Implementation and Analysis of a Free Water Protocol in Acute Trauma and Stroke Patients

We enrolled participants for this study from October 2012 to September 2015 and recruited them from a large urban acute care hospital (> 1000 beds) with a level I trauma center and primary stroke center. A subset of these participants was discharged within the hospital system to inpatient rehabilitation and was followed throughout their rehabilitation stay. We obtained informed consent or assent from each participant or the participant’s legally authorized representative. We acquired institutional review board approvals from the hospital system and collaborating university.

Attending physicians, nurses, or SLPs from the trauma or stroke critical care units or floors identified potential participants. Inclusion criteria were (1) English- or Spanish-speaking individuals aged 18 years and older; (2) admission with a primary diagnosis of acute stroke or acute trauma; (3) admittance to an acute care unit; (4) documented aspiration of thin liquids according to an objective examination (either a fiber-optic endoscopic evaluation of swallowing or a modified barium swallow study) obtained in the standard course of clinical care; and (5) mental status adequate to support oral intake of thin liquids as judged by a certified SLP. Exclusion criteria were (1) secondary diagnosis of congestive heart failure, Parkinson disease, or moderate to advanced dementia as determined by a neurologist; (2) physician-ordered fluid restrictions; (3) documented active oral infection (eg, thrush); and (4) excessive and/or uncomfortable coughing during attempted ingestion of liquids as noted by the SLP during objective or clinical evaluation. Criteria for dismissal from the study were discharge from the hospital system, diet upgrade to thin liquids, and request from the participant or the treating physician for withdrawal from the study.

Eligible patients received modified diets (by mouth or enteral feeding) and dysphagia management consistent with national best practice guidelines and the hospital’s standard clinical protocol. This protocol included facilitative oropharyngeal exercises, compensatory strategies, and dietary modifications including thickened liquids during meals and oral medication administration. We also administered to eligible patients the Aspiration Precaution Oral Care Program, which was previously developed and implemented by hospital nursing, SLP, and respiratory therapy teams for nonventilated, non– intensive care unit patients with dysphagia to mitigate hospital-acquired respiratory infections linked to bacterial colonization in the mouth and teeth (Figure 1).^30–37 

After obtaining consent or assent, we randomized participants via odd or even numbers from sealed envelopes into either (1) a control group with no access to thin liquids, including water or ice chips; or (2) an experimental group with access to water and ice chips, according to guidelines specified in the FWP. A physician order for aspiration precautions, modified diet, and/or participant access to ice and water was entered into the medical record. The research SLP placed written precautions concerning group assignment in each participant’s room to meet hospital requirements and to ensure adequate communication among all involved in participant care. As a result, blinding of the on-site researchers was not possible.

The FWP guidelines were modeled after the protocol implemented at Frazier Rehabilitation Center (Table 1). For participants in the experimental group, nursing and allied health staff offered water or ice with supervision as needed throughout each shift. To ensure adherence to FWP protocol guidelines, study investigators used in-services and written material to educate nursing staff on the purpose of the study and implementation of the FWP. Attendance was documented. The research SLP continued to provide individualized education sessions to participants, caregivers, and nurses when necessitated by staff turnover. See Figure 2 for additional information on implementation of the FWP in this study.

The research SLP entered information from the participant’s chart onto the data collection form. Patient information included date of admission, age, sex, primary and secondary diagnoses, ambulatory status, current diet, date and type of objective examination confirming aspiration of thin liquids, results of subsequent objective examinations, and date of dismissal from the study. The research SLP also recorded adverse events or significant changes in medical status, which were anticipated because of the severity of illness in this patient population. When a participant’s physician documented an adverse event in the medical record, the research SLP recorded the information on the data collection form, and the information was reported to and reviewed by the institutional review board.

Documenting water intake over the course of the study would have been ideal, but limited resources precluded this. Moreover, the hydration status of patients in trauma and stroke units is routinely monitored, and physicians recommend intravenous fluids as necessary. Whether hydration changes in the experimental group resulted from intravenous fluid administration or from increased water and ice intake was thus impossible to determine.^21,38 

The research SLP recorded positive and negative clinical outcomes daily during the business week. Events occurring during a weekend were recorded on the following Monday.

The 3 positive clinical outcomes were number of diet upgrades, fewer days to a diet upgrade, and fewer days in the study. Data recorded included diet on the date of consent to enter the study, diet upgrades, date of diet upgrade, and number of days between consent and dismissal from the study.

The 3 negative clinical indicators were intubation, pneumonia, and diet downgrade. Dates of intubation and pneumonia diagnosis were confirmed via physician documentation in the medical record. A diagnosis of pneumonia triggered completion of a follow-up form, which included confirmation of the diagnosis via chest radiography or objective examination, reason for the pneumonia (such as aspiration of liquids, solids, or tube feeds) as determined by the physician, and details of oral hygiene such as dependence on caregivers for oral hygiene and nonadherence with the oral care protocol.

We conducted a power analysis with a log-rank test based on a pilot study in which the mean number of days to a diet upgrade was 7 in a control group and 4 in an experimental FWP group. We determined that 80% power with a significance level of .05 would require 53 patients per group.^28,29  We compared patient demographic data, clinical information, positive clinical outcomes, and negative clinical indicators by using t tests or Wilcoxon tests for numerical variables, log-rank tests for time to upgrade, and c² or Fisher tests for categorical variables. We used multiple regression analysis to further analyze the positive clinical outcomes for all participants and for participants who received nothing by mouth. We used logistic regression to analyze diet upgrade, the Cox proportional hazards model to analyze days to diet upgrade, and negative binomial regression to analyze days in the study. All models included group assignment, diagnosis group, ambulatory status, and age as independent variables. We set a at .05 for each of these measures. Study personnel who entered, audited, and analyzed the data were all blinded to group assignment. We performed all analyses with data analysis software (SAS 9.4, SAS Institute Inc).

One hundred six patients consented to participate in the study. Of these, 70% had a trauma diagnosis and 30% had a stroke diagnosis. Two of the 106 participants, both from the experimental group, did not complete the study. Before data collection began, 1 participant was discharged and 1 participant died for reasons unrelated to the study, according to the physician review.

Table 2 shows demographic information and diagnoses of participants in the control and experimental groups. Age, sex, and primary diagnosis were not significantly different between the 2 groups. However, among the secondary diagnoses, the number of patients with a tracheostomy speaking valve was marginally higher in the experimental group than in the control group, suggesting that patients in the experimental group may have been more critically ill than those in the control group.

As expected, adverse events occurred during this study. The adverse events were 4 instances of mucous plugs, 2 changes in neurological status, 1 vertebral artery dissection, and 1 ileus and vomiting. These 7 participants, in conjunction with their adverse events, presented with negative clinical indicators. There were 3 occurrences of pneumonia in the experimental group and 2 occurrences in the control group. Physicians attributed these adverse events and negative clinical indicators to participants’ medical status rather than to study variables; therefore, the counts are excluded from our experimental and control group comparisons. Of note, 4 of the participants who had adverse events had a trauma diagnosis of spinal cord injury.

A subset of 15 participants, 6 in the control group and 9 in the experimental group, was discharged from acute care to inpatient rehabilitation and was followed until discharge from rehabilitation to collect data along the continuum of care. Seven of these patients (47%) had a trauma diagnosis and 8 (53%) had a stroke diagnosis. The ages of patients in the 2 groups did not significantly differ (mean [SD] ages: control group, 54.5 [26.6] years; experimental group, 54.4 [14.6] years). Men outnumbered women in both the control group (5 vs 1) and the experimental group (7 vs 2).

Positive clinical outcomes did not differ significantly between the 2 groups (Table 3). None of the regression models showed significant differences between the control and experimental groups (Table 4). Patients with trauma were more likely than those with stroke to have a diet upgrade and they also upgraded their diet in fewer days, whether the analysis included all patients or only those who began receiving nothing by mouth. Older age tended to be associated with lower odds of a diet upgrade, although this finding did not reach significance. Among patients who began receiving nothing by mouth, age was significantly associated with more days in the study.

Negative clinical indicators, unrelated to adverse events, were infrequent and did not differ significantly between the 2 groups, but this result must be interpreted with caution because statistical power to detect these infrequent events was likely low. Only 1 diet downgrade occurred, in a patient in the experimental group.

The 15 patients followed through their rehabilitation stays remained in the rehabilitation setting for a mean of 16 days (both groups). Five of 6 participants from the control group (83%) and 8 of 9 participants from the experimental group (89%) experienced a diet upgrade. Of the patients who were receiving nothing by mouth at the onset of the acute setting study, 2 of 3 from the control group (67%) and 9 of 9 from the experimental group (100%) were upgraded to an oral diet in the rehabilitation setting. No participants in either group experienced a diet downgrade, intubation, or pneumonia in the rehabilitation setting.

Among the 15 participants followed into rehabilitation, the mean number of days in the study from acute care admission through rehabilitation was 27. Participants in the experimental group remained in the study for a mean of 22 days; those in the control group remained in the study for a mean of 29 days.

Our study compared positive clinical outcomes and negative clinical indicators in 104 participants receiving a standard clinical protocol for dysphagia who were randomized either to have no access to thin liquids or to have access to water and ice under the guidelines of the FWP. We found no significant group differences in the 3 positive indicators. Diet upgrades, median days to diet upgrade, and median days in the study were similar in the 2 groups. With respect to negative clinical indicators, 1 patient in the experimental group experienced a diet downgrade.

Although the groups did not differ significantly in positive or negative clinical outcomes, 8 adverse events occurred in 7 participants over the course of the study. Four of the 7 participants had spinal cord injuries, and 5 of the 8 adverse events were attributed to these injuries. This finding is consistent with the critical illness and respiratory complications that typify this group of patients. Although these events were deemed to be related to patient diagnosis rather than to the FWP, further investigation of the FWP in this patient population is warranted.

In the subset of participants followed into the rehabilitation setting, our results showed that both the experimental and control groups experienced positive clinical outcomes. As expected, these patients, who were more medically stable than the others, experienced no negative clinical indicators during their stays. This result is consistent with previous research on the FWP in the rehabilitation setting.^1,20–25  Participants in the experimental group showed a trend toward decreased days to discharge from the study, but the small number of participants that were followed precluded statistical testing.

This study had limitations. Although our sample size was large relative to that of other FWP studies, the power calculation derived from our pilot studies was based on only 1 outcome, the number of days to diet upgrade.^28,29  Because effect sizes in our study were relatively small, statistical power of the study was low despite the size of our sample. Accordingly, additional studies will be needed before conclusions about the FWP can be drawn.

Two other limitations were inherent to studies conducted in the acute care setting: short stays and challenges in blinding on-site research staff to group assignment. Consistent with the short stay that is typical of the acute care setting, participants in this study were monitored for an average of 6 days after enrollment. Although lung changes have been reported as early as 48 hours after FWP implementation, monitoring outcomes over a longer time would be ideal.²⁰  We tried to address this limitation by following a subset of participants through their rehabilitation stays, but additional evidence is needed. The need to ensure communication and adhere to hospital requirements prevented us from blinding on-site research staff members to participant group, but off-site staff members, including the principal investigator, research assistants, and biostatistician, were blinded to patient group throughout the study.

The FWP has the potential to enhance patient care in the acute care setting. However, larger-scale studies are needed to reach definitive conclusions on the positive outcomes and negative clinical indicators of the protocol. As with our study, future research will require adherence to oral care and free water protocols and education of multidisciplinary staff members. Communication and collaboration among nurses, physicians, and SLPs will be necessary to monitor participant status across multiple floors and units. Targeting specific acute care populations at varying severity levels, possibly across multiple sites, could further establish an evidence base for evaluating the advantages and disadvantages of the FWP in the acute care setting.

The authors thank the medical staff at the Baylor University Medical Center and the research assistants from the University of Texas at Dallas. The authors also thank the research participants in the study as well as the following individuals for their input: Melissa Gibson, MA; Megan Flynn, MS; Miranda Tillson, MS; Erin O’Neill, MS; Kater Taylor, MS; Laura Harris, MS; Katlyn Nickels, BS; Simon Driver, PhD; Jan Lougeay, MS; Robert Stillman, PhD; and Thomas Campbell, PhD.