State of the art in aspiration assessment and the idea of a new non invasive predictive test for the risk of aspiration in stroke

Abstract

Background

Dysphagia is a common but severe complication in acute stroke. New bedside tests are necessary to assess the risk of aspiration and avoid unnecessary alimentary restrictions but they must be completed by evaluated instruments.

Objective

We evaluated the prognostic validity of the recognition of forms (rf) test related to the assessment by speech and language therapists (SLT).

Design

Double blinded screening tests.

Setting

Geriatric department of general hospital.

Subjects

50 patients with acute stroke admitted consecutively.

Methods

All patients were assessed by SLT, mini mental status test, short test for assessing deficits of memory and attention and abilities of daily living-test. The rf-test was performed twice with a two-week interval.

Results

Assessment by SLT demonstrated aspiration in 29 of 50 patients. 18 patients failed the rf-test, giving a specificity of 90% (positive predictive value 89%), a sensitivity of 55% (negative predictive value 59%). 5 of 7 patients with aspiration pneumonia failed the test. Test-retest-correlation was 0,827. Performance in the rf-test was related significantly to results of the activities of daily living-test.

Conclusions

Bedside tests to exclude aspiration in patients with acute stroke are necessary. No test so far combines high sensitivity with acceptable specificity. Compared to other bedside tests the rf-test has a high specificity while sensitivity is low. On behalf of good predictivity rf-test may be an interesting supplement to swallow tests in case of stroke and seems to be a candidate for more extended studies.

Introduction and review of state of the art

Acute stroke results in dysphagia in 37 – 78 % of patients, acute hemispheric stroke in 39 – 40 % (1). The frequency of swallowing problems increases with stroke severity (2, 3). Dysphagia resolves in up to 75 % of patients within 3 months (4). The consequences of dysphagia include dehydration, aspiration pneumonia and airway obstruction (5). Dysphagia during the first week after stroke is associated with poor outcome, especially during the first year after stroke (6, 7, 8). Nevertheless, prognostic value of dysphagia seems to be independent of the presence of aspiration (9).

Patients’ risk to develop pneumonia is increasing up to 4-fold with laryngeal aspiration, by 10-fold with tracheobronchial aspiration and by 13-fold with silent aspiration (10, 11). For aspiration pneumonia, mortality rates from 20 – 65 % have been reported (11).

Instituting a formal dysphagia screening protocol for all patients admitted with stroke can decrease the risk of pneumonia. The relative risk reduction for pneumonia exceeds 80 % and the relative risk reduction for mortality reaches 70% (12, 13, 14).

Before swallowing assessment, patients with acute stroke should not receive oral feeding or liquids. In consequence, early diagnosis of aspiration is essential.

Instrumental examination

According to the ACCP guidelines (5), patients with dysphagia should undergo videofluoroscopic swallow examination (VF) or fiberoptic endoscopic evaluation of swallowing (FEES). The VF, as adapted from the barium swallow by Logemann (15) has been considered to be the gold standard for a long time. In this dynamic study anatomic structures and their function in the oral and pharyngeal phases of swallowing can be examined. The patient is placed at 45° or 90° upright and asked to consume foods or drinks of defined consistencies mixed with barium. VF protocols vary, but should include testing potential compensatory techniques. VF is difficult to interpret and depends largely on the experience of the operator. The radiation exposure is acceptable for a diagnostic tool (16) but renders VF inappropriate for frequent repeat tests to monitor changes. VF protocols deviate from the usual conditions on wards, especially in posture. This may decrease diagnostic accuracy. Clayton et al. (17) recently demonstrated aspiration as a drop < 5.5 pH in tracheal monitoring after an acidic meal in 9 out of 32 stroke patients studied, all receiving a diet considered safe after VF.

Since 1988, when Langmore (18) formalized an examination of swallowing by flexible laryngoscopy, FEES has gradually become more widely used then VF (19). In 1998, Aviv et al. (20) augmented FEES with sensory testing into FEEST. In FEES and FEEST, aspiration cannot be seen directly, it is inferred from residue left after swallowing or ejected after coughing. The examination can be conducted at the bedside and has been considered as within the scope of practice of SLTs by the UK Royal College of Speech and Language Therapists (21). Just as VF, FEES and FEEST depend largely on the experience and skill of the assessor.

The U.S. Agency for Health Care Policy and Research published a report on the assessment of swallowing disorders in 1999 and concluded that neither VF nor FEES had been demonstrated as superior (22). Since then, reliable evidence as to which instrumental examination is more valuable in the assessment of dysphagia is still scarce. Full agreement between both tests can be expected in about half the patients, major disagreement in about one third (23). In a randomized trial, the incidence of pneumonia in the year following the examination was comparable for VF and FEES (n = 63 for each examination). In the subgroup of patients with acute stroke, the incidence of pneumonia was significantly less after FEES.

In both examination techniques, there is a considerable inter-observer variability. Acceptable levels of reliability from one observer to the next are reported for the rating of aspiration, in VF as well as FEES (24, 25).

Bedside swallow assessment

The main disadvantage of instrumental examination of swallowing is that these techniques are time consuming and not always readily available. Bedside tests are the key to early assessment of aspiration. They serve to assess the suitability of oral feeding and limit the number of patients who need to go on to instrumental examinations (26).

Several studies reported a correlation of weak cough, absent gag reflex, oral function or voice change with aspiration, though results have been limited by small patient numbers and limited statistics (26). The gag reflex alone is unreliable as it can be absent in as many as 44 % of healthy older volunteers (27). An intact gag reflex may be protective against longer-term swallowing problems (28). However, in a factor analysis of oral and pharyngeal function tests in 61 consecutive stroke patients, only cough and / or voice change on water swallowing belonged to the same factor as aspiration on VF (29).

Bed side tests have focused on observed water swallowing of different volumes, looking for dribbling, laryngeal movement, cough, voice change, and the time taken to finish the drink (30, 31, 32). These tests demonstrate aspiration rather than predicting it, consequently they tend to overlook silent aspiration. The prevalence of patients with silent aspiration has been reported as ranging from 20 – 72 % (33, 34). When tested against VF the sensitivity of water swallow tests ranged from 47 to 80 %, the specifity from 68 to 91 %. Until now there was no test combining high sensitivity with acceptable specifity. Interobserver and intraobserver reliability levels varied greatly and tended to be less for physicians and better for speech and language therapists (SLTs) (26). For a more detailed SLT-assessment related to videofluoroscopy, Mann et al. (35) demonstrated a sensitivity of 93% and a specificity of 63% in 128 patients with first onset of acute stroke without impairment of conscious level.

Recently, a swallowing provocation test has been developed, evaluating swallow time after instillation of water into the oropharynx just beneath the velum. This test fails to evaluate the oral phase of swallowing. As compared to FEES, the test rendered a sensitivity of 74 % and specificity of 100 % in 100 consecutive patients with acute stroke. Interestingly, the sensitivity was significantly decreased when a larger volume was instilled, raising doubts about the test design (36, 37).

Studies of the auscultation of swallowing suggest that audible cues to identify patients at risk for aspiration may be present. But inter- and intra-rater variability is too high to recommend the technique as a screening test (38, 39).

Trapl et al. (40) suggested a more detailed swallow protocol for the first assessment. Their GUSS-Tests starts with a preliminary assessment (vigilance, voluntary cough, deglutition of saliva, drooling, voice change) followed by repeated swallows of semisolid liquid (thickened water), water and solids, observing for deglutition, involuntary cough, drooling and voice change. The test is designed to be applied by nurses and scored in a point system demonstrating sensitivity of 100 % and a specificity of at least 50 % compared to FEES in the first 50 patients reported.

Pulse oximetry is a different approach to bedside aspiration assessment. Compared to swallow assessment, it should require less experience in the examiner and be able to detect silent aspiration. Hypoxia and desaturation after abnormal swallow has been attributed to reflex bronchoconstriction or poor breathing, resulting in ventilation-perfusion mismatching. Cutoff points for desaturation vary. For desaturation of > 2 % compared to VF, sensitivity ranged from 73 – 87 %, specificity from 39 – 87 % (26). One study compared pulse oximetry to FEES and reported better prognostic values. Positive predictive values were increased to 95 % by combining pulse oximetry with a bedside swallow test (41). In confirmatory studies there was no correlation between desaturation < 2 % and aspiration in VF (42, 43). In a larger study, Ramsey et al. (44) recently compared bedside swallowing tests and pulse oxymetry to VF in 189 patients with acute stroke. They reported sensitivity / specificity of 47 % / 72 % for bedside swallow assessment and 33 % / 62 % for desaturation. Combining both tests did not improve predictive values. Interestingly, none of 6 patients with silent aspiration suffered desaturation.

All methods used in bedside swallowing assessment diagnose the manifest aspiration, they do not predict aspiration. Neurological considerations suggested an evaluation of the test of oral stereognosis for the predictive assessment of aspiration in patients with acute stroke. Tests for oral stereognosis were developed in the 1960’s (45), in the 1970’s Landt (46) reevaluated these tests for patients with dental plates.

Methods

We examined 50 patients with acute stroke admitted consecutively to the geriatric department of the St. Bonifatius-Hospital in Lingen from March 1999 to January 2001. Patients from the age of 60 who consented to the study were included, without regard to conscious level, cognitive impairment, multiple handicaps and accompanying illness. Patients were excluded only if they did not cooperate in the recognition of forms-test (rf-test). Most patients excluded demonstrated global aphasia with severe impairment of lalognosis or severe cognitive impairment due to dementia.

Seven days after admission, aspiration was assessed by a specially trained SLT. Results of SLT-assessment were unknown to the examiner performing rf-tests. Rf-tests were performed promptly after SLT-assessment, the longest interval being seven days.

In the rf-test, 10 metal test forms of varied shapes with maximum size of 8 mm are placed in the patient’s mouth. The patient is asked to determine the shape of the test form with his tongue and identify the shape in a set of enlarged models of the test forms (figure 1). The rf-test forms were presented in random sequence. Patients passed the test if they identified 5 out of 10 shapes correctly. Aspiration of test forms was prevented by a safety-thread. The rf-test was performed twice with a two-week interval.

Figure 1.

Display of enlarged test forms

Immediately after the first rf-test the mini mental status test (MMSE by Cockrell and Folstein (47)), the short cognitive performance test (SKT by Erzigkeit (48)) and the activities of daily living-test (ADL = Barthel Scale by Lawton and Brody) (49)) were performed.

Relations between the data were analysed by Spearman test, differences between the first and second rf-tests by Wilcoxon tests for matched pairs. Specificity, sensitivity and predictive value were calculated for rf-test.

Results

We studied 50 patients with acute stroke (24 women, 26 men) aged 60-90 (median 78) years. SLT-assessment detected aspiration in 29 patients. 18 patients failed to identify at least 5 of 10 test forms in the rf-test by Landt (table 1).

Table 1.

Test results of patients with and without aspiration according to SLT-assessment

	>5 test forms identified	<5 test forms identified
Aspiration	13	16
No aspiration	19	2

The rf-test gave a sensitivity of 55%, a specificity of 90% related to SLT-assessment. The positive predictive value was 89%, the negative predictive value 59%. Seven patients developed aspiration pneumonia. Five of these failed in the rf-test.

Test-retest correlation for the number of correctly identified test forms in the rf-test was 0,827. Conditioning was demonstrated for the time needed to complete the test (241±155s for the first test versus 216±147s for the second test, p = 0,018).

The number of correctly identified test forms was significantly related to the results of the ADL (R=0,400, p=0,014). The time needed to complete the rf-test (R=0,603, p=0,00088) and the average time needed to correctly identify a test form (R=0,538, p=0,0038) were significantly related to the results of the MMSE.

Additionally, results of MMSE and SKT were significantly interrelated (table 2). No relations were found between the results of ADL and MMSE or SKT.

Table 2.

Correlations between results of performed tests (IF: identified test forms, Spearman tests:

	Rf-test Number IF	Rf-test Time / IF	MMSE	SKT	ADL
Rf-test, time	0,327*	0,8573	0,6022
Rf-test, number CF		-	-	-	0,400 *
Rf-test time / CF			0,5381	-	-
MMSE				-0,6112	-
SKT					-

*=p<0,05, 1=p<0,01, 2=p<0,001, 3=p<0,000001)

Conclusion

As dysphagia is common in acute stroke, we have to assess the aspiration risk and the suitability for oral feeding close to admission to the hospital.

VF and FEES are well validated instrumental investigations for assessment of aspiration in patients with dysphagia that provide details of anatomy and physiology of the swallowing process as well as information on potential therapeutic procedures. Unfortunately patients are exposed to radiation in VF, and the methods are not universally available. Instrumental assessment of aspiration risk in all patients with acute stroke cannot be achieved outside study settings.

For daily practice, simple bedside tests that can be administered by a range of health professionals are needed. Simple examination standards like abnormal gag reflex or reduced laryngopharyngeal sensation are associated with aspiration but do not predict the individual risk.

Bedside swallow tests are more reliable and can be performed by phyicians, SLTs and nurses. However, sensitivities and specificities of these tests vary considerably. Good sensitivity combined with an acceptable specificity, as would be expected in a screening test, has not been reliably demonstrated in any of the previous tests. A pragmatic dysphagia management considering available resources and experience is urgently needed.

Assessment of a water swallow of small and, if successful, larger volumes while observing for dribbling, laryngeal movement, time taken to finish the drink and especially cough and voice change seems to be the most wide spread bedside test used. Volumes are increased slowly. 30 ml seem to be enough, the test results do not improve with greater volumes [29]. Patients with silent aspiration will be missed. Pulse oximetry seems to give similar results. Whether a combination of water swallow test and pulse oximetry improves the diagnostic accuracy remains questionable. Pulse oximetry may not be as helpful as expected in diagnosing silent aspiration. It should be kept in mind that neither test has a sensitivity or specificity high enough to be reliable.

Patients with clinical symptoms of dysphagia or doubtful findings should be kept on “nil by mouth” until further swallowing assessment either by SLTs or by VF or FEES has been accomplished. This will usually mean that the patient is hydrated intravenously for the interval, as the FOOD trials did not show a significant advantage of early tube feeding [50].

The rf-test is a predictive test for aspiration with a specificity superior to SLT assessment. The test can be administered by physicians, nurses or SLTs and requires little experience in the examiner. It may serve to examine patients where dysphagia is suspected. If less than 5 forms are recognized, aspiration may be assumed to be present. Patients who pass the test may then be tested with non-predictive bedside swallow assessments.

Though bedside tests for aspiration are necessary in acute stroke, but until now no test as compares good sensitivity with acceptable specificity. This limitation should be kept in mind when applying bedside tests. More reliable tests are needed, and research must continue in this field.

Details of sources of research funding or any possible conflict of interest. The study was performed without any external sponsoring. Details of informed consent of patients studied and approval of an ethics committee. All patients including those with (partial) aphasia gave their informed consent to participate in the study.

Financial disclosure: None of the authors had any financial interest or support for this paper.