Search

The skepticism expressed in the article with regard to sulfonylureas is unjustified. In a recent Cochrane meta-analysis (1), second generation sulfonylureas had fewer cardiovascular side effects than metformin; mortality was the same. By comparison, the cited retrospective data provide low-level evidence. The article does not contain any concrete data on the adverse effects of sulfonylureas. Weight gain as a side effect of these drugs is clinically not relevant: 1.3 kg in 2 years (2) and 1.7 kg in 10 years (UKPDS). Hypoglycemia caused by sulfonylureas is rare: about 0.5/patient/year, for example, for glimepiride+metformin (2). These cases of hypoglycemia are mostly mild and occur in the first three months after the start of the treatment, when HbA1c values in the sulfonylurea groups drop more notably than in the comparator groups. The authors write that treatment with DPP-4 inhibitors does not carry a risk of hypoglycemia. In combination with insulin or insulinotropic substances, however, the risk of hypoglycemia is increased by DPP-4 inhibitors, and if required the treatment has to be adjusted. The evidence for the explained lifestyle interventions is low. An example: the lifestyle intervention in the Look AHEAD Study did not prevent cardiovascular end points. With regard to GLP-1 analogues, no benefit has been found with regard to clinical end points. There have so far not been sufficiently large studies that really prove or exclude the risk of pancreatic cancer in patients being treated with GLP-1 analogues. The increase in spontaneous reports of pancreatic cancer cases under GLP-1 treatment remain a cause for concern. The indication for this class of substances is altogether extremely rare. The benefits of intensified insulin treatment in patients with type 2 diabetes that the authors propose are elusive. A randomized comparison of the different forms of insulin therapy did not show any significant differences; quality of life is rather reduced under intensive treatment (3).

In their CME article on generalized anxiety disorder, the authors claim that 45% of such disorders are not recognized in primary care. Leaving aside the fact that the study cited in support of this statement is 20 years old, and that the diagnosis and treatment of anxiety disorders have changed over these two decades—from a general practitioner’s perspective, a detection rate of 55% for anxiety disorders, which often present somatically, seems a good result in primary care. The recommended drugs of first choice are SSRIs, SNRIs, and pregabalin. Among the SSRIs, escitalopram was favored. There are no scientifically proven differences in effectiveness between citalopram and the notably more expensive escitalopram as a levorotatory isomer. The authors did not discuss this in their article. However, the fact that the potential for addiction, which has been proven for pregabalin (1, 2), was not mentioned at all is almost criminal. Both substances are among those whose prescription has been viewed as uneconomical, according to publications by the Associations of Statutory Health Insurance Physicians. Their use could be justified in case of medical benefits, but not where those are lacking. Two years ago, the BMJ published a systematic review on the topic of generalized anxiety disorders (3), which was methodologically far more valuable than the results of a selective literature search presented in the CME article. The results: fluoxetine was the preferred drug in terms of effectiveness, sertraline in terms of tolerability. The authors did not mention this meta-analysis. In sum, the published article does not merit the sign of approval of a CME article; rather, it carries a sense of one-sided presentation. It is most welcome that authors publishing in Deutsches Arzteblatt have to declare their competing interests. The comprehensive list reported by the authors of this article gives readers a clear idea of what might be the cause of the authors’ bias.

A CME article published in Deutsches Arzteblatt gives rise to the expectation that it reflects confirmed results. The article by Fuhrer et al. clearly does not meet this expectation. The authors postulate that in every case of a thyroid nodule, scintigraphy should be undertaken at least once. The sensitivity and specificity of this investigation are poor with regard to malignancies (2, 3). It is easy to form an impression of unjustifiably excessive diagnostic testing (seeing that the material that is mass-handled is radioactive). The recommendation to measure calcitonin in every patient with nodular goiter is controversial, to say the least. This recommendation may be appropriate for specialist endocrinology outpatient clinics. For general practice, with its low incidence of thyroid malignancies, it is definitely excessive. The recommendation to measure thyroid antibodies is not sufficiently supported—especially with regard to the relevance of the result for subsequent clinical action. The authors finally recommend a combination of iodine and L-thyroxine. The LISA study cited by the authors (4) investigated iodine and L-thyroxin in isolation as well as in combination. The treated nodules were 1.47–1.96 mL in size. The relative reduction in size by 17.3% therefore corresponded to an absolute reduction of 0.25–0.33 mL—a completely irrelevant order of magnitude. The thyroid volume of 18.2–18.8mL decreased by 1.5 mL, which is similarly irrelevant. Clinically relevant end points, such as frequency of thyroidectomies or development of thyrotoxicosis, were not studied. The recommendation given in the article is not supported by this study. To publish a CME article full of unconfirmed, not evidence-based, recommendations for maximum diagnostic testing and treatment as a CME article in Deutsches Arzteblatt suggests, in my opinion unjustifiably, that the authors are describing the current gold standard.

Hartmann et al postulate that metformin should be substituted by sitagliptin in patients with an eGFR below 60 mL/min. This recommendation is surprising—on the one hand, compensated renal failure is increasingly being challenged as a contraindication to metformin (1). In the United Kingdom, only a creatinine concentration of >1.5 mg/dL is regarded as a contraindication. A recently published Cochrane review (2) did not find any indication that use of metformin leads to lactic acidosis more often than other antidiabetes drugs. On the other hand the recommendation is surprising because sitagliptin is a substance about which no more is known than the fact that it lowers blood glucose concentrations. In the meantime, many reports have been received about the potential of sitagliptin to cause harm (3). A short monograph from the Drug Commission of the German Medical Association does not recommend its use in moderate or severe renal failure because not enough studies exist. The UKPDS 34 showed that near-identical lowering of HbA1c benefits patients to extremely varying degrees, depending on whether metformin or other substances are used. Just now, the European Medicines Agency has ordered to take rosiglitazone off the market—another sign for the fact that lowering HbA1c by means of medication is a very weak indicator of its use. It is not clear why the article recommends sitagliptin, a substance about which we do not know whether its use affects patient-relevant end points positively or negatively. Why did the authors not vote in favor of insulin, just in case that metformin should really not be used in more severely restricted renal function?

The authors retrospectively studied adults who underwent tonsillectomies after recurrent tonsillitis. They used quality of life scores and measured number of days off work because of illness and the number of visits to doctors. It is surprising that they did not include the not insubstantial rate of perioperative and postoperative complications in the results of their study and the discussion section. The number of postoperative hemorrhages is 1.4% (1) to 3.3% (2). The fact that a simple PubMed search using the search terms “tonsillectomy” and “perioperative or postoperative complications” shows up 24 meta-analyses. To me, this confirms that the procedure is not trivial. Two years ago, Deutsches Arzteblatt published a CME article on the topic of tonsillectomy in children (3). The authors postulated rightly that the indication for this intervention should be very narrowly defined. The article by Senska et al is mostly of no value in this respect because it is subject to bias, especially in terms of the included patients: someone who has decided to have a tonsillectomy is bound to be convinced by this procedure and will give positive scores in order not to have to have to live with the feeling that they have risked health and life unnecessarily.

The authors recommend—and reinforce this recommendation with a question in the CME test—an optimal value for glycated haemoglobin (HbA1c) of below 7%, in order to counteract the development and progression of diabetic retinopathy. In support of this they cite the DCCT study (1), but this study was conducted in patients with type 1 diabetes. Whether the results translate to patients with type 2 diabetes is not only questionable but has been disproved by the results of the large studies of type 2 diabetes. The UKPDS 33 (2) is incorrectly cited: lowering HbA1c from 7.9% to 7.0% does not reduce the need for laser coagulation of the retina by one third but by a rather more modest amount—namely, from 11 per 1000 patient years to 7 per 1000 patient years. Doctors and patients will have to reach a joint decision about whether the price of achieving this objective is not rather too high when considering the additional rate of severe hypoglycemias (which require help from third parties) of 7% for glibenclamide and 11% for insulin per 10 years. HbA1c was lowered to less than 7% in only in 3 larger studies (3, 4, 5). The retinopathy rate did not fall in any of these three studies. Why the authors recommend a potentially dangerous treatment (hypoglycemias, excess mortality as in ACCORD) remains a mystery.

In his CME article on the treatment of sudden idiopathic sensorineural hearing loss, Professor Suckfull postulates several allegedly necessary diagnostic measures ranging from ear microscopy to tympanometry to acoustic evoked response audiometry. However, he does not deliver any proof of the benefits of these recommendations—decided on the basis of a consensus. He rightly points out that no confirmed proofs exist for the treatment of sudden hearing loss. In this context, the “selective literature search” takes its revenge, so to speak: the studies by Desloovere (1, 2), which—in addition to the cited one by Klemm et al—disprove the benefits of treatment with plasma expanders, are not mentioned. It is entirely unacceptable to extract subgroups from studies with negative evidence, with the intention of deducing positive recommendations. The recommendation to offer HAES infusions—which are known for their potential to cause renal damage (3) and lifelong persistent pruritus (4, 5)—as individual health services is completely misplaced in a CME article.

Pfitzenmaier and Altwein in their very readable article explain the advantage of hormone therapy in locally advanced prostate cancer with an unfavourable Gleason score. Further, they show in a table the extent to which a PSA increase may give an indication of a poorer prognosis. For the question that is crucial when recommending routine PSA measurements be used in tumor aftercare-namely, which treatment option should be chosen when PSA rises-the authors take recourse to the worst possible evidence-expert opinion ("EAU guideline 2008"). They do not provide a real proof for why after primary treatment for prostate cancer PSA should be measured and which order of benefit therapy would yield that is administered after a rise in PSA.

The literature cited by Professor Diener needs to be looked at in greater detail. The PROGRESS study – 14% of patients were a priori not evaluated because they did not tolerate the ACE inhibitor perindopril that was used – was based on a definition of hypertension as blood pressure (BP) > 160/90 mm Hg. It is not clear to me how from this, a therapeutic recommendation for hypertension patients is deduced according to the definition of BP > 140/90 mm Hg. The MOSES study compared a sartan with nitrendipine – a substance that has not been shown to lower mortality in hypertension patients. In spite of this, the number of deaths in the study was numerically higher in patients taking eprosartan. Further, nitrendipine was given at too low a dosage in the study. Claiming class effects of the sartans that exceed mere blood pressure lowering is incorrect. Especially the cited LIFE study shows a different extent of blood pressure lowering, which is crucial for the stroke risk. Blood pressure normalized in 49% of patients receiving losartan, but in only 46% receiving atenolol. In the SPARCL study, 1.9% fewer patients had a stroke while being treated with the maximum dosage of atorvastatin. None of my stroke patients would be willing to take atorvastatin as an additional medication if the NNT was 52 over 5 years. The dipyramidole studies ESPS2 and ESPRIT, which the author mentions, are based on a comparison with underdosed acetylsalicylic acid. In the meta-analysis of the Antithrombotic Trialists’ Collaboration, it clearly says that ASA dosages between 75 mg and 150 mg were investigated – the value of lower dosages remained unclear. I fear that these inconsistencies may be a pointer to a conflict of interests in the reporting.