Doxycycline is a second-generation, broad-spectrum, semi-synthetic tetracycline antibiotic derived from oxytetracycline (1, 2). It exhibits a high degree of lipophilicity, a long plasma half-life, and excellent oral bioavailability (3). Doxycycline acts primarily by binding to the 30S ribosomal subunit of bacteria, inhibiting protein synthesis and thereby exerting a bacteriostatic effect (4). Its favorable pharmacokinetics, including good tissue penetration and a lower affinity for calcium than that of earlier tetracyclines, make it a valuable agent for treating a wide range of infections, including respiratory tract infections, atypical pneumonia, rickettsial diseases, malaria (prophylaxis), sexually transmitted infections, and inflammatory conditions such as acne and rosacea (3, 5, 6).

Despite its therapeutic advantages, the use of doxycycline in children—particularly those under 8 years of age—has historically been limited due to concerns about adverse effects on developing teeth (7, 8). Tetracyclines, including doxycycline, are known to bind to calcium ions and can become incorporated into the hydroxyapatite matrix of teeth (9) and bones (10) during mineralization. This chelation process can lead to permanent tooth discoloration, typically presenting as yellow, brown, or gray staining, and, in severe cases, enamel hypoplasia (11). The mechanism is believed to involve oxidation of the tetracycline-calcium orthophosphate complex upon light exposure, resulting in pigmentation (12). Such discoloration is not only irreversible but may also lead to psychosocial distress and reduced quality of life, particularly during adolescence and adulthood (12).

While the association between tetracycline use and dental staining is well-documented, the risk specifically associated with doxycycline has been increasingly questioned. Doxycycline has a lower calcium-binding capacity than that of first-generation tetracyclines, and recent clinical observations suggest that the incidence of tooth discoloration may be much lower than previously assumed, especially when the antibiotic is used for short durations (8). In recognition of its effectiveness in life-threatening infections such as Rocky Mountain spotted fever, current guidelines from the Centers for Disease Control and Prevention (CDC) support the use of doxycycline in young children, including those under 8 years, when the benefits outweigh the risks (13). However, regulatory labeling and clinical practice in many regions remain cautious, often discouraging its use in children under 8 years of age. Given this clinical uncertainty, a comprehensive synthesis of the available evidence is warranted. Therefore, in this study, we performed a meta-analysis to estimate the overall incidence of tooth discoloration in children treated with doxycycline and to examine potential influencing factors through subgroup and meta-regression analyses.

This meta-analysis followed the PRISMA 2020 guidelines (14, 15) and the Cochrane Handbook for Systematic Reviews and Meta-Analyses (16) for protocol design, data extraction, statistical analysis, and results reporting. The protocol of the meta-analysis has been registered at PROSPERO with the identifier CRD420251009690.

This meta-analysis synthesized evidence from 17 studies to estimate the incidence of tooth discoloration in children treated with doxycycline and to explore potential influencing factors. The pooled incidence of tooth discoloration was 0.92% (95% CI: 0.34%–1.50%), based on data from diverse pediatric populations and study designs. While this suggests a low overall incidence, the result should be interpreted cautiously due to variability in patient characteristics and outcome assessment across studies. Subgroup analyses revealed no significant differences in incidence by geographic region, study design, age group, administration route, or method of outcome assessment. Furthermore, univariate meta-regression analyses demonstrated that factors such as publication year, sample size, mean age at exposure, treatment duration, doxycycline dose, follow-up duration, and study quality scores did not significantly affect the reported incidence. These findings suggest that doxycycline may be safer in children, including those less than 8 years of age, than previously believed.

Our findings are consistent with and supported by emerging literature that challenges the traditional avoidance of doxycycline in young children. In a recent 2024 pharmacovigilance analysis based on the FDA Adverse Event Reporting System (FAERS), Qiao et al. evaluated adverse reports involving doxycycline, minocycline, and tigecycline in pediatric patients (38). Their findings demonstrated a low incidence of dental adverse events with doxycycline relative to other tetracycline derivatives, reinforcing its relative safety and suitability for use in young children when clinically warranted (38). Building on this, a more recent large-scale pharmacovigilance analysis by Zhang et al. evaluated 21,561 adverse event reports associated with tetracyclines and found that among children under 8 years of age, only doxycycline, minocycline, and omadacycline generated positive signals for tooth discoloration (39). Doxycycline had lower signal strength and fewer reported events than minocycline. Notably, the majority of reported cases in children were associated with short treatment durations (median onset: 60 days), and only one case in a 2.5-year-old child involved a 10-day course of doxycycline (39). These pharmacovigilance data further support the notion that the actual risk of tooth discoloration from doxycycline in children is very low, particularly when used appropriately and for short durations. Moreover, this study highlights that previous safety concerns may have been overstated due to historical high-dose exposures and less refined surveillance systems (39).

Our subgroup analyses provide further context for interpreting the main findings. The incidence of tooth discoloration did not significantly differ between children exposed before and after 8 years of age, suggesting that early age alone may not be a strong risk factor for discoloration under the typical clinical scenarios and treatment durations reported in these studies. Similarly, the route of administration (oral vs. oral/IV) and the method of outcome detection (direct observation vs. self-report) did not seem to significantly influence the incidence. Although one might expect IV administration or self-reported outcomes to increase variability or reporting bias, our results remained robust across these subgroups. Notably, the low heterogeneity (I²=0%) across studies supports the consistency of the findings and increases confidence in the pooled estimate. The very low heterogeneity (I² = 0%) likely reflects the uniformly low incidence observed, similar short-course dosing patterns, and the use of a binary, clinically observed endpoint across studies. However, this apparent consistency may also be influenced by under-ascertainment of mild cases, retrospective data capture, and small-study effects, potentially masking true between-study variability. Our subgroup and meta-regression analyses did not identify study-level modifiers, and although Egger's test was non-significant, the funnel plot suggested possible small-study or reporting bias. Therefore, the apparent “safety” signal should be interpreted cautiously, and prospective IPD-based studies with standardized dental assessments are warranted. On the other hand, meta-regression analyses did not identify any study-level factors that significantly modified the incidence of tooth discoloration. This includes treatment duration, doxycycline dose, and age at exposure, which were evaluated in the regression model but showed no statistically significant associations with the risk of discoloration. These findings suggest that short courses of doxycycline, as commonly used in pediatric infections, may not meaningfully increase the risk of tooth staining. The lack of associations in our meta-regression might also be explained by the limited variation in dosing and duration across studies, or by the inherent limitations of study-level analyses that cannot fully capture individual-level risk factors (40). While this strengthens the argument for a low overall incidence, it also underscores the need for patient-level data to better elucidate whether certain subgroups of children may still be at slightly elevated risk (40). In addition, a sensitivity analysis excluding the Lochary 1998 study (22)—an early report with a relatively high incidence—yielded a nearly identical pooled estimate of 0.91% (95% CI: 0.33%–1.49%) and unchanged heterogeneity (I² = 0%). This consistency reinforces the reliability of our overall findings. The elevated rate in Lochary 1998 (22) may reflect differences in follow-up duration, diagnostic thoroughness, or historical treatment practices.

Several strengths of this meta-analysis merit attention. First, we performed a comprehensive and up-to-date literature search across five databases, including both English and Chinese-language sources, to ensure broad and inclusive study representation. Second, the analysis included over 1,000 pediatric patients from diverse geographic regions and clinical contexts, enhancing generalizability. Third, the use of multiple subgroup and meta-regression analyses enabled a nuanced exploration of potential modifiers of risk. These analytical approaches, rarely used in prior reviews on this topic, add methodological rigor and depth to our findings. Nonetheless, there are limitations to consider. Most included studies were retrospective in nature, and therefore susceptible to recall and selection bias (41). In particular, retrospective designs relying on clinical records or parental recall may underreport or inconsistently document tooth discoloration, especially if it was mild or not the primary focus of the study (41). Second, although we included studies from both English and Chinese databases, we excluded articles in other languages, which may have led to a degree of publication bias despite Egger's test suggesting no significant risk (42). Third, as with most meta-analyses of observational studies, our subgroup and meta-regression analyses were based on aggregated study-level data rather than individual patient data. This ecological approach limits the ability to account for confounding variables and may mask associations present at the individual level (43). In addition, follow-up durations varied considerably among the studies, and some were relatively short, potentially missing delayed or subtle manifestations of dental changes. Finally, although the included studies varied in patient age, clinical indications, and outcome assessment methods, our subgroup and meta-regression analyses consistently showed no significant differences in the incidence of tooth discoloration across these factors. Due to the lack of individual-level data, we were unable to directly assess risk during the specific window of permanent tooth formation. As an alternative, we stratified subgroup analyses using study-level median or mean age at exposure (e.g., ≤8 years vs. >8 years) to approximate the risk window. While this approach offers practical value, it may not precisely reflect individual variation in tooth development. Moreover, although minor shifts in the timing of permanent tooth eruption may have occurred across populations or over time, current evidence suggests that the general timeline for eruption has remained relatively stable in recent decades. Future individual participant data (IPD)-level studies would allow for more accurate stratification by dental developmental stage. In addition, we also recognize that tooth discoloration may involve not only enamel but also underlying dentin, and that the risk may be influenced by the timing of tooth bud formation—ranging from gestation through infancy for deciduous teeth, and from infancy through approximately 7–8 years for permanent teeth (44). However, due to the unavailability of IPD, we were unable to separately analyze outcomes in children exposed during these critical developmental periods. Instead, age-based stratification was performed at the study level using median or mean age, which may not precisely align with the biological windows of tooth bud formation. The overall statistical heterogeneity was low (I² = 0%), suggesting that the pooled estimate is stable despite variation in study characteristics. Nonetheless, we acknowledge that differences in study design and assessment approaches may introduce unmeasured bias and should be interpreted with caution. Finally, it is also important to note that while our analysis focused on tooth discoloration, doxycycline may theoretically carry a risk of enamel hypoplasia, particularly when administered during early stages of tooth development (26). However, none of the included studies reported enamel hypoplasia as a separate outcome, and most assessments relied on direct clinical observation or caregiver report, which are more likely to detect visible discoloration than structural enamel defects. As such, our findings may not capture the full spectrum of potential dental effects associated with doxycycline, underscoring the need for future studies with comprehensive dental evaluations.

From a clinical standpoint, the findings of this meta-analysis have important implications. In conditions where doxycycline is the most effective or recommended therapy—such as rickettsial infections, atypical pneumonia, or macrolide-resistant Mycoplasma pneumoniae—its use in children, including those under 8 years of age, should not be automatically precluded due to concerns about tooth discoloration alone. The low incidence reported here, together with supporting evidence from recent prospective and pharmacovigilance studies (38, 39), suggests that the historical restriction on pediatric doxycycline use may be overly cautious. This is consistent with updated CDC recommendations, which now endorse doxycycline for children of all ages in specific life-threatening infections (45, 46). Future research should focus on prospective, controlled studies with standardized dental assessments and sufficient follow-up to capture any late-onset effects. Ideally, studies should incorporate patient-level data to enable more precise risk stratification based on age, dosing, cumulative exposure, genetic factors, and dental development stage. Moreover, although our meta-analysis focused exclusively on doxycycline, we acknowledge that direct comparisons with other tetracyclines (e.g., tetracycline or minocycline) could offer valuable insight. However, such analyses were not feasible due to the lack of comparable incidence data or direct comparative studies. Future meta-analyses or head-to-head observational studies are warranted to better contextualize the relative dental risks across different tetracycline-class agents. In addition, exploration of parent and patient perceptions of dental side effects could provide insight into the psychosocial impact of tooth discoloration and inform shared decision-making in clinical practice.

In conclusion, this meta-analysis demonstrates that the incidence of tooth discoloration following doxycycline treatment in children is low, and not significantly affected by patient age, administration route, or study design. These findings support a re-evaluation of long-standing prescribing restrictions and provide evidence to guide safer, more informed use of doxycycline in pediatric care. With appropriate clinical judgment, doxycycline may be a reasonable therapeutic option even in children under 8 years of age, particularly when the benefits clearly outweigh the risks.