Recurrence of kidney stones: a meta-analysis of rates from randomized trials


Nephrolithiasis is a frequent condition in the general population, with an estimated prevalence of about 8% (“Croppi E – 2012” [1]). The risk of recurrence after the first episode is thought to be high; however, it has never been analyzed in a systematic way. Observational studies are often not adequate to obtain correct estimates of risk of recurrence; furthermore, the improvements in diagnosing and treating kidney stones as well as the changing trends in the disorder might have rendered figures from older studies obsolete. The aim of our study was to obtain estimates of recurrence risk for kidney stones from published randomized controlled trials, and to analyze them for temporal trends and in specific subgroups.


We searched the Pubmed Medline and Cochrane databases from inception to May 2013 for English-written randomized controlled trials in adults affected with idiopathic calcium stones. We retrieved 21 articles that were analyzed in full. Relevant data were extracted from these articles including publication year, type of intervention, previous history of stones, average age and proportion of males; when possible, data extraction was limited to patients who were stone-free at the beginning of the study (e.g., patients with residual stones or stone fragments at the beginning of the study were excluded). For each study, the length of follow-up was recorded together with either the proportion or the rate of recurrence, defined as a new stone event during follow-up. Recurrence risks were transformed into rates when needed.

We calculated recurrence rates with 95% confidence interval (CI) by means of a multilevel Poisson regression model clustered on study with weights based on sample size and follow-up length, with studies with larger sample sizes and longer follow-up being given larger weights.

To investigate linear and non-linear temporal trends and differences in overall recurrence rates by treatment type, previous history of stones, average age and proportion of males, we constructed models including terms for each of those variables.


Twenty-one studies were included in the analysis, contributing data from 2,168 participants in 46 study arms over a median follow-up of 3 years. One-hundred fifty six (7.2%), 1,528 (70%) and 484 (22%) of the participants were included in the placebo/no intervention (n = 4 study arms), dietary changes (n = 26 study arms) and drug treatment (n = 16 study arms) groups, respectively. The majority (71%) of the included studies reported a composite of clinical and imaging recurrence as the primary outcome, and 76% of the studies included participants with recurrent disease (history of more than one episode at enrollment in the trial). Publication years ranged from 1976 to 2011.

Overall, the recurrence rate was 18 per 100 person-years (95% CI 9.9, 33).

There was a borderline significant (p = 0.05) decrease of recurrence rates over time (figura 0); more complex models of time such as cubic splines did not show a significantly better fit (p = 0.11).

The recurrence rates in the placebo/no intervention, dietary changes and drug treatment groups were 22 (95% CI 9.3, 34), 19 (95% CI 8.8, 29) and 15 (95% CI 0, 35) per 100 person-years, respectively. The shape of recurrence rates over time appeared different in the dietary changes and drug treatment groups (figura 0), although the interaction was not statistically significant (p = 0.77). We did not include the placebo/no intervention studies in this analysis since they were too sparse over time (1 study in 1982, 1 in 1986, 1 in 1996 and 1 in 2006).

The number of stone episodes before enrolment in the trial was strongly associated with recurrence, with a rate of 5.9 (95% CI 2.9, 8.8) per 100 person-years in those with a single stone episode (first stone formers) compared with 22 (95% CI 8.4, 36) per 100 person-years in those with two or more stone episodes (p = 0.001).

Average age was directly associated with recurrence rates (p = 0.04), whereas proportion of males was not (p = 0.40).

In a full model including calendar time and average age, we evaluated the interaction between the number of stone episodes at enrollment and the effect of intervention, and found it to be significant (p < 0.001): using the most recent data available (2011), among those with one previous stone episode, the recurrence rates were 13, 3.6 and 26 per 100 person-years for patients untreated, treated with dietary changes and treated with drugs, respectively. Among those with two or more stone episodes, however, recurrence rates were 19, 14 and 11 per 100 person-years for patients untreated, treated with dietary changes and treated with drugs, respectively.


Our study suggests that characteristics such as number of previous stone episodes and type of treatment highly impact such recurrence rates. The data would also suggest that a watch and wait policy or dietary advice might be feasible in first idiopathic calcium stone formers whereas treatment, at least with drugs, could be reserved to recurrent or otherwise complicated patients. The recurrence rates reported in our paper might be of interest in the selection of patients to be enrolled in randomized controlled trials to maximize cost-effectiveness, as well as assist with therapeutic decisions for patients following their first stone episode or with recurrent stone disease.