Article Text
Abstract
We performed a nested convenience sample survey of harmful alcohol use among injured patients aged 18 years and older treated in the Komfo Anokye Teaching Hospital (Kumasi, Ghana) emergency department (ED). Data from the Alcohol Use Disorder Identification Test, alcohol breath or saliva tests, patient demographics and injury characteristics were collected from an administered survey and medical chart review. A total of 403 subjects were surveyed, and 107 (27%; 95% CI 22 to 31) reported harmful alcohol use. High rates of harmful alcohol use were found among males (35%), acutely alcohol-positive subjects (55%), drivers (32%), pedestrians (35%) and assault victims (43%). A substantial proportion of injured patients reported harmful alcohol use. The data obtained support routine screening of injured patients presenting to Ghanaian EDs for harmful alcohol use.
Statistics from Altmetric.com
Introduction
Injuries are responsible for 17% of the disease burden among adults globally with alcohol attributable for 15–20% of mortality.1 ,2 Risk of injury is increased not only by the volume, but pattern of alcohol consumption.3 Assessment of alcohol use characteristics among injured patients has been limited in large regions of the world, including Africa.4 ,5 We recently found that 35% of injured adults presenting to a Kumasi, Ghana, emergency department (ED) had positive alcohol breath tests or saliva tests.6 Using subject self-report, the WHO describes below-average per capita alcohol consumption in Ghana compared with the WHO Africa region.2 It estimates that 3.3% of adults in Ghana have an alcohol use disorder, including harmful alcohol use. Studies from the USA have found that injured patients presenting to the ED have a higher frequency of harmful alcohol use than the general population.7 They represent a high-risk population that could be targeted for interventions to prevent the harmful use of alcohol. However, there has been no assessment of self-reported alcohol use characteristics among injured patients in Ghana or the West African region.
On this premise, we conducted a study to assess the frequency and characteristics of harmful alcohol use, as determined by the Alcohol Use Disorder Identification Test (AUDIT), among injured adult patients presenting to the Komfo Anokye Teaching Hospital (KATH) ED in Kumasi, Ghana.
Materials and methods
We performed a nested convenience sample survey within a cross-sectional chart review of consecutive injured adult patients presenting to the KATH ED from 3 November 2014 through 11 April 2015. The objective of the cross-sectional study was to determine the frequency of positive alcohol tests among these patients.
All patients presenting to the ED are triaged using the South African Triage Scale (SATS) to assign urgency of care needed. SATS is a validated five-level triage acuity scale that assigns each patient a colour-coded acuity tier: (i) red—emergency; (ii) orange—very urgent; (iii) yellow—urgent; (iv) green—not urgent; and (v) blue—dead on arrival.8–10 Patients triaged green are referred to a separate clinic for further care. Triage nurse activities included alcohol testing for patients 18 years and older who present within 8 hours of injury and are managed in the ED.
The study's sample size was determined in order to assess the proportion of harmful alcohol use with a margin of error at most 5% for all surveyed subjects (95% CI). We calculated the sample size by assuming the true frequency of harmful alcohol use was 50%, which resulted in a sample size of 400 subjects.
Patients aged 18 years or older admitted to the KATH ED (ie, triaged yellow, orange or red) with injuries were included in the chart review if the injury occurred ≤8 hours from the time of alcohol testing in triage. Patients were excluded from the chart review if they had copious blood in the mouth that interfered with performing the alcohol test, very minor injuries (triaged green) or death prior to admission (triaged blue). The nested survey was administered to subjects included in the chart review if the subject was medically stable, clinically non-intoxicated and had capacity for informed consent.
Full-time trained research assistants collected medical record data from all patients included in the chart review using Epi Info 7 software (Georgia, USA) on computer tablets. Research assistants were proficient in both English and Twi, the local language. They identified patients for chart review by reviewing the triage admission book for chief complaint and alcohol-testing notations. Research assistants collected the following chart review variables from triage and physician sheets: alcohol result, date of admission, time of admission, prehospital time (time from injury to ED presentation), age, sex, SATS colour, cause of injury, intent of injury and ED disposition. The alcohol result was measured by the triage nurse with either a SureScreen Alcometer Breathalyzer (Derby, England) or an ALCO-Screen saliva alcohol test strip (Indiana, USA). Cause and intent of injury were defined by the promulgated CDC International Classification of Diseases, tenth revision, injury matrix.11
After collecting medical record data, the research assistants administered the survey to eligible and consenting subjects daily from 08:00 to 18:00. Most patients admitted during non-survey hours remained in the ED until research assistants were present the following day, allowing for sampling of subjects over 24 hours. The survey was administered orally due to significant illiteracy rates.12 The survey collected the following variables: religion (Christian, Muslim, other), current marital status (married or not married), education (junior high school or less, secondary school or greater) and the AUDIT. The AUDIT questionnaire is a validated measure of harmful alcohol use across diverse populations.7 ,13 ,14 Prior to performing the survey, we trained the research assistants to determine standard drinks (10 g of ethanol) for traditional Ghanaian drinks using estimated alcohol concentrations.15 We also calculated the number of standard drinks for commonly sold volumes of alcohol, such as a bottle of beer or a sachet of liquor. If a patient denied alcohol use, the research assistant would follow-up with specific questions about alcohol-containing beverages that may not be traditionally considered alcoholic.
Regarding excluded patients, a study investigator collected the following variables from the triage admission book: date and time of admission, sex, age, SATS colour and cause of injury.
Data were analysed using Stata V13 (College Station, Texas, USA). After the survey, the presence of harmful alcohol use was dichotomised by calculating AUDIT scores, with a score of ≥8 indicating harmful use.14 The primary measure of interest was harmful alcohol use among all surveyed subjects, with secondary measures pertaining to proportions of harmful alcohol use among age, sex, religion, marital status, highest education attained, acute alcohol use, cause and intent of injury, SATS colour and disposition. Proportions with 95% CIs were used for categorical variables and medians with IQRs performed on continuous variables.
In order to examine selection bias, we described time of admission, sex, age, SATS colour and cause of injury for surveyed patients, patients included in chart review only and patients excluded from chart review. Prehospital time and disposition were described for survey and chart review-only populations.
Results
A total of 2488 injured patients entered the sampling frame, with 1085 patients meeting criteria for chart review (43% of all patients) (figure 1). In total, 522 subjects were approached for administration of the survey. Of this number, 403 subjects were surveyed and included in this data analysis, 92 met exclusion criteria and 27 did not provide consent. Excluded patients and chart review-only subjects were not markedly different from those surveyed with respect to time of admission, age, sex and cause of injury (table 1). SATS colour distribution for surveyed subjects showed greater proportion triaged yellow compared with the other two populations. Notable differences in proportions between the surveyed and chart review-only subjects were observed for acute alcohol positivity, assaults and deaths in the ED.
Overall, 107 patients (27%; 95% CI 22 to 31) demonstrated harmful alcohol use and 192 patients (48%; 95% CI 43 to 53) endorsed consuming alcohol in the past 12 months. Table 2 describes the frequency of harmful alcohol use among all subjects and by subgroups of subject and injury characteristics. High rates of harmful alcohol use were found among males, subjects aged 25–54 years, those acutely alcohol positive, drivers, pedestrians and assault victims.
Table 3 compares subjects with and without harmful alcohol use. Subjects with harmful alcohol use were more often male (32% difference; 95% CI 24 to 40) and acutely alcohol positive (45% difference; 95% CI 35 to 55).
In total, 163 patients (40%; 95% CI 36 to 45) demonstrated either harmful alcohol use and/or acute alcohol positivity. Also, 38 of 107 (36%) subjects who reported harmful drinking tested negative for alcohol. And, 56 (45%) of 125 subjects testing positive for alcohol were not positive for harmful alcohol use.
Discussion
To our knowledge, this is the first study on harmful alcohol use and injury in West Africa. We found a substantial frequency of harmful alcohol use (27%) among study-eligible injured patients presenting to the ED. High proportions of harmful alcohol use were found among males (35%), those aged 25–54 years (31%), drivers (32%), pedestrians (35%), assault victims (43%) and those acutely alcohol positive (55%). However, harmful alcohol use was present among all subject and injury characteristic groups regardless of acute alcohol positivity, as shown in tables 2 and 3. Thus, targeting only patients with evidence of acute alcohol use for intervention will miss a substantial number of harmful drinkers: a finding previously noted.16
Furthermore, our study subjects had a rate of harmful alcohol use over eight times greater than described by WHO for the general population. We have also noted from a previous study that the results of positive alcohol testing among injured patients in the ED were higher than what might be expected when considering the WHO data on alcohol use in Ghana.6 Studies from other African countries have shown that harmful alcohol use among injured patients in the ED was higher than that in the general population estimates.2 ,4 ,5 These discrepancies highlight the need for individual low-and-middle-income countries (LMICs) to assess their specific burden of harmful alcohol use among the injured.
Though Ghana has a maximum blood alcohol concentration of 0.08% for drivers, the country has yet to develop a national action plan, legal minimum age for alcohol sales or broad access to alcohol use disorder treatment services.2 ,17 ,18 Successful health policy strategies include regulations for alcohol and driving, enforcement of these regulations, public education and treatment of harmful alcohol use.19
This study supports screening and brief intervention for harmful alcohol use in our population as 40% of subjects screened positive for acute and/or harmful alcohol use. However, such cost-effective interventions have not be widely used in LMIC EDs.19 Research is needed to determine the feasibility of implementing screening and intervention programmes at the KATH ED.
This study has several limitations. First, the survey was designed as a convenience sample, which increases the likelihood of selection bias. However, as shown in table 1, the similar times of admission, age, sex and mechanisms of injury provide evidence for limited potential selection bias incurred. This study also had a low proportion of patients refusing to participate, which likely mitigated further selection bias. Second, harmful alcohol use is associated with greater severity of injury.20 Since patients with minor injuries (ie, SATS green category) and severe injuries (ie, met other exclusion criteria) were not included in this study, our frequency of harmful drinking could be lower or higher than if these subjects were included. Finally, the identification of harmful alcohol use was predicated on accurate patient responses. Some subjects may not have been forthcoming with their alcohol use behaviours, though to an uncertain degree. Again, this suggests that our results provide a conservative estimate of harmful alcohol use.
What is already known on the subject?
Alcohol is attributable for 15–20% of injury-related mortality.
Acute and chronic alcohol use are independent risk factors for injury recidivism.
Screening for acute alcohol use among the injured will miss a significant number of patients with harmful alcohol use.
What this study adds?
This is the first known study to examine the frequency of harmful alcohol use among injured adults in the West African region.
We found a frequency of harmful alcohol use that was eight times greater than the general population estimate.
The magnitude of harmful alcohol use discovered among the injured surveyed has implications for or health policy and development of emergency department-based interventions.
Acknowledgments
The authors thank Sonia Cobold, Joycelyn Sarfo-Frimpong, Juliana Kwarteng, Anthony Fosu, Hagar Baidoo, Dorcas Boakye, Abena Boatemaa Boateng, Lydia Offeibea Sakye, Sylvia Adomako, Irene Kutin, Akosua Kwaa Marfo, Belinda Dufie Gyesi, Gifty Ofori, Priscilla Boateng, Zipporah Yong Kukubor Ntekor, Emmanuel Amoateng, Simon Manu, Margaret Agyemang Badu, Irene Nyanor, Vida Odum, Alice Adomako and Susan Sally Asabea for their support and contributions to the study.
Footnotes
Twitter Follow Kwasi Ofori-Anti at @princeghana
Contributors AG, PKF, GO and RFM designed the study, managed data collection, analysed results and made substantial contributions to writing the manuscript and subsequent revisions. DD, KOA and KO-A managed data collection, participated in data analysis and made substantial contributions to writing the manuscript and subsequent revisions.
Funding This study was funded by grant R25TW009345 from the US National Institutes of Health and Fogarty International Center (AG).
Competing interests None declared.
Patient consent Obtained.
Ethics approval Kwame Nkrumah University of Science and Technology and University of Michigan Medical School institutional review boards approved this study.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement AG, PKF, GO and RFM share the data pertaining to acute alcohol use among the parent study population and have jointly developed a separate manuscript for publication.