Characterization of risk for dexmedetomidine withdrawal in the intensive care unit

Introduction

Background

Dexmedetomidine is a centrally acting alpha-2 agonist, which decreases sympathetic tone by inhibiting norepinephrine release from presynaptic neurons in the periphery and attenuates central nervous system excitation centrally leading to sedative effects without depressing respiratory drive (1-5). Use as a sedative is approved by the Food and Drug Administration (FDA) for up to 24 h, with some post-marketing evidence suggesting safety and efficacy of continuous infusion dexmedetomidine for up to 5 days (4,6,7). Use in clinical settings often extends beyond the FDA-approved duration. Current guidance from the Society of Critical Care Medicine recommends the use of dexmedetomidine over benzodiazepines in critically ill, mechanically ventilated adults, citing primarily moderate benefits in time to light sedation and extubation when compared with benzodiazepines (8).

Abrupt discontinuation of dexmedetomidine may result in resurgence of norepinephrine binding to sympathetic nervous system receptors resulting in withdrawal symptoms. Dexmedetomidine withdrawal comprises symptoms consistent with sympathetic hyperactivity both centrally and peripherally including confusion, agitation, tachycardia, hypertension, nausea, diaphoresis, and tremors (9). While previous literature has identified withdrawal symptoms in pediatric patients receiving dexmedetomidine infusions for sedation, data exploring the identification of dexmedetomidine withdrawal in adult critically ill patients is more limited (2).

Clonidine, a centrally acting alpha-2 agonist with enteral formulation options, structurally similar to dexmedetomidine, has been used to successfully reduce withdrawal risk from dexmedetomidine due to the similar mechanism of action and longer half-life (10,11). Clonidine does have lesser affinity for the α2 receptor compared to dexmedetomidine (200:1 vs. 1,600:1) requiring more aggressive dosing (5,12). Most studies have described utilizing clonidine as a transitional tool when weaning patients from dexmedetomidine infusions, although variations in dosing strategies exist in the literature (10,11).

Rationale and knowledge gap

Although dexmedetomidine withdrawal symptoms are well defined in the literature, there remains a lack of consensus regarding which patients are at the highest risk of withdrawal after discontinuing a dexmedetomidine infusion (2,3). Bouajram and colleagues concluded withdrawal may be more common in patients receiving a high dose and/or duration of dexmedetomidine for over 72 h (9). These findings were not confirmed by Pathan and colleagues who reported no correlation between withdrawal incidence and dexmedetomidine dose, duration, or titration (2). Specific patient characteristics have not been clearly examined as potential factors for risk of withdrawal. Identification of risk factors for dexmedetomidine withdrawal may help proactively reduce the impacts of withdrawal on relevant clinical outcomes. Patients experiencing withdrawal symptoms may be more likely to have a prolonged intensive care unit (ICU) stay which can lead to worsened clinical outcomes.

Objective

The primary objective of our analysis was to identify patient and medication-related risk factors associated with the development of dexmedetomidine withdrawal. We present this article in accordance with the STROBE reporting checklist (available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-24-146/rc).

Methods

Study design and participants

This retrospective cohort study was conducted in a mixed medical/surgical ICU at a 350-bed community teaching hospital. The electronic medical record (Epic Systems, Verona, WI, USA) was utilized for identification of potential cases for the analysis.

Patients were eligible for screening if they were 18 years or older, admitted to the critical care service, and received dexmedetomidine continuous infusion for at least 12 consecutive hours between March 1, 2020 and July 31, 2023. Patients were excluded from the analysis if they were discharged from the critical care service or expired within 24 h after dexmedetomidine discontinuation, had a medical history of cirrhosis, or were initiated on dexmedetomidine for treatment of alcohol withdrawal.

Baseline patient characteristics collected included age, sex, creatinine clearance, and the acute physiology and chronic health evaluation II (APACHE-II) score (13). Data were also collected regarding baseline presence of atrial fibrillation or other tachyarrhythmias, hypertension, and generalized anxiety disorder, as well as baseline opioid, benzodiazepine or beta blocker use. Discontinuation of continuous opioid or benzodiazepine infusion within 24 h of dexmedetomidine discontinuation was also collected as discontinuation of these agents could have resulted in potential withdrawal from these agents. Data points were chosen based on the potential influence on dexmedetomidine withdrawal scoring. For example, patients with atrial fibrillation may be more prone to tachyarrhythmias, which is one of the criteria of dexmedetomidine withdrawal in this study. Opioid withdrawal has similar characteristics to dexmedetomidine withdrawal and could be considered a confounding factor. Details on clonidine dosing strategies were collected for any patients that received clonidine within 24 h of dexmedetomidine discontinuation for descriptive purposes and to assess for any potential consistency in tapering strategies.

Outcome measures

Patients were evaluated for withdrawal during a period that ranged from 24 h prior to dexmedetomidine discontinuation until 24 h after. Dexmedetomidine was considered discontinued when the infusion was stopped for more than 12 h and not re-initiated within 24 h of initial cessation. Dexmedetomidine was discontinued based on patient presentation and clinician discretion as no institutional policy exists regarding discontinuation of dexmedetomidine. The 48-h evaluation period was divided into 6-h time blocks to inform the assessment of withdrawal in order to encompass two half-lives of dexmedetomidine per evaluation and to include two assessments per 12-h nursing shift. Cardiovascular and neurological variables were analyzed, including blood pressure, heart rate, Richmond Agitation-Sedation Scale (RASS), and confusion assessment method for the intensive care unit (CAM-ICU) (14,15). Measures of central tendency for continuous data in each time block of pertinent factors were used to characterize withdrawal variables. Patients were considered positive for withdrawal if they had at least two signs of withdrawal during a single assessment upon chart review for the 24 h following dexmedetomidine discontinuation. Withdrawal signs included blood pressure greater than 160/90 mmHg, heart rate greater than 110 beats per minute (bpm), RASS score of +1 or higher, and positive CAM-ICU score (14,15). Presence of a positive CAM-ICU score or RASS score of +1 or higher at any point during the 6-h block was considered a withdrawal sign (14,15). Patients were divided into two groups, withdrawal present or absent, and were then compared to analyze secondary outcomes.

The primary outcome of this study was the incidence of dexmedetomidine withdrawal within 24 h after infusion cessation. Secondary outcomes included hospital and ICU length of stay, days of intubation, clonidine dosing strategies utilized for prevention or treatment of dexmedetomidine withdrawal, and incidence of new antihypertensive agent(s) prescribed at discharge. Beginning and ending dosing regimens, total daily dose, and total duration of clonidine administration was collected for any patients receiving clonidine in the timeframe of 24 h before or after dexmedetomidine discontinuation. Logistic regression was performed to identify characteristics associated with the development of withdrawal.

Statistical analysis

Patients eligible for screening were randomized electronically and screened until the end of the study timeframe. Descriptive statistics were analyzed using IBM SPSS Statistics software version 22 (IBM Inc., Armonk, NY, USA). Normality of data was determined by histogram assessment using SPSS software. Student’s t-test was used for normally distributed continuous data and Wilcoxon Rank Sum test was used for non-normally distributed continuous data. Chi-square or Fischer’s exact tests were used for nominal data. Statistical significance is defined as a P value of <0.05. Logistic regression was completed using R Statistical Software version 4.1.2 (Vienna, Austria). A forward and backward variable selection method, informed by the Akaike information criteria (AIC), was used to construct the logistic regression model. All baseline patient and dexmedetomidine infusion data points collected were considered candidate variables for inclusion into the final model.

Any missing data was imputed with patient-specific values of central tendency using surrounding data (e.g., a patient with a positive CAM-ICU score in one assessment was considered positive for the entirety of that 12-h shift). If no patient-specific value was available for a given assessment block, a neutral value was imputed (i.e., a missing variable in an APACHE-II score calculation was given a value of zero).

Ethical considerations

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional review board of Trinity Health Grand Rapids and Muskegon (Mercy Health Regional IRB—No. 23-1114-3) and individual consent for this retrospective analysis was waived.

Results

A total of 272 patients were screened (Figure 1) for study inclusion. Ninety-eight patients were included in the study, with 45 in the withdrawal positive group (46%) and 53 in the withdrawal negative group.

Figure 1 Patient screening and inclusion.

Baseline demographics

Baseline patient characteristics are outlined in Table 1. All baseline characteristics between the withdrawal positive and withdrawal negative groups were generally balanced, highlighting a median APACHE-II score of 26 noted in each group (13). The most common admission diagnosis in both groups was infectious, followed by neurologic. There was no statistically significant difference in the number of patients that had discontinuation of a continuous opioid infusion within the 24-h period after dexmedetomidine discontinuation (22% patients in the withdrawal positive group compared to 28% in withdrawal negative group; P=0.49).

Dexmedetomidine infusion characteristics

Infusion characteristics for dexmedetomidine are listed in Table 2. There were no significant differences regarding peak infusion rate (1.2 mcg/kg/h in the withdrawal positive group vs. 1.0 mcg/kg/h in the withdrawal negative group; P=0.09), infusion duration (52 vs. 49 h respectively; P=0.32) or time from peak rate to infusion discontinuation (12.5 vs. 12 h respectively; P=0.92). Notably, the range of the infusion duration in the withdrawal positive group extended past 5 days compared to less than 3 days in the withdrawal negative group (128 and 70 h, respectively, see Table 2). There were no differences in total cumulative dose for patients experiencing withdrawal, which averaged 2,562 mcg [interquartile range (IQR), 1,380–8,635 mcg], while patients not experiencing withdrawal averaged 2,092 mcg (IQR, 1,215–4,262 mcg); P=0.15.

Withdrawal symptoms

A comprehensive description of withdrawal symptoms noted in each group is presented in Table 3. Heart rate, RASS, and CAM-ICU are all noted to be statistically higher/more prevalent in the withdrawal positive group compared to the withdrawal negative group (14,15).

Secondary outcomes

A complete description of secondary outcomes is available in Table 4. ICU length of stay was significantly longer in patients who experience dexmedetomidine withdrawal [12 (IQR, 9–20) vs. 9 (IQR, 5–17) days, P=0.04]. Patients positive for withdrawal had a median duration of 7 days (IQR, 4–13 days) on mechanical ventilation, which was 2 days longer compared to the withdrawal negative group (median: 5 days, IQR, 2–12 days; P=0.07).

Logistic regression analysis

Following forward and backward AIC informed variable selection, the final model predicting dexmedetomidine withdrawal included five variables: atrial fibrillation, hypertension, opioid use, generalized anxiety disorder, and dexmedetomidine infusion duration (Table 5). Atrial fibrillation was determined to be the only independent risk factor for developing withdrawal [odds ratio (OR) =6.0; 95% confidence interval (CI): 1.49–31.3].

Clonidine characteristics

Table 6 details characteristics of clonidine used for either preventing or treating dexmedetomidine withdrawal. Patients who did not experience dexmedetomidine withdrawal averaged a higher total daily dosage, cumulative dosage, and longer duration of clonidine therapy. Additional information about patient-specific clonidine regimens can be found in Table S1.

Discussion

The estimated incidence of dexmedetomidine withdrawal based on previous analyses is approximately 30–64% (2,9). Withdrawal occurred in 46% of the patients we analyzed, similar to those previous reports, and was identified by a set of cardiovascular and neurologic symptoms. Prior studies have focused on the differentiation of dexmedetomidine withdrawal symptoms from comorbid conditions or impacts of being in an ICU (2,9). This study used the most commonly reported symptoms of increased heart rate and blood pressure, delirium presence, and positive agitation-sedation scores to identify withdrawal, based on the findings of previous analyses (2,9).

Through logistic regression analysis, we were able to evaluate several biologically plausible factors that may contribute to the development of dexmedetomidine withdrawal. These findings reveal that baseline atrial fibrillation was an independent risk factor for the development of dexmedetomidine withdrawal. Additionally, dexmedetomidine infusion duration and baseline opioid use were noted to be influential factors in the development of dexmedetomidine withdrawal based on the logistic regression results. Despite dexmedetomidine duration not being significantly different between the two groups of this study, the range of the withdrawal positive group extends over 2 days longer than the withdrawal negative group.

Atrial fibrillation was identified as an independent predictor of withdrawal which may have resulted in a higher incidence of heart rate symptoms related to withdrawal between the two groups. A study published by Pöyhiä and colleagues describes dexmedetomidine depressing sinus and AV nodal functions in patients undergoing atrial fibrillation ablation and an increasing AV nodal refractoriness (16). This increase in refractoriness may increase the likelihood of triggering atrial fibrillation when dexmedetomidine is discontinued, potentially causing more patients to trigger the heart rate criteria for withdrawal. As noted in Table 3, a significantly larger portion of patients in the withdrawal positive group had heart rate criteria met (42.2% vs. 17.0%, P=0.01). Future studies on a larger scale are needed to take a deeper look into the relationship between atrial fibrillation and risk of dexmedetomidine withdrawal.

Consistent with the duration seen previously, dexmedetomidine duration in our study was roughly equal between withdrawal groups (16). However, Pathan et al. did not find a correlation between duration and incidence of withdrawal (2). Logistic regression in our study identified dexmedetomidine duration as an impactful predictor of withdrawal despite the similar total duration in each patient group. The impact of dexmedetomidine duration on risk of withdrawal appears to be patient specific and larger future studies may be able to explore this area more. While our study is not able to define the specific duration of concern, limiting length of exposure to dexmedetomidine may be a viable intervention to reduce risk of developing withdrawal. Providers should assess patients on dexmedetomidine daily to assess duration of infusion in an attempt to limit a risk factor for dexmedetomidine withdrawal. Future studies would benefit from trying to identify the optimal duration and mechanism of tapering dexmedetomidine as well as patient characteristics and dexmedetomidine infusion characteristics that may increase risk of withdrawal.

The 2018 Pain, Agitation, Delirium, Immobility, and Sleep Disruption (PADIS) guidelines recommend dexmedetomidine use over benzodiazepines or ketamine for continuous sedation due to potential improvements in time to light sedation and extubation when compared with benzodiazepines, which has contributed to the increased utilization of dexmedetomidine in adult critically ill patients (8). These guidelines also recommend use of propofol over benzodiazepines (8). However, limited available evidence prevents formal guidance on choosing a preferred sedative agent between propofol and dexmedetomidine for mechanically ventilated patients (8). Without this clear guidance, benefits of dexmedetomidine, such as the absence of respiratory depression, compared to propofol, must be thoughtfully considered in light of the potential risks of withdrawal from dexmedetomidine to make the appropriate person-centered agent selection. This study showed an increase in ICU length of stay in patients positive for withdrawal, which has been associated with increased mortality, healthcare costs, and longer recovery post discharge (8,17). Daily assessments of sedative choice and likelihood of extubation should be completed for all patients in the ICU on mechanical ventilation.

Several studies have been published utilizing clonidine, an alpha-2 agonist with enteral options, as a transitional tool to blunt sympathetic hyperactivity when weaning patients from dexmedetomidine infusions, although variation in dosing strategies exist in the literature (10,18,19). Glaess et al. found dosing strategies ranging from 0.1 to 0.3 mg enterally every 6–8 h (10). Gagnon et al. utilized 0.3 mg enterally every 6 h and successfully weaned 75% of patients from dexmedetomidine within 48 h of initiating clonidine (19). In this study, clonidine was only utilized in 12% of patients, with 25% of these patients receiving clonidine after dexmedetomidine infusion was stopped. A wide variety of dosing strategies were noted in this study, which is indicative of the patient-specific nature of withdrawal and responsiveness to supplementation of alpha-2 agonists as well as the lack of institutional guidance for this indication. For withdrawal positive patients, clonidine dosing ranged from 0.1 to 0.2 mg every 6 to 24 h. In the withdrawal negative group, clonidine ranged from 0.1 to 0.2 mg every 6 to 12 h. Clonidine dosing strategies used in this trial are significantly lower than previous literature, which may have contributed to the incidence of dexmedetomidine withdrawal. For patients on dexmedetomidine infusions for longer than 48 h continuously, a clonidine taper may be beneficial to prevent withdrawal symptoms from occurring and treat any symptoms that may arise without needing to increase the dexmedetomidine rate, although this study was not able to narrow down a certain dosing regimen or duration that is most impactful. Institutions should consider a standardized approach to use of clonidine or other similar agents for dexmedetomidine tapers to ensure optimized dosing and a higher chance of success with dexmedetomidine discontinuation.

Limitations of the study include its retrospective nature, small sample size, and single center study location as documentation may be incomplete and confounding factors impact the results. Secondary outcomes were purely exploratory and should be interpreted as such given the risks posed by the potential for multiple comparisons to lead to discovery of statistically significant findings. Data on a number of potential confounding factors were collected and assessed lapses in documentation were also recorded. Using patient-specific averages or marking individual assessment markers negative for withdrawal in place of missing data points helped to not inflate the data. The small sample size of the study may have resulted in an underpowered analysis for some variables. Differences in incidence of certain variables between the two groups may affect the results of the logistic regression, including the result of hypertension and generalized anxiety disorder at baseline being protective against withdrawal, which does not appear to have a rational physiological basis. Although this study showed that duration of dexmedetomidine infusion is a factor associated with withdrawal, we were not able to determine the threshold for concern in relation to duration of dexmedetomidine therapy, as other studies have done (9). Appropriate use of dexmedetomidine in both groups would limit the ability to detect an influence of infusion duration on withdrawal development, another potential explanation for why dose and duration of infusion were not identified as independent risk factors for withdrawal. As with all studies containing ICU patients, a myriad of concomitant factors play into the overall clinical picture. Other comorbidities or concomitant medications that could influence withdrawal symptoms may have been missed. While withdrawal symptoms of tachycardia and hypertension have been well-defined in literature, the optimal definition of a certain value or increase from baseline has not been identified, which may have affected the number of patients screened positive for withdrawal (2,9). Critically ill patients can often require hemodynamic support, and the strict blood pressure and heart rate requirements for withdrawal assessment may have led to a number of patients being falsely negative for withdrawal. However, the heart rate and blood pressure threshold was chosen based on previous studies and to fit the timeframe of data collection and analysis. Although CAM-ICU can be used to detect hypoactive delirium, this is typically harder to detect than hyperactive delirium and some patients may have been inappropriately screened negative (15). Utilizing 6-h time blocks for withdrawal assessment allows two nursing shifts worth of data to be charted on for each patient, giving the best chance of detection as nurses are only required to chart CAM-ICU and RASS scores once per shift at a minimum (14,15).

Conclusions

Caution should be taken with prolonged durations of dexmedetomidine infusions, especially in patients with atrial fibrillation, to avoid risk of developing withdrawal symptoms, which can be assisted by continuous intentional assessment of sedation choice. An incidence of withdrawal of 46% presents an opportunity to prevent withdrawal or provide earlier identification in a number of patients. Future studies are needed to further explore patient, medication, and other characteristics that may be risk factors for developing dexmedetomidine withdrawal.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-24-146/rc

Data Sharing Statement: Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-24-146/dss

Peer Review File: Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-24-146/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-24-146/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the institutional review board of Trinity Health Grand Rapids and Muskegon (Mercy Health Regional IRB—No. 23-1114-3) and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict provision that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Simmons B, Kuo A. Analgesics, Tranquilizers, and Sedatives. In: Brown DL. editor. Cardiac Intensive Care, 3rd edition. Elsevier; 2019:421-31.
2. Pathan S, Kaplan JB, Adamczyk K, et al. Evaluation of dexmedetomidine withdrawal in critically ill adults. J Crit Care 2021;62:19-24. [Crossref] [PubMed]
3. Shehabi Y, Howe BD, Bellomo R, et al. Early Sedation with Dexmedetomidine in Critically Ill Patients. N Engl J Med 2019;380:2506-17. [Crossref] [PubMed]
4. Precedex (dexmedetomidine hydrochloride). FDA Prescribing Information. Hospira. 2013. Available online: https://www.accessdata.fda.gov/drugsatfda_docs/label/2013/021038s021lbl.pdf
5. Pichot C, Ghignone M, Quintin L. Dexmedetomidine and clonidine: from second- to first-line sedative agents in the critical care setting? J Intensive Care Med 2012;27:219-37. [Crossref] [PubMed]
6. Jakob SM, Ruokonen E, Grounds RM, et al. Dexmedetomidine vs midazolam or propofol for sedation during prolonged mechanical ventilation: two randomized controlled trials. JAMA 2012;307:1151-60. [Crossref] [PubMed]
7. Pandharipande PP, Pun BT, Herr DL, et al. Effect of sedation with dexmedetomidine vs lorazepam on acute brain dysfunction in mechanically ventilated patients: the MENDS randomized controlled trial. JAMA 2007;298:2644-53. [Crossref] [PubMed]
8. Devlin JW, Skrobik Y, Gélinas C, et al. Clinical Practice Guidelines for the Prevention and Management of Pain, Agitation/Sedation, Delirium, Immobility, and Sleep Disruption in Adult Patients in the ICU. Crit Care Med 2018;46:e825-73. [Crossref] [PubMed]
9. Bouajram RH, Bhatt K, Croci R, et al. Incidence of Dexmedetomidine Withdrawal in Adult Critically Ill Patients: A Pilot Study. Crit Care Explor 2019;1:e0035. [Crossref] [PubMed]
10. Glaess SS, Attridge RL, Christina Gutierrez G. Clonidine as a strategy for discontinuing dexmedetomidine sedation in critically ill patients: A narrative review. Am J Health Syst Pharm 2020;77:515-22. [Crossref] [PubMed]
11. Catapres (clonidine hydrochloride). FDA Prescribing Information. Boehringer Ingelheim. 2009. Available online: https://www.accessdata.fda.gov/drugsatfda_docs/label/2009/017407s034lbl.pdf
12. Thompson RZ, Gardner BM, Autry EB, et al. Survey of the Current Use of Dexmedetomidine and Management of Withdrawal Symptoms in Critically Ill Children. J Pediatr Pharmacol Ther 2019;24:16-21. [Crossref] [PubMed]
13. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818-29. [Crossref] [PubMed]
14. Sessler CN, Gosnell MS, Grap MJ, et al. The Richmond Agitation-Sedation Scale: validity and reliability in adult intensive care unit patients. Am J Respir Crit Care Med 2002;166:1338-44. [Crossref] [PubMed]
15. Ely EW, Inouye SK, Bernard GR, et al. Delirium in mechanically ventilated patients: validity and reliability of the confusion assessment method for the intensive care unit (CAM-ICU). JAMA 2001;286:2703-10. [Crossref] [PubMed]
16. Pöyhiä R, Nieminen T, Tuompo VWT, et al. Effects of Dexmedetomidine on Basic Cardiac Electrophysiology in Adults; a Descriptive Review and a Prospective Case Study. Pharmaceuticals (Basel) 2022;15:1372. [Crossref] [PubMed]
17. Wozniak H, Beckmann TS, Dos Santos Rocha A, et al. Long-stay ICU patients with frailty: mortality and recovery outcomes at 6 months. Ann Intensive Care 2024;14:31. [Crossref] [PubMed]
18. Bhatt K, Thompson Quan A, Baumgartner L, et al. Effects of a Clonidine Taper on Dexmedetomidine Use and Withdrawal in Adult Critically Ill Patients-A Pilot Study. Crit Care Explor 2020;2:e0245. [Crossref] [PubMed]
19. Gagnon DJ, Riker RR, Glisic EK, et al. Transition from dexmedetomidine to enteral clonidine for ICU sedation: an observational pilot study. Pharmacotherapy 2015;35:251-9. [Crossref] [PubMed]