Observations by a statistical watchdog

During more than five decades, the author has kept a critical eye on how statistical methods are (mis-)used in medical research. Some areas are presented where serious statistical mistakes are prevalent. Two investigations with erroneous conclusions are described in detail. Situations where outside authorities have tried to mute medical researchers are also commented upon. The authors own efforts to improve the use of statistical methods and the current situation with easily accessible statistical program packages are described. Finally, the importance of continued 'statistical cleansing' is stressed.

Are degenerative spondylolisthesis and further slippage postoperatively really issues in spinal stenosis surgery?

Background and aims Opinions diverge concerning the prognostic importance of preoperative degenerative spondylolisthesis in patients with lumbar spinal stenosis, as well as the significance of further slippage post-operatively following decompression alone. However, a slip is only one among several factors related to the topic, e.g. duration and intensity of back and leg pain, pre-operative walking ability, number of levels operated and not least the experience of the surgeon. Our aim was to take all of the above-mentioned factors into consideration when analysing the patients' clinical outcome, reported as Change in back pain, Change in leg pain, Overall satisfaction and Change in walking ability, with special emphasis on the possible importance of pre- and/or post-operative degenerative spondylolisthesis. Methods We studied 200 consecutive patients, mean follow-up time 81 months (range 62-108). Before treatment and on the follow-up occasion all patients answered the SF-36 questionnaire and assessed their back and leg pain on a visual analogue scale (VAS). At follow-up the patients were asked about possible changes in back and leg pain (completely free, much better, somewhat better, unchanged, somewhat worse, much worse) and whether they were; satisfied with the outcome, in doubt or not satisfied. Before treatment and at follow-up the presence or not of degenerative spondylolisthesis was determined in the lateral view on a plain X-ray or MRI. By use of a microsurgical technique decompression was achieved in all patients by bilateral laminotomy not sparing the midline ligaments, irrespective of a degenerative spondylolisthesis or not. Eight surgeons with different surgical experience performed the operations. Four separate multivariate analyses were conducted, one for each clinical outcome. The Lasso method was used for variable selection and multiple imputation was applied to handle missing values. Results At follow-up 78.5% of the patients were completely satisfied with the outcome. Minimal clinical important difference (MCID) was achieved for 69% of the patients. Before surgery 28 patients were able to walk more than 1 km compared to 111 at follow-up. The reoperation rate at 6.8 years was 12% further decompressions and 2.5% fusions at the index level. Post-operative slippage was equally common in patients with and without a preoperative slip (around 30%). There were no notable differences in outcome in patients with and without a preoperative slip and no effect of further slippage at the index or another level post-operatively. Nor could the statistical analysis show any of the other covariates (age, gender, duration and intensity of back and leg pain, pre-operative walking ability or number of levels operated) to be of statistically significant importance for predicting the outcome. In the univariate statistical analysis differences were found between the patients of individual surgeons regarding satisfaction, pain improvement, and reoperation rates in favour of surgical experience, which were, however, not statistically significant in the multivariate analysis. Conclusions None of the covariates, including pre-operative spondylolisthesis and further slippage post-operatively, were statistically significant for predicting the clinical outcome. Implication Our results provide no evidence for adding fusion to the decompression.

Superior outcomes following cervical fusion vs. multimodal rehabilitation in a subgroup of randomized Whiplash-Associated-Disorders (WAD) patients indicating somatic pain origin-Comparison of outcome assessments made by four examiners from different disciplines

BACKGROUND AND AIMS

Whiplash-Associated Disorders (WAD) are characterized by great variability in long-term symptoms. Patients with central neck and movement-induced stabbing pain participated in a randomized study comparing cervical fusion and multimodal rehabilitation. As reported in our previous paper, more patients treated by cervical fusion than by rehabilitation experienced pain relief. Although patient reported outcome measures are a core component of outcome evaluation, independent examiner has been recommended. Because of the heterogeneity of WAD complaints the patients in our study were examined at baseline and follow-up by four experts representing neurology, orthopedics, psychology and physical medicine. The aim was to compare the professional assessments of change both regarding the possible impact of the different examiners' perspectives on individual patient's outcome, and also on the analysis of possible outcome differences between the treatment groups.

METHODS

WAD patients with long-term neck pain as the predominant symptom after a traffic accident were eligible. The neck pain origin should be in the midline and perceived as dull and aching, with sudden movement inducing midline stabbing pain. Of the 1,052 patients in contact with our team, 49 were eligible. The overall treatment effect was evaluated on a global outcome transitional scale. The criteria for the scale categories were defined by each expert's professional perspective on change in the whiplash complaints. Statistical methods that take account of the non-metric properties of ordered categorical data were used. Observed inter-expert disagreement was evaluated by the Svensson method that identifies and measures systematic group-related disagreement separately from disagreement caused by individual variation. Possible differences in the distributions of assessments on the expert-specific outcome scales between the treatment groups were analyzed by the Kruskal-Wallis test.

RESULTS

The per-protocol evaluation showed that a majority of the 18 patients who underwent fusion surgery were assessed as somewhat or much better, ranging from 67% to 78% depending on the expert. Corresponding proportions of improvement in the 17 patients treated by multimodal rehabilitation ranged from 29% to 53%. The statistical analyses confirmed better outcomes in the patients treated by fusion surgery, with p-values ranging from 0.003 to 0.04. The experts' assessments of intra-patient change disagreed more or less for all patients. The analyses of the paired comparisons confirmed that these disagreements could most probably be explained by the different profession-specific operational definitions of the outcome scales rather than by individual variations in data.

CONCLUSIONS

The multi-dimensional complexity of WAD-related complaints was comprehensively demonstrated by the inter-disciplinary disagreements in assessing intra-patient outcomes. The superiority of positive treatment effects in patients who underwent cervical fusion compared with multimodal rehabilitation was evident to all experts.

IMPLICATIONS

The results strengthen our previous opinion that neck pain in this subgroup of WAD patients has a somatic origin. More than one examiner is recommended for multi-dimensional outcome assessments.

Symptoms and signs possibly indicating segmental, discogenic pain. A fusion study with 18 years of follow-up

BACKGROUND

Only two out of the five existing randomized studies have reported better results from fusion surgery for chronic low back pain (CLBP) compared to conservative treatment. In these studies the back symptoms of the patients were described simply as "chronic low back pain". One possible reason for the modest results of surgery is the lack of a description of specified symptoms that might be related to a painful segment/disc, and patient selection may therefore be more or less a matter of chance. Previous prospective studies including facet joint injections and discography and eventually MRI have failed to identify patients with a painful segment/disc that will benefit from fusion surgery.

PURPOSE

Our purpose was to analyse in detail the pre-operative symptoms and signs presented by patients who showed substantial relief from their back pain following spinal fusion surgery with the aim of possibly finding a pain pattern indicating segmental, discogenic pain.

METHODS

We analysed 40 consecutive patients, mean age 41 years, with a history of disabling low back pain for a mean of 7.7 years. Before surgery the patients completed a detailed questionnaire concerning various aspects of their back pain, and findings at clinical examination were thoroughly noted. Monosegmental posterior lumbar interbody fusion without internal fixation was performed using microsurgical technique. Outcome was assessed at 1, 2 and 4 years after surgery and finally at 18 years, using self-reporting measures and assessment by an independent examiner. Assessment at 18 years applied the Balanced Inventory for Spinal Disorders Questionnaire and the Roland-Morris Disability Questionnaire.

RESULTS

According to the independent observer's assessment at two years 27 of the 40 patients were much improved. Analysis of the pre-operative depiction of the back symptoms of this group revealed a rather uniform pattern, the most important being: dominating back pain originating in the midline of the spine, with a dull, aching character and stabbing pain in the same area provoked by sudden movements. Most patients in this group also had diffuse pain radiation of various extension down one or both legs and often bladder dysfunction with frequency. At clinical examination, localized interspinal tenderness was observed within the spinal area in question and the patient's back pain was provoked by pressure in that area and by tapping a neighbouring spinous process. At 18 years after surgery 19 patients assessed themselves as much improved. At that time 5 of them had pension due to age, 7 early pension, one worked full time and six patients part time. Eleven patients were re-operated due to defect bony healing.

CONCLUSIONS

The results may suggest that the use of a detailed symptom analysis and clinical examination may make it possible to select a subgroup of patients within the CLBP group likely to have better outcome following fusion surgery.

IMPLICATIONS

The next step would be to execute prospective studies and if our findings concerning back pain details and signs among CLPB patients can be confirmed this can provide for more accurate selection of patients suitable for fusion surgery.

Clinical outcome following anterior arthrodesis in patients with presumed sacroiliac joint pain

BACKGROUND

It has been reported that in 13-32% of patients with chronic low back pain, the pain may originate in the sacroiliac (SI) joints. When treatment of these patients with analgesics and physiotherapy has failed, a surgical solution may be discussed. Results of such surgery are often based on small series, retrospective analyses or studies using a minimal invasive technique, frequently sponsored by manufacturers.

PURPOSE

To report the clinical outcome concerning pain, function and quality of life following anterior arthrodesis in patients presumed to have SI joint pain using validated questionnaires pre- and post-operatively. An additional aim was to describe the symptoms of the patients included and the preoperative investigations performed.

MATERIAL AND METHODS

Over a 6 year period we treated 55 patients, all women, with a mean age of 45 years (range 28-65) and a mean pelvic pain duration of 9.1 years (range 2-30). The pain started in connection with minor trauma in seven patients, pregnancy in 20 and unspecified in 28. All patients had undergone long periods of treatment including physiotherapy, manipulation, needling, pelvic belt, massage and chiropractic without success, and 15 had been operated for various spinal diagnoses without improvement. The patients underwent thorough neurological investigation, plain X-ray and MRI of the spine and plain X-ray of the pelvis. They were investigated by seven clinical tests aimed at indicating pain from the SI joints. In addition, all patients underwent a percutaneous mechanical provocation test and extra-articular local anaesthetic blocks against the posterior part of the SI joints. Before surgery all patients answered the generic Short-Form-36 (SF-36) questionnaire, the disease specific Balanced Inventory for Spinal Disorders (BIS) questionnaire and rated their level of pelvic and leg pain (VAS, 0-100). At follow-up at a mean of 2 years 49 patients completed the same questionnaires (89%).

RESULTS

At follow-up 26 patients reported a lower level of pelvic pain than before surgery, 16 the same level and six a higher level. Applying Svensson's method RP_{pelvic pain}=0.3976, with 95% CI (0.2211, 0.5740) revealed a statistically significant systematic improvement in pelvic pain. At follow-up 28 patients reported a higher quality of life and 26 reported sleeping better than pre-operatively. In most patients the character of the pelvic pain was dull and aching, often accompanied by a stabbing component in connection with sudden movements. Referred pain down the leg/s even to the feet and toes was noted by half of the patients and 29 experienced frequency of micturition.

CONCLUSIONS

It is apparent that in some patients the SI joints may cause long-term pain that can be treated by arthrodesis. We speculate that continued pain despite a healed arthrodesis may be due to persistent pain from adjacent ligaments. The next step should be a prospective randomized study comparing posterior fusion and ligament resection with non-surgical treatment.

IMPLICATIONS

Anterior arthrodesis can apparently relieve pain in some patients with presumed SI joint pain. The problem is how to identify these patients within the low back pain group.

One- versus Two-View Mammography Screening

In order to compare mammography screening with one versus two views, a prospective population-based study was carried out. Mammograms from 12 636 women aged 40 to 54 years were examined first as one-view screening and later as two-view screening. The initial results showed that two-view screening resulted in 349 recalls, 72 surgical biopsies, and 32 breast cancers. One-view screening resulted in 542 recalls, 61 surgical biopsies, and 29 histologically proven breast cancers. The follow-up revealed 2 additional cancers detected by one-view screening. Thus, the use of two-view screening resulted in one more cancer case being detected. The difference in breast cancer detection was not significant.

Private and Non-Private Disc Herniation Patients: Do they Differ?

Objectives :

In the 2006 yearly report from the Swedish National Register for Lumbar Spine Surgery it was claimed that international studies show obvious differences between private and non-private patients with regard to results from back surgery. Therefore our aim was to reveal such possible differences by comparing the two categories of patients at a private clinic.

Material and Methods :

The material comprises 1184 patients operated on for lumbar disc herniation during the period of 1987 to 2007. Basic pre-operative data were obtained from the medical records and follow-up was performed by a questionnaire around 5 years post-operatively.

Results :

Small but statistically significant differences between private and non-private patients were seen pre-operatively regarding the proportions of a/ men and women in the samples, b/ those with physically demanding jobs, c/ those on sick leave and d/ those with lumbar pain. Over the years the admitted private patients had a decreasing mean duration of symptoms which was not seen in the non-private patients. No apparent differences (n.s.) were seen between the two categories of patients pre-operatively regarding age, presence and level of leg pain or the proportion who smoked. Post-operative improvement in leg and lumbar pain was very similar in private and non-private patients as was satisfaction with the results and the proportion of patients returning to work.

Conclusion :

Despite small pre-operative differences concerning some variables and a significant difference in symptom duration between private and non-private disc herniation patients, the final clinical results were very similar.

Contractile activity of human skeletal muscle cells prevents insulin resistance by inhibiting pro-inflammatory signalling pathways

AIMS/HYPOTHESIS

Obesity is closely associated with muscle insulin resistance and is a major risk factor for the pathogenesis of type 2 diabetes. Regular physical activity not only prevents obesity, but also considerably improves insulin sensitivity and skeletal muscle metabolism. We sought to establish and characterise an in vitro model of human skeletal muscle contraction, with a view to directly studying the signalling pathways and mechanisms that are involved in the beneficial effects of muscle activity.

METHODS

Contracting human skeletal muscle cell cultures were established by applying electrical pulse stimulation. To induce insulin resistance, skeletal muscle cells were incubated with human adipocyte-derived conditioned medium, monocyte chemotactic protein (MCP)-1 and chemerin.

RESULTS

Similarly to in exercising skeletal muscle in vivo, electrical pulse stimulation induced contractile activity in human skeletal muscle cells, combined with the formation of sarcomeres, activation of AMP-activated protein kinase (AMPK) and increased IL-6 secretion. Insulin-stimulated glucose uptake was substantially elevated in contracting cells compared with control. The incubation of skeletal muscle cells with adipocyte-conditioned media, chemerin and MCP-1 significantly reduced the insulin-stimulated phosphorylation of Akt. This effect was abrogated by concomitant pulse stimulation of the cells. Additionally, pro-inflammatory signalling by adipocyte-derived factors was completely prevented by electrical pulse stimulation of the myotubes.

CONCLUSIONS/INTERPRETATION

We showed that the effects of electrical pulse stimulation on skeletal muscle cells were similar to the effect of exercise on skeletal muscle in vivo in terms of enhanced AMPK activation and IL-6 secretion. In our model, muscle contractile activity eliminates insulin resistance by blocking pro-inflammatory signalling pathways. This novel model therefore provides a unique tool for investigating the molecular mechanisms that mediate the beneficial effects of muscle contraction.

Thyroid-associated ophthalmopathy after treatment for Graves' hyperthyroidism with antithyroid drugs or iodine-131

CONTEXT

Previous randomized trials have suggested an association between radioiodine treatment for Graves' hyperthyroidism and thyroid-associated ophthalmopathy (TAO).

OBJECTIVES

The aim of the study was to compare the occurrence of worsening or development of TAO in patients who were treated with radioiodine or antithyroid drugs.

DESIGN

We conducted a randomized trial (TT 96) with a follow-up of 4 yr. PATIENTS, SETTING, AND INTERVENTION: Patients with a recent diagnosis of Graves' hyperthyroidism were randomized to treatment with iodine-131 (163 patients) or 18 months of medical treatment (150 patients). Early substitution with T(4) was given in both groups.

MAIN OUTCOME MEASURE

Worsening or development of TAO was significantly more common in the iodine-131 treatment group (63 patients; 38.7%) compared with the medical treatment group (32 patients; 21.3%) (P < 0.001).

RESULTS

The risk for de novo development of TAO was greater in patients treated with iodine-131 (53 patients) than with medical treatment (23 patients). However, worsening of TAO in the 41 patients who had ophthalmopathy already before the start of treatment was not more common in the radioiodine group (10 patients) than in the medical group (nine patients). Smoking was shown to influence the risk of worsening or development of TAO, and smokers treated with radioiodine had the overall highest risk for TAO. However, in the group of smokers, worsening or development of TAO was not significantly associated with the choice of treatment for hyperthyroidism.

CONCLUSIONS

Radioiodine treatment is a significant risk factor for development of TAO in Graves' hyperthyroidism. Smokers run the highest risk for worsening or development of TAO irrespective of treatment modality.

Perspective on Roseroot (Rhodiola rosea) studies

Rhodiola rosea (roseroot) extract is a commercially successful product, primarily used to reduce the effect of fatigue on physical and mental performance. In this perspective we present our investigation of the most recent studies performed on human subjects. With a focus on the statistical methods we found considerable shortcomings in all but one of the studies that claim significant improvement from roseroot extract. Overall, the study designs have not been well explained. Experimental results have been confused and appear to be in some cases incorrect. Some of the conclusions are based on selected results and contradicting data have not been adequately taken into account. We point to other studies of higher quality performed on roseroot, several that found no significant effect and one that did. We conclude that the currently available evidence for the claimed effects is insufficient and that the effect of Rhodiola rosea is in need of further investigation before therapeutic claims can be made.

Master blood pressure variation in office, by self, and in sleep recordings

Variation inherent to each session in any recording situation has to be mastered to obtain informative mean values and differences. Variation due to time trend and unsteady consecutive readings versus inter-arm variations in blood pressure (BP) in the operator-initiated, self-initiated, and sleep recordings were studied by oscillometry. They were small in operator-initiated recordings, increased in self-recordings, and were paramount in sleep. A passive supine patient and simultaneous recordings with cuffs placed in equal horizontal positions are essential conditions for the estimation of a real mean inter-arm difference. The relative variation of pulse pressure (PP), compared with other components, was largest within each situation. Most of the systolic BP (SBP)/diastolic BP (DBP) variation in sleep was due to unequal vertical cuff distances to the heart level. Only sleep PP was independent of the vertical position of cuffs. A plot of inter-arm differences can visualize systematic errors in sequences of sleep SBP/DBP. As is well known, every doubling of variation (SD) requires quadrupling the number of independent recordings in order to retain the same size of confidence interval (CI) for a mean value.

Snoring and systemic hypertension--an epidemiological study

The prevalence of snoring, overweight and systemic hypertension was estimated in a random sample of 4,064 Swedish men, aged 30-69 years. Of the responders, 15.5% complained of habitual snoring and 29.6% of occasional snoring. Although there was an age-related increase in the prevalence of snoring up to 60 years, habitual snoring was found to be mainly related to body mass index (BMI) (p less than 0.0001) but not to age. Altogether 299 men (9.3%) reported hypertension, 21.5% of whom were habitual snorers, compared with 14.9% of the non-hypertensives (p less than 0.01). The hypertensives were also more often overweight. Logistic multiple regression analyses showed that among subjects 40-49 years old there was an average increase in the predicted prevalence of hypertension from 6.5% among non-snorers to 10.5% of habitual snorers in the same weight group. For the whole study population, however, the increase was mainly dependent on age and BMI. Thus, the importance of habitual snoring for the prevalence of hypertension differs in various age groups.

Cannabinoid type 1 receptors in human skeletal muscle cells participate in the negative crosstalk between fat and muscle

AIMS/HYPOTHESIS

Cannabinoid type 1 receptor (CB1R) antagonists such as rimonabant (Rim) represent a novel approach to treat obesity and related metabolic disorders. Recent data suggest that endocannabinoids are also produced by human adipocytes. Here we studied the potential involvement of endocannabinoids in the negative crosstalk between fat and muscle.

METHODS

The protein level of CB1R in human skeletal muscle cells (SkM) during differentiation was analysed using western blotting. SkM were treated with adipocyte-conditioned medium (CM) or anandamide (AEA) in combination with the CB1R antagonists Rim or AM251, and insulin-stimulated Akt phosphorylation and glucose uptake were determined. Furthermore, signalling pathways of CB1R were investigated.

RESULTS

We revealed an increase of CB1R protein in SkM during differentiation. Twenty-four hour incubation of SkM with CM or AEA impaired insulin-stimulated Akt(Ser473) phosphorylation by 60% and up to 40%, respectively. Pretreatment of cells with Rim or AM251 reduced the effect of CM by about one-half, while the effect of AEA could be prevented completely. The reduction of insulin-stimulated glucose uptake by CM was completely prevented by Rim. Short-time incubation with AEA activated extracellular regulated kinase 1/2 and p38 mitogen-activated protein kinase, and impaired insulin-stimulated Akt(Ser473) phosphorylation, but had no effect on Akt(Thr308) and glycogen synthase kinase 3alpha/beta phosphorylation. In addition, enhanced IRS-1 (Ser307) phosphorylation was observed.

CONCLUSIONS/INTERPRETATION

Our results show that the CB1R system may play a role in the development of insulin resistance in human SkM. The results obtained with CM support the notion that adipocytes may secrete factors which are able to activate the CB1R. Furthermore, we identified two stress kinases in the signalling pathway of AEA and enhanced IRS-1(Ser307) phosphorylation, potentially underlying the development of insulin resistance.

How to control measurement variation in estimating the mean intraocular pressure reducing effect at a specific point in time

OBJECTIVE

To clarify how to control measurement variation in estimating the mean intraocular pressure reducing effect at a specific point in time.

METHODS

A set of 629 open-angle glaucoma or ocular hypertensive patients, eligible for treatment in both eyes, previously untreated, were reanalyzed at 3 or 6 months of intraocular pressure (IOP) reducing treatment. The trial design had parallel treatment groups; one group receives treatment A and the other group receives treatment B. The standard deviation (SD) of the IOP response variable was calculated based on a single eye per patient, mean of right and left eye (bilaterally identical treatment within each group) or difference between right and left eye (paired-eye difference within each group; one eye receives active treatment and the other eye receives placebo). The effect of using mean of replicated recordings on the SD of the patient group mean IOP was also studied. The results were generalized by applying a measurement variation model.

RESULTS

The "paired-eye difference within each group" design was clearly most effective; the SD of the IOP response variable was 2.2 mm Hg. Using mean of the right and left eye in a "design with bilaterally identical treatment in each group" was the next best solution; the SD of the IOP was 3.0 mm Hg. The SD of the IOP was 3.2 to 3.4 mm Hg when based on a single eye per patient. The minimization of the SD gave important reduction of number of patients required to enroll in a new trial. Using a single IOP recording per patient produced similar SD of the patient group mean IOP as using mean of replicated IOP recordings when measured with Goldmann tonometry.

CONCLUSION

Calculating the mean IOP by using measurements from both eyes of the patient gives important reduction of the variability.

Which endpoint: mean intraocular pressure or proportion of successful patients?

A reduction in intraocular pressure in clinical trials can be determined through the mean intraocular pressure, through the proportion of patients who have the intraocular pressure reduced to a specific target intraocular pressure, or both. Since both these possible endpoints measure the shift of 2 intraocular pressure distributions, we recommend that only one of them be tested. In general, testing the difference between mean-values is much more efficient than testing the difference between proportions. However, proportions of successful patients are valuable in showing the clinical implication of a reduction in mean intraocular pressure, particularly when evaluating a moderate pressure reduction. The effect of a small mean intraocular pressure reduction on the probability to reach the target intraocular pressure is pointed out, particularly the fact that it can be substantial even if the mean reduction is smaller than the measurement error.

[Statistics must be seen and scrutinized! Fundamental to the progress of medical science--yet neglected in medical journals]

Even the most respectable medical journals contain questionable conclusions, frequently based on erroneous statistical analyses. A common attitude among medical researchers seems to be that a deeper knowledge of statistical theory is not necessary--it might suffice to be acquainted with the statistical program packages. The most common statistical errors in medical manuscripts are related either to the structure of the data or to misunderstandings regarding the logic in some basic statistical procedures, especially significance analysis. Editors should designate professional statisticians to sort out those manuscripts which require further scrutiny from a statistical point of view. This can be organized on a consultative basis outside the framework of the traditional, "blinded" peer-review system. The assessments made by the medical journals have come to play a very specific role in the Swedish academic system with respect to doctoral dissertations. Further discussion of present procedures in this area would be highly desirable.

["P" as in Program Package. About torturing data and significant "fishing expeditions"]

For the study of prognostic factors, the medical researchers have access to a number of advanced statistical techniques available in standard program packages. A tradition has developed where survival or time to relapse is analysed on the basis of statistical materials with few patients but a large number of possible explanatory variables. In statistical "fishing expeditions" the p-values are used to sort out potentially useful prognostic variables. Since the number of observations is small, all relevant prognostic factors do not give statistical significance. Since a large number of variables are tested there is a considerable risk for spurious significances. It is not enough to show that a prognostic factor seems to be efficient in the patient group where it was first found. The result must be verified in further studies of independent groups of similar patients.

Reproducibility of tumor response evaluation in patients with high-grade malignant non-Hodgkin's lymphoma

Estimation of complete response (CR) and partial response (PR) in patients with non-Hodgkin's lymphoma (NHL) is associated with a number of potential sources of error. The aim of this study was to define the reproducibility of response evaluation performed by an independent review committee (RC). In a phase III study of patients >60 yr with aggressive NHL, 60 patients who were already evaluated by the independent review committee (RC 1) for response were randomized to three groups and re-evaluated (RC 2). The assessment was classified into one of seven mutually exclusive categories, where the important borderlines with regard to one of the major end-points of the study, the CR rate, were between CR, "CR uncertain" (CR(U)), and PR. A discrepancy between RC 1 and 2 was found in 8/60 patients (13.3%), influencing the CR/CR(U) status in four of these patients. Two CR and two PR patients were reclassified as CR(U). Thus, CR/CR(U) was changed in 4/60 (6.7%). The reports of the local investigators were compared with that of RC 1 in 254 patients. The CR/CR(U) status was affected in 41 of these patients (16.1%). It is concluded that an independent RC is a major prerequisite for a uniform response evaluation in phase III clinical trials. However, the good RC reproducibility does not motivate a second assessment. Moreover, in the phase III setting end-points other than the CR rate, such as time to treatment failure, cause specific and overall survival are preferred.

Risk in numbers--difficulties in the transformation of genetic knowledge from research to people--the case of hereditary cancer

Difficulties in communicating diagnostic information are exacerbated when the 'diagnosis' is a 'genetic risk' for cancer. The risk estimation demanded in this situation differs from other types of probability estimations. Observations of participants in 45 consultation sessions between physicians and potential patients were conducted at a clinic for hereditary cancer to explore the communication of genetic information. Thirty-three sessions were audiotaped, transcribed verbatim and analyzed, along with notes from the other sessions. A dominant theme was found to be numerical discussion of risk. Further analysis resulted in the description of problems for practitioners in the process of translating scientific knowledge into clinical management. Problems in providing information include unclear aims of the consultation sessions, mixing various types of background information and probabilities, recognizing how low the predictive values are, and difficulties in communicating the relationship between probability and conclusions. Problems in communicating information about the genetic risk for cancer are of at least two types: dilemmas arising from uncertainties implicit in the nature of the information itself and difficulties in communicating information in a manner that those concerned can interpret. These issues need clarification, so that information with far-reaching consequences can be made as clear and comprehensible as possible for those involved.

[Count on your beliefs. Bayes--not the P value--measures credence]

This article (the second of two) describes traditional (frequentistic) statistical analysis in the context of the confidence interval. Bayesian analysis is described in two settings. In the choice between diagnostic alternatives, the bayesian approach offers useful integration of new information with previous knowledge. With regard to the evaluation of clinical trial data, this article exemplifies bayesian analysis as contrasted with traditional analysis, and advantages of the former are cited. A brief literature review exposes the wide applicability of bayesian analysis in medical statistics.

[Count on your beliefs. Bayes' theorem in diagnosis]

Bayesian analysis of data finds increasing use in medical statistics, diagnostic evaluation and decision analysis. The central element in bayesian analysis is a set of mathematical rules for integrated evaluation of prior knowledge and new information. In many situations this approach has superior ability to deliver dependable updated knowledge and to provide an optimal probability basis for decisions. This article (the first of two) presents Bayes' theorem and its application in diagnostic work. It is explained how likelihood ratios of diagnostic tests interact with the outcome of such tests in the conversion of initial information (prior odds) to enhanced information (posterior odds).

Interchangeability of ambulatory and office blood pressure: limitations of reproducibility and agreement

A possible equivalence of office (Off) patient-recorded blood pressure (BP) and ambulatory (Amb) BP was evaluated. The criteria were the between-visits reproducibility (R) of oscillometrically measured supine office (Off), seated Off, and 24-h Amb BP and the agreement between Off and Amb data. Randomly ordered sessions were completed within 4 months in 59 untreated patients with Amb 24-h BP 136/87 (SD 14/10) mmHg and HR 72 (SD 9). R improves as SD of differences between sessions decreases with the number (n) of values included in the average so that SDD = a n(-b), where a and b are constants for each method, data sampling strategy, and group. R of a few steady Off data is better than R of a few Amb data. As n of the averaged values increases, R of Off and Amb methods converges and with n approximately 24 becomes identical. Only further increase in n of Amb data makes R of the Amb method superior than the Off method. The variably elevated initial Off BP distorts R and agreement. After approximately 6 readings, Off BP stabilizes at a lower "steady" level. "Steady" data averaged over visits are close to the Amb 24-h average from two sessions. Supine Off "steady" level is close to supine levels before sleep. The Off versus Amb method agreement improves by averaging "steady" BP from one up to four Off sessions, up to 7 "steady" Off readings over a session, and BP in both arms. At best the SBP agreement is within +/- 7.6 mmHg in 90% of cases with 24-h Amb S/D BP ranging 115-155/75-105 mmHg. The casually elevated BP can be evaluated only by the demonstrated between-visits reproducible 24-h Amb average or by the nearly equivalent average of multiple "steady" data from two to four Off visits.

Classification of thyroid size by palpation and ultrasonography in field surveys

BACKGROUND

Goitre surveys are used to assess the degree of iodine deficiency in a population. The change of goitre classification made by WHO in 1994 implied that a smaller thyroid size should be regarded as goitre. Furthermore, the acceptable goitre prevalence was lowered from 10% to 5%, and ultrasonography was recommended as a more precise method for diagnosis of goitre. We studied the effects of the change of palpation system, and compared the precision of the old and new systems with that of ultrasonographic examination.

METHODS

We studied 225 schoolchildren (aged 7-14 years) in a highland village in Tanzania. The size of the thyroid was assessed in duplicate by ultrasonography and by WHO's 1960 and 1994 palpation systems. The latter were done by three examiners. Variations within and between examination methods and examiners were assessed, and measurement errors by ultrasonography were assessed from duplicate examinations. The sensitivity and specificity of the two palpation systems were calculated, with diagnosis by ultrasonography as the gold standard. Apparent palpation prevalences were calculated at a "true" 5% prevalence.

FINDINGS

The lowered criterion for goitre resulted in an extra 20-33% of children being diagnosed as having goitre by palpation. The variation between repeat examinations was only slightly smaller by ultrasonography (kappa=0.63) than by experienced examiners (kappa=0.57-0.58). The variation between thyroid volume estimation by ultrasonography and the true volume was about 50% due to both measurement error and variation in the shape of thyroid lobes. The new goitre criterion decreased specificity from 76% to 29%, whereas sensitivity rose from 56% to 80%. In contrast, a suggested sharpening of the old criterion increased specificity to 90%.

INTERPRETATION

A return to the old (1960) palpation criterion for goitre: "lobes larger than the terminal phalanxes of thumbs" and to an accepted palpation goitre prevalence of 10% can allow affordable monitoring of thyroid size through palpation in field surveys.

Numerical modeling of the temperature distribution within a bonded paper web when undergoing microwave heating in a waveguide

In this paper, a numerical technique is used to study the temperature distribution within the heated bonded paper web placed in a waveguide. The parameters of the problem were selected so as to best locate the test sample that would generate the desired absorbed power. The method of false transient was utilized to accelerate the numerical convergence and the equations were discretized with central difference in both time and space. The resulting system of equations was solved using an Alternating Direction Implicit scheme. Good agreement was achieved between the numerical solutions and the experimental validations.

A 10-year follow-up of snoring in men

STUDY OBJECTIVES

Little is known about the natural development of snoring, and this survey was conducted to study the development of snoring in men over a 10-year period.

DESIGN

Population-based prospective survey.

SETTING

The Municipality of Uppsala, Sweden.

PARTICIPANTS AND MEASUREMENTS

In 1984, 3,201 randomly selected men aged 30 to 69 years answered a questionnaire on snoring and sleep disturbances. Of the 2,975 survivors in 1994, 2,668 (89.7%) answered a new questionnaire with identical questions to those used at baseline. Questions about smoking habits, alcohol, and physical activity were also added.

RESULTS

Habitual snoring was reported by 393 men (15.0%) in 1984 and by 529 (20.4%) 10 years later. In both 1984 and 1994, the prevalence of snoring increased until age 50 to 60 years and then decreased. Risk factors for being a habitual snorer at the follow-up were investigated using multiple logistic regression with adjustments for previous snoring status, age, body mass index (BMI), weight gain, smoking habits, and physical activity. In men aged 30 to 49 years at baseline, the predictors of habitual snoring at the follow-up, in addition to previous snoring status, were as follows: persistent smoking (adjusted odds ratio, 95% confidence interval) (1.4, 1.1 to 1.9), BMI 1984 (1.1, 1.02 to 1.1/kg/m2) and weight gain (1.1, 1.03 to 1.2/kg/m2). Among men aged 50 to 69 years, after adjustments for previous snoring status and age, weight gain was the only significant risk factor for developing habitual snoring (1.2, 1.05 to 1.4/kg/m2).

CONCLUSIONS

In men, the prevalence of snoring increases up to the age of 50 to 60 years and is then followed by a decrease. Weight gain is a risk factor for snoring in all age groups, while smoking is mainly associated with snoring in men <60 years of age.

Quality of life aspects and costs in treatment of Graves' hyperthyroidism with antithyroid drugs, surgery, or radioiodine: results from a prospective, randomized study

The patients' views and costs of three different forms of treatment for Graves' hyperthyroidism were investigated. The study comprises 174 patients with Graves' hyperthyroidism who were stratified into two age groups: 20 to 34 years and 35 to 55 years. The younger group was randomly assigned to treatment with antithyroid drug plus thyroxine for 18 months or subtotal thyroidectomy, and in the older group iodine-131 was added as a third alternative. The patients' views of their therapy were based on a questionnaire formulated to identify possible differences between the three treatment forms. The costs were assessed by analyzing the official hospital reimbursement system for both outpatient and inpatient costs for a period of 2 years from the day of randomization. The results show that no significant differences in opinion were found between the five treatment groups with regard to any of the questions. Furthermore, only 10% of the patients expressed slight and 3% major hesitation to recommend the treatment form received to a friend with similar disease. Twenty percent of the patients with endocrine ophthalmopathy reported the eye problems to be much more troublesome and 14% somewhat more troublesome than the thyroid problems. The cost proportion between the medical and surgical treatment in the young group was 1:2.5 (1 = 1126 United States dollars [USD]) before and 1:1.3 (1 = 2284 USD) after inclusion of the relapse costs. The proportion between the medical, surgical, and iodine-131 treatment in the older group was 1:2.5:1.6 (1 = 1164 USD) before and 1:1.6:1.4 (1 = 1972 USD) after inclusion of the relapse costs.