- Record: found
- Abstract: found
- Article: not found
Does Restricting Pack Size of Paracetamol (Acetaminophen) Reduce Suicides?
other
Publication date (Electronic):
3 April 2007
Journal:
PLoS Medicine
Publisher:
Public Library of Science
Read this article at
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
The authors discuss a new study that examined the change in deaths attributed to paracetamol poisoning in England and Wales in the six years before and after a legislated reduction in the maximum pack size.
Related collections
Most cited references22
- Record: found
- Abstract: found
- Article: not found
UK legislation on analgesic packs: before and after study of long term effect on poisonings.
Keith Hawton, Sue Simkin, Jonathan Deeks … (2004)
To evaluate the long term effect of legislation limiting the size of packs of analgesics sold over the counter. Before and after study. Suicides in England and Wales, data from six liver units in England and Scotland and five general hospitals in England, and UK data on sales of analgesics, between September 1993 and September 2002. Office for National Statistics; six liver units in England and Scotland; monitoring systems in general hospitals in Oxford, Manchester, and Derby; and Intercontinental Medical Statistics Health UK. Deaths by suicidal overdose with paracetamol, salicylates, or ibuprofen; numbers of patients admitted to liver units, listed for liver transplant, and undergoing transplantations for paracetamol induced hepatotoxicity; non-fatal self poisonings with analgesics and numbers of tablets taken; and sales figures for analgesics. Suicidal deaths from paracetamol and salicylates were reduced by 22% (95% confidence interval 11% to 32%) in the year after the change in legislation on 16 September 1998, and this reduction persisted in the next two years. Liver unit admissions and liver transplants for paracetamol induced hepatotoxicity were reduced by around 30% in the four years after the legislation. Numbers of paracetamol and salicylate tablets in non-fatal overdoses were reduced in the three years after the legislation. Large overdoses were reduced by 20% (9% to 29%) for paracetamol and by 39% (14% to 57%) for salicylates in the second and third years after the legislation. Ibuprofen overdoses increased after the legislation, but with little or no effect on deaths. Legislation restricting pack sizes of analgesics in the United Kingdom has been beneficial. A further reduction in pack sizes could prevent more deaths.
- Record: found
- Abstract: found
- Article: not found
The coal gas story. United Kingdom suicide rates, 1960-71.
N B Kreitman (1976)
A detailed analysis of suicide rates between 1960 and 1971 for England and Wales and for Scotland confirms that all age-sex subgroups have shown a marked decline in suicide due to domestic gas, corresponding in time to the fall in the CO content. After considering data on the effects of the International Classification of Diseases (ICD) Eighth Revision, accident mortality, some personal characteristics of coal gas suicides, and the use of coal gas in parasuicide it was concluded that a simple casual explantation was likely. Suicide due to non-gas methods has in general increased, markedly so in some groups. It was suggested that neither improved psychiatric services nor voluntary agencies could have produced such changes. The 'compensatory' trend of gas and non-gas suicide rates was indicated for certain age-sex subgroups. The continuing need for suicide research was pointed out, and questions were raised concerning the psychological meaning of the epidemiological data.
- Record: found
- Abstract: found
- Article: not found
Interrupted Time-Series Analysis of Regulations to Reduce Paracetamol (Acetaminophen) Poisoning
Oliver W Morgan, Clare Griffiths, Azeem Majeed (2007)
Introduction Paracetamol (acetaminophen) is an effective, cheap, and widely available analgesic. However, poisoning due to paracetamol is a common problem worldwide [1] and is currently the most common cause of acute liver failure in UK [2] and the United States [3]. Each year in England and Wales, there are approximately 150 deaths and over 30,000 hospital admissions due to paracetamol overdose [4]. Successful interventions to reduce harm from paracetamol poisoning are needed. In September 1998, the Medicines Control Agency of the United Kingdom introduced legislation to limit the availability of paracetamol and thereby reduce mortality and morbidity due to paracetamol poisoning [5,6]. Availability of aspirin was similarly limited to avoid substitution. Pack sizes were limited to 16 tablets at general sales outlets and 32 tablets at pharmacies. The maximum number of tablets that can be sold in any single purchase without a prescription was limited to 100 tablets. Concurrently, the government agreed voluntary restrictions with the pharmaceutical industry and retailers to sell no more than 32 tablets in any single purchase. Several studies have shown that paracetamol poisoning deaths in England and Wales have decreased, although they are unable to convincingly attribute this to the 1998 regulations [7–9]. Here we use interrupted time-series analysis to assess whether the recent fall in the number of paracetamol deaths is different to trends in fatal poisoning involving aspirin, paracetamol compounds, antidepressants, or nondrug poisoning suicide. Methods Mortality Data The Office for National Statistics (ONS) maintains a dedicated database of drug-poisoning deaths in England and Wales since 1993. Drug-poisoning deaths are extracted from the national deaths database using specific International Classification of Diseases codes for the underlying cause of death [10]. In addition to data supplied in the cause of death section of the coroner's death certificate, the database also contains textual information supplied voluntarily and in confidence by coroners to ONS about circumstances of the death, which may include more detailed information from an inquest investigation about the drugs involved [10]. Two trained ONS coders independently code the textual information about all drugs mentioned by the coroner. Any discrepancies are resolved and discussed with a senior medical epidemiologist. The manual coding allows the database to be queried and all deaths identified where specific drugs were detected by the coroner. About 90% of drug-poisoning deaths have specific information about the drug(s) taken [4]. We identified deaths in which paracetamol was mentioned on the death certificate, with or without alcohol or other nonparacetamol drugs. Where death certificates mentioned paracetamol along with codeine, dihydrocodeine, dextropropoxyphene, or as part of a paracetamol-containing drug (e.g., Paramax and Propain), these were defined as being paracetamol compound-related deaths. To ascertain whether changes in mortality trends for paracetamol poisoning were coincidental rather than due to the intervention, we compared paracetamol deaths with paracetamol compounds, aspirin, antidepressant drugs, and suicide deaths involving methods other than drug poisoning [11]. We chose antidepressants because along with paracetamol, they are the most common substances mentioned in drug-related poisoning suicides [11]. Therefore, if changes in the time series for paracetamol were due to overall changes in the propensity to commit suicide by drug overdose rather than the regulations, we would expect to see similar changes for antidepressants. Furthermore, analysing nondrug poisoning suicides should give an indication of any effect of suicide trends overall. We did not compare deaths involving paracetamol with those involving opiates, even though they make up about half of all drug-poisoning deaths [4]. This was because the epidemiology of opiate-related deaths differs markedly from other drug-related deaths, having increased more than 6-fold between 1993 and 2000 [12]. Analysis We calculated annual directly age-standardised mortality rates using the European Standard population. For the interrupted time-series analysis, we used segmented linear regression, which divides a time series into pre- and postintervention segments [13]. As the regulations were introduced at the end of 1998, we chose 1999 as the intersection between segments (i.e., the intervention). A linear regression model has two parameters: the level and slope. Therefore the difference between the two segments can be quantified by testing the change in these two parameters (Equation 1). A change in level between the pre- and postintervention segments indicates a step-change, and a change in slope indicates a change in trend. β 0 estimates the baseline level of the outcome at the beginning of the time series. β 1 estimates the preintervention trend where time is a continuous variable indicating the time in years at time t from the start of the study period. β 2 estimates the change in level postintervention where intervention t = 0 before the intervention, and intervention t = 1 after the intervention. β 3 estimates the change in postintervention trend where time after intervention is a continuous variable indicating the number of years after the start of the intervention at time t and is coded as zero before the intervention. e t includes random error and autocorrelation. Time-series data are often autocorrelated (events closer together in a time series tend to be more similar than events further apart in time) [14]. Hence, the model residuals are not independent (a key assumption when using ordinary least-squares regression) [15]. Nonindependence of the residuals can lead to biased standard deviations, which can over- or underestimate tests of statistical significance [16]. We therefore corrected for autocorrelation effects by including a term in the regression model for the lagged residuals (i.e., the residual of the regression model moved to the previous time points in the time series). We considered whether changes in level and slope for paracetamol poisoning were different from the comparison series (compound paracetamol, aspirin, antidepressants, and nondrug poisoning suicide). We made this comparison by calculating a ratio for each year by dividing the age-standardised mortality rates for each comparison group by paracetamol and used segmented regression to assess changes in the level and slope of the ratios, again using 1999 as the intersection between segments. The null hypothesis was that changes in the age-standardised mortality rates for paracetamol and the comparison groups were similar. We rejected the null hypothesis if there was statistical evidence that the model parameters (level or slope) of the ratios were statistically different from 0 at the p < 0.05 level between the pre- and postintervention segments. The 95% confidence intervals of the ratio of the model parameters summarize the degree of difference between paracetamol and the control series that can be excluded as not being statistically different. All analysis was done using Stata 8.2 (Stata, http://www.stata.com). Results Between 1993 and 2004, 2,196 poisoning deaths occurred involving paracetamol in England and Wales. Number of deaths were similar among males and females, and hence we present results for males and females combined. Changes in age-specific mortality rates were similar across all age groups during the study period (unpublished data). For paracetamol poisoning the age-standardised mortality rate was 8.1 per million in 1993, peaking at 8.8 per million in 1997, followed by a decrease to 5.3 per million in 2004 (Figure 1; Table 1). The results of the segmented regression analysis for paracetamol poisoning are summarised in Table 2. There was evidence of first-order autocorrelation (p = 0.033), and so we included a term for it in the final model. There was a downward step-change in the annual age-standardised mortality rate, of −2.69 per million (p = 0.003) between the pre- and postintervention periods, while there was no evidence of a change in slope (p = 0.128) in the postintervention period. About twice as many deaths involved paracetamol compounds (n = 4,378), less than one-fourth of deaths involving aspirin (n = 457), two and a half times as many involving antidepressants (n = 5,602), and almost 20 times as many deaths due to nondrug poisoning suicide (n = 43,824) (Table 3). The median age at death for aspirin, 62 years, was higher than other groups (Table 3). A similar proportion of deaths involved males and females for all three drug-poisoning groups, but not nondrug suicide deaths, for which about 80% were male. Compared to paracetamol compounds and antidepressants, paracetamol deaths were less likely to mention other drugs, whereas aspirin deaths were more likely to mention other drugs. The majority of deaths were suicides for all groups except aspirin, among which only about half were considered suicide. Age-standardised mortality rates for paracetamol compounds, aspirin, antidepressants, and nonpoisoning suicides all declined over the study period (Figure 1; Table 1). In the preintervention period, rates for paracetamol compounds and antidepressants increased slightly, whereas paracetamol-poisoning deaths and nondrug poisoning suicides were relatively stable (Figure 1). In the postintervention period, rates for all series declined, with paracetamol compounds declining the most. A summary of the segmented regression models applied to the ratios of the comparison with paracetamol series is shown in Table 4. The time series for paracetamol compounds differed from paracetamol, having a greater step-change (coefficient = 0.81; p = 0.012) and greater postintervention decline (coefficient = −0.19; p = 0.031). In contrast, no statistical evidence suggested that the change in trends for aspirin or antidepressants was different to the change in paracetamol trends. The ratio of aspirin versus paracetamol showed a similar postintervention step-change (coefficient = 0.01; p < 0.779) and slope (coefficient = −0.003; p = 0.834). Likewise, antidepressants had a similar step-change (p = 0.108) and change in slope (p = 0.250). Nondrug poisoning suicide, on the other hand, showed a statistically marked upward step-change (coefficient = 8.44; p = 0.025) but a similar postintervention trend (p = 0.865). The upward step-change in the ratio of paracetamol versus nondrug poisoning suicide was due to a fall in paracetamol deaths. Examination of the number of deaths in Table 1 and the age-standardised rates in Figure 1 shows that only a small increase occurred in nondrug poisoning suicide between 1998 and 2000, which was within the bounds of measurement error and random annual variation. Discussion Between 1993 and 2004, there was a decline in age-standardised mortality rates from paracetamol poisoning. The decline did not occur gradually over the study period, but there was a downward step-change coinciding with the introduction of regulations to limit availability of paracetamol. This step-change was similar for poisoning deaths involving antidepressants and aspirin, but not paracetamol compounds or nondrug poisoning suicide. The 95% confidence intervals of the coefficients for aspirin and antidepressants show that statistical power was sufficient to detect a relatively small divergence in the ratios with paracetamol, thereby increasing our confidence in the conclusion that trends were similar to paracetamol poisoning. Strengths and Limitations Ideally, the impact of health care interventions should be assessed using an experimental study design, such as a randomised controlled trial [17]. However, where interventions are implemented across an entire population, researchers do not have control over intervention allocation, so nonexperimental study designs must be used instead [17,18]. Interrupted time-series methods have several advantages over other quasi-experimental studies, because they are less likely to be influenced by a number of biases [18]. For example, underlying increasing or decreasing secular trends may contribute to observed intervention effects. This variation may also occur where cyclical effects introduce localised upward or downward trends in the time series. Interventions that may be short-lived may erroneously report maximal effects if short time series are analysed. Finally, autocorrelated data means that adjacent data points can be more similar (positive autocorrelation) or dissimilar (negative autocorrelation), leading to under- or overestimates of effect, respectively. However, as with all nonexperimental designs, causal inference from interrupted time-series designs is limited because it is impossible to rule out alternative explanations for observed changes in time series [18]. One such explanation is a change in the way data are recorded (data artefact), and in our study, coding of deaths changed from International Classification of Diseases-9 to International Classification of Diseases-10 in 2001. However, we think this is an unlikely explanation for our findings: a bridge coding study showed no impact on the number of drug-poisoning deaths [19], and the ONS database does not use International Classification of Diseases codes to identify which drugs were involved [10]. Furthermore, as the coroner's primary function is to investigate the possibility of criminal involvement, information from death certificates may not be ideal for surveillance of drug-related deaths [10]. Information on the death certificate may be incomplete, as toxicological examination may not be conducted or may only test for a limited range of drugs; about 10% of drug-poisoning deaths have no specific information on the drug(s) taken [4]. Where multiple substances are mentioned on the death certificate, it is not possible to establish which drug was the most likely to have caused the death. This may lead to misclassification of some deaths as due to paracetamol. However, there is no evidence to suggest that the recording practice of coroners has changed over time, and no guidelines or recommendations about the investigation of poisoning deaths have been published that may have changed their recording with the ONS system. An alternative explanation is that an abrupt change in the population at risk coincided with the introduction of the intervention. This change could occur if, for example, the 1998 regulations inadvertently encouraged retailers to be more cautious about selling paracetamol to young girls, the group most frequently admitted to hospital for paracetamol poisoning [8]. Again, we think this is unlikely because age-specific mortality rates were similar across all age groups during the time series. A third explanation is that external influences may have caused longer-term cyclical trends that we have not taken into account, such as changing unemployment or other macrolevel effects [20]. Such influences could explain the similarity of the three drug-poisoning time series in our study. A major limitation of previous studies assessing the impact of the 1998 regulations on paracetamol-poisoning mortality is that no comparison group was used. However, identifying a suitable comparator is not straightforward, and hence we chose multiple comparison groups. Aspirin share many of the characteristics of paracetamol: it is a readily available analgesic and is freely available over the counter without prescription. However, aspirin is not an ideal comparator as it is less toxic than paracetamol, many more deaths involving aspirin also involve other drugs, the median age at death is greater, and fewer aspirin-related deaths are considered suicides. Furthermore, aspirin was also subject to pack-size restrictions, which may explain why trends in aspirin-related deaths are similar to paracetamol. Antidepressant-related deaths are far more similar to paracetamol in terms of median age at death and involvement in suicide, however, they too are frequently taken with other drugs and unlike paracetamol, are only available in prescription. Given these differences, it is even more striking that mortality trends involving these two drugs are similar. Paracetamol compounds include a number of different paracetamol-containing drugs. Like antidepressants, many are only available on prescription, and some are far more toxic in overdose than paracetamol. One such compound, co-proxamol, was recently banned in the United Kingdom because of its high toxicity and frequent involvement in poisoning deaths [21]. Conclusions We have shown that paracetamol-poisoning deaths decreased around the same time that the regulations were introduced. This observation is consistent with other studies [8,9,22]. However, we have also shown that poisoning deaths involving aspirin, antidepressants, and to a lesser degree, paracetamol compounds followed similar trends. This finding raises the question whether the concurrent introduction of the 1998 regulations and reduction in paracetamol-poisoning deaths following was coincidental rather than causal. We found little evidence to support the hypothesis that the 1998 regulations limiting pack size resulted in a greater reduction in poisoning deaths than occurred for other drugs or nondrug poisoning suicides.
Author and article information
Journal
Journal ID (nlm-ta): PLoS Med
Journal ID (publisher-id): pmed
Title:
PLoS Medicine
Publisher:
Public Library of Science
(San Francisco, USA
)
ISSN
(Print):
1549-1277
ISSN
(Electronic):
1549-1676
Publication date
(Print):
April
2007
Publication date
(Electronic):
3
April
2007
Volume: 4
Issue: 4
Electronic Location Identifier: e152
Author notes
* To whom correspondence should be addressed. E-mail:
Nick.Buckley@
123456act.gov.au
Article
Publisher ID: 07-PLME-P-0119DOI: 10.1371/journal.pmed.0040152
PMC ID: 1831716
PubMed ID: 17407389
SO-VID: f289a0fb-4b99-4467-8368-3a3fdd95550d
Copyright ©
Copyright: © 2007 Buckley and Gunnell. This is an open-access article distributed under the
terms of the Creative Commons Attribution License, which permits unrestricted use,
distribution, and reproduction in any medium, provided the original author and source
are credited.
History
Related
Interrupted Time-Series Analysis of Regulations to Reduce Paracetamol (Acetaminophen)
Poisoning
Page count
Pages: 2
Categories
Subject:
Perspectives
Subject:
Mental Health
Subject:
Mental Health
Subject:
Non-Clinical Medicine
Subject:
Non-Clinical Medicine
Subject:
Non-Clinical Medicine
Subject:
Pharmacology
Subject:
Pharmacology
Subject:
Public Health and Epidemiology
Subject:
Drugs and Adverse Drug Reactions
Subject:
Suicide
Subject:
Health Policy
Subject:
Pharmacology and Toxicology
Subject:
Regulation
Custom metadata
citation Buckley NA, Gunnell D (2007) Does restricting pack size of paracetamol (acetaminophen)
reduce suicides? PLoS Med 4(4): e152. doi:
10.1371/journal.pmed.0040152
ScienceOpen disciplines: Medicine
Data availability:
ScienceOpen disciplines: Medicine