Volume 23
Number 4
2002

[Table of Contents]

Do hospital E-codes consistently capture suicidal behaviour? 

Anne E Rhodes, Paul S Links, David L Streiner, Ian Dawe, Dan Cass and Samantha Janes 

Abstract 

Hospital separation data are used to study suicidal behaviour; however, there is little information about the appropriateness of these data for research and planning activities. The study purpose is to examine how consistently hospital separation E-code data reflect suicidal behaviours. Expert clinicians reviewed medical records of individuals who had a separation for self-poisoning to determine whether the self-poisoning was deliberate. Agreement among clinicians was evaluated and latent class analysis performed to derive a summary estimate of the prevalence of deliberate self-poisoning. This estimate was then compared to the prevalence of deliberate self-poisoning based on the external cause of injury (E-codes). Clinicians estimated the prevalence to be 63% higher than the E-code- based prevalence. Much larger discrepancies were apparent among older age groups, those whose care was primarily medical in nature and those with a longer length of hospital stay. In acute care settings, self-poisonings among the elderly may not receive adequate attention and/or documentation. Estimating the prevalence of admissions for suicidal behaviour using hospital separation data is of questionable validity, particularly among older age groups. 

Key words: hospitalization; poisoning reproducibility of results; suicide, attempted 

Introduction 

Suicidal behaviours are a serious public health problem contributing to morbidity, lost productivity, health care costs and premature mortality.^1–4 The impact on significant others can be devastating. In the year after an attempted suicide, 10–15% of individuals will make a repeat attempt and 1% will die. Within 10 years, about 3–10% will die by suicide. Continuous, population-based monitoring of suicidal behaviours, however, is rare.⁵ Nevertheless, hospital separation data are collected systematically in many countries and contain information about whether an injury is self-inflicted, characteristics of the method, and whether the person died during the hospital stay. A number of countries have used these data for descriptive and analytic studies of suicidal behaviours: the United States;^6–8 Canada;^1,9,10 New Zealand;¹¹ Australia^12,13 and the United Kingdom.^14–19 At issue is the paucity of information on the appropriateness of these data for population-based research and planning activities. The purpose of this study is to examine how consistently hospital separation data reflect suicidal behaviour within a universally insured Canadian setting. 

Some jurisdictions report that suicidal behaviours are one of the most common causes of acute care admissions,¹⁶ particularly among adolescents.²⁰ Mean annual prevalence estimates for these admissions range from 61 per 100,000 in the United States⁶ to approximately 200 per 100,000 in the United Kingdom.^16–18 Over 90% of these admissions are self-poisonings as opposed to injuries.^12,16 

Although wide variations in admitting practices occur in relation to suicidal behaviour,^16,21,22 hospital admission predicts future suicidal behaviour. Repetition often occurs within the first few months.²³ Within a year, approximately 10% will be readmitted after another attempt.^16,19 Deaths from all causes are elevated in this group. Mortality rates are about 2–4 times higher than in the general population.^8,9,19,20,23 The time spent in hospital and the post-discharge period, therefore, offer an opportunity to intervene and prevent further morbidity and mortality. 

Hospital separations connected to deliberate self-harm are identified by the presence of an external cause of injury or poisoning code, an “E-code”, according to the International Classification of Disease system (ICD). E-codes designate whether the injury or poisoning was “accidental”, “deliberate” or “cause undetermined”. Because of the sensitive, stigmatizing nature of suicidal behaviours and lack of knowledge about how to best treat these individuals, recording and practice patterns may vary.^24,25 E-codes are identified as “supplemental” within ICD-9 and their application is not mandatory in all settings.²⁶ In the United States, only about half of the states routinely collect E-coded hospital discharge data.⁶ Financial disincentives may be a reason if E-codes are not reimbursable.^27,28 If payment for treatment is tied to psychiatric or medical diagnoses where established guidelines exist, there would be little incentive to implement E-coding. In managed care settings, patients and providers may be reluctant to report/record these behaviours due to a possible loss of insurance benefits.⁶ Payment for admissions for suicidal behaviour have been disallowed retrospectively by managed care organizations.²⁹ 

In one study, conducted in Oxford, England, only 20% of 500 known cases of deliberate self-poisoning among 10–20-year-olds were captured as deliberate in the corresponding hospital separation records.³⁰ In contrast, in a study of adult subscribers to a health maintenance organization in California, chart reviews confirmed that 86% of hospitalizations assigned “deliberate” E-codes were suicide attempts.⁸ The potential for misclassification by relying on “deliberate” E-codes has been acknowledged and definitions of suicidal behaviour expanded by including “accidental” self-poisonings related to medications (E-codes: 850–859) and/or “undetermined” (E-codes 980–989).^15,16,18 

Widespread acceptance and application of these data for the purposes of monitoring the burden of suicidal behaviours and the effectiveness of interventions in reducing these behaviours within populations would seem to be premature and potentially harmful. For instance, if there is sizable under-reporting of these behaviours and/or systematic biases about who is affected, relying on these methods could lead to an inappropriate allocation of resources. On the other hand, the level of error in the data may be reasonable for some populations and could be an underutilized valuable resource. As more evidence comes in about beneficial treatments, and standards of practice are solidified, hospitals and/or communities might be able to use hospital separation data for quality assurance purposes. Those involved in the production and utilization of these data for research and planning purposes have a shared responsibility to the public to insure the data are managed and interpreted appropriately. 

Materials and methods 

Study setting and sample 

The setting for this study was an urban, teaching hospital that specializes in the care of trauma patients in a large Canadian city. In Canada, when a person is discharged from hospital, trained hospital personnel review the medical record for the index admission and assign discharge diagnoses according to standardized coding practices. There are 16 potential diagnostic fields. Whenever a principal or “N-code” diagnosis of self-poisoning or injury is present, an E-code must be applied.³¹ As hospital-based care in Canada is fully covered under provincial health insurance plans, financial disincentives in applying E-codes are less likely. In the study hospital, emergency and psychiatric staff are regularly exposed to persons who are admitted to hospital after a suicide attempt. Accordingly, this setting was chosen to represent a best-case scenario. Our reasoning was that if the data were inconsistently applied in this setting, then the problem is probably as likely or more prevalent elsewhere in Canada. 

This study received ethical approval from the Research Ethics Board of the participating hospital. In the fiscal year 1998/1999, there were 274 initial hospital separations for individuals that included a diagnosis of self-poisoning (ICD-9-CM 969–989) in at least one of the diagnostic fields. Most (89%) were poisonings by drugs, medicinal and biological substances. One hundred and four (38%) had self-inflicted injury codes assigned (E950–E959), and 21 (7.7%) had undetermined codes (E980–E989). For the purposes of this study, we selected a random sample of 181 separations from the original 274 (66.1%). 

Study measures and procedures 

In order to examine how consistently hospital separation data reflect deliberate self-poisonings, a definitive method is desired for comparison purposes. Regrettably, there is little consensus in the literature about how to define suicidal behaviours, and definitions vary across research studies.^32–34 Even when definitions are provided, often little or no information is given about who made the decisions and the reliability of the decisions made.³⁵ In the absence of a definitive method, we derived a best estimate of deliberate self-poisoning based on expert clinical judgement and latent class analyses (LCA). Latent class analysis answers the questions, “What would the diagnostic accuracy of the raters and the probability of deliberate self-poisoning need to be to produce the patterns of agreement and disagreement observed?”³⁶ 

To estimate the presence or absence of a latent class three or more ratings are necessary.³⁷ Accordingly, we conducted an inter-rater reliability study among three well trained physicians in emergency psychiatry and emergency medicine to estimate the latent class of deliberate self-poisoning. For each of the index admissions, the chart was copied and blinded in terms of identifying information and hospital separation codes by a research assistant. The blinded copies were then distributed to the physicians. The physicians examined each of the medical records for the stay in question, including records from the emergency department, and independently rated whether they believed the self-poisoning was deliberate or not. They then returned the materials to the research assistant for data entry. 

We also abstracted data often used in research and planning, readily available and reasonably accurate: age, sex and the length of hospital stay (LOS). To differentiate between those whose care was primarily medical in nature as opposed to psychiatric, we identified whether the most responsible physician was a psychiatrist and whether the most responsible diagnosis corresponded to an ICD-9 mental disorder (codes: 290–319).^38–41 

Statistical analyses 

Sample characteristics 

To examine whether the proportion of the select subject characteristics in our sample of 181 differed from the original 274, we conducted Chi Square tests. This analysis was repeated to determine whether subjects assigned a self-inflicted injury E-codes (n = 66) differed from those who were not (n = 115). 

Ratings between physicians 

The level of agreement between each physician pair was assessed according to the percent agreement, Kappa statistic and corresponding 95% confidence intervals. Kappa statistics were interpreted in relation to the guidelines of Landis & Koch.⁴² Using a maximum likelihood method⁴³ we estimated the following parameters and corresponding standard errors: the prevalence of the latent class of deliberate self-poisoning, and the false positive and the false negative rates of each physician overall. 

Consistency in the prevalence estimates of deliberate self-poisoning 

The proportions of deliberate self-poisoning (self-inflicted E-code vs. LCA) were compared overall and according to select sample characteristics. 95% confidence intervals around these estimates were calculated and overlap assessed.⁴⁴ 

Results 

Sample characteristics 

The sample of 181 did not differ from the group of 274 in terms of the subject characteristics (Table 1) or the proportion who died during their hospital stay (9%). The sample subjects tended to be middle-aged (mean age 49.4 years; SD ± 19.6). The diagnoses most responsible for the length of hospital stay (in the first diagnostic field) were typically medical in nature. Physicians most responsible for the care given during these hospital stays were usually not psychiatrists (82%). The mean length of stay (LOS) varied widely (mean LOS 16.6 days; SD ± 46.4). Approximately 25% of subjects had stays of two days or less; 50% with five days or less, and 75% 10 days or less. Two had a LOS that neared the one-year mark or exceeded it. 

When the sample was divided in terms of whether the E-code assigned identified the self-poisoning as deliberate or not, a different picture emerged (Table 2). Those identified as having deliberately poisoned themselves were more likely to be younger, have a shorter LOS, have a mental disorder as the most responsible diagnosis and to have a psychiatrist as the most responsible physician. 

Ratings between physicians 

Agreement beyond chance between each of the physician pairs was excellent and did not vary between physician pairs. In contrast to the other raters, Rater 3 identified fewer poisonings as deliberate in nature, i.e., this rater was less sensitive. The overall sensitivity and specificity of each of the raters was high ranging from 86.3% to 98.9%. 

Consistency in the prevalence estimates of deliberate self-poisoning 

The E-codes indicated that 36.5% of the self-poisonings were deliberate in nature (95% CI: 30%; 43%). In contrast, the LCA indicated that 59.5% were deliberate (95% CI: 50%; 70%). In comparison to the E-code estimate, the LCA estimate was 63% higher. 

Higher estimates of deliberate self-poisoning according to the LCA were particularly evident in certain subgroups. LCA estimates were about two to seven times higher than the comparable E-code ones in those aged 55 years or more, those whose most responsible diagnoses were medical in nature and those who had a longer LOS (Table 4). LCA estimates of deliberate self-poisoning were about 90% lower among those whose care was largely psychiatric in nature. 

Discussion 

In an urban, teaching hospital in Canada, clinicians estimated the prevalence of deliberate self-poisoning to be 63% higher than the prevalence as determined by E-codes among separations for self-poisoning. Much larger discrepancies were apparent among older age groups, those whose care was primarily medical in nature and those with a longer LOS. 

Before discussing the results in detail, some interpretive cautions are necessary. We selected a sample based on whether a diagnosis of self-poisoning occurred anywhere in the 16 possible diagnostic fields. Based on previous abstraction studies, coding for self-poisoning at the three-digit level was likely quite accurate.⁴⁵ Compared with other settings, our sample may have overrepresented persons with more lethal behaviours even though the sample was limited to self-poisonings effectively excluding trauma admissions. Our sample contained only a small number of subjects under age 25 and overall, men and women did not differ according to E-code or clinician-based estimates of deliberate self-poisoning. Most study samples have contained more women than men,^1,9,19,23 and in some, younger women (15–24 years of age) predominated.^1,6,8.13,17. However, eventual suicide was more likely in men and those in older age groups.^9,19 

Clinical relevance 

From a clinical standpoint, the findings are perhaps most relevant to hospital settings where older persons are medically treated for self-poisoning. A higher degree of suspicion about the presence of suicidal behaviours in admissions for self-poisoning and improved documentation may be necessary. Better screening and integration of medical and psychiatric care during and after the hospital stay may prevent future suicide attempts or completions. Alternatively, the quality of the care provided may be excellent but insufficient documentation of suicidal behaviours⁴⁶ contributes to missing E-code data or ambiguity noted in the elderly.²⁸ 

In this study, the prevalence of deliberate self-poisoning among younger persons was about 50%, whether E-codes or LCA methods were employed. This estimate did not change appreciably when the most responsible diagnosis of self-poisoning N-code (regardless of the nature of the E-codes) was applied. These findings suggest that in younger age groups, the prevalence of hospitalizations for deliberate self-poisoning may be estimated with more consistency than in older age groups in specific settings. This is further reinforced by the study of 10–20-year-olds in Oxford where nearly all hospitalizations for self-poisonings were verified as being deliberate. Nevertheless, only 20% of these hospital separations were deemed “deliberate” according to the E-codes³¹ in contrast to the near 50% in our study. Irribarren et al. confirmed 86% of the hospitalizations within a health maintenance organization, however it is not known whether “out of plan” use by various age/sex groups bear upon this estimate. 

Public health relevance 

From a public health standpoint, estimating prevalence rates of persons who are admitted to hospital for deliberate self-poisoning based on “deliberate” E-codes is of questionable validity, particularly among older age groups. Depending upon the underlying population structure and admitting/referral practices in a region, deliberate self-poisonings could be affected dramatically by relying only on these data. Time trend studies may be affected by changes in the underlying population over time but also by changes in coding practices. For example, E-coding may have improved over time in younger populations due to the heightened media exposure concerning suicide in youths. Reports that parasuicide admissions for men^14,47 particularly younger men¹⁸ have increased over time based on “deliberate” E-code data are suspect and could deflect attention away from the elderly. 

Clearly, the elderly need to be monitored given their greater access to medications in general,⁴⁸ greater fatality of behaviours⁴⁹ and the potential for an increasing burden arising from an aging population.⁵⁰ Elderly persons may be more likely to use prescribed drugs in suicide^17,49 and physicians may find themselves in the “unenviable position of having unwittingly prescribed the drugs used”.¹⁷ Prevention of further morbidity and mortality may be achieved through clinical and population based approaches concerning access to medications.^48,51–53 

References 

1.    Langlois S, Morrison P. Suicide deaths and suicide attempts. Health Reports 2001; 13(2): 9–22. 

2.    Clayton D, Barceló A. The cost of suicide mortality in New Brunswick, 1996. Chronic Dis Can 1999; 20(2):1–10. 

3.    Jenkins R. Nations for mental health. Soc Psychiatry Psychiatr Epidemiol 1997; 32: 309–311. 

4.    Satcher D. National Strategy for Suicide Prevention: Goals and Objectives for Action. U.S. Department of Health and Human Services PHS, editor. 2001. Rockville, Maryland. 

5.    National Health Centre for Reviews and Dissemination. Effective Health Care. Deliberate Self-Harm. 4[6], 1–11. 1998. University of York. 

6.    Spicer RS, Miller TR. Suicide acts in 8 States: Incidence and case fatality rates by demographics and method. Am J Public Health 2000; 90(12):1885–1891. 

7.    van Alast J, Shotts SD, Vitsky JL, Bass SM, Miller RS, Meador KG et al. Long-term follow-up of unsuccessful violent suicide attempts: risk factors for subsequent attempts. J Trauma 1992; 93(3):457–464. 

8.    Iribarren C, Sidney S, Jacobs DR Jr, Weisner C. Hospitalization for suicide attempt and completed suicide: epidemiological features in a managed care population. Soc Psychiatry Psychiatr Epidemiol 2000; 35:288–296. 

9.    Holley HL, Fick G, Love EJ. Suicide following an inpatient hospitalization for a suicide attempt: a Canadian follow-up study. Soc Psychiatry Psychiatr Epidemiol 1998; 33:543–551. 

10.    Neutel CI, Patten SB. Risk of suicide attempts after benzodiazepine and/or antidepressant use. Ann Epidemiol 1997; 1997(7):568–574. 

11.    Gould MS, Petrie K, Kleinman MH, Wallenstein S. Clustering of attempted suicide: New Zealand national data. Int J Epidemiol 1994; 23(6):1185–1189. 

12.    Watt GM, Ozanne-Smith J. Non-fatal injuries to young Victorians 1986–1991. Med J Aust 1994; 160:790–794. 

13.    Li L, Ozanne-Smith J. Injury hospitalization rates in Victoria, 1987–97: trends, age and gender patterns. Aust NZ J Public Health 2000; 24(2):158–165. 

14.    Stark C, Smith H, Hall D. Increase in parasuicide in Scotland. BMJ 1994; 308: 1569–1570. 

15.    Gunnell DJ, Peters TJ, Kammerling RM, Brooks J. Relation between parasuicide, suicide, psychiatric admissions, and socioeconomic deprivation. BMJ 1995; 311: 226–230. 

16.    Gunnell DJ, Brooks J, Peters TJ. Epidemiology and patterns of hospital use after parasuicide in the south west of England. J Epidemiol Community Health 1996; 50: 24–29. 

17.    Smith T. Differences between general practices in hospital admission rates for self-inflicted injury and self-poisoning: influence of socioeconomic factors. Br J Gen Pract 1995;458–462. 

18.    McLoone P, Crombie IK. Hospitalisation for deliberate self-poisoning in Scotland from 1981 to 1993: trends and rates and types of drugs used. Br J Psychiatry 1996; 169:81–5. 

19.    Hall DJ, O'Brien F, Stark C, Pelosi A, Smith H. Thirteen-year follow-up of deliberate self-harm, using linked data. Br J Psychiatry 1998; 172:239–242. 

20.    Sellar C, Hawton K, Goldacre MJ. Self-poisoning in adolescents. Hospital admissions and deaths in the Oxford Region 1980–85. Br J Psychiatry 1990; 156(866):870. 

21.    Kapur N, House A, Creed F, Feldman E, Friedman T, Guthrie E. Management of deliberate self-poisoning in adults in four teaching hospitals: descriptive study. BMJ 1998; 316:831-832. 

22.    Thomas SHL, Lewis S, Bhattacharyya S, Bramble MG, Chew K, Connolly J et al. Factors affecting hospital admission and length of stay of poisoned patients in the North East of England. Human and Experimental Toxicology 1996; 15(915):919. 

23.    Goodacre M, Hawton K. Repetition of self-poisoning and subsequent death in adolescents who take overdoses. Br J Psychiatry 1985; 146:395–398. 

24.    Dennis M, Beach M, Evans PA, Winston A, Friedman T. An examination of the accident and emergency management of deliberate self-poisoning. J Accid Emerg Med 1997; 14:311–315. 

25.    Hawton K, Arensman E, Townsend E, Bremner S, Feldman E, Goldney R et al. Deliberate self harm: systematic review of efficacy of psychosocial and pharmacological treatments in preventing repetition. BMJ 1998; 317:441–447. 

26.    Vespe Ozonoff V, Tan-Torres S, Barber CW. Assessment of E-coding practices and costs in Massachusetts hospitals. Public Health Rep 1993; 108(5):633–636. 

27.    Sniezek JE, Finklea JF, Gritcer PL. Injury coding and hospital discharge data. JAMA 989; 262(16):2270–2272. 

28.    Langlois JA, Buechner JS, O'Connor EA, Nacar EQ, Smith GS. Improving the E-coding of hospitalizations for injury: do hospital records contain adequate documentation? Am J Public Health 1995; 85(9):1261–1265. 

29.    Hall RCW, Platt DE, Hall RCW. Suicide risk assessment: A review of risk factors for suicide in 100 patients who made severe attempts. Evaluation of suicide risk in a time of managed care. Psychosomatics 1999; 40(1):18–27. 

30.    Sellar C, Goldacre MJ, Hawton K. Reliability of routine hospital data on poisoning as measures of deliberate self-poisoning in adolescents. J Epidemiol Community Health 1990; 44:313–315. 

31.    Canadian Institute for Health Information. Abstracting Manual. 1999. Ottawa, Ontario. 

32.    O'Carroll PW, Berman AL, Maris RW, Moscicki EK, Tanney BL, Silverman MM. Beyond the Tower of Babel: A nomenclature for suicidology. Suicide and Life-Threatening Behavior 1996; 26(3):237–252. 

33.    Arensman E, Kerkhof AJFM. Classification of attempted suicide: a review of empirical studies, 1963-1993. Suicide and Life-Threatening Behavior 1996; 26(1):46–67. 

34.    Leenaars AA, De Leo D, Diekstra RFW, Goldney RD, Kelleher MJ, Lester D et al. Consultations for research in suicidology. Archives of Suicide Research 1997; 3: 139–151. 

35.    Rhodes AE, Links PS. Suicide and Suicidal Behaviours: Implications for Mental Health Services. Can J Psychiatry 1998; 43(8): 785–791. 

36.    Uebersax JS, Gove WM. Latent class analysis of diagnostic agreement. Stat Med 1990; 9:559–572. 

37.    Uebersax JS. Latent class analysis of substance abuse patterns. [142], 64–80. 1994. NIDA Research Monograph. 

38.    Jarman B, Hirsch S, White P, Driscoll R. Predicting psychiatric admission rates. BMJ 1992; 304:1146–1151. 

39.    Tansella M, Bisoffi G, Thornicroft G. Are social deprivation and psychiatric service utilisation associated in neurotic disorders? A case register study in South Verona. Soc Psychiatry Psychiatr Epidemiol 1993; 28: 225–230. 

40.    Mustard CA, Derksen S, Tataryn D. Intensive use of mental health care. Can J Psychiatry 1996; 41(93):101. 

41.    Koppel S, McGuffin P. Socio-economic factors that predict psychiatric admissions at a local level. Psychol Med 1999; 29: 1235–1241. 

42.    Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159–174. 

43.    Walter SD. Measuring the reliability of clinical data: the case for using three observers. Rev Epidemiol et Santé Publ 1984; 32: 206–211. 

44.    Fleiss J. Statistical methods for rates and proportions. 2 ed. Toronto: John Wiley and Sons, 1981. 

45.    Macintyre CR, Ackland MJ, Chandraraj EJ. Accuracy of injury coding in Victorian hospital morbidity data. Aust NZ J Public Health 1997; 21(7):779–783. 

46.    Malone KM, Szanto K, Corbitt EM, Mann JJ. Clinical assessment versus research methods in the assessment of suicidal behavior. Am J Psychiatry 1995; 152(11): 1601–1607. 

47.    Beck P, Scorer R, Lazarus J, Smith P, Routledge P. Increasing rates of deliberate self-poisoning. BMJ 1994; 308:789. 

48.    Tamblyn RM, Jacques A, Laprise R, Huang A, Perreault R. The office of the future project: the integration of new technology into office practice. Academic detailing through the super highway. Quebec Research Group on Medication Use in the Elderly. Clinical Performance and Quality Health Care 1997; 5(2):104–108. 

49.    De Leo D, Padoani W, Scocco P, Lie D, Bille-Brahe U, Arensman E et al. Attempted and completed suicide in older subjects: results from the WHO/EURO Multicentre Study of Suicidal Behaviour. Int J Geriatr Psychiatry 2001; 16(3):300–310. 

50.    Statistics Canada. Population projections for Canada, provinces and territories 2000–2026. 91-520-XPB. 2001. Ottawa, Ontario. 

51.    Buckley NA, Whyte IM, Dawson AH, McManus PR, Ferguson NW. Correlations between prescriptions and drugs taken in self-poisoning. Implications for prescribers and drug regulation. Med J Aust 1995; 162:194–197. 

52.    Tamblyn R, Laprise R, Hanley JA, Abrahamowicz M, Scott S, Mayo N et al. Adverse effects associated with prescription drug cost-sharing among poor and elderly residents. JAMA 2001; 285(4): 421–29. 

53.    Crombie IK, McLoone P. Does the availability of prescribed drugs affect rates of self-poisoning? Br J Gen Pract 1998; 48(433):1505–1506.

Author References 

Mark Daniel, Department of Epidemiology and Department of Health Behaviour and Health Education, School of Public Health, The University of North Carolina at Chapel Hill, North Carolina, USA 

Lynne C Messer, Department of Health Behaviour and Health Education, School of Public Health, The University of North Carolina at Chapel Hill, North Carolina, USA. 

Correspondence: Mark Daniel, School of Public Health, The University of North Carolina at Chapel Hill, CB #7440, Rosenau Hall, Room 302, Chapel Hill, North Carolina 27599-7440, USA; Fax: (919) 966-2921; E-mail: danielm@email.unc.edu 

[Previous] [Table of Contents] [Next]