A CASE-CONTROL STUDY OF ALCOHOL AND SUBSTANCE USE DISORDERS AS RISK FACTORS FOR NON-FATAL INJURY

Abstract

Aims: While alcohol use is thought to be a major risk factor for both fatal and non-fatal injuries, the association of substance use disorders (alcohol use disorders, AUD and substance use disorders, SUD) with occurrence of injury has not received the same attention. To report the association of AUD and SUD, according to diagnostic and statistics manual of mental disorders-IV (DSM-IV) and international classification of diseases 10 (ICD-10) criteria, and the risk of non-fatal injuries. Methods: A case–control study: Cases included 653 injured patients, 18–65-years-old, who attended one emergency department (ED). Controls included 1131 subjects from a representative sample of residents of Mexico City, of the same age group. Information on drug and alcohol use was obtained by interview using the world mental health version of the composite international diagnostic interview (WMH-CIDI). Results: Among injured patients, the prevalence of substance abuse or dependence within the last 12 months was 12.3% for alcohol and 2.5% for other substances (marijuana, cocaine, tranquilizers, amphetamines, others). Among residents of Mexico City, these prevalences were 1.8 and 0.3%, respectively. Adjusted odds ratios (OR) of injury according to alcohol and substance use were 4.95 (95% confidence interval (CI): 2.87–8.52) for alcohol and 2.58 (0.73–9.17) for other substances. An important level of comorbid alcohol and substance use disorders was also found. Conclusions: Efforts in the ED should be carried out to treat and/or refer patients with alcohol and substance use disorders, and special care should be taken to address comorbid cases.

(Received 25 January 2005; first review notified 25 February 2005; in revised form 11 March 2005; accepted 13 March 2005)

INTRODUCTION

While alcohol use is thought to be a major risk factor for non-fatal injuries (Roizen, 1989; Cherpitel, 1993; Romelsjö, 1995), the association of other substance use with the occurrence of non-fatal injury has not received as much attention. Epidemiological studies of drug use and abuse in probability samples of emergency department (ED) patients have found prevalence rates for positive toxicology (excluding alcohol) ranging from 20% among both injured and non-injured ED patients (Rockett et al., 2000) to 45% among injured patients in a level 1 trauma centre (Soderstrom et al., 1997;Cornwell et al., 1998; Rockett et al., 2003). Illicit substance use in the previous 12 months was reported by 19.7% of unintentional injury patients and by 37.1% of intentional injury cases in two Canadian hospitals (Macdonald et al., 1999). Even fewer research exists that documents the role of alcohol use disorders (AUD) and substance use disorders (SUD) in ED populations as a risk factor for non-fatal injuries (Silverman et al., 1985; Cottrol and Frances, 1993; Borges et al., 1998; Cherpitel and Soderstrom, 2000). A study of patients in a level 1 trauma centre found that 28% were positive for a lifetime drug dependence diagnosis while 18% met the criteria for current drug dependence and a rate for current dependence six times higher than that found in the US general population (Kessler et al., 1994). Comorbidity of alcohol and substance use disorders is frequently found in surveys of the general population, including Mexico (Merikangas et al., 1998). A larger prevalence of drug use among those reporting alcohol use and among those meeting a diagnostic criteria for alcohol dependence was reported in an ED (Cherpitel and Borges, 2001), but this study did not report the prevalence of SUD.

In this report, we further explore the relationship between AUD and SUD among ED patients. Our goal here is to report a case–control study of AUD and SUD, according to the international classification of diseases 10 (ICD-10) (World Health Organization, 1992) and diagnostic and statistics manual of mental disorders-IV (DSM-IV) (American Psychiatric Association, 1994) criteria, and the risk of non-fatal injuries in one ED in Mexico City, Mexico.

SUBJECTS AND METHODS

This study is a case–control study (Rothman and Greenland, 2000). Cases included injuries seen in one ED of a general hospital from the south of Mexico City, and controls were a representative sample of residents from Mexico City. The ethics committee of the national institute of psychiatry reviewed and approved this work.

General population sample

The Mexican national comorbidity survey (M-NCS) presented here is part of the WHO-world mental health surveys consortium (Demyttenaere et al., 2004). All interviews were carried out face to face in the homes of respondents by trained lay interviewers. Interviews were conducted after a careful description of the study goals was given and informed consent was obtained. No financial incentive was given for respondents. The M-NCS is based on a stratified, multistage area probability sample of persons aged 18 to 65 years in the non-institutionalized population living in urban areas (population ≥2500) of Mexico. About 75% of the Mexican population has been defined as urban. Data collection took place from September 2001 through May 2002. The response rate for the survey was 76.6%. More details about the methods of this survey are available in Medina-Mora (Medina-Mora et al., 2003). One of the cities selected for the study was the metropolitan area of Mexico City. All results presented below are based on a total of 1131 respondents from Mexico City.

ED survey

A sample of adult patients, ≥18 years, admitted to the ED of a public hospital in the south of Mexico City was drawn from ED admission forms, which reflected consecutive patient arrival in the ED over a six and a half-week period (January and February 2002). All eligible patients from ED logs and medical records seen in the ED during a 24-h period were approached to be interviewed (with informed consent) as soon as possible after admission to the ED. Patients with severe mental disorders and patients in police custody were not included. Patients who were too severely injured to be interviewed in the ED were followed into the hospital and interviewed after their condition had stabilized. No financial incentive was given for respondents. Details of the methods and characteristics of the sample can be consulted elsewhere (Borges et al., 2004).

During the data collection period a total of 744 patients were approached, of whom 39 (5.2%) did not participate. The main reason for non-participation was refusal (2.2%) and patients being transferred before they could be interviewed (1.7%). In order to match the age range of the ED patients with the population sample, we restricted our population of patients to a total of 653 injured cases, aged 18–65 years.

Measures

A series of comparable questions were asked to obtain information on demographic variables and on alcohol and substance use in both surveys. Respondents who reported alcohol use and substance use (marijuana, cocaine, tranquilizers, amphetamines, others) use were further assessed for AUDs and SUDs.

Diagnostic: The instrument used was the world mental health version of the composite international diagnostic interview (WMH-CIDI) (Robins et al., 1988Kessler and Ustun, 2004). In the general population sample, this structured diagnostic interview was administered using a laptop computer version. In the ED sample, all interviews were performed using the paper and pencil version of the same modules of AUDs and SUDs. Our focus here is on either ICD-10 or DSM-IV AUDs and SUDs. For the purpose of this paper, all questions were related to the presence of symptoms during the past 12 months. A patient was considered positive for alcohol dependence if positive in three of the six domains of ICD-10 and/or three of the seven domains of DSM-IV. Patients were considered positive for alcohol abuse if positive on any one of the consequences of alcohol use items in either diagnostic scheme. A parallel series of questions on SUDs were also asked and respondents were classified with substance abuse and substance dependence (SUD) with similar criteria.

Analysis

Data are reported on the association of AUD and SUD with injury in the ED. As a result of the complex sample design and weighting used in the M-NCS, estimates of SEs for proportions were obtained by the Taylor series linearization method using the STATA software (STATA CORP, 2003). Logistic regression analysis (Hosmer and Lemeshow, 2000) was performed to study demographic correlates and the impact of AUD and SUD on injury. Estimates of SEs of odds ratios (ORs) from logistic regression coefficients were also obtained by STATA, and 95% confidence intervals (CI) have been adjusted to design effects.

RESULTS

Table 1 shows the distribution of basic demographic variables among the injured and the population sample from Mexico City. There were large differences between both groups on all variables. Injury cases were more likely to be male, young, single, to have a middle school education, and to be in ‘other’ (peasant, nun, sport player, etc.) employment category and less likely to be a homemaker.

Cases were more likely to have used alcohol and drugs during the past 12 months than the sample of Mexico City residents (Table 2). The injury cases were also more likely to use alcohol frequently (except for alcohol use ‘almost daily’) and to report larger quantities of alcohol consumed. Injury cases also reported a larger number of drugs used and were more likely to report the use of marijuana and cocaine, but not amphetamines/tranquilizers. Finally, the use of both alcohol and illicit drugs were almost eight times more likely to be reported by injury cases than by the sample of Mexico City residents. We further examined the combined association between alcohol use and substance use with injury. Substance use alone was uncommon in the sample of ED patients, since 49 out of the 51 patients who reported any substance use in the past 12 months also used alcohol. A multiple logistic regression that included both current drinker and substance use in the past 12 months, along with demographic variables, resulted in an OR of 2.5 (CI = 2.0–3.3) for current drinker in the past 12 months and 2.1 (CI = 1.2–3.5) for substance use in the past 12 months (data not shown).

The top of Table 3 shows the OR estimates for AUD, and the bottom shows the estimates for SUD. Three models are presented: a univariate (crude) estimate, a multiple model that adjusts AUD or SUD for sociodemographics variables (gender, marital status, age, education and employment) and finally, a model that includes both AUD and SUD controlling for sociodemographic variables. The prevalence of both AUD and SUD were much higher among the injured than among the residents of Mexico City. Univariate estimates for both AUD and SUD showed large OR estimates for the association with injuries (7.6 for AUD and 8.9 for SUD) that decreased when adjusted by sociodemographic variables (5.4 and 5.5, in the same order), but are still very large (Model A). Finally, when both AUD and SUD were considered simultaneously, more than half of the effect of the association between SUD and injuries was owing to comorbid AUD (Model B). In this model, the OR for SUD dropped from 5.5 to 2.6. On the other hand, the OR for AUD showed only a small change when comorbid SUD was controlled, decreasing from 5.4 to 4.9.

DISCUSSION

In this study, we found a greater prevalence of alcohol and substance use among cases of injury than among the residents of Mexico City. The most conservative estimate suggests that AUD may increase the odds of an injury by 4.9 times, and SUD may increase the odds of injury by 2.6 times. Smaller but similar effects were seen for the association between a 12-month alcohol use and substance use with injury (OR = 2.5 and 2.1, for alcohol and substance use).

The prevalence of SUDs has been rarely reported in ED populations (Cottrol and Frances, 1993). A larger ED study reported a prevalence of 17.7% of DSM-III-R current (6 months) drug use diagnoses and a large current comorbid prevalence of alcohol and drug use disorders (8.3%) (Soderstrom et al., 1997). A study conducted in seven Tennessee general hospitals EDs, found that 4% of patients met the DSM-IV current drug dependence criteria and 27% were assessed as requiring substance abuse treatment (Rockett et al., 2003). We found a lower prevalence of drug use disorders in this sample of injured patients, possibly owing to the lower baseline prevalence of substance dependence in Mexico when compared with the US (Vega et al., 2002) and, in part, on account of this being a level 1 trauma centre.

The prevalence of drug use in combination with alcohol has also been found not to be inconsequential, with rates in probability samples of ED patients ranging from 16% (Soderstrom et al., 1997) to 22% (Cornwell et al., 1998). A positive screen for either alcohol or other drugs has been found to be associated with a 40% higher rate of positivity for the other substance in selected ED samples (Rivara et al., 1989; Buchfuhrer and Radecki, 1996), and rates for drug use alone have been found to be lower than for drug use in combination with alcohol across all classes of drugs tested (Madan et al., 1999). In this sample of ED patients drug users were usually also heavy drinkers.

A previous study from our group among injured patients in three EDs in Pachuca found an OR of 3.25 for any drug use and an OR of 2.88 for alcohol use disorders (Borges et al., 1998). Both estimates are comparable with the ones of the current report. A previous US study among Mexican Americans reported an OR for injury (compared with non-injury) of 1.66 for any drug use in the past year among ED patients (Cherpitel and Borges, 2001). In addition, alcohol use and alcohol use disorders showed larger ORs than substance use and substance use disorders, when both variables were considered together.

Limitations

A major limitation is the use of self-reported alcohol and substance use data, and the retrospective nature of the CIDI questions. However, previous studies in the ED in Mexico have found that self-reported acute alcohol has good agreement with breath test (Cherpitel et al., 1992), and although there are no data in EDs in Mexico for concordance between self-reported substance use and biological specimens, studies among Mexican migrants in the US have shown good agreement (Vega et al., 1997). Validity of the CIDI for AUDs and SUDs has been shown to be adequate in the international context, including Mexico (Vega et al., 2002), but ED patients maybe more likely to deny the use of illicit substances (Hser et al., 1999). If the latter happened in our study, our calculations of ORs for substance use would be underestimated. Further research on the validity of self-reported substance use and SUDs in the context of the ED is an important step for future research.

Generalizability of results

A single ED facility was used and although injury cases in our sample are representative of this facility, they may not represent the full spectrum of ED patients from Mexico City. Controls used for this comparison were sampled from the greater Mexico City area, and it is possible that residents from the catchments area served by this ED may report differences in alcohol and substance use than reported for the entire city. Since the M-NCS reported no differences in the 12-month prevalence of both AUDs and SUDs across the six regions of Mexico (Medina-Mora et al., 2003), it seems unlikely that there are differences across smaller areas in Mexico City alone.

CONCLUSIONS

Our data suggests that alcohol and substance use disorders were important risk factors for injuries in Mexico. Comorbid AUDs and SUDs were common, but alcohol was more important than drug use in injury. Efforts in the ED should be carried out to treat and/or refer patients with AUDs and SUDs, and special care should be taken to address comorbid cases. AUDs and SUDs are strong risk factors for non-fatal injury among ED patients, when compared with a sample of the general population.

Large levels of comorbid AUDs and SUDs were reported, and AUDs showed a larger involvement with injury than SUDs.

This paper is partially based on the data and experience obtained during the WHO Collaborative Study on Alcohol and Injuries — a project, sponsored by the World Health Organization (http://www.who.int/substance_abuse/activities/injuries/en/) We appreciate the efforts and support from the authorities, patients and the staff of the Hospital Dr Manuel Gea González.

This study was supported by the Consejo Nacional de Ciencia y Tecnología (CONACyT) (39607-H) and a National Alcohol Research Centre grant AA 05595–12 from the U.S. National Institute on Alcohol Abuse and Alcoholism. Data from the Emergency Department was collected under the World Health Organization (Geneva) and the National Institute of Psychiatry Ramon de la Fuente (4275P) (Mexico). Support for the Mexican National Comorbidity Survey (M-NCS) data collection came from The National Institute of Psychiatry Ramon de la Fuente (INPRFM-DIES 4280) and by the National Council on Science and Technology (CONACyT-G30544-H), with supplemental support from the Pan American Health Organization (PAHO).

REFERENCES

Author notes

1Metropolitan Autonomous University-Xochimilco, Mexico City, Mexico

2Division of Epidemiological and Psychosocial Research, National Institute of Psychiatry, Calz. Mexico-Xochimilco No.101, Col. San Lorenzo Huipulco, CP.14370, Mexico

3Alcohol Research Group, Berkeley CA, USA