Danube prob

From Prosit
Jump to navigationJump to search

Vlc.png This page contains development contents which may not yet represent final PROSIT Content.


New probabilities lead to new model versions

The Danube A model is using a set of probabilities p1-p13 which together with the model structure build the "model logic". In both cases changes to the structure or to the probabilites lead to new versions of the disease model.

Be aware that such new versions need an official labelling as described in the Versioning Section of this forum.

Evidence Grading with SIGN Checklists

Shortly we will provide the results of our literature grading using the SIGN checklists.

Background

We prefer to grade the level of evidence of literature by means of the Scottish Intercollegiate Guidelines Network (SIGN). We could as well have chosen guidelines of german institutions but there occurred several problems when trying to find fitting guidelines with DIMDI or gmds. Within a development thread modelers may agree on different grading schemes such as the MERGE or the GRADE scheme or others. Within one thread one method must be applied.

Goals

By grading the levels of evidence of the literature used in our models, we want to publish the way we work and give the chance to measure the quality of our models. One important argument is that we want to make our models as transparent as possible. We achieve this by publication of the literature we used to build these models. Anyone can read the literature and reproduce the way we build our models. Another argument is that by grading the levels of evidence of literature we unfold that the the evidence of literature used is high and appropriate for our models. We also want to unfold the results of grading the levels of evidence of literature that anyone can check the evidence himself or herself. Therefore all sources are open to anyone. By using the checklists published by SIGN we give the chance to do so, because these checklists can be used for free.

Methods

SIGN = Scottish Intercollegiate Guidelines Network:

What is SIGN?

The Scottish Intercollegiate Guidelines Network (SIGN) was established to improve the scottish health care for patients by developing national clinical guidelines. These guidelines should be based on current knowledge.

What are guidelines?

Guidelines are documents developed by a group of people that explain how to help patients in a cheaper and more efficient way. This way has to be verified by research.

How to develop guidelines?

SIGN guidelines have to be based on systematic review of acknowledgments and critical evaluation of current scientific literature. By doing so SIGN wants to reduce possible sources of error.

Publication

All guidelines are available from the website of SIGN.

How to grade the level of evidence of literature?

First of all you need to find out which checklist to use. There are checklists for systematic reviews and meta-analyses, randomized controlled trials, cohort studies, case-control studies, diagnostic studies and economic studies. After that the fitting checklist has to be filled in. To get the grade given by the level of evidence of literature one has to look up the table of grades which can be found in.


Finding used literature in PubMed:

Before we could search for the literature we used in DanubeA we had to extract it from our model and list it in a database. After that all titles were entered one after another into the search engine on PubMed. Normally there were more than one results, so we had to do an manual search on these results.

Vancouver Standard:

All of our literature had to be written down in a standard style so one could easily find and check on them. Vancouver Standard is a standard for formatting information such as author, title, kind of literature, PubMed ID and so on. In this way we could organize all of our literature in a database and order it standardized.

Literature used in the Danube A model

In the following part all references used in DanubeA are listed in Vancouver-Style und linked with the article's abstract in PubMed.

Articles in Journals:

  • Adler AI, Stevens RJ, Manley SE, Bilous RW, Cull CA, Holman RR; UKPDS

Group; Development and progression of nephropathy in type 2 diabetes: The United Kingdom Prospective Diabetes Study (UKPDS 64); 2002; Kidney International; p.225-232. 1

  • Krolewski AS, Laffel LM, Krolewski M, Quinn M, Warram JH. Glycosylated hemoglobin and the risk of microalbuminuria in patients with insulin-dependent diabetes mellitus. N Engl J Med. 1995 May 11;332(19):1251-5.2
  • Forsblom CM, Groop PH, Ekstrand A, Tötterman KJ, Sane T, Saloranta C, Groop L. Predictors of progression from normoalbuminuria to microalbuminuria in NIDDM. Diabetes Care. 1998 Nov;21(11):1932-8.3
  • Jermendy G, Ruggenenti P. Preventing microalbuminuria in patients with type 2 diabetes. Diabetes Metab Res Rev. 2007 Feb;23(2):100-10.4
  • Newman DJ, Mattock MB, Dawnay AB, Kerry S, McGuire A, Yaqoob M, Hitman GA, Hawke C. Systematic review on urine albumin testing for early detection of diabetic complications. Health Technol Assess. 2005 Aug;9(30):iii-vi, xiii-163.5
  • Gaede P, Tarnow L, Vedel P, Parving HH, Pedersen O. Remission to normoalbuminuria during multifactorial treatment preserves kidney function in patients with type 2 diabetes and microalbuminuria. Nephrol Dial Transplant. 2004 Nov;19(11):2784-8. Epub 2004 Aug 24.6
  • Strippoli GF, Bonifati C, Craig M, Navaneethan SD, Craig JC. Angiotensin converting enzyme inhibitors and angiotensin II receptor antagonists for preventing the progression of diabetic kidney disease. Cochrane Database Syst Rev. 2006 Oct 18;(4):CD006257.7
  • Deferrari G, Repetto M, Calvi C, Ciabattoni M, Rossi C, Robaudo C. Diabetic nephropathy: from micro- to macroalbuminuria. Nephrol Dial Transplant. 1998;13 Suppl 8:11-5.8
  • Hovind P, Rossing P, Tarnow L, Smidt UM, Parving HH. Remission and regression in the nephropathy of type 1 diabetes when blood pressure is controlled aggressively. Kidney Int. 2001 Jul;60(1):277-83.9
  • Valmadrid CT, Klein R, Moss SE, Klein BE. The risk of cardiovascular disease mortality associated with microalbuminuria and gross proteinuria in persons with older-onset diabetes mellitus. Arch Intern Med. 2000 Apr 24;160(8):1093-100.10
  • Van Dijk PC, Jager KJ, Stengel B, Grönhagen-Riska C, Feest TG, Briggs JD. Renal replacement therapy for diabetic end-stage renal disease: data from 10 registries in Europe (1991-2000). Kidney Int. 2005 Apr;67(4):1489-99.11
  • Roper NA, Bilous RW, Kelly WF, Unwin NC, Connolly VM; South Tees Diabetes Mortality Study. Cause-specific mortality in a population with diabetes: South Tees Diabetes Mortality Study. Diabetes Care. 2002 Jan;25(1):43-8.12
  • Ray JA, Valentine WJ, Secnik K, Oglesby AK, Cordony A, Gordois A, Davey P, Palmer AJ. Review of the cost of diabetes complications in Australia, Canada, France, Germany, Italy and Spain. Curr Med Res Opin. 2005 Oct;21(10):1617-29.13

Articles on the Internet:

  • Frei U, Schober-Halstenberg H-J. Nierenersatztherapie in Deutschland. QUASI Niere gGmbH, Berlin. 2005 [cited 08.10.2007]. Available from: 14

Definitions

Grading

'Grading' means giving objects an order and a value. Here 'grading' means to value the level of evidence of literature. All possible levels of evidence are listed in SIGN50 section6 published by SIGN. These possible levels of evidence are:


table 2: levels of evidence according to SIGN
grade description
1++ High quality meta analyses, systematic reviews of RCTs, or RCTs with a very low

risk of bias

1+ Well conducted meta analyses, systematic reviews of RCTs, or RCTs with a low

risk of bias

1- Meta analyses, systematic reviews of RCTs, or RCTs with a high risk of bias
2++ High quality systematic reviews of case-control or cohort or studies
2+ Well conducted systematic reviews of case-control or cohort or studies
2- systematic reviews of case-control or cohort or studies
3 Non-analytic studies, e.g. case reports, case series
4 Expert opinion

Results

Thirteen out of the fourteen articles listed above were graded using the SIGN checklists. Only [14] was not graded with the SIGN checklists. The results will be described in the following, starting with those to each single article, and a final result in the end. The filled in checklists can be found below.

Development and progression of nephropathy in type 2 diabetes: The United Kingdom Prospective Diabetes Study (UKPDS 64) [1]

For this study the checklist for randomised controlled trials had to be used. It had an appropriate and clearly focused question that was answered by assigning patients to treatment groups by their status of diabetic nephropathy. The outcome measures were the disease status after about fifteen years of follow-up and ten years Markov model prediction. The measured states were no nephropathy, microalbuminuria, marcoalbuminuria, elevated plasma creatinin or renal replacement therapy and death form any of these states. Considering all these criteria it has to be said that the study was altered in order to minimize the bias to a low level. One can conclude from the above mentioned that the level of evidence of this article is - considering that it is a randomised controlled trial with a low risk of bias - 1+.

Diabetic nephropathy: from micro- to macroalbuminuria [8]

For this study the checklist for systematic reviews and meta-analyses was used. It addresses an appropriate and clearly focused question by reviewing randomised controlled trials, clinical controlled trials and cohort studies. For our key question it helped by giving a rough estimate (6-8%) for progression from microalbuminuria to macroalbuminuria. As the methodology used in this study was poorly described, the quality of the studies that were reviewed was not addressed at all and there were not many similarities in these reviewed studies to make combining them reasonable the bias had to be graded as high. After comparing with the table, listing all the levels of evidence in combination with the kind of study and the level of bias, the result of this study had to be set as 2-.

Predictors of progression from normoalbuminuria to microalbuminuria in NIDDM [3]

Forsblom et al.'s study population was a cohort so the checklist for cohort studies had to be used. The addressed question was to find out about predictors of progression from normoalbuminuria to microalbuminuria in NIDDM. This question was clearly focused and appropriate. It was ensured that the two groups that had been selected from source populations were comparable in all respects. In the article there was clearly noted how many people of the ones that were asked took part in each of the groups being studied. It can be said that the selection of patients for this study was done well. All outcomes were clearly defined and Forsblom et al. recognized that where blinding was not possible the knowledge of the exposure status could have influenced the assessment of outcome of the study. But none the less the measure of outcome is reliable because it can be compared to other sources that the method of outcome assessment is valid and reliable. Keeping all this in mind the risk of bias of this study is very low. The table used to look up for the levels of evidence says for a cohort study with a very low risk of bias that the level is 2++.

Remission to normoalbuminuria during multifactorial treatment preserves kidney function in patients with type 2 diabetes and microalbuminuria [6]

Gaede et al.'s study is about the remission to normoalbuminuria during multifactorial treatment. The study population was in a randomised controlled trial. The question Gaede et al. addressed was appropriate and clearly focused and the patients were assigned to the treatment groups randomly. An adequate concealment method was not used and if patients and investigators were kept blind in this study is never mentioned explicitly. Nothing is said about the rate of similarity of the treatment groups before the start of the trial. The conclusion of all these criteria is that there is a high risk of bias that might affect the study results giving a better result as really is. Regarding all these points the level of evidence of Gaede et al. is graded 1-.

Remission and regression in the nephropathy of type 1 diabetes when blood pressure is controlled aggressively [9]

Hovind et al. wanted to know if there are remission and regression in the nephropathy of type 1 diabetes when blood pressure is controlled aggressively by following a cohort of 301 patients. Their study addressed an appropriate and clearly focused question with the two groups being studied were selected from source populations that are comparable in nearly all factors. 271 patients received antihypertensive treatment and 30 patients received normotensive. Hence it can be said that the study clearly says how many of the patients that were asked to take part did so, in each of the groups being studied. Although there were 80 drop-outs there comparison was made between full participants and those lost to follow up, by exposure status. The outcomes are clearly defined but assessment of outcome is not made blind to exposure status and this was not even recognized as an influence on the assessment of outcome. Because evidence from other sources was used to demonstrate the validity and reliability of outcome assessment the risk of bias is not high or very low, but still low. So the level of evidence of Hovind et al. is 2+.

Glycosylated hemoglobin and the risk of microalbuminuria in patients with insulin-dependent diabetes mellitus [2]

Krolewski et al. tried to find a connection between glycosylated hemoglobin and the risk of microalbuminuria in patients with type 1 diabetes by examining a cohort of 1795 patients. This study like all the others above addressed an appropriate and clearly focused question but here the two groups that were studied could not be compared very well. It is exlicitly stated how many people took part in each of the groups being studied. Also the likelihood that some suitable patients might have had the outcome at the time of registration for the study is assessed and taken into account in the analysis. Krolewski et al. did not make any difference between full participants and those who dropped out during the follow-up. The assessment part was done pretty good all over the study and so the level of evidence can - regarding the low risk of bias - be set 2+.

South Tees Diabetes Mortality Study. Cause-specific mortality in a population with diabetes: South Tees Diabetes Mortality Study [12]

Roper et al. analysed the cause-specific mortality in a diabetes population. Their study addressed this appropriate and clearly focused question. The groups they studied were selected from source populations that are comparable very good. It is told how many patients took part in each of the study groups. Roper et al. took into account in the analysis and assessed that it is possible that some eligible patients might have had the outcome at the time of the enrollment. Only 0.45% of the patient could not be followed-up the whole time and comparison is made between full participants and those 0.45% by exposure status. Although the outcomes were clearly defined there was no blinding and it was not even some recognition that the knowledge of the exposure status could have an influence in the assessment of outcome. On the other hand Roper et al. used evidence from other sources for demonstration of the validity and reliability of the method of outcome assessment. Regarding all these criteria the level of evidence is 2+.

Angiotensin converting enzyme inhibitors and angiotensin II receptor antagonists for preventing the progression of diabetic kidney disease [7]

Strippoli et al. took 43 randomised controlled trials to find out the differences between ACE inhibitors and placebo and ACE inhibitors and AIIRAs (angiotensin II receptor antagonists). The description of the used methodology was done very well. For identifying all the relevant studies Strippoli et al. searched the most important data bases (Medline, Embase, Cochrane Renal Group trial register, Cochrane central register of randomised controlled trials). While extracting the data they assessed and took the study quality into account. Combination could be done very well between the studies found because there were enough similarities. Considering all these criteria it has to be said that the risk of bias is very low in this systematic review. Hence the level of evidence is 1++.

The risk of cardiovascular disease mortality associated with microalbuminuria and gross proteinuria in persons with older-onset diabetes mellitus [10]

Valmadrid et al. conducted a cohort study of 840 patients with type 2 diabetes. The two groups being studied were selected from very good comparable source populations and the article explicitly says how many patients took part in the study. It is mentioned that possibly some patients who took part in the study might have had the outcome at the time of registration and this is considered in the analysis. There was also made comparison between full study participants and those who dropped out. With clearly defined outcomes and its assessment being made blind to exposure status one can say as well that the measure of assessment has been reliable. Valmadrid et al. as well used evidence from other sources to demonstrate that the method of outcome assessment is valid and reliable. That leads to the conclusion that the study has a very low risk of bias and level of evidence of 2++.

Renal replacement therapy for diabetic end-stage renal disease: data from 10 registries in Europe (1991-2000) [11]

Van Dijk et al. collected data of patients with type 1 or 2 diabetic nephropathy from 10 registries in Europe and got a cohort of 17113 patients (7469 patients with type 1 diabetes mellitus, 9404 patients with type 2 diabetes mellitus and 240 patients with undefined type of diabetes mellitus). They included a description of the methodology being used with all the registries they searched and the criteria for including patients. Furthermore the study quality was not taken into account very much and if there was any possibility to compare the patients being selected from the registers was not mentioned explicitly. It can be said that there is a low risk of bias and considering that the level of evidence is 2+.

Preventing microalbuminuria in patients with type 2 diabetes [4]

Jermendy et al. did a double-blind, randomized, placebo-controlled trial with only one shortcoming: It is never mentioned if patients from more than one site took part in the study. They had 1204 patients in this study. Jermendy et al. wanted to find out whether angiotensin-converting-enzyme inhibitors and non-dihydropyridine calcium- channel blockers, alone or in combination, could prevent microalbuminuria in patients with hypertension, type 2 diabetes mellitus and normoalbuminuria. They found out that angiotensin-converting-enzyme inhibitors and non-dihydropyridine calcium- channel blockers in combination and angiotensin-converting-enzyme inhibitors alone decreased the incidence of microalbuminuria in a similar way. However non- dihydropyridine calcium-channel blockers had the same effect like placebo. Considering all that and the very low risk of bias in this study the level of evidence is 1++.

Systematic review on urine albumin testing for early detection of diabetic complications [5]

Newman et al. searched electronic databases up until January 2002 and completed their search by searching manually in major journals. The description of the used methodology is very detailed and well done. They even used special methods just like funnel plots, trim and fill methods and meta-regression to assess the bias and the sources of heterogeneity of the used publications. Newman et al. only included studies fulfilling the following criteria:

  • determining rate and predictors of development and factors involved in regression
  • carrying out economic evaluations of different screening strategies
  • investigating the effects of screening on patients
  • standardizing screening tests to enable use of common reference ranges
  • evaluating the effects of lipid-lowering therapy
  • using to modulate antihypertensive therapy

Hence it can be said that there are enough similarities to make combining the selected studies reasonable. From all this it follows that the risk of bias is very low and this systematic review can be graded a level of evidence of 2++ because it was reviewing mostly cohort studies.

Review of the cost of diabetes complications in Australia, Canada, France, Germany, Italy and Spain [13]

Ray et al. reviewed the cost of diabetes complications in Australia, Canada, France, Germany, Italy and Spain. To do so they search for peer-reviewed journals that were listed in PubMed and other not named databases. The methodology they used is included and described very well. In researching literature they could have done a bit better by researching more than just PubMed. On the other hand Ray et al. admitted that the quality of the studies was not always good for their specific question but none the less used these studies. By often using non-diabetes-specific values the bias might effect the study results into negative direction. Not all parameters that were listed by Ray et al. had much similarity but most did so. From all this it can be said that there is only a low risk of bias and hence the level of evidence is 1+.

Summary of results

Of the 14 articles only 13 could be graded because there was no fitting checklist for the article 'Nierenersatztherapie in Deutschland'. There were 3 studies using randomised controlled trials (RCT); 1 study using systematic review of RCT, clinical controlled trials (CCT) and cohort studies; 6 studies using cohort studies; 2 studies using systematic reviews of RCT; and 1 study using a systematic review of cohort. In table 3 the results of grading the 13 articles are listed.


table 3: overview of results
key question no level of evidence kind of study
1 1+ rct
2 2- systematic review of rct, cct, cohort
3 2++ cohort study
4 1- rct
5 2+ cohort study
6 2+ cohort study
7 2+ cohort study
8 1++ systematic review of rct
9 2++ cohort study
10 2+ cohort study
11 1++ rct
12 2++ systematic review of cohort
13 1+ systematic review of rct


In the column 'Key Question No' the internal number of the study is listed. This number is the same as in the field 'Key question no.' of the associated checklist. In the column 'Level of Evidence' the grades of the single articles are listed just as they were graded using the checklists. Those grades are as well listed above with the result of every specific article. In the last column, named 'kind of study', can be read which study graded has been of which kind. The levels of evidence of the literature used in DanubeA were all between 1++ and 2-.

levels of evidence

As can be seen in figure 1, there were two articles rated 1++, two were rated 1+, one got an 1-, three were rated 2++, four were rated 2+ and only one got a 2-. Of all literature that has been graded evidence of literature in DanubeA there are two studies, a systematic review of RCT [7] and a RCT [4], that are rated 1++ and are directly applicable to a population consisting of diabetes mellitus patients. Furthermore there are two studies, a RCT [1] and a systematic review of RCT [13], that are rated 1+ and are directly applicable to a population consisting of diabetes mellitus patients. There is as well given an overall consistency of results. Taking all that into account and including table 4 the conclusion is that the grade of recommendation in DanubeA is A.

Bschreiweis 17:15, 9 July 2009 (CEST)