Long-term trends in the incidence of urolithiasis in the United States and China: a joinpoint regression and age-period-cohort analysis based on GBD 2021

Introduction

Urolithiasis remains a common global public health issue defined as stone formation along the genitourinary tract. It affects all age groups, including children, adults and the elderly, impacting 10%–15% of the global population.1 In 2019, the number of new cases of urolithiasis exceeded 1 billion.2

The clinical symptoms of urolithiasis mainly include haematuria and pain; moreover, urolithiasis is associated with increased risks of fracture,3 cardiovascular disease4 and chronic kidney disease.5 Estimates indicate that in industrialised countries, the lifetime risk of urolithiasis exceeds 12% in males and 6% in females.6 Although surgical techniques for stone removal have improved, these procedures are often expensive. Furthermore, as a lifelong condition, once urolithiasis occurs, it often recurs, with most cases experiencing a recurrence rate of up to 50% within 5 years after the initial stone event.7 The high number of new and recurring cases, frequent surgical interventions and the introduction of new technologies impose significant personal and societal burdens. According to reports, the global economic burden of urolithiasis reached $5.3 billion in the year 2014, with indications that this amount has been increasing over time.8

Although previous studies reported epidemiological trends of urolithiasis at global scale, they often lacked targeted analyses for specific countries.9 The United States and China are at varying stages of social development, each with distinct food cultures, population structures, geographical locations and healthcare systems.

In this study, we employed the latest GBD data to describe the temporal trends in the incidence of urolithiasis in the United States and China from 1992 to 2021, evaluating the effects of age, period and cohort. By comparing the urolithiasis trends in different countries, our findings may provide a scientific basis for the relevant department to formulate effective prevention and control measures.

Method

Data sources

In our study, the data utilised were sourced from the GBD 2021 public database, available at https://ghdx.healthdata.org/.10 The GBD 2021 offers a comprehensive assessment of the global disease burden for 371 diseases and injuries across 204 countries and regions from 1990 to 2021. The dataset is systematically organised by geographical region, calendar year, sex and age group. For our analysis, annual incidence data of urolithiasis in the United States and China were extracted for the period 1992 through 2021. Specifically, the data included the annual number of incident cases and incidence rates stratified by sex (male, female and both sexes combined) across all available age groups. Additionally, overall age-standardised incidence rates (ASIR) were obtained for each sex category. Each estimate was accompanied by its respective 95% CI, as provided by the GBD 2021 dataset.

Statistical analysis

To assess the trends of urolithiasis incidence in the United States and China from 1992 to 2021, we performed Joinpoint Regression Analysis using the ASIR as input data. The model identifies specific calendar years (‘joinpoints’) at which statistically significant changes in trend occur. We specified a maximum of nine join points, as recommended by the software, with the final number determined by a permutation test involving 4499 permutations and an overall significance level of 0.05. The analysis produces two primary metrics: the annual percent change (APC), which quantifies the rate of change in ASIR within each segment defined by joinpoints, and the average annual percent change (AAPC), which summarises the overall trend across the entire study period. A p value less than 0.05 for the APC indicates a statistically significant increasing (APC>0) or decreasing (APC<0) trend during that segment. Similarly, a statistically significant AAPC (p<0.05) reflects a meaningful overall increase or decrease in ASIR from 1992 to 2021.

To examine the impact of age, period and cohort on the incidence of urolithiasis, we used the Age-Period-Cohort model (APC model), available at https://analystools.cancer.gov/apc/.11 This analysis was conducted separately for the United States and China to account for country-specific heterogeneity in trends. This tool implements the APC analysis using a generalised linear model, assuming a Poisson distribution and a log link function. By generating estimable functions, the model circumvents the issue of the collinearity among age, period and cohort effects. To implement the APC model, we coded age, period and cohort into data with continuous 5-year intervals. Age was divided into 20 groups, ranging from 0 to 4, 5–9, 10–14, up to 95+years. The period was divided into six groups, spanning from 1992 to 1996, 1997–2001, up to 2017–2021. Based on the intersection of age and period, 25 birth cohorts were identified, beginning from 1897 to 1901, 1902–1906, through to 2017–2021. The central age group, period and birth cohort within each interval were defined as the reference group in all APC model analyses. For categories with an even number of groups, the two lower central values were set as the reference group. The primary estimators used in our analysis included the longitudinal age curves, period (or cohort) relative risk (RR) and the net and local drifts. The longitudinal age curves, which were adjusted for periodical bias in the reference cohort, reflected the age-related trends of disease. The RR for each period (or cohort) denoted the risk in a particular period or birth cohort relative to the reference groups. Net drift is defined as the overall trend of annual percentage change observed during the study; local drift is defined as the annual percentage change in incidence within each age group over time. The Wald χ² test was employed to assess the statistical significance of the estimated model parameters. All p values were two-sided and significant at the α=0.05 level.

To forecast the incidence rates and number of new urolithiasis cases in the United States and China for the next decade, we utilised the Nordpred model, an APC log-linear Poisson regression model.12 The model was applied separately to the datasets of each country to generate country-specific projections. The corresponding population projection data were also sourced from the GBD study. The analysis was conducted in R (version 4.2.2) using the Nordpred package. Compared with other forecasting tools, the Nordpred method—which smooths exponential growth using a power function and adjusts trends—has demonstrated strong performance in 10-year projections and effectively avoids overestimation from exponential growth.12 For this analysis, the model parameters were specified as follows: a power5 link function was used (linkfunc = ‘power5’), and the recent period for trend extrapolation was defined as the last 4 to 6 5-year intervals (noperiods=4:6). The starting age group for both estimation and usage in the projection was set to the third group (startestage=3, startuseage=3). To ensure plausible projections, we applied a trend modification using cuttrend=c(0, .25). By setting the recent parameter to NULL, we allowed the model to automatically select the most appropriate trend from recent data based on the noperiods specification, rather than forcing the use of a fixed number of recent cohorts.

To assess the robustness of our projections, we conducted a series of sensitivity analyses to test the model’s tolerance to changes in key assumptions. These included: (1) varying standard populations: using population projections from the WHO as an alternative standard to examine the impact of different demographic forecasts; (2) different drift assumptions: forcing the recent period for trend extrapolation to be fixed to the last three birth cohorts (recent=3) to evaluate the model’s sensitivity to the definition of recent trends. (3) alternative model parameters: altering the model’s starting parameter (start.stage=1) to test its influence on the projection curve.

Result

Descriptive analysis

The incidence of urolithiasis in the United States was 1582.70 per 100 000 in 1992 (95% CI: 1297.11 to 1954.69), initially decreased and then increased gradually to 954.04 per 100 000 in 2021 (95% CI: 816.99 to 1106.92) (figure 1A–B) (online supplemental table S1), with the lowest point at 879.42 per 100 000 in 2012 (95% CI: 793.31 to 971.62). During the same period, China experienced a continuous decline in urolithiasis incidence rates, from 1732.74 per 100 000 in 1992 (95% CI: 1396.54 to 2161.67) to 964.70 per 100 000 in 2021 (95% CI: 801.26 to −1175.09).

Supplemental material

Figure 1
Figure 1

Urolithiasis incidence in the United States and China, 1992–2021. (A) Trend in age-standardised rate, United States; (B) trend in age-standardised rate, China; (C) age-specific rates by sex, United States; (D) age-specific rates by sex, China.

In terms of gender difference, from 1992 to 2021, urolithiasis incidence in males in both countries consistently exceeded that in females, although the gender gap gradually narrowed. In the United States, the gender ratio of urolithiasis incidence decreased from 1.61 in 1992 to 1.43 in 2021; in China, the gender ratio decreased from 2.66 in 1992 to 2.18 in 2021. In terms of age patterns, the highest incidence of urolithiasis was observed among the middle-aged and elderly populations. In the United States, the peak incidence in 1992 occurred in the 55–59 age group, with males at 5282.13 (95% CI: 3199.27 to 8094.39) and females at 2432.42 (95% CI: 1478.35 to 3667.51) per 100 000. By 2021, the peak incidence had shifted to the 65–69 age group, with males at 3898.91 (95% CI: 2522 to 5595.71) and females at 1910.52 per 100 000 (95% CI: 1237.22 to 2727.1). In contrast, in China, the highest incidence rates in 1992 were in the 65–69 age group, with males at 9283.74 (95% CI: 5726.28 to 13 650.25) and females at 2559.99 per 100 000 (95% CI: 1541.88 to 3806.36). By 2021, the peak shifted earlier to the 55–59 age group, where males reached an incidence of 3792.58 (95% CI: 5317.17 to 12 618.97) and females 1914.04 per 100 000 (95% CI: 1190.83 to 2817.16) (figure 1C–D) (online supplemental table S2).

Joinpoint regression analysis

From 1992 to 2021, the ASIR of urolithiasis in the United States demonstrated an AAPC of −1.72% (95% CI: −1.74% to −1.69%) (figure 2). This decline was more pronounced in males (AAPC: −2.03%, 95% CI: −2.06% to −1.98%) than in females (AAPC: −1.33%, 95% CI: −1.35% to −1.31%) (figure 2). The downward trend was steepest from 1992 to 1994 (APC: −4.32%, 95% CI: −4.73% to −3.79%) and 2001–2005 (APC: −5.07%, 95% CI: −5.24% to −2.55%). In contrast, the trend later reversed, culminating in a significant increase from 2015 to 2021 (APC: 1.36%, 95% CI: 1.21% to 1.60%).

Figure 2
Figure 2

Joinpoint regression analysis of age-standardised incidence rates of urolithiasis from 1992 to 2021 in (A) the United States and (B) China. AAPC, average annual percent change; APC, annual percent change.

Similarly, the ASIR of urolithiasis in China showed an AAPC of −2.01% from 1992 to 2021 (95% CI: −2.04% to −1.98%). The rate of decline was greater for males (AAPC: −2.19%, 95% CI: −2.22% to −2.15%) than for females (AAPC: −1.52%, 95% CI: −1.54% to −1.50%). As detailed in figure 2, the most significant period of decline was observed from 2005 to 2010, with an APC of −5.66% (95% CI: −5.83% to −5.49%).

The effects of age, period and cohort on incidence

The age-effect curves demonstrated a unimodal trend, with the incidence of urolithiasis peaking in a specific age range in both the United States and China (figure 3A) (online supplemental table S3). In the United States, the three age groups with the highest incidence rates were 55–59 years (2024.58 per 100 000; 95% CI: 1953.80 to 2097.92), 60–64 years (2007.57 per 100 000; 95% CI: 1935.48 to 2082.35) and 50–54 years (1997.47 per 100 000; 95% CI: 1928.19 to 2069.23). A similar pattern was observed in China, where the highest incidence rates were found in the 55–59 age group (3077.98 per 100 000; 95% CI: 3021.70 to 3135.32), followed by the 50–54 age group (3054.94 per 100 000; 95% CI: 3001.02 to 3109.82) and the 60–64 age group (2638.46 per 100 000; 95% CI: 2586.10 to 2691.88). These data quantitatively establish that the burden of urolithiasis is most heavily concentrated in the 50–64 age demographic.

Figure 3
Figure 3

Age-Period-Cohort analysis of urolithiasis incidence in the United States and China. (A) Longitudinal age curves showing the age effect. (B, C) Period and cohort effects presented as rate ratio. (D) Net drift and local drifts in incidence rates. Blue and red lines represent data for the United States and China, respectively.

During 1992–2021, the period RR of urolithiasis incidence in China showed a decreasing trend, indicating a low risk of urolithiasis incidence. The period effect in the United States showed a trend of initially declining then rising, with a turning point around 2009. Although the period RR in some years remained below the reference period, in recent years the relative risk of urolithiasis in the United States has been on an upward trend (figure 3B) (online supplemental table S4).

Cohort analyses of urolithiasis incidence in the United States revealed elevated trends peaking around 1930 (Cohort RR: 1.50, 95% CI: 1.42 to 1.58), followed by a gradual decline, yet demonstrating an upward trajectory in more recent years. This suggested that in the United States, the birth cohort around 1930 faced the highest relative risk of urolithiasis incidence. Under the effect of the birth cohort, the trend of urolithiasis incidence in China was decreasing, suggesting that the risk of urolithiasis incidence in China decreased with the later cohorts (figure 3C) (online supplemental table S5).

Analysis of the data showed that the net drift in the incidence of urolithiasis in the United States and China was −1.40 (95% CI: −1.57 to −1.22) and −2.33 (95% CI: −2.55 to −2.11), indicating a decreasing trend in the incidence of urolithiasis in the United States and China, with a more pronounced decrease in China (figure 3D) (online supplemental table S6).

In China, the analysis indicated a declining trend in the incidence risk of urolithiasis across nearly all age groups from 1992 to 2021. With the exception of the 95+ age group, all other age groups presented a negative local drift, with their 95% CI being entirely below zero (figure 3D) (online supplemental table S7). The most significant decline was observed in the 65–69 age group, which had a local drift of −3.66 (95% CI: −3.77 to −3.55). In contrast, the United States exhibited a more complex, age-differentiated pattern. For older populations aged 80–94 years (encompassing the 80–84, 85–89 and 90–94 age groups), the local drift values and their corresponding 95% CI were all greater than zero, indicating an increasing risk trend over the period. Conversely, a decreasing risk was noted for individuals between 10 and 75 years of age, evidenced by negative local drift values. Within this range, the most substantial decrease occurred in the 30–34 age group (local drift: −2.75, 95% CI: −2.97 to −2.54). Among all age groups, the most pronounced increase was recorded in the 90–94 age group (local drift: 2.98, 95% CI: 1.21 to 4.79).

Prediction of incidence of urolithiasis

Over the next decade, the incidence of urolithiasis was projected to increase in both the United States and China. By 2031, the ASIR in the United States was estimated to be 1011.15 per 100 000, with the number of cases rising to 4 856 243(figure 4) (online supplemental table S8). In the same year, the ASIR in China was expected to reach 970.49 per 100 000, totalling 2 035 0877 cases. Furthermore, we performed several sensitivity analyses to test the robustness of our projections. As detailed in the Methods section, these involved altering key assumptions and data sources. The results remained broadly consistent with our primary findings, supporting their stability (online supplemental file 1) (online supplemental tables S9-S11).

Figure 4
Figure 4

Sensitivity analyses of projected urolithiasis incidence rates in the United States and China. (A) Primary projection using the main model parameters. (B) Sensitivity analysis using the WHO standard population. (C) Sensitivity analysis using an alternative drift assumption (fixing the recent period to the last three birth cohorts). (D) Sensitivity analysis using an alternative model starting parameter (start.stage=1).

Discussion

This study utilised joinpoint regression analysis and APC analysis to explore the temporal trends in the incidence of urolithiasis in the United States and China from 1992 to 2021. The findings of this study provide theoretical support for the prevention of urolithiasis in the future.

Asymptomatic stones are a ‘subclinical or hidden’ stage of urolithiasis, typically not requiring treatment. When stones involve the urinary tract structure, urolithiasis can affect patients’ normal activities, with some leading to life-threatening complications in end stages, including pyonephrosis, perinephric abscess and sepsis.13 A higher incidence of urolithiasis was observed among males than females. In 2021, the gender ratio in China was 2.18, and in the United States, it was 1.43. This variation may be attributed to multiple factors, notably androgen, which has been widely discussed as a risk factor for urolithiasis and involved in multiple signalling pathways promoting the formation of urinary calculi.14 We similarly observed a narrowing gender disparity in the incidence of urolithiasis among males and females in the United States and China. In the United States, the gender ratio of urolithiasis incidence decreased from 1.61 in 1992 to 1.43 in 2021; in China, the gender ratio declined from 2.66 in 1992 to 2.18 in 2021. These findings are consistent with previous studies and may be associated with changes in diet and lifestyle in the workplace among females.15–17

Over the past three decades, the overall incidence of urolithiasis has declined in both China and the United States. This downward trend is consistently confirmed by two complementary analytic approaches: the AAPC estimated via Joinpoint regression and the net drift calculated from APC models, both of which yielded negative values. This indicates that the long-term effectiveness of prevention and control measures is clear, both at the level of observed annual data and after controlling for period and cohort effects. However, this overall downward trend does not manifest uniformly across different age groups in the two countries. In China, the incidence rate shows a decreasing trend across nearly all age groups, suggesting that public health interventions may have provided widespread protective effects. In contrast, the United States exhibits marked age disparities: while the incidence rate is declining among individuals aged 10–75, the risk is rising among the elderly population aged 80–95. This phenomenon reveals new challenges in urolithiasis prevention and control in the US, which may be associated with more complex comorbidities, polypharmacy and the more frequent and sensitive use of medical imaging in the elderly population.18–20 Moreover, age-effect curves indicate that disease burden in both countries is highly concentrated in midlife, peaking at 50–64 years. This finding aligns with the results of other scholars.21 22 The increased susceptibility to urolithiasis among middle-aged individuals is attributed to several factors. Typically, this age group completed most of the strenuous tasks in their daily lives, often resulting in inadequate fluid intake and higher rates of dehydration.23 Furthermore, their lifestyle choices—characterised by irregular diet or staying up late—and professional stress contribute to this heightened risk.24 Identifying this high-incidence age window provides a critical basis for devising targeted public-health screening and prevention strategies.

The period or cohort effect has different impacts on the incidence of urolithiasis in different countries. In the United States, after a rapid decline, the period effects showed a rising trend in recent years. The cohort effects initially increased, peaking around 1930, followed by a sustained decline, but had shown an upward trend in recent years. This trend might be linked to socioeconomic levels, lifestyle changes and dietary habits. In the 1920s, preceding the Great Depression, the United States experienced a period of rapid economic expansion following victory in World War I. This economic expansion was associated with increases in smoking and alcohol consumption, a reduction in sleep and a rise in work-related stress linked to more strenuous labour.25 These factors collectively contribute to a heightened relative risk of urolithiasis.

Metabolic factors likely played a major role in the recent period or cohort increases observed in the United States. In the United States, the prevalence of metabolic syndrome increased from 32.5% in 2011 to 36.9% in 2016, and obesity rates climbed from 30.5% in 2000 to 42.4% in 2018.26 27 Metabolic syndrome involved the concurrent presence of metabolic abnormalities, including obesity, hypertension, dyslipidaemia and hyperglycaemia, which elevated the risk of urolithiasis, particularly uric acid and calcium oxalate stones.28 The risk of stone formation increases with the accumulation of these traits, and previous studies have reported that metabolic syndrome can elevate the average risk of urolithiasis by 1.769 times.29 30 Additionally, the misuse of medications such as antibiotics,31 which can lead to drug-induced stones,32 also contributes to the recent relative risk increase; conversely, in China, both period and cohort effects showed a declining trend. Despite experiencing industrialisation and lifestyle changes, there are still notable differences in dietary patterns between China and the United States. In China, during the same period, the prevalence of metabolic syndrome in China was 31.1%, with an obesity rate of 8.1%,33 34 significantly lower than that in the United States. Chinese residents consume less protein and fat35 36 and more vegetables and fruits,37 and this dietary pattern is associated with a reduced risk of developing urolithiasis.38

The Nordpred model forecasts that over the next 10 years, the incidence of urolithiasis will continue to rise in the United States and China. Therefore, the focus on urolithiasis remains on prevention. Increasing fluid intake to lower urine concentration is a long-standing and widely recognised recommendation for preventing urolithiasis. Higher urine volumes also reduce the risk of recurrent urolithiasis.39 Lifestyle and habit modifications can also decrease the risk of urolithiasis, such as dietary restrictions and exercise to maintain a normal body mass index. A large cohort study involving nearly 85 000 postmenopausal women demonstrated that physical activity could prevent urolithiasis, and higher caloric intake increases the risk of incident stones.40 Dietary management is equally crucial, including limiting sodium intake, avoiding foods high in oxalates, abstaining from vitamin D supplements and reducing animal protein consumption.41 Ideally, preventive measures should involve personalised plans and identifying risk factors through detailed metabolic, nutritional and lifestyle screening. Although comprehensive screening programmes, as recommended by guidelines such as those from the International Alliance of Urolithiasis,42 may initially seem costly, in the long term, these measures can reduce the disease burden for healthcare authorities. A UK study indicates that by preventing just one episode of urolithiasis, the National Health Service can save approximately £2000,43 and similar calculations in the US indicate substantial cost savings.44 Beyond the direct cost savings associated with treating urolithiasis, the potential socioeconomic benefits are incalculable. These include improved quality of life for the population due to the avoidance of pain and discomfort, as well as the prevention of complications associated with kidney stone diseases, such as chronic kidney disease, which may take years to manifest but can lead to substantial treatment costs.

However, our analysis has limitations. The estimation process for GBD data involves collecting relevant disease datasets from various governments, international organisations and GBD collaborators, with the collected data undergoing correction and being modelled through standardised tools to produce the final estimates. Thus, the quality of the original data, the methods used for data correction and the reliability of the standardised models could all impact the usability of the final data.10 Second, the GBD study does not distinguish between different anatomical locations and chemical compositions of urolithiasis, whose epidemiological trends may vary due to climate, geographical location or level of social development. Third, the observed rising incidence in countries like the United States should be interpreted with caution, as this trend may be partly an artefact of detection bias due to the increased use of advanced medical imaging for discovering asymptomatic stones, rather than solely a true increase in disease occurrence. Methodologically, our statistical analysis relied on a web-based APC tool that could not test for or adjust for potential overdispersion, which can lead to artificially narrow CIs. Additionally, while our projections are based on established models, the Nordpred model used for forecasting does not provide CIs for its estimates, which limits the assessment of uncertainty for those specific projections. Finally, our projections to 2031 are based on the extrapolation of historical trends, an assumption challenged by the COVID-19 pandemic. Disruptions in healthcare access and lifestyle changes during the pandemic mean the actual future burden may deviate from our estimates, which should therefore be viewed with considerable caution.

This post was originally published on https://bmjopen.bmj.com