Immortal time bias and survival in patients who self-monitor blood glucose in the Retrolective Study: self-monitoring of Blood Glucose and Outcome in Patients with Type 2 Diabetes (ROSSO).

AIMS/HYPOTHESIS: In the February 2006 issue of Diabetologia, the observational Retrolective Study: Self-monitoring of Blood Glucose and Outcome in Patients with Type 2 Diabetes (ROSSO) reported a 51% reduction in the risk of all-cause mortality in patients with type 2 diabetes who performed self-monitoring of blood glucose (SMBG). However, these impressive benefits conflict with results from observational studies and randomised controlled trials. We aimed to show that these findings are caused by a flawed design that introduced immortal time bias. METHODS: We illustrate the bias in the ROSSO study and demonstrate that it is large enough to completely explain the apparently protective effect of SMBG on all-cause mortality. RESULTS: In the ROSSO study, patients were classified as exposed to SMBG for their whole follow-up time if they performed self-monitoring for at least 1 year during the study period. Thus, the time between cohort entry and the date after 1 year self-monitoring was performed is unavoidably 'immortal' for patients with SMBG. Patients had to survive at least 1 year to be classified as exposed to this intervention and were artificially 'protected' from death. Based on published information, the total amount of misclassified immortal person-time in the SMBG group is at least 5,082 of 9,248 person-years at risk (55%). After re-classification of immortal person-time as unexposed, the unadjusted relative risk changed from 0.59 to 1.95. CONCLUSIONS/INTERPRETATION: The apparently protective effect of SMBG on all-cause mortality observed in the ROSSO study is completely explained by immortal time bias.

[Coding problems in German mortality statistics as illustrated by ischaemic heart disease]. / Kodierungsprobleme in der deutschen Todesursachenstatistik am Beispiel ischämischer Herzkrankheiten.

BACKGROUND: The German mortality statistics are an important data source in terms of research and health policy but might be influenced by different sources of error such as ICD-10 coding by regional authorities. The aim of this study was to identify state-specific coding problems using the example of ischaemic heart disease (IHD) and myocardial infarction (MI). METHODS: After obtaining age-standardised mortality rates (1998-2008), outlier analyses were used to identify German states with suspect changes in mortality from IHD or MI over time or suspect proportional mortality from an unspecific cause-of-death (ICD-10 R00-R99). Values outside the 3-fold interquartile range (IQR) from the IQR boundaries were considered as outliers. The impact of these discrepancies was studied by using an ecological correlation between state-specific poverty rates and mortality from IHD or MI as an example. RESULTS: During the study period, mortality rates dropped by 38.1% for IHD and by 38.7% for MI. Suspect mortality trends or proportional mortality rates were identified in Berlin, Bremen, Hamburg, Schleswig-Holstein and Saarland. The association between state-specific poverty rates and mortality from IHD or MI increased after exclusion of the suspect states. CONCLUSIONS: The ICD-10 coding by regional authorities seems to be an important source of error. A standardisation of coding procedures is urgently needed.

Drug-induced acute pancreatitis: results from the hospital-based Berlin case-control surveillance study of 102 cases.

BACKGROUND: Drug toxicity is a well-known cause of acute pancreatitis (AP). Although many drugs have been associated with AP, the magnitude of the risk of most of them remains largely unknown. AIM: To determine the pancreatotoxic risk of a wide range of drugs. METHODS: The hospital-based Berlin case-control surveillance study, including all 51 Berlin hospitals in a hospital network, ascertained 102 cases with idiopathic AP (IAP) and 750 controls between 2002 and 2011. Patients with IAP were thoroughly validated using anamnestic, clinical or laboratory data. Drug exposure was obtained in a face-to-face interview. Possible drug aetiology was assessed in individual patients through a standardised causality assessment applying the criteria of the World Health Organization. Drug risks were further quantified [odds ratios (OR) with 95% confidence intervals (CI)] in a case-control design with unconditional logistic regression analysis. RESULTS: The pancreatotoxic risk of several drugs, including azathioprine (OR 5.1; 95% CI 1.9-13.5), fenofibrate (OR 12.2; 95% CI 2.3-69.1), mesalazine (OR 3.3; 95% CI 1.1-9.5) or angiotensin-converting enzyme inhibitors, was corroborated by case-control analysis and causality assessment. Causality assessment suggested a pancreatotoxic potential, among others, for mercaptopurine or the seldom reported leflunomide, and alluded to a novel risk for tocilizumab. Case-control analysis showed an increased risk for two phytotherapeutics: harpagophytum and valerian radix. CONCLUSIONS: Our study quantified the pancreatotoxic risk of different drugs and phytotherapeutics. The findings corroborate previous results from the literature but also indicate risks for substances not previously reported, highlighting the need for further controlled studies on pancreatic toxicity.

[Pharmacoepidemiological research with large health databases]. / Pharmakoepidemiologische Forschung mit Routinedaten des Gesundheitswesens.

Over the years there has been an increase in the number of pharmacoepidemiological studies using secondary data from large health databases. Administrative health databases consist of data recorded within the health care system for reasons of billing. Physician-based databases use data derived from electronic medical records. In both types of databases, data are recorded prospectively and may include demographic information, lifestyle information, ambulatory consultations, drug prescriptions, ambulatory and in-hospital diagnoses, ambulatory services, laboratory values, hospitalizations and information on death. Health databases are valuable for research on drug utilization and drug effects, but they are also increasingly used for disease epidemiology studies. In recent years, most new drugs within the European Union have been approved with the requirement of active post-marketing surveillance for investigation of drug utilization or monitoring of drug safety. This implies an increasing need for valid data sources. Large health databases are important instruments for the detection of unknown drug risks and for the investigation of new safety signals derived from spontaneous reporting systems.

[Validation of secondary data. Strengths and limitations]. / Validierung von Sekundärdaten. Grenzen und Möglichkeiten.

Medical records databases (such as the General Practice Research Database-GPRD) and administrative databases (such as German Statutory Health Insurance (SHI) claims data) are useful sources for pharmacoepidemiology and health services research. However, these data are not primarily collected for research purposes. Validation studies are needed to examine their completeness and accuracy depending on the corresponding research question. This article reviews strategies for checks of internal consistency within the data from one SHI as well as between data from several SHIs and possibilities of internal data validation. Descriptive analyses of consistency can help to determine the integrity of data. The aim of internal validation is to separate uncertain from true cases based on information from secondary data alone or to reproduce known associations within the database. In addition external validation of secondary data is desirable using original prescriptions, medical records, hospital discharge letters and/or patient or physician interviews as a gold standard. A considerable number of external validation studies of diagnostic coding have been conducted within the GPRD. In contrast, such validation studies of German SHI claims data are mostly lacking and are urgently needed in the near future.

Thrombotic thrombocytopenic purpura/haemolytic uraemic syndrome associated with clopidogrel: report of two new cases.

Clopidogrel has been reported to be safe and effective in reducing vascular events. Nevertheless, there is growing evidence that clopidogrel may cause thrombotic thrombocytopenic purpura/haemolytic uraemic syndrome (TTP/HUS). This association has been debated, since in several cases alternative causes could not be excluded. Two new cases of TTP/HUS associated with clopidogrel are reported here. After discontinuation of clopidogrel and treatment with plasma exchange, both patients had a complete and sustained recovery from TTP/HUS. These cases corroborate previous observations that clopidogrel may indeed be a rare cause of TTP/HUS.

Amlodipine-induced immune thrombocytopenia.

BACKGROUND AND OBJECTIVES: Drug-induced thrombocytopenia is a serious, but rare, side-effect of treatment with a number of drugs. In this report, we investigate the suspicion that amlodipine, a calcium-channel blocker, was responsible for immune thrombocytopenia in a 79-year-old patient. PATIENT AND METHODS: Our patient experienced two attacks of thrombocytopenic purpura after 10 years of treatment with amlodipine. Antibodies to platelets were tested by standard methods. RESULTS: Initially, the platelet count increased owing to treatment with prednisolone and intravenous immunoglobulin G, but decreased shortly after discontinuation of this treatment. The patient's serum was found to contain amlodipine-dependent antibodies to platelets, and he recovered after stopping the drug. CONCLUSIONS: Amlodipine can induce immune thrombocytopenia, which may strongly resemble autoimmune thrombocytopenia.

Involvement of plasma atrial natriuretic peptide in protracted alcohol withdrawal.

OBJECTIVE: Atrial natriuretic peptide (ANP) has been shown to inhibit the effects of corticotrophin releasing hormone, corticotrophin and cortisol, and to influence affective and anxiety symptoms in man. We tested the hypothesis of whether ANP is associated with endocrine and psychopathological disturbances during acute alcohol withdrawal. METHOD: ANP and cortisol plasma concentrations were studied in alcoholics during in-patient detoxification and in healthy controls. Additionally, craving, depressive mood and anxiety were assessed. RESULTS: Although mean ANP levels increased significantly in alcoholics between days 1 and 14, they remained diminished compared to controls. Separating a subgroup of alcoholics with a decrease of ANP levels during withdrawal, these individuals revealed significantly elevated scores for mean and maximum craving and a trend to an elevated self-rated anxiety on day 14. CONCLUSION: We suggest that a dysregulation of ANP plasma levels during alcohol withdrawal may contribute to symptoms of protracted withdrawal such as craving and anxiety.

Long-term effects of pharmacotherapy on relapse prevention in alcohol dependence.

BACKGROUND: There is growing evidence that pharmacological treatment with two of the best validated anticraving drugs, acamprosate and naltrexone, is efficacious in promoting abstinence in recently detoxified alcohol-dependent subjects. OBJECTIVE: The stability of effects after termination of treatment remains to be answered, especially when combining both the drugs. METHOD: After detoxification, 160 alcohol-dependent subjects participated in a randomized, double-blind, placebo-controlled trial. Patients received naltrexone or acamprosate or a combination of naltrexone and acamprosate or placebo for 12 weeks. Patients were assessed weekly by interview, self-report, questionnaires and laboratory screening. Additionally, follow-up evaluation based on telephone interview of participants, general practitioners and relatives was conducted 12 weeks after terminating the medication. RESULTS: At week 12, the proportion of subjects relapsing to heavy drinking was significantly lower in the group with combined medication compared with both placebo and acamprosate (P < 0.05). No difference was detectable between acamprosate and naltrexone, both of which were superior to placebo (P < 0.05). Relapse rates were 28% (combined medication), 35% (naltrexone), 50% (acamprosate) and 75% (placebo). After follow-up (week 24), combined medication led to relapse rates significantly lower than placebo, but not lower than acamprosate. Again, both naltrexone and acamprosate were superior to placebo. Relapse rates were 80% (placebo), 54% (acamprosate), 53% (naltrexone) and 34% (combined medication). CONCLUSIONS: The results of this study highlight the stability of effects of pharmacotherapy on relapse prevention in alcohol dependence.