Identifying risk factors for adverse drug reactions in HIV/AIDS patients using LASSO regression modeling Authors: Shiqiong Huang, Liang Gong, Hui Qi, Guoqiang Zhou, Ji Sun Published: 2026 ABSTRACT ObjectiveWe perform a real-world study to investigate the incidence, clinical manifestations, suspected drugs, and associated risk factors of adverse drug reactions (ADRs) among Acquired Immunodeficiency Syndrome (AIDS) patients in the Chinese population.Methods918 patients were enrolled in this retrospective study. All suspected ADRs were assessed for causality using the World Health Organization-Uppsala Monitoring Centre system. We employed least absolute shrinkage and selection operator (LASSO) regression to identify risk factors for ADRs in patients with HIV/AIDS.Results210 of 918 patients experienced ADRs, with an incidence rate of 22.88%. Subsequently, standardized classification of ADRs was conducted using the Medical Dictionary for Regulatory Activities (MedDRA), revealing that blood and lymphatic system disorders were the most predominant manifestation, accounting for 5.77%. As to the type of drugs, Antibacterial (26.34%) and antiviral drugs (24.55%) were the most frequent causes for ADRs. Through the LASSO regression, the number of medications administered during hospitalization (odds ratio [OR] = 1.64, 95% confidence interval [CI]: 1.25–2.19) was identified as a significant influencing factor of ADRs in HIV patients.ConclusionThis study analyzed the specific manifestations of ADRs and their suspected drugs in hospitalized AIDS patients. More importantly, it revealed the potential correlation between the occurrence of ADRs and the number of medications administered during hospitalization in those patients. These findings provide a basis for clinical risk stratification management and offer reference data for optimizing treatment regimens. Introduction Adverse Drug Reactions (ADRs) emerged as significant factors contributing to the increasing hospitalization rates and disease burden in both developed and developing nations, posing substantial challenges to global healthcare systems [1]. Acquired Immunodeficiency Syndrome (AIDS), a chronic infectious disease caused by the Human Immunodeficiency Virus (HIV), represents a major public health threat worldwide due to its rapid transmission, prolonged latency period, and high mortality rate [2]. According to the latest data from UNAIDS, as of 2023, there were approximately 39.9 million people suffering from HIV globally, with 1.3 million new infections and 630,000 AIDS-related deaths annually. Antiretroviral Therapy (ART), as the cornerstone strategy for HIV/AIDS, has significantly prolonged patient survival and improved life quality for AIDS patients. However, the associated ADRs emerged as critical factors affecting treatment adherence and therapeutic efficacy [3]. Although extensive international researches focused on the mechanisms, types, and influencing factors of ADRs associated with antiretroviral drugs [4, 5], studies on the Chinese HIV/AIDS patient population beared few fruits, mainly focusing on the hepatotoxicity induced by antiviral medications [6, 7]. In recent years, with the continuous advancements in anti-HIV drug development, vast progression was seen in the clinical application of novel antiretroviral agents, and the optimization of therapeutic strategies against the incidence and severity of ADRs [8]. Nevertheless, due to the complex nature of HIV/AIDS, patients often require combination therapy involving multiple drugs, including antiretroviral agents, antimicrobials, and symptomatic treatments for comorbidities, which further increases the risk and complexity of ADRs. Therefore, this study aims to reveal the occurrence of ADRs and their potential risk factors among hospitalized HIV/AIDS patients based on a real-world study, providing scientific evidence for optimizing HIV treatment strategies and enhancing therapeutic safety and efficacy. Methods Study population and ethics approval This study was a retrospective observational analysis, involving a total of 918 patients with AIDS who were discharged from our hospital between January 1, 2022 and June 30, 2022. All cases met the diagnostic criteria established in the “Chinese Guidelines for HIV/AIDS Diagnosis and Treatment”. Based on the occurrence of ADRs during treatment, the patients were categorized into ADRs group and ADRs-free group. This research protocol has been approved by the Ethics Committee of the First Hospital of Changsha (Approval No. KXKX-2020048). Given the retrospective nature of the study, the Ethics Committee granted an exemption from the requirement for informed consent. To safeguard patient privacy, all collected data were encoded to achieve anonymization. Following the completion of data analysis, all identifiable personal information will be permanently deleted to further ensure the privacy and data security of study participants. Data collection and ADR assessment This study systematically collected epidemiological and clinical data from HIV/AIDS patients, including information such as hospitalization duration, gender, age, underlying diseases, CD4 cell counts, and history of drug allergies. References should be made to our previously published studies for the evaluation of all suspected ADRs [9]. Two clinical pharmacists and the attending physician jointly assessed the occurrence of ADRs in HIV patients based on established evaluation criteria. ADRs are systematically classified and described in accordance with the Medical Dictionary for Regulatory Activities (MedDRA), with corresponding categorization applied to the suspected medicinal products implicated in the occurrence of these reactions. The World Health Organization-Uppsala Monitoring Centre (WHO-UMC) system was employed to assess the causal relationship of suspected ADRs. The associations were classified into six levels: certain, probable, possible, unlikely, conditional, and unassessable. Only the cases falling into the categories of certain, probable, and possible were included. The severity was determined according to the ICH E2A guidelines [10], which included death, life-threatening situations, requirement for hospitalization/extension of hospitalization, significant disability/functional impairment, congenital malformation/birth defects, and requirement for intervention to prevent serious consequences. The clinical outcome indicators were death, cure, improvement, recovery with sequelae, non-cure, and unknown. Clinical cure was defined as the disappearance of symptoms or the return of abnormal indicators to normal. Statistical analysis Statistical analysis was conducted using SPSS 25. Categorical variables were expressed as frequencies, while continuous variables were presented as mean ± standard deviation (x ± SD). In this study, the glmnet package in R 4.4.2 was employed for variable selection in Least Absolute Shrinkage and Selection Operator (LASSO) regression. The data were stratified and randomly partitioned into a training set (n = 642) and a test set (n = 275) at a ratio of 7:3 based on the outcome variable (ADR), with a random seed set to 123 (caret: createDataPartition). After inspection, no missing values were found in the data. Five continuous variables were standardized based on the mean and standard deviation of the training set, and the same parameters were applied to the test set. LASSO regression employs ten-fold cross-validation to select the optimal penalty parameter λ, while simultaneously examining lambda.min and lambda.1se. Given that lambda.min (retaining 8 variables) does not result in a significant performance improvement, the more parsimonious lambda.1se (retaining 3 variables) is ultimately chosen as the final model parameter. It should be noted that in the original process, Synthetic Minority Oversampling Technique (SMOTE) was placed before standardization. To avoid the influence of different scales on the neighbor synthesis based on Euclidean distance, we adjusted the order as follows: first standardize the training set, and then perform SMOTE. When evaluating the predictive performance of the model in the test set, the binomial deviance was calculated. Due to the complete separation of the variable “allergies” in the test set, we adopted Firth penalized logistic regression to estimate the odds ratios (OR), their 95% confidence intervals, and p-values for each variable to ensure the stability and interpretability of the estimates. Results Characteristics of patients We enrolled a total of 918 patients and divided them into two cohorts based on the occurrence of ADRs (the ADRs group and the ADRs-free group). Comparative analysis revealed that the ADRs group exhibited a greater number of patients taking anti-tuberculosis drugs (P < 0.0001) compared with the control group. Furthermore, the ADRs group demonstrated a significantly prolonged length of stay (21.61 ± 15.36 vs. 16.73 ± 22.11, P = 0.003). The ADRs group also showed significantly lower CD4 count levels (196.35 ± 231.34 vs. 246.68 ± 226.17, P = 0.005), a higher number of medications administered during hospitalization (28.95 ± 13.38 vs. 20.50 ± 11.82, P<0.0001) and a shorter duration of HIV infection (1.90 ± 2.77 vs. 2.50 ± 3.34, P = 0.018) compared with the control group. No statistically significant differences were observed between the two groups in terms of mean age, gender, or history of drug allergies (Table 1). Table 1Characteristics of patient between those with ADRs and ADRs-free groupCharacteristicADRs Group(n = 210)ADRs-free Group(n = 708) P Age (years)47.45 ± 14.7548.26 ± 14.730.50Male173 (82.38%)580 (63.18%)0.88History of drug allergies1 (0.48%)16 (2.26%)0.14Underlying disease83 (39.52%)353 (49.86)0.317Taking anti-tuberculosis drugs93 (44.29%)200 (28.25%)<0.0001Length of stay (days)21.61 ± 15.3616.73 ± 22.110.003CD4 counts (cells/µL)196.35 ± 231.34246.68 ± 226.170.005The number of medications administered during hospitalization28.95 ± 13.3820.50 ± 11.82<0.0001Duration of HIV infection (years)1.90 ± 2.772.50 ± 3.340.018 Table 1Characteristics of patient between those with ADRs and ADRs-free groupCharacteristicADRs Group(n = 210)ADRs-free Group(n = 708) P Age (years)47.45 ± 14.7548.26 ± 14.730.50Male173 (82.38%)580 (63.18%)0.88History of drug allergies1 (0.48%)16 (2.26%)0.14Underlying disease83 (39.52%)353 (49.86)0.317Taking anti-tuberculosis drugs93 (44.29%)200 (28.25%)<0.0001Length of stay (days)21.61 ± 15.3616.73 ± 22.110.003CD4 counts (cells/µL)196.35 ± 231.34246.68 ± 226.170.005The number of medications administered during hospitalization28.95 ± 13.3820.50 ± 11.82<0.0001Duration of HIV infection (years)1.90 ± 2.772.50 ± 3.340.018 Characteristics of patient between those with ADRs and ADRs-free group Risk factors for ADRs in AIDS patients by LASSO regression LASSO regression is employed for variable selection, and the optimal penalty parameter λ is determined through tenfold cross-validation. (Fig. 1). The two dashed lines in Fig. 1 correspond to λmin and λ1se: λmin represents the λ value at which the model deviance is minimized, whereas λ1se denotes the λ value that yields the most parsimonious model with the fewest variables within one standard error of λ-min. We further investigated the associations between ADRs risks in HIV/AIDS patients and multiple factors, including age, sex, history of drug allergy, length of hospital stay, underlying comorbidities, concurrent use of anti-tuberculosis drugs, CD4 count, the number of medications administered during hospitalization, and duration of HIV infection. Fig. 1LASSO regression and ten-fold cross-validation for predictor screening. (A) Coefficient path diagram: This diagram illustrates the trajectory of regression coefficients for nine candidate variables as a function of the penalty parameter log(λ). The horizontal axis represents log(λ), while the vertical axis denotes the variable coefficients. (Legend1: History of drug allergies; 2: The number of medications administered during hospitalization; 3: Underlying disease; 4: Taking anti-tuberculosis drugs; 5: Male; 6: Length of stay; 7: Duration of HIV infection (years); 8: Age (years); 9: CD4 counts). (B) 10-Fold Cross-Validation Plot: Illustrating the variation of model binomial deviance with respect to log(λ). Two vertical dashed lines indicate the optimal log(λ) value (corresponding to minimum model deviance) and the log(λ) associated with the most parsimonious model within one standard error. Red scatter points represent binomial deviance, with error bars denoting confidence intervals LASSO regression and ten-fold cross-validation for predictor screening. (A) Coefficient path diagram: This diagram illustrates the trajectory of regression coefficients for nine candidate variables as a function of the penalty parameter log(λ). The horizontal axis represents log(λ), while the vertical axis denotes the variable coefficients. (Legend1: History of drug allergies; 2: The number of medications administered during hospitalization; 3: Underlying disease; 4: Taking anti-tuberculosis drugs; 5: Male; 6: Length of stay; 7: Duration of HIV infection (years); 8: Age (years); 9: CD4 counts). (B) 10-Fold Cross-Validation Plot: Illustrating the variation of model binomial deviance with respect to log(λ). Two vertical dashed lines indicate the optimal log(λ) value (corresponding to minimum model deviance) and the log(λ) associated with the most parsimonious model within one standard error. Red scatter points represent binomial deviance, with error bars denoting confidence intervals Through LASSO regression, we found that the number of medications administered during hospitalization (odds ratio [OR] = 1.64, 95% confidence interval [CI]: 1.25–2.19) was an important factor influencing the occurrence of ADRs in HIV patients. The estimated coefficient of the number of medications administered during hospitalization was 0.4959 (with bootstrap standard error). The logarithmic loss of this model was 0.616, and the area under the curve (AUC) was 0.692 (95% CI 0.621–0.763). The sensitivity and specificity reached 0.672 and 0.692, respectively. Characteristics of ADRs Among the 918 patients, 210 cases experienced ADRs (with an incidence rate of 22.88%). Since the same patient could have multiple ADRs, a total of 259 adverse events occurred in these 210 patients. The specific manifestations of these ADRs and their corresponding system organ class were shown in Table 2. Blood and lymphatic system disorders (5.77%) were the most frequently observed ADRs, followed by skin and subcutaneous tissue disorders (4.79%), investigations (4.14%), hepatic and hepatobiliary disorders (3.27%), and gastrointestinal disorders (3.05%) (Table 2). Table 2Involved organs and systems of ADRs for the AIDS patientsInvolved organs and systemsIncidence of ADR (%)Blood and lymphatic system disorders53 (5.77%) Myelosuppression37 (4.03%) Thrombocytopenia11(1.20%) Granulocytopenia3 (0.33%) Red blood cell abnormality2 (0.22%)Skin and subcutaneous tissue disorders44 (4.79%) Rash39 (4.25%) Pruritus4 (0.44%) Erythema1 (0.11%)Investigations38 (4.14%) White blood cell count decreased18 (1.96%) Blood creatinine increased14 (1.53%) Myocardial necrosis marker increased4 (0.44%) Protein urine present1 (0.11%) Blood creatine phosphokinase MB abnormal1 (0.11%)Hepatic and hepatobiliary disorders30 (3.27%) Hepatic function abnormal29 (3.16%) Hypertransaminasaemia1 (0.11%)Gastrointestinal disorders28 (3.05%) Dyspepsia7 (0.76%) Nausea and Vomiting6 (0.65%) Anorexia7 (0.76%) Retching4 (0.44%) Lip ulceration3 (0.33%) Gastritis1 (0.11%)Nervous system disorders15 (1.64%) Paralysis8 (0.87%) Dizziness6 (0.65%) Neuritis1 (0.11%)Musculoskeletal and connective tissue disorders13 (1.42%) Muscular weakness7 (0.76%) Osteoporosis4 (0.44%) Joint swelling1 (0.11%) Myalgia1 (0.11%)General disorders and administration site conditions12 (1.31%) Malaise10 (1.09%) Fatigue2 (0.22%)Metabolism and nutrition disorders11 (1.20%) Hyperuricaemia9 (0.98%) Hyperlipasaemia1 (0.11%) Hypokalaemic syndrome1 (0.11%)Psychiatric disorders6 (0.65%) Delusional perception2 (0.22%) Insomnia2 (0.22%) Mental disorder1 (0.11%) Abnormal dreams1 (0.11%)Surgical and medical procedures6 (0.65%) Anaphylaxis treatment6 (0.65%)Reproductive system and breast disorders1 (0.11%) Gynaecomastia1 (0.11%)Eye disorders1 (0.11%) Blindness1 (0.11%)Respiratory, thoracic and mediastinal disorders1 (0.11%) Dyspnoea1 (0.11%)Total259Data are n (%)∗The denominator for calculating the percentages in the figure is the total of 918 inpatients Table 2Involved organs and systems of ADRs for the AIDS patientsInvolved organs and systemsIncidence of ADR (%)Blood and lymphatic system disorders53 (5.77%) Myelosuppression37 (4.03%) Thrombocytopenia11(1.20%) Granulocytopenia3 (0.33%) Red blood cell abnormality2 (0.22%)Skin and subcutaneous tissue disorders44 (4.79%) Rash39 (4.25%) Pruritus4 (0.44%) Erythema1 (0.11%)Investigations38 (4.14%) White blood cell count decreased18 (1.96%) Blood creatinine increased14 (1.53%) Myocardial necrosis marker increased4 (0.44%) Protein urine present1 (0.11%) Blood creatine phosphokinase MB abnormal1 (0.11%)Hepatic and hepatobiliary disorders30 (3.27%) Hepatic function abnormal29 (3.16%) Hypertransaminasaemia1 (0.11%)Gastrointestinal disorders28 (3.05%) Dyspepsia7 (0.76%) Nausea and Vomiting6 (0.65%) Anorexia7 (0.76%) Retching4 (0.44%) Lip ulceration3 (0.33%) Gastritis1 (0.11%)Nervous system disorders15 (1.64%) Paralysis8 (0.87%) Dizziness6 (0.65%) Neuritis1 (0.11%)Musculoskeletal and connective tissue disorders13 (1.42%) Muscular weakness7 (0.76%) Osteoporosis4 (0.44%) Joint swelling1 (0.11%) Myalgia1 (0.11%)General disorders and administration site conditions12 (1.31%) Malaise10 (1.09%) Fatigue2 (0.22%)Metabolism and nutrition disorders11 (1.20%) Hyperuricaemia9 (0.98%) Hyperlipasaemia1 (0.11%) Hypokalaemic syndrome1 (0.11%)Psychiatric disorders6 (0.65%) Delusional perception2 (0.22%) Insomnia2 (0.22%) Mental disorder1 (0.11%) Abnormal dreams1 (0.11%)Surgical and medical procedures6 (0.65%) Anaphylaxis treatment6 (0.65%)Reproductive system and breast disorders1 (0.11%) Gynaecomastia1 (0.11%)Eye disorders1 (0.11%) Blindness1 (0.11%)Respiratory, thoracic and mediastinal disorders1 (0.11%) Dyspnoea1 (0.11%)Total259Data are n (%)∗The denominator for calculating the percentages in the figure is the total of 918 inpatients Involved organs and systems of ADRs for the AIDS patients Data are n (%) ∗The denominator for calculating the percentages in the figure is the total of 918 inpatients Concurrently, our analysis of suspected drugs associated with ADRs revealed that the majority were antibacterial (26.34%), antiviral drugs (24.55%), anti-tuberculosis drugs (23.66%), and anti-tumor drugs (15.63%) (Table 3). The incidence of severe ADRs caused by antibacterial was the highest, approximately 7.14%, while 92.41% of ADRs resulted in clinical improvement. Table 3Suspicious drugs, casualty assessment and prognosis of ADRs for the AIDS patientsSuspicious DrugsADRsCasualty AssessmentPrognosisAll ADRsSADRsCertainProbablePossibleCureImprovementUnknownDeathAntibacterial59 (26.34)16 (7.14)2 (0.89)38 (16.96)19 (8.48)3 (1.34)54 (24.11)1 (0.45)1 (0.45)Antiviral drugs55 (24.55)14 (6.25)1 (0.45)28 (12.50)26 (11.61)0 (0.00)50 (22.32)5 (2.23)0Anti-tuberculosis drugs53 (23.66)13 (5.80)0 (0.00)31 (13.84)22 (9.82)1 (0.45)50 (22.32)2 (0.89)0Anti-tumor drugs35 (15.63)12 (5.36)8 (3.57)24 (10.71)3 (1.34)0 (0.00)33 (14.73)2 (0.89)0Other drugs15 (6.70)6 (2.68)0 (0.00)1 (0.45)14 (6.25)2 (0.89)13 (5.80)00Drugs unknown7 (3.12)2 (0.89)0 (0.00)7 (3.13)0 (0.00)0 (0.00)7 (3.13)00Total2246311129846207101Data are n (%)Frequency was calculated as number/224*100%. (A single ADR may be associated with multiple drugs, and a single patient may experience multiple ADRs. Therefore, a total of 224 drugs is suspected) Table 3Suspicious drugs, casualty assessment and prognosis of ADRs for the AIDS patientsSuspicious DrugsADRsCasualty AssessmentPrognosisAll ADRsSADRsCertainProbablePossibleCureImprovementUnknownDeathAntibacterial59 (26.34)16 (7.14)2 (0.89)38 (16.96)19 (8.48)3 (1.34)54 (24.11)1 (0.45)1 (0.45)Antiviral drugs55 (24.55)14 (6.25)1 (0.45)28 (12.50)26 (11.61)0 (0.00)50 (22.32)5 (2.23)0Anti-tuberculosis drugs53 (23.66)13 (5.80)0 (0.00)31 (13.84)22 (9.82)1 (0.45)50 (22.32)2 (0.89)0Anti-tumor drugs35 (15.63)12 (5.36)8 (3.57)24 (10.71)3 (1.34)0 (0.00)33 (14.73)2 (0.89)0Other drugs15 (6.70)6 (2.68)0 (0.00)1 (0.45)14 (6.25)2 (0.89)13 (5.80)00Drugs unknown7 (3.12)2 (0.89)0 (0.00)7 (3.13)0 (0.00)0 (0.00)7 (3.13)00Total2246311129846207101Data are n (%)Frequency was calculated as number/224*100%. (A single ADR may be associated with multiple drugs, and a single patient may experience multiple ADRs. Therefore, a total of 224 drugs is suspected) Suspicious drugs, casualty assessment and prognosis of ADRs for the AIDS patients Data are n (%) Frequency was calculated as number/224*100%. (A single ADR may be associated with multiple drugs, and a single patient may experience multiple ADRs. Therefore, a total of 224 drugs is suspected) Discussion At present, there is a substantial of researches investigating the adverse effects of antiretroviral drugs on HIV/AIDS patients, with a particular focus on the types and mechanisms of ADRs, as well as the variations in ADRs induced by different drug combinations [4–6, 11]. However, comprehensive studies examining the actual incidence of ADRs across the entire HIV/AIDS patient population remain relatively scarce. Particularly within the specific context of the HIV/AIDS patient population in China, relevant researches are even more limited. A previous study demonstrated that HIV-positive patients, due to the immune system compromised by the virus, exhibited a higher probability of ADRs [12]. This finding underscores the necessity for heightened vigilance regarding ADRs in HIV/AIDS patients undergoing pharmacological treatment. Consequently, our research mainly focuses on the manifestations of ADRs in this patient in the real world, as well as the risk factors for the occurrence of ADRs. ART, as the pivotal treatment modality for controlling HIV progression, prolonging patient survival and enhancing quality of life, exhibited distinct characteristics of ADRs across different regions and countries [13, 14]. In developing nations, the incidence of ADRs associated with antiretroviral medications ranged from 11% to 35.9% [15, 16]. However, the results of this study indicated that among 918 patients, a total of 210 cases (22.88%) experienced ADRs. These 210 patients were cumulatively associated with 224 kinds of suspected drugs, among which antiviral drugs accounted for 24.55%, while the proportion directly attributable to antiretroviral drugs was only 4.14%. This study primarily focused on the occurrence of ADRs during patients’ hospitalization periods. Due to the limited observation window, some patients might experience related ADRs prior to admission or post-discharge, which were not captured in this investigation. This limitation may potentially lead to an underestimation of the incidence rate of ADRs associated with antiviral drug therapy. Considering the significant impact of ADRs on clinical outcomes and treatment adherence in HIV patients, this study employed the LASSO regression to precisely identify influencing factors for ADR occurrence. This approach effectively addressed potential multicollinearity among multiple variables, while simultaneously performing variable selection and regression coefficient estimation under control of known clinical risk factors [17, 18]. Model constraints and variable selection were conducted through 10-fold cross-validation, wherein the LASSO model automatically excluded variables with weak power. The optimized final model revealed that, among numerous potential factors, the number of medications administered during hospitalization was retained as a risk factor, indicating its close association with the occurrence of ADRs in HIV patients. A substantial body of literature indicated a correlation between increased medication utilization and the incidence of ADRs [19–21]. Numerous studies further demonstrated that polypharmacy significantly elevated the risk of severe ADRs in patients. Specifically, for each additional drug administered, the probability of experiencing adverse reactions increased by a factor of 1.08 [21]. This study further analyzed the drug categories associated with ADRs in HIV patients, revealing that antibacterial, antiviral, and anti-tuberculosis agents were the predominant contributors. This pattern was likely attributable to the heightened incidence of opportunistic infections in individuals with HIV, which often necessitated combination drug therapies and consequently elevated the potential risk of ADR occurrence [4]. It should be noted that concomitant administration of highly active antiretroviral therapy (HAART) with anti-tuberculosis regimens might induce pharmacokinetic interactions that interfered with the normal metabolic processes of anti-tuberculosis drugs. This could result in reduced metabolic clearance rates, leading to increased systemic exposure and elevated risks of ADRs [22, 23]. Literature reports indicated that both HIV infection and ART were independent risk factors for ADR-related hospitalizations [24]. In addition, some literature reported that in HIV-infected individuals, polypharmacy, drug interactions, and ADRs were significant factors influencing health-related quality of life and increasing the treatment burden [25]. Therefore, it is particularly crucial to implement systematic medication management for such groups of patients. Researches indicated that pharmacist-led Medication Therapy Management (MTM) and Multidisciplinary Team (MDT) collaboration could reduce medication-related problems and the incidence of ADRs, thereby improving the quality of life of patients [26, 27]. It is recommended to identify and address drug-related issues through MTM and, at the same time, utilize prescription simplification strategies to optimize the medication regimen, thereby reducing the risk of ADRs and improving the treatment efficacy. This study has several limitations: the patients were from specific regional medical institutions, resulting in limited representativeness; the observation was limited to the hospitalization period, lacking long-term follow-up; the retrospective design may be affected by incomplete medical records, and it is impossible to distinguish the reverse causality between the quantity of drugs and ADRs. To address the aforementioned limitations, we will conduct multicenter studies, extend the observation period, and adopt a prospective cohort study design to further validate the causal relationship between increased medication use and the occurrence of ADRs. Conclusion In this study, we performed a retrospective analysis to systematically investigate the relevant influencing factors of ADRs in AIDS patients, successfully identifying their risk factors. Furthermore, the study delineated the common types of ADRs and the associated suspected drugs during patient treatment, suggesting that clinicians should pay heightened attention to potential ADRs risks when prescribing these medications, and strengthen medication monitoring and evaluation.