Article Text

Download PDFPDF

Poisoning hospitalisations among reproductive-aged women in the USA, 1998–2006
  1. Shanna Cox1,
  2. Cassie Kuo1,
  3. Denise J Jamieson1,
  4. Athena P Kourtis1,
  5. Melissa L McPheeters2,
  6. Susan F Meikle3,
  7. Samuel F Posner1
  1. 1Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, Georgia, USA
  2. 2Department of Obstetrics and Gynecology, Vanderbilt University Medical Center, Nashville, Tennessee, USA
  3. 3Office of Research on Women's Health, National Institutes of Health, Bethesda, Maryland, USA
  1. Correspondence to Dr Shanna Cox, Division of Reproductive Health, Centers for Disease Control and Prevention, 4770 Buford Highway MS K-20, Atlanta, Georgia 30341, USA; shanna.cox{at}cdc.hhs.gov

Abstract

Objective To describe poisoning hospitalisations among reproductive-aged women from 1998 to 2006.

Methods 1998–2006 data from the Nationwide Inpatient Sample of the Healthcare Cost and Utilisation Project were used to identify hospitalisations for poisonings among US women aged 15–44 years. Differences in hospitalisation characteristics were compared by intent using χ2 statistics. Trends in poisoning hospitalisation rates were calculated overall and by subgroup.

Results There were approximately 636 000 poisoning hospitalisations in women aged 15–44 years during 1998–2006. Hospitalisations for intentionally self-inflicted poisonings had a higher proportion of women aged 15–24 years and privately insured women than did unintentional poisonings (p<0.001). Poisoning hospitalisations in rural areas and those that resulted in death were more likely to be of undetermined intent than those for which intent was specified (p<0.001). Co-diagnoses of substance abuse (34.5%) or mental disorders (66.5%) were high. The rate of poisoning hospitalisations overall and unintentional poisoning hospitalisations increased 6% and 22%, respectively, during this period (p<0.001). The most frequently diagnosed poisoning agent was acetaminophen. Poisonings attributable to acetaminophen, opioids, central nervous system stimulants and benzodiazepines increased, while poisonings attributable to antidepressants decreased (p<0.05).

Conclusions The increase in unintentional poisoning hospitalisations among women aged 15–44 years and the changing profile of poisoning agents should inform the healthcare community's poisoning prevention strategies. Poisoning prevention strategies should include a component to address substance abuse and mental health disorders among reproductive-age women.

  • Benzodiazepines
  • gender
  • opioids
  • poisoning hospitalisations
  • poison see ingestion
  • public health
  • reproductive-aged women
  • surveillance
  • trends

Statistics from Altmetric.com

Poisonings are injuries resulting from excess ingestion, inhalation, injection, or absorption of a substance. In the USA, poisoning accounts for approximately 1% of injury episodes, 4% of injury emergency department visits, 10% of injury hospitalisations and 18% of injury deaths.1 While surveillance systems for poisoning episodes,2 emergency department visits3 and deaths4 are in place, poisoning cases that lead to hospitalisation are less systematically investigated.

There have been documented increases in the non-medical use of prescription drugs.5 Women are more likely than men to engage in non-medical prescription drug use, primarily for narcotic analgesics and minor tranquillisers.6 Women are at higher risk than men for non-fatal self-poisoning, and poisoning is the most common method of intentionally self-inflicted injuries among women.7 8 Unintentional poisoning deaths have increased from 1999 to 2004, with the largest increases among women.4 Because rates of poisoning injury among women are highest among reproductive-aged women (15–44 years),9 and because of the high rate of unintentional pregnancy, possible exposure to certain agents during pregnancy may be especially problematical.

In this study, we reviewed poisoning hospitalisations of reproductive-age women (15–44 years) during 1998–2006. We describe the characteristics of these hospitalisations, including trends in poisoning intent and agent over the 8-year study period.

Methods

We obtained hospital discharge data from the Healthcare Cost and Utilisation Project (HCUP), Nationwide Inpatient Sample (NIS). The NIS is a research database produced annually through a partnership between the Agency for Healthcare Research and Quality (AHRQ) and public and private state-level data-collection organisations to provide national estimates of inpatient care in the USA.

Using a stratified, probability design, the NIS approximates a 20% sample of all US community hospitals as defined by the American Hospital Association (AHA). The AHA defines community hospitals as all non-federal short-term (average length of stay <30 days) general and speciality hospitals whose facilities are open to the public. The sampling frame consists of state-specific hospital discharge data provided to HCUP. In 2006, the sampling frame included approximately 90% of all hospital discharges in the USA and participating states included 90% of the US population. Five stratification parameters are used to select hospitals: rural/urban location; bed size; geographical region; teaching status and ownership. The NIS includes all inpatient data from sampled institutions and annually includes over 900 hospitals and approximately 7 million discharge records. To generate nationwide estimates of inpatient hospitalisation the data are weighted. The weighting process accounts for changes in the number of states included over the years as well as different hospitals sampled each year. The Centers for Disease Control and Prevention (CDC) classified the project as research not involving human subjects because the administrative dataset does not include any personal identifying information.

All hospitalisations of reproductive-aged women (15–44 years) that listed poisoning as the primary diagnosis on the discharge record were included in this study. International Classification of Diseases, 9th revision, clinical modification (ICD-9-CM) diagnosis codes 960–979.9 defined poisoning diagnoses based on the Barell injury diagnosis matrix.10 These diagnosis codes document acute poisonings including the use of drugs or chemicals for medical or recreational purposes in excessive amounts, such as an ‘overdose’. These diagnosis codes do not include substance abuse codes, toxic effects of substances chiefly non-medicinal as to source or codes describing adverse events related to drugs in therapeutic use. A poisoning that occurs inadvertently is defined as ‘unintentional’ and a poisoning that results from a conscious, wilful decision (such as for suicide) is defined as ‘intentional’. When the intent is unclear, poisonings are usually labelled ‘undetermined’. We used the CDC-recommended framework of E-code groupings to describe the intent of injury.11 When combining poisoning hospitalisation that did not have an E-code to describe intent with those categorised as ‘undetermined’, the results remained the same; therefore, we present results for these hospitalisations together.

We first described the characteristics of hospitalisations with a poisoning diagnosis, overall and by intent. Characteristics examined include age group, primary expected payer (public insurance, private insurance, self-pay or other), location (rural vs urban), pregnancy status, inhospital death, substance abuse comorbidity and mental disorder comorbidity. We identified pregnancy hospitalisations by the presence of any pregnancy-related diagnostic code in any field (ICD-9-CM 630–677, V22–24, and V27–28). We classified substance abuse and mental disorder comorbidity using AHRQ's comorbidity software, version 3.5.12 The comorbidity software is an algorithm designed to identify comorbidities using ICD-9-CM diagnosis codes and separate them from the primary reason for hospitalisation. This algorithm was developed and validated independently of this analysis. Substance abuse comorbidity included alcohol and drug abuse, and mental disorder comorbidity included psychoses and depression. We compared hospitalisation characteristics by intent using χ2 statistics. We compared the characteristics of unintentional poisoning hospitalisations with intentionally self-inflicted poisonings. We compared the characteristics of poisoning hospitalisations of undetermined intent with those with specified intent.

The second analysis examines the trends in the rates of poisoning hospitalisations overall, by intent and by age. We calculated discharges per 1000 women aged 15–44 years by dividing estimated total discharges by the appropriate age-specific population estimates from the census bureau.13 Linear trends over the study period were tested using a method of weighted least squares.14

We investigated trends in pregnancy hospitalisations with a poisoning diagnosis using hospitalisation ratios (the number of hospitalisations per 1000 deliveries). Hospitalisation ratios describe inpatient pregnancy-related morbidity in relation to the number of deliveries each year.15 We used the SUDAAN analytical procedure PROC Ratio to calculate hospitalisation ratios for poisoning diagnoses and to test for linear trends.

Next, we described the most commonly diagnosed poisoning agents overall, by intention and pregnancy status. Compounds were classified using ICD-9-CM codes according to their main pharmacological properties. We used the first listed ICD-9-CM code to classify poisoning agents (appendix A). We examined linear trends in the rates of most commonly diagnosed poisoning agents overall, by intent and by pregnancy status.

We present national estimates of poisoning hospitalisations using standard methods for analysing weighted survey data using SUDAAN version 9. All calculated p values were two-sided. Two independent researchers confirmed programming and data results.

Results

During the study period, of the approximately 69.1 million hospitalisations of reproductive age, an estimated 635 886 (0.9%) included a primary diagnosis of poisoning. Poisoning accounted for 35% of injury hospitalisations among women of reproductive age in the USA. Sixty-eight per cent (n=433 877) of poisoning hospitalisations among reproductive-age women were intentionally self-inflicted, 16.0% (n=101 680) were unintentional, 15.7% (n=99 706) were of undetermined intent and 0.1% (n=623) were classified as assault.

There was a higher proportion of women aged 15–24 years and privately insured women among intentionally self-inflicted poisoning hospitalisations than among unintentional poisoning hospitalisations (table 1). One per cent of poisoning hospitalisations among reproductive-age women included a notation of pregnancy, with a notation of pregnancy more common among intentionally self-inflicted poisonings than unintentional poisonings. Overall, an estimated 4225 deaths (0.7%) occurred; deaths were more common among unintentional poisonings than intentionally self-inflicted poisonings. Substance abuse and mental disorder comorbidity was high (34.5% and 66.5%, respectively). Substance abuse comorbidity was higher among unintentional poisoning than intentionally self-inflicted poisonings. Mental disorder comorbidity was higher among intentionally self-inflicted poisoning than unintentional poisonings.

Table 1

Characteristics of poisoning hospitalisations among women 15–44 years of age, 1998–2006

There was a higher proportion of women aged 35–44 years and hospitals in rural areas among poisonings for which the intent was undetermined than among poisoning hospitalisations for which the intent was specified. Privately insured women were less likely to be hospitalised for poisonings of undetermined intent compared with when the intent was specified; 1.6% of poisoning hospitalisations of undetermined intent resulted in death, compared with 0.5% of poisoning hospitalisations with intent specified. Substance abuse comorbidity was higher among poisoning hospitalisations of undetermined intent, compared with poisoning hospitalisations with intent specified. Mental disorder comorbidity was higher among poisoning hospitalisations with intent specified than poisoning hospitalisations of undetermined intent

There was a slight but significant increase in the overall rate of poisoning hospitalisations among women aged 15–44 years, increasing from 1.10 per 1000 women aged 15–44 years in 1998 to 1.18 per 1000 women aged 15–44 years in 2006 (p=0.0003; table 2). The rate of unintentional poisoning increased 22% (p<0.0001), whereas the rate of intentionally self-inflicted poisonings or poisoning of undetermined intent did not change significantly. The rate of poisoning hospitalisations increased significantly for women aged 25–44 years (p<0.0001), whereas there was no significant change in the rate of poisoning hospitalisations for women aged 15–24 years of age. The rate of poisoning hospitalisations with noted pregnancy decreased significantly from 0.24 per 1000 deliveries in 1998 to 0.16 per 1000 deliveries in 2006 (p<0.0001).

Table 2

Trends in rates of poisoning hospitalisations among women 15–44 years of age, 1998–2006

The five most common poisoning agents accounted for over 60% of poisoning diagnoses. The most common poisoning agents associated with poisoning hospitalisations of women aged 15–44 years were acetaminophen (17.7%), benzodiazepines (16.1%), antidepressants (14.1%), opiates (6.5%) and other specified psychotropics (5.8%; table 3). This ranking differed by intent and pregnancy status. For example, opiates were the most common poisoning agent among unintentional poisonings (14.6%), whereas benzodiazepines were the most common agent among poisonings of undetermined intent (16.3%).

Table 3

Most common poisoning agents, by intent and pregnancy status

Acetaminophen poisoning hospitalisations increased overall, among unintentional and intentional poisonings, and among pregnancy hospitalisations (p<0.0001). Benzodiazepine poisoning hospitalisations increased overall (p=0.03) and among unintentional poisonings (p=0.0004; figure 1 and figure 2). Antidepressant poisoning hospitalisations decreased irrespective of intent (p<0.01), but not among pregnancy hospitalisations. Opiate poisoning hospitalisations increased overall and among poisonings of unintentional or undetermined intent (p<0.0001). Other psychotropic and tranquilliser poisoning hospitalisations increased overall and among intentionally self-inflicted poisonings (p<0.0001). Although rare, there was a striking increase in poisoning hospitalisations attributable to central nervous system (CNS) stimulants among the unintentional poisoning group (from 0.002 per 1000 women aged 15–44 years to 0.03 per 1000 women aged 15–44 years, p<0.0001).

Figure 1

Trends in most common agents associated with poisoning hospitalisations of women 15–44 years, 1998–2006.

Figure 2

Trends in most common agents associated with unintentional poisoning hospitalisations of women 15–44 years, 1998–2006.

Discussion

Poisonings account for a substantial share of injury hospitalisations among reproductive-aged women when compared with other gender and age groups; supplementary analyses using HCUP NIS data revealed that during the same study period, poisoning accounted for 10% of injury hospitalisations among adult men and 13% of injury hospitalisations among adult women, compared with 35% among reproductive-aged women. Our findings that intentional self-inflicted poisonings are more common than unintentional poisonings are consistent with the previous reports regarding poisoning hospitalisations.16 17 We found an increase in the rate of poisoning hospitalisations among reproductive-aged women; specifically among unintentional poisonings. It is important to identify differences in the characteristics of poisoning hospitalisations by intent so that preventive strategies target those most at risk.

Previous research found that E coding inhospital discharge data are relatively complete and accurate;18 19 however, in our study, the intent of 16% of poisoning episodes was undetermined or missing. Poisonings are more likely to be coded as undetermined when compared with other injury types.20 We found significant differences in whether a poisoning hospitalisation included specified intent by hospitalisation characteristics. Systematic misclassification of intent may skew research findings and impede the identification of target populations for intervention or policy development. Recommendations exist to improve the quality of E-coded hospitalisation to develop a more complete understanding of injuries among various populations.21

Acetaminophen was the most commonly listed agent among poisoning hospitalisations for reproductive-aged women, and rates of acetaminophen poisoning increased over the study period. This finding is analogous to an analysis of the epidemiology of acute poisonings in women of reproductive age in the state of California.17 The common availability of acetaminophen may explain why it is the most common substance indicated in intentionally self-inflicted poisoning hospitalisations. Women may rely on non-prescription analgesics such as acetaminophen for gynaecological conditions such as primary dysmenorrhoea. During pregnancy, acetaminophen is the recommendation for pain relief compared with other analgesics available without prescription such as ibuprofen or aspirin. An increasing number of multiple ingredient formulations such as over-the counter products to treat symptoms of the common cold and prescription narcotic pain medicines contain acetaminophen. Consumers may attempt to treat different conditions or symptoms at the same time with more than one product containing acetaminophen without being aware of the risk of unintentional poisoning. Acetaminophen poisoning is the leading cause of acute liver failure, further emphasising the need for strategies to educate patients about the dangers of improper acetaminophen use.22 Policy-level interventions such as the US Food and Drug Administration's recent ruling requiring stronger label warnings and dosage limits on drugs containing acetaminophen may play a role in reducing acetaminophen poisonings.23

Similar to the trends that we identified, increases in emergency department visits and hospitalisations for poisonings attributable to opiate-related compounds, benzodiazepines and CNS stimulants have been demonstrated elsewhere.3 24–26 In our study, increases in opiate-related poisonings were largely driven by agents in the ‘other’ category, which includes narcotic analgesics such as morphine, oxycodone and hydrocodone, for which there have been documented increases in prescribing patterns and non-medical use. While there have been reports documenting the increase in the non-medical use of prescription opioids, increases in benzodiazepine poisonings have not received the same level of attention.25 Treatment for anxiety and mood disorders, conditions that are more common among women than men, often includes the use of benzodiazepines. Our finding that benzodiazepines were the second most common poisoning agent among reproductive-age women highlights the need to document the misuse of this drug class in this population. The majority of poisoning hospitalisations attributable to CNS stimulants were categorised as ‘other specified’, making interpretation of results difficult. Increases in the diagnosis and pharmaceutical treatment of attention deficit hyperactivity disorders, as well as the diversion of these drugs for non-medical use, may partly explain the increase in poisoning attributable to CNS stimulants.27 In our study we found a significant decline in the rate of antidepressant poisoning hospitalisations. A previous study reported similar declines in antidepressant poisoning hospitalisations,28 although there have been general increases in antidepressant poisoning episodes and emergency department visits.

Only a small percentage of the poisoning hospitalisations of reproductive-age women included a notation of pregnancy; however, the consequences of poisoning in this case not only include the health of the mother but also the impact on fetal development.29 A study of attempted suicide during pregnancy in the state of California revealed that attempted suicides were associated with an increase in adverse maternal and neonatal outcomes, and that 86% of suicide attempts during pregnancy involved poisoning.30 It is important for healthcare providers to be familiar with common poisoning agents in pregnant women and the clinical risks they present. There are limited data on the teratogenic risks associated with poisoning agents or antidotes.31 Data generated from surveillance systems such as the Vaccine and Medication in Pregnancy Surveillance System may further our understanding of the effect of these substances on fetal development and pregnancy outcome.32

Substance abuse and mental health disorder comorbidities were strikingly high among reproductive-age women hospitalised for poisoning. Previous research has documented that substance abuse and mental health disorders may be the best identifiers for acute poisonings among women of reproductive-age and during pregnancy.17 30 The extent of concomitant mental health disorders and intentionally self-inflicted poisonings documented in our study re-emphasise the need to address mental health treatment and poisoning prevention concurrently.

There are limitations in using hospital discharge data for poisoning surveillance. The use of ICD-9-CM coding makes it difficult to tease out specific compounds of interest and investigate particular trends. For example, there is no differentiation between serotonin-specific reuptake inhibitors or monoamine oxidase inhibitors among ICD-9-CM codes for antidepressant poisonings. The use of the primary diagnosis code to identify poisoning hospitalisations is a standard, yet conservative measure of hospitalisations for which a poisoning diagnosis may have played a role.33 The primary diagnosis field is reserved for the code corresponding to ‘the reason for which, after study by the attending physician or nurse, the patient was admitted’. The primary diagnosis may not be poisoning among some patients presenting with acute conditions that are resultant of poisoning such as respiratory or cardiac distress, leading to an underestimation of the impact of poisoning among this population. There is no standard for classifying poisoning agents. Classification schemes vary by surveillance system, making direct comparisons difficult. We assigned poisoning agents based on first-listed poisoning diagnoses, which precluded evaluating multiple agents. Thirty-four per cent of poisoning hospitalisations had more than one poisoning agent listed; however, there is no standard in prioritising which agents are the primary causes of a hospitalisation without knowing the clinical symptoms of the poisoning. We did not include toxic effects of substances chiefly non-medicinal in source in the definition of poisoning; therefore, this analysis excludes injuries that involve the ingestion of substances such as alcohol or household cleaners. Our data regarding poisoning hospitalisations among pregnant women are limited. In early pregnancy, a woman may not know she is pregnant; therefore, there will be no notation of pregnancy on the discharge record, leading to an underestimate of poisoning hospitalisations during pregnancy. Hospitalisation ratios only provide a relative measure of inpatient pregnancy-related morbidity to the number of deliveries and do not capture all pregnancies, particularly those that result in fetal death outside of the inpatient hospital setting. Psychiatric hospitals and alcohol/chemical dependency treatment facilities are not included in NIS; therefore, our results are limited to describing poisoning hospitalisations that occur at the community hospital setting.

Conclusions

We described the epidemiology of poisoning hospitalisations among reproductive-age women to provide data for use to inform poisoning prevention strategies. Providers who treat reproductive-age women should be aware of the impact of poisoning injuries in this population when prescribing medications as trends in self-poisoning often mirror trends in prescription patterns.24 34 However, restrictions on the availability of drugs must balance the therapeutic benefit received by many against inappropriate use by a few. Policy-level interventions such as package size limits on analgesics available over the counter have been found to reduce suicidal overdoses.35 Other preventive strategies for reducing the non-medical use of medications include modifying prescriptions to reduce their potential for misuse, screening for substance abuse and mental health disorders, and discussing proper medication use with patients in an effort to avoid unintentional overdose.36

What is already known on the subject

  • Poisoning is the most common form of intentionally self-inflicted injuries among women.

  • The rates of poisoning injury among women are highest among reproductive-aged women (15–44 years).

  • The rate of unintentional poisoning deaths (overdoses) have tripled among women since 1999.

What this study adds

  • The overall rate of poisoning hospitalisations and unintentional poisoning hospitalisations increased over the 8-year study period.

  • The profile of poisoning agents changed, with increases in poisoning hospitalisations related to opiates, CNS stimulants and benzodiazepines and decreases in poisoning hospitalisations related to antidepressants.

Acknowledgments

The authors gratefully acknowledge the contributions of the AHRQ-CDC HCUP Collaboration on Women's and Children's Health.

References

View Abstract

Footnotes

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.