Effectiveness of Nursing Interventions to Prevent Dry Eye in Critically Ill Patients

Diego Dias de Araujo, PhD, MSN, RN; Daniel Vinicius Alves Silva; Carolina Amaral Oliveira Rodrigues; Patricia Oliveira Silva; Tamara Goncalves Rezende Macieira, BSN; Tania Couto Machado Chianca, PhD, MSN, RN


Am J Crit Care. 2019;28(4):299-306. 

In This Article


This study was registered in ClinicalTrials.gov (Identifier: NCT02767258) and in the Brazilian Clinical Trials Registry (ReBec) (Identifier: RBR-5r8syp). Ethical approval was obtained from the institutional review board of the Universidade Federal de Minas Gerais before the study was begun. We followed the Consolidated Standards of Reporting Trials (CONSORT) guidelines for nonpharmacological interventions.[11] Written informed consent was obtained from each patient's family member or next of kin before recruitment.


This was a double-blind (patients, outcome assessor) randomized controlled trial with 2 parallel groups. The data reported here were collected between January 14, 2016, and March 14, 2017, in a 10-bed ICU at a large tertiary care, nonprofit hospital in Brazil.

Patients recruited for the study met the following inclusion criteria: age of 18 years or older, no diagnosis of dry eye at ICU admission, receipt of mechanical ventilation, blink rate of less than 5 times per minute, and a Glasgow Coma Scale score of 7 or lower.[2] Patients were excluded if they had an ICU stay of less than 48 hours or were admitted to the unit with a diagnosis of brain death. Failure to document the delivery of nursing interventions (liquid artificial tears or artificial tears gel) at the correct time resulted in the participant's exclusion from the study and discontinuation of treatment.

Sample Size and Randomization

We performed a pilot study involving 30 patients between November and December 2015, with 10 patients allocated to each of 3 groups (liquid artificial tears, artificial tears gel, and 0.9% sodium chloride solution), to estimate sample size. In the pilot study, 40% of the patients treated with liquid artificial tears had dry eye develop, compared with 10% of those treated with artificial tears gel (P = .01).

Power analysis using the proportion of unfavorable results in the pilot study (40%), a significance level of .05, power of 80%, and a relative risk (RR) of 0.5 in favor of artificial tears gel (or RR reduction of 20%) resulted in an estimated sample size of 134 patients: 67 patients for each of the 2 intervention groups. If any participants were lost during the study, more would be recruited until at least 67 patients were allocated to each group.

The initial study population consisted of 546 medical or surgical patients who had been admitted to the ICU of the target hospital. Of the 546 patients assessed for eligibility, 406 were excluded according to the inclusion and exclusion criteria. The resulting final sample consisted of 140 patients, 70 in each group (see Figure).


Flowchart illustrating the 4 phases of the study, following the Consolidated Standards of Reporting Trials (CONSORT) recommendations for nonpharmacological interventions.

Despite allocation to a third group in the pilot study, we decided not to treat patients with 0.9% sodium chloride solution in this study. The results of the pilot study showed that 60% of patients allocated to this intervention group had dry eye develop. The findings of previous studies support our decision not to use 0.9% sodium chloride solution as a study intervention.[1,12]

A statistician performed block randomization of patients using the computer software R-3.2.3. The randomized list was subdivided every 10 patients into 2 groups in a 1:1 ratio. The list was sent directly to the research coordinator of the study (T.C.M.C.) and to 2 undergraduate research assistants (D.V.A.S., C.A.O.R.) who were responsible for the allocation of the patients.

"Two lubricant eye drops were used as interventions: liquid artificial tears (Lacribell) and artificial tears gel (Visidic Gel)."


Two types of lubricating eye drops—liquid artificial tears (Lacribell) and artificial tears gel (Vidisic Gel)—were used as the study interventions. After a patient was recruited for the study, the ICU nurses were notified through an information center which of the 2 interventions would be used for that patient. The intervention was prepared by a nurse and stored in a brown envelope. The ICU's nursing technicians delivered the intervention twice a day (at 8:00 AM and 8:00 PM) for 5 consecutive days. The nursing technicians followed a protocol for cleaning the patient's eyes with 0.9% sodium chloride before administering 2 drops of the predetermined intervention to each eye.

Before the study was begun, we trained the nursing team in the study protocols and procedures. The training consisted of an explanation of the study problem; an overview of the study methods; description of the inclusion and exclusion criteria; instruction on when to discontinue the interventions; explanation of informed consent, its importance, and how to obtain it; and the techniques for application of each intervention. To increase the chances of recruiting participants, nurses were given the responsibility for obtaining informed consent because of the study personnel's inability to be present on the unit for 24 consecutive hours.

"The ocular assessment consisted of the Schirmer test and the fluorescein test."


The study outcome was the development of dry eye. Potentially confounding variables included in the data analysis were age, sex, unit of origin, Nursing Activities Score, Acute Physiology and Chronic Health Evaluation II score; patient type (medical condition only or postsurgical), death, length of stay, referral unit, medical diagnosis at admission, sedation, Ramsay Sedation Scale score, Glasgow Coma Scale score, intubation, tracheostomy, mechanical ventilation, days of mechanical ventilation, mode of mechanical ventilation, fraction of inspired oxygen, positive end-expiratory pressure, other ventilatory assistance device, blink rate per minute, ocular surface exposure, edema, severity of corneal ulcer, medications, and positioning (degree of head elevation).

Data Collection

On 5 consecutive days, one of the researchers (D.D.A.) collected data and performed ocular assessment for each participant included in the sample. This 5-day period was established on the basis of the reported mean time of 3.5 days for development of dry eye in critically ill patients.[5] Before ocular assessment, the nursing technicians cleaned the patient's eyes with 0.9% sodium chloride solution to remove any traces of the intervention substances, ensuring that the researcher remained unaware of the treatment allocation of each patient.

The ocular assessment consisted of the Schirmer test and the fluorescein eye stain test. The Schirmer test was used to analyze tear volume. This test involved placing a strip of Whatman filter paper grade 41 or 50 measuring 5 mm wide and 35 mm long with the tip folded (about 5 mm) in the bottom of the lower conjunctival sac in the temporal region (outer corner of the lower eyelid). After 5 minutes, the strip was removed and the moistened part was measured and the result documented.[13] The fluorescein eye stain test was used to evaluate the cornea for possible abnormalities. A drop of fluorescein was placed in each of the patient's eyes; after 1 to 2 minutes, under low-light conditions, the cornea was examined using an ophthalmoscope with a cobalt blue light filter and a magnifying glass.[13]

Data Analysis

Two of the researchers (C.A.O.R., P.O.S.) independently entered the data into the Epi Info software program, version 3.5.1. The data entered were checked for consistency and then extracted and analyzed in the R-3.2.3 software. Frequency, central tendency (average), and standard deviation were measured. Categorical variables in the 2 intervention groups were compared using the Fisher exact test. Continuous variables were compared using the Mann-Whitney test. The assumption that the distribution of the continuous variables was normal was tested using the Shapiro-Wilk test. The incidence of dry eye and the effect of the nursing interventions were analyzed using the Fisher exact test. The results were presented with a 95% CI. Poisson regression was used to present the results, with the model adjusted for potential confounders (the risk factors of age, sex, and ocular surface exposure). Statistical significance was set at P ≤ .05.