C. D. J. Holman

The Causes of Malignant Melanoma: Results from the West Australian Lions Melanoma Research Project

In Australia malignant melanoma ranks forth among cancers as a cause of morbidity and premature death (Armstrong 1985). For this reason, and because the State of Western Australia has one of the highest incidence rates of melanoma in the world (Holman et al. 1980), we undertook a multidisciplinary research program embracing the epidemiology, histopathology, and clinical management of melanoma. A major component of our research was a population-based case-control study in which constitutional traits, sunlight exposure, hormones, diet and other possible causal factors were evaluated. We have already published several reports of the results relating to particular subject areas (Holman and Armstrong 1983, 1984 a, b; Holman et al. 1984 b, 1985; English et al. 1985). In this report we aim to bring together in summary form the key results of the project to date and to present new data on diet and other factors.

Ought low alcohol intake to be promoted for health reasons

There is increasingly widespread acceptance that alcohol taken in moderation by the population aged 35 years or older reduces the risks of ischaemic heart disease and all-cause mortality. Ten causal criteria are used to evaluate the scientific evidence for a protective effect of low alcohol intake on ischaemic heart disease. Inferences for public policy are then assessed using the principles of beneficence, non-maleficence, justice and autonomy to support a framework of nine ethical considerations: intervention versus causation; effect modification by gender, smoking, biogenetic and other factors; inappropriate adoption of recommendations; competing hazards between atherosclerotic disease and cancer; opportunity cost; equity of access; the value system used to judge outcomes; the degree of social influence warranted; and consent and responsibility. We conclude that in the absence of more adequate scientific knowledge and informed community debate it is unethical to promote low alcohol intake as a preventive health measure.

A Comparison of Mental Health Hospital Admissions in a Cohort of Heroin Users Prior to and After Rapid Opiate Detoxification and Oral Naltrexone Maintenance

Mental health (MH) hospital admissions were investigated in a cohort (N = 1184) of heroin dependent persons using linked health records. All MH in-patient admissions were extracted 36 months before to 36 months after commencing rapid opioid detoxification (ROD) and oral naltrexone. Results show that the incidence rate ratio (IRR) of drug-related and other MH admissions peaked in the 3 months immediately prior to treatment. All categories subsequently declined to baseline levels by 36 months following treatment. The authors conclude that treatment for heroin dependence reduces risk of MH admissions.

Identifying cases or early breast cancer in Australia using prescription and other health services claims and self-report

Background: Several propensity score (PS) balance measures have been compared in simulated data with normally distributed covariates. Comparisons in data with binary or mixed covariate distributions and rare outcomes, typical of pharmacoepidemiologic data sets, are scarce. Objectives: To compare balance measures in simulated data with various covariate distributions and rare outcomes. Methods: We performed Monte Carlo simulations to examine the relative ability of different balance measures to select PS models that yielded the least biased estimates. In different simulations, covariates were binary, normal or gamma distributed, considering sample sizes of n = 400, 1,600, and 3,000, incidence of outcomes of 10% and 25%, and strength of exposure-outcome association of OR = 1 and 2. Bias was estimated as the difference between the true marginal effect and the effect estimate obtained from a logistic regression model with PS as a covariate. The balance of covariates between treatment groups was assessed using the standardised difference (SD), Kolmogorov Smirnov (KS) distance, Levy distance (Levy) and overlapping coefficient (OVL). Pearsons correlation coefficients (r) between these balance measures and bias were calculated. Results: With large sample sizes, all balance measures were similarly correlated with bias irrespective of covariate distributions, strength of the effect, and prevalence of outcome (e.g., when all covariates were binary, OR = 2.0, n = 3,000 and incidence of outcome = 25%: correlations were 0.76, 0.79, 0.79, and -0.79 for SD, KSD, Levy and OVL, respectively). These correlations decreased with smaller sample sizes (e.g., for n = 400: 0.51, 0.20, 0.17, and -0.43, for SD, KSD, Levy distance and OVL, respectively). Incidence of the outcome and strength of the exposure-outcome relation didnt have much impact. For sample sizes smaller than 800, SD showed a better correlation with bias than the other balance measures. Conclusions: The SD or KS performed best across different simulation scenarios and are recommended for reporting the amount of balance reached and selecting the final PS model.Background: The effect of guideline changes on trends of prescription drug use are commonly studied by age and over time period. This masks the birth cohort dimension which affects the age-specific trends in each time period. Objectives: We investigated whether including the birth cohort dimension in time series analysis leads to a more accurate estimation of the effect of a guideline change on the trend of benzodiazepine use. Methods: Outpatient pharmacy data from a drug prescription database in the Netherlands (IADB.nl) were used to obtain the age- (18-85) and sex -specific number of users of benzodiazepine (ATC: N05BA and N05CD) per 1,000 population (prevalence) per quarter year from 1998 to 2008. We studied the prevalence over time by age groups and by birth cohorts. Interrupted time series analyses were performed to the de-seasonalized trend to estimate the effect of the guideline change in 2001. Results: From 1998 to 2008 the overall age-standardized prevalence of benzodiazepine use per 1,000 population declined from ∼55 for males and ∼105 for females to ∼42 for males and ∼78 for females, this decline strengthened in 2001 for both sexes (significant slope change of p <0.05). Similar patterns over time, including slope changes (p <0.05), can be found within the majority of age categories, providing no additional information. Within birth cohorts the prevalence first increased over time but after 2001 this increase stopped or weakened (p <0.05), indicating a reduction in starters or less chronic use of benzodiazepine. Conclusions: Studying trends within birth cohorts can be a useful addition to time series analyses because the same or similar individuals are followed over time, making analysis more intuitive. This is not the case for trends within age-categories, potentially leading to less informed conclusions about guideline effects.Background: Epidemiological research has been criticized as being unreliable. Scientific evidence is strengthened when the study procedures of important findings are transparent, open for review, and easily reproduced by different investigators and in various settings. Studies often have common scientific workflows. The development of generalized execution engines, reusing epidemiological software/script program code for specific clinical questions, can serve as a valuable tool for transparent and reproducible research. Objectives: Learn about standards for transparent, reproducible and reusable research and how it is being applied in pharmacoepidemiolgy. Description: The focus of the symposium will be transparent, reproducible and reusable research. Principles of reproducible research in the context of the Medication History Estimator will be discussed. In addition, an overview of the IMI-PROTECT WP2: Framework for pharmacoepidemiologic studies and the Observational Medical Outcomes Partnership will be given. Outline: 1. Principles of reliability (Helga Gardarsdottir): Introduction to basic principles of transparent reproducible and reusable research. 2. Standardization (Huifang Liang): Standardization of data for drug utilization studies. A discussion of steps involved to convert the raw data into the readily usable data, including how to impute certain fields with examples. 3. Demonstration of the Medication History Estimator (MHE) and a description of the VINCI EpiTools (Brian Sauer). The MHE will be presented to demonstrate concepts of transparency, reproducibility and reuse. 4. The IMI-PROTECT project (Olaf Klungel & Robert Reynolds). Experiences with developing, testing and disseminating methodological standards for the design, conduct and analysis of database studies will be discussed. 5. Lessons from the Observational Medical Outcomes Partnership (Patrick Ryan, PhD). Standardized analytics tools developed by the OMOP community to characterize, visualize, and explore the effects of medical products within a distributed network of observational databases will be presented. 6. Closing summary/discussion. Chairs: Helga Gardarsdottir & Brian Sauer.Background: The Netherlands Pharmacovigilance Centre Lareb received six reports of hearing impairment in association with oral terbinafine use. This study describes these cases and provides support for this association from the Lareb database of spontaneous ADR reporting and from Vigibase, the adverse drug reaction database of the WHO UppsalaMonitoring Centre. Objectives: The objective of the study is to identify whether the observed association between oral terbinafine use and hearing impairment, based on several cases received by Lareb, supports a safety signal. Methods: Cases of hearing impairment in oral terbinafine users are described. In a case/non-case analysis, the strength of the association in Vigibase and the Lareb database was determined by calculating the reporting odds ratios (ROR), adjusted for possible confounding by age, sex and possibly ototoxic concomitant medication. RORs are calculated for deafness, hypoacusis, and the combination of both, defined as hearing impairment. Results: In the Lareb database, six reports concerning individuals aged 31-82 years, who developed hearing impairment after starting oral terbinafine, are present. The use of oral terbinafine is disproportionally associated with hypoacusis in both the Lareb database (adjusted ROR = 3.9, 95% CI: 1.7-9.0), and in Vigibase (adjusted ROR = 1.7, 95% CI: 1.0-2.8). Deafness is not disproportionally present in either of the databases. The pharmacological action of terbinafine is based on the inhibition of squalene epoxidase, an enzyme present in both fungal and human cells. This inhibition might result in decreased cholesterol levels in, among others, the outer hair cells of the cochlea, possibly leading to impaired cochlear function and hearing impairment. Conclusions: To our knowledge, hearing impairment associated with oral terbinafine use has not been described before. A causal relationship between the use of oral terbinafine and hearing impairment is possible, based on statistical analysis of reported cases in different databases and a possible pathophysiological explanation.Background: In observational studies of time-varying exposure and confounders, the use of propensity score (PS) is limited to assigning weights as in marginal structural models (MSMs). Stratification and conditioning on time-varying cofounders which are also intermediates can induce collider-stratification bias and adjust-away the (indirect) effect of exposure. Similar bias could be expected when one conditions on time-dependent PS. Objectives: We explored collider-stratification and confounding bias due to conditioning or stratifying on timedependent PS in a clinical example on the effect of inhaled short and long-acting beta2-agonist use (SABA and LABA, respectively) on coronary heart disease (CHD). Methods: A cohort of patients with an indication for SABA and/or LABA use was extracted from the Netherlands University Medical Center Utrecht General Practitioner Research Network. Information from 1995 to 2005 was used. SABA and LABA use and potential confounders were ascertained on 3 month intervals. Follow-up began the first day of diagnosis of bronchitis, asthma, or COPD and ended at the occurrence of CHD, death, unregistration with the GP, or end of the study, whichever occurred first. HR were estimated using PS stratification as well as covariate adjustment and compared with those of MSMs in both SABA and LABA separately. In MSMs, censoring was accounted for by including inverse probability of censoring weights. Results: The crude HR of CHD was 0.90 [95% CI: 0.63, 1.28] and 1.55 [95% CI: 1.06, 2.62] in SABA and LABA users respectively. When PS stratification, adjustment using PS, and MSMs were used, the HRs were 1.09 [95%CI: 0.74, 1.61], 1.07 [95% CI: 0.72, 1.60], and 0.86 [95% CI: 0.55, 1.34] for SABA, and 1.09 [95%CI: 0.74, 1.62], 1.13 [95%CI: 0.76, 1.67], 0.77 [95% CI: 0.45, 1.33] for LABA, respectively. Conclusions: Results were similar for different PS methods, but systematically higher than those of MSMs. When treatment and confounders vary during follow-up, conditioning or stratification on time-dependent PS may induce substantial collider-stratification or confounding bias. Hence, the use of methods such as MSMs is recommended.Background: Compared to RCTs, observational (nonrandomized) studies often comprise larger sample sizes, which gives adequate power to study interaction. Observational studies, however, are prone to confounding. Objectives: To determine the validity of subgroup and interaction effects (differences between subgroups) for different study designs. Methods: We compared effects of medical interventions based on observational studies, RCTs, and Individual Patient Data Meta-Analysis of RCTs (IPDMAs; reference) on three different clinical topics: (1) mammography screening effects on breast cancer mortality; (2) CABG and all-cause mortality; (3) statins and the incidence of major coronary events. Main, subgroup, and interaction effects were compared. Results: Main and subgroup effects were comparable with respect to the direction of effects for IPDMAs, RCTs, and observational studies. Small differences in the magnitude of subgroup effects in observational studies yielded different interactions compared to IPDMA. In the mammography example the Ratio of Risk Ratios (RRR) (i.e., interaction effect) among observational studies was 1.46 (95% CI 1.09;1.96) compared to an IPDMA effect of 110 (95% CI 0.89;1.37). For the CABG studies the observational RRR was 1.03 (95% CI 0.84;1.26), whereas in the IPDMA this was 1.40 (95% CI 1.08;11.81). Finally, in the statin example the RRR was 1.35 (95% CI 113;1.61) for observational studies, in the IPDMA this was 0.90 (95% CI 0.84;0.97). Conclusions: Main and subgroup effects based on observational data are in line with main and subgroup effects in IPDMAs based on RCTs, yet interactions may differ substantially.Background: Both antidepressants (AD) and benzodiazepines (BZD) have been associated with an increased risk for hip fracture. However, the hazard function is not constant over exposure time and differs for these two medication classes. Hence, the timing of initiation of co-use will determine the overall hazard function. Objectives: The aim of this study was to describe timing of BZD co-medication use among AD users. Methods: The study population included patients from the Netherlands Information Network of General Practice who initiated ADs between 2002 and 2009. AD-treatment episodes were constructed for each patient assuming the start of a new episode when 90 days elapsed between the theoretical end-date of a prior AD prescription (Rx) and the start-date of the next AD Rx. Within the first AD episode, three groups of BZD co-use were defined: “simultaneous” (simultaneous start of AD and BZD, no BZD during 182 days prior to start), “before” and “during” starters. These groups were described according to intensity of BZD use (number of Rxs), mono/polytherapy of AD and duration of AD episode. Results: The study population consisted of 16,617 AD users. The mean age was 50 years (SD = 18) and 63.2% were female. The median duration of the AD episode was 80 days (IQR = 259). Of 28.5% of patients used both AD and BZD. Of these, 57.2% started BZD before the AD, 19.0% were “simultaneous” starters and 23.8% started BZD during their first AD episode. In general, “simultaneous” starters were younger (mean = 45.3 vs. 56.4 years) yet had less intensity of BZD use (mean = 4.3 vs. 7.5 Rxs) compared to “before” starters. “After” starters were slightly older (mean = 48.0 vs. 45.3 years), had more AD polytherapy (20.2% vs. 14.9%) however less intensive BZD use (mean = 3.7 vs. 4.3 Rxs) compared to “simultaneous” starters. Conclusions: Timing of initiating BZD use in AD users is important, as each BZD co-use group is expected to have a different overall hazard function for hip fracture as opposed to when co-medication is defined as constant over time. To calculate accurate hazard function, it is important to take into consideration the timing of initiation of co-medication use in pharmacoepidemiological studies.Background: Pregnancy outcomes in women with preexisting diabetes are known to be worse than in the healthy population: rates of congenital malformations have been reported to be 2 to 10-fold higher. This is due to poor glycaemic control. Objectives: To determine rates of congenital malformations in babies of mothers with type1 (DM1) and type2 (DM2) diabetes and compare with the healthy population. Methods: Patients with pre-existing DM1 or DM2 during pregnancy were identified on the General Practice Research Database using diagnoses, prescribing, use of testing equipment and referral records. Mothers were matched to babies using registration records with the same family number, within 2 months of delivery date. Major congenital malformations were identified in the baby’s record and malformations verified with at least two related medical records; free text entries were checked. Results: Between January 1992 and March 2007 1,057 DM1 patients had 1,329 pregnancies and 365 DM2 patients had 441 pregnancies that resulted in a live birth. Eighty-four major malformations were found in 78 babies of DM1 mothers: 41 were cardiac, 11 were urogenital and eight were limb defects. Twenty-six major malformations were identified in 22 babies of DM2 mothers (seven cardiac defects). Overall, 5.9% of babies of DM1 mothers and 5.0% of babies of DM2 mothers had a malformation compared to 2–3% in the healthy population. The malformation rate for babies of DM2 mothers was comparable with another study (5.8%) but was lower for DM1 (8.2%). The proportion of cardiac malformations in babies of DM1 mothers was three times higher than in the healthy population. Conclusions: Babies whose mothers had DM1 or DM2 during pregnancy had double the rate of malformations. Pregnancy loss due to malformation has not been included but previous work found a greater proportion of terminations for medical reasons (1% in DM1, 1.8% in DM2) compared to healthy pregnancies (0.8%). Limitations include potentially differential recording of malformations between babies of mothers with DM1 and DM2 and the lack of records of glycaemic control. Differences in rates of malformations between treatments including analogue and human insulin will be evaluated next.