Comparison of the statistical efficiency of case-crossover and case-control designs: application to severe cutaneous adverse reactions

Hot weather and death related to acute cocaine, opioid and amphetamine toxicity in British Columbia, Canada: a time-stratified case-crossover study

BACKGROUND

Previous research has shown that cocaine-associated deaths occur more frequently in hot weather, which has not been described for other illicit drugs or combinations of drugs. The study objective was to evaluate the relation between temperature and risk of death related to cocaine, opioids and amphetamines in British Columbia, Canada.

METHODS

We extracted data on all deaths with cocaine, opioid or amphetamine toxicity recorded as an underlying or contributing cause from BC vital statistics for 1998-2017. We used a time-stratified case-crossover design to estimate the effect of temperature on the risk of death associated with acute drug toxicity during the warmer months (May through September). Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated for each 10°C increase in the 2-day average maximum temperature at the residential location.

RESULTS

We included 4913 deaths in the analyses. A 10°C increase in the 2-day average maximum temperature was associated with an OR of 1.43 (95% CI 1.11-1.86) for deaths with only cocaine toxicity recorded (n = 561), an OR of 1.15 (95% CI 0.99-1.33) for deaths with opioids only (n = 1682) and an OR of 1.11 (95% CI 0.60-2.04) for deaths with amphetamines only (n = 133). There were also elevated effects when toxicity from multiple drugs was recorded. Sensitivity analyses showed differences in the ORs by sex, by climatic region, and when the location of death was used instead of the location of residence.

INTERPRETATION

Increasing temperatures were associated with higher odds of death due to drug toxicity, especially for cocaine alone and combined with other drugs. Targeted interventions are necessary to prevent death associated with toxic drug use during hot weather.

The case-crossover design via penalized regression

BACKGROUND

The case-crossover design is an attractive alternative to the classical case-control design which can be used to study the onset of acute events if the risk factors of interest vary in time. By comparing exposures within cases at different time periods, the case-crossover design does not rely on control subjects which can be difficult to acquire. However, using the standard method of maximum likelihood, resulting risk estimates can be heavily biased when the prevalence to risk factors is very low (or very high).

METHODS

To overcome the problem of low risk factor prevalences, penalized conditional logistic regression via the lasso (least absolute shrinkage and selection operator) has been proposed in the literature as well as related methods such as the Firth correction. We apply and compare several penalized regression approaches in the context of a case-crossover analysis of the European Study of Severe Cutaneous Adverse Reactions (EuroSCAR; 1997-2001).

RESULTS

Out of 30 drugs, standard methods only correctly classified 17 drugs (including some highly implausible risk estimates), while penalized methods correctly classified 22 drugs.

CONCLUSION

Penalized methods generally yield better risk classifications and much more plausible risk estimates for the EuroSCAR study than standard methods. As these novel techniques can be easily implemented using available R packages, we encourage routine use of penalized conditional logistic regression for case-crossover data.

Medications associated with clinical deterioration in hospitalized children

BACKGROUND

Medical emergency teams have been shown to reduce mortality in children's hospitals, but there are many potential barriers to their activation. Surveillance tools using electronic health record data help identify children at risk of deterioration. Existing early warning scores primarily include vital signs, but may benefit from the incorporation of medications.

OBJECTIVE

We aimed to identify the therapeutic classes of medications temporally associated with clinical deterioration that could be incorporated with vital signs into surveillance tools.

DESIGN

Case-crossover study.

SETTING

The Children's Hospital of Philadelphia.

PATIENTS

Children with clinical deterioration, defined as cardiopulmonary arrest, acute respiratory compromise, or urgent intensive care unit transfer while hospitalized on pediatric wards (n = 141).

EXPOSURES

Intravenous administrations of medications from therapeutic classes administered in ≥5% of control periods.

RESULTS

Nine therapeutic classes were significantly associated with clinical deterioration: glycopeptide antibiotics, anaerobic antibiotics, third-generation and fourth-generation cephalosporins, aminoglycoside antibiotics, systemic corticosteroids, benzodiazepines, loop diuretics, narcotic analgesics (full opioid agonists), and antidotes to hypersensitivity reactions.

CONCLUSIONS

We identified a set of therapeutic classes associated with increased risk of clinical deterioration. Future work should focus on evaluating whether including these therapeutic classes in multivariable models improves their accuracy in detecting early, evolving deterioration.

Case-only designs in pharmacoepidemiology: a systematic review

Background

Case-only designs have been used since late 1980’s. In these, as opposed to case-control or cohort studies for instance, only cases are required and are self-controlled, eliminating selection biases and confounding related to control subjects, and time-invariant characteristics. The objectives of this systematic review were to analyze how the two main case-only designs – case-crossover (CC) and self-controlled case series (SCCS) – have been applied and reported in pharmacoepidemiology literature, in terms of applicability assumptions and specificities of these designs.

Methodology/Principal Findings

We systematically selected all reports in this field involving case-only designs from MEDLINE and EMBASE up to September 15, 2010. Data were extracted using a standardized form. The analysis included 93 reports 50 (54%) of CC and 45 (48%) SCCS, 2 reports combined both designs. In 12 (24%) CC and 18 (40%) SCCS articles, all applicable validity assumptions of the designs were fulfilled, respectively. Fifty (54%) articles (15 CC (30%) and 35 (78%) SCCS) adequately addressed the specificities of the case-only analyses in the way they reported results.

Conclusions/Significance

Our systematic review underlines that implementation of CC and SCCS designs needs to be more rigorous with regard to validity assumptions, as well as improvement in results reporting.

Case-crossover design: an alternative strategy for detecting drug-induced liver injury

OBJECTIVE

Our observational study was conducted to assess if the case--crossover design could be applied to detect the risk of hepatoxic drugs on liver injury in the automated databases.

STUDY DESIGN AND SETTING

The study was conducted on approximately 22 million people enrolled in Taiwan's National Health Insurance database from 1997 to 2004. We applied case--crossover and case--control designs to assess the estimated risks of liver injury related to well-known hepatoxic drugs, including isoniazid, rifampicin, erythromycin, and diclofenac. Using case--crossover and case--control designs, we analyzed to explore the association between hospitalization and our target drugs through a conditional logistic regression model.

RESULTS

The adjusted odds ratios (ORs) of isoniazid, rifampicin, erythromycin, and diclofenac showed 24.4 (confidence interval [CI] =10.7-55.5), 30.8 (CI=14.1-67.1), 2.1 (CI=1.4-3.1), and 2.9 (CI=2.4-3.5) among 4,413 hospitalized liver injury patients during the 30-day exposure window by the case--crossover designs. Most adjusted ORs by case--crossover design were more conservative than ORs by case--control design.

CONCLUSIONS

In addition to a case--control design, the case--crossover design is a suitable method, for detecting the potential hepatotoxicity of drugs.

Within-person study designs had lower precision and greater susceptibility to bias because of trends in exposure than cohort and nested case-control designs

OBJECTIVE

To compare precision and apparent bias between cohort, nested case-control, self-controlled case series, case-crossover, and case-time-control study designs.

STUDY DESIGN AND SETTING

Study designs were implemented to evaluate the association between thiazolidinediones (TZDs) and heart failure, TZDs and fracture, and liver enzyme-inducing anticonvulsants and fracture.

RESULTS

Effect estimates were similar for the cohort and case-control study; for the association between TZDs and fracture in women, the hazard ratio was 1.36 (1.18, 1.56) and odds ratio (OR) was 1.44 (1.21, 1.70). For this clinical example, the self-controlled case series gave upward bias when follow-up was censored at the outcome (incidence rate ratio [IRR], 7.08; 4.96, 10.09) but was otherwise unbiased (IRR, 1.41; 1.14, 1.75). The retrospective case-crossover OR was 3.24 (2.18, 4.80), which was reduced by either bidirectional sampling (OR, 1.20; 0.98, 1.46) or with the case-time-control design (OR, 1.40; 1.09, 1.81). Findings on apparent bias were similar for the other two clinical examples. In each clinical example, within-person designs had considerably lower precision than the cohort or case-control study designs.

CONCLUSION

When long-term exposures are analyzed, within-person study designs may have lower precision and greater susceptibility to bias. Bias may be reduced by sampling follow-up both before and after the outcome or with the case-time-control study design.

Clinical features and risk factors for upper gastrointestinal bleeding in children: a case-crossover study

OBJECTIVE

This population-based survey was conducted to provide a formal description of upper gastrointestinal bleeding (UGIB) in children on a nationwide basis and assess the contribution of risk factors, principally nonsteroidal anti-inflammatory drugs (NSAID).

METHODS

A case-crossover study of UGIB patients aged between 2 months and 16 years was conducted in France. Medical data were collected by physicians, and personal risk factors and exposure to drugs during the month preceding the onset of the bleeding was ascertained by a standardised telephone interview with parents. The odds ratios for UGIB and NSAID was assessed by comparing exposure during the 7 days preceding the date of hospitalisation and the 21st to the 28th days before that date.

RESULTS

A total of 177 children with UGIB were included over 2 years. Eighty-three children had taken at least one NSAID before the index date, among which 58 were ibuprofen, 26 aspirin and nine others. The adjusted odds ratio (OR) of exposure was 8.2 [95% confidence interval (CI) 2.6-26.0] for NSAIDs altogether, and this was 10.0 (95% CI 2.0-51.0) for ibuprofen and 7.3 (95% CI 0.9-59.4) for aspirin. There was no increased risk associated with NSAIDS for oesophageal lesion [OR = 1.0 [(5% CI:0.2-7.2)].

CONCLUSION

The study confirms that UGIB is rare but that some cases may be avoided, as one third of the cases was attributable to exposure to NSAID at doses used for analgesic or antipyretic purposes, which may be attained with alternative therapy. The findings from this study call for more caution in prescribing NSAIDS to children.

Study of partner-related and situational risk factors for symptomatic male urethritis

During the last decade, the incidence of male urethritis stopped declining in France. Risk factors associated with unprotected intercourse have been extensively studied in men who have sex with men, but not in men in general. The purpose of the study was to determine major risk factors for urethritis among men and to describe the sociodemographic and medical characteristics of this population in 2005. We conducted a prospective case-crossover study of sexual behaviors among men with acute urethritis attending at general practitioners or sexually transmitted infection (STI) clinics in France. Each patient filled out a selfcompleted questionnaire focusing on sociodemographic characteristics, and on sexual behaviors for the month before urethritis onset and for the preceding 3 months. The doctor reported medical information on a separate questionnaire. Between January and September 2005, 121 cases of male urethritis, defined as recent-onset pain on micturition and/or purulent or mucoid discharge, were included. Median age was 33 years, 22.3% were MSM, 55.1% were single, and 72.0% had at least high school education. Conditional logistic regression analysis showed that intercourse with only casual partners or with both casual and steady partners (OR = 2.6, CI 95%: 0.8-8.7, and OR = 8.7, CI 95%: 2.7-28.0), as well as inconsistent condom use (OR = 5.8, CI 95%: 1.7-19.2) significantly increased the risk of male urethritis. STI prevention campaigns should continue to focus on consistent condom use and should not neglect men over 30 years of age.

Benzodiazepines and elderly drivers: a comparison of pharmacoepidemiological study designs

PURPOSE

Contradictory results were published from two studies in the late 1990s about the effects of long half-life benzodiazepine use on the risk of motor vehicle crashes (MVCs) in the elderly. The use of different study designs could explain the differences observed in these studies.

METHODS

The results of an unmatched case-control study were compared to those of a case-crossover study using the same prescription claims database to determine whether the current use of benzodiazepines increased the risk of MVCs.

RESULTS

There were 5579 cases and 12 911 controls identified between the years 1990 and 1993 in the province of Quebec, Canada. The case-control approach demonstrated an increased rate of injurious MVC associated with the current use of long-acting benzodiazepines [odds ratio (OR) 1.45; 95% confidence interval (CI): 1.12-1.88]. The case-crossover approach applied to all cases did not show any association [OR 0.99; 95%CI: 0.83-1.19]. However, among the cases restricted to subjects with four or less prescriptions filled in the previous year, corresponding more to transient exposures, the OR was elevated [OR 1.53; 95%CI: 1.08-2.16].

CONCLUSIONS

Differences in study design and analysis may explain some of the discrepancies in previous results. Both study designs provide evidence that long-acting benzodiazepines appear to be associated with an increased risk of MVC.