- Email: [email protected]
Effectiveness of Statin Prescribing on Reducing Mortality in South Asian, Chinese, and White Patients With Diabetes
Canadian Journal of Cardiology 29 (2013) 920e926
Clinical Research
Effectiveness of Statin Prescribing on Reducing Mortality in South Asian, Chinese, and White Patients With Diabetes Nathan W. Brunner, MD,a Krishnan Ramanathan, MD,a Hong Wang, MSc,b Hude Quan, PhD,c and Nadia A. Khan, MD, MScb,d a b
Division of Cardiology, University of British Columbia, Vancouver, British Columbia, Canada
Centre for Health Evaluation and Outcomes Sciences, St Paul’s Hospital, Vancouver, British Columbia, Canada c d
Department of Community Health Sciences, University of Calgary, Calgary, Alberta, Canada
Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
ABSTRACT
RESUM E
Background: Clinical trials have shown that 3-hydroxy-3-methylgutaryl coenzyme A reductase inhibitors (statins) reduce mortality in patients with diabetes. However, as these trials were conducted in largely white populations, it is unknown whether the benefits of statins can be extended to other ethnic populations in which the incidence of diabetes is rising sharply. We investigated associations between statin prescription and outcomes in a multiethnic population with diabetes. Methods: We identified all patients with newly diagnosed diabetes in British Columbia, Canada (1993-2006), using administrative data. Validated surname analysis was used to identify South Asian and Chinese patients. Statin prescribing was defined as any prescription filled within 1 year of diabetes diagnosis according to a provincial pharmacy database. Median length of follow-up was 4 years. Cox proportional hazards models were constructed for each ethnic group to determine the association of statin prescribing with time to death, adjusting for covariates including age, sex, socioeconomic status, and comorbid conditions. Results: There were 143,630 white, 9529 South Asian, and 14,084 Chinese persons with newly diagnosed diabetes. White patients were older and had more comorbidity than the other groups. Statin prescribing was associated with lower mortality compared with no prescribing within each ethnic group: South Asian (Hazard Ratio [HR], 0.69; 95% confidence interval [CI], 0.55-0.86; P ¼ 0.001), Chinese
montre que les inhibiteurs Introduction : Les essais cliniques ont de ductase de la HMG-CoA (3-hydroxy-3-methylglutaryl coenzyme A) re duisent la mortalite chez les patients ayant le diabète. (statines) re te mene s chez des populations en Cependant, comme ces essais ont e ne fices des statines peuvent grande partie blanches, on ignore si les be tendre à d’autres ethnies chez qui l’incidence de diabète est en s’e les associations entre l’ordonnance forte hausse. Nous avons examine sultats chez une population multiethnique ayant le de statines et les re diabète. thodes : Nous avons de termine tous les patients ayant nouvelleMe ment reçu un diagnostic de diabète en Colombie-Britannique, au es administratives. La Canada (1993-2006), en utilisant les donne te utilise e pour validation de l’analyse du nom de famille a e terminer les patients sud-asiatiques et chinois. L’ordonnance de de te de finie comme toute ordonnance remplie dans la 1re statines a e e du diagnostic du diabète selon une base de donne es pharmaanne e me diane du suivi a e te de 4 ans. Les ceutique provinciale. La dure te e labore s pour chamodèles de risques proportionnels de Cox ont e terminer le lien entre l’ordonnance cun des groupes ethniques pour de de statines et le moment de la mort, l’ajustement de covariables conomique et les facteurs de incluant l’âge, le sexe, le statut socioe . comorbidite sultats : Il avait 143 630 personnes blanches, 9 529 personnes Re
With increased urbanization and development, the prevalence of type 2 diabetes mellitus is rising to epidemic proportions worldwide. The greatest increase is occurring in Asia and the Indian subcontinent, where it is expected to affect > 130 million individuals in the next 2 decades.1 Recent research also demonstrates that the South Asian and Chinese diasporas
to the West similarly experience a high prevalence of type 2 diabetes.2,3 This is a critical global health issue, given the staggering health and economic costs associated with diabetes. Diabetes is associated with a 2- to 4-fold increase in risk of premature death, myocardial infarction, and stroke.4-6 In 2003, developing countries spent between 2.5% and 15% of their direct health care budgets on diabetes-related treatments.7 Substantial strides have been made in lowering the risk of death and cardiovascular outcomes in patients with diabetes. Clinical trials have shown that therapy with coenzyme A (CoA) reductase inhibitors (statins) results in a relative risk reduction
Received for publication July 27, 2012. Accepted October 6, 2012. Corresponding author: Dr Nadia Khan, 620 B, 1081 Burrard St, St Paul’s Hospital, Vancouver, British Columbia V6Z 1Y6, Canada. E-mail: [email protected] See page 924 for disclosure information.
0828-282X/$ - see front matter Ó 2013 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cjca.2012.10.012
Brunner et al. Statins in Diabetes by Ethnicity
921
(HR, 0.60; 95% CI, 0.49-0.72; P < 0.0001), and white (HR, 0.65; 95% CI, 0.63-0.67; P < 0.0001). Conclusions: Statin prescribing is associated with lower mortality in white, Chinese, and South Asian patients with newly diagnosed diabetes, confirming the benefits of statins across these ethnic groups.
sud-asiatiques et 14 084 personnes chinoises ayant nouvellement taient plus âge s et reçu un diagnostic de diabète. Les patients blancs e s que ceux des autres groupes. L’ordonavaient plus de comorbidite te associe e à une plus faible mortalite comnance de statines a e parativement à la non-ordonnance au sein de chacun des groupes ethniques : les Sud-Asiatiques (rapport de risque [RR], 0,69; intervalle de confiance [IC] à 95 %, 0,55-0,86; P ¼ 0,001), les Chinois (RR, 0,60; IC à 95 %, 0,49-0,72; P < 0,0001) et les blancs (RR, 0,65; IC à 95 %, 0,63-0,67; P < 0,0001). e à une plus faible Conclusions : L’ordonnance de statines est associe chez les patients blancs, chinois et sud-asiatiques ayant mortalite nouvellement reçu un diagnostic de diabète, ce qui confirme les ne fices des statines au sein de ces groupes ethniques. be
of 9% to 27% for death, 22% to 37% for major coronary event, and 21% to 47% for stroke in primary and secondary prevention populations with diabetes.8-10 The majority of patients with type 2 diabetes mellitus are at high risk for vascular events, and control of low-density lipoprotein cholesterol to less than 2.0 mmol/L is indicated for this population.11 Although the incidence and mortality associated with diabetes are now in epidemic proportions in Asia, clinical trials demonstrating benefit with statin therapy were conducted in primarily white populations.9,10,12 Moreover, it is unlikely that these trials will be repeated in Asian populations. Although some studies found that statin therapy can effectively modulate levels of low-density lipoprotein cholesterol and high-density lipoprotein cholesterol in South Asian patients to a similar degree as in white populations,13,14 other studies documented increased plasma statin levels in Chinese and South Asian populations, compared with white populations.1,15,16 Thus, whether South Asian and Chinese patients also derive net benefits in terms of mortality and cardiovascular morbidity risk reductions from statin therapy is not known. Ethnic variations in response to statin therapy, if they exist, would have substantial implications for both primary and secondary prevention efforts in these increasingly prevalent populations. We sought to identify differences in mortality in a population cohort of South Asian, Chinese. and whitepatients with newly diagnosed diabetes as a function of statin prescribing. We also investigated risk of the composite of death, hospitalization for acute myocardial infarction (AMI), or stroke, as well as the individual components in these groups, as secondary end points for this analysis.
Classification of Diseases, Ninth and Tenth Revisions (ICD-9 and ICD-10). Physician claims files provided diagnoses and listings of services provided for all primary care clinic and hospital encounters for all residents of BC. Demographic data were obtained through linkage with the universal provincial health care insurance registry. We limited our analysis to persons with newly diagnosed diabetes in order to identify a population of patients at similar and early stages of disease. All newly diagnosed diabetes cases were identified with a diabetes coding algorithm, which was validated in a population of 3362 patients in Canada to have a sensitivity of 92.3% and a specificity of 96.9% for physiciandiagnosed diabetes.18 Patients with diabetes were defined as those with any of the following codes: ICD-9, Clinical Modification codes 250.x and ICD-10 codes E109, E119, E139, E149, E101, E111, E131, E141, E105, E115, E135, or E145 in at least 1 hospital discharge abstract or 2 physician claims within 2 years. In this manner, virtually all patients with a diagnosis of diabetes in the province of BC were identified. New cases were defined as patients with a minimum of 3 preceding years without coding for diabetes. Only those with valid provincial health insurance for at minimum of 3 years prior to their diagnosis were included, in order to avoid the misclassification of newly arrived immigrants with diabetes as new cases. To reduce the misclassification of type 2 diabetes as type 1 diabetes, exclusion criteria consisted of patients younger than 35 years. Patients with gestational diabetes, identified with coding for an obstetrical event within 5 months of diabetes diagnosis, were also excluded. To control for severity of illness at time of diabetes diagnosis, we measured clinical variables from the Charlson comorbidity index as this index is associated with mortality in patients with type 2 diabetes.19,20 For each patient, administrative codes for these variables were extracted from physician service claims and hospital discharge abstracts dating from 1993 until the date of diabetes diagnosis. Area level median income was determined from postal codes and the Canadian Census 2001 socioeconomic file.
Methods This study was a retrospective cohort study that used administrative data from the province of British Columbia (BC) Canada. Study population We identified all patients newly diagnosed with diabetes residing in the province of BC from hospital discharge abstracts and physician claims files collected from 1993 through 2006, of which the details are described elsewhere.17 Hospital discharge abstract data contain information on all hospital visits occurring in BC for all residents and include up to 25 diagnosis fields expressed according to the International
Identification of ethnicity Ethnicity is not recorded in administrative data in Canada. We categorized ethnicity using validated surname analysis algorithms as South Asian (from Pakistan, India, or Bangladesh) or Chinese (ancestry from China, Taiwan, or Hong Kong). Patient surnames, recorded in provincial
922
registries, were merged with Quan’s Chinese name list21 and the Nam Pehchan computer program22 to define Chinese and South Asian ethnicity. Compared with self-report, the sensitivity for Quan’s surname algorithm was 78%, specificity was 99.7%, and the positive predictive value (PPV) was 81%.21 Validation studies for the Nam Pehchan algorithm report a sensitivity of 90% to 94%, specificity of 99.4%, and PPV of 63% to 96%.23 In BC, blacks account for 0.7% of the population, and after excluding Chinese and South Asians, 93% of the remaining population is white.24 Thus, we categorized all noneSouth Asian and non-Chinese patients as white. Determination of 3-hydroxy-3-methylglutaryl CoA reductase inhibitor prescribing The BC PharmaNet database contains complete information on all prescription drug claims filled at community pharmacies throughout BC. Information is included on all patients, regardless of age or amount of copay. Prescription drug claims contain information on date of prescription, drug name and dosage, and days supplied of each medication. Accuracy in Canadian provincial drug databases is high, with a coding error rate of only 0.7% reported in an analogous database.25 Patients were taken as being prescribed statins if they filled a prescription for any 3-hydroxy-3-methylglutaryl CoA inhibitor within 1 year following their diagnosis of diabetes. In BC, prescription drug costs are paid fully by the patient up to an income-based deductible. For drug expenses exceeding the deductible, 70% of prescription drug costs are subsidized by the PharmaCare program. If patient drug expenditures exceed a certain percentage of their net income, additional prescriptions are fully covered. In order to facilitate comparison between different statins, statins were expressed in terms of an equivalent atorvastatin dose, in a manner similar to that employed by Gupta and colleagues.14 Ten milligrams of atorvastatin was taken to be equivalent to 5 mg of rosuvastatin, 40 mg of lovastatin, 40 mg of pravastatin, 40 mg of fluvastatin, and 0.8 mg of ceruvastatin, respectively. Clinical outcomes The primary outcome was time to death. We obtained date of death through linkage with BC vital statistics, a mandatory provincial database of mortality that is updated daily. Secondary end points included (1) time to composite risk of death or (2) hospitalization for acute stroke or AMI, whichever was experienced first, as well as the individual components of AMI and stroke. Hospitalization for acute stroke and AMI was determined by the ICD-9 and ICD-10 codes from hospital discharge abstracts. The PPV for AMI and stroke with these algorithms is 0.95 (95% CI, 0.91-0.98) and 0.85 to 0.98, respectively.26 Time to event was assessed up to March 31, 2006, with a maximum duration of follow-up of 9 years (median follow-up, 4.0 years). Patients were censored if they moved out of province or reached the end of the observation period. Statistical analysis Comparisons of baseline characteristics for each ethnic group were made using the c2 test for categorical variables and
Canadian Journal of Cardiology Volume 29 2013
analysis of variance for continuous variables. To determine the association between statin prescribing and time to death, we constructed Cox proportional hazards models for each ethnic group, comparing those prescribed and not prescribed statin within 1 year of diagnosis. We used the risk adjustment method to construct each model, adjusting for age in 5-year categories, income quintile, year at diagnosis, sex, and Charlson comorbidities including hypertension, AMI, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, renal insufficiency, and malignancy. Initial statin dosages were compared between ethnic groups with a multivariate model that included age, sex, and socioeconomic status. For the secondary end point of AMI, we excluded those with a previous AMI or coronary revascularization prior to their statin prescription in order to lesson confounding associated with the inclusion of patients who were started on statins for established coronary artery disease rather than diabetes alone. For all models, we additionally performed propensity score analysis with a greedy match algorithm.27 We calculated propensity scores for statin prescribing, using a nonparsimonious multivariate logistic regression model fitting age, sex, socioeconomic status, and baseline clinical variables in the model. We then used the propensity scores to obtain 1-to-1 matches between subjects who received a prescription for statin therapy and those who did not, using a caliper width of 0.2 standard deviation of the log odds of the propensity score, age ( 1 year), sex, and index date of diabetes diagnosis ( 1.2 year). Standard differences were used to compare the baseline characteristics between matched groups, and a standard difference of less than 10% was taken as good balance. Within matched pairs, we used a stratified Cox proportional hazards model to compare time to event for each ethnicity, using a restricted cubic spline method. Findings derived from propensity score models were similar to the main models (see Appendix I), and the main adjusted Cox regression models are presented herein. All P values presented are 2-tailed, and a value of less than 0.05 was considered significant. Analyses were performed with SAS version 9.2 (SAS Institute Inc, Cary, NC). The study was approved by the local ethics institutional review board. Results Our total cohort consisted of 167,243 patients with newly diagnosed diabetes. Of these, 9529 (5.7%) were identified as South Asian and 14,084 (8.4%) as Chinese. White patients were significantly older at diagnosis than were South Asian or Chinese patients (Table 1). White patients were more likely to be of higher socioeconomic status compared with South Asians and Chinese. There were significant differences in prevalence of comorbidities for each ethnic group. Myocardial infraction and congestive heart failure were more common among South Asians, whereas the prevalence of hypertension in South Asians was half that of whites and Chinese. Statins were prescribed to 25% of South Asians, 21% of Chinese, and 27% of white patients, respectively. The mean initial statin doses for each ethnicity, expressed as equivalent atorvastatin doses, are shown in Table 1. Compared with the white patients, the initial statin dose was significantly lower for Chinese patients (P < 0.001) but not for South Asian
Brunner et al. Statins in Diabetes by Ethnicity
923
Table 1. Baseline characteristics among diabetes patients aged 35 years and older in British Columbia Characteristics Age at diagnosis, n (%) Mean age, years (SD) 35-49 years 50-64 years 65-79 years 80 years Female, n (%) Socioeconomic quintile, n (%) First quintile (low) Second quintile Third quintile Fourth quintile Fifth quintile (high) Unknown Comorbidity (Charlson measurements), n (%) Hypertension Myocardial infarction Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic pulmonary disease Renal disease Cancer Outcomes, n (%) Death AMI Stroke Proportion of total receiving statin therapy Initial statin dose (atorvastatin equivalent)
South Asian (n ¼ 9,529)
Chinese (n ¼ 14,084)
White (n ¼ 143,630)
P value
56.1 (12.3) 3157 (33.1) 3890 (40.8) 2143 (22.5) 339(3.6) 6749 (47.9)
59.4 3585 5331 4275 893 4268
(12.7) (25.5) (37.9) (30.4) (6.3) (44.8)
61.6 28,177 55,815 46,574 13,064 6749
(13.0) (19.6) (38.9) (32.4) (9.1) (47.9)
<0.001 <0.001
2459 2673 1911 1278 838 370
(25.8) (28.1) (20.1) (13.4) (8.8) (3.9)
3756 3423 2422 2059 2047 377
(26.7) (24.3) (17.2) (14.6) (14.5) (2.7)
32,196 27,741 25,993 24,857 23,543 9300
(22.4) (19.3) (18.1) (17.3) (16.4) (6.5)
<0.001
2422 2059 2047 377 217 48 1225 101 215
(17.2) (14.6) (14.5) (2.7) (2.3) (0.5) (12.9) (1.1) (2.3)
5605 176 329 211 342 87 1032 230 456
(39.8) (1.3) (2.3) (1.5) (2.4) (0.6) (7.3) (1.6) (3.2)
55,866 5045 7898 3494 5260 2028 17,098 2624 8656
(38.9) (3.5) (5.5) (2.4) (3.7) (1.4) (11.9) (1.8) (6.0)
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
<0.001
551 (5.8) 185 (1.9) 129 (1.4) 25%
920 (6.5) 103 (0.7) 222 (1.6) 21%
19,997 (13.9) 3344 (2.2) 3225 (2.3) 27%
<0.001 <0.001 <0.001 <0.001
14.2 (9.0)
11.6 (7.5)
13.8 (9.8)
<0.001
AMI, acute myocardial infarction; SD, standard deviation.
patients (P ¼ 0.4), after adjusting for age, sex, and socioeconomic status. Clinical outcomes associated with statin prescribing During the observation period, there were 21,468 deaths, 3632 nonfatal AMIs, and 3576 stroke events among the study cohort. As shown in Table 1, the prevalence of death and stroke during the study period was higher in whites than in South Asians or Chinese. Rates of AMI were lowest in Chinese patients, compared with whites and South Asians. As
shown in Table 2, statin prescribing was associated with a significant decrease in mortality for all ethnic groups, even after we adjusted for baseline socio-demographic and clinical characteristics. Statin prescribing was also associated with a reduced risk of the composite end point of death, AMI, or stroke for Chinese and white patients. This was also seen in South Asians in our main analysis, although the difference became nonsignificant in the propensity score analysis. There were no significant differences in risk of hospitalization for AMI in those prescribed statins compared with those who
Table 2. Adjusted outcomes associated with statin prescribing for South Asian, Chinese, and white patients with newly diagnosed diabetes, as obtained from Cox proportional hazards models Outcome HR (95% CI), P value Total mortality Composite* outcome AMIy Stroke
South Asian
Chinese
White
0.69 (0.55-0.86) P ¼ 0.001 0.83 (0.70-0.99) P ¼ 0.04 1.16 (0.83-1.64) P ¼ 0.37 1.38 (0.94-2.04) P ¼ 0.1
0.60 (0.49-0.72) P < 0.0001 0.57 (0.48-0.68) P < 0.0001 1.07 (0.67-1.7) P ¼ 0.8 0.54 (0.38-0.79) P ¼ 0.001
0.65 (0.63-0.67) P < 0.0001 0.66 (0.64-0.68) P < 0.0001 0.93 (0.86-1.01) P ¼ 0.1 0.94 (0.86-1.01) P ¼ 0.1
All Cox regression models are adjusted for age, sex, income quintile, hypertension, congestive heart failure, peripheral vascular disease, dementia, chronic pulmonary disease, renal insufficiency, and malignancy. AMI, acute myocardial infarction; CI, confidence interval; HR, hazard ratio. * Composite of mortality, AMI, and stroke. y Excluded patients with prior percutaneous coronary intervention or coronary artery bypass graft.
924
were not. Among Chinese patients with diabetes, statin prescribing was associated with significantly reduced risk of hospitalization for acute stroke. However, there was no difference with risk of acute stroke in South Asian or white patients. Interpretation In this analysis of a large administrative database of persons newly diagnosed with diabetes, statin prescribing was associated with a reduction in mortality in South Asian, Chinese, and white populations. The magnitude of the reduction in mortality was generally similar for all 3 ethnicities, suggesting they derived similar benefits from statin therapy. South Asian, Chinese, and white patients had substantial differences in their baseline characteristics. South Asians were much more likely to have a history of prior MI compared with the other groups. This finding is consistent with other studies done in healthy multiethnic populations in Canada.2,28 Statin starting dosage was low for all ethnicities studied and was significantly lower for Chinese patients compared with South Asian and white patients. This seems consistent with the emergence of studies demonstrating ethnic differences in rosuvastatin metabolism in 2005, prompting a Health Advisory from Health Canada.15,29,30 However, many of our patients began statin therapy prior to 2005, suggesting that physicians were cautious about statin therapy in Chinese populations even before then. The mortality findings in our study mirror those of randomized trials carried out in primarily white populations. The Collaborative Atorvastatin Diabetes Study (CARDS) and the Heart Protection Study (HPS), which included predominately white patients, reported a 20% to 27% relative risk reduction in death for primary prevention patients with diabetes. Our large population-based study suggests that the mortality benefits of statins seen in white populations may extend to South Asian and Chinese populations as well. In our analysis of the secondary end points, fewer differences in outcomes were noted with statin prescribing. Although our study showed an association between statin prescribing and lower incidence of stroke in the Chinese population, this was not observed in white or South Asian patients. In white populations, the HPS and CARDS studies both demonstrated a reduction in stroke with statin therapy.10,13 There was also no association between statin prescribing and AMI in any ethnic groups. This lack of association between statin usage and stroke and AMI contrasts with randomized controlled trials and is puzzling. It may be partially a result of the lower event rates in our study. In our AMI analysis, all patients with prior AMI or coronary revascularization were excluded in order to avoid confounding due to statin prescription for prior cardiac event. Therefore, this was a purely primary prevention population, explaining the modest incidence of stroke and AMI during follow-up. Given that significant differences exist in the development and disease course of diabetes in various ethnicities, we limited our analysis to incident cases only. Although the exclusion of prior coronary revascularization in our AMI analysis lessened confounding by indication, it may also have introduced some selection bias by preferentially including patients with few prior medical contacts. Moreover,
Canadian Journal of Cardiology Volume 29 2013
excluding these patients lowered our statistical power, especially in the South Asian and Chinese groups. Also, it is possible that statins may have been prescribed preferentially to patients with higher cardiac risk not accounted for in our analysis. This would have attenuated our observed cardiovascular benefit of statins. The major strength of our study is it represents one of the largest retrospective studies to investigate associations between statin prescription and clinical end points in a “real-world” multiethnic population. In addition, we used propensity score matching to minimize the effects of confounding. However, there are several limitations to studies of this nature. As this was a retrospective observational study, we were not able to adjust for all potentially relevant covariates including A1c level, smoking, and other health behaviours. We could not evaluate resultant changes in lipid levels associated with statin prescribing. Stroke and nonfatal AMI occurring outside of the hospital setting would not be included in our results. In our study, ethnicity was determined by surname analysis. While the gold standard for the determination of ethnicity remains self-report,31 the algorithms used have been validated and correlate well with the gold standard. Finally, we used prescription claims data as a surrogate for medication ingestion. However, prescription claims data are shown to correlate with home inventory pill count as well as serum measures of drug presence.32-34 Conclusions Our study provides new evidence that statin prescribing is associated with lower mortality in South Asian, Chinese, and white patients with newly diagnosed diabetes. These results should reassure prescribers that the mortality benefits of statin observed in landmark clinical trials in diabetes can likely be generalized to South Asian and Chinese populations with diabetes. However, further study evaluating effectiveness of statin prescribing on halting the progression of cardiac atherosclerosis and stroke are greatly needed in these ethnic groups. Funding Sources This study was funded by the Canadian Institutes of Health Research (CIHR). N.A.K. received support from CIHR New Investigator Award and the Michael Smith Foundation for Health Research Career Scientist award. H.Q. receives funding from the Alberta Heritage Foundation for Medical Research Senior Population. Disclosures The authors have no conflicts of interest to disclose. References 1. Ramaraj R, Chellappa P. Cardiovascular risk in South Asians. Postgrad Med J 2008;84:518-23. 2. Anand SS, Yusuf S, Vuksan V, et al. Differences in risk factors, atherosclerosis, and cardiovascular disease between ethnic groups in Canada: the Study of Health Assessment and Risk in Ethnic groups (SHARE). Lancet 2000;356:279-84.
Brunner et al. Statins in Diabetes by Ethnicity 3. Rajpathak SN, Gupta LS, Waddell EN, et al. Elevated risk of type 2 diabetes and metabolic syndrome among Asians and South Asians: results from the 2004 New York City HANES. Ethn Dis 2010;20: 225-30. 4. Nichols GA, Gullion CM, Koro CE, Ephross SA, Brown JB. The incidence of congestive heart failure in type 2 diabetes: an update. Diabetes Care 2004;27:1879-84. 5. Lee WL, Cheung AM, Cape D, Zinman B. Impact of diabetes on coronary artery disease in women and men: a meta-analysis of prospective studies. Diabetes Care 2000;23:962-8. 6. Kissela BM, Khoury J, Kleindorfer D, et al. Epidemiology of ischemic stroke in patients with diabetes: the greater Cincinnati/Northern Kentucky Stroke Study. Diabetes Care 2005;28:355-9. 7. Narayan KMV, Zhang P, Kanaya AM, et al. Diabetes: the pandemic and potential solutions. In: Jamison DT, Breman JG, Measham AR, et al., eds. Disease Control Priorities in Developing Countries. 2nd ed. Washington, DC: World Bank, 2006:591. 8. Cholesterol Treatment Trialists’ (CTT) Collaborators, Kearney PM, Blackwell L, Collins R, et al. Efficacy of cholesterol-lowering therapy in 18,686 people with diabetes in 14 randomised trials of statins: a metaanalysis. Lancet 2008;371:117-25. 9. Colhoun HM, Betteridge DJ, Durrington PN, et al.; CARDS investigators. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo-controlled trial. Lancet 2004;364: 685-96. 10. Sacks FM, Tonkin AM, Craven T, et al. Coronary heart disease in patients with low LDL-cholesterol: benefit of pravastatin in diabetics and enhanced role for HDL-cholesterol and triglycerides as risk factors. Circulation 2002;105:1424-8. 11. Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Canadian Diabetes Association 2008 clinical practice guidelines for the prevention and management of diabetes in Canada. Can J Diabetes 2008;32(suppl 1):S1-S201. 12. Collins R, Armitage J, Parish S, Sleigh P, Peto R; Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterollowering with simvastatin in 5963 people with diabetes: a randomised placebo-controlled trial. Lancet 2003;361:2005-16. 13. Patel JV, Gupta S, Lie F, Hughes EA. Efficacy and safety of atorvastatin in South Asian patients with dyslipidemia: an open label noncomparative pilot study. Vasc Health Risk Manag 2005;1:351-6. 14. Gupta M, Braga MF, Teoh H, Tsigoulis M, Verma S. Statin effects on LDL and HDL cholesterol in South Asian and white populations. J Clin Pharmacol 2009;49:831-7. 15. Lee E, Ryan S, Birmingham B, et al. Rosuvastatin pharmacokinetics and pharmacogenetics in white and Asian subjects residing in the same environment. Clin Pharmacol Ther 2005;78:330-41. 16. Tzeng TB, Schneck DW, Birmingham BK, et al. Population pharmacokinetics of rosuvastatin: implications of renal impairment, race, and dyslipidaemia. Curr Med Res Opin 2008;24:2575-85. 17. Khan NA, Wang H, Anand S, et al. Ethnicity and sex affect diabetes incidence and outcomes. Diabetes Care 2011;34:96-101.
925
18. Chen G, Khan N, Walker R, Quan H. Validating ICD coding algorithms for diabetes mellitus from administrative data. Diabetes Res Clin Pract 2010;89:189-95. 19. Monami M, Lambertucci L, Lamanna C, et al. Are comorbidity indices useful in predicting all-cause mortality in Type 2 diabetic patients? comparison between Charlson index and disease count. Aging Clin Exp Res 2007;19:492-6. 20. Tao LS, MacKenzie CR, Charlson ME. Predictors of postoperative complications in the patient with diabetes mellitus. J Diabetes Complications 2008;22:24-8. 21. Quan H, Wang F, Schopflocher D, et al. Development and validation of a surname list to define Chinese ethnicity. Med Care 2006;44:328-33. 22. Macfarlane GJ, Lunt M, Palmer B, Afzal C, Silman AJ, Esmail A. Determining aspects of ethnicity amongst persons of South Asian origin: the use of a surname-classification programme (Nam Pehchan). Public Health 2007;121:231-6. 23. Cummins C, Winter H, Cheng KK, Maric R, Silcocks P, Varghese C. An assessment of the Nam Pehchan computer program for the identification of names of South Asian ethnic origin. J Public Health Med 1999;21: 401-6. 24. Statistics Canada. 2001 Census data: British Columbia. Available at: http://www12.statcan.ca/english/census01/products/standard/prprofile/ prprofile.cfm?G=59. Accessed December 6, 2012. 25. Levy AR, O’Brien BJ, Sellors C, Grootendorst P, Willison D. Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 2003;10:67-71. 26. Tu JV, Austin PC, Naylor CD, Iron K, Zhang H. Acute myocardial infarction outcomes in Ontario. In: Naylor CD, Slaughter PM, eds. Cardiovascular Health and Services in Ontario: An ICES Atlas. Toronto, Canada: Institute for Clinical Evaluative Studies, 1999:83-110. 27. Austin PC. The performance of different propensity-score methods for estimating differences in proportions (risk differences or absolute risk reductions) in observational studies. Stat Med 2010;29: 2137-48. 28. Chiu M, Austin PC, Manuel DG, Tu JV. Comparison of cardiovascular risk profiles among ethnic groups using population health surveys between 1996 and 2007. CMAJ 2010;182:E301-10. 29. Liao JK. Safety and efficacy of statins in Asians. Am J Cardiol 2007;99: 410-4. 30. Health Canada. Health Canada advises patients about the risk of CrestorÒ (rosuvastatin). Available at: http://www.hc-sc.gc.ca/ahc-asc/ media/advisories-avis/_2005/2005_10-eng.php. Accessed October 2, 2012. 31. Mays VM, Ponce NA, Washington DL, Cochran SD. Classification of race and ethnicity: implications for public health. Annu Rev Public Health 2003;24:83-110. 32. Lau HS, de Boer A, Beuning KS, Porsius A. Validation of pharmacy records in drug exposure assessment. J Clin Epidemiol 1997;50:619-25. 33. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105-16. 34. West SL, Savitz DA, Koch G, Strom BL, Guess HA, Hartzema A. Recall accuracy for prescription medications: self-report compared with database information. Am J Epidemiol 1995;142:1103-12.
926
Canadian Journal of Cardiology Volume 29 2013
Appendix I. Adjusted outcomes associated with statin prescribing for South Asian, Chinese, and White Patients with newly diagnosed diabetes, as obtained from propensity score analysis Outcome HR (95% CI), P value Total mortality Composite* outcome AMIy Stroke
South Asian
Chinese
White
0.70 (0.50-0.97) P ¼ 0.03 0.89 (0.69-1.14) P ¼ 0.35 1.01 (0.64-1.59) P ¼ 0.98 1.29 (0.74-2.26) P ¼ 0.37
0.58 (0.44-0.77) P ¼ 0.0002 0.60 (0.47-0.77) P < 0.0001 0.95 (0.49-1.84) P ¼ 0.87 0.53 (0.32-0.87) P ¼ 0.012
0.62 (0.59-0.66) P < 0.0001 0.67 (0.64-0.71) P < 0.0001 0.95 (0.86-1.06) P ¼ 0.34 (0.90-1.12) P ¼ 0.89
AMI, acute myocardial infarction; CI, confidence interval; HR, hazard ratio. * Composite of mortality, AMI, and stroke. y Excluded patients with prior percutaneous coronary intervention or coronary artery bypass graft.
Clinical Research
Effectiveness of Statin Prescribing on Reducing Mortality in South Asian, Chinese, and White Patients With Diabetes Nathan W. Brunner, MD,a Krishnan Ramanathan, MD,a Hong Wang, MSc,b Hude Quan, PhD,c and Nadia A. Khan, MD, MScb,d a b
Division of Cardiology, University of British Columbia, Vancouver, British Columbia, Canada
Centre for Health Evaluation and Outcomes Sciences, St Paul’s Hospital, Vancouver, British Columbia, Canada c d
Department of Community Health Sciences, University of Calgary, Calgary, Alberta, Canada
Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
ABSTRACT
RESUM E
Background: Clinical trials have shown that 3-hydroxy-3-methylgutaryl coenzyme A reductase inhibitors (statins) reduce mortality in patients with diabetes. However, as these trials were conducted in largely white populations, it is unknown whether the benefits of statins can be extended to other ethnic populations in which the incidence of diabetes is rising sharply. We investigated associations between statin prescription and outcomes in a multiethnic population with diabetes. Methods: We identified all patients with newly diagnosed diabetes in British Columbia, Canada (1993-2006), using administrative data. Validated surname analysis was used to identify South Asian and Chinese patients. Statin prescribing was defined as any prescription filled within 1 year of diabetes diagnosis according to a provincial pharmacy database. Median length of follow-up was 4 years. Cox proportional hazards models were constructed for each ethnic group to determine the association of statin prescribing with time to death, adjusting for covariates including age, sex, socioeconomic status, and comorbid conditions. Results: There were 143,630 white, 9529 South Asian, and 14,084 Chinese persons with newly diagnosed diabetes. White patients were older and had more comorbidity than the other groups. Statin prescribing was associated with lower mortality compared with no prescribing within each ethnic group: South Asian (Hazard Ratio [HR], 0.69; 95% confidence interval [CI], 0.55-0.86; P ¼ 0.001), Chinese
montre que les inhibiteurs Introduction : Les essais cliniques ont de ductase de la HMG-CoA (3-hydroxy-3-methylglutaryl coenzyme A) re duisent la mortalite chez les patients ayant le diabète. (statines) re te mene s chez des populations en Cependant, comme ces essais ont e ne fices des statines peuvent grande partie blanches, on ignore si les be tendre à d’autres ethnies chez qui l’incidence de diabète est en s’e les associations entre l’ordonnance forte hausse. Nous avons examine sultats chez une population multiethnique ayant le de statines et les re diabète. thodes : Nous avons de termine tous les patients ayant nouvelleMe ment reçu un diagnostic de diabète en Colombie-Britannique, au es administratives. La Canada (1993-2006), en utilisant les donne te utilise e pour validation de l’analyse du nom de famille a e terminer les patients sud-asiatiques et chinois. L’ordonnance de de te de finie comme toute ordonnance remplie dans la 1re statines a e e du diagnostic du diabète selon une base de donne es pharmaanne e me diane du suivi a e te de 4 ans. Les ceutique provinciale. La dure te e labore s pour chamodèles de risques proportionnels de Cox ont e terminer le lien entre l’ordonnance cun des groupes ethniques pour de de statines et le moment de la mort, l’ajustement de covariables conomique et les facteurs de incluant l’âge, le sexe, le statut socioe . comorbidite sultats : Il avait 143 630 personnes blanches, 9 529 personnes Re
With increased urbanization and development, the prevalence of type 2 diabetes mellitus is rising to epidemic proportions worldwide. The greatest increase is occurring in Asia and the Indian subcontinent, where it is expected to affect > 130 million individuals in the next 2 decades.1 Recent research also demonstrates that the South Asian and Chinese diasporas
to the West similarly experience a high prevalence of type 2 diabetes.2,3 This is a critical global health issue, given the staggering health and economic costs associated with diabetes. Diabetes is associated with a 2- to 4-fold increase in risk of premature death, myocardial infarction, and stroke.4-6 In 2003, developing countries spent between 2.5% and 15% of their direct health care budgets on diabetes-related treatments.7 Substantial strides have been made in lowering the risk of death and cardiovascular outcomes in patients with diabetes. Clinical trials have shown that therapy with coenzyme A (CoA) reductase inhibitors (statins) results in a relative risk reduction
Received for publication July 27, 2012. Accepted October 6, 2012. Corresponding author: Dr Nadia Khan, 620 B, 1081 Burrard St, St Paul’s Hospital, Vancouver, British Columbia V6Z 1Y6, Canada. E-mail: [email protected] See page 924 for disclosure information.
0828-282X/$ - see front matter Ó 2013 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.cjca.2012.10.012
Brunner et al. Statins in Diabetes by Ethnicity
921
(HR, 0.60; 95% CI, 0.49-0.72; P < 0.0001), and white (HR, 0.65; 95% CI, 0.63-0.67; P < 0.0001). Conclusions: Statin prescribing is associated with lower mortality in white, Chinese, and South Asian patients with newly diagnosed diabetes, confirming the benefits of statins across these ethnic groups.
sud-asiatiques et 14 084 personnes chinoises ayant nouvellement taient plus âge s et reçu un diagnostic de diabète. Les patients blancs e s que ceux des autres groupes. L’ordonavaient plus de comorbidite te associe e à une plus faible mortalite comnance de statines a e parativement à la non-ordonnance au sein de chacun des groupes ethniques : les Sud-Asiatiques (rapport de risque [RR], 0,69; intervalle de confiance [IC] à 95 %, 0,55-0,86; P ¼ 0,001), les Chinois (RR, 0,60; IC à 95 %, 0,49-0,72; P < 0,0001) et les blancs (RR, 0,65; IC à 95 %, 0,63-0,67; P < 0,0001). e à une plus faible Conclusions : L’ordonnance de statines est associe chez les patients blancs, chinois et sud-asiatiques ayant mortalite nouvellement reçu un diagnostic de diabète, ce qui confirme les ne fices des statines au sein de ces groupes ethniques. be
of 9% to 27% for death, 22% to 37% for major coronary event, and 21% to 47% for stroke in primary and secondary prevention populations with diabetes.8-10 The majority of patients with type 2 diabetes mellitus are at high risk for vascular events, and control of low-density lipoprotein cholesterol to less than 2.0 mmol/L is indicated for this population.11 Although the incidence and mortality associated with diabetes are now in epidemic proportions in Asia, clinical trials demonstrating benefit with statin therapy were conducted in primarily white populations.9,10,12 Moreover, it is unlikely that these trials will be repeated in Asian populations. Although some studies found that statin therapy can effectively modulate levels of low-density lipoprotein cholesterol and high-density lipoprotein cholesterol in South Asian patients to a similar degree as in white populations,13,14 other studies documented increased plasma statin levels in Chinese and South Asian populations, compared with white populations.1,15,16 Thus, whether South Asian and Chinese patients also derive net benefits in terms of mortality and cardiovascular morbidity risk reductions from statin therapy is not known. Ethnic variations in response to statin therapy, if they exist, would have substantial implications for both primary and secondary prevention efforts in these increasingly prevalent populations. We sought to identify differences in mortality in a population cohort of South Asian, Chinese. and whitepatients with newly diagnosed diabetes as a function of statin prescribing. We also investigated risk of the composite of death, hospitalization for acute myocardial infarction (AMI), or stroke, as well as the individual components in these groups, as secondary end points for this analysis.
Classification of Diseases, Ninth and Tenth Revisions (ICD-9 and ICD-10). Physician claims files provided diagnoses and listings of services provided for all primary care clinic and hospital encounters for all residents of BC. Demographic data were obtained through linkage with the universal provincial health care insurance registry. We limited our analysis to persons with newly diagnosed diabetes in order to identify a population of patients at similar and early stages of disease. All newly diagnosed diabetes cases were identified with a diabetes coding algorithm, which was validated in a population of 3362 patients in Canada to have a sensitivity of 92.3% and a specificity of 96.9% for physiciandiagnosed diabetes.18 Patients with diabetes were defined as those with any of the following codes: ICD-9, Clinical Modification codes 250.x and ICD-10 codes E109, E119, E139, E149, E101, E111, E131, E141, E105, E115, E135, or E145 in at least 1 hospital discharge abstract or 2 physician claims within 2 years. In this manner, virtually all patients with a diagnosis of diabetes in the province of BC were identified. New cases were defined as patients with a minimum of 3 preceding years without coding for diabetes. Only those with valid provincial health insurance for at minimum of 3 years prior to their diagnosis were included, in order to avoid the misclassification of newly arrived immigrants with diabetes as new cases. To reduce the misclassification of type 2 diabetes as type 1 diabetes, exclusion criteria consisted of patients younger than 35 years. Patients with gestational diabetes, identified with coding for an obstetrical event within 5 months of diabetes diagnosis, were also excluded. To control for severity of illness at time of diabetes diagnosis, we measured clinical variables from the Charlson comorbidity index as this index is associated with mortality in patients with type 2 diabetes.19,20 For each patient, administrative codes for these variables were extracted from physician service claims and hospital discharge abstracts dating from 1993 until the date of diabetes diagnosis. Area level median income was determined from postal codes and the Canadian Census 2001 socioeconomic file.
Methods This study was a retrospective cohort study that used administrative data from the province of British Columbia (BC) Canada. Study population We identified all patients newly diagnosed with diabetes residing in the province of BC from hospital discharge abstracts and physician claims files collected from 1993 through 2006, of which the details are described elsewhere.17 Hospital discharge abstract data contain information on all hospital visits occurring in BC for all residents and include up to 25 diagnosis fields expressed according to the International
Identification of ethnicity Ethnicity is not recorded in administrative data in Canada. We categorized ethnicity using validated surname analysis algorithms as South Asian (from Pakistan, India, or Bangladesh) or Chinese (ancestry from China, Taiwan, or Hong Kong). Patient surnames, recorded in provincial
922
registries, were merged with Quan’s Chinese name list21 and the Nam Pehchan computer program22 to define Chinese and South Asian ethnicity. Compared with self-report, the sensitivity for Quan’s surname algorithm was 78%, specificity was 99.7%, and the positive predictive value (PPV) was 81%.21 Validation studies for the Nam Pehchan algorithm report a sensitivity of 90% to 94%, specificity of 99.4%, and PPV of 63% to 96%.23 In BC, blacks account for 0.7% of the population, and after excluding Chinese and South Asians, 93% of the remaining population is white.24 Thus, we categorized all noneSouth Asian and non-Chinese patients as white. Determination of 3-hydroxy-3-methylglutaryl CoA reductase inhibitor prescribing The BC PharmaNet database contains complete information on all prescription drug claims filled at community pharmacies throughout BC. Information is included on all patients, regardless of age or amount of copay. Prescription drug claims contain information on date of prescription, drug name and dosage, and days supplied of each medication. Accuracy in Canadian provincial drug databases is high, with a coding error rate of only 0.7% reported in an analogous database.25 Patients were taken as being prescribed statins if they filled a prescription for any 3-hydroxy-3-methylglutaryl CoA inhibitor within 1 year following their diagnosis of diabetes. In BC, prescription drug costs are paid fully by the patient up to an income-based deductible. For drug expenses exceeding the deductible, 70% of prescription drug costs are subsidized by the PharmaCare program. If patient drug expenditures exceed a certain percentage of their net income, additional prescriptions are fully covered. In order to facilitate comparison between different statins, statins were expressed in terms of an equivalent atorvastatin dose, in a manner similar to that employed by Gupta and colleagues.14 Ten milligrams of atorvastatin was taken to be equivalent to 5 mg of rosuvastatin, 40 mg of lovastatin, 40 mg of pravastatin, 40 mg of fluvastatin, and 0.8 mg of ceruvastatin, respectively. Clinical outcomes The primary outcome was time to death. We obtained date of death through linkage with BC vital statistics, a mandatory provincial database of mortality that is updated daily. Secondary end points included (1) time to composite risk of death or (2) hospitalization for acute stroke or AMI, whichever was experienced first, as well as the individual components of AMI and stroke. Hospitalization for acute stroke and AMI was determined by the ICD-9 and ICD-10 codes from hospital discharge abstracts. The PPV for AMI and stroke with these algorithms is 0.95 (95% CI, 0.91-0.98) and 0.85 to 0.98, respectively.26 Time to event was assessed up to March 31, 2006, with a maximum duration of follow-up of 9 years (median follow-up, 4.0 years). Patients were censored if they moved out of province or reached the end of the observation period. Statistical analysis Comparisons of baseline characteristics for each ethnic group were made using the c2 test for categorical variables and
Canadian Journal of Cardiology Volume 29 2013
analysis of variance for continuous variables. To determine the association between statin prescribing and time to death, we constructed Cox proportional hazards models for each ethnic group, comparing those prescribed and not prescribed statin within 1 year of diagnosis. We used the risk adjustment method to construct each model, adjusting for age in 5-year categories, income quintile, year at diagnosis, sex, and Charlson comorbidities including hypertension, AMI, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, renal insufficiency, and malignancy. Initial statin dosages were compared between ethnic groups with a multivariate model that included age, sex, and socioeconomic status. For the secondary end point of AMI, we excluded those with a previous AMI or coronary revascularization prior to their statin prescription in order to lesson confounding associated with the inclusion of patients who were started on statins for established coronary artery disease rather than diabetes alone. For all models, we additionally performed propensity score analysis with a greedy match algorithm.27 We calculated propensity scores for statin prescribing, using a nonparsimonious multivariate logistic regression model fitting age, sex, socioeconomic status, and baseline clinical variables in the model. We then used the propensity scores to obtain 1-to-1 matches between subjects who received a prescription for statin therapy and those who did not, using a caliper width of 0.2 standard deviation of the log odds of the propensity score, age ( 1 year), sex, and index date of diabetes diagnosis ( 1.2 year). Standard differences were used to compare the baseline characteristics between matched groups, and a standard difference of less than 10% was taken as good balance. Within matched pairs, we used a stratified Cox proportional hazards model to compare time to event for each ethnicity, using a restricted cubic spline method. Findings derived from propensity score models were similar to the main models (see Appendix I), and the main adjusted Cox regression models are presented herein. All P values presented are 2-tailed, and a value of less than 0.05 was considered significant. Analyses were performed with SAS version 9.2 (SAS Institute Inc, Cary, NC). The study was approved by the local ethics institutional review board. Results Our total cohort consisted of 167,243 patients with newly diagnosed diabetes. Of these, 9529 (5.7%) were identified as South Asian and 14,084 (8.4%) as Chinese. White patients were significantly older at diagnosis than were South Asian or Chinese patients (Table 1). White patients were more likely to be of higher socioeconomic status compared with South Asians and Chinese. There were significant differences in prevalence of comorbidities for each ethnic group. Myocardial infraction and congestive heart failure were more common among South Asians, whereas the prevalence of hypertension in South Asians was half that of whites and Chinese. Statins were prescribed to 25% of South Asians, 21% of Chinese, and 27% of white patients, respectively. The mean initial statin doses for each ethnicity, expressed as equivalent atorvastatin doses, are shown in Table 1. Compared with the white patients, the initial statin dose was significantly lower for Chinese patients (P < 0.001) but not for South Asian
Brunner et al. Statins in Diabetes by Ethnicity
923
Table 1. Baseline characteristics among diabetes patients aged 35 years and older in British Columbia Characteristics Age at diagnosis, n (%) Mean age, years (SD) 35-49 years 50-64 years 65-79 years 80 years Female, n (%) Socioeconomic quintile, n (%) First quintile (low) Second quintile Third quintile Fourth quintile Fifth quintile (high) Unknown Comorbidity (Charlson measurements), n (%) Hypertension Myocardial infarction Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic pulmonary disease Renal disease Cancer Outcomes, n (%) Death AMI Stroke Proportion of total receiving statin therapy Initial statin dose (atorvastatin equivalent)
South Asian (n ¼ 9,529)
Chinese (n ¼ 14,084)
White (n ¼ 143,630)
P value
56.1 (12.3) 3157 (33.1) 3890 (40.8) 2143 (22.5) 339(3.6) 6749 (47.9)
59.4 3585 5331 4275 893 4268
(12.7) (25.5) (37.9) (30.4) (6.3) (44.8)
61.6 28,177 55,815 46,574 13,064 6749
(13.0) (19.6) (38.9) (32.4) (9.1) (47.9)
<0.001 <0.001
2459 2673 1911 1278 838 370
(25.8) (28.1) (20.1) (13.4) (8.8) (3.9)
3756 3423 2422 2059 2047 377
(26.7) (24.3) (17.2) (14.6) (14.5) (2.7)
32,196 27,741 25,993 24,857 23,543 9300
(22.4) (19.3) (18.1) (17.3) (16.4) (6.5)
<0.001
2422 2059 2047 377 217 48 1225 101 215
(17.2) (14.6) (14.5) (2.7) (2.3) (0.5) (12.9) (1.1) (2.3)
5605 176 329 211 342 87 1032 230 456
(39.8) (1.3) (2.3) (1.5) (2.4) (0.6) (7.3) (1.6) (3.2)
55,866 5045 7898 3494 5260 2028 17,098 2624 8656
(38.9) (3.5) (5.5) (2.4) (3.7) (1.4) (11.9) (1.8) (6.0)
<0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
<0.001
551 (5.8) 185 (1.9) 129 (1.4) 25%
920 (6.5) 103 (0.7) 222 (1.6) 21%
19,997 (13.9) 3344 (2.2) 3225 (2.3) 27%
<0.001 <0.001 <0.001 <0.001
14.2 (9.0)
11.6 (7.5)
13.8 (9.8)
<0.001
AMI, acute myocardial infarction; SD, standard deviation.
patients (P ¼ 0.4), after adjusting for age, sex, and socioeconomic status. Clinical outcomes associated with statin prescribing During the observation period, there were 21,468 deaths, 3632 nonfatal AMIs, and 3576 stroke events among the study cohort. As shown in Table 1, the prevalence of death and stroke during the study period was higher in whites than in South Asians or Chinese. Rates of AMI were lowest in Chinese patients, compared with whites and South Asians. As
shown in Table 2, statin prescribing was associated with a significant decrease in mortality for all ethnic groups, even after we adjusted for baseline socio-demographic and clinical characteristics. Statin prescribing was also associated with a reduced risk of the composite end point of death, AMI, or stroke for Chinese and white patients. This was also seen in South Asians in our main analysis, although the difference became nonsignificant in the propensity score analysis. There were no significant differences in risk of hospitalization for AMI in those prescribed statins compared with those who
Table 2. Adjusted outcomes associated with statin prescribing for South Asian, Chinese, and white patients with newly diagnosed diabetes, as obtained from Cox proportional hazards models Outcome HR (95% CI), P value Total mortality Composite* outcome AMIy Stroke
South Asian
Chinese
White
0.69 (0.55-0.86) P ¼ 0.001 0.83 (0.70-0.99) P ¼ 0.04 1.16 (0.83-1.64) P ¼ 0.37 1.38 (0.94-2.04) P ¼ 0.1
0.60 (0.49-0.72) P < 0.0001 0.57 (0.48-0.68) P < 0.0001 1.07 (0.67-1.7) P ¼ 0.8 0.54 (0.38-0.79) P ¼ 0.001
0.65 (0.63-0.67) P < 0.0001 0.66 (0.64-0.68) P < 0.0001 0.93 (0.86-1.01) P ¼ 0.1 0.94 (0.86-1.01) P ¼ 0.1
All Cox regression models are adjusted for age, sex, income quintile, hypertension, congestive heart failure, peripheral vascular disease, dementia, chronic pulmonary disease, renal insufficiency, and malignancy. AMI, acute myocardial infarction; CI, confidence interval; HR, hazard ratio. * Composite of mortality, AMI, and stroke. y Excluded patients with prior percutaneous coronary intervention or coronary artery bypass graft.
924
were not. Among Chinese patients with diabetes, statin prescribing was associated with significantly reduced risk of hospitalization for acute stroke. However, there was no difference with risk of acute stroke in South Asian or white patients. Interpretation In this analysis of a large administrative database of persons newly diagnosed with diabetes, statin prescribing was associated with a reduction in mortality in South Asian, Chinese, and white populations. The magnitude of the reduction in mortality was generally similar for all 3 ethnicities, suggesting they derived similar benefits from statin therapy. South Asian, Chinese, and white patients had substantial differences in their baseline characteristics. South Asians were much more likely to have a history of prior MI compared with the other groups. This finding is consistent with other studies done in healthy multiethnic populations in Canada.2,28 Statin starting dosage was low for all ethnicities studied and was significantly lower for Chinese patients compared with South Asian and white patients. This seems consistent with the emergence of studies demonstrating ethnic differences in rosuvastatin metabolism in 2005, prompting a Health Advisory from Health Canada.15,29,30 However, many of our patients began statin therapy prior to 2005, suggesting that physicians were cautious about statin therapy in Chinese populations even before then. The mortality findings in our study mirror those of randomized trials carried out in primarily white populations. The Collaborative Atorvastatin Diabetes Study (CARDS) and the Heart Protection Study (HPS), which included predominately white patients, reported a 20% to 27% relative risk reduction in death for primary prevention patients with diabetes. Our large population-based study suggests that the mortality benefits of statins seen in white populations may extend to South Asian and Chinese populations as well. In our analysis of the secondary end points, fewer differences in outcomes were noted with statin prescribing. Although our study showed an association between statin prescribing and lower incidence of stroke in the Chinese population, this was not observed in white or South Asian patients. In white populations, the HPS and CARDS studies both demonstrated a reduction in stroke with statin therapy.10,13 There was also no association between statin prescribing and AMI in any ethnic groups. This lack of association between statin usage and stroke and AMI contrasts with randomized controlled trials and is puzzling. It may be partially a result of the lower event rates in our study. In our AMI analysis, all patients with prior AMI or coronary revascularization were excluded in order to avoid confounding due to statin prescription for prior cardiac event. Therefore, this was a purely primary prevention population, explaining the modest incidence of stroke and AMI during follow-up. Given that significant differences exist in the development and disease course of diabetes in various ethnicities, we limited our analysis to incident cases only. Although the exclusion of prior coronary revascularization in our AMI analysis lessened confounding by indication, it may also have introduced some selection bias by preferentially including patients with few prior medical contacts. Moreover,
Canadian Journal of Cardiology Volume 29 2013
excluding these patients lowered our statistical power, especially in the South Asian and Chinese groups. Also, it is possible that statins may have been prescribed preferentially to patients with higher cardiac risk not accounted for in our analysis. This would have attenuated our observed cardiovascular benefit of statins. The major strength of our study is it represents one of the largest retrospective studies to investigate associations between statin prescription and clinical end points in a “real-world” multiethnic population. In addition, we used propensity score matching to minimize the effects of confounding. However, there are several limitations to studies of this nature. As this was a retrospective observational study, we were not able to adjust for all potentially relevant covariates including A1c level, smoking, and other health behaviours. We could not evaluate resultant changes in lipid levels associated with statin prescribing. Stroke and nonfatal AMI occurring outside of the hospital setting would not be included in our results. In our study, ethnicity was determined by surname analysis. While the gold standard for the determination of ethnicity remains self-report,31 the algorithms used have been validated and correlate well with the gold standard. Finally, we used prescription claims data as a surrogate for medication ingestion. However, prescription claims data are shown to correlate with home inventory pill count as well as serum measures of drug presence.32-34 Conclusions Our study provides new evidence that statin prescribing is associated with lower mortality in South Asian, Chinese, and white patients with newly diagnosed diabetes. These results should reassure prescribers that the mortality benefits of statin observed in landmark clinical trials in diabetes can likely be generalized to South Asian and Chinese populations with diabetes. However, further study evaluating effectiveness of statin prescribing on halting the progression of cardiac atherosclerosis and stroke are greatly needed in these ethnic groups. Funding Sources This study was funded by the Canadian Institutes of Health Research (CIHR). N.A.K. received support from CIHR New Investigator Award and the Michael Smith Foundation for Health Research Career Scientist award. H.Q. receives funding from the Alberta Heritage Foundation for Medical Research Senior Population. Disclosures The authors have no conflicts of interest to disclose. References 1. Ramaraj R, Chellappa P. Cardiovascular risk in South Asians. Postgrad Med J 2008;84:518-23. 2. Anand SS, Yusuf S, Vuksan V, et al. Differences in risk factors, atherosclerosis, and cardiovascular disease between ethnic groups in Canada: the Study of Health Assessment and Risk in Ethnic groups (SHARE). Lancet 2000;356:279-84.
Brunner et al. Statins in Diabetes by Ethnicity 3. Rajpathak SN, Gupta LS, Waddell EN, et al. Elevated risk of type 2 diabetes and metabolic syndrome among Asians and South Asians: results from the 2004 New York City HANES. Ethn Dis 2010;20: 225-30. 4. Nichols GA, Gullion CM, Koro CE, Ephross SA, Brown JB. The incidence of congestive heart failure in type 2 diabetes: an update. Diabetes Care 2004;27:1879-84. 5. Lee WL, Cheung AM, Cape D, Zinman B. Impact of diabetes on coronary artery disease in women and men: a meta-analysis of prospective studies. Diabetes Care 2000;23:962-8. 6. Kissela BM, Khoury J, Kleindorfer D, et al. Epidemiology of ischemic stroke in patients with diabetes: the greater Cincinnati/Northern Kentucky Stroke Study. Diabetes Care 2005;28:355-9. 7. Narayan KMV, Zhang P, Kanaya AM, et al. Diabetes: the pandemic and potential solutions. In: Jamison DT, Breman JG, Measham AR, et al., eds. Disease Control Priorities in Developing Countries. 2nd ed. Washington, DC: World Bank, 2006:591. 8. Cholesterol Treatment Trialists’ (CTT) Collaborators, Kearney PM, Blackwell L, Collins R, et al. Efficacy of cholesterol-lowering therapy in 18,686 people with diabetes in 14 randomised trials of statins: a metaanalysis. Lancet 2008;371:117-25. 9. Colhoun HM, Betteridge DJ, Durrington PN, et al.; CARDS investigators. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo-controlled trial. Lancet 2004;364: 685-96. 10. Sacks FM, Tonkin AM, Craven T, et al. Coronary heart disease in patients with low LDL-cholesterol: benefit of pravastatin in diabetics and enhanced role for HDL-cholesterol and triglycerides as risk factors. Circulation 2002;105:1424-8. 11. Canadian Diabetes Association Clinical Practice Guidelines Expert Committee. Canadian Diabetes Association 2008 clinical practice guidelines for the prevention and management of diabetes in Canada. Can J Diabetes 2008;32(suppl 1):S1-S201. 12. Collins R, Armitage J, Parish S, Sleigh P, Peto R; Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterollowering with simvastatin in 5963 people with diabetes: a randomised placebo-controlled trial. Lancet 2003;361:2005-16. 13. Patel JV, Gupta S, Lie F, Hughes EA. Efficacy and safety of atorvastatin in South Asian patients with dyslipidemia: an open label noncomparative pilot study. Vasc Health Risk Manag 2005;1:351-6. 14. Gupta M, Braga MF, Teoh H, Tsigoulis M, Verma S. Statin effects on LDL and HDL cholesterol in South Asian and white populations. J Clin Pharmacol 2009;49:831-7. 15. Lee E, Ryan S, Birmingham B, et al. Rosuvastatin pharmacokinetics and pharmacogenetics in white and Asian subjects residing in the same environment. Clin Pharmacol Ther 2005;78:330-41. 16. Tzeng TB, Schneck DW, Birmingham BK, et al. Population pharmacokinetics of rosuvastatin: implications of renal impairment, race, and dyslipidaemia. Curr Med Res Opin 2008;24:2575-85. 17. Khan NA, Wang H, Anand S, et al. Ethnicity and sex affect diabetes incidence and outcomes. Diabetes Care 2011;34:96-101.
925
18. Chen G, Khan N, Walker R, Quan H. Validating ICD coding algorithms for diabetes mellitus from administrative data. Diabetes Res Clin Pract 2010;89:189-95. 19. Monami M, Lambertucci L, Lamanna C, et al. Are comorbidity indices useful in predicting all-cause mortality in Type 2 diabetic patients? comparison between Charlson index and disease count. Aging Clin Exp Res 2007;19:492-6. 20. Tao LS, MacKenzie CR, Charlson ME. Predictors of postoperative complications in the patient with diabetes mellitus. J Diabetes Complications 2008;22:24-8. 21. Quan H, Wang F, Schopflocher D, et al. Development and validation of a surname list to define Chinese ethnicity. Med Care 2006;44:328-33. 22. Macfarlane GJ, Lunt M, Palmer B, Afzal C, Silman AJ, Esmail A. Determining aspects of ethnicity amongst persons of South Asian origin: the use of a surname-classification programme (Nam Pehchan). Public Health 2007;121:231-6. 23. Cummins C, Winter H, Cheng KK, Maric R, Silcocks P, Varghese C. An assessment of the Nam Pehchan computer program for the identification of names of South Asian ethnic origin. J Public Health Med 1999;21: 401-6. 24. Statistics Canada. 2001 Census data: British Columbia. Available at: http://www12.statcan.ca/english/census01/products/standard/prprofile/ prprofile.cfm?G=59. Accessed December 6, 2012. 25. Levy AR, O’Brien BJ, Sellors C, Grootendorst P, Willison D. Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 2003;10:67-71. 26. Tu JV, Austin PC, Naylor CD, Iron K, Zhang H. Acute myocardial infarction outcomes in Ontario. In: Naylor CD, Slaughter PM, eds. Cardiovascular Health and Services in Ontario: An ICES Atlas. Toronto, Canada: Institute for Clinical Evaluative Studies, 1999:83-110. 27. Austin PC. The performance of different propensity-score methods for estimating differences in proportions (risk differences or absolute risk reductions) in observational studies. Stat Med 2010;29: 2137-48. 28. Chiu M, Austin PC, Manuel DG, Tu JV. Comparison of cardiovascular risk profiles among ethnic groups using population health surveys between 1996 and 2007. CMAJ 2010;182:E301-10. 29. Liao JK. Safety and efficacy of statins in Asians. Am J Cardiol 2007;99: 410-4. 30. Health Canada. Health Canada advises patients about the risk of CrestorÒ (rosuvastatin). Available at: http://www.hc-sc.gc.ca/ahc-asc/ media/advisories-avis/_2005/2005_10-eng.php. Accessed October 2, 2012. 31. Mays VM, Ponce NA, Washington DL, Cochran SD. Classification of race and ethnicity: implications for public health. Annu Rev Public Health 2003;24:83-110. 32. Lau HS, de Boer A, Beuning KS, Porsius A. Validation of pharmacy records in drug exposure assessment. J Clin Epidemiol 1997;50:619-25. 33. Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol 1997;50:105-16. 34. West SL, Savitz DA, Koch G, Strom BL, Guess HA, Hartzema A. Recall accuracy for prescription medications: self-report compared with database information. Am J Epidemiol 1995;142:1103-12.
926
Canadian Journal of Cardiology Volume 29 2013
Appendix I. Adjusted outcomes associated with statin prescribing for South Asian, Chinese, and White Patients with newly diagnosed diabetes, as obtained from propensity score analysis Outcome HR (95% CI), P value Total mortality Composite* outcome AMIy Stroke
South Asian
Chinese
White
0.70 (0.50-0.97) P ¼ 0.03 0.89 (0.69-1.14) P ¼ 0.35 1.01 (0.64-1.59) P ¼ 0.98 1.29 (0.74-2.26) P ¼ 0.37
0.58 (0.44-0.77) P ¼ 0.0002 0.60 (0.47-0.77) P < 0.0001 0.95 (0.49-1.84) P ¼ 0.87 0.53 (0.32-0.87) P ¼ 0.012
0.62 (0.59-0.66) P < 0.0001 0.67 (0.64-0.71) P < 0.0001 0.95 (0.86-1.06) P ¼ 0.34 (0.90-1.12) P ¼ 0.89
AMI, acute myocardial infarction; CI, confidence interval; HR, hazard ratio. * Composite of mortality, AMI, and stroke. y Excluded patients with prior percutaneous coronary intervention or coronary artery bypass graft.