Reply to: Comment on revisiting the association between anticholinergic burden and frailty in people with HIV

We thank Dr Wang for the careful appraisal of our work and for the constructive comments. ¹ , ²

First, we acknowledge that the cross‐sectional design precludes causal inference and raises the possibility of confounding by indication. Anticholinergic medications are frequently prescribed for conditions that may themselves contribute to frailty. ³ In our analysis, we adjusted for multimorbidity and relevant clinical covariates, and we performed perturbation and bootstrap analyses, which showed consistent associations between anticholinergic burden (ACB) and frailty across models. Nevertheless, we agree that residual confounding, particularly related to symptom severity, cannot be fully excluded.

At the same time, several elements argue against confounding by indication as the sole explanation of our findings, including the consistency of the associations across two complementary frailty constructs (Fried and Rockwood), the dose–response relationship with ACB, and the robustness of estimates across multiple sensitivity analyses. Furthermore, anticholinergic exposure was also independently associated with the Rockwood index, which is less directly driven by symptom‐level components, further supporting the robustness of the observed relationship. While interaction analyses do not establish causality, the presence of biologically coherent effect modification by age and prior AIDS argues against a purely uniform confounding mechanism and supports the plausibility of a differential vulnerability to anticholinergic exposure. We still agree that approaches such as propensity score methods or analyses accounting more directly for symptom severity would provide additional insight and represent important directions for future longitudinal studies.

Second, we appreciate the comment regarding the interaction between ACB and prior AIDS. We interpreted this finding as reflecting increased vulnerability in individuals with reduced physiological and neurocognitive reserve. ⁴ , ⁵ While we agree that cumulative exposure to neurotoxic agents may contribute to vulnerability, we would emphasize that ACB captures a specific pharmacological construct, the muscarinic receptor–mediated effects, which is not directly reflected by antiretroviral neurotoxicity. We would also note that greater historical exposure to neurotoxic antiretrovirals cannot be assumed to be specific to participants with AIDS. Many people with HIV without a history of AIDS were treated during the same eras and were likewise exposed to agents such as efavirenz. In addition, participants with prior AIDS in our cohort represent a clinically heterogeneous group, including both individuals with long‐standing HIV infection and historical exposure to earlier regimens, as well as more recent late presenters. This variability in exposure, both in terms of antiretrovirals and their duration, makes it less likely that a uniform, unmeasured neuropharmacological burden could fully explain the observed interaction. Therefore, although current different ACB scales do not incorporate antiretroviral drugs, contain different drugs, with different correlations with the presence of anticholinergic symptoms, ⁶ this is unlikely to fully account for the observed interaction. Rather, we believe our findings support the presence of effect modification in a biologically vulnerable subgroup, though further studies are warranted to better characterize this mechanism, considering also the prevalence of the ACB in ageing people with HIV. ⁷

Lastly, our intention was not to advocate indiscriminate deprescribing, but rather to support structured medication review, individualized risk–benefit assessment and, where appropriate, substitution with alternative agents with lower ACB rather than simple discontinuation. ACB should be considered within a comprehensive clinical framework, and any deprescribing strategy should be carefully monitored to avoid unintended negative effects on symptom control. In this context, our findings should be viewed as a rationale for prospective and interventional studies.

The association between ACB and frailty is likely multifactorial and potentially bidirectional. However, the consistency of the findings across models supports the robustness of the observed relationship. We identified age and prior AIDS as effect modifiers of the association between ACB and frailty. In this context, the identification of such higher‐risk subgroups, particularly those with a history of AIDS, in whom increased susceptibility may be less apparent than with ageing alone, has clinical relevance. Our findings support a more tailored approach to medication review in people with HIV while underscoring the need for longitudinal studies and refined measurement tools to better define causality and guide interventions.