|Home | About | Journals | Submit | Contact Us | Français|
Data from the Project for Improvement of Stroke Care Management in Minnesota were used to characterize provider practice in the management of hypertension in hospitalized acute ischemic stroke patients in relation to guidelines from the American Stroke Association (ASA) and the European Stroke Initiative (EUSI). Among 1181 patients, 129 received as-needed medication in direct response to elevated blood pressure in the first 24 hours after admission. Of these 129, 56% were overtreated according to ASA guidelines and 24% by EUSI guidelines. Of the 1052 patients not treated, 16% were undertreated by ASA guidelines and 3% by EUSI guidelines. In contrast, discharge hypertension was well treated, with 93% (424 of 454) of patients likely to have chronic hypertension appropriately treated with antihypertensive medication on discharge. These findings support the need for more research to fill identified gaps in evidence, improve treatment guidelines, and reduce variability in patient management.
Among stroke survivors, the benefit of long-term treatment of chronic hypertension is well supported by clinical trial evidence.1-3 This is reflected in clear, consistent guidelines advocating long-term blood pressure (BP) treatment in this patient population.4,5 In contrast, the optimal approach to BP management during the acute phase of ischemic stroke is less clear. BP elevation is common among patients with acute ischemic stroke even among those without prior history of hypertension.6 On the basis of anecdotal experience and expert opinion, it is widely believed that rapid lowering of elevated BP in the acute phase may result in neurologic deterioration due to decreased perfusion pressure to ischemic brain tissue.7-9 Existing guidelines recommend that elevated BP in the acute phase after ischemic stroke be treated cautiously and only when substantially elevated.4,10,11 These recommendations are not based on level I evidence, and their recommended BP thresholds for initiating treatment differ. Furthermore, none of the guidelines address the issue of when new antihypertensive regimens for long-term management should be initiated after stroke or whether prestroke antihypertensive regimens should be stopped in the acute phase (and, if so, when they should be restarted). There are clinical trials underway to address these issues.12-14 In the interim, gaps in the immediate poststroke BP treatment guidelines make patient BP management in the immediate aftermath of ischemic stroke difficult for practitioners.15
In the absence of clear treatment guidelines, provider behavior is likely to vary substantially. In this study, we sought to determine the extent of variability in providers’ management of hypertension in acute ischemic stroke patients, using data from a population-based study (Project for Improvement of Stroke Care Management in Minnesota [PrISMM]).
We examined provider behavior during different stages of in-hospital stroke care: the acute phase (first 24 hours after admission), the remaining predischarge time, and at discharge. We expected to find the greatest variability in provider behavior when guidelines were absent or weakly supported by empirical evidence. Observed practice was compared with applicable guidelines from the American Stroke Association (ASA) and the European Stroke Initiative (EUSI), which are similar but not identical in their recommendations for poststroke BP management.
Our goal in conducting this descriptive study was to characterize a baseline of the variability in provider behavior and guideline adherence for BP management after acute ischemic stroke. This baseline will be useful for assessing change when more complete and stronger evidence-based guidelines become available.
A review of the literature on the acute-phase management of BP after ischemic stroke yielded 3 sets of guidelines: 1 from the American Heart Association/ASA,10 a second set from the International Society of Hypertension (ISH),11 and a third from the EUSI.4 The thresholds recommended by the ASA and the ISH are identical. The EUSI recommendations allow a lower treatment threshold. The specific treatment thresholds are as follows:
Both guidelines focus on the avoidance of antihypertensive medications below the recommended thresholds (ie, the avoidance of overtreatment). Neither guideline clearly addresses the issue of undertreatment. For example, the EUSI guideline does not state that elevated BP beyond 180/105 mm Hg should be treated and in fact allows for higher treatment thresholds up to 220/140 mm Hg. The EUSI guideline therefore allows a range of BP elevations (180–220 mm Hg systolic; 105–140 mm Hg diastolic) as acceptable, while the ASA guideline has a more rigid cutoff (220 mm Hg systolic; 120 mm Hg diastolic). The EUSI undertreatment threshold was 220 mm Hg SBP or 140 mm Hg DBP and was different from its overtreatment threshold of 180/105 mm Hg. The ASA overtreatment and undertreatment thresholds were identical at 220 mm Hg SBP or 120 mm Hg DBP or 130 mm Hg mean BP.
Neither the ASA nor the EUSI guidelines specify a time window during which their BP treatment thresholds would apply (ie, it is unclear whether the pressures should be allowed to be elevated for the first 24 hours, 48 hours, or even longer). Both guidelines list hypertensive encephalopathy, aortic dissection, acute myocardial infarction (MI), acute renal failure, and acute pulmonary edema as potential exceptions to the thresholds listed above.
BP control in the postacute, predischarge period is also not explicitly addressed in the guidelines. Neither guideline comments on the exact timing for initiation of new, scheduled antihypertensive medications after stroke. The timing for resumption of prestroke regimens is similarly ignored.
Finally, the long-term treatment of hypertension for secondary prevention of recurrent stroke is clearly advocated by both guidelines. In fact, the latest iteration of the ASA stroke prevention guidelines (published after our study)5 recommends postacute-phase, long-term BP treatment regardless of a diagnosis of hypertension.
In addition to the stroke-specific guidelines discussed above, poststroke BP management is addressed by the broader, comprehensive report of the Joint National Committee on the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7).16 This report, published after our study, cites the ASA guideline for acutephase treatment thresholds10 and also offers similar recommendations for long-term treatment.5
The published express version of the JNC 7,17 however, recommends that the BP in the acute setting be maintained at about 160/100 mm Hg, a lower level than either the ASA or the EUSI guideline thresholds.
The PrISMM project was designed to examine acute ischemic stroke care in Minnesota and included 19 hospitals in the Minneapolis and St Paul metropolitan and surrounding areas. Approval was obtained from the institutional review board of each participating hospital. The PrISMM study included patients between the ages of 30 and 84 years identified by hospital discharges with International Classification of Diseases, Ninth Revision (ICD-9) codes (434, 436) of acute ischemic stroke between July 1 and December 31, 2000. The diagnosis of acute ischemic stroke was further confirmed by chart review. Only the first admission was included for patients who had multiple stroke admissions during the 6-month study period.
We used data from the PrISMM study to assess provider behaviors in the management of hypertension in acute ischemic stroke patients. Patients who received thrombolytic therapy after stroke and patients whose stroke occurred during hospitalization for other illnesses were excluded from this analysis because the poststroke BP thresholds are different for the former, and BP management may be influenced by comorbid conditions for the latter. Information compiled in the PrISMM database included patient history, acute stroke care, clinical variables, and demographic information. Variables pertaining to hypertension and therefore relevant to this study were as follows: history of hypertension, prestroke antihypertensive medications, systolic and diastolic BP measurements in the first 24 hours after admission, scheduled medications and as-needed (ie, in immediate response to an elevated BP) medications given during hospitalization, and medications prescribed on discharge. Assessments of the patients’ clinical features on presentation were also collected. Trained personnel collected data on laptop computers according to instructions in detailed manuals. All medication information was entered using comprehensive pick-lists with the option of noting medications not on the pick-list. Randomly selected charts from each abstractor were reabstracted by an experienced reference abstractor. Interrater reliability for abstractors vs the reference abstractor was >95%.
Data on provider behaviors with respect to BP management were analyzed for the first 24 hours after admission (acute phase), after the first 24 hours but before discharge, and at discharge. In defining the acute phase, we used admission time rather than time of symptom onset because many patients did not have symptom onset noted in their charts. As mentioned earlier, the guidelines do not specify a time window for defining the acute phase during which elevated BPs would be untreated until they exceeded a high treatment threshold. For our study, we chose a 24-hour time window. We reasoned that in this amount of time, practitioners would likely have acquired enough clinical and radiographic data to determine their patients’ neurologic stability and tailor BP management accordingly.
Antihypertensive medication orders in the acute phase were classified as “as needed” (ie, given as needed in response to an elevated BP for immediate effect) or “scheduled” (medications prescribed for regular use at set times) orders. For each use of as-needed medication, we identified the type of medication prescribed and the BP triggering treatment. Using the BP thresholds recommended by the ASA and EUSI guidelines, we determined episodes of overtreatment and undertreatment. As described earlier, the EUSI overtreatment threshold was 180 mm Hg SBP or 105 mm Hg DBP and undertreatment threshold was 220 mm Hg SBP or 140 mm Hg DBP. The ASA overtreatment and undertreatment thresholds were identical at 220 mm Hg SBP or 120 mm Hg DBP or 130 mm Hg mean BP.
An episode of overtreatment was counted if a patient received as-needed antihypertensive medication in response to a BP less than the guideline overtreatment threshold and if there were no BP values meeting the guideline criteria within a 2-hour time window before medication administration. This time window was intended for instances where a physician may have ordered a medication in response to BP elevation, but the arrival and administration of medications may have been delayed. Patients with 1 or more episodes of overtreatment were counted as being overtreated according to the guidelines. Our database contained information on MI and acute renal failure; thus, we excluded cases with these conditions from our analyses of overtreatment. An episode of undertreatment was counted if medication was withheld in the face of elevated BP exceeding the undertreatment threshold. We examined provider behavior in terms of not treating BPs beyond guideline-recommended thresholds with the caveat that neither guideline clearly recommended against undertreatment.
Overtreatment and undertreatment could potentially occur in the same patient. Patients frequently had multiple BP measurements in the first 24 hours and sometimes received multiple medication doses during that period. A problem during analysis was the potential assignment of a medication dose given in response to multiple BP measurements within the qualifying 2-hour time window. In this situation, the adjudication of whether a given medication usage was overtreatment and whether there was undertreatment of certain BPs would be ambiguous. To avoid this, we defined separate and nonoverlapping eligible subsets of patients or denominators for characterizing each treatment pattern. Figure 1 describes the separate denominators.
Each guideline was analyzed separately. We examined the association between overtreatment and patient age; sex; history of MI; clinical features at presentation, such as limb weakness, level of consciousness, aphasia; stroke subtype; and involvement of a neurologist either as a consulting or managing physician. Similar analyses were done for undertreatment. The chi-square test was used for categoric variables and the t test for continuous variables. We applied the Bonferroni correction to our results to guard against spurious positive associations from the multiple comparisons performed.
We determined rates of initiation/resumption of antihypertensive medications in the postacute, predischarge period. We could not classify their use as being compliant with recommended practice due to lack of guidelines addressing this issue.
We used 2 criteria to examine antihypertensive use on discharge: (1) history of preadmission antihypertensive medication use by the patient, and (2) presence of sustained BP elevation during hospitalization. Sustained elevation was defined as BP values ≥140 SBP mm Hg or ≥90 DBP mm Hg on 2 or more consecutive occasions at an interval of 30 minutes or longer after the first 24 hours. Finally, we examined the different types of antihypertensive medications prescribed on discharge and compared them with guideline recommendations.
The PrISMM database contains 1413 acute ischemic stroke hospitalizations. Only the first admission was included for 28 patients with multiple stroke admissions. Thirty-seven patients who received thrombolytic therapy and 167 inpatient stroke cases were excluded, leaving 1181 patients for our analysis. Patients’ mean age was 70 years (SD, 11.4 years) and median age was 73 years. The sex distribution of the sample was nearly equal, with 51.5% men and 48.5% women. The race distribution was as follows: 78% white, 4% black, 1% Hispanic, 2% Asian, and 1% other, including multiracial. Race was not recorded for 14% of patients.
Before admission, 73% (861 of 1181) of all patients had a history of hypertension (Table I), and 60% (713 of 1181) were being treated with at least 1 antihypertensive medication. (Each patient had up to 4 preadmission BP medications recorded.) Also in Table I are the number and percentage of patients who experienced acute-phase or sustained BP elevation (as defined in the Methods section) after being admitted for acute stroke. There were 60 in-hospital deaths. Of the 1121 patients who survived and were discharged, 672 (60%) were on preadmission antihypertensive treatment.
In the acute phase, there were 245 instances of asneeded medications given to 129 patients (11% of the total sample) (Table II). These 129 patients formed the denominator for our examination of overtreatment by guideline definitions. The remaining 1052 patients formed the denominator for our examination of undertreatment. Also in the acute phase (first 24 hours), new scheduled medications were administered to 25% of all patients (294 of 1181), and complete or partial resumption of prestroke medications occurred in 66% of patients who were on preadmission regimens (468 of 713).
Figure 2 summarizes the results of overtreatment and undertreatment analyses. Of the 129 patients who received as-needed treatment, 1 patient had acute MI and 2 had acute renal failure; these patients were removed from analysis of overtreatment. Of the remaining 126 patients, treatment was always consistent with ASA guidelines in 56 (44%) patients and always consistent with the EUSI guidelines in 96 (76%) patients. Hence, 56% of patients who received as-needed medications had at least 1 episode of overtreatment according to the ASA guidelines, and 24% had at least 1 episode of overtreatment according to EUSI guidelines. Figure 3 shows the distribution of BPs that triggered the use of acute-phase, as-needed antihypertensive medications. The undertreatment denominator consisted of 1052 patients. Only 3% of the patients (29 of 1052) were undertreated by EUSI guidelines, and 16% (167 of 1052) by ASA guidelines.
There was no association between overtreatment or undertreatment by either guideline and any of the following variables: patient age; sex; history of MI; or clinical presentation, including level of consciousness, aphasia, limb weakness; or stroke subtype. There was similarly no association between overtreatment or undertreatment by either guideline and involvement of a neurologist in patient care.
There was considerable variation in the choice of as-needed antihypertensive agents used in the acute phase. Many patients received agents recommended by the guidelines, however, including labetalol (36 of 129; 28%) and nitroprusside (26 of 129; 20%) (Table II).
As noted above, 713 patients were taking antihypertensive medications prestroke. These regimens were completely or partially restarted in 72% of patients (516 of 713) within 48 hours of admission and in 78% of patients (553 of 713) at some point during hospitalization. New scheduled antihypertensive regimens were started in 36% of patients (428 of 1181) within 48 hours of admission and in 49% of patients (579 of 1181) at some point during hospitalization.
Figure 4 illustrates the discharge treatment of the 1121 patients who survived hospitalization. To summarize, 93% of patients who were on preadmission (prestroke) antihypertensive medications and had sustained BP elevation after the acute phase were discharged on antihypertensive medications. Of note, 26% of patients with neither criteria (ie, neither prestroke antihypertensive medication use nor postacute phase sustained BP elevation) were discharged on antihypertensive medications. The rates for patients meeting 1 of the 2 criteria are shown in Figure 4.
A total of 822 of 1121 (73%) patients were discharged on antihypertensive medications. Of these, 572 were discharged on more than 1 agent. Table III shows the frequency of different classes of antihypertensive agents and guideline recommendations.
BP management in the immediate aftermath of stroke has been examined in single-hospital studies.18-20 A study by Lindenauer and colleagues18 looked at poststroke antihypertensive medication use at a large community-based teaching hospital for the same time period as that presented here and reported overtreatment rates of up to 74% according to ASA guidelines. The authors of that report did not separate as-needed and scheduled medication use as in our study. Hence, their rates of overtreatment are higher than our rates. A Canadian study by Kanji and associates19 described considerable variability in poststroke management of BP and concluded that guidelines had little influence on prescribing patterns. This was attributed to lack of evidence from randomized controlled trials. In a similar vein, a Brazilian study reported no relation between antihypertensive medication use in the acute phase and BP levels despite local guidelines advocating high treatment thresholds.20 Our results and conclusions, while consistent with other hospital reports, show that this issue is not an isolated problem in single hospitals; rather, it is widespread. Our data describe prescription patterns across 19 acute-care hospitals in a population-based setting. One caveat is that the hospitals in our study are all located in Minnesota. Hence, the practice variations shown here cannot be assumed to represent provider behavior in other parts of the country. The single hospital reports from geographically distant areas such as Massachusetts,18 Canada,19 and Brazil20 suggest that our results are not unique to Minnesota. There is, nevertheless, a dearth of population-based studies examining practice variations in the acute-phase management of BP after ischemic stroke.
Our results raise the obvious question as to possible explanatory variables underlying the observed contrast between acute-phase management with its high overtreatment rates based on present guidelines and the more guideline-compliant discharge treatment. One hypothesis is that provider behavior is influenced by the strength of the evidence underlying management guidelines. The recommendation regarding the long-term use of antihypertensive medications after stroke is well supported by level I evidence.1-3 Correspondingly, providers in our study showed a very high level of compliance with this guideline when the diagnosis of hypertension was clear (Figure 4). The prescription of antihypertensive medications on discharge varied depending on whether the patient was on prestroke antihypertensive medication and had sustained postacute phase elevated BPs, compared with individuals not meeting one or both criteria. This likely reflects the provider’s certainty in diagnosing hypertension in patients who need chronic treatment. Our results suggest that providers are primarily prescribing discharge antihypertensive medications in patients who would be more likely to carry the diagnosis of chronic hypertension compared with those unlikely to do so. Whether providers will modify their behavior in response to the new, broader recommendations5 to treat all stroke patients regardless of a diagnosis of hypertension remains to be seen.
In contrast to the treatment of hypertension on discharge, we observed significant variability in the management of elevated BP in the acute phase—a period for which there is weak evidence and incomplete guideline recommendations on BP management. Overtreatment in the first 24 hours was 56% in the case of more stringent guidelines10,11 and 24% for the more liberal recommendations.4 None of the guidelines for antihypertensive use are based on level I evidence, and existing studies offer conflicting viewpoints. To illustrate, Castillo and colleagues21 reported an association between poor prognosis and both high and low BPs after stroke; a randomized trial of intravenous nimodipine22 showed neurologic deterioration after diastolic (but not systolic) BP reduction. Likewise, Oliveira-Filho and associates20 reported an independent association between acute-phase BP reduction and poor 3-month outcome. A recent systematic review23 concluded, however, that high SBP, DBP, and mean BP in acute ischemic stroke were associated with poor outcomes and argued for moderate lowering of elevated BP in the acute phase. A post hoc analysis of National Institute of Neurological Disorders and Stroke (NINDS) rt-PA trial data24 also suggested that gentle BP reduction to 180/105 mm Hg, a threshold closest to the EUSI limits, might be well tolerated. Expert opinion reflects the contradictory nature of existing evidence. Some experts9,15 caution against acute-phase BP lowering and others advocate a carefully titrated reduction.25,26 Practitioners are thus faced with conflicting evidence, differing expert opinions, and consequently weak guidelines. This may explain the observed variability in the acute-phase management patterns.
The high overtreatment rates may also be a consequence of gaps in provider knowledge, namely, guideline unfamiliarity. Providers may be overgeneralizing known facts regarding the role of hypertension as a stroke risk factor and the importance of long-term poststroke BP control into the acute phase. They may therefore aggressively control BP in the early stages of stroke. This study does not directly test hypotheses linking provider knowledge and behavior; however, our analysis showed a lack of significant association between the involvement of a neurologist in patient care and guideline adherence. This may suggest that the variability was not due to guideline familiarity or lack thereof, if one assumes neurologists to be more familiar with stroke guidelines as compared with other practitioners caring for ischemic stroke patients. A different explanation for the high overtreatment rates may be provider discomfort with letting BPs run very high (despite guideline familiarity), possibly due to fear of end organ damage leading to renal failure, MI, or intracerebral hemorrhage. The modest undertreatment rates reported in our study are compatible with, but do not prove, this explanation. Although avoidance of undertreatment was not emphasized by the guidelines, the prevalence of undertreatment was considerably less than that of overtreatment.
A limitation of our study is that although we have identified gaps in practice guidelines and show a corresponding variability in actual practice patterns, as yet, we can offer no prescription to guide physicians on navigating this issue. We contend that a desirable reduction in the practice variation illustrated here will only follow when well-designed randomized trials are conducted—trials that can robustly ascertain which treatment approaches in the acute, predischarge time periods lead to optimal outcomes in patients hospitalized with acute ischemic stroke.
Acknowledgments and disclosures: The authors acknowledge A. M. Weber-Main for her critical review and editing of manuscript drafts and D. Quick for figure preparation. Significant contributors to this article include N. E. Morris, P. A. Janey, A. Boese, and B. McLaughlin, who assisted in data collection, data management, and analyses. AHRQ U18HS11073–01 and NIH/NINDS R01NS39028 funded the PrISMM study. Dr Lakshminarayan was supported by a clinical research training fellowship (#84500–2002) from the American Academy of Neurology Foundation and NIH/NCRR 1K12RR023247.