Global prevalence and incidence of hallux valgus: a systematic review and meta-analysis

The literature search was performed by the first author of this paper by using six electronic databases (PubMed, Embase, Cochrane Library, CNKI, China Online Journals and CQVIP) from their inception to November 16, 2022. The searches of this systematic review and meta-analysis were conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines [7, 8]. The search terms included “hallux valgus or bunion and prevalence or incidence or epidemiology” (the search strategy was provided in eAppendix 1 in the Supplementary File). All languages were included.

The inclusion criteria were formulated as follows: (1) original research articles worldwide; (2) all studies with a cross-sectional, prospective cohort study, case-control, or cohort design; and (3) only studies that reported or sufficient data to calculate the prevalence of HV were included. There were no restrictions regarding languages. If there was more than one study among the same population, only the largest and latest one was brought into the review. Studies were excluded if they had (1) no relation to HV, foot deformities, or prevalence; (2) descriptions of operative or non-operative interventions; (3) studies related to specific disease groups (Rheumatoid Arthritis, diabetes, thyroid, and neuromuscular disorders), discussions of traumatic injury to the first toe joint; (4) they were review articles, conference abstracts, short communications, case reports, clinical opinions, letters, posters, or laboratory researches; (5) they did not report sufficient data, and efforts to contact the authors were unsuccessful; or (6) no access to obtain full text.

Titles and abstracts were screened by two review authors (Y.T.C and Y.K.S). If the study was of relevance, it was selected for full-text review. Reviewers collected data from the full text independently and cross-checked the data. This was to ensure that the correct article was identified for full-text screening. Queries were discussed and resolved by a third review author (H.W.) if there was a disagreement. The following collected data included authors, the year of publication, countries or area in which countries were studied, continents, sample size, gender and age range, year of investigation, diagnosis, the number of participants affected by HV, and prevalence of HV. The authors were contacted if more information was required or unclear .

The risk of bias of the included studies was independently assessed by using the Joanna Briggs Institute Critical Appraisal Instrument for Studies Reporting Prevalence Data (JBI) [8] by two of the authors (Y.T.C and Y.K.S), which was proven to be a reliable and valid tool for assessing observational studies. Risk of bias was categorized as high when the study reaches up to 49% score “yes”, moderate when the study reached 50% to 69% score “yes”, and low when the study reached more than 70% score “yes” (eTable 1 in the Supplement).

The metaprop module in the R statistical software package, version 4.2.2, was used for the summary statistic. The logit method was chosen to transform each prevalence proportion of study or subgroup with HV, then adopted an inverse-variance-weighted random-effects meta-analysis. HV prevalence (with 95% CIs) for the overall population and subgroups were calculated. The results were visualized using forest plots. The I² statistic for heterogeneity and the P value for heterogeneity (Cochrane Q statistic) were showed in forest plots. A funnel plot was used to assess publication bias by using the Egger test. Since the study populations were diverse, prevalence estimates were only pooled between studies with similar continents, gender and age characteristics. The prevalence of HV among the overall population and among subgroups were reported.

The search of the databases resulted in 906 records from the six databases. Search history was shown in the PRISMA flow diagram (Fig. 1). A total of 44 papers [9‐52] were included (n=186,262,669). These epidemiological studies came from five continents and 17 different national regions (Table 1). There was a wide variation in study characteristics in terms of the population studied and the methodology employed. Three studies were conducted in Africa, 15 studies in Asia, 18 studies in Europe, five studies in North America and four studies in Oceania. Among these 45 studies, nine reported the prevalence of HV in China, one in France, one in Germany, three in Nigeria, one in Indonesia, three in Japan, one in Korea, one in Netherlands, one in New Guinea, one in Greece, one in Poland, four in Spain, one in Turkey, nine in UK, five in USA, and three reported the prevalence of HV in Australia.

The pooled prevalence estimates for HV were shown in Fig. 2. The studies reported prevalence and incidence of HV ranged from 1% to 82%. The random-effects overall pooled estimated prevalence was 19% (95% CI, 13% to 25%). The heterogeneity was very high (I²=100%; P=0). It was found that no study disproportionately affected the overall result. According to the Egger test (P=0.45), there was no publication bias existed.

To study the relationship between the year of publication and the prevalence of HV, an association analysis on publication year was conducted. The results indicated that the prevalence of HV was different in publication year from Fig. 3, ranging from 81.82% in 2003 to 0.03% in 2009. According to the figure, there was no consistent trend apparent and no statistically significant relationship between HV prevalence and year of publication.

In the comprehensive review of the studies, a global distribution was observed. A significant proportion of the epidemiological investigations originated from Europe, encompassing seven countries (41.18%) and 18 research articles (38.93%), with a collective sample size of 114,526. In contrast, Africa was represented by a mere three studies (6.67%), involving 5,872 participants. Meanwhile, Asia contributed 15 investigations (33.33%) with a total of 30,911 subjects, and North America was the source of five studies (11.11%) with an extensive sample population of 186,109,693. Lastly, Oceania was represented by four studies (8.89%), encompassing 1,667 participants. The largest sample of epidemiological survey was in the United States from 1978 to 1979. The prevalence of HV was 3% (95% CI, 0% to 15%) in Africa (with very high heterogeneity [I²=99%; P<0.01]), 21.96% (95% CI, 10.95% to 35.46%) in Asia (with very high heterogeneity [I² =100%; P=0]), 18.35% (95% CI, 11.65% to 26.16%) in Europe (with very high heterogeneity [I²=100%; P=0]), 16.1% (95% CI, 5.9% to 30.05%) in North America (with very high heterogeneity [I²=100%; P=0]) and 29.26% (95% CI, 4.8% to 63.26%) in Oceania (with very high heterogeneity [I²=99%; P<0.01]) (eFigures 1-5 in the Supplement). The global prevalence and incidence of HV was illustrated in Fig. 4 in various areas of the world from the geographical distribution map.

Meta-analysis by gender in the subgroups consistently showed gender differences in the prevalence of HV. Only 24 studies reported the number of men and women with and without HV. Only two studies were all females. There was a total of 28,330 females (55.62%) and 22,608 males (44.38%). The pooled prevalence of HV by gender was 23.74% (95% CI, 16.21% to 32.21%) for females (with very high heterogeneity [I²=100%; P=0]) and 11.43% (95% CI, 6.18% to 18%) for males (with very high heterogeneity [I²=99%; P=0]) (eFigures 6, 7 in the Supplement).

The results of studies were grouped by age of study population. 10 studies reported individuals younger than 20 years old, seven studies included individuals 20 to 60 years old, and eight studies among aged over 60 years old. The prevalence was 11% (95% CI, 2% to 26%) in individuals younger than 20 years old (with very high heterogeneity [I²=99%; P<0.01]), 12.22% (95% CI, 5.86% to 20.46%) in adults aged 20 to 60 years old (with very high heterogeneity [I²=97%; P<0.01]) and 22.7% (95% CI, 13.1% to 33.98%) in elderly people aged over 60 years old (with very high heterogeneity [I²=100%; P=0]) (eFigures 8-10 in the Supplement).

Overall, the quality of the studies included in this systematic review was moderate to high. The majority of studies reached more than 70% score “yes”, indicating that they were of high methodological quality. However, some studies had limitations, such as small sample sizes or inadequate reporting of results, which may have affected the validity of their findings. To ensure the robustness of this meta-analysis, sensitivity analyses were conducted to exclude studies that scored low on the JBI quality assessment. A funnel plot was used to assess publication bias using the Egger test. According to the Egger test (P=0.45), it showed that there was no statistically significant publication bias (P>0.05) (eFigure 11 in the Supplement). These sensitivity analyses did not materially affect the overall results, indicating that the findings of this meta-analysis were robust to variations in the quality of the included studies. In conclusion, the quality of the studies included in this systematic review and meta-analysis was generally good, with most studies meeting the majority of the JBI quality assessment criteria.