[Show abstract][Hide abstract] ABSTRACT: Purpose:
Research on medication safety in pregnancy often utilizes health plan and birth certificate records. This study discusses methods used to link mothers with infants, a crucial step in such research.
We describe how eight sites participating in the Medication Exposure in Pregnancy Risk Evaluation Program created linkages between deliveries, infants and birth certificates for the 2001-2007 birth cohorts. We describe linkage rates across sites, and for two sites, we compare the characteristics of populations linked using different methods.
Of 299,260 deliveries, 256,563 (86%; range by site, 74-99%) could be linked to infants using a deterministic algorithm. At two sites, using birth certificate data to augment mother-infant linkage increased the representation of mothers who were Hispanic or non-White, younger, Medicaid recipients, or had low educational level. A total of 236,460 (92%; range by site, 82-100%) deliveries could be linked to a birth certificate.
Tailored approaches enabled linking most deliveries to infants and to birth certificates, even when data systems differed. The methods used may affect the composition of the population identified. Linkages established with such methods can support sound pharmacoepidemiology studies of maternal drug exposure outside the context of a formal registry.
Pharmacoepidemiology and Drug Safety 07/2013; 22(7). DOI:10.1002/pds.3443 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Purpose:
To validate an algorithm that uses delivery date and diagnosis codes to define gestational age at birth in electronic health plan databases.
Using data from 225,384 live born deliveries to women aged 15-45 years in 2001-2007 within eight of the 11 health plans participating in the Medication Exposure in Pregnancy Risk Evaluation Program, we compared (1) the algorithm-derived gestational age versus the "gold-standard" gestational age obtained from the infant birth certificate file and (2) the prenatal exposure status of two antidepressants (fluoxetine and sertraline) and two antibiotics (amoxicillin and azithromycin) as determined by the algorithm-derived versus the gold-standard gestational age.
The mean algorithm-derived gestational age at birth was lower than the mean obtained from the birth certificate file among singleton deliveries (267.9 vs 273.5 days) but not among multiple-gestation deliveries (253.9 vs 252.6 days). The algorithm-derived prenatal exposure to the antidepressants had a sensitivity and a positive predictive value of ≥95%, and a specificity and a negative predictive value of almost 100%. Sensitivity and positive predictive value were both ≥90%, and specificity and negative predictive value were both >99% for the antibiotics.
A gestational age algorithm based upon electronic health plan data correctly classified medication exposure status in most live born deliveries, but trimester-specific misclassification may be higher for drugs typically used for short durations.
Pharmacoepidemiology and Drug Safety 05/2013; 22(5). DOI:10.1002/pds.3407 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: In post-marketing drug safety surveillance, data mining can potentially detect rare but serious adverse events. Assessing an entire collection of drug–event pairs is traditionally performed on a predefined level of granularity. It is unknown a priori whether a drug causes a very specific or a set of related adverse events, such as mitral valve disorders, all valve disorders, or different types of heart disease. This methodological paper evaluates the tree-based scan statistic data mining method to enhance drug safety surveillance.
We use a three-million-member electronic health records database from the HMO Research Network. Using the tree-based scan statistic, we assess the safety of selected antifungal and diabetes drugs, simultaneously evaluating overlapping diagnosis groups at different granularity levels, adjusting for multiple testing. Expected and observed adverse event counts were adjusted for age, sex, and health plan, producing a log likelihood ratio test statistic.
Out of 732 evaluated disease groupings, 24 were statistically significant, divided among 10 non-overlapping disease categories. Five of the 10 signals are known adverse effects, four are likely due to confounding by indication, while one may warrant further investigation.
The tree-based scan statistic can be successfully applied as a data mining tool in drug safety surveillance using observational data. The total number of statistical signals was modest and does not imply a causal relationship. Rather, data mining results should be used to generate candidate drug–event pairs for rigorous epidemiological studies to evaluate the individual and comparative safety profiles of drugs. Copyright
Pharmacoepidemiology and Drug Safety 05/2013; DOI:10.1002/pds.3423 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Background:
Drug adverse event (AE) signal detection using the Gamma Poisson Shrinker (GPS) is commonly applied in spontaneous reporting. AE signal detection using large observational health plan databases can expand medication safety surveillance.
Using data from nine health plans, we conducted a pilot study to evaluate the implementation and findings of the GPS approach for two antifungal drugs, terbinafine and itraconazole, and two diabetes drugs, pioglitazone and rosiglitazone. We evaluated 1676 diagnosis codes grouped into 183 different clinical concepts and four levels of granularity. Several signaling thresholds were assessed. GPS results were compared to findings from a companion study using the identical analytic dataset but an alternative statistical method-the tree-based scan statistic (TreeScan).
We identified 71 statistical signals across two signaling thresholds and two methods, including closely-related signals of overlapping diagnosis definitions. Initial review found that most signals represented known adverse drug reactions or confounding. About 31% of signals met the highest signaling threshold.
The GPS method was successfully applied to observational health plan data in a distributed data environment as a drug safety data mining method. There was substantial concordance between the GPS and TreeScan approaches. Key method implementation decisions relate to defining exposures and outcomes and informed choice of signaling thresholds.
[Show abstract][Hide abstract] ABSTRACT: This study aims to estimate the prevalence of and temporal trends in prenatal antipsychotic medication use within a cohort of pregnant women in the U.S. We identified live born deliveries to women aged 15-45 years in 2001-2007 from 11 U.S. health plans participating in the Medication Exposure in Pregnancy Risk Evaluation Program. We ascertained prenatal exposure to antipsychotics from health plan pharmacy dispensing files, gestational age from linked infant birth certificate files, and ICD-9-CM diagnosis codes from health plan claims files. We calculated the prevalence of prenatal use of atypical and typical antipsychotics according to year of delivery, trimester of pregnancy, and mental health diagnosis. Among 585,615 qualifying deliveries, 4,223 (0.72 %) were to women who received an atypical antipsychotic and 548 (0.09 %) were to women receiving a typical antipsychotic any time from 60 days before pregnancy through delivery. There was a 2.5-fold increase in atypical antipsychotic use during the study period, from 0.33 % (95 % confidence interval: 0.29 %, 0.37 %) in 2001 to 0.82 % (0.76 %, 0.88 %) in 2007, while the use of typical antipsychotics remained stable. Depression was the most common mental health diagnosis among deliveries to women with atypical antipsychotic use (63 %), followed by bipolar disorder (43 %) and schizophrenia (13 %). The number and proportion of pregnancies exposed to atypical antipsychotics has increased dramatically in recent years. Studies are needed to examine the comparative safety and effectiveness of these medications relative to other therapeutic options in pregnancy.
Archives of Women s Mental Health 02/2013; 16(2). DOI:10.1007/s00737-013-0330-6 · 2.16 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To evaluate the validity of health plan and birth certificate data for pregnancy research.
A retrospective study was conducted using administrative and claims data from 11 U.S. health plans and corresponding birth certificate data from state health departments. Diagnoses, drug dispensings, and procedure codes were used to identify infant outcomes (cardiac defects, anencephaly, preterm birth, and neonatal intensive care unit [NICU] admission) and maternal diagnoses (asthma and systemic lupus erythematosus [SLE]) recorded in the health plan data for live born deliveries between January 2001 and December 2007. A random sample of medical charts (n = 802) was abstracted for infants and mothers identified with the specified outcomes. Information on newborn, maternal, and paternal characteristics (gestational age at birth, birth weight, previous pregnancies and live births, race/ethnicity) was also abstracted and compared to birth certificate data. Positive predictive values (PPVs) were calculated with documentation in the medical chart serving as the gold standard.
PPVs were 71% for cardiac defects, 37% for anencephaly, 87% for preterm birth, and 92% for NICU admission. PPVs for algorithms to identify maternal diagnoses of asthma and SLE were ≥ 93%. Our findings indicated considerable agreement (PPVs > 90%) between birth certificate and medical record data for measures related to birth weight, gestational age, prior obstetrical history, and race/ethnicity.
Health plan and birth certificate data can be useful to accurately identify some infant outcomes, maternal diagnoses, and newborn, maternal, and paternal characteristics. Other outcomes and variables may require medical record review for validation. Copyright
Pharmacoepidemiology and Drug Safety 01/2013; 22(1). DOI:10.1002/pds.3319 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: A new meningococcal conjugate vaccine (MCV4) was introduced in 2005. Shortly after, case reports of Guillain–Barré syndrome (GBS), a serious demyelinating disease, began to be reported to the Vaccine Adverse Event Reporting System. In 2006, the Centers for Disease Control and Prevention and the Food and Drug Administration requested the evaluation of GBS risk after MCV4 vaccination. We conducted a study to assess the risk of GBS after MCV4 vaccination using health plan administrative and claims data together with the review of primary medical records of potential cases.
Retrospective cohort study among 12.6 million 11- to 21-year-old members of five US health plans with a total membership of 50 million. Automated enrollment and medical claims data from March 2005 through August 2008 were used to identify the population, the vaccinations administered, and the medical services associated with possible GBS. Medical records were reviewed and adjudicated by a neurologist panel to confirm cases of GBS. The study used distributed data analysis methods that minimized sharing of protected health information.
We confirmed 99 GBS cases during 18,322,800 person-years (5.4/1,000,000 person-years). More than 1.4 million MCV4 vaccinations were observed. No confirmed cases of GBS occurred within 6 weeks after vaccination. The upper 95% CI for the attributable risk of GBS associated with MCV4 is estimated as 1.5 cases per 1,000,000 doses.
Among members of five US health plans, MCV4 vaccination was not associated with increased GBS risk. Copyright
Pharmacoepidemiology and Drug Safety 12/2012; 21(12). DOI:10.1002/pds.3321 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Background/Aims Active post-marketing drug safety surveillance has traditionally focused on predefined drug-adverse event (AE) pairs. However, evaluating predefined pairs does not illuminate unsuspected potential AEs. Drug safety data mining is a surveillance approach that formally evaluates the relationship between medications and a very large number of AEs. We used varying specifications of the tree-based scan statistic data mining method (TreeScan) to search for adverse events among clozapine drug users. Methods Electronic health records from three HMO Research Network health plans were assessed. We used TreeScan to evaluate a hierarchical clinical classification to identify signals of excess risk during prevalent drug exposure as compared to unexposed time. The test statistic - a Poisson based log likelihood ratio - is adjusted for multiple testing inherent in the many potential AEs evaluated. Four alternate specifications were incorporated: ramp-up periods of 180 and 400 days and outcome definitions using inpatient plus outpatient diagnoses and inpatient diagnoses only. For each drug and specification, we calculated expected and observed counts for each level of the hierarchical tree, adjusting for age, sex, and health plan. Results We identified 242,000 to 580,000 exposed clozapine days and 150 to 345 exposed outcomes across the different specifications. Both ramp-up periods found 17 unique statistical signals using inpatient plus outpatient diagnoses. Of those, several represent confounding by indication (three signals in mental health, two for injury and poisoning) and others are known AEs (e.g., convulsions, hypotension, GI system). Limiting outcomes to the inpatient setting reduced the number of signals to 14 (180-day ramp-up) and 10 (400-day ramp-up). Overall, the inpatient-only AE signals were in the same clinical systems, with some exceptions. The inpatient specification signaled for circulatory events and diseases of the heart, but hypotension was no longer found. Genitourinary AEs signaled using inpatient plus outpatient diagnoses but were not identified using inpatient diagnoses only. Conclusions Data mining using electronic health records is an important complement to other post-marketing drug safety research. Once specifications are finalized, TreeScan will be applied to assess the safety of over 100 oral outpatient medications.
Clinical Medicine & Research 11/2011; 9(3-4):180. DOI:10.3121/cmr.2011.1020.c-c5-01
[Show abstract][Hide abstract] ABSTRACT: To evaluate the validity of health plan administrative and claims data to identify pre-gestational and gestational diabetes, obesity, and ultrasounds among pregnant women.
A retrospective study was conducted using the administrative and claims data of three health plans participating in the HMO Research Network. Diagnoses, drug dispensings, and procedure codes were used to identify diabetes, obesity, and ultrasounds among women who were pregnant between January 2006 and December 2008. A random sample of medical charts (n = 222) were abstracted. Positive predictive values (PPVs) were calculated. Sensitivity also was calculated for obesity among women for whom body mass index data were available in electronic medical records at two sites.
Overall, 190 of 222 cases of diabetes (86%) were confirmed (82% for gestational diabetes and 74% for pre-gestational diabetes). The PPV for codes to identify ultrasounds was 80%. Whereas the PPV for obesity-related diagnosis codes was high (93%), and the sensitivity was low (33%).
Health plan administrative and claims data can be used to accurately identify pre-gestational and gestational diabetes and ultrasounds. Obesity is not consistently coded.
Pharmacoepidemiology and Drug Safety 11/2011; 20(11):1168-76. DOI:10.1002/pds.2217 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: To describe a program to study medication safety in pregnancy, the Medication Exposure in Pregnancy Risk Evaluation Program (MEPREP). MEPREP is a multi-site collaborative research program developed to enable the conduct of studies of medication use and outcomes in pregnancy. Collaborators include the U.S. Food and Drug Administration and researchers at the HMO Research Network, Kaiser Permanente Northern and Southern California, and Vanderbilt University. Datasets have been created at each site linking healthcare data for women delivering an infant between January 1, 2001 and December 31, 2008 and infants born to these women. Standardized data files include maternal and infant characteristics, medication use, and medical care at 11 health plans within 9 states; birth certificate data were obtained from the state departments of public health. MEPREP currently involves more than 20 medication safety researchers and includes data for 1,221,156 children delivered to 933,917 mothers. Current studies include evaluations of the prevalence and patterns of use of specific medications and a validation study of data elements in the administrative and birth certificate data files. MEPREP can support multiple studies by providing information on a large, ethnically and geographically diverse population. This partnership combines clinical and research expertise and data resources to enable the evaluation of outcomes associated with medication use during pregnancy.
Maternal and Child Health Journal 10/2011; 16(7):1349-54. DOI:10.1007/s10995-011-0902-x · 2.24 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: Black box warnings (BBWs) are the Food and Drug Administration's (FDA) strongest labeling requirements for high-risk medicines. It is unknown how frequently physicians prescribe BBW drugs and whether they do so in compliance with the warnings. The purpose of the present study was to assess the frequency of use of BBW medications in ambulatory care and prescribing compliance with BBW recommendations.
This retrospective study used automated claims data of 929 958 enrollees in 10 geographically diverse health plans in the United States to estimate frequency of use in ambulatory care of 216 BBW drugs/drug groups between 1/1/99 and 31/6/01. We assessed dispensing compliance with the BBW requirements for selected drugs.
During a 30-month period, more than 40% of enrollees received at least one medication that carried a BBW that could potentially apply to them. We found few instances of prescribing during pregnancy of BBW drugs absolutely contra-indicated in pregnancy. There was almost no co-prescribing of contra-indicated drugs with the two QT-interval-prolonging BBW drugs evaluated. Most non-compliance occurred with recommendations for baseline laboratory monitoring (49.6% of all therapy initiations that should have been accompanied by baseline laboratory monitoring were not).
Many individuals receive drugs considered to carry the potential for serious risk. For some of these drugs, use is largely consistent with their BBW, while for others it is not. Since it will not be possible to avoid certain drug- associated risks, it will be important to develop effective methods to use BBWs and other methods to minimize risks.
Pharmacoepidemiology and Drug Safety 06/2006; 15(6):369-86. DOI:10.1002/pds.1193 · 2.94 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: The advent of domestic bioterrorism has emphasized the need for enhanced detection of clusters of acute illness. We describe a monitoring system operational in eastern Massachusetts, based on diagnoses obtained from electronic records of ambulatory-care encounters. Within 24 hours, ambulatory and telephone encounters recording patients with diagnoses of interest are identified and merged into major syndrome groups. Counts of new episodes of illness, rates calculated from health insurance records, and estimates of the probability of observing at least this number of new episodes are reported for syndrome surveillance. Census tracts with unusually large counts are identified by comparing observed with expected syndrome frequencies. During 1996-1999, weekly counts of new cases of lower respiratory syndrome were highly correlated with weekly hospital admissions. This system complements emergency room- and hospital-based surveillance by adding the capacity to rapidly identify clusters of illness, including potential bioterrorism events.
[Show abstract][Hide abstract] ABSTRACT: Gaps in disease surveillance capacity, particularly for emerging infections and bioterrorist attack, highlight a need for efficient, real time identification of diseases.
We studied automated records from 1996 through 1999 of approximately 250,000 health plan members in greater Boston.
We identified 152,435 lower respiratory infection illness visits, comprising 106,670 episodes during 1,143,208 person-years. Three diagnoses, cough (ICD9CM 786.2), pneumonia not otherwise specified (ICD9CM 486) and acute bronchitis (ICD9CM 466.0) accounted for 91% of these visits, with expected age and sex distributions. Variation of weekly occurrences corresponded closely to national pneumonia and influenza mortality data. There was substantial variation in geographic location of the cases.
This information complements existing surveillance programs by assessing the large majority of episodes of illness for which no etiologic agents are identified. Additional advantages include: a) sensitivity, uniformity and efficiency, since detection of events does not depend on clinicians' to actively report diagnoses, b) timeliness, the data are available within a day of the clinical event; and c) ease of integration into automated surveillance systems. These features facilitate early detection of conditions of public health importance, including regularly occurring events like seasonal respiratory illness, as well as unusual occurrences, such as a bioterrorist attack that first manifests as respiratory symptoms. These methods should also be applicable to other infectious and non-infectious conditions. Knowledge of disease patterns in real time may also help clinicians to manage patients, and assist health plan administrators in allocating resources efficiently.
BMC Public Health 02/2001; 1(1):9. DOI:10.1186/1471-2458-1-9 · 2.26 Impact Factor
[Show abstract][Hide abstract] ABSTRACT: The SAS system is known not to support any more or less developed Bayesian method. At the same time a Bayesian framework is the ideal environment for resolving the problem of model selection uncertainty (which is important for getting proper inference based on the model), though at a price of very complex and time-consuming algorithms. In this presentation, which is a continuation of our SUGI'2000 paper, the possibility of avoiding the complexities of fully developed Bayesian methods is discussed. A Bayesian-like approach to resolving the problem of model selection uncertainty in PROC LOGISTIC and PROC GENMOD is developed, while staying completely within the maximum-likelihood methodology. Only standard elements of the output are used, such as the likelihood, the Akaike information criterion, and the Schwarz information criterion, etc., or some equivalent R 2 measures discussed in the above mentioned Shtatland, Moore & Barton (2000). The proposed approach uses some averaging and improves the model selection process by taking model uncertainty into account. The average of a (usually small) number of 'good' models is often better than any one model alone. The improvement is seen in terms of the quality of predictions, more realistic confidence intervals, etc. Applications to some medical studies are discussed.