Hospitalized Stroke Surveillance in the Community of Durango, Mexico
The Brain Attack Surveillance in Durango Study
Background and Purpose— Vascular conditions are becoming the greatest cause of morbidity and mortality in developing countries. Few studies exist in Latin America. We aimed to perform a rigorous stroke surveillance study in Durango, Mexico.
Methods— Active and passive surveillance were used to identify all patients with potential stroke presenting to Durango Municipality hospitals from August 2007 to July 2008. Exclusion criteria were subjects younger than 25 years old, stroke attributable to head trauma, and non-Durango Municipality residents. Brain Attack Surveillance in Durango-trained neurologists validated cases as stroke using source documentation. Stroke hospitalization rates were defined to include patients examined in the emergency department or admitted to the hospital.
Results— Abstractors identified 435 potential cases; 309 (71%) were validated as stroke. Of the validated stroke cases, the median age was 71 and 49% were female. Subtypes were 61.5% ischemic stroke, 20.7% intracerebral hemorrhage, 7.4% subarachnoid hemorrhage, and 10.4% undetermined. Overall initial NIHSS was a median of 11 (interquartile range, 7–17); in-hospital mortality was 39%. When adjusted to the world population, the age-adjusted hospitalization rate of first-ever stroke was 118.2 per 100 000; rates by type were: ischemic stroke, 69.1 (95% CI, 57.5–80.7); intracerebral hemorrhage, 26.7 (95% CI, 19.6–33.8); subarachnoid hemorrhage, 9.5 (95% CI, 5.3–13.8); and unknown, 12.3 (95% CI, 7.4–17.3). Of 190 patients with validated ischemic stroke, 44.2% received lipid testing and 7.4% received carotid imaging and echocardiography; 1.1% received tissue plasminogen activator.
Conclusion— To our knowledge, this is the first estimate of stroke hospitalization rates in a Mexican community and it provides information important for design of interventions to prevent and treat stroke. This information is critical to reduce Mexico’s stroke burden.
In high-income countries, stroke has long-been recognized as a main cause of death and disability. The burden of stroke on developing countries has more recently been highlighted. The Global Burden of Disease study showed that in 2005, 87% of worldwide stroke deaths were in low- and middle-income countries.1 In Mexico, stroke has become a major public health problem with the prevalence of chronic, nontransmissible diseases, such as hypertension, diabetes, and obesity, experiencing exponential growth and increasing above the prevalence of infectious diseases in adults.2,3 The socioeconomic impact of this “epidemiological transition” threatens to overwhelm health systems by giving rise to devastating consequences: cardiovascular disease and stroke. Stroke is now the third leading cause of death in those aged 65 years and older (381.3/100 000).4
Before planning and monitoring effective public health interventions targeted against this epidemiological transition, high-quality surveillance systems are needed. Despite stroke being a major health problem in Latin America,5 there are only a few such epidemiological stroke studies in the region: 2 in Brazil6,7 and 1 in a Chilean Hispanic Mestizo population (a racial admixture of Spaniards and Amerindians).8 Mexico has one of the largest Hispanic Mestizo populations. In 2008, the population was 106 680 000 people, of which 90.3% were Hispanic Mestizo; the remaining 9.7% comprise mainly different indigenous ethnic groups. The Brain Attack Surveillance in Durango (BASID) study was established to build the vital infrastructure necessary to develop a surveillance system for cerebrovascular diseases in Mexico.
Subjects and Methods
The methods used in BASID’s parent project, the Brain Attack in Corpus Christi (BASIC), which takes place in Corpus Christi, Texas, were adapted to meet requirements of Durango, Mexico.9,10 BASID took place in Durango Municipality, which is located in northern Mexico (Figure) and encompasses 1 large city and 112 small towns and villages. Total population was 527 157 in 2005 (target population aged 25 years or older: 247 665 habitants), with ≈90% of the population residing within Durango City, the capital of Durango State. The population mainly belongs to the Hispanic Mestizo ethnic group; indigenous ethnic groups are uncommon in this community. Medical care in Durango is highly self-contained, with the nearest urban centers >200 km away from Durango City. There are 3 large, acute care, public hospitals and 12 small, private hospitals whose combined catchment areas encompass Durango residents and the sparse populations of nearby municipalities. Therefore, the likelihood of referral outside Durango for first medical contact for acute stroke is extremely low. The Durango public hospitals have ≈50 adult beds each. CT is available 24 hours per day at all 3 public hospitals, with MRI at the largest public hospital. The small, private hospitals in Durango mainly provide obstetric care (all have 10–20 adult beds); 3 have CT and 1 has MRI. There is only 1 nursing home and no CT/MRI scanners outside the hospitals. The Durango medical community includes 169 general physicians, 25 internists, 12 cardiologists, 8 neurologists, and 8 neurosurgeons.
Medical Insurance and the Hospital System in Mexico
The public health care system in Mexico is complex and provides services for insured and uninsured populations. The insured care system includes 2 institutions that cover ≈45% to 50% of the Mexican population: the Mexican Institute of Social Security serves employees working in the formal private sector (≈40% of the Mexican population) and the Instituto de Seguridad y Servicios Sociales de los Trabajadores del Estado (ISSSTE) (Health and Social Security Institute for State Workers) covers government workers (≈10% of Mexicans). The other 45% to 50% of Mexicans have informal employment or are self-employed and uninsured. Most of them (≈50 million people) have to pay for prescription drugs and their health care services, including hospitalizations. These patients receive care at the third large pubic hospital in Durango, the General Hospital. Private hospitals give medical care to ≈10% of the population.
During the first year of the project, begun in September 2006, we exhaustively met with hospital officials to determine the ideal way to ascertain cases. We assessed the quality of medical record documentation and the means by which cases are tracked within each hospital. Computerized case report forms were developed and pilot-tested.
The BASID protocol involved formalized data collection procedures. Abstractors underwent standardized training before starting the hospital surveillance phase. They were trained to review medical records and abstract data into an electronic case report form. All data collection procedures were outlined in a procedure manual in which standard definitions for variables were clearly described to ensure consistent data collection.
Ascertainment of Hospital Cases
Case identification of potential stroke cases was accomplished by using both active (“hot pursuit”) and passive (“cold pursuit”) surveillance at all Durango Municipality hospitals. The World Health Organization suggests the combination of active and passive surveillance to ensure the most complete identification of stroke cases.11 BASIC investigators also found the combination to be superior to either method alone for complete case ascertainment.10 Exclusion criteria were subjects younger than age 25 years, individuals who had a stroke as a result of head trauma, and non-Durango Municipality residents.
Emergency departments, admission logs, hospital wards, and intensive care units were reviewed every weekday by study abstractors, with weekend logs reviewed on Mondays. Logs were combed for any symptoms or diagnostic terms previously validated as predictors of acute stroke in BASIC.10 Abstractors reviewed the medical record of any patient presenting with one of the screening terms. All data were entered directly into an on-site laptop-based database. Abstractors then de-identified the record, digitalized the source documents, and sent them to the field office for validation.
Lists of discharges with Stroke International Classification of Diseases, Tenth Revision (ICD-10) codes 160 to 169 (cerebrovascular disease) were obtained monthly from each hospital and from municipality death certificates. Any newly identified cases were reviewed and abstracted as described.
Stroke Validation Process
All potential stroke cases were reviewed by a BASID neurologist (A.R., G.Q., J.A.) who was trained to validate stroke cases according to study standards using all available source documentation. Ten percent of the validated charts were revalidated by the principal investigator in Mexico (C.C.B.).
Strokes were defined based on published criteria12 as an acute onset of focal neurological deficit attributable to a cerebrovascular distribution, persisting for >24 hours (except in the case of sudden death), and not attributable to another disease process. Based on the source documents, strokes were further classified as ischemic, intracerebral hemorrhage, subarachnoid hemorrhage, or unknown type. The classifications scheme required for ischemic stroke were: infarct on imaging (or a normal head CT result in the case of early imaging); for intracerebral hemorrhage, parenchymal blood on imaging or at autopsy (not attributable to hemorrhagic conversion of ischemic infarction); and for subarachnoid hemorrhage, sudden onset of severe headache or altered consciousness (with or without focal neurological deficits) associated with evidence of subarachnoid blood on imaging, cerebrospinal fluid studies, or autopsy. A stroke was classified as unknown type if it met clinical definition for stroke but imaging or autopsy was not performed to identify hemorrhage or ischemia.
The BASID project was approved by the Institutional Review Board of the University of Michigan and the National Institute of Medical Sciences and Nutrition Salvador Zubiran in Mexico, as well as the 3 public hospitals in Durango. All study personnel, including physician validators and study abstractors, completed United States NIH Human Subjects Protection Training translated to Spanish.
Descriptive statistics were calculated for all study variables in patients with validated stroke. Stroke hospitalization rate was defined to include patients with validated first-ever stroke who presented to the emergency department and were either admitted or sent home and those who were directly admitted to the hospital. TIA was excluded. Cumulative hospitalization rates of stroke were calculated overall, by stroke type, by gender, and by age group. Numerators were calculated from the case surveillance; denominators were calculated from the 2005 Mexican Census with population counts for those aged 25 or older. Rates were standardized to the population in Mexico (2005 Mexican Census counts), the United States (2000 United States Census population counts), and Segi13 and WHO populations14 using direct standardization.15 We used classical direct standardization formulas: sum (wk * Ik)/Sum(wk), where wkrefers to a weight assigned to each 10-year age group in the reference population and Ik refers to the observed rate in Durango within the age group. To perform a cross-border comparison with the BASIC data, we also calculated first-ever and recurrent stroke hospitalization rates in Durango within the same age categories published for BASIC data16 (45–59, 60–74, and 75 years or older) and compared these to annual BASIC rates (published rate16 divided by 3 to convert to annual rate). Ninety-five percent confidence intervals were calculated for all measures of incidence.
Data on evaluation and management of first-ever and recurrent ischemic strokes were summarized with frequencies and percents. Clinical outcome, as measured at hospital discharge, was summarized overall and by stroke type using medians and interquartile ranges.
From August 1, 2007 to July 31, 2008, 435 patients with signs possibly attributable to stroke were identified. Of these, 309 were validated as stroke. Of these, 89.5% underwent neuroimaging. There were 238 first-ever strokes and 71 recurrent strokes with the following stroke subtypes: 190 ischemic strokes (61.5%), 64 intracerebral hemorrhages (20.7%), 23 subarachnoid hemorrhages (7.4%), and 32 unknown type (10.4%). Among patients with validated stroke, there were 153 women (49.5%); median age was 71 years (interquartile range, 58–82 years). The majority were from the city of Durango (88.4%). Most patients (92.2%) presented to public hospitals.
The prevalence of vascular risk factors among first-ever or recurrent strokes was: arterial hypertension (70.1%), diabetes mellitus (38.9%), excessive alcohol use (16.1%), current smoker (16.4%), hypercholesterolemia (10.7%), coronary artery disease (9.8%), atrial fibrillation (11.2%), and first-degree family history of stroke (10.7%). Previous stroke or TIA was reported in 25.9%.
The crude annual hospitalization rate of any stroke (first-ever or recurrent) was 143 per 100 000 population (95% CI, 128–158). Table 1 describes the annual first-ever stroke hospitalization rates by age groups. Table 2 describes the annual hospitalization rates by stroke type. The crude annual hospitalization rate of first-ever stroke was 96.1 per 100 000 (95% CI, 84.6–109.1) in Durango and 97.3 per 100 000 (95% CI, 84.6, 109.7) when adjusted to the Mexican population. The age-adjusted hospitalization rates of first-ever stroke were 159.9 per 100 000 and 118.2 per 100 000 when adjusted to the United States and world populations, respectively. There was no statistical difference in rates by gender.
Table 3 describes the evaluation and management of the 190 patients with validated ischemic strokes. Only 7.4% received any form of carotid imaging, and echocardiography was rare. Lipid profile was conducted in nearly half of patients (44.2%). Only 1.1% received intravenous tissue plasminogen activator. Among ischemic stroke patients who survived to discharge, prestroke use of antithrombotics or anticoagulants was 21.7% and increased to 78.2% at discharge. Use of lipid-lowering agents was low before stroke but doubled during hospitalization from 12.4% to 27.9%.
Table 4 shows that stroke severity on admission as measured by NIHSS was a median of 11 points (interquartile range, 7–17). The length of stay was a median of 6 days (interquartile range, 3–12). Overall in-hospital case fatality rate was 39% and was as high as 72% in stroke of unknown type; it was 52% in subarachnoid hemorrhage.
Table 5 compares the age- and ethnic-specific annual stroke hospitalization rates per 100 000 for total completed strokes (first-ever and recurrent completed ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage) that occurred in non-Hispanic whites and Mexican Americans in Corpus Christi, Texas (BASIC study)17 and those in Mexican Mestizos in Durango (BASID study, current report). At all ages, stroke hospitalization rates were higher in Mexican Americans and non-Hispanic whites living in the United States than in Mexicans living in Mexico.
This study provides a broad framework to address the quantity of stroke and quality of stroke treatment and secondary prevention efforts in Mexico. Despite an epidemic of stroke in developing countries, the paucity of stroke epidemiology data from Latin America hinders efforts to reverse this troubling trend.5
The BASID stroke hospitalization rate of 96.1 per 100 000 is similar to the in- and out-of-hospital stroke incidences reported in recent population-based studies in developing countries, with the following rates per 100 000 and hospital admission rates: 86 in Iquique, Chile (71%);8 103 in Tbilisi, Georgia (66%);18 105 in Joinville, Brazil (84%);7 130 in Matao, Brazil (97%);6 and 151 in Mumbai, India (67%).19 A systematic review of population-based studies from 1970 to 2008 showed a 42% decrease in stroke incidence in high-income countries and >100% increase in stroke incidence in low- to middle-income countries over the past 4 decades.20
Although the BASID hospitalization rate in Mexico is consistent with other developing countries, Mexican Americans living in the United States had a 2-fold higher stroke hospitalization rate compared with Mexicans living in Mexico (Table 4). This means that although overall stroke hospitalization rates in low- to middle-income countries are converging with those seen in high-income countries,20 there are places such as our cross-border comparison in which stroke hospitalization is still higher in the more developed country.
There are 2 main possibilities to explain the discrepancy between stroke hospital presentation rate in Durango and in Corpus Christi. First, despite the same Mexican ancestry, Mexicans living in United States have a higher likelihood of stroke developing than do Mexicans living in Mexico because of changes in environment or lifestyle that alter stroke susceptibility. This has been seen in epidemiological studies of coronary heart disease in Japanese men living in Japan, Hawaii, and California, where a striking increase in the incidence of myocardial infarction occurred in Japanese immigrants to the United States.21,22 In this regard, Durango serves as a valuable comparison community to Corpus Christi, Texas, by providing continued monitoring of international trends and identifying stroke-related genetic, other biological, and social factors in those of Mexican ancestry on both sides of the border. The second possibility to explain the excess of stroke hospitalization rates in Mexican Americans is that a high proportion of stroke patients living in Mexico do not seek medical care in hospitals or their emergency departments, and receive medical care only as outpatients, or do not receive care at all. In the BASIC study, out-of-hospital strokes accounted for 5.6% of all first-ever and recurrent ischemic strokes (5.3% in Mexican Americans; 6.0% in non-Hispanic whites).17 Community-based surveillance is the only way to determine the proportion of subjects missed by limiting surveillance in Durango to hospitals and their emergency departments. Door-to-door surveillance is 1 community-based method to rigorously ascertain missed cases; the WHO has outlined other possible methods that utilize primary care facilities11 to accomplish community-based surveillance.
This work has demonstrated clear opportunities for quality improvement in stroke care in Durango. Whereas it is well-recognized in guidelines that determination of stroke mechanism is critical for secondary stroke prevention,23 carotid artery imaging and echocardiography are performed rarely in Durango. Development and implementation of specific pathways of care followed by performance measures may increase stroke care quality.
Limitations of this study include that it is hospital-based and may have missed patients with stroke that did not seek inpatient care, as discussed. BASID screening terms were adopted from BASIC and have not been validated in Spanish. Although 25 cases of first-ever stroke (10.5%) had no neuroimaging and thus were not able to be typed, this rate easily meets the Feigin criteria24 of 80% neuroimaging rates for accurate diagnosis of stroke types in stroke surveillance studies. However, the paucity of additional diagnostic tests, such as vessel and cardiac imaging, meant that further subtyping of the ischemic strokes was not possible.
In conclusion, we found that hospital surveillance is feasible despite the differences in the documentation and methods of ascertaining stroke cases in hospitals of the various health systems in Mexico. BASID is now poised through door-to-door surveillance to study the proportion of patients with stroke who do not present for medical attention and then to identify specific targets for public health interventions to reduce the stroke burden in Mexico.
The authors are especially grateful to our colleagues and the staff of the hospitals of Durango, Mexico.
Sources of Funding
This project was funded by National Institutes of Health R03 NS059056 and R01 NS38916.
- Received December 30, 2009.
- Accepted January 26, 2010.
Velazquez-Monroy O, Rosas Peralta M, Lara Esqueda A, Pastelin Hernandez G, Sanchez-Castillo C, Attie F, Tapia Conyer R. [Prevalence and interrelations of noncommunicable chronic diseases and cardiovascular risk factors in mexico. Final outcomes from the national health survey 2000]. Arch Cardiol Mex. 2003; 73: 62–77.
Jimenez-Cruz A, Bacardi-Gascon M. The fattening burden of type 2 diabetes on Mexicans: Projections from early growth to adulthood. Diabetes Care. 2004; 27: 1213–1215.
Minelli C, Fen LF, Minelli DP. Stroke incidence, prognosis, 30-day, and 1-year case fatality rates in Mateo, Brazil: A population-based prospective study. Stroke. 2007; 38: 2906–2911.
Cabral NL, Goncalves AR, Longo AL, Moro CH, Costa G, Amaral CH, Fonseca LA, Eluf-Neto J. Incidence of stroke subtypes, prognosis and prevalence of risk factors in Joinville, Brazil: A 2 year community based study. J Neurol Neurosurg Psychiatry. 2009; 80: 755–761.
Piriyawat P, Smajsova M, Smith MA, Pallegar S, Al-Wabil A, Garcia NM, Risser JM, Moye LA, Morgenstern LB. Comparison of active and passive surveillance for cerebrovascular disease: The brain attack surveillance in Corpus Christi (BASIC) project. Am J Epidemiol. 2002; 156: 1062–1069.
Stepwise approach to surveillance (steps).World Health Organization. Available at: http://www.WHO.Int/chp/steps/en/. Accessed December 16, 2009.
Hatano S. Experience of a multicenter stroke register: A preliminary report. Bull Word Health Org. 1976; 54: 541–553.
Segi M. Cancer mortality for selected sites in 24 countries (1950–57). Sendai, Japan: Tohoku University of Medicine; 1960.
Ahmed OB, Pinto CB, Lopez AD, Murray CJL, Lozano R, Inoue M. Age standardization of rates: A new WHO standard; GPE discussion paper series, no. 31. EIP/GPE/EVD Geneva: World Health Organization; 2000.
Rothman K, Greenland S. Modern epidemiology, II ed. Philadelphia: Lippincott-Raven; 1998.
Morgenstern LB, Smith Mexican American, Lisabeth LD, Risser JM, Uchino K, Garcia N, Longwell PJ, McFarling DA, Akuwumi O, Al-Wabil A, Al-Senani F, Brown DL, Moye LA. Excess stroke in Mexican Americans compared with non-Hispanic whites: The brain attack surveillance in Corpus Christi project. Am J Epidemiol. 2004; 160: 376–383.
Tsiskaridze A, Djibuti M, van Melle G, Lomidze G, Apridonidze S, Gauarashvili I, Piechowski-Jozwiak B, Shakarishvili R, Bogousslavsky J. Stroke incidence and 30-day case-fatality in a suburb of Tbilisi: Results of the first prospective population-based study in Georgia. Stroke. 2004; 35: 2523–2528.
Worth RM, Kato H, Rhoads GG, Kagan K, Syme SL. Epidemiologic studies of coronary heart disease and stroke in japanese men living in Japan, Hawaii and California: Mortality. Am J Epidemiol. 1975; 102: 481–490.
Robertson TL, Kato H, Rhoads GG, Kagan A, Marmot M, Syme SL, Gordon T, Worth RM, Belsky JL, Dock DS, Miyanishi M, Kawamoto S. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California. Incidence of myocardial infarction and death from coronary heart disease. Am J Cardiol. 1977; 39: 239–243.
Sacco RL, Adams R, Albers G, Alberts MJ, Benavente O, Furie K, Goldstein LB, Gorelick P, Halperin J, Harbaugh R, Johnston SC, Katzan I, Kelly-Hayes M, Kenton EJ, Marks M, Schwamm LH, Tomsick T. Guidelines for prevention of stroke in patients with ischemic stroke or transient ischemic attack: A statement for healthcare professionals from the American heart association/American stroke association council on stroke: Co-sponsored by the council on cardiovascular radiology and intervention: The American academy of neurology affirms the value of this guideline. Stroke. 2006; 37: 577–617.
Feigin VL, Carter K. Editorial comment–stroke incidence studies one step closer to the elusive gold standard? Stroke. 2004; 35: 2045–2047.