Development of a minimal set of prescribing quality indicators for diabetes management on a general practice level

pharmacoepidemiology and drug safety (2011)Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/pds.2248 Development of a minimal set of prescribing quality indicators fordiabetes management on a general practice level Liana Martirosyan1*, Flora M. Haaijer-Ruskamp1, Jozé Braspenning2 and Petra Denig1 1 Department of Clinical Pharmacology, University Medical Center Groningen, University of Groningen, The Netherlands2 Scientific Institute for Quality of Healthcare, Radboud University Nijmegen Medical Centre, The Netherlands Objective To identify the relevant prescribing quality domains of type 2 diabetes mellitus care as a basis for the selection of a minimal setof prescribing quality indicators from a set of previously validated indicators.
Methods We used the principal factor analysis to identify the underlying dimensions or domains of prescribing quality for 76 generalpractitioners participating to the Groningen Initiative to Analyse Type 2 Diabetes Treatment project in the Netherlands. From a set of 10prescribing quality indicators covering various aspects of cardiovascular and metabolic management, we selected a subset of indicators withthe highest loading within each identified domain. Next, we evaluated the effect of using this subset on the quintile ranking of practices ontheir prescribing quality scores.
Results We identified five prescribing quality domains in our data set: two assessing initiation of pharmacotherapy for different risk factorsin diabetic patients, two on stepwise intensification of treatment, and one on treatment of patients with cardiovascular disease. A compositescore comprising the indicators selected from each of the domains showed good agreement with the composite score comprising allindicators with 82% of general practitioners either not changing their position or shifting their ranking by only one quintile.
Conclusions We showed that a minimal set of prescribing quality indicators for type 2 diabetes mellitus care should not just focus on themanagement of different clinical risk factors but also reflect different steps of treatment intensification. The results of our study are relevant forstakeholders when selecting quality indicators to assess the quality of prescribing in diabetic patients. Copyright 2011 John Wiley & Sons, Ltd.
key words—quality health care indicator; type 2 diabetes mellitus; quality of healthcare assessment; drug prescribing Received 14 March 2011; Revised 3 August 2011; Accepted 5 August 2011 Stakeholders using quality information, such as health care providers, policy makers, and payers, have The demand for accountability in health care and the to deal with a large number of quality indicators be- need for improving the quality of provided care have cause of the growing number of different quality- resulted in the development of a large number of qual- reporting programs. The number of quality indicators ity indicators for an increasing number of diseases.1,2 included in national sets is varying from country to Quality measurement and reporting have the potential country. For example, the number of quality measures to improve the quality of care and reduce health care included in Healthcare Effectiveness Data and Infor- costs but can also cause administrative and financial mation Set 2010 set in the USA is about half of the burden of collecting and reporting quality information.
number of indicators included in Quality and Outcome To minimize this burden, it is important to seek strate- Framework in the UK.6,7 Although both sets are com- gies to reduce the number of quality indicators used.3–5 prehensive, there is lack of understanding on what the This article describes the process and results of selecting number of indicators in such sets should be. Besides a minimal set of prescribing quality indicators (PQIs) for such comprehensive programs, many sets of quality treatment of type 2 diabetes mellitus (T2DM).
indicators exist focusing on specific diseases, forexample, diabetes management.8–14 The availabilityof various quality indicators creates a possibility to *Correspondence to: L. Martirosyan, The Netherlands Institute for Health choose the most appropriate indicators for speci Services Research (NIVEL), PO Box 1568, 3500 BN Utrecht, The Netherlands.
user aims. However, this also introduces the challenge Copyright 2011 John Wiley & Sons, Ltd.
of selecting the indicators while maintaining the com- The inclusion criterion for our study was having eligi- ble patients for all tested indicators.
Diabetes is a chronic condition with increasing prev- alence in the world. Although lifestyle modification plays an important role in treatment of T2DM patients, All participating GPs used electronic health records most of them eventually require pharmacotherapy and prescribe electronically, which means that the data because of progressive nature of the disease.15,16 set includes full information regarding the prescribed Currently, to evaluate and improve the quality of drug medication. We collected data from the Groningen treatment in T2DM patients, a large number of PQIs Initiative to Analyse Type 2 diabetes Treatment data- base, which contains information from the electronic Combining measures to a composite score is one health records of all T2DM patients registered in the way to reduce the number of indicators included in participating practices.24 In addition, survey information quality assessment. Composite scores provide an is collected yearly regarding the practice characteristics, advantage of quick overview of the provided quality including practice size and supporting personnel. The of care in a certain area.18 However, they do not patient data set for our study included information on reduce the burden related to collecting and reporting the demographics, prescribed medication, comorbidities, much data on an individual indicator level.
physical examinations and laboratory measurements as Several approaches are available to make a selection of relevant prescribing indicators from a larger set.
One can start choosing indicators on the basis of stake- holders’ specific preferences and areas of interest.4,19It is possible to further narrow down the choice of Previously, a set of 14 indicators for assessing indicators on the basis of clinimetric characteristics, prescribing quality in T2DM was developed on the such as the grade of evidence supporting the indica- basis of several national and international diabetes tors, the concurrent and predictive validity, and the guidelines.14 This set of indicators was selected after availability of data,17,20 or discard all indicators that assessment in two panels of experts on face and con- do not show room for improvement.21 Another ap- tent validity. In short, the indicators cover adequate proach to systematically minimize the number of qual- and timely treatment of relevant cardiovascular, renal, ity indicators is the use of data reduction techniques, and metabolic risk factors as well as prescribing of such as factor analysis, allowing to uncover hidden metformin in overweight patients. The PQIs were calculated by dividing the number of eligible patients The aim of this study is to identify relevant prescrib- who were prescribed the recommended treatment by ing quality domains of T2DM management that can the total number of eligible patients as specified by serve a basis for selecting a minimal set of PQIs to the PQIs, and the percentages were obtained by multi- be applied on a general practice level.
plying the received ratio by 100. The operationaldefinitions of the PQIs are described elsewhere.14 Two indicators were discarded from this original setbecause of a lack of eligible patients at the general practice level, that is, one focusing on “prescription In the Netherlands, patients are registered with a single of statins in patients younger than 40 with a history general practitioner (GP) who has a gatekeeper role in of cardiovascular disease (CVD)” and one on coordinating their medical care. Most patients with “prescription of metformin in incident T2DM patients T2DM are managed in general practice. Many GPs who are overweight.” In addition, we modified one have a diabetes nurse or assistant who will conduct indicator focusing on “prescription of statins in all the routine three-monthly examinations of patients.
diabetic patients with increased cardiovascular risk” In our study region, in the north of the Netherlands, to “prescription of statins in patients with dyslipidemia” there is also a regional diabetes facility that offers sup- to reflect recommendations of the Dutch diabetes guide- port to GPs by providing thrice a month and yearly di- lines regarding prescription of statins for the study abetes follow-up examination of patients. In all cases, time.25 The validated set of the indicators included three the GP is responsible for the patients’ management PQIs focusing on the management of albuminuria with and for prescribing their medication. All GPs partici- a renin–angiotensin–aldosterone system inhibitor in pating in a regional program for monitoring diabetes mutually exclusive subpopulations of T2DM patients, care by the end of 2007 were eligible for inclusion.23 that is, patients without hypertension or with incident Copyright 2011 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety, (2011) quality indicators for diabetes management hypertension or with prevalent hypertension. Because Next, we selected the PQIs with the highest loading only 17 GPs had eligible patients for all three indica- within each factor to represent that specific domain of tors, we combined them to one indicator to increase prescribing quality. To evaluate the effect of selecting the number of eligible patients per practice for the this subset of PQIs on prescribing quality assessment at general practice level, we assessed changes in theranking of GPs using all or only this subset of indica- tors. For this, we calculated two composite scores foreach GP averaging scores of individual indicators Descriptive statistics are presented for practice and and ranked GPs on these scores. The first composite patient characteristics in the Table 1. We calculated score included all 10 initial PQIs, and the second one the scores of PQIs and their 95% confidence intervals included PQIs selected using the factor analysis. To (midP) using an individual GP as unit of analysis. An compare shifts in ranking, the rankings on both com- exploratory factor analysis was conducted to identify posite scores were divided in quintiles. We considered the number of possible underlying dimensions or a rank shift of 0–1 quintile as satisfactory agreement.
domains. We used principal factor analysis to model A shift of two quintiles was considered as intermediate the correlation between indicators and to show the agreement, whereas more than two quintiles were extent to which they reflect the same underlying concepts. We evaluated models with different numbers Finally, we explored the association of practice of factors and selected the model with best conceptual characteristics, that is, practice size and availability coherence, total variance explained, and communali- of supporting personnel (such as diabetes nurse or ties of the PQIs. The communality of each indicator, diabetes assistant), with the scores of PQIs using linear that is, the sum of the squared factor loadings for all regression. The Statistical Package for the Social factors for a given variable, shows the amount of Sciences for Windows (version 16.0; SPSS Inc., variance in a given PQI explained by the selected Chicago, IL) was used for all analyses.
factors: PQI loading across the same identified domainsas for the total population. We repeated the analysis in a subpopulation of GPs that had at least 70 T2DMpatients to assess whether population size would influ- From the GPs participating to the Groningen Initiative ence the observed domain structure and factor loadings.
to Analyse Type 2 diabetes Treatment project at the This cutoff excludes 16 practices in the lowest practice end of 2007, we included 76 (70%) practices that had eligible patients for all tested indicators coveringa total of 7944 T2DM patients. The characteristics of General characteristics of GPs and patient population the included practices and their patient population aresummarized in the Table 1. The scores of the PQIs calculated on a general practice level varied from General practice characteristics (n = 76) 11% (SE 18) to 79% (SE 9) (Table 2).
We carried out the principal factor analysis with two-, three-, four-, and five-factor solutions and No. T2DM patients visiting diabetes facility considered the five-factor model as being the best Percentage of all T2DM patients per practice interpretable and conceptually meaningful. The factors Percentage of practices with diabetes nurse or assistant explained a substantial part of the total variance with a cumulative variance of 16% (one factor), 30% (two factors), 43% (three factors), 56% (four factors), and 67% (five factors) (Table 3). No PQI was ex- cluded from the analysis because all indicators loaded across the factors with correlation coeffi- cients greater than 0.5. Communalities were 0.6 or The first two factors focused on the general first-step drug treatment recommendations for majority of T2DM *Body mass index: weight in kilograms divided by height in square meters first factor named “starting treatment I” History of CVD included history of myocardial infarction, ischemic heart disease, transient cerebral ischemia, stroke/cerebrovascular accident, and atherosclerosis/peripheral vascular disease as registered by GPs.
most patients, such as prescription of metformin, statin, Copyright 2011 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety, (2011) Mean general practice scores of PQIs for T2DM management (n = 76) T2DM patients with systolic blood pressure ≥140 mmHg and prescribed any antihypertensive drug T2DM patients prescribed a second antihypertensive drug from a different class if systolic blood pressure remained ≥ 140 mmHg with first class of antihypertensive drug T2DM patients with albuminuria prescribed RAAS-inhibitor PQIs merged to PQI 3 (%)T2DM patients without hypertension with albuminuria prescribed RAAS-inhibitor T2DM incident for hypertension patients with albuminuria prescribed RAAS-inhibitor T2DM prevalent for hypertension patients with albuminuria prescribed a multiple drug regime T2DM patients with a history of ischemic heart disease or myocardial infarction prescribed b-blocker Nonincident T2DM patients with HbA1c >7 % and prescribed any oral antihyperglycemic agent or insulin Nonincident T2DM patients not receiving insulin prescribed a second oral antihyperglycemic drug from a different class if with one oral antihyperglycemic drug HbA1c remained >7% T2DM patients who are prescribed insulin if with combination of two oral drugs HbA1c remained >7% Overweight prevalent T2DM patients prescribed a multiple drug regime containing metformin T2DM patients with LDL ≥ 2.5 mmol/L or TC ≥ 4.5 mmol/L who are prescribed a statin T2DM patients with a history of CVD prescribed acetyl salicylic acid RAAS inhibitor, renin–angiotensin–aldosterone system inhibitor; HbA1c, glycosilated hemoglobin; LDL, low-density lipoprotein; TC, total cholesterol.
and any antihypertensive medication (Table 3). The on prescribing quality assessment, we ranked GPs second factor, “starting treatment II,” consisted of two using composite score based on the initial 10 PQIs other PQIs focusing on the treatment initiation of and the composite score on the selected 5 PQIs, T2DM patients with specific risk factors, that is, prescrib- that is, PQIs 1, 4. 5, 6, 7. The comparison of these ing glucose lowering medication in patients with elevated composite scores showed that 81.5% of GPs had HbA1c levels and prescribing renin–angiotensin system an acceptable shift by either remaining within the inhibitors in T2DM patients with albuminuria. The third same quintile or shifting only by one quintile, identified factor reflected “treatment of CVD” in T2DM 10.5% had an intermediate shift by two quintiles, patients and comprised the two PQIs from our set of and only 8% had poor agreement because they indicators concerning patients with a history of CVD, shifted by more than two quintiles (Table 4).
focusing on prescription of beta blockers and acetyl We found no significant associations between salicylic acid. Finally, there were two factors focusing practice size, having a diabetes nurse or diabetes as- on next steps of treatment intensification. The factor sistant, or making use of the diabetes facility and the named “step 1 treatment intensification” included only composite scores of the PQIs. Also regarding the in- one PQI focusing on adding a second drug in patients dividual PQI scores, no significant or meaningful with hyperglycemia despite monotherapy with oral associations were found with these general practice glucose lowering medication. The “step 2 treatment intensification” factor comprised a PQI focusing onadding a second class antihypertensive medication if one class was not sufficient to control the blood pressureand a PQI on prescribing insulin in patients with uncon- Using factor analysis, we identified five prescribing trolled HbA1c levels despite oral glucose-lowering treat- quality domains for T2DM within our data set: two ment. Additional analysis limited to GPs that had at on initiation of treatment, two on treatment intensifica- least 70 T2DM patients showed similar results with tion steps, and one on the treatment of T2DM patients PQI loading across the same identified domains as for with known CVD. We selected a subset of five indica- tors, representing each of these domains. On a general Within each domain, we selected the indicator with practice level, the prescribing quality assessed with the highest loading as the PQIs that could represent this subset adequately reflected the overall prescribing that domain, that is, PQI 1 for “starting treatment I,” quality using the initial set of 10 indicators.
PQI 5 for “starting treatment II,” PQI 4 for “treatment One might expect that the PQIs focusing on the of CVD,” PQI 6 for “step 1 treatment intensification,” management of the same risk factor, for example, and PQI 7 for “step 2 treatment intensification” hypertension or hyperglycemia, would correlate highly (Table 3). To assess the effect of this selection and would therefore constitute one domain. Our study, Copyright 2011 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety, (2011) quality indicators for diabetes management however, showed that the PQIs that loaded on the samefactor often represented the management of different clinical risk factors related to diabetes. Previous studieshave shown that the relationship between prescribing indicators is often unpredictable with very differentprescribing indicators correlating to a high degree.26 Instead of correlations within a risk factor, we observed seemed to be linked to the different steps of treatmentintensity. This probably occurred because the prescrib- ing behavior of GPs is influenced by clinical triggers rather the medical history of individual patients. Recent studies have shown that medication burden and presence of comorbidities can have similar effects on the start and intensification of treatment for the management of hyperglycemia, hypertension, and Our selection of indicators was based on the highest factor loadings within each identified domain. The reduction we propose would result in a set of 5 instead of 10 indicators. Although this may seem a relative small reduction, it can reduce the data noise and administrative burden associated with reporting on large indicator sets for many practices. In addition, the use of a composite score on the basis of a small set of indicators can increase the comparability of quality measures across different general practices.
The indicators comprising the selected minimal set are supported by the highest level of scientific evidence,14 are accepted by the professionals in the field worldwide, and are operationally feasible.17 It has been shown that use of quality indicators contributes to improved quality of provided care and patient outcomes.29,30 In particular, indicators focusing on adequate and timely drug treatment were found to be predictive of better In our study, we have used complete individual level data on medication prescriptions made by GPs for all their T2DM patients. Although we had a large data set comprising electronic health records of 76 GPs with more than 7944 T2DM patients, our results may not be generalized to other settings. Prescribing patterns of primary care doctors in different countries which may vary both across and within countries.34,35 data sets and countries is recommended.
To our knowledge, this is the first study to look at the domains of prescribing the quality of T2DM care.
Our study showed that factor analysis can be functional for minimizing the number of indicators.
Copyright 2011 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety, (2011) Agreement between the composite scores per general practice* Additional supporting information may be found in Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quinile 5 Operational definitions for calculation of prescribing quality indicators (PQI) for type 2 diabetes mellitus Please Note: Wiley‐Blackwell are not responsible for the content or functionality of any supportingmaterials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
*Rows represent the quintile distribution of GPs based on a composite score of initial 10 PQIs; columns represent the quintile distribution ofGPs based on a composite score of the selected five PQIs. Dark gray cells represent GPs with satisfactory agreement between two composite pre- scribing scores; intermediate gray cells represent GPs with intermediate perspective from US researchers. Int J Qual Health Care 2000 Aug 1; 12(4): agreement between two composite prescribing scores; light gray cells rep- resent GPs with poor agreement between two composite prescribing 2. Mattke S, Seid M, Ma S. Evidence for the effect of disease management: is $1 billion a year a good investment? Am J Manag Care 2007 Dec; 13(12): 670–676.
3. Health Care Financing and Organization. Reducing the Administrative Burden of The results indicate that when making indicator selec- 4. National Quality Measures Clearing House. Selecting measures. tions, it is important to include prescribing indicators that represent different levels of treatment intensity.
5. Leas BF, Berman B, Kash KM, Crawford AG, Toner RW, Goldfarb NI, et al.
Quality measurement in diabetes care. Popul Health Manag 2009 Oct; 12(5): For the management of diabetes, this selection should include indicators focusing on the start of treatment, 6. Washington, D C : National Committee for Quality Assurance. HEDIS 2010 summary table of measures, product lines and changes. the intensification of treatment, and the management of patients with additional CVDs. The proposed set 7. Indicators for quality improvement. Full indicators list. NHS Information Center for health and social care. (accessed 20 June 2011) of indicators can then be used as a tool to monitor 8. Campbell SM, Hann M, Hacker J, Durie A, Thapar A, Roland MO. Quality and improve the quality of pharmacological manage- assessment for three common conditions in primary care: validity and reliabilityof review criteria developed by expert panels for angina, asthma and type 2 diabetes. Qual Saf Health Care 2002 Jun; 11(2): 125–130.
9. Nicolucci A, Greenfield S, Mattke S. Selecting indicators for the quality of diabetes care at the health systems level in OECD countries. Int J Qual HealthCare 2006 Sep; 18(Suppl 1): 26–30.
10. Shekelle PG, Vijan S. Quality indicators for the management of diabetes mellitus in vulnerable elders. JAGS 2007 200755(S2): S312–S317.
11. Voorham J, Denig P, Wolffenbuttel BH, Haaijer-Ruskamp FM. Cross-sectional The authors declare no conflict of interest.
versus sequential quality indicators of risk factor management in patients withtype 2 diabetes. Med Care 2008 Feb; 46(2): 133–141.
12. Wens J, Dirven K, Mathieu C, Paulus D, Van RP. Quality indicators for type-2 diabetes care in practice guidelines: an example from six European countries.
Prim Care Diabetes 2007 Feb; 1(1): 17–23.
13. The National Quality Forum. National Voluntary Consensus Standards for Adult Diabetes Care: 2005 Update. (accessed 20 • Using factor analysis we selected five PQI indica- 14. Martirosyan L, Braspenning J, Denig P, de Grauw WJ, Bouma M, Storms F, tors representing different dimensions of prescrib- et al. Prescribing quality indicators of type 2 diabetes mellitus ambulatory care.
ing quality in our dataset. This subset of indicators Qual Saf Health Care 2008 Oct; 17(5): 318–323.
15. Orozco LJ, Buchleitner AM, Gimenez-Perez G, Roque IF, Richter B, Mauricio adequately reflected the overall prescribing qual- D. Exercise or exercise and diet for preventing type 2 diabetes mellitus.
ity as assessed with a larger set of PQI at a family Cochrane Database Syst Rev 2008; (3): CD003054.
16. Turner RC, Cull CA, Frighi V, Holman RR. Glycemic control with diet, sulfonylurea, metformin, or insulin in patients with type 2 diabetes mellitus: • Factor analysis is helpful for selecting a mini- progressive requirement for multiple therapies (UKPDS 49). UK ProspectiveDiabetes Study (UKPDS) Group. JAMA 1999 Jun 2; 281(21): 2005–2012.
mum set of prescribing quality indicators through 17. Martirosyan L, Voorham J, Haaijer-Ruskamp FM, Braspenning J, Wolffenbuttel identification of underlying dimensions of pre- BH, Denig P. A systematic literature review: prescribing indicators related totype 2 diabetes mellitus and cardiovascular risk management. Pharmacoepidemiol Drug Saf 2010 Apr; 19(4): 319–334.
• A minimal set of PQI for T2DM care should in- 18. Houghton G, Rouse A. Are NHS primary care performance indicator scores acceptable as markers of general practitioner quality? Br J Gen Pract 2004 tegrate different treatment intensity levels of clin- 19. Rasmussen HM, Sondergaard J, Kampmann JP, Andersen M. General practitioners prefer prescribing indicators based on detailed information on individual patients: aDelphi study. Eur J Clin Pharmacol 2005 May; 61(3): 237–241.
Copyright 2011 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety, (2011) quality indicators for diabetes management 20. Campbell SM, Braspenning J, Hutchinson A, Marshall M. Research methods 29. de Grauw WJ, van Gerwen WH, van de Lisdonk EH, van den Hoogen HJ, van used in developing and applying quality indicators in primary care. Qual Saf den Bosch WJ, van Weel C. Outcomes of audit-enhanced monitoring of patients Health Care 2002 Dec; 11(4): 358–364.
with type 2 diabetes. J Fam Pract 2002 May; 51(5): 459–464.
21. Lee TH. Eulogy for a Quality Measure. N Engl J Med 2007 Sep 20; 357(12): 30. Kiefe CI, Allison JJ, Williams OD, Person SD, Weaver MT, Weissman NW.
Improving quality improvement using achievable benchmarks for physician feed- 22. Skyggedal Rasmussen H, Sondergaard J, Sokolowski I, Kampmann J, Andersen back: a randomized controlled trial. JAMA 2001 Jun 13; 285(22): 2871–2879.
M. Factor analysis improves the selection of prescribing indicators. Eur J Clin 31. Sidorenkov G, Haaijer-Ruskamp FM, de ZD, Bilo H, Denig P. Review: re- Pharmacol 2006 Nov 1; 62(11): 953–958.
lation between quality-of-care indicators for diabetes and patient out- 23. Groningen initiative to analyze Type 2 diabetes treatment. comes: a systematic literature review. Med Care Res Rev 2011 Jun; 68(3): 24. Voorham J, Denig P. Computerized extraction of information on the quality of 32. Bjerrum L, Boada A, Cots JM, Llor C, Fores GD, Gahrn-Hansen B, et al.
diabetes care from free text in electronic patient records of general practitioners.
Respiratory tract infections in general practice: considerable differences in J Am Med Inform Assoc 2007 May; 14(3): 349–354.
prescribing habits between general practitioners in Denmark and Spain. Eur J 25. The Dutch College of General Practitioners. Management of Type 2 diabetes Clin Pharmacol 2004 Mar; 60(1): 23–28.
33. Rosman S, Le Vaillant M, Schellevis F, Clerc P, Verheij R, Pelletier-Fleury N.
Prescribing patterns for upper respiratory tract infections in general practice 26. Avery AJ, Heron T, Lloyd D, Harris CM, Roberts D. Investigating relationships in France and in the Netherlands. Eur J Public Health 2008 Jun 1; 18(3): between a range of potential indicators of general practice prescribing: an observational study. J Clin Pharm Ther 1998 Dec; 23(6): 441–450.
34. Burgers JS, Bailey JV, Klazinga NS, Van der Bij AK, Grol R, Feder G. Inside 27. Voorham J, Haaijer-Ruskamp FM, Wolffenbuttel BH, Stolk RP, Denig P.
Guidelines. Diabetes Care 2002 Nov; 25(11): 1933–1939.
Cardiometabolic treatment decisions in patients with type 2 diabetes: the role 35. Clark MJ, Jr., Sterrett JJ, Carson DS. Diabetes guidelines: a summary and comparison of repeated measurements and medication burden. Qual Saf Health Care 2010 of the recommendations of the American Diabetes Association, Veterans Health Administration, and American Association of Clinical Endocrinologists. Clin Ther 28. Voorham J. Assessing cardiometabolic treatment quality in general practice 2010.
Copyright 2011 John Wiley & Sons, Ltd.
Pharmacoepidemiology and Drug Safety, (2011)

Source: http://www.giantt.nl/Martirosyan_factoranalysePDS2011.pdf