(2980)4(119,500) Australian 2(3215)2(2980)4(6195) Oxford 1(10,396)3(505,299)4(515,695) CCC ECD PVES NICE f N/Ag1(59,101)1(1874)2(60,975) Country (13 countries) Western27(516,617)17(473,009)(34)/44(989,626) Asian5(23,197)5(70,192)(10)/10(93,389) Othersh1(1087)1(1874)(2)/2(2961) Diagnostic method Interview (medical test −)19(111,943)3(68,848)(19)/22(180,791) Interview (medical test +)9(57,339)14(17,445)(18)/23(74,784) Physician diagnosis1(114,000)5(435,782)(4)/6(549,782) Medical records4(257,619)1(23,000)(5)/5(280,619)aThe number of studies is larger (n = 46) than the number of article as one article included two studiesbSome articles included multiple applications of case definitions; thus, the number is larger than the total number of studiescTwenty-one studies (24 prevalence data points) included information about participant sexdThe mean age of the participants whose sex was known (12 studies) was estimated using either the reported mean age for each sex or the mean age for both. Children and adolescents were excludedeSpecific groups included nurses, Gulf War veterans, livestock workers, company employees, etcfCCC, Canadian Consensus Criteria; ECD, epidemiological case definition; PVES, post viral exhaustion syndrome; and NICE, National Institute for Health and Care Excellence guideline 2007gTwo N/A cases had no verification of case definition or defined criteriahOther countries included India and Nigeria

A total of 8 case definitions were adopted in 46 studies (56 data); the most frequently used definitions were CDC-1994 (34 data) and Holmes (8 data). The majority of the studies (37 studies/45 data) used interview-based diagnoses with/without medical tests, and the remaining studies (9 studies/11 data) were based on a physicians’ diagnosis/determination and medical record reviews. The average age of all participants except children and adolescents, based on the mean of the means/median ages reported in the articles, was 40.0 ± 9.9 years (Table 1).

Overview of CFS/ME prevalence: The average prevalence of CFS/ME based on the 56 prevalence data reported was 1.40 ± 1.57% (95% CI: 0.98–1.82), and the pooled prevalence was 0.39% (5370 CFS/ME patients of 1387,787 participants) (Table 2). The meta-analysis yielded an estimate of 0.68% (95% CI: 0.48–0.97) with high heterogeneity I2 = 99.4% (Table 3, Fig. 2). For the general population (excluding data for children and adolescents and specific populations), the average prevalence was 1.45 ± 1.68% (95% CI: 1.01–1.89) and 0.38% (4724 CFS/ME patients of 1258,337 participants) and 0.65% (95% CI: 0.43–0.99) in the meta-analysis based on 44 prevalence data (Tables 2 and 3, and Fig. 3).Table 2Prevalence of CFS/ME by subgroup (± SD)GroupaMale (%)Female (%)Total (%, M/F)Total (%)(21 studies/24 data)(46 studies/56 data)Average prevalence of all studies1.11 ± 1.052.24 ± 2.591.67 ± 2.061.40 ± 1.57Pooled prevalence of all studies (Total N. of CFS/N. of participantsb)0.74 (451/61,069)1.92 (1308/68,124)1.37 (1778/129,780)0.39 (5370/1387,787)Pooled prevalence of the general population (N. of adult CFS/N. of adult participants)0.75 (451/60,432)1.92 (1304/67,790)1.38 (1774/128,809)0.38 (4724/1258,337)Mean age of CFS patients (12 studies)39.3 ± 7.839.1 ± 7.640.4 ± 7.740.4 ± 7.7Prevalence by subgroup (N. of studies that reported sex) Study participants Community (16)1.03 ± 1.132.31 ± 2.881.67 ± 2.281.56 ± 1.80 Pooled prevalence (N. of CFS/N. of participantsb)0.70 (404/57,751)1.94 (1190/61,365)1.34 (1594/119,116)0.73 (4014/548,461) Primary care (5)1.39 ± 0.561.96 ± 0.741.68 ± 0.721.16 ± 1.13 Pooled prevalence (N. of CFS/N. of participantsb)1.42 (47/3318)1.75 (118/6759)1.64 (165/10,077)0.21 (1739/839,326) Publication year 1990–2000 (9)0.62 ± 0.581.26 ± 1.420.94 ± 1.130.96 ± 0.91 2001–2010 (10)1.35 ± 1.182.99 ± 3.112.17 ± 2.492.08 ± 2.01 2011–2018 (2)1.75 ± 0.831.78 ± 0.711.76 ± 0.770.84 ± 0.86 Population group General population (≥ 18 years) (19)1.39 ± 1.052.83 ± 2.612.11 ± 2.071.45 ± 1.68 Children/adolescents (< 18 years) (0)–––0.89 ± 0.82 Specific populationc(1)0.120.060.09 ± 0.031.62 ± 1.17 Case definition (8 case definitions) CDC-1994 (16)1.24 ± 1.042.61 ± 2.751.93 ± 2.191.46 ± 1.34 Holmes (4)0.07 ± 0.050.14 ± 0.150.11 ± 0.120.34 ± 0.40 Australian (1)2.655.233.94 ± 1.292.52 ± 2.99 Oxford (2)1.23 ± 0.641.76 ± 1.221.51 ± 1.001.73 ± 1.35 CCC ECD PVES NICE d(0)–––0.53 ± 0.77 Country (13 countries) Western (12)1.14 ± 0.972.40 ± 2.861.77 ± 2.221.32 ± 1.45 Asian (8)1.23 ± 2.922.06 ± 1.851.65 ± 1.641.51 ± 1.74 Otherse(1)0.110.500.31 ± 0.202.65 ± 2.37 Diagnostic method Interview (medical test -) (9)1.70 ± 1.214.32 ± 3.243.01 ± 2.632.03 ± 2.13 Interview (medical test +) (10)0.86 ± 0.701.23 ± 0.931.05 ± 0.841.17 ± 0.77 Physician diagnosis (0)–––0.10 ± 0.05 Medical records (1)2.572.492.53 ± 0.061.25 ± 1.00aThe prevalence by sex was estimated from studies that reported both the number of participants and the number with CFSbParticipant number was applied to individual prevalence data for the studies with multiple case definitionscSpecific groups included nurses, Gulf War veterans, livestock workers, company employees, etcdCCC, Canadian Consensus Criteria; ECD, epidemiological case definition; PVES, post viral exhaustion syndrome; and NICE, National Institute for Health and Care Excellence guideline 2007; two studies with no verification of case definition and defined criteria were excludedeOther countries included India and NigeriaTable 3Meta-analysis results for the prevalence of CFS/MEGroupN. of dataaRandom effects model (%)Heterogeneity (P < 0.01, P = 0%)Prevalence95% CIQT2I2 (%)Total560.68[0.48; 0.97]8602.901.719999.4Sex Male240.86[0.58; 1.27]279.890.666691.8 Female241.36[0.91; 2.02]822.640.800397.2 Total481.04[0.76; 1.41]1481.760.947196.8Study participants Community330.76[0.53; 1.10]3286.471.036399.0 Primary care230.63[0.37; 1.10]2732.101.774599.2Population group General population (≥ 18 years) Male230.89[0.60; 1.32]276.230.660792.0 Female231.36[0.91; 2.04]818.300.800997.3 Total440.65[0.43; 0.99]7717.651.851899.4 Children/adolescents (< 18 years)70.55[0.22; 1.35]538.131.431998.9 Specific population51.31[0.61; 2.78]80.540.665795.0Case definition CDC-1994 340.89[0.60; 1.33]3871.641.369199.1 Holmes 80.17[0.06; 0.49]101.721.889093.1 Australian 40.79[0.05; 12.55]1002.437.886099.7 Oxford 41.41[0.68; 2.93]35.170.446891.5 CCC ECD PVES NICE b40.17[0.04; 0.83]1200.672.586499.8Diagnostic method Interview (medical test −)221.14[0.76; 1.72]1675.910.826998.7 Interview(medical test +)230.95[0.69; 1.31]365.720.520894.0 Physician diagnosis60.09[0.05; 0.13]200.490.295297.5 Medical records50.52[0.16; 1.71]1197.281.836099.7Refer to supplementary Additional file: Figs. 1, 2, 3, 4, 5, 6, 7aTotal number of prevalence data pointsbCCC, Canadian Consensus Criteria; ECD, epidemiological case definition; PVES, post viral exhaustion syndrome; and NICE, National Institute for Health and Care Excellence guideline 2007; two studies with no verification of case definition and defined criteria were excludedFig. 2Meta-analysis of the total CFS/ME prevalance. CFS chronic fatigue syndrome. ME myalgic encephalitisFig. 3Meta-analysis of the CFS/ME prevalence in adults (≥ 18 years). CFS chronic fatigue syndrome, ME myalgic encephalitis

Regarding gender-related differences, 24 data (from 21 studies that included information about gender) indicated an approximately 2.0-fold preponderance of females of 2.24 ± 2.59% vs. 1.11 ± 1.05% for the total population and 2.83 vs. 1.39% for the general population; the prevalence was 1.5-fold higher according to the meta-analysis, 1.36 vs. 0.89% (Table 2 and 3, Fig. 4).Fig. 4Meta-analysis of the CFS/ME prevalence in males a, females b. CFS chronic fatigue syndrome, ME myalgic encephalitis

Prevalence by study participants: When we analyzed the CFS/ME prevalence according to study participants (community-based sample vs. primary care sample), the 30 community studies (33 data) showed a prevalence of 1.56 ± 1.80% (95% CI: 1.08–2.04), and the 16 primary care studies (23 data) showed a prevalence of 1.16 ± 1.13% (95% CI: 0.86–1.46). A higher prevalence in community studies was also observed in the pooled prevalence results of 0.73% (4014 CFS/ME patients out of 548,461 participants) vs. 0.21% (1739 out of 839,326) and in the meta-analysis findings of 0.76% (95% CI: 0.53–1.10) vs. 0.63% (95% CI: 0.37–1.10), respectively (Table 2, 3, and Additional file 2). From the results of 21 studies (24 data) with known gender-related data, the prevalence of female preponderates in both community (2.31% vs. 1.03%) and primary care settings (1.96 vs. 1.39%) (Table 2).

Prevalence by study population group: The study population was divided into 3 groups: general population (n = 956,526, ≥ 18 years), children and adolescents (n = 117,307, < 18 years), and specific populations (n = 12,143, e.g., nurses, Gulf War veterans) (Table 1). The prevalence in the general population (1.45%) and in specific populations (1.62%) were higher than that in children and adolescents (0.89%) (Table 2). The meta-analysis indicated prevalence of 0.65% (95% CI: 0.43–0.99) in the general population, 1.31% (95% CI: 0.61–2.78) in specific populations and 0.55% (95% CI: 0.22–1.35) in children and adolescents (Table 3, Fig. 3, and Additional file 3). In the general population, a female predominance was shown in both the averaged (2.83 ± 2.61 vs. 1.39 ± 1.05%), and the meta-analysis (1.36%, 95% CI: 0.91–2.04 vs. 0.89, 95% CI: 0.60–1.32) (Table 2, 3 and Fig. 4).

Prevalence by case definitions: Eight case definitions were adopted for 44 studies (54 data); 2 studies used unknown case definitions. The total prevalence was notably different according to case definition:, the prevalence was the highest with Australian (2.52 ± 2.99%), then in descending order, Oxford (1.73 ± 1.35%), CDC-1994 (1.46 ± 1.34%) and Holmes definitions (0.34 ± 0.40%), but in the meta-analysis, the orders changed to Oxford (1.41%, 95% CI: 0.68–2.93), CDC-1994 (0.89%, 95% CI: 0.60–1.33), Australian (0.79%, 95% CI: 0.05–12.55) and Holmes (0.17%, 95% CI: 0.06–0.49) (Table 2, 3 and Additional file 4, 5).

Prevalence by diagnostic method: In this study, the diagnostic methods could be classified into four groups, and the prevalence data differed significantly among them (P < 0.05), as follows: interview without a medical test (survey and/or questionnaire, 19 studies, averaged prevalence 2.03 ± 2.13%), interview with a medical test (18 studies, 1.17 ± 0.77%), review of medical records (5 studies, 1.25 ± 1.00%) and physician diagnosis (4 studies, 0.10 ± 0.05%) (Table 1, 2).

The meta-analysis also showed different prevalence, as follows: interviews without medical tests (1.14%, 95% CI: 0.76–1.72), interviews with medical tests (0.95% 95% CI: 0.69–1.31), medical record review (0.52% 95% CI: 0.16–1.71), and physician diagnoses (0.09%, 95% CI: 0.05–0.13) (Table 3, and Additional file 6, 7).

Prevalence by country and publication year: The majority of the studies were conducted in Western countries (34 studies/44 prevalence data from 8 countries, 989,626 participants), followed by Asian countries (10 studies/10 data, 3 countries, 93,389), and others (2 studies/2 data, 2 countries, 2961). The total prevalence reported for Western and Asian populations were comparable (1.32 ± 1.45% vs. 1.51 ± 1.74%) (Tables 1, 2 and Additional file 1). The majority (38 of 46) of the studies were published between 1990 and 2010. Of those studies, 10 (of 18) studies were for primary care population conducted in 1990s, whereas 16 (of 20) for community in 2000s. More studies on community-based were conducted in 2000s than 1990s (Table 1). The total prevalence in 2000s (2.08 ± 2.01) was approximately twofold higher than 1990s (0.96 ± 0.91) (Table 2).

Background: Myalgic Encephalomyelitis/Chronic Fatigue Syndrome (ME/CFS) is a complex condition with no reliable diagnostic biomarkers. Studies have shown evidence of autonomic dysfunction in patients with ME/CFS, but results have been equivocal. Heart rate (HR) parameters can reflect changes in autonomic function in healthy individuals; however, this has not been thoroughly evaluated in ME/CFS.

Methods: A systematic database search for case-control literature was performed. Meta-analysis was performed to determine differences in HR parameters between ME/CFS patients and controls.

Results: Sixty-four articles were included in the systematic review. HR parameters assessed in ME/CFS patients and controls were grouped into ten categories: resting HR (RHR), maximal HR (HRmax), HR during submaximal exercise, HR response to head-up tilt testing (HRtilt), resting HR variability (HRVrest), HR variability during head-up tilt testing (HRVtilt), orthostatic HR response (HROR), HR during mental task(s) (HRmentaltask), daily average HR (HRdailyaverage), and HR recovery (HRR) Meta-analysis revealed RHR (MD ± 95% CI = 4.14 ± 1.38, P <.001), HRtilt (SMD ± 95% CI = 0.92 ± 0.24, P <.001), HROR (0.50 ± 0.27, P <.001), and the ratio of low frequency power to high frequency power of HRVrest (0.39 ± 0.22, P <.001) were higher in ME/CFS patients compared to controls, while HRmax (MD ± 95% CI = –13.81 ± 4.15, P <.001), HR at anaerobic threshold (SMD ± 95% CI = –0.44 ± 0.30, P = 0.005) and the high frequency portion of HRVrest (–0.34 ± 0.22, P =.002) were lower in ME/CFS patients.

Conclusions: The differences in HR parameters identified by the meta-analysis indicate that ME/CFS patients have altered autonomic cardiac regulation when compared to healthy controls. These alterations in HR parameters may be symptomatic of the condition.

The literature search initially identified 412 unique records. The 317 studies did not meet the inclusion criteria and were excluded. Of the remaining 95 studies, data were inadequately reported in 31. None of the authors provided missing data upon request, resulting in a total of 64 studies for inclusion. A summary of the search outcomes, including the number of studies excluded at various phases is shown in Figure 1.

Critical appraisal of included studies: Inter-rater agreement for critical appraisal was high (κ = 0.75). Critical appraisal revealed that the majority of included studies were of moderate or good quality. Seventeen articles were found to be of ‘moderate’ quality (score between 50% and 59%), while 30 studies were found to be of ‘good’ quality (score between 60% and 69%). Four articles were found to be of ‘low’ quality (score below 50%), with each article scoring 45%. Thirteen articles were found to be ‘high’ quality (score over 70%) The most common questions which resulted in a response of ‘No’ during critical appraisal were “Were confounding factors identified?” (performed in 20% of studies), and “Were strategies to deal with confounding factors stated?” (23%). Few studies (30%) reported on the specific length of exposure to the condition so it was difficult to assess the question “Was the exposure period of interest long enough to be meaningful?”. However, as this review only included studies which diagnosed ME/CFS through accepted criteria, and all of these criteria require the presence of the condition for at least 6 months, it is likely that all included studies reported on a meaningful exposure. That said, studies which did not explicitly report on the length of exposure to the condition were graded as “not available” for that question. Generally, patients and controls within the included studies were comparable, and both the exposure to the condition and all relevant HR parameters were assessed in a standard, valid and reliable manner. Appropriate statistical analysis were used in the vast majority of studies, however statistical analyses were unclear in 7 studies.

Participants: In total, the 64 included studies reported on 2286 ME/CFS patients and 1758 healthy controls who had HR parameters assessed. Of these studies, 14 recruited exclusively female participants, and 50 studies recruited a mixed gender sample. Overall, 79% of ME/CFS patients were female (n = 1803), compared to 73% for controls (n = 1283), while 1 study did not specify the gender of participants.

Diagnostic criteria: The 1994 CDC criteria were the most commonly used diagnostic criteria (47 studies), followed by the 1988 CDC criteria (8 studies), the Oxford criteria (5 studies), and the ICC (5 studies). Four studies employed the CCC, in each case in addition to another form of criteria (CDC alone, CDC+ICC).

Study outcomes: Overall, HR parameters extracted from the included studies were sorted into 10 categories, and included studies for each parameter are presented in Table 1 – resting HR (RHR), maximal HR (HRmax) HR during submaximal exercise, heart rate during HUTT (HRtilt), Orthostatic HR response (HROR), resting HRV (HRVrest), HRV during HUTT (HRVtilt), HR during mental task (HRmentaltask), daily average HR (HRdailyaverage), and HRR

Resting heart rate: The most commonly assessed HR parameter was (RHR), which was assessed in 43 studies (patients n = 1766, controls n = 1291). Studies varied regarding the body position used to assess RHR, with 16 studies (patients n = 1059, controls n = 548) using a seated/reclined position (RHRseated), and 26 (patients n = 707, controls n = 743) using a supine position (RHRsupine). Meta-analysis revealed ME/CFS patients had a RHR that was ∼4 beats/minute faster compared to controls (MD ± 95% CI = 4.14 ± 1.30, P <.001; Fig. 2), but with significant heterogeneity between studies (P <.001, I2 = 62%). Subgroup analysis revealed a similar difference between ME/CFS patients and controls for RHRsupine (4.02 ± 2.11, P <.001), and RHRseated (4.53 ± 2.40, P <.001), with significant heterogeneity amongst studies evaluating RHRsupine (P <.001, I2 = 73%).

Maximal heart rate: HRmax was reported in 20 studies (patients n = 989, controls n = 568), which used 2 different exercise modalities (cycling, 13 studies; and treadmill walking, 7 studies) to elicit a maximal response. Five studies (patients n = 111, controls n = 123) – four of which employed a cycling protocol – based the attainment of a maximal response on accepted criteria and assessed true HRmax, while the remaining 15 studies (patients n = 878, controls n = 445) only reported the highest HR achieved during testing, and is likely to represent symptom limited peak HR (HRpeak), rather than a true HRmax.

Meta-analysis showed HRmax values were found to be lower for ME/CFS patients compared to controls (–13.81 ± 4.14, P <.001, Fig. 3), but were also affected by significant heterogeneity (P <.001, I2 = 78%). Subgroup analysis showed the difference between ME/CFS patients and controls was considerable in studies which measured symptom limited HRpeak (–16.62 ± 4.68, P <.001), with significant heterogeneity between studies (P <.001, I2 = 74%). Comparatively, in studies which assessed true HRmax the difference between ME/CFS patients and controls was smaller, albeit still significant (–5.81 ± 3.34, P <.001), with no heterogeneity present (P =.88, I2 = 0%).

Heart rate during submaximal exercise: HR during submaximal exercise was reported by 14 studies (patients n = 837, controls n = 496), with 10 studies using cycling exercise, and 4 studies using treadmill walking exercise (Table 1 ). Six studies (patients n = 539, controls n = 313) reported HR at a particular submaximal exercise ‘threshold’ (anaerobic threshold [AT], ventilatory threshold [VT], or lactate threshold [LT]), which were representative of the transition from predominately aerobic to predominately anaerobic energy systems (HRthreshold). In addition, nine studies reported a steady state HR response (HRsteadystate) at a given submaximal workload, with Blackwood et al reporting HR's at both a light absolute workload, and at a workload designed to elicit 85% of maximal HR. Seven (patients n = 244, controls n = 129) studies reported HRsteadystate during workloads that were likely below submaximal exercise threshold (AT, VT or LT), while 3 studies (patients n = 54, controls n = 54) reported a HRsteadystate that was likely to be at or above submaximal exercise threshold (AT, VT, or LT).

Given the variations in the ways exercise HR was recorded and reported, with different thresholds being assessed in addition to varying exercise intensities at, above or below those thresholds, meta-analysis was performed separately for each HR parameter assessed during submaximal exercise. Analysis of HRthreshold revealed a moderate difference, with HRs being lower in ME/CFS patients compared to healthy controls (SMD ± 95% CI = –0.44 ± 0.31, P =.005), albeit with significant heterogeneity (P =.04, I2 = 56%; Fig. 4). No difference was found between ME/CFS patients and controls for HRsteadystate for workloads that were either likely below (0.09 ± 0.30, P =.59) or likely above submaximal exercise threshold (AT, VT, or LT) (0.30 ± 0.69, P =.40).

Heart rate during head-up tilt testing: Nine studies (patients n = 285, controls n = 257) reported HR in response to HUTT (HRtilt) (Table 1 ). Eight studies (patients n = 208, controls n = 188) reported an absolute HR value during HUTT (peak HR, end HR, etc) and 3 studies (patients n = 77, controls n = 69) reported the ΔHR in response to HUTT. Meta-analysis revealed HRtilt was higher in ME/CFS patients than it was in controls (SMD ± 95% CI = 0.92 ± 0.24, P <.001; Fig. 5). Subgroup analysis revealed that HRtilt was increased in ME/CFS patients compared to controls, regardless of whether data were reported as the peak HR during a short (1.23 ± 0.40, P <.001) or long (0.86 ± 0.36, P <.001) duration HUTT, or if data were reported as the ΔHR during HUTT (0.74 ± 0.44, P =.001).

Orthostatic heart rate response: The orthostatic HR response (HROR) was reported in 6 studies (patients n = 197, controls n = 131) (Table 1 ). Five studies reported HROR in response to standing up after a period of supine rest, while 1 study reported the effect of standing up following sitting. Meta-analysis showed HROR was higher for ME/CFS patients compared to controls (SMD ± 95% CI = 0.50 ± 0.27, P <.001; Fig. 6). Subgroup analysis revealed there was a moderate difference when HROR was reported as the ΔHR upon moving from lying/sitting to standing, with ME/CFS patients having a higher ΔHR than controls (0.86 ± 0.46, P <.001), with a similar finding, albeit of a smaller magnitude, also found for the peak HR obtained upon standing (0.34 ± 0.27, P =.01).

Resting heart rate variability: Measures of resting HRV (HRVrest) were reported in 13 studies (patients n = 294, controls n = 270) (Table 1 ). Seven studies recorded HRVrest in a supine position, while 3 recorded HRVrest while seated/reclined. Two studies recorded HRVrest during overnight sleep. Frequency domain HRVrest measures were used in 12 studies, with 1 study reporting only time domain analyses, and 2 studies reporting both frequency and time domain analyses.

Parameters of HRVrest from included studies were meta-analysed separately. ME/CFS patients had a higher ratio between low frequency and high frequency power (LF/HF) (SMD ± 95% CI = 0.20 ± 0.25, P =.11) than controls, and a lower high frequency power (HFP) (–0.34 ± 0.22, P =.002) and root mean of the sum of squares of beat to beat deviations (RMSSD) (–0.37 ± 0.32, P =.02) (Fig. 7). No differences were found for low frequency power (LFP) (0.39 ± 0.22, P <.001), Vagal Power (–1.86 ± 2.08, P =.08), or sleeping LF/HF ratio (–0.52 ± 2.48, P =.68), with significant heterogeneity for Vagal Power (P =.006, I2 = 87%), and sleeping LF/HF (P <.001, I2 = 96%).

Heart rate variability during head-up tilt testing: Two studies (patients n = 40, controls n = 24) reported HRV during HUTT (HRVtilt) (Table 1 ). Both studies reported exclusively frequency domain parameters, with each reporting on LFP, HFP, and LF/HF ratio. Meta-analysis found no difference between ME/CFS patients and controls for LFP (SMD ± 95% CI = 0.44 ± 0.87, P =.32; Fig. 8), HF power (0.29 ± 0.51, P =.27), or LF/HF ratio (0.35 ± 0.51, P =.18).

Heart rate during mental task(s): Five studies (patients n = 107, controls n = 116) reported the HR response to a mental task (HRmentaltask) (Table 1 ). Subgroup analysis revealed no significant effect for HRmentaltask when reported as the average HR during the tasks in 4 studies 6451, (SMD ± 95% CI = 0.25 ± 0.32, P =.57) (Fig. 9). One study reported HRmentaltask as the ΔHR during task, again with no significant difference between ME/CFS and controls (-0.29 ± 0.45, P =.21).

Daily average heart rate: Daily average HR (HRdailyaverage) was reported in 2 studies (patients n = 80, controls n = 162) (Table 1 ). Both studies used the same measurement techniques and reported data in the same way – HR data were measured through ECG monitoring, and both studies reported the average HR over 24 hours. Meta-analysis revealed no difference between the ME/CFS patients and controls (SMD ± 95% CI = 0.11 ± 0.27, P =.45; Fig. 10), with no heterogeneity between the studies (P =.56, I2 = 0%).

Post-exertional heart rate recovery: HRR was reported in only 2 articles (patients n = 108, controls n = 72), both of which used similar methods to elicit and record HRR (Table 1 ). Meta-analysis found no difference in HRR between ME/CFS patients and controls (MD ± 95% CI = –11.78 ± 13.72, P =.09; Fig. 11), with results affected by significant statistical heterogeneity (P =.003, I2 = 89%).

Background: As the etiology of chronic fatigue syndrome (CFS) is unclear and the treatment is still a big issue. There exists a wide range of literature about acupuncture and moxibustion (AM) for CFS in traditional Chinese medicine (TCM). But there are certain doubts as well in the effectiveness of its treatment due to the lack of a comprehensive and evidence-based medical proof to dispel the misgivings. Current study evaluated systematically the effectiveness of acupuncture and moxibustion treatments on CFS, and clarified the difference among them and Chinese herbal medicine, western medicine and sham-acupuncture.

Methods: We comprehensively reviewed literature including PubMed, EMBASE, Cochrane library, CBM (Chinese Biomedical Literature Database) and CNKI (China National Knowledge Infrastructure) up to May 2016, for RCT clinical research on CFS treated by acupuncture and moxibustion. Traditional direct meta-analysis was adopted to analyze the difference between AM and other treatments. Analysis was performed based on the treatment in experiment and control groups. Network meta-analysis was adopted to make comprehensive comparisons between any two kinds of treatments. The primary outcome was total effective rate, while relative risks (RR) and 95% confidence intervals (CI) were used as the final pooled statistics.

Results: A total of 31 randomized controlled trials (RCTs) were enrolled in analyses. In traditional direct meta-analysis, we found that in comparison to Chinese herbal medicine, CbAM (combined acupuncture and moxibustion, which meant two or more types of acupuncture and moxibustion were adopted) had a higher total effective rate (RR (95% CI), 1.17 (1.09 ~ 1.25)). Compared with Chinese herbal medicine, western medicine and sham-acupuncture, SAM (single acupuncture or single moxibustion) had a higher total effective rate, with RR (95% CI) of 1.22 (1.14 ~ 1.30), 1.51 (1.31–1.74), 5.90 (3.64–9.56). In addition, compared with SAM, CbAM had a higher total effective rate (RR (95% CI), 1.23 (1.12 ~ 1.36)). In network meta-analyses, similar results were recorded. Subsequently, we ranked all treatments from high to low effective rate and the order was CbAM, SAM, Chinese herbal medicine, western medicine and sham-acupuncture.

Conclusions: In the treatment of CFS, CbAM and SAM may have better effect than other treatments. However, the included trials have relatively poor quality, hence high quality studies are needed to confirm our finding.

Electronic supplementary material: The online version of this article (doi:10.1186/s12906-017-1647-x) contains supplementary material, which is available to authorized users.

Searching result: A total of 1345 potentially relevant citations were identified, 47 of which were degree research thesis. 157 duplicate papers were removed firstly, and 1000 papers were excluded after scanning their titles and abstracts. After screening the full texts of the included articles, 158 studies were excluded for the following reasons: no relevant data (n = 35), inapposite treatments set (n = 73), patients with tuberculosis (n = 1), duplicate reports (n = 12), not RCT (n = 3), unclear diagnose criteria (n = 7), inapposite criteria of invalid evaluation (n = 24), inappropriate western medicine (n = 3). However, one additional article was identified during screening. Finally, 31 studies (including1 three-arm study ) were included for further analysis (Fig. 1).Fig. 1Flow chart of study selection

Studies description: We included 31studies (Additional file 1: Table S1), all of these were conducted in China and published in Chinese. The sample size of these studies ranged from 30 to 200, and the total number of patients was 2255. The age of patients ranged from 18 to 78, and the duration of treatment was from 10 to 90 days. All studies adopted the diagnosis criteria of American CDC (1988 or 1994 edition). As acupuncture and moxibustion had many forms, such as auricular acupuncture, electroacupuncture, which were acupuncture, we treated them all as acupuncturein the present study. Therefore, the control treatments were Chinese herbal medicine, western medicine, sham-acupuncture and SAM (single acupuncture or single moxibustion).

All the included studies were divided into five groups based on the intervention in experiment and control groups: (1) CbAM (combined acupuncture and moxibustion, which meant two or more types of acupuncture and moxibustion were adopted) versus Chinese herbal medicine (n = 8); (2) SAM versus Chinese herbal medicine (n = 12); (3) SAM versus western medicine (n = 4); (4) SAM versus sham-acupuncture (or placebo treatment) (n = 4); (5) CbAM versus SAM (n = 5).

Quality of the included studies: After assessing the quality of studies based on the basis of Cochrane risk of bias, we revealed that 17 trials reported random sequence generation, 2 studies provided information on allocation concealment, 4 trials described blinding of participants (single or double),1 trial described blinding of outcome assessment,3 trials had unclear bias of complete data and 5 trials had unclear other biases (defined as baseline data comparability). Selective reporting bias was unclear (Fig. 2).Fig. 2The Cochrane risk of bias

Synthesis of results / CbAM versus Chinese herbal medicine: Eight trials, 515 participants (259 in experiment group and 256 in control) were included in this group. The pooled results of the 8 trials that compared CbAM with Chinese herbal medicine were of RR (95% CI), 1.17 (1.09-1.25), indicating that CbAM was more effective than Chinese herbal medicine (Fig. 3). There were no heterogeneity (I 2 = 0%, p = 0.83).Fig. 3Forest plot of CbAM versus Chinese herbal medicine

SAM versus Chinese herbal medicine: Twelve trials, 951 participants (478 in experiment group and 473 in control) were included in this group. With only not important heterogeneity existed (I 2 = 29.8%, p = 0.15), the pooled result of the 8 trials was performed with fixed effect model and the result (RR (95% CI), 1.22 (1.14–1.30) showed that SAM had better effects than Chinese herbal medicine (Fig. 4). There was no publication bias after egger’s test (p = 0.82).Fig. 4Forest plot of SAM versus Chinese herbal medicine

SAM versus western medicine: A total of 4 trials (286 participants, 142 in experiment group and 144 in control) were adopted in this group. The pooled results indicated that SAM had a better effect than western medicine (RR (95% CI), 1.51 (1.31–1.74)), with no heterogeneity among the studies (I 2 = 0%, p = 0.59) (Fig. 5).Fig. 5Forest plot of SAM versus western medicine

SAM versus placebo treatment (sham-acupuncture): Four trials (246 participants, 124 in experiment group and 122 in control) were conducted to compare SAM with sham-acupuncture. The pooled result indicated that SAM was much more effective than sham-acupuncture (RR (95% CI), 5.90 (3.64–9.56)), with no heterogeneity (I 2 = 0%, p = 0.45) (Fig. 6).Fig. 6Forest plot of SAM versus placebo treatment (sham-acupuncture)

CbAM versus SAM: A total of 5 trials, 356 participants (178 in experiment group and 178 in control) were involved in this group. The pooled results revealed that CbAM was more effective than SAM (RR (95% CI), 1.23 (1.12–1.36)) (Fig. 7). The statistic I 2 for homogeneity of RR showed there was no heterogeneity among the studies (I 2 = 13.1%, p = 0.33).Fig. 7Forest plot of CbAM versus SAM

Sensitivity analyses were performed in all traditional meta-analysis. No small sample effect existed after removing any single trial. Random effect model was also adopted and the results were similar to the fixed random effect, showing stable results (Figs. 3, 4, 5, 6 and 7).

Network meta-analysis: The results of network meta-analysis were similar with the traditional meta-analysis (Table 1), showed that all treatments were more effective than sham-acupuncture, and the comprehensive rank (from high to low effective rate) was CbAM, SAM, Chinese herbal medicine, western medicine and sham-acupuncture (Figs. 8 and 9)Table 1Comparison of network meta-analysis and traditional meta-analysisCbAMSAMChinese medicineWestern medicinePlacebo (Sham-acupuncture)TreatmentT-metaa N-metab T-metaa N-metab T-metaa N-metab T-metaa N-metab T-metaa N-metab CbAM––1.23 (1.12–1.36)1.10 (1.04–1.17)1.17 (1.09–1.25)1.23 (1.15–1.32)–1.68 (1.29–1.69)–5.84 (1.19–12.21)SAM–0.91 (0.85–0.96)––1.22 (1.14–1.30)1.12 (1.03–1.21)1.51 (1.31–1.74)1.53 (1.17–2.43)5.90 (3.64–9.56)5.31 (1.74–11.07)Chinese medicine–0.81 (0.77–0.87)–0.89 (0.83–0.97)–––1.20 (1.10–1.50)–4.00 (2.50–6.90)Western medicine–0.62 (0.37–0.77)–0.68 (0.41–0.85)–0.76 (0.46–0.96)–––3.20 (1.90–5.50)placebo (sham-acupuncture)–0.21 (0.08–0.53)–0.24 (0.09–0.58)–0.26 (0.10–0.65)–0.37 (0.12–0.99)–– astands for traditional meta-analysis bstands for network meta-analysis Fig. 8Network of involved treatments. CbAM, combined acupuncture and moxibustion; SAM, single acupuncture or single moxibustion; CHM, Chinese herbal medicine; WM, western medicine; PT, placebo treatment (sham-acupuncture) Fig. 9Comprehensive comparison of all treatments. Left, probabilities of all treatments in each rank; right, absolute risk (effective rate) of all treatments

Adverse events: Of the conducted trials, adverse events happened in four studies. In total, there was fainting during acupuncture (6 cases), feeling of acid bilges on the back (6 cases),subcutaneous hematoma (1 case), burn during moxibustion (1 case). No serious adverse events happened.

Acupuncture point: The acupoints were chosen with two ways, constant point and constant point plus points based on the basis of symptoms. After analysis of points adopted in trial, we found Walking Three Miles (ST36), Spleen Locus (BL20), and Liver Locus (BL18) were the three points most commonly used (Additional file 2: Figure S1).