Self-controlled Case Series Method Research Articles

Comparative Effectiveness of Generic Imatinib and Brand-Name Imatinib for the Treatment of Chronic Myeloid Leukemia

Background: Generic versions of imatinib (GEN) have been approved for use in Canada for chronic myeloid leukemia, chronic phase (CML-CP) on the basis of bioequivalence studies and were reimbursed in Québec starting from October 2013. Molecular responses with GEN have not yet been examined in detail. This study assesses the risk of diminished molecular response in switchers from brand-name imatinib (BN) to GEN and compares the effectiveness of initiating first-line GEN and first-line BN.Methods: Prospective individual patient data were available from nine hospitals participating in the Québec CML registry. To allow equipoise between GEN and BN, we focused on stable BN users at the time of GEN market entry, of which some were subsequently switched to GEN. We further selected only those who had a 1-log rise in international reporting scale (IS) BCR-ABL1 transcript level and conducted a self-controlled case series study (SCSS).1 Using SCSS, each patient contributed follow-up for BN use, and for GEN use (if a switch occurred). The analysis used pooled BN and GEN person-time and compared the odds ratio (OR) of 1-log rise during GEN and BN treatment using conditional Poisson regression. A second analysis used a cohort of initiators of BN and GEN from 2013 and onwards. Kaplan Meier (KM) analyses were used to estimate the cumulative incidence of early molecular response (EMR) corresponding to < 10% IS. Cox proportional hazards models were used to estimate age-adjusted hazard ratio (HR) with 95% confidence intervals (CI) for EMR with GEN use, when compared with BN use.Results: We identified 184 patients treated with BN, 38 who were switched from BN to GEN, and 5 who used GEN only. For the SCSS analysis we included 23 patients, of which 17 had 1-log rise during BN use and 6 during GEN use. All patients had achieved major molecular response (MMR) prior to cohort entry (Table). Mean follow-up was 1.45±0.43 years. Overall, the use of GEN was associated with an increased incidence of 1- log rise (OR: 3.34, 95% CI: 0.33-33.68), although not reaching statistical significance. Ten of 23 rises in BCR-ABL1 levels were subsequently confirmed (7 in BN and 3 in GEN). Eleven patients lost MMR (IS>0.1%), 9 during BN use and 2 during GEN use. The cohort of first-line imatinib included 11 patients, 4 GEN and 7 BN. GEN users were slightly older (61 vs 53, GEN vs BN), and Sokal scores were comparable (low, 2 vs 4; intermediate, 2 vs 3). There was no clear separation of the EMR curves (Figure). However, the adjusted HR of EMR with GEN was 0.38 (95% CI: 0.07-2.15), compared with BN.Conclusions: While these analyses are preliminary, our results call for an initiative on a larger scale to examine the clinical effectiveness of generic imatinib for CML-CP.1. Whitaker HJ, Farrington CP, Spiessens B, Musonda P. Tutorial in biostatistics: the self-controlled case series method. Stat Med. 2006;25(10):1768-1797.Table 1Baseline characteristics of the self-controlled case series cohort (n=23)CharacteristicValueMean age, years (SD)62.87 (15.4)Female sex (n, %)9 (39.1)Mean number of concomitant medications (SD)1.07 (2.2)Mean years of brand-name imatinib use (SD)6.85 (2.8)Number major molecular response (%)23 (100.0) [Display omitted] DisclosuresChamakhi:Pfizer: Consultancy; BMS: Consultancy; Novartis: Consultancy. Delage:Pfizer: Consultancy; BMS: Consultancy; Novartis: Consultancy. Laneuville:Pfizer: Consultancy; BMS: Consultancy; Novartis: Consultancy. Mollica:Pfizer: Consultancy; Novartis: Consultancy; BMS: Consultancy. Olney:Cellgene: Honoraria; Pfizer: Consultancy; BMS: Consultancy; Novartis: Consultancy. Busque:Novartis: Consultancy, Honoraria, Research Funding, Speakers Bureau; BMS: Consultancy, Honoraria; Pfizer: Consultancy, Honoraria. Assouline:Pfizer: Consultancy; BMS: Consultancy; Novartis: Consultancy.

Febrile seizures after 2009 influenza A (H1N1) vaccination and infection: a nationwide registry-based study.

BackgroundDuring the 2009 influenza A (H1N1) pandemic, a monovalent pandemic strain vaccine containing the oil-in-water adjuvant AS03 (Pandemrix®) was offered to the Norwegian population. The coverage among children reached 54 %. Our aim was to estimate the risk of febrile seizure in children after exposure to pandemic influenza vaccination or infection.MethodsThe study population comprised 226,889 children born 2006–2009 resident in Norway per October 1st, 2009. Febrile seizure episodes were defined by emergency hospital admissions / emergency outpatient hospital care with International Classification of Diseases, Version 10, codes R56.0 or R56.8. The self-controlled case series method was applied to estimate incidence rate ratios (IRRs) in pre-defined risk periods compared to the background period. The total observation window was ± 180 days from exposure day. Among 113,068 vaccinated children, 656 (0.6 %) had at least one febrile seizure episode.ResultsThe IRR of febrile seizures 1–3 days after vaccination was 2.00 (95 % confidence interval [CI]: 1.15–3.51). In the period 4–7 days after vaccination, no increased risk was observed. Among the 8172 children diagnosed with pandemic influenza, 84 (1.0 %) had at least one febrile seizure episode. The IRR of febrile seizures on the same day as a diagnosis of influenza was 116.70 (95 % CI: 62.81–216.90). In the period 1–3 days after a diagnosis of influenza, a tenfold increased risk was observed (IRR 10.12, 95 % CI: 3.82 – 26.82).ConclusionsIn this large population-based study with precise timing of exposures and outcomes, we found a twofold increased risk of febrile seizures 1–3 days after pandemic influenza vaccination. However, we found that pandemic influenza infection was associated with a much stronger increase in risk of febrile seizures.

Active Surveillance for Adverse Events After a Mass Vaccination Campaign With a Group A Meningococcal Conjugate Vaccine (PsA-TT) in Mali.

Background. The monovalent meningococcal A conjugate vaccine (PsA-TT, MenAfriVac) was developed for use in the “meningitis belt” of sub-Saharan Africa. Mali was 1 of 3 countries selected for early introduction. As this is a new vaccine, postlicensure surveillance is particularly important to identify and characterize possible safety issues.Methods. The national vaccination campaign was phased from September 2010 to November 2011. We conducted postlicensure safety surveillance for PsA-TT in 40 government clinics from southern Mali serving approximately 400 000 people 1–29 years of age. We conducted analyses with individual-level data and population-level data, and we calculated rates of adverse events using the conditional exact test, a modified vaccine cohort risk interval method, and a modified self-controlled case series method for each outcome of interest, including 18 prespecified adverse events and 18 syndromic categories.Results. An increased rate of clinic visits for fever within 3 days after vaccination was found using multiple methods for all age groups. Although other signals were found with some methods, complete assessment of all other prespecified outcomes and syndromic categories did not reveal that PsA-TT was consistently associated with any other health problem.Conclusions. No new safety concerns were identified in this study. These results are consistent with prelicensure data and other studies indicating that PsA-TT is safe. The approach presented could serve as a model for future active postlicensure vaccine safety monitoring associated with large-scale immunization campaigns in low-income countries.

Flexible modelling of vaccine effect in self-controlled case series models.

The self-controlled case series (SCCS) method, commonly used to investigate the safety of vaccines, requires information on cases only and automatically controls all age-independent multiplicative confounders, while allowing for an age-dependent baseline incidence. Currently, the SCCS method represents the time-varying exposures using step functions with pre-determined cut points. A less prescriptive approach may be beneficial when the shape of the relative risk function associated with exposure is not known a priori, especially when exposure effects can be long-lasting. We therefore propose to model exposure effects using flexible smooth functions. Specifically, we used a linear combination of cubic M-splines which, in addition to giving plausible shapes, avoids the integral in the log-likelihood function of the SCCS model. The methods, though developed specifically for vaccines, are applicable more widely. Simulations showed that the new approach generally performs better than the step function method. We applied the new method to two data sets, on febrile convulsion and exposure to MMR vaccine, and on fractures and thiazolidinedione use.

Risk and Risk Periods for Stroke and Acute Myocardial Infarction in Patients with Central Retinal Artery Occlusion

To investigate the risk of stroke and acute myocardial infarction (AMI) in patients with incident central retinal artery occlusion (CRAO). A self-controlled case series (SCCS) study. Patients with incident CRAO from the entire Korean population of 48 million individuals. We used the Korean national claim database (2007-2011) for analyses. After identifying patients with incident CRAO, the relative incidence rate ratios (IRRs) for stroke and AMI in risk periods were measured in these patients using a SCCS method. The IRRs of stroke and AMI by risk periods. Of 1655 patients with incident CRAO in 2009-2010, 165 had stroke/AMI (ischemic stroke in 139, hemorrhagic stroke in 13, and AMI in 15) in the observation period spanning 365 days before and after the occurrence of CRAO. The IRR of stroke/AMI 1 to 30 days after CRAO occurrence significantly increased (14.0; 95% confidence interval [CI], 8.90-22.00); the IRR peaked during the 1 to 7 days after CRAO occurrence (44.51; 95% CI, 27.07-73.20), and the increased risk was present for the first 30 days. The IRR of stroke/AMI also significantly increased 1 to 30 days (6.82; 95% CI, 4.01-11.60) and 31 to 90 days (2.86; 95% CI, 1.66-4.93) before CRAO occurrence. Subanalysis for only ischemic stroke showed similar, magnified IRRs in the risk periods compared with all events. The IRRs were not significantly different between sexes or age groups (<65 vs. ≥65 years). Patients with incident CRAO are at increased risk of ischemic stroke just after CRAO occurrence, and the risk is particularly increased during the first week immediately after the CRAO occurrence. The results suggest that patients with incident CRAO require immediate neurologic evaluation and preventive treatment to reduce mortality and morbidity.

A vaccine study design selection framework for the postlicensure rapid immunization safety monitoring program.

The Postlicensure Rapid Immunization Safety Monitoring Program, the vaccination safety monitoring component of the US Food and Drug Administration's Mini-Sentinel project, is currently the largest cohort in the US general population for vaccine safety surveillance. We developed a study design selection framework to provide a roadmap and description of methods that may be utilized to evaluate potential associations between vaccines and health outcomes of interest in the Postlicensure Rapid Immunization Safety Monitoring Program and other systems using administrative data. The strengths and weaknesses of designs for vaccine safety monitoring, including the cohort design, the case-centered design, the risk interval design, the case-control design, the self-controlled risk interval design, the self-controlled case series method, and the case-crossover design, are described and summarized in tabular form. A structured decision table is provided to aid in planning of future vaccine safety monitoring activities, and the data components comprising the structured decision table are delineated. The study design selection framework provides a starting point for planning vaccine safety evaluations using claims-based data sources.

Automatic identification of Web-based risk markers for health events.

BackgroundThe escalating cost of global health care is driving the development of new technologies to identify early indicators of an individual’s risk of disease. Traditionally, epidemiologists have identified such risk factors using medical databases and lengthy clinical studies but these are often limited in size and cost and can fail to take full account of diseases where there are social stigmas or to identify transient acute risk factors.ObjectiveHere we report that Web search engine queries coupled with information on Wikipedia access patterns can be used to infer health events associated with an individual user and automatically generate Web-based risk markers for some of the common medical conditions worldwide, from cardiovascular disease to sexually transmitted infections and mental health conditions, as well as pregnancy.MethodsUsing anonymized datasets, we present methods to first distinguish individuals likely to have experienced specific health events, and classify them into distinct categories. We then use the self-controlled case series method to find the incidence of health events in risk periods directly following a user’s search for a query category, and compare to the incidence during other periods for the same individuals.ResultsSearches for pet stores were risk markers for allergy. We also identified some possible new risk markers; for example: searching for fast food and theme restaurants was associated with a transient increase in risk of myocardial infarction, suggesting this exposure goes beyond a long-term risk factor but may also act as an acute trigger of myocardial infarction. Dating and adult content websites were risk markers for sexually transmitted infections, such as human immunodeficiency virus (HIV).ConclusionsWeb-based methods provide a powerful, low-cost approach to automatically identify risk factors, and support more timely and personalized public health efforts to bring human and economic benefits.

The case-case-time-control study design

Although the 'self-matched case-only studies' (such as the case-cross-over or self-controlled case-series method) can control the time-invariant confounders (measured or unmeasured) through design of the study, however, they can not control those confounders that vary with time. A bidirectional case-crossover design can be used to adjust the exposure-time trends. In the areas of pharmaco-epidemiology, illness often influence the future use of medications, making a bidirectional study design problematic. Suissa's case-time-control design combines the case-crossover and the case-control design which could adjust for exposure-trend bias, but the control group may reintroduce selection bias, if the matching does not go well. We propose a "case-case-time-control" design which is an extension of the case-time-control design. However, rather than using a sample of external controls, we choose those future cases as controls for current cases to counter the bias that arising from temporal trends caused by exposure to the target of interest. In the end of this article we will discuss the strength and limitations of this design based on an applied example.

Psychological stress as a trigger for herpes zoster: might the conventional wisdom be wrong?

The causes for zoster remain largely unknown. Psychological stress is one commonly considered risk factor. We used self-controlled case series methods to look for increases in zoster following death or catastrophic health event occurring in a previously healthy spouse. We found no increase, although this stressor led to increased mental health visits.

Febrile seizures following measles and varicella vaccines in young children in Australia

BackgroundFebrile seizures (FS) are common in childhood with incidence peaking in the second year of life when measles and varicella-containing vaccines are administered. This study aimed to examine the vaccine-attributable risk of FS following separate administration of MMR and monovalent varicella vaccines (VV) prior to a planned change to MMRV as the second dose of measles-containing vaccine at 18 months of age. MethodsAll FS cases in children aged <5 years from 1st January 2012 to 30th April 2013 were identified from emergency department (ED) and inpatient databases at five Australian tertiary paediatric hospitals participating in PAEDS (Paediatric Active Enhanced Disease Surveillance). Immunization records were obtained from the Australian Childhood Immunization Register (ACIR). The relative incidence (RI) of FS following MMR dose 1 (MMR1) and VV in children aged 11–23 months was determined using the self-controlled case series (SCCS) method and used to calculate attributable risk. ResultsThere were 2013 FS episodes in 1761 children. The peak age at FS was 18 months. The risk of FS was significantly increased 5–12 days post receipt of MMR1 at 12 months (RI=1.9 [95% CI: 1.3–2.9]), but not after VV at 18 months (RI=0.6 [95% CI: 0.3–1.2]. The estimated excess annual number of FS post MMR1 was 24 per 100,000 vaccinated children aged 11–23 months (95% CI=7–49 cases per 100,000) or 1 per 4167 doses. ConclusionsOur study detected the expected increased FS risk post MMR1 vaccine at 12 months, but monovalent varicella vaccine at age 18 months was not associated with increased risk of FS. This provides baseline data to assess the risk of FS post MMRV, introduced in Australia as the second dose of measles-containing vaccine at 18 months of age in July 2013.

Safety of inactivated monovalent pandemic (H1N1) 2009 vaccination during pregnancy: A population-based study in Taiwan

BackgroundPregnant women were prioritized for H1N1 vaccination during the 2009–2010 pandemic. Safety concerns exist with vaccinating pregnant women, particularly in their first trimesters. MethodsWe linked computerized data on H1N1 vaccination, National Health Insurance, and Taiwan Birth Registry and identified events of spontaneous abortions (SABs) and all singleton births that occurred/delivered during November 1, 2009–September 30, 2010. The observation period for each case of SAB (6–19 weeks gestation) was divided into period at risk (1–28 days after vaccination) and control periods (the remaining person-days until SAB). The self-controlled case series method for truncated observational periods assessed the incidence rate ratio (IRR) of SAB during the 1–28 days compared with those in the control period. The case-control design matched each case of adverse fetal outcomes to up to 10 controls on fetal sex and year/month of pregnancy onset, and calculated matched odds ratio (OR) on H1N1 vaccination at <14 or ≥14 weeks gestation. ResultsSixty-five women with SAB had received H1N1 vaccination at 6–19 weeks gestation. The IRR of SAB for the risk period 1–28 days was 1.03 (95% confidence interval [CI] 0.55–1.93). Among the 147,294 live births and 1354 stillbirths, maternal H1N1 vaccine receipt at <14 weeks gestation was associated with significantly reduced likelihood of small for gestational age (SGA) birth (OR 0.72, 95% CI 0.61–0.84) and birth defect (OR 0.46, 95% CI 0.22–1.00), whereas receipt at ≥14 weeks gestation was associated with significantly reduced likelihood of stillbirth (OR 0.63, 95% CI 0.46–0.86), prematurity (OR 0.90, 95% CI 0.83–0.97), low birth weight (OR 0.81, 95% CI 0.74–0.88), and SGA birth (OR 0.90, 95% CI 0.84–0.97). ConclusionsH1N1 vaccination during pregnancy did not increase risk of SAB or adverse fetal outcomes.

Risk of stroke following herpes zoster: a self-controlled case-series study.

Herpes zoster is common and vaccine preventable. Stroke risk may be increased following zoster, but evidence is sparse and could be explained by differences between people with and without zoster. Our objective was to determine if stroke risk is increased following zoster. Within-person comparisons were undertaken using the self-controlled case-series method and data from the UK Clinical Practice Research Datalink (1987-2012). Participants had a first-ever diagnosis of zoster and stroke within the study period. Stroke incidence in periods following zoster was compared with incidence in other time periods. Age-adjusted incidence ratios (IRs) and 95% confidence intervals (CIs) were calculated. A total of 6584 individuals were included. Stroke rate was increased following zoster compared with the baseline unexposed period, then gradually reduced over 6 months: weeks 1-4 (age-adjusted IR, 1.63; 95% CI, 1.32-2.02), weeks 5-12 (IR, 1.42; 95% CI, 1.21-1.68), and weeks 13-26 (IR, 1.23; 95% CI, 1.07-1.42), with no increase thereafter. A stronger effect was observed for individuals with zoster ophthalmicus, rising to a >3-fold rate 5-12 weeks after zoster. Oral antivirals were given to 55% of individuals: IRs for stroke were lower among those receiving antivirals compared with those not treated, suggesting a protective effect. We have established an increased stroke rate within 6 months following zoster. Findings have implications for zoster vaccination programs, which may reduce stroke risk following zoster. The low antiviral prescribing rate needs to be improved; our data suggest that antiviral therapy may lead to a reduced stroke risk following zoster.

Increased risk of acute myocardial infarction and stroke during hemorrhagic fever with renal syndrome: a self-controlled case series study.

We recently observed that cardiovascular causes of death are common in patients with hemorrhagic fever with renal syndrome (HFRS), which is caused by hantaviruses. However, it is not known whether HFRS is a risk factor for the acute cardiovascular events of acute myocardial infarction (AMI) and stroke. Personal identification numbers from the Swedish HFRS patient database (1997-2012; n=6643) were cross-linked with the National Patient Register from 1987 to 2011. Using the self-controlled case series method, we calculated the incidence rate ratio of AMI/stroke in the 21 days after HFRS against 2 different control periods either excluding (analysis 1) or including (analysis 2) fatal AMI/stroke events. The incidence rate ratios for analyses 1 and 2 for all AMI events were 5.53 (95% confidence interval [CI], 2.6-11.8) and 6.02 (95% CI, 2.95-12.3) and for first AMI events were 3.53 (95% CI, 1.25-9.96) and 4.64 (95% CI, 1.83-11.77). The incidence rate ratios for analyses 1 and 2 for all stroke events were 12.93 (95% CI, 5.62-29.74) and 15.16 (95% CI, 7.21-31.87) and for first stroke events were 14.54 (95% CI, 5.87-36.04) and 17.09 (95% CI, 7.49-38.96). The majority of stroke events occurred in the first week after HFRS. Seasonal effects were not observed, and apart from 1 study, neither sex nor age interacted with the associations observed in this study. There is a significantly increased risk for AMI and stroke in the immediate time period after HFRS. Therefore, HFRS patients should be carefully monitored during the acute phase of disease to ensure early recognition of symptoms of impending AMI or stroke.

Circulation : Clinical Summaries

<i>Circulation</i> : Clinical Summaries

Application of a Self-Controlled Case Series Study to a Database Study in Children

Post-marketing surveillance activities are particularly important for safety issues in children, the elderly, and patients with severe comorbidities since these populations are usually excluded from clinical trials. In addition, using electronic databases for monitoring of safety of marketed products has been of considerable interest. This study aimed to clarify the advantages and difficulties of the self-controlled case series method relative to cohort studies in pharmacoepidemiological studies in children, using an administrative database, and to explore the impact on results of handling the period eligible for analysis and recurrent events in different ways. Datasets of only individuals who had the outcome of interest were derived from an anonymized hospital administrative database in Japan from April 2003 through August 2011. We calculated incidence rate ratios (IRRs) and their 95 % confidence intervals (CIs) for the risks of diarrhea, bronchitis, and eczema related to palivizumab treatment in young children. The analysis included 'first diagnosed' events or 'multiple' events during an eligible period. An eligible period was defined in two ways: first-time inpatient periods of more than 3 continuous days (EPA); and a continuous period in cases where the interval between visits was below the 75th percentile of the interval between visits for patients with the same diagnosis (EPB). We extracted data for 70,771 patients and identified 641 who were exposed to palivizumab. The age-adjusted IRRs for diarrhea, bronchitis, and eczema were 3.0 (95 % CI 1.7-5.4), 10.3 (95 % CI 8.0-13.2), and 16.9 (95 % CI 12-23), respectively, in multiple events and the EPB eligible period. The IRRs varied greatly between the two eligible periods. This method could be a useful tool in pharmacoepidemiological studies in children. Careful consideration in the handling of inpatient and outpatient periods, including sensitivity analyses, is necessary because this method is a within-individual comparison.

Glucagon-like peptide 1-based therapies and risk of pancreatitis: a self-controlled case series analysis.

Previous studies have suggested a link between glucagon-like peptide 1 (GLP-1)-based therapies and acute pancreatitis, while other studies have found no association. Because differences in diabetes severity may confound this relationship, a self-controlled case series (SCCS) analysis has been suggested as a means to control for individual-level confounding. We evaluated the relationship between GLP-1-based therapies and pancreatitis by SCCS method using a large observational database. We calculated the incidence density ratio of pancreatitis for exposure versus non-exposure to each drug. To examine the robustness of our findings, we performed sensitivity analyses by varying risk windows, using two pancreatitis definitions and including incident pancreatitis or all occurrences. From dispensing data on 1.2 million patients, we found 7992 sitagliptin-exposed patients and 3552 exenatide-exposed patients between 2004 and 2009. Using an ICD9/CPT-based case definition of pancreatitis, we identified 207 sitagliptin and 82 exenatide cases. Augmenting this definition with laboratory criteria increased our cohort to 245 sitagliptin and 96 exenatide cases. For sitagliptin and exenatide cases, respectively, the mean duration of observation was 5.2 and 5.5 years, and the mean duration of drug exposure was 0.7 and 0.5 years. For all analyses (including different pancreatitis definitions, risk periods, and incident or recurrent events), the incidence density ratios for development of pancreatitis during exposure versus non-exposure ranged from 0.68 to 1.46, with all having 95% confidence intervals containing 1. We found no association between the use of GLP-1-based therapies and pancreatitis using SCCS analysis in a large observational database.

Chickenpox and Risk of Stroke: A Self-controlled Case Series Analysis

There is good evidence that respiratory and other infections that cause systemic inflammation can trigger strokes; however, the role of specific infections is unclear. Case reports have highlighted chickenpox as a possible risk factor for arterial ischemic stroke, particularly in children, but rigorous studies are needed to determine and quantify any increased risk. We used anonymized electronic health records totaling >100 million person-years of observation from 4 UK primary care databases to identify individuals who had documented clinical chickenpox and a stroke or transient ischemic attack (TIA). Self-controlled case series methods were used to quantify any increased risk of first stroke or TIA in the 0-6 and 7-12 months following chickenpox compared to other observed time periods. We analyzed data within each database, and performed meta-analyses to obtain summary age-adjusted incidence ratios (IRs) separately for adults and children. Five hundred sixty eligible individuals (including 60 children) were identified who experienced chickenpox and a stroke or TIA during follow-up. Among children, there was a 4-fold increased risk of stroke in the 0-6 months after chickenpox (summary IR = 4.07; 95% confidence interval [CI], 1.96-8.45; I(2) = 0%). Among adults, there was a less marked increased risk with moderate between-database heterogeneity (random-effects summary IR = 2.13; 95% CI, 1.05-4.36; I(2) = 51%). There was no significant increased risk of stroke in the 7-12 months after chickenpox in children or adults, nor was there evidence of increased risk of TIA in either time period. Our study provides new evidence that children who experience chickenpox are at increased risk of stroke in the subsequent 6 months.

The association between multidose vaccinations and death: comparing case series methods when the first exposure changes the general risk of an event

Many case-control studies have shown a protective effect of vaccinations on the risk of sudden unexplained death (SUD). In this paper we compare the properties of different statistical methods in this situation, when the first vaccination appears to reduce the overall risk of an event (SUD). The first method is the self controlled case series (SCCS) method, which considers only subjects with an event during the observation time. This method yields unbiased estimates in the situation of non-censoring events. We show by simulation studies that the second method considered, the adjusted SCCS method, underestimates the parameter of interest, the effect of the first dose, when the general risk of SUD is lower in control periods after the first vaccination than in the period before vaccination. This type of bias could be eliminated by considering only cases who had received at least one vaccination. Additionally, we compare the adjusted SCCS method with the Cox model as a third method. Cox regression can take into account the time before the first vaccination, and this method yields unbiased estimates at modest effect sizes and short risk periods. We applied the SCCS method and Cox regression to data from the German study on sudden infant death (GeSID) and to a case series study from Italy (the HERA study) examining sudden unexpected deaths and vaccinations during the first 2 years of life. We show that the adjusted SCCS analysis with all cases underestimates the vaccination effect, as expected from the simulation analyses. Using Cox regression, we examined the general risk reduction after vaccination, as was the focus of the above mentioned studies. With a relative incidence of 0.8, our results were less pronounced than in the case-control analysis of the GeSID study (adjusted odds ratio: 0.51). SCCS analyses of both the GeSID and HERA studies yielded very similar estimates for the first and second vaccine doses.

Bias Correction of Risk Estimates in Vaccine Safety Studies With Rare Adverse Events Using a Self-controlled Case Series Design

The self-controlled case series (SCCS) method is often used to examine the temporal association between vaccination and adverse events using only data from patients who experienced such events. Conditional Poisson regression models are used to estimate incidence rate ratios, and these models perform well with large or medium-sized case samples. However, in some vaccine safety studies, the adverse events studied are rare and the maximum likelihood estimates may be biased. Several bias correction methods have been examined in case-control studies using conditional logistic regression, but none of these methods have been evaluated in studies using the SCCS design. In this study, we used simulations to evaluate 2 bias correction approaches-the Firth penalized maximum likelihood method and Cordeiro and McCullagh's bias reduction after maximum likelihood estimation-with small sample sizes in studies using the SCCS design. The simulations showed that the bias under the SCCS design with a small number of cases can be large and is also sensitive to a short risk period. The Firth correction method provides finite and less biased estimates than the maximum likelihood method and Cordeiro and McCullagh's method. However, limitations still exist when the risk period in the SCCS design is short relative to the entire observation period.

A New Self-Controlled Case Series Method for Analyzing Spontaneous Reports of Adverse Events After Vaccination

In this paper, we propose new methods for analyzing cases of vaccine adverse events spontaneously reported to a surveillance database. The methods use the self-controlled case series approach, extended in several ways with parametric and nonparametric assumptions to account for the specific features of the data (large amount of underreporting and variation of reporting with time since vaccination). This work was motivated by the documented risk of intussusception after RotaShield vaccination (Wyeth-Lederle Vaccines, Radnor, Pennsylvania) and used worldwide spontaneous reports of intussusception occurring after Rotarix vaccination (GlaxoSmithKline Biologics, Research Triangle Park, North Carolina) collected between January 2004 and February 2010. The estimated risk during the 3- to 7-day period after vaccination was approximately 5 times higher after dose 1 of Rotarix than after dose 2, which is similar to published findings on the same topic. We undertook a large simulation study to evaluate the performance of the method in different scenarios, including its robustness to different sample sizes and time-dependent reporting functions. The bias was generally small, the type I error rate was correctly controlled, and the power to detect a risk ratio of 4 was satisfactory, provided that the sample size was over 100. The proposed methods are an effective way to explore and quantify vaccine safety signals from spontaneous reports.