This retrospective study utilized data from the TriNetX Research Network, a federated health research network that provides access to electronic health records from healthcare organizations worldwide. The platform provides access to real-world clinical data encompassing comprehensive patient information, including demographic characteristics, laboratory values, diagnoses coded using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), medical procedures, and prescribed medications, with standardized drug coding systems. The TriNetX database has been extensively used in numerous peer-reviewed publications, supporting its reliability as a source of real-world evidence (26–28). The study was approved by the Institutional Review Board of Chi Mei Medical Center (IRB number: 11310-E04), which waived the requirement for informed consent owing to the retrospective nature of the study and the use of de-identified data, which posed minimal risk to the participants.

The study population comprised patients aged ≥18 years who underwent zinc testing between January 1, 2010, and June 30, 2023. Participants were categorized into two distinct groups based on their serum zinc concentrations: the zinc deficiency group (ZD group) included patients with serum zinc levels below 70 μg/dL, while the control group consisted of patients with serum zinc levels between 70–120 μg/dL, representing the widely accepted normal physiological range (16, 25). Accordingly, values >120 μg/dL were classified as high zinc levels, reflecting concentrations above the upper normal limit. The date of zinc testing served as the index date for each patient, establishing a clear temporal reference point for subsequent outcome assessment and ensuring uniform follow-up periods across the study cohort.

All eligible patients must have had an established diagnosis of COPD documented prior to the index date, ensuring that the study population represented individuals with existing respiratory compromise, who might be particularly vulnerable to zinc deficiency-related complications. The diagnosis of COPD in this study was identified using the ICD-10-CM codes documented in the electronic health records of participating institutions. While the GOLD guidelines provide standardized clinical and spirometric criteria for COPD diagnosis, our retrospective design necessitated reliance on ICD-10 coding to ensure consistent identification of patients across the TriNetX network.

To ensure a homogeneous study population and minimize confounding variables, patients with a pre-existing history of end-stage renal disease (ESRD) or chronic kidney disease stage 4 or 5 were excluded to avoid the confounding effects of advanced renal dysfunction on both zinc metabolism and clinical outcomes. Additionally, patients with a history of hemodialysis before the index date were excluded, as dialysis can significantly alter zinc homeostasis and overall prognosis. We also excluded patients with a documented history of malignant neoplasms of the bronchus and lung, heart or lung transplantation, cirrhosis of the liver, or respiratory tuberculosis prior to the index date, as these conditions could independently influence both the zinc status and respiratory outcomes. Furthermore, to eliminate the potential impact of acute illness on zinc levels, we excluded patients who experienced acute conditions within 1 month before the index date, including acute kidney injury, pneumonia, COPD with acute exacerbation, and sepsis.

To minimize selection bias and ensure a balanced comparison between the groups, we applied a 1:1 propensity score-matching approach. This strategy incorporated not only demographic and clinical variables but also detailed laboratory and therapeutic data potentially affecting zinc status and clinical outcomes. Baseline characteristics were extracted from the 3 years prior to the index date and included age, sex, race, body mass index (BMI), serum albumin, hemoglobin A1c (HbA1c), and hemoglobin levels.

The Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage was not available in the database, and FEV₁ values were infrequently recorded, making it difficult to incorporate direct COPD severity measures into matching. To minimize confounding by COPD severity and disease progression, we additionally matched patients for the presence of COPD-related complications, including the use of supplemental oxygen, history of COPD exacerbations, and presence of pulmonary hypertension. Furthermore, we controlled for the use of therapeutic agents that may affect COPD prognosis, specifically bronchodilators, glucagon-like peptide-1 receptor agonists (GLP-1 RAs), and sodium-glucose cotransporter-2 inhibitors (SGLT2is), which could potentially modify disease outcomes independent of the zinc status.

To further address potential treatment bias and ensure comparable therapeutic backgrounds, we matched for zinc supplementation use, as well as the utilization of angiotensin-converting enzyme (ACE) inhibitors and angiotensin II receptor blockers (ARBs), medications that could influence both renal function and overall cardiovascular health. Although C-reactive protein (CRP) levels are available in the TriNetX database, these measurements are not consistently obtained simultaneously with zinc testing. Although zinc concentrations can be transiently influenced by systemic inflammation, the lack of temporal alignment between CRP and zinc assessments could have introduced misclassification. Therefore, CRP was not incorporated as a covariate in our matching strategy.

The primary outcome was the risk of overall mortality at the 6-month follow-up. Secondary outcomes included the 6-month risk of incident COPD exacerbation, pneumonia occurrence, intensive care unit (ICU) admission, episodes of high glucose levels (defined as glucose concentrations exceeding 180 mg/dL), and hospital admission. Glucose levels and pneumonia were included as secondary outcomes, as zinc deficiency may contribute to hyperglycemia and increase susceptibility to respiratory infections such as pneumonia, a common complication in COPD. To examine the long-term effects of zinc deficiency on clinical outcomes and disease progression, we analyzed all outcomes at the 12-month follow-up.

To assess potential dose-dependent effects, we conducted a separate analysis comparing patients with severe zinc deficiency (serum zinc ≤ 50 μg/dL) to those with normal zinc levels (70–120 μg/dL). This threshold represents a more profound degree of zinc depletion, which might be associated with more pronounced clinical consequences. All outcomes were evaluated at both the 6-month and 12-month follow-up periods to assess whether severe zinc deficiency demonstrated a dose-dependent relationship with adverse clinical outcomes.

As zinc status exhibits a U-shaped dose–response relationship in many biological systems, where both deficiency and excess can be harmful, we conducted a separate analysis examining patients with high zinc levels (≥120 μg/dL) compared to those with normal zinc levels (70–120 μg/dL) at both the 6-month and 12-month follow-up periods.

To explore potential effect modifications and identify patient populations that might be at a particularly high risk for zinc deficiency-related complications, we performed prespecified subgroup analyses. These analyses were stratified by clinically relevant characteristics, including sex, presence or absence of hypertension, dyslipidemia, overweight/obesity status, asthma, and diabetes mellitus.

Baseline characteristics were summarized using appropriate descriptive statistics, with continuous variables presented as means with standard deviations and categorical variables expressed as frequencies with percentages. To achieve an optimal balance in baseline characteristics between the comparison groups, we implemented propensity score matching using a greedy nearest-neighbor algorithm with appropriate caliper settings. The quality of matching was assessed using standardized mean differences (SMD), with values less than 0.1 indicating adequate balance between groups, and visual inspection of propensity score distributions to ensure appropriate overlap.

Time-to-event outcomes were analyzed using the Kaplan–Meier method, with between-group differences assessed via the log-rank test to evaluate statistical significance. The association between zinc status and clinical outcomes was quantified using Cox proportional hazard regression models to calculate hazard ratios (HRs) with 95% confidence intervals (CIs) accounting for the time-to-event nature of the data. For subgroup analyses, the statistical significance of the differences between subgroups was evaluated by examining the confidence interval overlap. All statistical analyses were conducted using the TriNetX platform, with a two-sided p-value < 0.05 considered statistically significant for all comparisons.

The systematic patient selection process is illustrated in Figure 1. Initially, the unmatched cohort comprised 4,101 patients in the zinc deficiency group and 4,723 patients in the control group. Through propensity score matching, we successfully created two well-balanced cohorts of 3,525 patients each, ensuring optimal comparability for subsequent analyses. The effectiveness of our matching strategy is visually demonstrated in Figure 2, where the propensity score density distributions show an improvement in the overlap between groups after matching.

Table 1 presents the baseline characteristics before and after propensity score matching. Prior to matching, several differences existed between groups, including body mass index (40.0 ± 10.6 kg/m² in zinc deficiency versus 32.2 ± 10.2 kg/m² in controls), dyslipidemia prevalence, overweight and obesity, anemia, heart failure, and malnutrition rates. After matching, these differences were effectively eliminated, with all standardized mean differences falling below 0.1, indicating excellent balance. The final matched cohorts were comparable across all key demographics, with a mean age of approximately 61 years, predominantly female composition (approximately 63%), and similar prevalence of major comorbidities, including diabetes mellitus (37%), hypertension (69%), and COPD-related complications.

As shown in Table 2, zinc deficiency was associated with a nearly two-fold increase in 6-month mortality risk, with 163 deaths (4.6%) in the ZD group compared to 86 deaths (2.4%) in the control group (HR:1.94, 95% CI: 1.49–2.52, p < 0.001), underscoring the clinical relevance of zinc status in managing COPD. The secondary outcomes also demonstrated a consistent pattern of increased morbidity associated with zinc deficiency. Pneumonia risk was modestly but significantly elevated in the ZD group (HR 1.24, 95% CI: 1.02–1.50, p = 0.031), supporting the established role of zinc in immune function. Episodes of hyperglycemia (glucose >180 mg/dL) also occurred more frequently in zinc-deficient patients (HR 1.28, 95% CI: 1.13–1.43, p < 0.001), reflecting the crucial role of zinc in glucose metabolism. Regarding health source utilization, the risk of ICU admission (HR 1.61, 95% CI: 1.28–2.03, p < 0.001) and hospital admissions were also significantly increased (HR 1.28, 95% CI: 1.17–1.39, p < 0.001). Interestingly, there was no increased risk of COPD exacerbations in patients with zinc deficiency compared to the control group (HR 1.17, 95% CI: 0.95–1.45, p = 0.142).

Analysis at the 12-month follow-up (Table 3) demonstrated that the adverse effects of zinc deficiency persisted over extended follow-up periods, with a pattern remarkably similar to the 6-month findings. Mortality remained significantly elevated in the ZD group (HR 1.65, 95% CI: 1.34–2.03, p < 0.001), although with a somewhat attenuated effect size compared to the 6-month analysis. All other outcomes, including pneumonia (HR 1.19, p = 0.023), hyperglycemia (HR 1.23, p < 0.001), ICU admission (HR 1.61, p < 0.001), and hospital admission (HR 1.24, p < 0.001), showed statistically significant associations with zinc deficiency. The consistency of these findings across both time periods reinforces the sustained clinical impact of zinc status on COPD outcomes, suggesting that zinc deficiency is a persistent rather than a transient risk factor.

The analysis of severe zinc deficiency revealed a dose–response relationship (Table 4), providing evidence for the biological plausibility of our findings. Patients with severe zinc deficiency demonstrated markedly elevated risks across all measured outcomes compared to those with normal zinc levels. The mortality risk was particularly pronounced at 6 months (HR: 1.96, 95% CI: 1.35–2.86, p < 0.001) and at 12 months (HR: 1.90, 95% CI: 1.41–2.55, p < 0.001).

Most remarkably, severe zinc deficiency was associated with a more than two-fold increase in COPD exacerbation risk at 6 months (HR 2.42, 95% CI: 1.63–3.58, p < 0.001). This suggests that the relationship between zinc status and respiratory stability may have a critical threshold effect. ICU admission risks were dramatically elevated, with hazard ratios exceeding 2.5 at both time points. These findings collectively support a dose-dependent relationship between the severity of zinc deficiency and adverse clinical outcomes.

The analysis of patients with elevated zinc levels revealed an important U-shaped relationship between zinc status and outcomes (Table 5). Contrary to the assumption that higher zinc levels might be protective, we observed significantly increased mortality risk in patients with high zinc levels compared to those with normal levels (HR: 1.74, 95% CI: 1.18–2.55, p = 0.005) at 6 months and at 12 months (HR: 1.79, 95% CI: 1.32–2.43, p < 0.001). ICU admission risk was also elevated in patients with high zinc status at the 6- and 12-m follow-up. In contrast, other outcomes, including COPD exacerbations, pneumonia, hyperglycemia, and hospital admissions, showed no significant associations with high zinc levels.

The subgroup analysis (Table 6) examined whether the impact of zinc deficiency on 6-month mortality varied across different patient populations. Although numerical differences in risk were observed between the subgroups, none of the interaction p-values reached statistical significance (all p > 0.05). These findings indicate that the adverse impact of zinc deficiency on mortality risk is consistent across sex, comorbidity status (hypertension, diabetes, dyslipidemia, obesity), and asthma presence, suggesting that the clinical importance of zinc status applies broadly to patients with COPD, regardless of their baseline characteristics.

Multivariable analysis (Table 7) confirmed zinc deficiency as an independent predictor of 6-month mortality after adjusting for multiple potential confounders. Zinc deficiency emerged as one of the strongest modifiable risk factors (HR: 2.03, 95% CI: 1.59–2.58, p < 0.001). Other significant independent predictors included male sex (HR 1.56, p < 0.001), advanced age (HR 1.05 per year, p < 0.001), heart failure (HR 2.54, p < 0.001), nicotine dependence (HR 1.44, p = 0.003), and malnutrition (HR 2.24, p < 0.001). Notably, overweight and obesity showed a protective effect (HR 0.54, 95% CI: 0.40–0.72, p < 0.001). The persistent association between zinc deficiency and adverse outcomes, even after adjusting for malnutrition, suggests that the effects of zinc extend beyond general nutritional status, underscoring its distinct biological role in COPD.