Lipid profile changes through nurse-led cardiac rehabilitation: evidence from a systematic review and meta-analysis

Abstract

Background

Cardiovascular diseases (CVDs) continue to be the leading cause of death worldwide, with dyslipidemia as one of their main risk factors. Nurse-led cardiac rehabilitation (NLCR) has been shown to improve the quality of life of cardiac patients. We aim to measure its effect on lipid levels.

Methods

We conducted a systematic review and meta-analysis of clinical trials comparing NLCR with usual care. Major electronic databases and grey literature sources were searched through July 31, 2025. Eligible studies reported at least one lipid outcome, including triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C). Risk of bias was assessed using RoB 2.0 for randomized controlled trials (RCTs) and ROBINS-I for non-RCTs. Random-effects models were used to calculate pooled standardized mean differences (SMDs) and mean differences (MDs) with 95% confidence intervals (CIs).

Results

Ten studies were included in the meta-analysis. NLCR was associated with a significant reduction in TG levels (SMD = − 0.19; 95% CI: −0.36 to − 0.02), corresponding to an absolute reduction of 8.38 mg/dL (MD = − 8.38; 95% CI: −13.82 to − 2.94). Changes in TC (MD = − 6.03 mg/dL; 95% CI: −14.57 to 2.51), LDL-C (MD = − 4.78 mg/dL; 95% CI: −9.89 to 0.33), and HDL-C (MD = 1.90 mg/dL; 95% CI: −2.30 to 6.10) did not reach statistical significance in the overall analysis. Subgroup analyses demonstrated a significant TG reduction among RCTs (MD = − 9.25 mg/dL; 95% CI: −14.84 to − 3.65). According to follow-up duration, studies with follow-up durations of ≤ 3 months showed a significant reduction in TC levels (MD = − 9.81 mg/dL; 95% CI: −14.13 to − 5.49), whereas studies with follow-up durations of > 3 months demonstrated a significant reduction in TG levels (MD = − 8.02 mg/dL; 95% CI: −13.95 to − 2.09). Meta-regression analyses identified study design as a significant predictor of TG and LDL changes.

Conclusion

NLCR was associated with a significant reduction in TG levels among patients with CVD, while changes in TC, LDL-C, and HDL-C were not statistically significant. Subgroup and meta-regression analyses suggested that study design and follow-up duration may influence lipid outcomes. Further high-quality RCTs with longer follow-up are needed to confirm these findings and evaluate their long-term clinical implications.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12872-026-06218-5.

Introduction

It is estimated that about 80% of all deaths worldwide in 2021 will be due to cardiovascular diseases (CVDs), which account for over 20.5 million deaths worldwide [1]. The upward trend in the incidence of heart failure, stroke, and other cardiovascular events suggests a substantial increase in global disease burden by 2050 [2]. Among the major modifiable risk factors contributing to atherosclerosis and ischemic events is dyslipidemia, characterized by elevated levels of low-density lipoprotein cholesterol (LDL-C), high triglycerides (TGs), and reduced high-density lipoprotein cholesterol (HDL-C) [3, 4]. LDL-C promotes atherogenesis by penetrating the vascular endothelium and initiating plaque formation, thereby impeding blood flow and increasing ischemic risk. HDL-C has traditionally been considered a favorable lipid parameter because of its role in reverse cholesterol transport and its inverse association with cardiovascular risk. However, contemporary evidence suggests that HDL-C may be more appropriately regarded as a marker of cardiovascular risk rather than a direct causal therapeutic target [4, 5]. In addition, TGs and TG-rich lipoproteins, particularly remnant particles, have been identified as potent atherogenic agents [6]. Consequently, optimizing lipid profiles remains a cornerstone in both the prevention and management of CVD.

Nurse-led cardiac rehabilitation (NLCR) represents an innovative and evidence-based approach to the management of CVDs, integrating structured education, continuous monitoring, behavioral counseling, and psychological support to enhance both physical and mental well-being in patients [7]. Through their proactive involvement in treatment follow-up, lifestyle modification, and promotion of medication adherence, nurses play a pivotal role in mitigating metabolic risk factors such as dyslipidemia [8]. However, despite existing evidence of its potential benefits, significant gaps remain in the understanding of how NLCR influences multiple lipid parameters in CVD management. Current evidence indicates that NLCR interventions can lead to reductions in LDL-C, TG, and blood pressure, while also influencing HDL-C levels [9, 10]. Moreover, the supportive and motivational framework of these programs encourages long-term dietary and physical activity changes, which are linked to improved endothelial function and decreased atherosclerotic burden [11, 12].

Recent clinical findings have further underscored the significance of NLCR as a valuable intervention in cardiovascular care. A 12-week randomized controlled trial (RCT) demonstrated that NLCR was associated with notable reductions in total cholesterol (TC), LDL-C, TGs levels, along with an increase in HDL-C concentrations [13]. Similarly, a study involving post-myocardial infarction patients reported that nurse-managed care, in comparison to standard care, resulted in decreased LDL-C and TC levels and elevated HDL-C levels [14]. In a distinct intervention study conducted by Page et al., patients who underwent coronary artery bypass grafting (CABG) demonstrated a significant reduction in TC, in addition to notable improvements in both body mass index (BMI) and blood pressure levels [15]. Collectively, these preliminary studies highlight the potential clinical value of NLCR in optimizing cardiovascular risk profiles.

Given the substantial global burden of CVDs and the critical role of dyslipidemia in its pathogenesis, implementing accessible, cost-effective, and evidence-based interventions such as NLCR may play a crucial role in mitigating atherosclerotic risk and improving overall clinical outcomes. This study aims to contribute valuable insights into this area by evaluating the comprehensive impact of NLCR on lipid profiles across a range of key parameters, including LDL-C, TGs, TC, and HDL-C. By synthesizing both RCTs and non-RCTs, this meta-analysis seeks to provide a more robust understanding of the effects of NLCR in diverse patient populations. Additionally, the study’s inclusion of subgroup analyses, which account for influencing factors such as follow-up duration and study quality, represents a novel approach that has been underexplored in prior reviews. Through this systematic review and meta-analysis, this study aims to provide evidence that may inform the optimization of NLCR programs and contribute to improved cardiovascular risk management and patient outcomes.

Materials and methods

Study design

This study was carried out as a systematic review and meta-analysis, following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, to ensure methodological clarity and rigor in the reporting process (Table S1) [16]. The study protocol was registered in the PROSPERO under the registration number CRD420261332730.

Search strategy and data sources

A thorough literature review was performed utilizing a range of prominent electronic databases, such as PubMed/MEDLINE, Scopus, Embase, the Cochrane Library, Web of Science, and Google Scholar. Following the identification of relevant keywords through a combination of published articles, expert consultation, and Medical Subject Headings (MeSH), a tailored search strategy was developed and applied to each database. The search covered all available records from database inception through July 31, 2025, and incorporated terms such as “Cardiac Rehabilitation,” “Nurse,” “Trials,” “HDL,” “LDL,” “VLDL,” “Cholesterol,” and “Triglyceride.” Also, the publication date or language was not restricted. To enhance the inclusiveness of the review, relevant grey literature was also screened and evaluated (Table S2).

Study selection process

Following the development of customized search strategies for each database, two independent reviewers imported all retrieved records into EndNote software (version 2020). Duplicate entries were identified and removed using the software’s automated de-duplication feature. Two independent authors identified studies that met the inclusion criteria, and disagreements were settled by a corresponding author.

Eligible studies included published clinical trials comparing NLCR with either standard care or non-intervention control groups. To qualify for inclusion, studies were required to report at least one relevant lipid profile outcome, such as TGs, TC, HDL-C, LDL-C, or VLDL. Exclusion criteria encompassed review articles, case reports, case series, ecological studies, observational designs (including cohort, case-control, and cross-sectional studies), laboratory-based investigations, and trials lacking a control group. Additionally, studies were excluded if the intervention targeted non-cardiac populations, if nurses were not the primary providers of the rehabilitation program, or if lipid-related outcomes of interest were not reported.

Data extraction and variables

An independent data extraction was carried out by two authors in accordance with the earlier stage. If discrepancies occurred, a third investigator was consulted to resolve them. Extracted data were then compiled using Microsoft Excel 2019 (Microsoft, Redmond, WA, USA). Lipid profile outcomes were initially extracted in mg/dL as reported in the primary studies, and when values were provided in mmol/L, they were converted using the “Meta Converter” tool to ensure consistency across datasets [17].

The final data extraction form captured the following items: first author and year of publication, countries, study design, mean age of intervention and control groups, year of data collection, sample size in each group, study quality assessment, follow up duration, type of cardiac condition, and relevant lipid parameters.

Quality appraisal

The methodological quality of the included studies was assessed using the Cochrane RoB 2.0 tool for RCTs and the ROBINS-I tool for non-RCTs [18]. RoB 2.0 evaluates five core domains of bias, categorizing them into low, high, or uncertain risk, while ROBINS-I examines seven domains, including participant selection and outcome reporting. All assessments were conducted independently by two reviewers, with any discrepancies resolved by a third investigator.

Statistical analyses

Data analysis was carried out using STATA software (version 18.0, StataCorp, College Station, TX, USA). A random-effects model using restricted maximum likelihood (REML) estimation was applied a priori as a conservative analytical approach. This model was selected because variability in study populations, intervention characteristics, and implementation strategies was anticipated across the included studies, and the random-effects framework allows for the incorporation of potential between-study heterogeneity. Pooled effect sizes were calculated as standardized mean differences (SMDs; Hedges’ g) with 95% confidence intervals (CIs). To enhance clinical interpretability, mean differences (MDs) with corresponding 95% CIs were additionally calculated for lipid outcomes reported on a common scale (mg/dL). Because all lipid outcomes were reported using the same measurement unit (mg/dL), MDs were considered particularly useful for facilitating clinical interpretation of the pooled findings. The analysis focused on change scores (the differences between pre- and post-intervention values), which were derived from baseline and follow-up data extracted from the included studies and calculated using Microsoft Excel [19]. Forest plots were generated to present both SMD- and MD-based pooled estimates alongside individual study results. Statistical significance was set at p < 0.05. To assess heterogeneity, both Cochran’s Q test and Higgins’ I² statistic were used, categorizing I² values as low (< 25%), moderate (25–50%), or high (> 50%) heterogeneity [20]. If studies did not report standard deviations (SDs) of changes in continuous outcomes from baseline, we calculated them using the formula:

assuming a correlation coefficient (R) of 0.5 [21, 22]. Subgroup analyses were conducted to further explore potential sources of heterogeneity. These analyses were performed according to study design (RCTs vs. non-RCTs), follow-up duration (≤ 3 months vs. > 3 months), and risk of bias classification (low risk vs. some concerns/moderate risk). The selection of follow-up duration categories was based on distinguishing short-term (≤ 3 months) versus relatively longer-term (> 3 months) intervention effects, which is a commonly used threshold in cardiac rehabilitation and lifestyle intervention studies. In addition, stratification based on risk of bias was performed to assess the potential influence of study quality on the pooled estimates, in accordance with methodological recommendations for systematic reviews and meta-analyses. For meta-regression analyses, the influence of various study variables, including study design, follow-up duration, published year, start year of data collection, and study quality, on lipid changes was evaluated. The meta-regression models were used to explain the variability in lipid outcomes across studies, with R-squared values indicating the proportion of heterogeneity explained by the study variables. Coefficients, p-values, and I² statistics were reported for both univariable and multivariable analyses to assess the robustness of the findings. Sensitivity analyses, including the leave-one-out method, were conducted to evaluate the robustness of the effect estimates. Funnel plot visualization and Egger’s weighted regression test were employed to assess publication bias, with a p-value > 0.05 indicating no significant bias [23].

Results

Study selection

A total of 881 studies were initially identified through comprehensive searches of international databases. After the removal of 210 duplicate entries, 671 records remained for title and abstract screening. Following this screening process, 19 full-text articles were assessed for eligibility. Ultimately, 10 studies, comprising 12 individual records, met the inclusion criteria and were incorporated into the final quantitative meta-analysis (Fig. 1).

Fig. 1.

PRISMA flowchart

Study characteristics and risk of bias assessment

All ten included studies consisted of RCTs and non-RCTs controlled trials conducted among patients with CVDs. The years of publication ranged from 2002 to 2022, with data collection periods spanning from 2000 to 2017. The mean age of participants was generally comparable between the intervention and control groups across studies. The majority of trials were conducted in China (n = 2) and India (n = 2). Additional study characteristics are summarized in Table 1. Because only one study reported the effects of NLCR on VLDL levels [13], this outcome was not included in the quantitative meta-analysis due to insufficient data. Based on the applied evaluation tools, six studies were deemed to be of “some concern” or “moderate risk” regarding bias, while four were considered to have low risk of bias (Table S3).

Table 1.

Basic information

ID	Author	Type of Trial	Duration of data collection	country	Mean-age of intervention	SD-age of intervention	Mean-age of control	SD-age of control	Rehabilitation programs						Lipid-lowering medication use (number)	Diagnosis of patients	Follow up time	Sample size of Intervention	Sample size of control
									Education	exercise	Telephonic/ home visits follow-up	psychological support	dietary counseling	lifestyle counseling
1	Savage et al. (2002) [24]	Non-RCT	1997–2000	USA	63.40	9.70	59.70	11.50	✔	✔	✔	✔	✔	-	NA	Coronary heart disease	≤ 3 Months	27	55
2	Kim et al. (2014) [25]	Non-RCT	2010–2011	Republic of Korea	63.64	12.26	60.39	16.97	✔	✔	✔	✔	✔	-	NA	Coronary heart disease	≤ 3 months and > 3 months*	32	29
3	Fernandez et al. (2009) [26]	RCT	2006–2006	Australia	57.14	8.96	56.41	8.73	✔	✔	-	-	✔	-	NA	Acute coronary syndrome	≤ 3 months	29	22
4	Jiang et al. (2006) [27]	RCT	2002–2002	China	62.11	7.44	61.37	7.61	✔	-	✔	-	-	✔	Intervention: 72 / Control: 70	Coronary heart disease	≤ 3 months and > 3 months*	83	84
5	Arjunan et al. (2021) [28]	RCT	2011–2014	India	All participants: age mean = 59.57, age SD = 11.16				✔	✔	✔	✔	-	-	NA	Chronic heart failure	≤ 3 months	100	100
6	Ruiz-Bustillo et al. (2019) [29]	RCT	2012–2013	Spain	63.33	13.85	61.00	13.08	✔	✔	✔	✔	-	-	Intervention: 39 / Control: 37	Ischemic Heart disease	> 3 months	39	39
7	Premkumar et al. (2022) [13]	RCT	2017–2017	India	45.1% of participants in 50–65 years old		58.0% of participants in 50–65 years old		✔	✔	✔	-	✔	-	NA	Cardiovascular disease	≤ 3 months	31	31
8	Zutz et al. (2006) [30]	RCT	2006–2006	Canada	58	4	59	12	✔	✔	-	✔	✔	✔	NA	Cardiovascular disease	≤ 3 months	8	7
9	Köhler et al. (2020) [31]	RCT	2011–2015	Sweden	68.5	9.2	68.9	7.7	-	-	✔	✔	-	✔	NA	Coronary heart disease	> 3 months	79	78
10	Jiegang et al. (2019) [32]	RCT	2017–2017	China	53.3	4.2	54.6	4.7	✔	✔	✔	-	✔	-	NA	Coronary heart disease	≤ 3 months	43	42

NA Not available, SD Standard deviation, RCT Randomized Clinical Trial

* These studies were sub grouped for this variable

Results of the meta-analysis

Lipid profiles changes through NLCR

The meta-analysis of nine clinical trials indicated a significant reduction in TG levels among participants receiving NLCR compared with those in the control groups (SMD = − 0.19; 95% CI: −0.36 to − 0.02; I² = 29.07%) (MD = − 8.38; 95% CI: −13.82 to − 2.94; I² = 0.00%) (Fig. 2). Also, Pooled results from eight trials showed no statistically significant difference in TC levels between the NLCR and control groups (SMD = − 0.20 (95% CI: −0.48 to 0.07; I² = 63.15%) (MD = − 6.03; 95% CI: −14.57 to 2.51; I² = 52.68%) (Fig. 3). Likewise, no statistically significant difference in LDL-C levels was observed between groups (SMD = − 0.18; 95% CI: −0.41 to 0.05; I² = 61.72%) (MD = − 4.78; 95% CI: −9.89 to 0.33; I² = 43.26%) (Fig. 4). Furthermore, no statistically significant difference in HDL-C levels was observed between NLCR and control groups (SMD = 0.14; 95% CI: −0.20 to 0.49; I² = 79.07%). The pooled MD was 1.90 mg/dL (95% CI: −2.30 to 6.10; I² = 87.49%) (Fig. 5).

Fig. 2.

Forest plot of effect of NLCR on TGs compared to control group, A; Standard mean difference and B; Mean difference

Fig. 3.

Forest plot of effect of NLCR on TC compared to control group, A; Standard mean difference and B; Mean difference

Fig. 4.

Forest plot of effect of NLCR on LDL compared to control group, A; Standard mean difference and B; Mean difference

Fig. 5.

Forest plot of effect of NLCR on HDL compared to control group, A; Standard mean difference and B; Mean difference

Subgroup analyses based on follow-up duration, risk of bias, and study design

Subgroup analyses based on follow-up duration showed no statistically significant differences in TG (SMD = − 0.20; 95% CI: −0.44 to 0.05), LDL-C (SMD = − 0.22; 95% CI: −0.51 to 0.08), TC (SMD = − 0.29; 95% CI: −0.60 to 0.02), or HDL-C levels (SMD = 0.12; 95% CI: −0.38 to 0.62) in studies with follow-up durations of ≤ 3 months. In contrast, analysis using MD demonstrated a significant reduction in TC levels in studies with ≤ 3 months of follow-up (MD = − 9.81 mg/dL; 95% CI: −14.13 to − 5.49), whereas no statistically significant differences were observed for TG, LDL-C, or HDL-C. Among studies with follow-up durations > 3 months, a significant reduction in TG levels was observed in the MD analysis (MD = − 8.02 mg/dL; 95% CI: −13.95 to − 2.09), whereas no statistically significant differences were identified for the remaining lipid parameters. Subgroup analysis according to study quality showed a statistically significant difference for HDL-C in the SMD analysis among studies with low risk of bias (SMD = 0.42; 95% CI: 0.04 to 0.79). However, the corresponding MD estimate did not reach statistical significance (MD = 5.13 mg/dL; 95% CI: −0.51 to 10.77). Furthermore, significant reductions in TG levels were observed in low-risk studies when assessed using MDs (MD = − 7.97 mg/dL; 95% CI: −13.88 to − 2.05), whereas the corresponding SMD estimate did not reach statistical significance. Additionally, subgroup analysis by study design revealed a statistically significant reduction in TG levels among RCTs compared with non-RCTs in both the SMD (SMD = − 0.24; 95% CI: −0.42 to − 0.05) and MD analyses (MD = − 9.25 mg/dL; 95% CI: −14.84 to − 3.65). Similarly, LDL-C levels showed a significant reduction in RCTs when analyzed using MDs (MD = − 6.23 mg/dL; 95% CI: −11.64 to − 0.83), although the corresponding SMD estimate did not reach statistical significance. No significant subgroup differences were observed for the remaining lipid parameters (Table 2).

Table 2.

Subgroup analysis of lipid profile changes based on follow up length, risk of bias, and study type

Variable subgroup	Number of datasets	SMD (95% CI)	Difference between groups	Heterogeneity		MD (95% CI)	Difference between groups	Heterogeneity
				Q	I2 (%)			Q	I2 (%)
TG
Follow up length			0.925				0.964
≤ 3 Months	7	-0.20 (-0.44 to 0.05)		13.34	54.30	-7.71 (-19.87 to 4.45)		10.08	40.38
> 3 Months	4	-0.18 ( -0.44 to 0.08)		5.39	42.79	-8.02 (-13.95 to -2.09) *		3.15	0.00
Risk of bias			0.941				0.663
Low	4	-0.20 (-0.46 to 0.06)		5.65	46.27	-7.97 (-13.88 to -2.05) *		2.77	0.00
Some concerns	5	-0.19 (-0.43 to 0.06)		5.77	27.27	-12.51 (-32.10 to 7.07)		6.75	43.74
Type of study			0.296				0.478
Randomized	7	-0.24 (-0.42 to -0.05) *		7.97	29.70	-9.25 (-14.84 to -3.65) *		6.37	0.00
Non-Randomized	2	0.03 (-0.43 to 0.49)		1.87	46.49	2.96 (-30.32 to 36.23)		1.58	36.54
TC
Follow up length			0.517				0.228
≤ 3 Months	6	-0.29 (-0.60 to 0.02)		10.99	55.05	-9.81 (-14.13 to -5.49) *		5.66	0.00
> 3 Months	4	-0.11 (-0.57 to 0.36)		19.57	81.95	-1.29 (-14.46 to 11.88)		12.66	71.50
Risk of bias			0.980				0.543
Low	3	-0.21 (-0.88 to 0.47)		16.44	83.85	-2.77 (-19.95 to 14.40)		9.81	75.94
Some concerns	5	-0.20 (-0.43 to 0.04)		4.54	13.65	-8.79 (-17.77 to 0.18)		3.73	5.81
Type of study			0.834				0.869
Randomized	6	-0.22 (-0.55 to 0.11)		16.91	65.29	-6.25 (-16.48 to 3.98)		9.98	54.42
Non-Randomized	2	-0.14 (-0.83 to 0.54)		4.06	75.39	-4.14 (-27.13 to 18.85)		3.44	70.90
LDL
Follow up length			0.850				0.696
≤ 3 Months	7	-0.22 (-0.51 to 0.08)		19.36	67.62	-5.52 (-10.77 to -0.27)		9.80	40.06
> 3 Months	4	-0.16 (-0.63 to 0.30)		18.16	81.95	-3.11 (-14.03 to 7.81)		11.37	71.68
Risk of bias			0.809				0.998
Low	4	-0.21 (-0.64 to 0.21)		14.56	78.47	-4.57 (-12.73 to 3.59)		7.49	59.11
Some concerns	5	-0.15 (-0.42 to 0.12)		6.50	37.55	-4.56 (-11.59 to 2.47)		5.88	27.01
Type of study			0.349				0.360
Randomized	7	-0.24 (-0.50 to 0.02)		16.70	62.60	-6.23 (-11.64 to -0.83) *		9.11	39.35
Non-Randomized	2	0.04 (-0.49 to 0.58)		2.51	60.14	1.56 (-14.24 to 17.36)		2.56	60.95
HDL
Follow up length			0.807				0.972
≤ 3 Months	5	0.12 (-0.38 to 0.62)		28.60	84.92	1.20 (-4.19 to 6.59)		39.13	92.25
> 3 Months	3	0.19 (-0.02 to 0.39)		1.34	0.00	1.31 (-1.66 to 4.27)		3.33	32.08
Risk of bias			0.020*				0.024 *
Low	4	0.42 (0.04 to 0.79) *		12.01	72.33	5.13 (-0.51 to 10.77)		33.85	86.42
Some concerns	3	-0.19 (-0.53 to 0.15)		3.28	38.18	-2.22 (-5.21 to 0.76)		2.31	17.34
Type of study			0.335				0.285
Randomized	6	0.18 (-0.21 to 0.58)		27.63	81.57	2.49 (-2.39 to 7.37)		41.08	88.84
Non-Randomized	1	-0.11 (-0.57 to 0.34)		-	-	-1.10 (-5.53 to 3.33)		-	-

SMD Standard Mean Difference, MD Mean Difference, CI Confidence Interval, TG Triglyceride, TC Total cholesterol, LDL Low-density lipoprotein, HDL High-density lipoprotein

*Significant association (p < 0.05) were found

Meta-regression analysis of lipid changes due to NLCR

Meta-regression analyses were conducted to explore potential sources of heterogeneity in lipid outcomes (Table S4). For TG, multivariable analysis based on SMD identified study design as the only significant predictor. RCTs were associated with significantly greater reductions in TG levels compared with non-RCTs (coefficient = − 0.7061; 95% CI: −1.3371 to − 0.0750; p = 0.028). The multivariable model explained 80.11% of the between-study heterogeneity, with residual heterogeneity substantially reduced (I² = 5.28%). In contrast, no study-level variable was significantly associated with TG changes in the MD-based multivariable analysis. Although the MD model explained 38.80% of heterogeneity, none of the examined covariates reached statistical significance. Regarding TC, no study-level variables were significantly associated with TC changes in either the SMD- or MD-based univariable and multivariable analyses. The final SMD model explained 17.58% of the heterogeneity, with substantial residual heterogeneity remaining (I² = 52.76%). Similarly, the MD-based multivariable model did not identify any significant predictors of TC changes despite explaining all observed heterogeneity (R² = 100%; I² = 0%). For LDL-C, study design again emerged as a significant determinant in both SMD- and MD-based multivariable analyses. In the SMD model, RCTs demonstrated significantly greater LDL-C reductions than non-RCTs (coefficient = − 0.8904; 95% CI: −1.5017 to − 0.2791; p = 0.004). Likewise, the MD model showed that RCTs were associated with significantly greater reductions in LDL-C levels (coefficient = − 25.2112; 95% CI: −45.5672 to − 4.8552; p = 0.015). The SMD model accounted for all observed heterogeneity (R² = 100%), with no residual heterogeneity (I² = 0%). In the case of HDL-C, multivariable analysis based on SMD identified publication year as the only significant predictor (coefficient = 0.1299; 95% CI: 0.0026 to 0.2572; p = 0.045), indicating that the estimated HDL-C effect size was larger in more recently published studies. Although study quality was significantly associated with HDL-C changes in the SMD-based univariable analysis (p = 0.026), this association did not persist after adjustment (p = 0.392). Similarly, in the MD-based multivariable analysis, publication year remained the only significant predictor of HDL-C changes (coefficient = 1.6208; 95% CI: 0.3430 to 2.8985; p = 0.013), whereas no other study-level variable reached statistical significance after adjustment. The SMD-based multivariable model explained 73.88% of heterogeneity, with moderate residual heterogeneity (I² = 32.56%), while the MD-based model explained all observed heterogeneity (R² = 100%; I² = 0%).

Publication bias and sensitivity analysis

Sensitivity analyses for the lipid parameters demonstrated that the overall findings remained consistent across all outcomes when each study was sequentially excluded. The pooled estimates for TG, TC, LDL-C, and HDL-C were not substantially influenced by the removal of any single study, indicating the robustness of the results (Figures S1–S4).

Furthermore, Egger’s regression test did not reveal any evidence of publication bias for TG (SMD: bias = -1.16, SE = 1.272, p = 0.361, MD: bias = -0.86, SE = 0.744, p = 0.250), TC (SMD: bias = 0.55, SE = 1.789, p = 0.760, MD: bias = 0.38, SE = 1.003, p = 0.703), LDL-C (SMD: bias = 0.95, SE = 1.577, p = 0.546, MD: bias = 1.86, SE = 0.732, p = 0.11), or HDL-C (SMD: bias = -1.85, SE = 1.809, p = 0.308, MD: bias = 0.22, SE = 1.234, p = 0.961). These findings were visually supported by symmetrical funnel plots for all lipid-related outcomes (Figures S5–S8).

Discussion

The present meta-analysis evaluated the effects of NLCR on lipid profiles in patients with CVDs. Overall, the pooled analysis demonstrated a statistically significant reduction in TG levels, while no statistically significant differences were observed for TC and LDL-C. No statistically significant change was observed for HDL-C. Subgroup and meta-regression analyses further suggested that study design may influence the observed treatment effects. The TG-lowering effect was consistently more pronounced in RCTs, and study design emerged as a significant determinant of both TG and LDL-C changes in meta-regression analyses. These findings suggest that rigorously designed NLCR programs may contribute to improved lipid management through lifestyle modification, enhanced treatment adherence, and structured follow-up. The availability of pooled MD provides additional clinical context for interpreting these findings. NLCR was associated with a reduction of 8.38 mg/dL in TG levels. Although the pooled MD for TC (− 6.0 mg/dL) and LDL-C (− 4.8 mg/dL) were numerically lower in the NLCR group, these differences did not reach statistical significance. The observed TG reduction may still be clinically relevant in secondary prevention settings, where lifestyle-based interventions can provide additive benefits alongside evidence-based pharmacological therapies. Accordingly, NLCR should be viewed as a complementary rather than alternative strategy for cardiovascular risk management. Observed heterogeneity may partly reflect differences in intervention characteristics, delivery modalities, and concurrent lipid-lowering therapies. Subgroup analyses demonstrated more favorable lipid responses among RCTs and studies with lower risk of bias. In many included studies, participants were receiving statins, which may have attenuated the independent effect of NLCR. Furthermore, substantial variation existed in program duration, intensity, educational content, and delivery methods, all of which may have influenced patient adherence and lipid outcomes. Such variability in program design may have influenced patient engagement, adherence to lifestyle recommendations, and ultimately the magnitude of lipid-related outcomes. These differences likely contributed to the moderate to high heterogeneity observed in some pooled estimates. Although HDL-C has traditionally been considered a favorable lipid parameter, contemporary evidence suggests that HDL-C is more appropriately regarded as a cardiovascular risk marker rather than a direct causal therapeutic target [33–35]. The lack of a statistically significant increase in HDL-C may be of limited clinical relevance, given that contemporary evidence increasingly considers HDL-C a marker of cardiovascular risk rather than a direct therapeutic target. Further high-quality RCTs with longer follow-up durations are needed to determine whether these improvements translate into reductions in major cardiovascular events and long-term mortality.

Recent systematic reviews and meta-analyses have also highlighted the benefits of structured cardiac rehabilitation. For example, a meta-analysis of NLCR in patients with coronary artery disease reported significant reductions in TC and LDL-C, along with increases in HDL-C [36]. Similarly, exercise-based cardiac rehabilitation has been associated with improved lipid profiles and cardiovascular outcomes, while participation in rehabilitation programs has been linked to lower cardiovascular mortality and recurrent events [37, 38]. Although these studies evaluated broader rehabilitation models rather than exclusively nurse-led programs, their findings generally support the present results. In our analysis, NLCR was associated with a significant reduction in TG levels, whereas no statistically significant differences were observed for TC, LDL-C, or HDL-C. These findings support the role of NLCR as an adjunctive strategy for cardiovascular risk management through lifestyle counseling and adherence support.

Our findings are, consistent with previous studies, suggesting that NLCR may improve lipid-related cardiovascular risk factors. In a RCT, Premkumar et al. reported a significant reduction in TG levels after three months in patients receiving NLCR compared with usual care [13]. Similarly, NLCR have been associated with improved LDL-C target attainment, blood pressure control, and quality of life [29, 39]. Additional evidence from CABG populations has shown that NLCR programs incorporating education, counseling, and structured follow-up can lead to significant reductions in LDL-C, TC, and TG, with no statistically significant change in HDL-C [40]. Collectively, these studies support the potential role of NLCR in secondary prevention. Although statistically significant effects in the present meta-analysis were observed primarily for TG, nurse-led programs may provide broader benefits through improved adherence, lifestyle modification, patient engagement, and risk-factor monitoring [41].

In the subgroup analyses, most lipid outcomes did not differ significantly according to follow-up duration. However, MD-based analyses demonstrated a significant reduction in TC among studies with follow-up durations of ≤ 3 months and a significant reduction in TG among studies with follow-up durations of > 3 months, although the overall pattern remained inconsistent across lipid parameters. Previous studies have reported inconsistent findings regarding the impact of follow-up duration on cardiovascular risk factors. For example, a nurse-led self-management program with 12 months of follow-up improved health behaviors, LDL-C control, and BMI [42], while Michelsen et al. reported improvements in TC, LDL-C, and systolic blood pressure but no significant differences in HDL-C or TG [14]. Other studies have shown benefits extending beyond lipid control, including improved medication adherence, dietary behaviors, and reduced hospital readmissions [43]. Likewise, Arjunan et al. reported a significant increase in HDL-C, with no significant effect on LDL-C after three months [28]. Overall, these findings suggest that while improvements in TG levels may be observed within relatively short intervention periods, the broader benefits of NLCR likely extend beyond lipid modification and may become more apparent with longer-term follow-up. Given the chronic nature of CVD, sustained nurse-led follow-up may support long-term risk factor control through continuous education and adherence support [44, 45].

This study has several notable strengths. First, the present meta-analysis provides a comprehensive synthesis of available evidence on the effects of NLCR on lipid profiles by including both RCTs and non-RCTs. The inclusion of different study designs allowed for a broader evaluation of NLCR interventions across diverse clinical settings. In addition, subgroup and meta-regression analyses were conducted to explore potential sources of heterogeneity, including study design, follow-up duration, and risk of bias, thereby providing further insight into factors that may influence lipid outcomes. In addition, by reporting both SMD and MD, the present study provides not only a standardized assessment of intervention effects but also clinically interpretable estimates of lipid changes expressed in conventional units (mg/dL). However, several limitations should be acknowledged. First, moderate to substantial heterogeneity was observed across some lipid outcomes, which may reflect differences in patient populations, intervention characteristics, program duration, follow-up periods, and implementation strategies across studies. In addition, concurrent pharmacological treatments, particularly the use of lipid-lowering medications such as statins, may have influenced lipid outcomes and potentially obscured the independent effects of NLCRs. Furthermore, although MD were calculated to improve clinical interpretability, the relatively lipid changes and the variability in baseline lipid levels across studies should be considered when interpreting the clinical significance of pooled estimates. Second, the relatively small sample sizes in several included trials may have limited the statistical power to detect intervention effects. Third, many of the included studies had relatively short follow-up durations, which may not be sufficient to capture long-term changes in lipid metabolism or sustained behavioral modifications associated with rehabilitation programs. Moreover, the number of studies contributing to some meta-regression models was relatively small. Although meta-regression analyses were performed to explore potential sources of heterogeneity, fewer than ten studies were available for several outcomes, which may limit the robustness and interpretability of these exploratory analyses. Additionally, one included study delivered the cardiac rehabilitation intervention entirely through a telemedicine platform, whereas most other studies used in-person or hybrid nurse-led programs. Differences in delivery modality may influence patient engagement, adherence to behavioral recommendations, and the overall effectiveness of rehabilitation programs, potentially contributing to clinical heterogeneity in the pooled estimates. Furthermore, considerable variability existed in the structure of NLCR programs, including differences in intervention duration, intensity, educational components, and delivery modalities (e.g., in-person, telephone-based, or hybrid formats), which may have contributed to the observed heterogeneity in effect estimates. Another limitation is that non-HDL-C, an increasingly recognized marker of atherogenic lipoprotein burden and cardiovascular risk, was not consistently reported across the included studies. Consequently, the effect of NLCR on non-HDL-C could not be evaluated in the present meta-analysis. Another important limitation relates to the lack of standardized reporting of intervention intensity (“dose”) across the included studies. Intervention intensity is a key determinant of effectiveness in cardiac rehabilitation and is increasingly described using structured frameworks such as the FITT principle (Frequency, Intensity, Time, and Type). However, most of the included studies did not provide sufficient detail on these components to allow consistent characterization or comparison of intervention dose. This limitation restricted our ability to explore dose–response relationships or to determine whether variations in intervention intensity contributed to differences in lipid outcomes. The absence of FITT-based reporting highlights an important gap in the current evidence base and limits the reproducibility and comparability of NLCR programs. Future studies should adopt standardized frameworks for describing intervention components to enhance the methodological rigor and clinical applicability of research in this field.

Future research should focus on conducting larger, well-designed multicenter RCTs with longer follow-up periods to better evaluate the long-term impact of NLCR on lipid control, achievement of guideline-recommended lipid targets, and cardiovascular outcomes. Moreover, integrating NLCR with personalized risk management strategies, such as tailoring interventions according to patients’ lipid profiles, comorbidities, and treatment adherence, may further enhance the effectiveness of these programs. The incorporation of emerging technologies, including telemedicine platforms and digital monitoring tools, may also improve patient engagement, facilitate continuous follow-up, and increase accessibility to rehabilitation services.

Conclusion

This meta-analysis found that NLCR was associated with a significant reduction in TG levels, corresponding to a decrease of 8.38 mg/dL. No statistically significant effects were observed for total cholesterol, LDL-C, or HDL-C. These findings suggest that NLCR may contribute to improving selected atherogenic lipid parameters, particularly TGs, as part of comprehensive cardiovascular risk management. Further large-scale studies with standardized intervention protocols and longer follow-up are needed to determine the sustainability and clinical significance of these effects.

Abbreviations

BMI

Body Mass Index

CABG

Coronary Artery Bypass Grafting

CI

Confidence Interval

CVD

Cardiovascular Disease

CVDs

Cardiovascular Diseases

HDL-C

High-Density Lipoprotein Cholesterol

LDL-C

Low-Density Lipoprotein Cholesterol

MD

Mean Difference

MeSH

Medical Subject Headings

NLCR

Nurse-Led Cardiac Rehabilitation

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

PROSPERO

International Prospective Register of Systematic Reviews

RCT

Randomized Controlled Trial

REML

Restricted Maximum Likelihood

RoB 2.0

Risk of Bias 2 Tool

ROBINS-I

Risk Of Bias In Non-randomized Studies of Interventions

SD

Standard Deviation

SMD

Standardized Mean Difference

TC

Total Cholesterol

TG

Triglyceride

TGs

Triglycerides

VLDL

Very Low-Density Lipoprotein

Authors’ contributions

Y.H. and K.P. was responsible for the initial conception and design of the study. The literature search and data extraction were conducted jointly by Y.H. and A.A.N. They also performed the data analysis and interpretation, under the academic supervision and oversight of A.A.N. The manuscript was written and critically revised through a collaborative effort among all authors. All authors (Y.H., A.A.N., and K.P.) provided substantial contributions to the final version of the manuscript through their feedback and revisions, and all approved the final submitted version.

Funding

None.

Data availability

All relevant data underpinning the findings of this study are thoroughly documented in the main manuscript and accompanying supplementary materials.

Declarations

Ethics approval and consent to participate

Ethical approval and informed consent were not required for this study, given its design and nature.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.