Gender Differences in the Colorado Stroke Registry
Background and Purpose— Gender differences in stroke are matters of increasing interest. The American Stroke Association’s patient management tool, Get with the Guidelines–Stroke (GGS) is widely used to increase adherence to quality indicators in stroke care, but it also provides an opportunity to analyze gender differences in the acute stroke setting.
Methods— We used a state-wide database, based on GGS, to explore gender differences in stroke in Colorado. We analyze demographics, risk factors, lifestyles, treatments, and responses to treatment.
Results— Of 126 data elements examined, statistically significant gender differences were noted for 47 (37%). As compared to men, women in Colorado were older and more significantly impacted by acute stroke. Risk factor profiles differed between the 2 genders, with men having a higher incidence of coronary artery disease, dyslipidemia, diabetes, carotid stenosis and tobacco smoking, while women had a higher incidence of atrial fibrillation and hypertension. Lipids were less aggressively treated and antithrombotics were less commonly used in women. Overall, acute stroke treatment of women appeared “less aggressive” than for men.
Conclusions— GGS may be used not only for quality improvement initiatives in individual hospitals. It can also give an overview of clinical aspects of stroke at a state level, and may shed light on gender differences.
Interest in gender differences for stroke is growing partly as a result of the American Heart Association’s Go Red for Women campaign. The Colorado Stroke Alliance (CSA) is a grassroots nonprofit consortium, consisting of 36 urban and rural hospitals, formed with the aim of improving stroke care in Colorado.1 Central to the CSA’s efforts is the Colorado Stroke Registry (CSR), wherein clinically relevant data are prospectively collected, using Get with the Guidelines–Stroke (GGS).2 A “superuser account” with Outcome Science (the tool’s contracted vendor) allows CSA to analyze data for all participating hospitals.
Our database affords a unique opportunity to compare and contrast stroke in women and men, as they are managed in Colorado hospitals. We have access to patient-level clinical data, drawn from the acute care setting. These data pertain to issues of age, risk factors, lifestyles, treatments, and responses to treatment. Because the CSR captures >80% of the strokes in Colorado, we believe our data fairly represent the Colorado experience with acute stroke care.
Subjects and Methods
Our sample size was 6690 (male 3028, 45.3%; female 3662, 54.7%). Of these, 3530 (52.9%) had ischemic stroke; 1814 (27.2%) had transient ischemic attack; 266 (4%) had subarachnoid hemorrhage; 786 (11.8%) had intracerebral hemorrhage; 74 (1.1%) were not stroke, but stroke mimics; 220 (3.3%) did not have stroke type specified. We explored gender differences in 126 data elements, without regard to stroke-type classification.
The patients were admitted to the hospital between July 2006 and September 2008. Statistical analysis was performed using JMP 7.0.2 (©2007 SAS Institute, Inc). For continuous variables, men and women were compared using Students t test. For nominal variables, the χ2 test was used. Because our analysis was exploratory and hypothesis-generating, we did not adjust for multiple comparisons. Data are displayed in tabular form, sorted according to probability values, and grouped into clinically oriented sets. Note that because data points were not available for every element in every patient we chose to present our summary of nominal data as #Yes/#No (%Yes) and continuous data as Mean (95% CI).
As Table 1 shows, women were more likely to have Medicare or private insurance than were men. Women were less likely to have a stroke consult or to be admitted by a neurologist. Women and men differed in stroke-related risk factors and in current medications at the time of admission (Table 2). Women were more likely to have had atrial fibrillation and hypertension, but less likely to have had coronary artery disease, a history of dyslipidemia, diabetes, and carotid stenosis. They were also less likely to have been tobacco smokers within the past year. A noteworthy item is that 28% of both genders had a history of prior stroke or transient ischemic attack. In agreement with the medical history, women were more likely to have been on antihypertensive medications, but less likely to have been on treatment for dyslipidemia or diabetes.
Expressed as (mean, 95% CI), women were older than men (71.5, 71.0 to 72.0 versus 67.5, 66.9 to 68.0 years, P<0.001). They also had higher levels of total cholesterol (4.50, 4.45 to 4.56 versus 4.23, 4.17 to 4.29 mmol/L, P<0.001), LDL cholesterol (2.67, 2.62 to 2.71 versus 2.51, 2.46 to 2.56 mmol/L, P<0.001), and HDL cholesterol (1.14, 1.15 to 1.18 versus 1.10, 0.99 to 1.03 mmol/L, P<0.001). Triglyceride levels for women were lower (1.44, 1.39 to 1.49 versus 1.55, 1.48 to 1.61 mmol/L, P<0.001) as was their body mass index (26.4, 26.2 to 26.7 versus 27.1, 26.9 to 27.3, P=0.001). Length of hospital stay was shorter for women (4.8, 4.5 to 5.0 versus 5.1, 4.8 to 5.4 days, P=0.026). When treated with thrombolytics, arrival to treatment time was longer for women (89.5, 83.5 to 95.5 versus 82.0, 76.2 to 87.8 minutes, P=0.012), despite their having a slightly higher National Institutes of Health Stroke Scale score (7.1, 6.7 to 7.6 versus 6.5, 6.1 to 7.0, P=0.05). Values for onset-to-arrival, fasting blood sugar and glycohemoglobin were not significantly different for women and men.
The National Institutes of Health Stroke Scale was performed less often for women, but imaging was done with equal frequency. Results of initial imaging studies, general stroke type, and rate of symptom resolution were not significantly different from those noted in men. Thrombolytic treatment for ischemic stroke was used at similar rates for both genders (Table 3).
Concerns listed as reasons for not using thrombolytics differed in that age (5.1% versus 2.7, P<0.001), delayed arrival (76.7% versus 71%, P<0.001), and refusal (2.8% versus 1.7%, P=0.033) were cited more often in women. Seizure at onset (0.8% versus 1.7%, P=0.01) and mild/improving symptoms (46.4% versus 49.8%, P=0.044) were noted more often in men.
By the second hospital day, women were more likely to be made “comfort care” and they were less likely to be ambulatory. They were less likely to develop pneumonia. They were not significantly different from men in regard to hemorrhagic complications of thrombolytic treatment (Table 4).
By the time of discharge, women differed from men in multiple aspects (Table 5⇓). They were more likely to be designated “comfort care” and to have died or been discharged to hospice. They were less likely to have been discharged to home and less likely to be ambulatory. They were more likely to receive antihypertensives, but less likely to receive antithrombotics. They were also less likely to receive lipid-lowering agents and diabetic medications. They were more likely to have been noted to have atrial fibrillation and to be assessed for rehabilitation. They were less likely to be given lifestyle recommendations or antismoking treatment.
Despite these differences at discharge, adherence to the consensus stroke quality indicators was not significantly different for the 2 genders, except that women were less likely to receive lipid-lowering treatment for elevated LDL levels (85.5% versus 90.4%, P<0.001).
GGS is widely used to improve adherence to stroke quality indicators within individual hospitals. We have found that it may also be used to give a state-wide picture of acute stroke and acute stroke care. In this report, we have examined 126 data elements. Of these, we found that 47 (37%) showed statistically significant gender-related differences at the 0.05 level. This “yield” is no doubt inflated because the data elements are not necessarily independent of one another, and because we did not correct for multiple comparisons. Nevertheless, fairly consistent patterns appear in this analysis. The most interesting of these are summarized in Table 6.
In general, it appears that, as compared to men, women in Colorado are older and more significantly affected by acute stroke. Risk factor profiles differ between the 2 genders. Lipids are less aggressively treated and antithrombotics are less commonly used in women. Less frequently obtained stroke consultation and National Institutes of Health Stroke Scale measurement, along with more frequent assignment to comfort care, may suggest that overall, the treatment of women is “less aggressive” than for men in Colorado. In part, this may be explained by the women being significantly older than the men. Determining whether this is the case, and whether there are other explanatory factors will require additional analysis. These data are observational and exploratory in nature. Further research is needed to determine whether the differences noted here are genuine and clinically important.
Acknowledgments of Participating Hospitals: Boulder Community Hospital, Boulder; Community Hospital, Grand Junction; Conejos County Hospital, La Jara; Denver Health, Denver; Exempla St Joseph’s Hospital, Denver; Family Health West, Fruita; Haxtun Hospital District, Haxtun; Littleton Adventist Hospital, Littleton; Medical Center of Aurora, Aurora; Medical Center of Rockies, Loveland; Melissa Memorial, Holyoke; Memorial Hospital, Colorado Springs; Mt San Rafael Hospital, Trinidad; North Colorado Medical Center, Greeley; North Suburban Medical Center, Thornton; Parkview Medical Center, Pueblo; Penrose St Francis Hospital, Colorado Springs; Porter Adventist Hospital, Denver; Poudre Valley Hospital, Ft Collins; Presbyterian/St. Luke’s Medical Center, Denver; Rangely Hospital District, Rangely; Rose Medical Center, Denver; San Luis Valley Regional Medical Center, Alamosa; Sky Ridge Medical Center, Lonetree; Southeast Colorado Hospital District, Springfield; Spanish Peaks Regional Health Center, Walsenburg; St Anthony Central Hospital, Denver; St Anthony Hospital - North, Westminster; St Mary Corwin Hospital, Pueblo; St Mary’s Hospital, Grand Junction; St Vincent General Hospital, Leadville; Swedish Medical Center, Englewood; The Memorial Hospital, Craig; University Hospital, Aurora; Wray Community District Hospital, Wray; Yuma Hospital District, Yuma.
Source of Funding
This manuscript is sponsored by the Cancer, Cardiovascular Disease and Pulmonary Disease Grants Program at the Colorado Department of Public Health and Environment.
- Received November 19, 2008.
- Revision received December 18, 2008.
- Accepted December 19, 2008.