Dysphagia 9:7-11 (1994) Dysphagia �9 Springer-Verlag New York Inc. 1994

Videofluoroscopic Evidence of Aspiration Predicts Pneumonia and Death but not Dehydration Following Stroke

John Schmidt, MD, Marlene Holas, MS, Kathryn Halvorson, MS, and Michael Reding, MD Cornell University Medical College at The Burke Rehabilitation Center, White Plains, New York, USA

Abstract. In order to assess the risk of pneumonia, de- hydration, and death associated with videofluoroscopic evidence of aspiration following stroke, the clinical records of 26 patients with aspiration and 33 randomly selected, case-matched, dysphagic controls without video- fluoroscopic evidence of aspiration were reviewed. The videofluoroscopic modified barium swallow technique included 5 ml-thin and thick liquid barium, 5 ml barium pudding, and V4 cookie coated with barium, plus addi- tional 20 and 30 ml of thin liquid barium. Patients were assessed a mean of 2 + 1 SD months poststroke and were followed for a mean of 16 + 8 SD months poststroke. The odds ratio for developing pneumonia was 7.6 times greater for those who aspirated any amount of barium irrespective of its consistency (p = 0.05). The odds ratio for developing pneumonia was 5.6 times greater for those who aspirated thickened liquids or more solid consisten- cies compared with those who did not aspirate, or who aspirated thin liquids only (p = 0.06). Dehydration was unrelated to the presence or absence of aspiration. The odds ratio for death was 9.2 times greater for those aspi- rating thickened liquids or more solid consistencies com- pared with those who did not aspirate or who aspirated thin liquids only (p = 0.0l). Aspiration documented by modified videofluoroscopic barium swallow technique is associated with a significant increase in risk of pneumo- nia and death but not dehydration following stroke.

Key words: Cerebrovascular disorders - - Dyspha- gia - - Pneumonia - - Mortality - - Epidemiology - - Deglutition - - Deglutition disorders.

Address offprint requests to: Michael Reding, M.D., 785 Mamaroneck Avenue, White Plains, NY 10605, USA

The frequency of dysphagia following stroke varies from 14% for patients with unilateral hemispheric involvement to 71% for those with bilateral hemispheric or brainstem involvement [1].

Dysphagia following stroke is usually confined to difficulty with the oral, pharyngeal, or combined oral- pharyngeal phases of swallowing. Problems with the oral phase of swallowing are usually apparent to the trained observer watching the patient eat and drink. Bedside observation of patients swallowing 50 ml of water was used by several British authors who found that 45% of all patients with new stroke in a community-based popula- tion were dysphagic [2,3]. Problems with the pharyngeal phase are often missed at the time of bedside observation of a patient's swallowing function. A number of studies based on the modified videofluoroscopic barium swallow (MBS) technique published by Logemann have docu- mented "silent aspiration" in up to 58% of dysphagic patients [4,5]. The frequency of pharyngeal phase dys- phagia is therefore probably underestimated by simple bedside observation of swallowing function. Veis and Logemann [6], using MBS examination to study their hospital-based sample of dysphagic stroke patients, found that 50% had oral, 82% had pharyngeal, and 76% had mixed oral-pharyngeal dysphagia.

The medical complications of dysphagia are probably different for oral, pharyngeal, and mixed oral- pharyngeal dysphagia. An obvious example is upper air- way obstruction in patients with oral phase dysfunction. These patients have difficulty with bolus control and may attempt to swallow incompletely masticated food. We are not aware of any data concerning the prevalence of upper airway obstruction for stroke patients with vs. those with- out dysphagia. Pharyngeal phase dysphagia with residue, and inhalation of food after the swallow, or inadequate closure of the airway, is expected to be associated with an increased incidence of aspiration pneumonia.

8 J. Schmidt et al.: Aspiration Following Stroke

Only one study has focused on the frequency of

pneumonia in patients with dysphagia vs. those without

dysphagia fol lowing stroke. Gordon et al. [2] defined dysphagia on the basis of a bedside swal lowing evalua-

t ion, and found the frequency of pneumon ia to be 19%

for dysphagic vs. 8% for nondysphagic patients. There are no studies based on MBS examinat ion that have re-

ported the prevalence of pneumon ia for stroke patients

with aspiration vs. those without aspiration. The relation-

ship between the amount and consis tency of bar ium test

material aspirated and subsequent development of aspira-

t ion pneumon ia has not yet been studied.

The role of dysphagia in the development of de-

hydrat ion fol lowing stroke has been studied only for pa- tients with dysphagia diagnosed on the basis of the bed-

side swal lowing examinat ion. Two studies that used the hematocri t to define dehydrat ion showed no signif icant

difference in this parameter for dysphagic vs. nondys- phagic patients [2]. One of these studies also used the

Blood Urea Nit rogen (BUN) and found it to be signifi-

cantly elevated in the dysphagic patients. Neither study

controlled for the concomitant use of in t ravenous hydra-

t ion or non-ora l feeding. The data presented in these two

studies do not allow one to estimate the frequency of dehydrat ion among patients with dysphagia fol lowing

stroke. The death rate of stroke patients with dysphagia

as diagnosed by the bedside swal lowing evaluat ion, has been reported to be from 40% to 63% over the first 6

months [3,7]. The death rate of all stroke patients, 6

months post onset has been reported to be 21% [8]. We

are not aware of any studies reporting the mortali ty of

patients with vs. those without aspiration based on MBS

examinat ion. The present study focuses on MBS evidence of

aspiration and its effect on the relative risk of developing

pneumonia , dehydrat ion, and death. Risk of developing

these medical complicat ions was computed indepen- dent ly for two aspects of aspiration: (1) amount of mate- rial aspirated, and (2) consis tency of material aspirated.

Methods

All subjects were residents on an inpatient stroke rehabilitation unit and had a diagnosis of stroke based on clinical history, neurologic examina- tion, and computed transaxial tornography scan or magnetic resonance imaging scan. Patients with stroke were referred for videofluoroscopic MBS if they had one or more of the following signs indicating possible dysphagia: (1) complaints of trouble swallowing, (2) coughing associ- ated with feeding observed by any of the rehabilitation staff members, (3) failure to consume more than half of the prescribed diet tray, (4) prolonged time required for feeding, (5) bilateral hemispheric stroke, (6) brainstem stroke, (7) history of pneumonia during the acute phase poststroke and (8) non-oral feeding program in progress.

A modification of the barium swallow technique outlined by Logemann [9] was used. Subjects were given 5 ml thin liquid barium, 5

ml thick liquid barium, 5 ml pudding-consistency barium, I/4 piece of cookie coated with barium, 20 rnl thin liquid barium to be taken in one swallow, and 30 ml thin liquid barium to be taken in successive swal- lows. Patients were seated upright and viewed in the lateral position. In addition, patients were given 5 ml thin liquid barium to be viewed in the anterior-posterior position. The fluoroscopic studies were recorded on a Panasonic AG-6200 videocassette recorder with a video countertimer and reviewed by a team of speech-language pathologists to determine the presence or absence of aspiration, the approximate amount of mate- rial aspirated, and the consistency of material aspirated.

Twenty-six patients with MBS-documented aspiration were compared with 33 randomly selected patients whose MBS showed no evidence of aspiration. Medical records were reviewed blinded to the results of MBS evaluation to determine the relative frequency of pneu- monia, dehydration, and death for both groups. The presence of com- plications following discharge from the inpatient rehabilitation unit was determined by phone follow-up with either the patient or next of kin.

Pneumonia occurring during the inpatient rehabilitation hospi- talization was diagnosed if the patient had three or more of the follow- ing: fever, positive ausculatory findings on chest exam, productive cough, Pa O z below 70 mmHg or a 10 mmHg decrease from the patient's baseline, purulent sputum Gram's stain or pathogen isolated from the sputum culture, positive chest x-ray film. Pneumonia occur- ring at the acute care hospital prior to rehabilitation referral or subse- quent to discharge from the rehabilitation unit was recorded as present or absent based on hospital transfer records or phone follow-up report, respectively.

Dehydration during inpatient rehabilitation hospitalization was defined as a serum sodium > 145 or BUN >35 without evidence of primary renal dysfunction. Dehydration occurring at the acute-care hospital or subsequent to discharge from the rehabilitation unit was recorded as present or absent based on hospital transfer records or the need for non-oral hydration as reported by the patient or next of kin.

Death was determined by rehabilitation hospital record review or by phone follow-up with next of kin.

Categorical outcome results were compared using Chi-square analysis. Fisher's exact test was used when Chi-square analysis gave three or fewer expected events in any cell. The "odds ratio" statistic was calculated rather than the "relative risk" because of the retrospective case-control design of our study. The odds ratios and confidence inter- vals were determined as described by Rimm et al. [10].

Results

Demographic features of the aspiration and nonaspira t ion groups are shown in Table 1. The mean fol low-up inter- val was 16 --- 8 SD months poststroke. Age, sex, prior

history for stroke; t ransient ischemic attack; chronic ob-

structive pu lmonary disease; surgery of tongue, palate, or larynx; interval poststroke to the date of videofluoro- scopic evaluation; and admiss ion Barthel scores were not significantly different for the aspiration vs. nonaspira t ion groups. The frequency of aspiration was not significantly different for patients with right vs. left hemisphere strokes. Those with bilateral lesions or bra ins tem lesions were signif icantly more likely to show aspiration on MBS than those with unilateral hemispheric stroke

(p < 0.05). The frequency of pneumon ia for those with vs

those without aspiration of any consis tency at the t ime of

J. Schmidt et al.: Aspiration Following Stroke 9

Table 1. Demographic features of the study population

Aspirators Nonaspirators (n = 26) (n = 33)

Age a (yrs) 73 + 2 70 • 2 Male/Female ratio 17/9 15/18 Previous stroke or TIA 9 (35%) 8 (24%) Previous pneumonia 0 (0%) 1 (3%) History of chronic obstructive 6 (23%) 4 (12%)

pulmonary disease History of surgery to the tongue, 2 (8%) 1 (3%)

palate, or larynx Site of lesion

Right hemisphere (19) 5 (26%) b 14 (74%) b Left hemisphere (28) 12 (42%) b 16 (58%) b Bilateral hemispheric or 9 (75%) b 3 (25%) b brainstem (12)

Post-stroke interval (days to date of video) 62 • 9 62 • 16

Admission Barthel score" 38 • 4 37 + 4 Mean follow-up interval (months 18 -+ 4 13 -+ 2

post stroke) a

aMean • SEM. bp < 0.05, X 2 analysis of 2 • 3 table.

Table 2. Frequency of pneumonia associated with videofluoroscopic evidence of aspiration (any amount, any consistency)

Aspirators Nonaspirators (n = 26) (n = 33)

Pneumonia 5 1 No pneumonia 21 32

Fisher's exact test p = 0.05. The odds ratio for developing pneumonia was 7.6 times greater for those with aspiration.

MBS was significantly different for the two groups (Fish-

er 's exact test p = 0.05) (see Table 2). Of the 26 patients

who aspirated any amount, irrespective of its consis- tency, 5 developed pneumonia. Of the 33 patients who did not aspirate, only 1 developed pneumonia. The odds

ratio for developing pneumonia was 7.6 times greater for

patients who aspirated any amount of test material irre- spective of its consistency compared with those without aspiration (95% confidence limit 1-70).

The relationship between aspiration of thickened liquids or more solid consistency test material and pneu-

monia is shown in Table 3. Of the 11 patients who aspi-

rated thickened liquid or more solid consistency test ma- terial, 3 developed pneumonia. Of 48 patients who did not aspirate or only aspirated thin liquids, 3 developed pneumonia. The difference between the two groups ap- proached statistical significance (Fisher's exact test p = 0.06). The odds ratio for developing pneumonia was 5,6 times greater for patients who aspirated thickened liquids or more solid consistencies compared with those

Table 3. Frequency of pneumonia associated with videofluoroscopic evidence of aspiration (thick liquids or more solid consistencies)

Aspiration of thickened No aspiration or liquids or more aspiration of thin solid consistencies liquids only (n = 11) (n = 48)

Pneumonia 3 3 No pneumonia 8 45

Fisher's exact testp = 0.06. The odds ratio for developing pneumonia was 5.6 times greater for those aspirating thickened liquids or more solid consistencies.

Table 4. Frequency of death associated with videofluoroscopic evi- dence of aspiration (thickened liquids or more solid consistencies)

Aspiration of thickened No aspiration or liquids or more aspiration of thin solid consistencies liquids only (n= 11) (n=48)

Dead 5 4 Alive 6 44

Fisher's exact testp = 0.01. Odds ratio = 9.2 times greater for those aspirating thickened liquids or more solid consistencies.

who did not aspirate or who aspirated only thin liquids. The 95% confidence interval for this odds ratio is from 1 to 28.

The frequency of dehydration was not statistically different for the two groups: 5 of 26 who aspirated any

amount or consistency vs. 3 of 33 without aspiration (Fisher's exact testp = 0.20).

Information on the relationship between evidence

of aspiration and death is listed in Table 4. Of the 11

patients who aspirated thickened liquids or more solid consistencies, 5 died. Of the 48 patients who did not

aspirate or who only aspirated thin liquids, 4 died. This difference between the two groups was statistically sig-

nificant (Fisher's exact testp = 0.01). The odds ratio for

death was 9.2 times greater for patients who aspirated thickened liquids or more solid consistencies compared with those who did not aspirate or who only aspirated thin liquids (95% confidence interval 1-16).

D i s c u s s i o n

Literature reviewed in the introduction underscores the minimal information available concerning medical com- plications associated with aspiration during MBS exami- nation. Our data provide a quantitative estimate of the risk of three of the most obvious complications: pneumo-

10 J. Schmidt et al.: Aspiration Following Stroke

nia, dehydration, and death. Our patient sample, stroke patients selected for inpatient rehabilitation, and the ret- rospective nature of this study do not allow us to generate prevalence rates for the outcome variables in question. Such an analysis would require a prospective commu- nity-based study. The odds ratio technique which we have used to analyze our data is less definitive than prev- alence data but is ideally suited for a case-controlled retrospective study such as is presented here.

The odds ratio for developing pneumonia was 7.6 times greater for those patients referred for MBS who aspirated any amount of test material irrespective of its consistency. We were unable to define a graded risk which increased with the amount of material aspirated. Our estimate of the amount of material aspirated was qualitative, assessing the amount as trace, moderate, or large. More quantitative MBS assessment scales do not exist.

The increased risk of pneumonia for those aspirat- ing thickened liquid or more solid consistency test mate- rial fell short of reaching statistical significance (p = 0.06). Our clinical experience indicates that the aspiration of thickened liquids or more solid consistency test material signals more severe dysphagia. This dis- crepancy between statistical and clinical significance probably reflects our small sample size (only 11 patients aspirated thickened liquids or more solid consistencies).

Our criteria for the diagnosis of pneumonia are based on both patient physical examination and radiolog- ical criteria. If only radiologically confirmed pneumonia is studied, one would expect to seriously underestimate the prevalence of aspiration pneumonia. Dysphagic pa- tients with fever, appropriate auscultatory changes on chest examination, and purulent sputum have tracheo- bronchitis irrespective of findings on chest x-ray film. The chest x-ray film is not expected to be positive unless there is significant bronchial-alveolar inflammation. The only previous study of pneumonia in stroke patients with vs. those without dysphagia also used either clinical or radiological evidence of pneumonia as their endpoint [2]. Our diagnosis of pneumonia is therefore consistent with clinical practice and with previous studies in this area.

We were unable to identify an increased risk of dehydration for those patients with aspiration vs. those without. As all patients with dysphagia received inten- sive intervention, including diet modification and train- ing in both direct and indirect treatment strategies, we may have reduced the prevalence of dehydration in our dysphagic patients. State of hydration may also have been influenced by other factors such as cognitive status, medications prescribed, or perception of thirst, not stud- ied in this protocol. There are no other studies of dehy- dration following stroke based on MBS evaluation with which to compare our results.

The risk of death was significantly increased only for the group that aspirated thickened liquids or more solid consistency test material. Cause of death could not be reliably determined based on phone follow-up with family members. Responses regarding cause of death were often vague or uncertain. The high death rate for those aspirating thickened liquids or more solid consis- tency test material may indicate more severe dysphagia, putting the patient at higher risk for developing pneumo- nia, inanition, and death.

There is one previous report of an increased fre- quency of aspiration for patients with right hemisphere stroke [ 11]. We did not find a difference in frequency of aspiration for patients with right vs. left hemisphere strokes. Our data did reveal a greater frequency of aspira- tion for patients with bilateral hemispheric strokes and for those with brainstem involvement. This has been reported previously by others [1].

All dysphagic patients in this study were enrolled in a dysphagia treatment protocol. The diet consistency and compensatory feeding techniques used were based on the MBS examination. The dysphagia therapist saw pa- tients 5 days per week during their inpatient rehabilitation hospitalization to teach and reinforce recommended treatment interventions. Our outcome results therefore reflect pneumonia, dehydration, and death for patients with stroke given focused dysphagia evaluation and treat- ment.

Acknowledgments. The authors thank Kathleen DePippo, M.S. for her comments and suggestions concerning this manuscript. This study was supported by U.S. Public Health service grant No. 1-R01-DC00885- 01, and funds from the Winifred Masterson Burke Stroke Research Fund.

References

1. Kuhlemeier K, Rieve J, Kirby N, Siebens A: Clinical correlates of dysphagia in stroke patients. Arch Phys Med Rehabil. 70: (special annual meeting issue, A-56) 1989

2. Gordon C, Hewer R, Wade D: Dysphagia in acute stroke. Br Med J 295:411414, 1987

3. Wade D, Hewer L: Motor loss and swallowing difficulty after stroke: frequency, recovery and prognosis. Acta Neurol Scand 76:50-54, 1987

4. Homer F, Massey E, Riski J, Lathrop D, Chase K: Aspiration following stroke: clinical correlates and outcome. Neurol 38:1359-1362, 1988

5. Linden P, Siebens A: Dysphagia: predicting laryngeal penetra- tion. Arch Phys Med Rehabi164:281-284, 1983

6. Veis S, Logemann J: Swallowing disorders in patients with cerebrovascular accident. Arch Phys Med Rehabi166:372-375, 1985

7. Barer D: The natural history and functional consequences of dysphagia after hemispheric stroke. JNeurol Neurosurg Psych# atry 52:236-241, 1989

J. Schmidt et al.: Aspiration Following Stroke 11

8. Sacco R, Wolf P, Kannel W, McNamara P: Survival and recur- rence following stroke: the framingham study. Stroke 13:290- 295, 1982

9. Logemann J: Evaluation and Treatment of Swallowing Disor- ders. Boston: Little Brown and Co., 1983

10. Rirnm A, Hartz A, Kalbfleisch J, Anderson A, Hoffmann R: Basic Biostatistics in Medicine and Epidemiology. London: Ap- pleton-Century-Crofts, 1980, pp 296-299

11. Robbins J, Levine R: Swal/owing after unilateral stroke of the cerebral cortex: preliminary experience. Dysphagia 3:11 - 17, 1988