Abstract
Background:
Emergency departments represent a unique opportunity to intervene in mental health and substance use crises. For people residing in frontier and remote locations (> 60 min from cities of 50,000 people), emergency departments may be a critical source of mental healthcare, given limited local access to mental health professionals. The purpose of the current study was to investigate emergency department usage for substance use disorders and suicidal ideation in patients residing in frontier and non-frontier locations.
Methods:
South Dakotan syndromic surveillance data from 2017-2018 were obtained for this cross-sectional study. ICD-10 codes were queried to identify substance use disorder and suicidal ideation during emergency department visits. Differences in substance use visits were investigated in frontier and non-frontier patients. Additionally, logistic regression was used to predict suicidal ideation in cases and age- and sex-matched controls.
Results:
Frontier patients had a higher percentage of emergency department visits with a diagnosed nicotine use disorder. Conversely, non-frontier patients were more likely to use cocaine. Substance use for other categories of substances was similar between the frontier and non-frontier patients. Alcohol, cannabis, nicotine, opioid, stimulant, and psychoactive substance diagnoses all increased the patient's odds of receiving a suicidal ideation diagnosis. Further, residing in a frontier location also increased the odds of suicidal ideation.
Conclusions:
Patients residing in frontier locations differed in some forms of substance use disorders and in suicidal ideation. Increasing access to mental health and substance use treatment may be critical for those residing in these remote locations.
Keywords: Substance Use, Suicidal Ideation, Frontier, South Dakota
Introduction
Substance use disorders and suicide are often co-occurring.1 In the United States, suicide and substance use deaths have increased dramatically over the past decade. Specifically, unintentional drug overdoses have increased approximately five-fold from 1999 to 2017.2 Similarly, in 2018, suicide was the 10th leading cause of death in the United States.3,4 Because these preventable deaths are increasing, understanding trends and the opportunities to intervene are critical.
Rural health disparities exist in these measures of preventable deaths. Overall, suicide rates were 45% higher in rural than metro areas.3 Although suicide rates have increased in both rural and urban settings, the rate of increase was higher for males and females in rural settings compared to urban.3 Drug overdose deaths have increased in both urban and rural settings, but small metros and large fringe metros had slightly higher deaths per capita than rural and large central metros.2
For certain types of substances, overdose mortality differed across the rural/urban continuum.5 Methamphetamine overdoses and alcohol-related deaths were significantly higher in rural settings, but heroin and cocaine-related deaths were higher in urban environments.6,7 Finally, emergency department visits for drug poisoning were highest among residents in micropolitan communities and lowest in large central metropolitan communities.8 Understanding these trends is critical for developing tailored interventions.
Frontier living represents unique challenges to healthcare. Approximately 12.2 million people live in geographical locations considered to be frontier and remote level one.10 This is defined as a geographical location that is more than 60 minutes away from urban areas of 50,000 or more people.9 South Dakota ranks in the top 5 states for the proportion of the population living in frontier and remote locations and in the proportion of land classified as frontier and remote. These locations are far from healthcare facilities that provide advanced medical procedures; thus, this population may be at a higher risk for medical and mental health conditions given the reduced access to treatment.
The CDC reported that 3 out of 5 patients who died of a drug overdose had prior intervention opportunities.11 The emergency department provides a unique opportunity for interventions to reduce preventable deaths. Understanding trends in emergency department usage for substance use and suicidal ideation is critical for training frontline healthcare workers and developing tailored interventions. In the current study, emergency department utilization was examined in South Dakota for substance use disorders and suicidal ideation in frontier and non-frontier patients. One objective of this study was to identify differences in emergency department visits for substance use disorders and suicidal ideation in patients residing in frontier and non-frontier settings. A second objective was to determine if substance use, frontier residence, or health professional shortage areas increased the odds of suicidal ideation.
Methods
Syndromic surveillance data was retrieved from the South Dakota Department of Health from January 1, 2017 to December 31, 2018, for this retrospective cross-sectional study. Syndromic surveillance conducted by the South Dakota Department of Health is an integral part of public health monitoring. Briefly, electronic health records from emergency departments are sent to the state for review to enable near real-time monitoring of diseases and other indicators of health. This statewide sample was chosen to reduce selection bias. Although most emergency departments participated, not all emergency departments in the state participated in the syndromic surveillance system at the time of data collection. Data included 575,904 records from contributing emergency departments in South Dakota. Of those records, approximately 99% had either a completed discharge diagnosis or chief complaint. This included 97% with a non-missing chief complaint and 66% with a non-missing discharge diagnosis. Given the missing discharge diagnosis fields, initial attempts were made to categorize observations based on chief complaints. However, chief complaints are open text fields to document the reason a person seeks medical attention, not necessarily what the person was diagnosed with. These fields often contain non-standard abbreviations, misspellings, or symptoms instead of diagnoses.22 However, discharge diagnosis, where standardized codes are used, may be delayed or omitted because emergency departments are awaiting final test results such as drug testing. Still, prior studies have found that using discharge diagnosis from syndromic surveillance when compared to chart reviews offers both high accuracy and sensitivity of diagnosis.23,24 Chief complaints alone resulted in the lowest accuracy and sensitivity.24 Based on these previous studies and to maximize the accuracy of our findings, records with no ICD-10 or invalid discharge diagnosis were removed from the dataset resulting in 372,959 records. Of all emergency department visits not included in this analysis, 23.5% were from frontier areas, while 37.2% were from non-frontier locations. ICD-10 codes were then queried and labeled for substance use disorder-related events or suicidal ideation events via the Python SNOTRA package. The ICD-10 codes queried were similar to those described in prior studies.12 Codes included alcohol (F10), opioids (F11), cannabis (F12), sedatives (F13), cocaine (F14), stimulants (F15), hallucinogens (F16), nicotine (F17), inhalants (F18), psychoactive substances (F19), and suicidal ideation (R45.851). Age, gender, and patient zip code were included in the dataset. Frontier level one status10 and health professional shortage areas status (HPSAs)13 were derived from the patient residence zip codes.
Chi-squared analysis was utilized to assess for differences in substance use disorders and suicidal ideation measures by frontier status using SAS Studio (372,959 syndromic surveillance records included). Posthoc analysis was completed using a Bonferroni adjustment. For logistic regression analysis, suicidal ideation cases (n = 4108) were matched by age and sex. Only cases where both age and sex were reported in the syndromic surveillance record were included, resulting in the exclusion of 8 cases and 127 potential controls. Stepwise selection of predictor variables was performed. Findings were considered significant when p ≤ 0.05.
Results
Demographic information is found in Table 1. Results suggested the patients utilizing the emergency department from frontier areas are more likely to be older (X2(7) = 1796.84, p < 0.05). A slightly higher proportion of emergency department visits were male patients from frontier areas compared to non-frontier areas (X2(2) = 92.93, p < 0.05).
Table 1:
Demographic information for records included in analyses. Cells contain the sample size (% of visits).
| Demographic Information |
Number of Non- Frontier Visits (% of 239043 Visits) |
Number of Frontier Visits (% of 133916 Visits) |
Number Visits with No Suicidal Ideation (% of 4108 Visits) |
Number of Visits Suicidal Ideation (% of 4108 Visits) |
|---|---|---|---|---|
| Sex | ||||
| Female | 130458 (54.58) | 70889 (52.94) | 2160 (52.58) | 2160 (52.58) |
| Male | 108581 (45.42) | 63025 (47.06) | 1948 (47.42) | 1948 (47.42) |
| Unknown | 4 (< 0.01) | 2 (< 0.01) | ||
| Age Category | ||||
| 0 - 14 years old | 39204 (16.4) | 22459 (16.77) | 531 (12.93) | 531 (12.93) |
| 15 - 24 years old | 32933 (13.78) | 16927 (12.64) | 1420 (34.57) | 1420 (34.57) |
| 25 - 34 years old | 35703 (14.94) | 17442 (13.02) | 849 (20.67) | 849 (20.67) |
| 35 - 44 years old | 30181 (12.63) | 14540 (10.86) | 555 (13.51) | 555 (13.51) |
| 45 - 54 years old | 26202 (10.96) | 13205 (9.86) | 383 (9.32) | 383 (9.32) |
| 55 - 64 years old | 25961 (10.86) | 14879 (11.11) | 254 (6.18) | 254 (6.18) |
| 65 + years old | 48662 (20.36) | 34436 (25.71) | 116 (2.82) | 116 (2.82) |
| Unknown | 197 (0.08) | 28 (0.02) | ||
Overall, emergency department visits for substance use-related reasons differed between patients residing in frontier vs. non-frontier regions (X2(2) = 594.38, p < 0.05). Posthoc analysis revealed a higher proportion of visits with a single substance use disorder (Frontier: n = 7,750, 5.79% vs. Non-Frontier: n = 9,696, 4.06%) and polysubstance use (Frontier: n = 607, 0.45% vs. Non-Frontier: n = 885, 0.37%) in patients residing in frontier locations. Most types of substance use disorder diagnoses were similar in patients who live in frontier locations compared to non-frontier locals (Table 2). The exception was nicotine use, where the percentage of patients with nicotine use (F17) was 2.4 times higher in people residing in frontier locations than non-frontier. Psychoactive substance use and cocaine use diagnoses were slightly higher in patients living in non-frontier areas than frontier locations. Opioid, inhalant, hallucinogen, sedatives, cannabis, and stimulant use diagnoses made up similar percentages of emergency department visits in frontier and non-frontier patients. A trend towards lower cannabis use in frontier patients also was observed (p = 0.07). Of interest, alcohol (F10) was noted in approximately 2.5% of visits from both frontier and non-frontier patients.
Table 2:
The number (percentage) of emergency department visits for substance use disorders in patients residing in frontier and non-frontier settings.
| Substance Category (ICD 10 Code) |
Number of Non-Frontier Visits (% of 239043 Visits) |
Number of Frontier Visits (% of 133916 Visits) |
X2(1) | P-Value |
|---|---|---|---|---|
| Alcohol (F10) | 6028 (2.52) | 3399 (2.54) | 0.09 | ns |
| Opioid (F11) | 310 (0.13) | 181 (0.14) | 0.2 | ns |
| Cannabis (F12) | 504 (0.21) | 245 (0.18) | 3.33 | p = 0.07 |
| Sedatives (F13) | 66 (0.03) | 27 (0.02) | 1.91 | ns |
| Cocaine (F14) | 52 (0.02) | 13 (0.01) | 7.15 | p < 0.05 |
| Stimulant (F15) | 681 (0.28) | 388 (0.29) | 0.07 | ns |
| Hallucinogen (F16) | 16 (0.01) | 8 (0.01) | 0.07 | ns |
| Nicotine (F17) | 3435 (1.44) | 4528 (3.38) | 1552.77 | p < 0.05 |
| Inhalant (F18) | 24 (0.01) | 15 (0.01) | 0.11 | ns |
| Psychoactive (F19) | 479 (0.2) | 229 (0.17) | 3.91 | p < 0.05 |
Suicidal ideation was also significantly higher in patients residing in frontier locations (X2(1) =34.62, p < 0.05; Odds Ratio: 1.21, 95% Confidence Limits: 1.13 to 1.28). Suicidal ideation was reported in approximately 1.24% (n = 1,658) of emergency department visits in patients residing in frontier locations compared to 1.03% (n = 2,458) of visits in patients residing in non-frontier locations.
To further explore the suicidal ideation emergency department visits, logistic regression was performed using patients with a suicidal ideation diagnosis and controls matched by age and sex (1 case: 1 control). Table 1 describes the demographic information of the cases and controls selected. Stepwise variable selection resulted in the inclusion of psychoactive substance use, nicotine use, cannabis use, alcohol use, opioid use, stimulant use, and frontier status as predictor variables (X2(7) = 451.51, p < 0.05). Substance use increased the odds that the patient exhibited suicidal ideation (Table 3). Patients residing in frontier locations also exhibited higher odds of expressing suicidal ideations.
Table 3:
Predictor variable and odds ratio estimate of variables included in logistic regression of suicidal ideation.
| Model Variables | Number of Visits with No Suicidal Ideation (% of 4108 Visits) |
Number of Visits with Suicidal Ideation (% of 4108 Visits) |
Odds Ratio Estimate |
95% Wald Confidence Limits of Odds Ratio |
|
|---|---|---|---|---|---|
| Alcohol (F10) | |||||
| No | 3973 (96.71) | 3502 (85.25) | |||
| Yes | 135 (3.29) | 606 (14.75) | 4.58 | 3.77 | 5.57 |
| Opioid (F11) | |||||
| No | 4104 (99.90) | 4054 (98.69) | |||
| Yes | 4 (0.1) | 54 (1.31) | 10.25 | 3.65 | 28.77 |
| Cannabis (F12) | |||||
| No | 4092 (99.61) | 3976 (96.79) | |||
| Yes | 16 (0.39) | 132 (3.21) | 5.86 | 3.43 | 10.02 |
| Stimulant (F15) | |||||
| No | 4098 (99.76) | 3947 (96.08) | |||
| Yes | 10 (0.24) | 161 (3.92) | 11.99 | 6.27 | 22.95 |
| Nicotine (F17) | |||||
| No | 4001 (97.4) | 3857 (93.89) | |||
| Yes | 107 (2.6) | 251 (6.11) | 1.84 | 1.44 | 2.34 |
| Psychoactive (F19) | |||||
| No | 4098 (99.76) | 4031 (98.13) | |||
| Yes | 10 (0.24) | 77 (1.87) | 5.99 | 3.05 | 11.78 |
| Frontier | |||||
| No | 2719 (66.19) | 2451 (59.66) | |||
| Yes | 1389 (33.81) | 1657 (40.34) | 1.28 | 1.17 | 1.41 |
Discussion
The current study sought to understand emergency department utilization for substance use and suicidal ideation in South Dakota. Overall, patients residing in frontier locations had a higher percentage of visits for a single substance use disorder and polysubstance use. This increase was likely mediated by higher nicotine use among patients residing in frontier and remote areas. The South Dakota Youth Tobacco Survey found higher e-cigarette uses among rural students than urban students.14 Similarly, rural high school students had higher cigarette and smokeless tobacco use than students in urban locations.15 Disparities in nicotine use among rural, frontier, and remote locations may emerge in youth and persist leading to an overall higher nicotine use rate in rural settings.
Two categories of substance use disorders were higher among patients residing in non-frontier settings. Cocaine emergency department visits, although rare overall, were higher in non-frontier settings. This is consistent with prior studies showing that cocaine overdose mortality was higher in urban settings than in rural areas.16 Psychoactive substance use emergency department visits were also higher in non-frontier residing patients compared to frontier patients. A trend towards a higher percentage of cannabis emergency department visits was observed in non-frontier patients compared to frontier patients. As South Dakota readies to legalize marijuana, increases in cannabis-related emergency department visits will likely occur.
A substance use disorder diagnosis increased the odds of patients exhibiting suicidal ideation. Alcohol, opioid, cannabis, stimulant, nicotine, and psychoactive substance use all increased these odds. With the legalization and decriminalization of cannabis in this state and others, monitoring the co-occurrence of suicidal ideation and cannabis use will become increasingly important. In Colorado, where marijuana is legal, an increasing proportion of suicides have tested positive for the drug.17 Similarly, among those who have died by suicide in 34 states across the United States, 40% tested positive for alcohol, 12% tested positive for amphetamines, 22.6% tested positive for marijuana, and 25.2% tested positive for opioids.18 Systematic reviews of suicide deaths also found that substance use is a risk factor for suicide.19 Emergency department visits may represent a critical opportunity for interventions for both serious mental illness and substance use disorders.
Barriers to healthcare access also increased the odds of suicidal ideation. Frontier status increased the odds of the patient seeking emergency attention for suicidal ideation. Access to mental health resources is often limited in rural areas.20 Among rural non-core areas, 80% of these rural counties are without a psychiatrist, 61% are without a psychologist, 35% lack a social worker, and 24% lack a counselor.21 Reduced access to mental health professionals may influence the need to seek treatment in emergency settings. With expanded use of telehealth, monitoring these trends will become increasingly important.
Limitations
The current study is not without its limitations. Only South Dakota emergency department visits were included in this study. As discussed above, not all emergency departments participated in the syndromic surveillance program during this time or may have utilized alternative coding systems such as SNOMED CT, especially in non-frontier settings. The state has a high proportion of residents living in frontier designated areas and may not generalize to other areas of the country. Lastly, emergency departments are only one source of treatment for substance use and mental health crises. Future studies are necessary to examine if these findings persist when examining other treatment locations, such as office visits.
Conclusions
The current study sought to investigate emergency department usage related to substance use disorders and suicidal ideation in frontier and non-frontier settings in South Dakota. Overall, patients residing in frontier classified zip codes had higher nicotine use and lower cocaine use than non-frontier settings as measured by ICD-10 codes. However, other categories of substance use disorders were similar between the frontier and non-frontier patients. Finally, substance use and frontier status increased the patient's odds of reporting suicidal ideation in the emergency department. Given most forms of substance use and suicidal ideation were similar or higher in frontier patients and access to mental healthcare may be limited, novel strategies may be needed to address these needs. Expanding access to treatment for substance use and mental health, especially in frontier and remote areas, is critical to aid patients in these geographical regions. The expansion of telehealth may represent an opportunity to improve access to the treatment of mental health and substance use disorders in these underserved areas.20
Acknowledgements:
We would like to thank the South Dakota Department of Health for providing access to this data.
Funding Sources:
This project was supported by the National Institute on Drug Abuse (DA033674) and the National Institute of General Medical Sciences (GM121341). The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies.
Footnotes
Conflict of Interest: The authors have no conflicts to disclose.
Contributor Information
Alexandra Dolezal, Division of Basic Biomedical Science, Sanford School of Medicine, University of South Dakota, Vermillion, SD.
Marie Severson, Division of Basic Biomedical Science, Sanford School of Medicine, University of South Dakota, Vermillion, SD.
Rusul Ali, Division of Basic Biomedical Science, Sanford School of Medicine, University of South Dakota, Vermillion, SD.
Phil Dohn, Division of Basic Biomedical Science, Sanford School of Medicine, University of South Dakota, Vermillion, SD.
Lisa McFadden, Division of Basic Biomedical Science, Sanford School of Medicine, University of South Dakota, Vermillion, SD; Sanford Research, Sioux Falls, SD.
References
- 1.Land ME, Wetzel M, Geller RJ, Steck AR, Grunwell JR. Adult opioid poisonings by drug, intent, and resource use from the United States National Poison Data System from 2005-2018. Clin Toxicol. 2021;59: 142–151. [DOI] [PubMed] [Google Scholar]
- 2.Olaisen RH, Rossen LM, Olaisen RH, Rossen LM, Warner M, Anderson RN. Unintentional Injury Death Rates in Rural and Urban Areas: United States, 1999-2017. NCHS Data Brief. 2019;343:1–8. [PubMed] [Google Scholar]
- 3.Pettrone K, Curtin SC. Urban-rural Differences in Suicide Rates, by Sex and Three Leading Methods: United States, 2000-2018. NCHS Data Brief. 2020;373:1–8. [PubMed] [Google Scholar]
- 4.Xu J, Murphy SL, Kockanek KD, Arias E. Mortality in the United States, 2018. NCHS Data Brief. 2020;355:1–8. [PubMed] [Google Scholar]
- 5.Hedegaard H, Spencer MR, Garnett MF. Increase in Drug Overdose Deaths Involving Cocaine: United States, 2009-2018. NCHS Data Brief. 2020;384:1–8. [PubMed] [Google Scholar]
- 6.Hedegaard H, Miniño AM, Warner M. Urban-rural Differences in Drug Overdose Death Rates, by Sex, Age, and Type of Drugs Involved, 2017. NCHS Data Brief. 2019;345:1–8. [PubMed] [Google Scholar]
- 7.Spencer MR, Curtin SC, Hedegaard H. Rates of alcohol-induced deaths among adults aged 25 and over in rural and urban areas: United States, 2000–2018. NCHS Data Brief. 2020;383:1–8. [PubMed] [Google Scholar]
- 8.Centers for Disease Control and Prevention. 2019 Annual Surveillance Report of Drug-Related Risks and Outcomes — United States Surveillance Special Report. Centers for Disease Control and Prevention, U.S. Department of Health and Human Services. https://www.cdc.gov/drugoverdose/pdf/pubs/2019-cdc-drug-surveillancereport.pdf. Published November 1, 2019. [Google Scholar]
- 9.United States Department of Agriculture. Frontier and Remote Area Codes. Economic Research Service, United States Department of Agriculture. https://www.ers.usda.gov/data-products/frontier-and-remote-area-codes.aspx. Updated August 20, 2019. [Google Scholar]
- 10.United States Department of Agriculture. Documentation- 2010 Frontier and Remote (FAR) Area Codes. Economic Research Service, United States Department of Agriculture. https://www.ers.usda.gov/data-products/frontier-and-remote-area-codes/documentation/. Updated August 20, 2019. [Google Scholar]
- 11.O'Donnell J, Gladden RM, Mattson CL, Hunter CT, Davis NL. Vital Signs: Characteristics of Drug Overdose Deaths Involving Opioids and Stimulants - 24 States and the District of Columbia, January-June 2019. MMWR Morb Mortal Wkly Rep. 2020;69:189–1197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Di Rico R, Nambiar D, Stoové M, Dietze P. Drug overdose in the ED: a record linkage study examining emergency department ICD-10 coding practices in a cohort of people who inject drugs. BMC Health Serv Res. 2018;18:945. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Centers for Medicare & Medicaid Services. HPSA Zip Code Listings. Qualified Health Plan Certification Information and Guidance. https://www.qhpcertification.cms.gov/s/Service%20Area. Published 2020. [Google Scholar]
- 14.Da Rosa P, Parsons J, Miller A, Jodozi-Molengraaf C, Kerkvliet J. E-cigarette Use Among Youth: Results from the 2019 South Dakota Youth Tobacco Survey. S D Med. 2020;73:484–488. [PubMed] [Google Scholar]
- 15.Wiggins AT, Huntington-Moskos L, Rayens EA, et al. Tobacco Use Among Rural and Urban US Middle and High School Students: National Youth Tobacco Survey, 2011-2016. J Rural Health. 2020;36:48–54. [DOI] [PubMed] [Google Scholar]
- 16.Kariisa M, Scholl L, Wilson N, Seth P, Hoots B. Drug Overdose Deaths Involving Cocaine and Psychostimulants with Abuse Potential - United States, 2003-2017. MMWR Morb Mortal Wkly Rep. 2019;68:388–395. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Roberts BA. Legalized Cannabis in Colorado Emergency Departments: A Cautionary Review of Negative Health and Safety Effects. West J Emerg Med. 2019;20(4):557–572. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Petrosky E, Ertl A, Sheats KJ, Wilson R, Betz CJ, Blair JM. Surveillance for Violent Deaths - National Violent Death Reporting System, 34 States, Four California Counties, the District of Columbia, and Puerto Rico, 2017. MMWR Surveill Summ. 2020;69:1–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mohatt NV, Kreisel CJ, Hoffberg AS, Mph LW, Beehler SJ. A Systematic Review of Factors Impacting Suicide Risk Among Rural Adults in the United States. J Rural Health. 2021;37(3):565–575. [DOI] [PubMed] [Google Scholar]
- 20.Warfield SC, Pack RP, Degenhardt L, et al. The next wave? Mental health comorbidities and patients with substance use disorders in under-resourced and rural areas. J Subst Abuse Treat. 2021;121:108189. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Larson EH, Patterson DG, Garberson LA, Andrilla CHA. Supply and Distribution of the Behavioral Health Workforce in Rural America. Data Brief #160. Seattle, WA: WWAMI Rural Health Research Center, University of Washington; 2016. [Google Scholar]
- 22.Virginia Department of Health. Limitations of Syndromic Surveillance Data. Syndromic-Surveillance-Limitations_For-Website_2-26-21.pdf (virginia.gov). Published 2021. [Google Scholar]
- 23.Liljeqvist HT, Muscatello D, Sara G, Dinh M, Lawrence GL. Accuracy of automatic syndromic classification of coded emergency department diagnoses in identifying mental health-related presentations for public health surveillance. BMC Med Inform Decis Mak. 2014;14:84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Reis BY, Mandl KD. Syndromic surveillance: the effects of syndrome grouping on model accuracy and outbreak detection. Ann Emerg Med. 2004;44(3):235–41 [DOI] [PubMed] [Google Scholar]