Prevalence rate of primary osteoporosis in China: a meta-analysis – BMC Public Health

Study selection

The literature screening process and its outcome are shown in Fig. 1. In brief, 5,713 articles were initially identified, and 1,874 articles were removed for duplication. 3,303 unrelated articles were further excluded after title and abstract screening. After reviewing 536 full-text articles with detailed reading, finally, leaving 45 articles for final analyses.

Characteristics of included studies

Characteristics of the included 45 studies [21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65] are shown in Additional file 1. These studies were published between 2002 and 2023, while data were collected from 1997 to 2021. Among these, 25 studies were conducted in southern China [21,22,23,24,25, 29, 30, 33, 38, 41, 42, 44, 45, 47,48,49, 51,52,53, 55, 56, 59,60,61, 64], 19 studies were in northern China [26,27,28, 31, 32, 34,35,36,37, 39, 40, 46, 50, 54, 57, 58, 62, 63, 65], and only one study was conducted nationwide [43]. Moreover, a total of 23 studies reported the prevalence of the entire population (both men and women) over the age of 20 years [21,22,23, 26,27,28,29, 33, 35,36,37,38,39, 44, 47, 49, 52, 54, 58,59,60, 62, 63]. These 23 studies were analysed for overall data as well as subgroups (period of data collection, gender, area, BMD measurement positions) prevalence. Nineteen studies reported prevalence only of the middle-aged or older population (aged ≥ 40 years) and were thus used for age-specific group analyses [24, 25, 30, 31, 34, 40,41,42, 45, 46, 48, 50, 51, 55,56,57, 61, 64, 65]. Only two studies exclusively reported the prevalence rate among men [32, 65], while seven studies focused on postmenopausal women [25, 30, 40, 48, 55, 57, 61].

The sample size per study ranged from 151 to 75,321 creating a total population of 241,813 participants for this meta-analysis, including 113,613 men and 128,200 women. The point prevalence of POP varied between 1.1% and 63.1%. All studies had a cross-sectional design and all response rates for the survey were above 90%.

Risk of bias in included studies

According to the Agency for Healthcare Research and Quality (AHRQ), among the 45 studies included, nine were categorized as high-quality, while 29 were deemed moderate-quality cross-sectional studies. Additionally, seven studies scored less than five points, indicating low quality. Overall, 84.44% of the included studies demonstrated acceptable quality, falling within the moderate or high categories.

Pooled prevalence rates of osteoporosis

Overall

The meta-analysis of total prevalence estimates of studies with participants aged ≥ 20 years (n = 23, N = 106,725) revealed the prevalence rate of POP in China at 18.2% (95% CI: 14.7–21.7%, Fig. 2), with a high level of heterogeneity (99.6%). Numbers for men were n = 24 (N = 53,746), with 11.5% (95% CI: 9.1–13.9%, Fig. 3) and women were n = 23 (N = 61,879) at 23.4% (95% CI: 18.3–28.5%, Fig. 4), with significant differences between them.

Forest plot of prevalence of primary osteoporosis for total population

Forest plot of prevalence of primary osteoporosis for men

Forest plot of prevalence of primary osteoporosis for women

Period of data collection

Period of Data Collection was divided into two groups: before 2010 with the 13 studies and 2011–2020 with eight studies. Meta-analysis results indicated a general upward trend (see Additional file 2). The prevalence rates from studies with data before 2010, and from 2011 to 2020 were 16.9% (95%CI: 11.9–21.8%) and 20.3% (15.1–25.4%), respectively. These data indicated that POP prevalence has increased over the past decade.

Area

The prevalence rate of POP in southern China was 16.4% (95% CI: 12.9–19.9%), which was lower than that in northern China 20.2% (13.4–21.0%) (see Additional file 2).

BMD measurement positions

Most studies measured BMD of lumbar spine, femoral neck, Ward’s triangle and greater trochanter, while only a few studies measured other sites (forearm distal 1/3 of ulna and radius). However, a subgroup analyses of included lumbar spine measurement and results from other anatomical sites revealed an almost similar prevalence rate (18.3% vs. 18.2%) (see Additional file 2).

Sex- and age-specific groups

Among age-specific groups, the prevalence rates of POP generally increased with age, except for the first two age groups (20–29 and 30–39 years old). Specifically, the prevalence was lowest in the 30 to 39 years age group (1.2%), while it peaked in those aged 80 years and older (53.9%). This increasing trend was also evident in other age brackets: 1.4% for 20–29 years, 4.9% for 40–49 years, 16.8% for 50–59 years, 35.2% for 60–69 years, and 44.1% for 70–79 years.

In delineating sex- and age-specific cohorts, the prevalence rates of POP exhibit a discernible consistency across genders and age brackets. Among men, prevalence rates stand at 1.6%, 1.1%, 4.2%, 9.9%, 19.0%, 27.6%, and 38.3%, respectively. Correspondingly, for women, rates are observed at 1.5%, 1.4%, 5.2%, 22.6%, 44.1%, 55.2%, and 63.1%, respectively. Notably, a prevailing trend emerges with higher prevalence rates in women compared to men across most age strata, with the exception noted in the 20–29 age group (see Additional file 2).

Heterogeneity analysis: meta-regression analysis

When assessed collectively, a notable degree of heterogeneity was evident among the studies, with I² statistics ranging from 63.7 to 99.9%. An exception was observed in the age-specific group 20–29 and men aged 20–29, where the I² values were 8.3% and 0%, respectively (see Additional file 2). Meta-regression analysis (Table 1) revealed that the overall prevalence estimates remained largely unaffected by the period of data collection, geographical area, or BMD measurement positions. However, sample size was found to exert a significant impact on the point prevalence estimate (p = 0.037).

The results of Meta-regression still not fully explain the high level of heterogeneity observed. Alternatively, high heterogeneity may be caused by multiple factors (e.g., sample size, measuring instrument, etc.) with synergistic effects.

Sensitivity analysis

A sensitivity analysis was conducted by systematically excluding each study and re-evaluating the merged statistics. No significant directional changes were noted, indicating the stability of the results (Fig. 5).

Publication bias

No evidence of asymmetry was observed in the funnel plot constructed using data from the 23 studies with participants over the age of 20 (Fig. 6). However, the Begg test (p = 0.086) and Egger test (p = 0.058) suggested a low likelihood of publication bias, although exceptions were noted for certain subgroups (see Additional file 2). The region parameter (urban or rural) was not included as a subgroup in this meta-analysis due to the lack of reporting on participants’ regions of residence in most studies. Additionally, as many studies reported a prevalence of zero for men and women aged 20 to 29 years, it was not possible to calculate publication bias for this age group.

Funnel plot for publication bias