Background and Purpose Clinical trials of new drugs that reverse neurological deficits when used in the first hours of stroke onset suggest that early hospital admission is important. We analyzed a database of patients with acute stroke to determine the factors that delay hospital admission.
Methods We analyzed all patients with their first stroke during 1993 in the province of Central Finland (population, 256 000). Patients referred to the Central Hospital, the only tertiary referral hospital in the area, were included in the study.
Results Of the patients with first stroke, 363 (79%) were admitted to the Central Hospital. The stroke subtype was confirmed in 356 (98%) patients with CT scan, and the patient population included 272 (75%) with brain infarction, 51 (14%) with intracerebral hemorrhage, and 40 (11%) with subarachnoid hemorrhage. The most important factor associated with a delay in reaching the hospital was the referral pattern. The median delay was 2 hours for patients brought directly to the Central Hospital, 8 hours if a physician at the local health center was consulted, and 47 hours if the patient was first admitted to the health center for observation. Other factors associated with a delay were ischemic stroke and stroke onset in the evening or night or during the weekend.
Conclusions The majority of patients who are candidates for acute stroke trials arrive at the hospital after prolonged delays for multiple reasons. Public and medical personnel education could result in significant reduction in these delays.
The importance of early hospital admission of patients with acute stroke is being increasingly recognized because of ongoing clinical trials of new promising drugs with narrow therapeutic time windows. Recruitment of patients for these trials is hampered by delays in hospital arrival. Until now, only a few studies have addressed the factors that determine the time interval from stroke onset to hospital admission.1 2 3 4 5 6 Our aim in the present population-based study was to analyze factors that delay hospital admission.
Subjects and Methods
We identified all patients in the province of Central Finland (population, 256 000) who had their first stroke in 1993. Patients were traced from the discharge lists of the Central Hospital and 18 health centers, death certificates, and autopsy reports. Because stroke patients are traditionally treated in the hospital in Finland, we believe that virtually all patients were identified from these sources. If stroke (International Classification of Diseases, 9th Revision, codes 430 to 438) appeared on the discharge list or death certificate, all medical records of the patient were reviewed. Stroke was defined as an acute focal or global neurological deficit of presumed vascular origin lasting for more than 24 hours.
Time of stroke onset was defined as the time the patient or an observer first noted a neurological deficit. In many cases, the hour of onset was available, but in others the onset was often listed as having occurred in the morning, afternoon, evening, or night. In these cases, we estimated the time of onset as 9 am, 3 pm, 9 pm, and 3 am, respectively. If the symptoms were present on waking, we assumed that stroke had occurred during the night, and 3 am was coded as the time of onset. For some patients, only the day of onset could be obtained because of disturbed consciousness or dysphasia, especially if the patient had lived alone or there was diffuse onset of stroke symptoms. The exact time of arrival at the Central Hospital is routinely marked on the forms of the ED. The length of delay from onset to admission was calculated on the basis of the exact or estimated hour of onset and the time of arrival in the ED.
Because of the skewed distribution of the delays, the nonparametric M-W test and K-W ANOVA were used in testing if significant differences existed between the group medians.
During 1993, a total of 459 persons in the study population had their first stroke. Of these, 318 (69%) patients were first admitted to the Central Hospital, 135 (29%) were admitted to a health center, and 6 (1%) died before admission. Forty-five of the 135 patients first admitted to a health center were later transferred to the Central Hospital, and a total of 363 patients eventually treated in the Central Hospital were the subject of our study.
CT scan was performed on 356 (98%) patients: 148 (42%) within 24 hours, 268 (75%) within 48 hours, and 301 (85%) within 72 hours of onset. Fifty-six (15%) patients underwent angiography, and 30 (8%) deceased were autopsied. The distribution of the patients by sex, age, and diagnosis is shown in the Table⇓.
The time of onset was known in 312 (86%) of the 363 patients. In 146 patients, the hour was known and in 166 patients the quarter of the day. The distribution of time of onset was as follows: midnight to 6 am (83 patients), 6 am to noon (111 patients), noon to 6 pm (67 patients), and 6 pm to midnight (51 patients). In 51 patients only the day of onset was known, and in 32 (63%) admission was greatly delayed (from 1 to 14 days after onset). Nineteen of these 51 patients were admitted during the first day, and among them the time of onset could be estimated to a 12-hour period.
Most of the 343 patients admitted to the ED arrived between 10am and 3 pm, and the admission rate was minimal during the night and early morning hours (Fig 1⇓). Twenty patients were not admitted through the ED: 15 patients had stroke onset in the hospital while hospitalized for other reasons, 2 were admitted directly to the hospital ward, and 3 were remitted to the neurological outpatient department. For these 20 cases, the time of neurological evaluation was taken as time of admission.
Of the 363 patients, 43% were admitted within 6 hours, 60% within 12 hours, 71% within 24 hours, and 84% within 48 hours. Patients with hemorrhagic strokes were admitted after much shorter delays than those with INF (Fig 2⇓). The median delay for subarachnoid hemorrhage was 3 hours, for intracerebral hemorrhage 4 hours, and for INF 10 hours (K-W, P=.0069). Correspondingly, patients with disturbed consciousness, severe headache, and/or vomiting at onset presented significantly earlier (M-W, P=.002; median, 5 hours) than those without these symptoms (median, 11 hours). The earlier admitted hemorrhagic stroke patients were younger than patients with INF. However, among INF patients aged <70 years, the median delay (10 hours) was the same as for those aged ≥70 years (9.5 hours) (M-W, P=.67). Previous transient ischemic attacks (61 patients), use of acetylsalicylic acid (94 patients), or the distance between the Central Hospital and the residence of the patient had no measurable effects on the delay.
The quarter of day when the onset occurred had a significant (K-W, P<.0001) impact on the delay: it was longer when onset occurred during the night (median, 10 hours) or evening (median, 12 hours) compared with delay for those who had onset in the morning (median, 4 hours) or afternoon (median, 3 hours). We also observed that onset during the weekend was associated with longer median delay (11 to 16 hours) than onset on a weekday (4 to 8.5 hours) (K-W, P=.0496). One explanation was that patients who had onset during the weekend were often first admitted to the local health center.
One very important factor deciding the length of the delay was the referral pattern. If the patient was brought directly to the ED, the median delay was only 2 hours, but if the patient was first assessed at the outpatient department of the local health center and afterward referred to the Central Hospital, several hours were lost and the median delay was 8 hours. Patients who were admitted to the health center for observation and later transferred to the Central Hospital had the longest delay (median, 47 hours), and none of the patients was admitted to the ED within 12 hours of onset. In the case of INF, the situation was similar to that for the total patient population (Fig 3⇓).
The reported time interval from stroke onset to hospital admission, especially concerning the first 6 hours, has varied widely in earlier studies: from 50% admitted within 4 hours,6 and 48% to 77% within 6 hours,1 4 5 to only 42% admitted within the first 24 hours.3 Several factors delaying admission have been identified: INF compared with hemorrhagic stroke,3 6 stroke occurring during the night,4 5 living alone,4 age more than 70 years,4 admission to a geriatric instead of a medical unit,4 admission through a bed-allocation bureau,4 stroke occurring at home,5 and first medical contact with a personal physician or study hospital compared with an emergency service.5
In half of the patients we studied, only a relatively crude estimate, quarter of the day of the time of onset, could be determined. For an individual patient, the use of these estimates in the calculations would not give the exact length of the delay; however, for the total patient population, the results may be reliable. Had the study been prospective, the hour of onset might have been known more often, but diffuse onset of stroke, onset during sleep, decreased level of consciousness, and dysphasia are among the factors that do not allow a more precise estimate.
The generally more fulminant onset of intracerebral and subarachnoid hemorrhage seems a logical explanation for why patients with hemorrhagic strokes were admitted after shorter delays than patients with INF. In accordance with earlier studies,4 5 the delay from onset to admission was significantly longer in strokes occurring during the night. One explanation is that they are often evident first on awakening, and the time of onset was coded as 3 am. On the other hand, people who notice the symptoms during the night often are hesitant to seek medical advice before morning (Fig 1⇑). Onset during the night also constitutes a therapeutic problem: the time interval from stroke onset to awakening may vary from 0 to 8 to 9 hours, and this uncertainty may prevent thrombolytic treatment, for example.
The most important finding in our study was the role of the referral pattern. If the patients came directly to the ED, 50% were admitted within 2 hours and 71% within 6 hours after onset. If the physician at the local health center was first consulted, only 16% were admitted within 2 hours and 42% within 6 hours (Fig 2⇑). The corresponding proportions for INF were 39% and 65%, respectively, if the patients were admitted directly to the ED and 4% and 35%, respectively, if the health center was first consulted (Fig 3⇑). On the other hand, the distance from the patient’s home to the Central Hospital did not have a measurable effect on the length of the delay.
Before treatment is begun, stroke subtype must be confirmed. In our study, only 42% of patients underwent CT scan within the first 24 hours, and three fourths had CT within the first 48 hours, which is in accordance with a recent study from New Zealand.6 If some of the new drugs with narrow therapeutic windows prove effective, earlier timing of CT scanning becomes mandatory. Immediately after arrival in the ED, patients should be sent for CT scan; after this, treatment can begin in the ED.
Despite the fact that we presently do not have any effective treatment for acute ischemic stroke,7 an early admission policy must be promoted to recruit patients into trials of promising new drugs. Additionally, we must be prepared for the day when we have an effective treatment at our disposal. This can be accomplished by informing the public and medical personnel of the symptoms and signs of cerebrovascular disorders and the fact that stroke is an emergency situation similar to myocardial infarction. Patients must also be informed about the importance of seeking a hospital with high-quality stroke diagnosis and treatment facilities.
Selected Abbreviations and Acronyms
This study was financially supported by the Health Care District of Central Finland. We thank Lawrence C. Sellin, PhD, and Cormac O’Donovan, MD, for reviewing the manuscript and checking the language.
Reprint requests to Rainer Fogelholm, Soidintie 3 as 6, 40630 Jyväskylä, Finland.
- Received October 24, 1995.
- Revision received November 13, 1995.
- Accepted November 30, 1995.
- Copyright © 1996 by American Heart Association
Foulkes MA, Wolf PA, Price TR, Mohr JP, Hier DB. The Stroke Data Bank: design, methods, and baseline characteristics. Stroke. 1988;19:547-554.
Harper GD, Haig RA, Potter JF, Castleden CM. Factors delaying hospital admission after stroke in Leicestershire. Stroke. 1992;23:835-838.
Anderson NE, Broad JB, Bonita R. Delays in hospital admission and investigation in acute stroke. BMJ. 1995;311:162.
Bath PMW. Treating acute ischaemic stroke: still no effective drug treatment. BMJ. 1995;311:139-140.