Genetic causal association between heart failure, frailty and poisoning by narcotics and psychodysleptics: A two‐sample Mendelian randomization

Abstract

Aims

Observational studies have suggested associations between heart failure (HF), frailty and poisoning by narcotics and psychodysleptics (PNP). However, establishing causal relationships has been challenging. This study used a two‐sample Mendelian randomization (MR) approach to investigate the genetically proxied causality between HF, frailty and PNP.

Methods

Summary‐level data from genome‐wide association studies (GWAS) were utilized to investigate the causal relationship between frailty index (FI) and PNP risk, PNP and HF risk, as well as the bidirectional relationship between FI and HF. Various MR methods, including inverse‐variance weighted (IVW), MR‐Egger, weighted median and weighted mode, were employed. Horizontal pleiotropy, heterogeneities and the robustness of genetic variants were assessed using MR‐Egger intercept tests, Cochran's Q test and leave‐one‐out analyses. The MR‐PRESSO outlier test was applied to identify and remove outlier variants to mitigate potential pleiotropy.

Results

Significant genetic causal associations were observed between FI and HF (IVW: OR = 1.42, 95% CI: 1.16–1.74) and between HF and FI (IVW: OR = 1.09, 95% CI: 1.05–1.14). However, no causal relationships were found between other variables. Sensitivity analyses demonstrated no evidence of horizontal pleiotropy or heterogeneity, confirming the robustness of the results.

Conclusions

This MR study provides genetic evidence of a bidirectional causal relationship between FI and HF, highlighting the intertwined nature of frailty and heart failure. No genetically proxied causal associations were observed between FI and PNP or between PNP and HF. Further research, including age‐stratified and longitudinal studies, is needed to validate these findings and explore the underlying mechanisms.

What is already known about this subject

• Previous observational studies showed a link between heart failure (HF), frailty and poisoning by narcotics and psychodysleptics (PNP), suggesting a possible association.

• This study aimed to provide clarity by investigating genetic causality using a two‐sample Mendelian randomization approach.

What this study adds

• This study found no genetic causal association between frailty (FI) and poisoning by narcotics and psychodysleptics (PNP) risk, and PNP and heart failure (HF) risk.

• The results indicate the need for further research to understand the complex interplay between genetic factors, frailty, PNP and HF.

1. INTRODUCTION

Heart failure (HF) is a global public health issue, especially common in the elderly population. The prevalence of HF in individuals aged 35 and above is 1.38%, rising to 7.55% in those aged 80 and above. ¹ , ² , ³ The main causes of HF include coronary artery disease, hypertension and dilated cardiomyopathy. Clinically, HF patients may present symptoms such as dyspnoea, ankle oedema and fatigue, ⁴ as well as signs such as elevated jugular venous pressure, pulmonary rales and peripheral oedema. ⁵ , ⁶ Understanding its pathogenesis is crucial for collecting optimal evidence for targeted therapy and effectively guiding clinical decision‐making.

Frailty is a common geriatric syndrome characterized by multisystem damage and decreased resistance to external stressors. ⁷ , ⁸ The prevalence of frailty in the population aged 60 and above is approximately 7.0%, and it is more common in the elderly, women and those with lower education levels. ⁹ , ¹⁰ Frailty becomes more prevalent with increasing age, but the onset of frailty in the elderly is not constant. Some elderly individuals remain at relatively low levels of frailty until late in life, while others become frail at an earlier age. ¹¹ Frailty, as a multifactorial syndrome, is characterized by clinical features such as fatigue, weakness and decreased exercise tolerance, ¹² which significantly disturb patients' daily lives and work. Its development is associated with various biopsychosocial factors. ¹³ Frailty may increase the risk of diseases such as HF by affecting patients' immune system function, metabolic activity and cardiovascular health. ¹⁴ Furthermore, frailty is linked to increased susceptibility to medication use and drug‐related adverse effects, including poisoning by narcotics and psychodysleptics (PNP). Elderly individuals with frailty often experience reduced drug metabolism and excretion, resulting in prolonged drug activity and heightened toxicity risk. These physiological vulnerabilities make frailty a critical factor in understanding PNP risk.

PNP is a preventable drug‐related disease often associated with drug abuse or misuse. ¹⁵ These drugs are widely used in clinical settings for sedation, anaesthesia and depression treatment, but long‐term or high‐dose use may result in poisoning. Clinical manifestations of PNP vary, including confusion, drowsiness and ataxia, with severe cases potentially causing respiratory depression, arrhythmia and life‐threatening complications. ¹⁶ Such poisoning may increase the risk of cardiovascular diseases such as heart failure by directly impairing cardiac and circulatory system function. The intersection between frailty and PNP risk lies in shared mechanisms such as impaired organ function, decreased mobility and altered cognitive states, which exacerbate the effects of drug misuse. This relationship underscores the biological plausibility of considering the frailty index (FI) as an exposure for PNP in Mendelian randomization (MR) analyses. Certain drugs, like digoxin, may improve haemodynamics in heart failure patients by blocking Na⁺‐K⁺‐ATPase. ¹⁷ However, causal association is unclear between FI, PNP and HF.

Mendelian randomization is a more robust approach to establish causal relationships, addressing the limitations of observational studies. ¹⁸ MR utilizes genetic variations strongly linked to exposure as instrumental variables, effectively minimizing the influence of confounding factors and reverse causation. ¹⁹ The extensive identification of numerous variants associated with complex exposures through genome‐wide association studies (GWAS) has greatly enhanced the application of MR in research. ²⁰ In this study, we employed a two‐sample MR method to investigate the reciprocal causality between FI, PNP and HF.

2. METHODS

2.1. Study design overview

This study investigates the relationships between FI, PNP and HF using a bidirectional two‐sample MR approach. All statistics involved in the analysis were derived from publicly available GWAS. Single‐nucleotide polymorphisms (SNPs) associated with FI, HF and PNP were extracted as instrumental variables (IVs). Based on the summary‐level data from GWAS of FI, HF and PNP, we conducted a two‐sample MR analysis. MR operates under three critical assumptions: (a) the selected genetic IVs have a strong association with the exposure; (b) these genetic IVs are independent of any confounders that might influence both the exposure and the outcome; and (c) the effect of the genetic IVs on the outcome is exclusively mediated through the exposure, with no alternative biological pathways. The flowchart of the study is shown in Figure 1.

FIGURE 1.

Flowchart of the study.

2.2. Data sources

Atkins et al. ⁷ reported the most comprehensive exploration of genetic influences on the frailty index so far by performing a genome‐wide association study (GWAS) meta‐analysis of the frailty index data in European descent UK Biobank participants ( n  = 164 610, 60–70 years old) and Swedish Twin Gene participants ( n  = 10 616, 41–87 years old). FinnGen is a large public‐private partnership aiming to collect and analyse genome and health data from 500 000 Finnish biobank participants. PNP summary statistics were obtained from the FinnGen study, which involved 16 380 433 European ancestry individuals (270 cases and 213 568 controls). The GWAS summary statistics of genetic associations for HF were extracted from the largest GWAS comprising 47 309 cases and 930 014 controls of European ancestry across 26 studies from the Heart Failure Molecular Epidemiology for Therapeutic Targets (HERMES) Consortium. ²¹ These datasets were used to evaluate the relationships between FI and PNP, PNP and HF, and also FI and HF, applying a consistent methodological approach. Full summary statistics for the FI, PNP and HF GWAS are available for download from the publicly available GWAS catalogue website (https://gwas.mrcieu.ac.uk).

2.3. Genetic instruments selection and harmonization

In this study, the IVs selected must meet the following criteria: (1) Initially, SNPs that demonstrate a P‐value < 5 × 10⁻⁸ (FI). Since there were no SNPs with P‐value less than 5 × 10–8 for PNP, we broadened the threshold to P‐value < 5 × 10⁻⁶ (PNP) to select eligible instrumental variables. (2) SNPs with a minimum minor allele frequency (MAF) > 0.01 undergo screening. (3) SNP linkage disequilibrium (LD) effects are eliminated by applying the criterion R ² < 0.001 within a window size of 10 000 kb. ²² (4) If an IV is not present in the summary data for the outcome, alternative SNPs with high LD (R ² > 0.8) to that IV are sought and used as substitutes. (5) The strength of each SNP in the IV is evaluated by calculating its F‐value to exclude any potential bias rising from weak instruments between the IV and exposure factor using this formula: F = R ² × (N − 2)/(1 − R ²), where R ² represents the proportion of exposure variation explained by SNPs in the IV. The requirement for an acceptable F‐value is set at >10. ²³

2.4. Statistical analyses

We employed MR analyses utilizing the random‐effects inverse‐variance weighted (IVW) ²⁴ method as our primary approach. ²⁰ This approach offers a robust estimation even in the absence of directional pleiotropy. Moreover, supplementary analyses were performed using weighted median, simple mode, weight mode and MR‐Egger methods. ²⁵ , ²⁶ The presence of significant heterogeneity among the individual SNPs was evaluated using the Cochran Q test, where significant values indicated notable heterogeneity. The leave‐one‐out analysis was used to detect the robustness and consistency of the results. ²⁷ Horizontal pleiotropy was identified by conducting the intercept test of MR Egger and MR pleiotropy residual sum and outlier (MR‐PRESSO), with a P‐value greater than 0.05 suggesting the absence of such pleiotropy. ²⁸ MR‐PRESSO analysis was carried out to identify outliers within the MR analysis. ²⁹ All statistical analyses were carried out using TwoSampleMR packages in R.

3. RESULTS

3.1. Selection of instrumental variables

In this study, MR analyses were conducted using FI, PNP and HF as exposures and outcomes. A total of 15 SNPs were selected for FI, 7 SNPs for PNP, and 12 SNPs for HF. FI‐associated SNPs (e.g., rs374943348) were not available in the summary statistics of the PNP outcome, and two PNP‐associated SNPs (rs149192666 and rs3171722) were unavailable in the HF outcome data. The detailed information on instrumental variables utilized in the Mendelian randomization analyses is provided. The F‐statistics for all IV‐exposure associations were much greater than 10, demonstrating a low likelihood of weak instrument bias. The R ² values for each analysis indicated appropriate instrument strength (Supplementary Table S1 and Supplementary Table S2).

3.2. Causal effect analyses

The MR analysis revealed no statistically significant genetic causal association between FI and PNP risk. The IVW method yielded an odds ratio (OR) of 2.04 (95% CI: 0.22–18.65, P = 0.53), with consistent results observed across MR‐Egger, weighted median and weighted mode methods (Table 1). Similarly, MR analysis showed no statistically significant genetic causal association between PNP and FI risk. The scatter plot and forest plots for the causal effect of FI on PNP risk is shown in Figures 2A,B and 3A,B.

TABLE 1.

Association between genetic predictions of causal risk for poisoning by narcotics and psychodysleptics, frailty index and heart failure.

Exposure	Outcome	SNPs (n)	Methods	OR (95% CI)	P
Heart failure	Frailty index	10	Inverse variance weighted	1.09 (1.05–1.14)	2.45E‐05
			MR Egger	1.13 (1.01–1.27)	0.06
			Weighted median	1.08 (1.02–1.15)	0.007
			Weighted mode	1.06 (0.96–1.16)	0.26
Narcotics	Frailty index	7	Inverse variance weighted	0.997 (0.99–1.01)	0.49
			MR Egger	1.00 (0.97–1.04)	0.8
			Weighted median	0.99 (0.99–1.00)	0.3
			Weighted mode	0.99 (0.97–1.01)	0.35
Frailty index	Heart failure	14	Inverse variance weighted	1.42 (1.16–1.74)	0.0006
			MR Egger	1.57 (0.62–4.02)	0.36
			Weighted median	1.37 (1.03–1.83)	0.03
			Weighted mode	1.28 (0.82–1.98)	0.29
Poisoning by narcotics and psychodysleptics	Heart failure	5	IVW	1.02 (1–1.05)	0.11
			MR Egger	1.02 (0.89–1.17)	0.81
			Weighted median	1.01 (0.99–1.04)	0.26
			Weighted mode	0.99 (0.95–1.04)	0.82
Frailty index	Poisoning by narcotics and psychodysleptics	14	Inverse variance weighted	2.04 (0.22–18.65)	0.53
			MR Egger	11.84 (0–278974.13)	0.64
			Weighted median	0.91 (0.05–18.15)	0.95
			Weighted mode	0.78 (0.01–60.44)	0.91
Heart failure	Poisoning by narcotics and psychodysleptics	10	Inverse variance weighted	0.70 (0.14–3.57)	0.67
			MR Egger	0.52 (0.003–84.96)	0.81
			Weighted median	1.21 (0.24–6.17)	0.81
				4.61 (0.53–40.26)	0.2
Heart failure	Poisoning by narcotics and psychodysleptics (the outlier rs17617337 was excluded)	9	Inverse variance weighted	1.12 (0.27–4.63)	0.88
			MR Egger	0.34 (0.01–22.75)	0.63
			Weighted median	1.96 (0.40–9.56)	0.41
			Weighted mode	7.83 (0.80–76.56)	0.11

Abbreviation: MR, Mendelian randomization.

FIGURE 2.

The scatter plots of the association. A. Causal effect of FI on PNP risk. B. Causal effect of PNP on FI risk. C. Causal effect of PNP on HF risk. D. Causal effect of HF on PNP risk. E. Causal effect of FI on HF risk. F. Causal effect of HF on FI risk. G. Causal effect of HF on PNP risk (the outlier rs17617337 was excluded). PNP, poisoning by narcotics and psychodysleptics; FI, frailty index; HF, heart failure.

FIGURE 3.

Forest plots analysis. A. Causal effect of FI on PNP risk. B. Causal effect of PNP on FI risk. C. Causal effect of PNP on HF risk. D. Causal effect of HF on PNP risk. E. Causal effect of FI on HF risk. F. Causal effect of HF on FI risk. G. Causal effect of HF on PNP risk (the outlier rs17617337 was excluded). PNP, poisoning by narcotics and psychodysleptics; FI, frailty index; HF, heart failure.

No significant association was observed between PNP and HF risk. The IVW method produced an OR of 1.02 (95% CI: 1.00–1.05, P = 0.11), with supplementary MR methods confirming this result (Table 1, Figure 2C, Figure 3C). Similarly, reverse MR analysis of the risk of HF and PNP did not find a genetic causal association (OR_IVW = 0.70: 0.14–3.57, P = 0.67) (Table 1, Figure 2D, Figure 3D).

The additional analysis assessing the direct genetic association between FI and HF revealed a significant causal relationship. The IVW method estimated an OR of 1.42 (95% CI: 1.16–1.74, P = 0.0006). This result was supported by the weighted median method (OR = 1.37, 95% CI: 1.03–1.83, P = 0.03), while the MR‐Egger and weighted mode methods yielded non‐significant results. The scatter plot and forest plots for the causal effect of FI on HF is shown in Figures 2E and 3E. In the reverse analysis, HF was associated with an increased risk of FI. The IVW method estimated an OR of 1.09 (95% CI: 1.05–1.14, P = 2.45 × 10⁻⁵), with consistent results across MR‐Egger, weighted median and weighted mode methods. The scatter plot and forest plots for the causal effect of HF on FI is shown in Figures 2F and 3F.

3.3. Pleiotropy, heterogeneity and sensitivity analysis

Heterogeneity analyses based on Cochran's Q test indicated no significant heterogeneity in most causal relationships (all P > 0.05), except for HF on PNP (Q statistic = 21.004, P = 0.013; Table 2). No evidence of horizontal pleiotropy was observed, as all MR‐Egger intercept P‐values were greater than 0.05. The MR‐PRESSO outlier test identified one outlier SNP (rs17617337) in the HF‐PNP analysis. After removing the outlier, no significant change in the causal estimates was observed. The distortion test P‐value was 0.141, indicating the robustness of the results after adjustment (Table 3). Leave‐one‐out analyses demonstrated that no single SNP disproportionately influenced the causal estimates for any of the tested relationships. Forest plots and funnel plots illustrating these findings are shown in Supplementary Figures S1‐S2.

TABLE 2.

Results of heterogeneity and pleiotropy for instrumental variables.

Exposure	Outcome	Heterogeneity		Pleiotropy
		Q statistic (IVW)	P‐value	MR‐Egger intercept	P‐value
Heart failure	Frailty index	8.977	0.439	−0.003	0.478
Poisoning by narcotics and psychodysleptics	Frailty index	9.97	0.126	−0.004	0.662
Frailty index	Heart failure	10.572	0.647	−0.002	0.834
Poisoning by narcotics and psychodysleptics	Heart failure	7.977	0.092	0.001	0.986
Frailty index	Poisoning by narcotics and psychodysleptics	11.173	0.596	−0.04	0.732
Heart failure	Poisoning by narcotics and psychodysleptics	21.004	0.013	0.02	0.907
Heart failure	Poisoning by narcotics and psychodysleptics (the outlier rs17617337 was excluded)	12.98	0.11	0.08	0.57

Abbreviation: IVW, inverse variance weighted; MR, Mendelian randomization.

TABLE 3.

MR‐PRESSO test results.

Exposure	Outcome	Raw		Outlier corrected		Global P	Number of outliers	Distortion P
		OR (95% CI)	P	OR ( 95% CI)	P
Heart failure	Frailty index	1.1175 (1.0671–1.1703)	0.001	NA		0.131	0	NA
Poisoning by narcotics and psychodysleptics	Frailty index	0.997 (0.99–1.00)	0.514	NA		0.137	0	NA
Frailty index	Heart failure	1.42 (1.19–1.71)	0.002	NA		0.679	0	NA
Poisoning by narcotics and psychodysleptics	Heart failure	1.02 (1–1.05)	0.189	NA	NA	0.171	0	NA
Frailty index	Poisoning by narcotics and psychodysleptics	2.04 (0.26–15.87)	0.508	NA	NA	0.6	0
Heart failure	Poisoning by narcotics and psychodysleptics	0.79 (0.20–3.01)	0.731	1.16 (0.37–3.63)	0.81	0.04	1	0.141
Heart failure	Poisoning by narcotics and psychodysleptics (The outlier rs17617337 was excluded)	1.16 (0.37–3.63)	0.81	NA		0.26	0	NA

4. DISCUSSION

Our findings provide important insights into the relationships between HF, FI and PNP. Although no genetically proxied causal relationships were identified between PNP and HF or between FI and PNP, we observed significant bidirectional genetically proxied causal associations between FI and HF. These results highlight the need for a nuanced understanding of the interplay among these conditions, with implications for clinical practice and future research.

The association between frailty and PNP is one of the highly discussed topics in the field of medicine. ³⁰ With the increasing ageing population and drug abuse issues, particularly among older adults, research on this association has become increasingly important. Frailty, as a comprehensive indicator of physical function and health status, is closely related to PNP. ³¹ , ³² Due to physiological changes and decreased ability to metabolize and excrete drugs, elderly patients are more susceptible to the effects of medications, thereby increasing their risk of poisoning. Anaesthetics and tranquillizers are widely used in anaesthesia, pain management and depression treatment for older adults; however, long‐term or excessive use may lead to adverse reactions including toxicity. ³³ While our study did not identify a genetically proxied causal relationship between FI and PNP, this does not exclude the possibility of a complex bidirectional relationship influenced by external factors.

Mechanistically, PNP may contribute to frailty by impairing nutritional status, reducing mobility or causing cognitive decline. ³⁴ Conversely, frailty, characterized by reduced physiological reserves and organ function, may predispose individuals to drug‐related toxicities. These interactions suggest a multifactorial relationship rather than a singular cause–effect pathway. Furthermore, age is a significant determinant of frailty, and exploring age‐specific associations may provide additional insights. However, due to the lack of age‐stratified GWAS data for FI, such analyses could not be performed in this study. This limitation underscores the need for future studies incorporating age‐specific data to clarify the impact of age on the frailty–PNP relationship.

Our findings support a bidirectional genetically proxied causal relationship between FI and HF, consistent with previous research suggesting shared pathophysiological mechanisms. Frailty may predispose individuals to HF through systemic inflammation, impaired cardiovascular function and reduced resilience to stressors. Conversely, HF may contribute to frailty via mechanisms such as reduced mobility, chronic fatigue and muscle atrophy, highlighting the intertwined nature of these conditions. ³⁵ Clinically, these findings emphasize the importance of addressing frailty as part of HF management and vice versa. Early identification of frailty in HF patients may enable targeted interventions, including physical rehabilitation and nutritional support, to improve outcomes. Similarly, optimizing HF treatment might mitigate the development or progression of frailty. Future research should focus on identifying specific pathways linking these conditions to develop more effective prevention and treatment strategies.

Although HF patients are known to have reduced drug metabolism and excretion capabilities, making them more susceptible to PNP, our study did not find a genetically proxied causal relationship between these traits. This may reflect the influence of external factors such as comorbidities, treatment regimens, or healthcare practices that were not accounted for in the genetic analysis. Additionally, the possibility that PNP is a consequence rather than a cause of HF cannot be excluded.

The MR‐PRESSO analysis identified and adjusted for an outlier SNP in the HF‐PNP analysis, confirming the robustness of the findings. Nevertheless, given the clinical relevance of PNP in HF patients, healthcare professionals should remain vigilant in monitoring and managing medication use in this population, particularly for those with impaired renal or hepatic function. ³⁶ Enhanced education on medication safety and regular assessments of drug efficacy and toxicity are critical for minimizing the risk of PNP in HF patients. ³⁷

This study has several strengths. By using genetic variants as proxies for exposures, MR provides robust evidence for causal inference, minimizing confounding and reverse causation. The use of large, publicly available GWAS datasets enhances the statistical power and reliability of our findings. Additionally, sensitivity analyses confirmed the robustness of the results, with no evidence of horizontal pleiotropy.

However, there are limitations that should be acknowledged. Firstly, the study focused on participants of European ancestry, which may limit the generalizability of the findings to other populations. Secondly, due to the use of GWAS summary statistics, subgroup analyses based on individual characteristics such as age, gender or ethnicity were not possible. These constraints may restrict the generalizability of our findings. Future research should address this by incorporating diverse populations and age‐stratified analyses. Finally, the genetic instruments used in MR reflect lifelong average effects, which may not fully capture the dynamic nature of frailty or its interactions with other conditions such as PNP and HF. Further research using alternative approaches, such as longitudinal studies or individual‐level data, is warranted to validate and extend these findings.

5. CONCLUSION

In conclusion, this MR study provides genetic evidence for a bidirectional causal relationship between FI and HF, highlighting the close interplay between frailty and heart failure. However, no genetically proxied causal relationships were observed between FI and PNP or between PNP and HF. These findings underscore the complexity of the relationships among these conditions and suggest that external factors, such as age, comorbidities or treatment regimens, may play a significant role. Further research, including age‐stratified and longitudinal studies, is needed to validate these findings and to explore the underlying mechanisms in more diverse populations.

AUTHOR CONTRIBUTIONS

BW carried out the studies, participated in collecting data and drafted the manuscript. HY and XX performed the statistical analysis and participated in its design. BW and MC participated in acquisition, analysis or interpretation of data and drafted the manuscript. All authors read and approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no competing interests.

Supporting information

Figure S1 The funnel plots. A. Causal effect of FI on PNP risk. B. Causal effect of PNP on FI risk. C. Causal effect of PNP on HF risk. D. Causal effect of HF on PNP risk. E. Causal effect of FI on HF risk. F. Causal effect of HF on FI risk. G. Causal effect of HF on PNP risk (the outlier rs17617337 was excluded). PNP, poisoning by narcotics and psychodysleptics; FI, frailty index; HF, heart failure.

Figure S2 Leave‐one‐out sensitivity analysis. A. Causal effect of FI on PNP risk. B. Causal effect of PNP on FI risk. C. Causal effect of PNP on HF risk. D. Causal effect of HF on PNP risk. E. Causal effect of FI on HF risk. F. Causal effect of HF on FI risk. G. Causal effect of HF on PNP risk (the outlier rs17617337 was excluded). PNP, poisoning by narcotics and psychodysleptics; FI, frailty index; HF, heart failure.

Table S1. Detailed information of instrumental variables utilized in the Mendelian randomization analysis.

Table S2. Instrumental variable extraction procedure used in the Mendelian randomization analysis.

Wang B, Yu H, Cai M, Xie X. Genetic causal association between heart failure, frailty and poisoning by narcotics and psychodysleptics: A two‐sample Mendelian randomization. Br J Clin Pharmacol. 2025;91(9):2696‐2704. doi: 10.1002/bcp.70091

DATA AVAILABILITY STATEMENT

All data generated or analysed during this study are included in this published article and its supplementary information files.