French, Dustin D. PhD*; Margo, Curtis E. MD†; Harman, Lynn E. MD‡
The interplay of risk factors for primary open-angle glaucoma (age, race, positive family history, intraocular pressure, cup size, corneal thickness, etc.) is incompletely understood. The identification of other risk factors may someday further our understanding of the pathogenesis of the disease and better guide treatment. One of the authors of this study (L.E.H.) has sensed a high prevalence of recreational cocaine use among patients with glaucoma and thought it may contribute to the development or progression of the disease beyond the issue of compliance. Such a hypothesis is biologically feasible given the complex interactions cocaine has on central nervous system blood flow and the plasma membrane transport systems of dopamine, norepineprhine, and serotonin.1,2 Although the long-term affects of cocaine use on intraocular pressure are unknown, limited experimental data in animals have shown intravenous cocaine can effect aqueous dynamics.3
We used a case-control study design to test the hypothesis that persons with open-angle glaucoma (both primary open-angle glaucoma and low-tension glaucoma) have a greater exposure risk to cocaine than a comparable population of patients without glaucoma.
MATERIALS AND METHODS
The Veterans Health Administration (VHA) national encounter database was used to collect information on medical diagnoses and was linked to national pharmacy files (inpatient and outpatient) using a scrambled identifier. The database consolidates information from approximately 1300 sites of care throughout the nation on over 5.3 million veterans. Over 95% of VHA enrollees are men, and roughly 73% are White. Patient-specific information from the medical records was not available for review.
Potential case patients were identified using International Classification of Diseases (9th revision), Clinical Modification (ICD-9-CM) for fiscal year (FY) 2009. Two ICD-9-CM diagnostic categories of open-angle glaucoma were used to screen for case patients: primary open-angle glaucoma (ICD-9-CM code 365.11) and low-tension or normal tension glaucoma (ICD-9-CM code 365.12). After screening, inclusion criteria for glaucoma required consecutive prescriptions for 1 or more of the following topical antiglaucoma medications identified through the national pharmacy database: prostaglandins (and prostamides), carbonic anhydrase inhibitors, β-blockers, pilocarpine, or α-2 adrenergic agonists.
Cocaine use was identified through the following ICD-9-CM codes: 304.2 (cocaine dependence); 304.20 (cocaine dependence unspecific use); 304.21 (cocaine dependence continuous use); and 304.22 (cocaine dependence episodic use); 304.23 (cocaine dependence in remission); 305.60 (nondependent cocaine abuse unspecified use); 305.61 (nondependent cocaine abuse continuous use): 305.62 (nondependent cocaine abuse episodic use); and 305.63 (nondependent cocaine abuse in remission).
ICD-9-CM coding of cocaine dependency follows the nomenclature of drug dependency in the Diagnostic and Statistical Manual of Mental Diseases, Fourth Edition (DSM-IV).4 The DSM-IV criteria for drug dependency has been tested against other assessment standards and found reliable.4–6
To facilitate communication in this paper, the term “cocaine exposure” refers to ICD-9-CM codes for dependency, abuse, and remission. It is the most inclusive group of cocaine-related diagnoses. “Cocaine dependency” refers the 4 ICD-9-CM codes 304.2, 304.20, 304.21, and 304.22. The term “cocaine use” includes the ICD-9-CM codes for abuse and those for remission. The same terminology and coding patterns are applied to other illicit drugs (opioids, amphetamines, etc.). The term “substance use disorder” describes illicit drug use in general regardless of type or number.
The control group came from the same source population as case patients. Control patients consisted of national VHA beneficiaries who used the outpatient clinic for 1 or more clinical evaluations (any medical or surgical clinic) during FY 2009. A 5-year look-back period (through FY 2004) was used to exclude beneficiaries with ICD-9-CM codes for glaucoma (365.0 to 365.9).
Since any association between glaucoma and cocaine could be confounded by illicit drug use in general, we created a binary variable “other drugs” to assess the impact of substance use disorder. ICD-9-CM codes within the 304 to 305 range for illicit abuse and dependency were selected for opioids, sedatives, hypnotics, anxiolytics, cannabis, hallucinogens, amphetamines, and antidepressants (codes not listed but are available on request). For the purposes of this paper, poly-drug exposure refers to 2 or more ICD-9-CM codes for illicit drug use. Other illicit drugs refer to the above drug groups exclusive of cocaine. Alcohol and tobacco use were not included. An interaction variable for cocaine and other illicit drugs was used in logistic regression models to test for independence of effect on glaucoma.
We used a standard logistic regression model with 2 outcome variables (open-angle glaucoma and none) to test our primary hypothesis, and a multinomial logistic regression model with a 3-level outcome variable: 1, primary open-angle; 2, low-tension; and 3, neither or none (reference group), to test the secondary hypothesis. Independent variables included cocaine exposure, other illicit drugs exposures, and selected drug interactions (determined from preliminary models). The odds ratios of cocaine exposure, other illicit drug exposure, and significant drug interaction effects were assessed for open-angle glaucoma (primary hypothesis) and for 2 glaucoma subtypes (ie, primary open-angle glaucoma and low-tension glaucoma) (secondary hypothesis), adjusted for age and reported by sex.
As the VHA dataset for race are incomplete, we conducted a nested case-control study from the FY 2009 population of beneficiaries limited to patients whose racial identity was available. Preliminary analysis of those records showed the composition of the sample was similar in racial distribution to the overall VHA enrollment. As the number of Asians and Native Americans was less than 1%, they were not included in the analysis. The same inclusion and exclusion criteria were applied. The nested subset of patients was examined with a race categorical variable included in the logistic regression model described above. White race was the reference group. Results were reported according to sex.
Levels of cocaine exposure were examined using a surrogate measure of severity of exposure based on ICD-9-CM codes for drug-dependency status (ie, dependency vs user).
We used the Statistical Analysis Software Version 9.1 Cary, NC and specifically the procedure CATMOD for multinomial logistic regression model. Documentation of the CATMOD procedure is available through Statistical Analysis Software on-line.7 Odds ratios and 95% confidence intervals (CIs) were obtained by calculating the exponential of the parameter estimates obtained from the CATMOD procedure.
This study was approved by the Indiana University Institutional Review Board and the Veterans Administration Research and Development Committee for Compliance with Human Protection.
For FY 2009 there were 5,373,205 individual VHA enrollees who used the outpatient clinics. Among these VHA beneficiaries, 82,990 (1.5%) had a primary ICD-9-CM diagnosis coded as either primary open-angle glaucoma or low-tension glaucoma and pharmacy claims for at least 1 topical antiglaucoma medication. During the same FY, there were 177,929 patients (3.3%) with ICD-9-CM coded diagnoses of cocaine abuse or dependency. Among these beneficiaries, 105,023 (2.0%) were listed as drug (cocaine) dependent. Women made up 9.1% of the outpatient beneficiaries, and 4.3% of all cocaine-exposed patients. Classification of race was available on 1,340,360 beneficiaries, 264,360 of whom were Black (19.7%). Less than 1% were Asian, Native American, and other.
During this same time period, there were 712,264 beneficiaries (13.4%) who had ICD-9-CM codes for substance use disorder (illicit use of opioids, sedative, hypnotics, anxiolytic, cannalbis, hallucinogen, amphetamines, antidepressants and other drugs, exclusive of sole cocaine use). Of these, 94.3% were men. Cocaine exposure and other illicit drug use were documented among 93,098 veterans (1.7%).
The proportion of men with primary open-angle glaucoma and low-tension glaucoma was 97.6% and 96.4%, respectively. The mean ages of veterans without a drug exposure history was 73.0 years (standard deviation [SD]:10.9y) for primary open-angle glaucoma, and 71 years (SD: 11.1y) for low-tension glaucoma. Patients with the corresponding glaucomas and a positive history of cocaine exposure were 54.1 years (SD: 7.0 y) and 53.1 years (SD 6.5 y), respectively (Table 1). The average age of veterans without glaucoma who used the VHA outpatient clinics in FY 2009 (n=5,290,215) was 61.0 years (SD: 16.3 y).
Reduced logistic regression models showed a significant association between open-angle glaucoma and exposure to cocaine, but there also were significant interactions between cocaine use and 4 other illicit drug types (opioids, hallucinogens, amphetamines, and cannabis). Independent variables in the full logistic model included illicit drug type, significant drug interaction variables, and age. For the sake of brevity, adjustment for other illicit drugs will hereafter include adjustments for drug interaction variables although not always stated. The model for the nested study also included race.
The age-adjusted and other illicit drug-adjusted risk of exposure to cocaine was significantly increased for open-angle glaucoma for both men [3.52 (95% CI: 3.21-3.86)] and women [1.87 (95% CI: 1.79-1.96)] (Table 2). Significant risks of exposure were found for primary open-angle and low-tension glaucoma for both sexes, although the magnitude of risk was greater for men than women (Table 2). The odds ratios of cocaine exposure ranged from a low of 1.87 (95% CI: 1.61-2.18) for women with low-tension glaucoma to a high of 3.62 (95% CI: 3.29-3.99) for men with primary open-angle glaucoma. The risks of cocaine exposure did not differ significantly by dependency status (data not shown).
Statistically significant exposures to opioids, amphetamines, and cannabis were also found for men with primary open-angle glaucoma, and for cannabis for women with primary open-angle glaucoma (Table 2). The interaction variables for cocaine and opioids, and for cocaine and amphetamines for men remained statistically significant (interactive variables not shown), but the interaction terms for cocaine and cannabis for men and women was not statistically significant.
In the nested study, the odds ratio of cocaine exposure after controlling for age, other illicit drugs, and race remained statistically significant for men [1.45 (95% CI: 1.27-1.66)], but not for women [0.98 (95% CI: 0.91-1.04)] (Table 3). Although the magnitude of risk of cocaine exposure was substantially less in men with open-angle glaucoma after adjustment, this risk was due to primary open-angle glaucoma [1.49 (95% CI: 1.30-1.71)] and not low-tension glaucoma [1.26 (95% CI: 0.85-1.86)] (Table 3).
The mean age of men and women with primary open-angle glaucoma with exposure to opioids, amphetamine and cannabis was 56.4 (SD 7.2 y), or about 16 years younger than glaucoma patients without a drug history (Table 1).
This study found an age-adjusted increased risk of cocaine exposure in both men and women veterans with open-angle glaucoma. A similar risk was also observed for both primary open-angle and low-tension glaucoma when compared with the national VHA outpatient population without glaucoma (Table 2). The nested study conducted to control for race showed a lesser but still statistically significant 45% increased exposure risk for men with open-angle glaucoma (Table 3). Men with open-angle glaucoma also had a statistically significant—but lesser magnitude—risk of exposure to amphetamines and cannabis in the nested study (Table 3). We did not find any significant difference in drug exposure risk according to drug-dependency status, a surrogate means of quantifying drug use.4 In particular, patients with open-angle glaucoma and exposure history to cocaine were about 18 years younger than glaucoma patients without a drug exposure history.
One of the inescapable hazards of epidemiological investigation of illicit drug use is the inability to measure individual drug effect because poly-drug abuse is so common among persons with substance use disorder.8,9 That was certainly the case encountered in the VHA, where over half as many beneficiaries were identified as cocaine-poly-drug users as cocaine users (93,098 vs 177,929). The proportion of true poly-drug users is likely far greater, which makes disentangling inferences about single drug exposures doubtful.10,11
Historically, information on race has been incomplete in the VHA dataset with some recent interest in improving future acquisition. We believe, however, that incomplete racial acquisition has been a random process. Our nested sample included 19.7% African Americans, which corresponds to the VHA estimate of 15.9% for this general cohort of beneficiaries.12 The slight increase in the proportion of Blacks in the nested cohort would tend to drive the odds ratio to the null, if it were to have any effect from the higher prevalence of glaucoma in African Americans.
The association of open-angle glaucoma and cannabis exposure is complicated by the fact that veterans could use cannabis to treat their glaucoma.13 Rather than being a risk factor for glaucoma, it could be a self-prescribed medication. Given the retrospective study design and inability to review patient-specific records, we cannot determine the immediate reasons for cannabis use.
Although the temporal relationship between exposure and disease cannot be determined in a case-control study, it is improbable that glaucoma preceded the use of other illicit drugs in this study, as substance use disorder in the vast majority of all Americans begins in the teens or twenties.11
In the civilian population, the behavioral aspects of substance use disorder tends to adversely affect access to medical care because it results in a delay of medical diagnoses rather than early detection.11 We suspect the antisocial behavior that accompanies drug use would also delay diagnosis of glaucoma in the VHA rather than result in earlier recognition of the disease. However, this traditional line of thinking could be challenged. As the VHA provides comprehensive drug rehabilitation and supportive general medical services, it is conceivable that veterans with drug abuse problems actually have better access to medical care (including eye care) in the VHA than an age-matched cohort without substance use disorder.
The strengths of this study include large numbers of cases and controls, and, what we believe is a reliable means of identifying patients with substance use disorder. Although the reliability of obtaining a illicit drug use history when not volunteered is generally poor, in the VHA admitted drug use tends to correlate strongly with objective tests for drug exposure.14,15 The VHA monitors substance use disorder programs closely and mandates thorough documentation of related diagnoses in the electronic medical record.16–18 The VHA substance use disorder treatment program is the largest and most comprehensive program of its kind in the country owing to the fact that chronic physical conditions and illicit drug use are so highly prevalent among military veterans.19–22 Published reports from nonveterans hospitals have shown that ICD-9-CM coding for drug abuse is 99% specific (low false-positive rate) for recording the correct diagnosis based on structured chart review.23 We have no reason to think that VHA ICD-9-CM coding of cocaine exposure is any less specific than it is in the civilian sector.
There are a variety of limitations related to the study of administrative databases, including inability to verify patient-specific diagnoses and drug use, inability to confirm the temporal relationship of exposure and outcome, as well as the possibility of unrecognized confounding.24 Although the results of any study using administrative databases must be interpreted cautiously, the technique can be valuable for generating clinically important hypotheses that require a large study population. Positive results typically required more rigorous methodologies for confirmation.
The hypothesis that chronic cocaine use is associated with open-angle glaucoma has not been tested earlier. Although this retrospective case-control study found an association between open-angle glaucoma and cocaine use in men independent of age and race, the interaction of cocaine use with poly-drug abuse leaves open the question of whether cocaine exposure alone or substance use disorder, in general, is the crucial variable. The lack of a significant association in women in the nested study could be due to the small number of female beneficiaries in the subset.
Cocaine exposure alone points to a biological pathway, whereas substance use disorder speaks in favor of a behavioral mechanism. The observation that glaucoma in drug-exposed patients is diagnosed 18 years before it is detected in nonexposed patients is easier to understand in terms of a biological predisposition rather than a bias toward early diagnosis.11,25
Given the estimated lifetime prevalence of drug abuse in the United States is 7.7%, with trends of persistent exposure continuing among persons over 50, the potential impact of illicit drug use on glaucoma in the general population could be substantial.10,11,26 The association of illicit drug use with open-angle glaucoma requires confirmation, but if the relationship is real, it could lead to new strategies to prevent vision loss.
1. Carrera MRA, Meijler MM, Janda KD. Cocaine pharmacology and current pharmacotherapies for its abuse. Bioorganic Med Chem. 2004;12:5019–5030
2. Gorelick DARies RK. The pharmacology of cocaine, amphetamines and other stimulants. Principles of Addiction Medicine. 2009;104th ed Philadelphia Wolters Kluwer:133–158
3. Artu AA. Trabecular outflow facility and formation rate of aqueous humor during intravenous cocaine or lidocaine in rabbits. Anesth Analg. 1994;78:889–896
4. Task Force on DSM-IV. Diagnostic and Statistical Manual of Mental Disorders Test Revision. 20004th ed Washington, DC American Psychiatric Association:242–243
5. Pierucci-Lagha A, Gelernter J, Chan G, et al. Reliability of DSM-IV diagnostic criteria using the semi-structured assessment of drug dependence and alcoholism (SSADDA). Drug Alcohol Depend. 2007;91:85–90
6. Ustün B, Compton W, Mager D, et al. WHO Study on the reliability and validity of the alcohol and drug use disorder instruments: overview of methods and results. Drug Alcohol Depend. 1997;47:161–169
8. Agrawal A, Lynskey MT, Madden PA, et al. A latent class analysis of illicit drug abuse/dependence: results from the National Epidemiological Survey on Alcohol and Related Conditions. Addiction. 2007;102:94–104
9. Gfroerer JC, Adams EH, Moien M. Drug Abuse discharges from non-federal short-stay hospitals. Am J Public Health. 1988;78:1559–1562
10. Compton WM, Thomas YF, Stinson FS, et al. Prevalence, correlates, disability and comorbidity of DSM-IV drug abuse and dependence in the United States: results from the national epidemiologic survey on alcohol and related conditions. Arch Gen Psychiatry. 2007;64:566–576
11. Brook JS, Pahl K, Rubenstone EGalanter M, Kleber HD. Epidemiology of addiction. Textbook of Substance Abuse and Treatment. 20084th ed Washington, DC American Psychiatric Publishing:29–38
12. Shen Y, Hendricks A, Zhang S, et al. VHA enrollees' health care coverage and use of care. Med Care Res Rev. 2003;60:253–267
13. Voth EA. A peek into Pandora's box: the medical excuse marijuana controversy. J Addict Dis. 2003;22:27–46
14. McNagny SE, Parker RM. High prevalence of recent cocaine use and the unreliability of patient self-report in an inner-city walk-in clinic. JAMA. 1992;267:1106–1108
15. Calhoun PS, Sampson WS, Bosworth HB, et al. Drug use and validity of substance use self-reports in veterans seeking help for posttraumatic stress disorder. J Consult Clin Psychol. 2000;68:923–927
16. Harris AH, Humphreys K, Bowe T, et al. Measuring the quality of substance use disorder treatment: evaluating the validity of the Department of Veterans Affairs continuity of care performance measure. J Subst Abuse Treat. 2009;36:294–305
19. Banerjea R, Findley PA, Smith B, et al. Co-occurring medical and mental illness and substance use disorders among veteran clinic users with spinal cord injury patients with complexities. Spinal Cord. 2009;47:789–795
21. Department of Health and Human Services: Substance Abuse and Mental Health Services Administration (SAMHSA). All Reports of Veterans. Available at http://www.oas.samhsa.gov/Veterans.htm
[Accessed on November 5, 2009].
22. Yang JC, Huang D, Hser YI. Long-term morbidity and mortality among a sample of cocaine-dependent black and white veterans. J Urban Health. 2006;83:926–940
23. Quan H, Li B, Saunders LD, et al. Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database. Health Serv Res. 2008;43:1424–1441
24. Coleman AL, Morgenstern H. Use of insurance claims databases to evaluate the outcomes of ophthalmic surgery. Surv Ophthalmol. 1997;42:271–278
25. Warner LA, Kessler RC, Hughes M, et al. Prevalence and correlates of drug use and dependence in the United States. Results from the National Comorbidity Survey. Arch Gen Psychiatry. 1995;52:219–229
26. Beynon CM. Drug use and ageing: older people do take drugs! Age Ageing. 2009;38:8–10
© 2011 Lippincott Williams & Wilkins, Inc.