Short-Term Therapeutic Adherence of Hospitalized Older Patients with Oropharyngeal Dysphagia after an Education Intervention: Analysis of Compliance Rates, Risk Factors and Associated Complications

Oropharyngeal dysphagia (OD) is associated with adverse outcomes that require a multidisciplinary approach with different strategies. Our aim was to assess the adherence of older patients to dysphagia management recommendations during hospitalization, after a specific nurse guided dysphagia education intervention and to identify short term complications of OD and their relationship with short-term adherence. We carried out a prospective observational study in an acute and an orthogeriatric unit of a university hospital over ten months with a one-month follow-up. Four hundred and forty-seven patients (mean age 92 years, 70.7% women) were diagnosed with dysphagia using Volume-Viscosity Swallow Test (V-VST). Compensatory measures and individualized recommendations were explained in detail by trained nurse. Therapeutic adherence was directly observed during hospital admission, after an education intervention, and self-reported after one-month. We also recorded the following reported complications at one month, including respiratory infection, use of antibiotics, weight loss, transfers to the emergency department, or hospitalization). Postural measures and liquid volume were advised to all patients, followed by modified texture food (95.5%), fluid thickeners (32.7%), and delivery method (12.5%). The in-hospital compliance rate with all recommendations was 37.1% and one-month after hospital discharge was 76.4%. Both compliance rates were interrelated and were lower in patients with dementia, malnutrition, and safety signs. Higher compliance rates were observed for sitting feeding and food texture, and an increase in adherence after discharge in the liquid volume and use of thickeners. Multivariate logistic regression analysis showed that adherence to recommendations during the month after discharge was associated with lower short-term mortality and complications (i.e., respiratory infection, use of antibiotics, weight loss, transfers to the emergency department, or hospitalization). One-third of our participants followed recommendations during hospitalization and three-quarters one month after admission, with higher compliance for posture and food texture. Compliance should be routinely assessed and fostered in older patients with dysphagia.


Study Design
This was a prospective, observational, pragmatic study on patients consecutively admitted to an acute geriatric unit or an orthogeriatric unit of a university hospital over ten months, with one-month follow up, to describe the therapeutic adherence of older patients with OD. During admission, patients received OD management recommendations according to clinical dysphagia assessment. The study included at least two visits during hospitalization and a telephone assessment one month after discharge, as it is presented in Figure 1.

Study Population
We included 447 older patients between October 2018 and July 2019. Inclusion criteria: patients over 70 years old admitted to an acute geriatric or orthogeriatric unit with OD diagnosis during admission with the Volume-Viscosity Swallowing Test (V-VST) [12]. Exclusion criteria: end-of life situations, permanent low level of consciousness that precluded performance of the V-VST, enteral nutrition administered by nasogastric tube or percutaneous endoscopic gastrostomy, and refusal to participate. The study population inclusion procedure has also been published elsewhere [2,19].
All patients were informed about the study and gave their written consent. The study protocol was approved by the ethics review board of Hospital Universitario Ramón y Cajal (protocol Number 120/18) and was conducted following the principles and rules of the Declaration of Helsinki. percutaneous endoscopic gastrostomy, and refusal to participate. The study population inclusion procedure has also been published elsewhere [2,19].
All patients were informed about the study and gave their written consent. The study protocol was approved by the ethics review board of Hospital Universitario Ramón y Cajal (protocol Number 120/18) and was conducted following the principles and rules of the Declaration of Helsinki.

Health Status, Comorbidity and Geriatric Assessment, and In-Hospital Evolution
During hospital admission, a trained nurse collected demographic information, previous comorbidities and pharmacological treatment related to OD, geriatric assessment measures, the reason for hospitalization, presence of delirium during hospitalization, length of stay, and place of discharge, previously described in detail [2,19].
Regarding geriatric assessment, the nurse collected functional status with Barthel index of basic activities of daily living (ADL), Lawton index of instrumental ADL and Functional Ambulation Classification of gait; cognitive status using Global Deterioration Scale; and nutritional status through Mini Nutritional Assessment-Short Form and Body Mass Index, and calculated the body mass index.

OD Assessment
All patients were clinically assessed for OD by a trained nurse by a general exam and the V-VST, within 48 h of admission in the acute geriatric unit and 72 h post-surgery in hip fracture patients. This evaluation was performed in the patients´ room and in the presence of the family or caregiver. If the patient presented low level of consciousness, the

Health Status, Comorbidity and Geriatric Assessment, and In-Hospital Evolution
During hospital admission, a trained nurse collected demographic information, previous comorbidities and pharmacological treatment related to OD, geriatric assessment measures, the reason for hospitalization, presence of delirium during hospitalization, length of stay, and place of discharge, previously described in detail [2,19].
Regarding geriatric assessment, the nurse collected functional status with Barthel index of basic activities of daily living (ADL), Lawton index of instrumental ADL and Functional Ambulation Classification of gait; cognitive status using Global Deterioration Scale; and nutritional status through Mini Nutritional Assessment-Short Form and Body Mass Index, and calculated the body mass index.

OD Assessment
All patients were clinically assessed for OD by a trained nurse by a general exam and the V-VST, within 48 h of admission in the acute geriatric unit and 72 h post-surgery in hip fracture patients. This evaluation was performed in the patients room and in the presence of the family or caregiver. If the patient presented low level of consciousness, the assessment was delayed, and in case of no improvement of the level of consciousness, the patient was excluded, as is described in Figure 1. Previous to the study, the nurse received theoretical and practical training about clinical diagnosis and management of OD in older persons by a Speech-Language Pathologist.
The VVS-T uses different volumes and viscosities for evaluating safety signs of swallowing (cough, decreased oxygen saturation, and voice changes) and efficacy signs (poor labial seal, multiple swallows, and pharyngeal residue). This test is not only a clinical tool for screening, but also for diagnosis of OD with high sensitivity (93.2%) and a specificity (81.4%) for clinical diagnosis of OD, including severity, and it is simple and fast to perform [1,20]. This tool can be used to provide accurate indications on the optimal bolus Nutrients 2022, 14, 413 4 of 13 volume and viscosity for dysphagia patients [1], as it was used in the recent study of older patients with dementia and dysphagia [18].

Dysphagia Management, Educational Intervention and In-Hospital Adherence Assessment
When OD was detected by the V-VST, the nurse orally informed of the diagnosis and the protocol design. She also explained the treatment recommendations, as well as the reason for these measures, both to the patient and to the caregiver and gave written specific, individualized recommendations about food texture (regular diet, soft diet, pureed diet, or dysphagia diet), liquid texture (water, nectar or pudding), liquid volume (5, 10 and or 20 mL), posture (sitting, chin tuck, head rotation) and delivery method advised (spoon or dysphagia cup).
In-hospital adherence to these five recommendations was evaluated two days after diagnosis during mealtime by the same nurse s direct observation in a dichotomous way (yes or no). This evaluation was performed after explaining the study, being the patient and the caregiver aware of the observation, as is presented in Figure 1.

One-Month Self-Reported Adherence and OD-Related Complications Assessment
The nurse recorded the self-reported adherence to the measures recommended during hospital admission on a 3-point scale (always, sometimes, and never) by a telephone interview with patients or caregivers one month after hospital discharge, as summarized in Figure 1. She collected in the same call any reported complications, including respiratory infection, use of antibiotics, weight loss (≥3 kg), transfers to the emergency department, hospitalization, and mortality. Reported adherence and complications was also assessed with the main caregiver in those patients who had died one month after discharge

Data Analysis
Statistical analysis was performed using SPSS version 20. Data were described by relative and absolute frequencies or mean ± standard deviation as appropriate, and compared with the chi-square test, the Fisher exact test, the Mann-Whitney U test, and the Student's t-test.
In-hospital and one-month adherence rates were calculated for the five treatments: modified diet, texture and volume of liquid, feeding posture, and delivery method. As it was described previously, in-hospital adherence was assessed in a dichotomous was, while one-month adhere on a 3-point scale that was codified in two categories as it was performed by Low [21].
A patient was considered compliant when he/she was following the suggested recommendations, including those that should continue using a regular diet, water, or no delivery methods). However, compliance rates for individual treatments are reported only for patients with changes in these indications in order to avoid overestimating of these rates. We also used a univariate and multivariate regression model to analyze one-month complications risk factors. Statistical significance was set at p < 0.05.

Sample Characteristics
A total of 447 patients presented OD (97.8% with low efficacy signs and 64% with safety signs) and accepted participation in the study. Among included patients, 271 participants were admitted to the acute geriatric unit and 176 to the orthogeriatric unit, and 403 were followed after one month. The mean age was 92 years, 70.7% were women, and 25.5% lived in a nursing home before admission. Most had impaired functional and mental status, malnutrition, a high number of comorbidities, and polypharmacy. Of the patients, 60.6% had been admitted to the acute geriatric unit, mainly for a respiratory infection in 32.2%. Baseline information is reported in Table 1.

OD Management Recommendations and Post-Discharge Complications
The most frequent management recommendations were sitting when feeding and volume of liquid, which were recommended to all patients (n = 447), followed by the use of modified texture food (95.5% n = 427), self-feeding (56.8% n = 254), use of fluid thickeners (32.7% n = 146), and delivery method (12.5% n = 56), as is shown in Figure 2. During hospitalization, 60.2% of patients presented delirium, and the mean length of stay was eight days for the overall sample. At discharge, 19.9% received oral nutritional supplements (ONS), and 31.5% were transferred to a nursing home, as is reported in Table 1.
stay was eight days for the overall sample. At discharge, 19.9% received oral nutritio supplements (ONS), and 31.5% were transferred to a nursing home, as is reported in Ta 1.
Among patients assessed one month after discharge, 17.6% had been transferred the emergency department or readmitted to the hospital, 15.6% reported weight loss, 8 a chest infection, and 6% had died. One-month complications are described in Table 1 Figure 2. Management recommendations compliance rates during hospitalization and at one mo of discharge.
Non-compliant patients had a significantly higher prevalence of dementia (33.1% 17.5%, p < 0.001) and stroke (16.4% vs. 9.0%, p = 0.029). Admission in the acute geria unit vs. the orthogeriatric ward was also associated with non-adherence (68.3% vs. 47. p < 0.001). Respiratory infection, one of the main OD complications, as the main reason admission was related to reduced adherence (35.9% vs. 25.9%, p = 0.028). Non-complia was higher in patients with dysphagia safety signs (75.4% vs. 44.6%, <0.001), while th was a higher in-hospital adherence in patients advised to self-feed (80.7% vs. 42.7%, 0.001). However, there were no differences in adherence to OD measures according to presence of efficacy signs, delirium during hospitalization, or one-month complication Among patients assessed one month after discharge, 17.6% had been transferred to the emergency department or readmitted to the hospital, 15.6% reported weight loss, 8.9% a chest infection, and 6% had died. One-month complications are described in Table 1.
Non-compliant patients had a significantly higher prevalence of dementia (33.1% vs. 17.5%, p < 0.001) and stroke (16.4% vs. 9.0%, p = 0.029). Admission in the acute geriatric unit vs. the orthogeriatric ward was also associated with non-adherence (68.3% vs. 47.6%, p < 0.001). Respiratory infection, one of the main OD complications, as the main reason for admission was related to reduced adherence (35.9% vs. 25.9%, p = 0.028). Non-compliance was higher in patients with dysphagia safety signs (75.4% vs. 44.6%, <0.001), while there was a higher in-hospital adherence in patients advised to self-feed (80.7% vs. 42.7%, p < 0.001). However, there were no differences in adherence to OD measures according to the presence of efficacy signs, delirium during hospitalization, or one-month complications.

Discussion
This study found a low observed compliance with individualized recommendations to manage dysphagia in very old hospitalized patients with OD (37.1%) with improved self-reported compliance in the first month after hospital discharge (76.4%). We are not aware of other studies that report both compliance rates. We found a higher adherence than that described in the systematic review of Krekeler (21.9-51.9) [17]. However, studies with similar methods found similar adherence rates during hospitalization (35.6-56.5%) [22][23][24] and after-hospitalization (>50-79%) [21,25,26]. The difference between in-hospital and after discharge compliance might be explained by the assessment method used: direct observation [22][23][24] vs. self-reported [18,21] or not described [25,26]. The low compliance rates found in our patients is worrying, especially in older patients admitted for a respiratory infection, one of the main complications of OD in older patients. Recently, the Mataró research group developed a modified Mediterranean diet to increase patient compliance [27].
Compliance with recommendations varied with the content: they were higher for feeding posture and food texture, and low for liquid volume and use of thickeners. Rosenvigne also found higher compliance during hospital assessments for dietary modifications (82% and 78%) than for liquid volume (35.3% and 69%) or consistency of fluids (48% and 64%) [22]. Shim also found low thickener compliance (56.5%) [23], while Espinosa Val found lower compliance to fluid adaptation (50.4%) than to diet adaptation (88.3%) [18]. In comparison, Low registered high reported compliance rates to consistency of diet (79%) and liquid (84%) [21], similar to our one-month self-report compliance rates of modified diet and liquid (94.1% and 75.8%). The lower adherence rates to the adaptation of fluid (quantity and texture), described in our study may be caused by the previously reported dissatisfaction of thickeners [13]. Still, there is also a high difference in observed and self-reported adherence rates to liquid volume in our study (43.3% vs. 87.6%), without any apparent reason except for the assessment method.
We included older patients admitted to hospital with OD due to different causes like other studies [18,[21][22][23][24][25], but with essential differences in the methodology. Most of the previous studies were retrospective or case series [21][22][23][24] and included OD patients diagnosed by videofluoroscopy [21,23,24], while we performed a prospective study after diagnosis of OD with a validated and accessible clinical assessment, the V-VST [18,20].
We found lower adherence rates in patients with dementia (GDS ≥ 4), malnutrition (MNA-SF ≤ 7), and safety signs of OD. This association has not been reported previously. Patients with moderate to severe dementia depend on their caregivers and may not change their feeding habits due to their cognitive deficits [24]. This may be partially improved with professional caregivers. The association with malnutrition may be a consequence of undetected OD. The association between low adherence and safety signs is unexpected, as patients with safety concerns would be expected to have a drive to improve compliance. This might be explained due to the complexity of management recommendations for patients with impaired safety and more frequent recommendations in patients with OD safety signs of modified diets (97.6% vs. 91.9%, p = 0.006), fluid thickeners (49.7% vs. 2.5%, p < 0.001) and delivery method (15.7% vs. 6.8%, p = 0.006).
There was also an inverse association between compliance rates during and after hospitalization and living in a nursing home, which may be explained by the different assessment methods (observed vs. self-reported) and the presence of family caregivers of nursing home patients during hospital admission. We also found an association between in-hospital adherence with Barthel index, own feeding advice and age. Previously Low reported that adherence to OD management is associated with age and living at home [21], but Shim did not find any risk factor of measures compliance [23].
To our best knowledge, this is the first study that has shown a relationship between compliance one month after discharge, rated on a 3-point scale, and in-hospital adherence, rated dichotomously (43.2% vs. 28.4%, p = 0.010). Recently, Espinosa-Val observed that noncompliance patients during follow up reported non-compliance during admission, without adherence assessment during hospitalization [18]. This fact may allow focusing on training management during hospital admission in patients with an expected low compliance.
We found a lower one-month mortality in patients with good adherence to treatment recommendations in a multivariate analysis and higher risk among patients with low BMI or Barthel index. This is reassuring as compliance had not been included in previous mortality studies in patients with OD [5,6,25,26], except the study performed by Espinosa-Val, which did not observe an association between therapeutic compliance after hospitalization and mortality or other complications [18]. Moreover, 31% of our patients presented other complications in the month after hospital discharge. Multivariate analysis showed that risk was higher in patients with a low BMI and lower in women and in those with selfreported adherence to recommendations. Lower loss of weight in adherent patients to our recommendations could be justified by better oral intakes following hospital indications. Low previously found a lower risk or hospital admissions for chest infections in adherent patients [21], and recently Martín reported lower readmissions in patients with a minimalmassive intervention based on diet adaptation and oral hygiene [25].
Therefore, assessment and implementation of recommendations to treat OD may help in reducing complications after a hospital admission where OD is detected. However, a systematic review published in 2018 [13] proposed a weak recommendation against the use of texture-modified liquid according to the results of two trials [18,20]. The first study did not find differences in the incidence of pneumonia with the use of thickened liquid or chin down posture [26]. In contrast, the second study described a lower risk of aspiration with honey-thickened liquid [28]. While this review supported using texturemodified foods in patients with OD as a GCP [13], based on the IDSSI Framework that developed an international terminology for texture-modified foods and thickened fluids for OD patients [29].
Another Cochrane review [30] evaluated the same topic in patients with dysphagia and dementia, including the same trials [26,28], that were part of a large multicenter trial based on videofluoroscopy, with a high risk of bias, concluding the need for further high-quality trials. While the systematic review performed by Wu et al. in 2021, also remembered the low quality of evidence and the need for high-quality research [31] and recently, different authors started a debate about the use of modified diets highlighting the importance of shared decision-making [32] and patient compliance [33].
In our study, staff investment was low. A single nurse was trained to perform V-VST, explain management, and collect data. Previous studies usually involved speech and language therapists in the diagnosis of OD [21,22,24], but nurses have shown to effectively diagnose OD [2,5,6,19]. Also, a current systematic review described that 49% of the articles include nurses among the participants, mainly in screening and performing a clinical assessment of OD [34]. According to our study, the nurse can be a key member in managing OD, performing diagnosis, indicating individualized recommendations and training patients and caregivers.
Our study has some strengths: it is a prospective study, includes a high number of patients, and measures a range of demographic, clinical, and functional variables. We also used a validated assessment of dysphagia and two different methods to assess adherence (observation vs self-reported) that have to be considered in the compliance rate comparison. We recorded hospital adherence based on one-time direct observation and one-month adherence self-reported. The main limitation of the study is the subjectivity of this self-reported adherence to measures, especially in patients who died during follow-up. However, most studies described in the systematic review assessed compliance through journals, checklists or diaries [17]. We also performed a single telephone follow-up, fewer than other similar studies [5,6,25,26], limiting the multivariate analysis due to the low number of one-month complications, which precluded individual analysis of each complication. We also based OD diagnosis in V-VST, with a lack of instrumental assessment due to the complex access of this acute population [35] and based recommendations on clinical assessment that was performed by one trained nurse that also evaluated the adherence during and after hospitalization. Despite our efforts on the educational intervention to patients and caregivers about OD therapeutic adherence, assessing adherence to each recommendation during hospitalization and after one month, there is still room for improvement when adapting OD recommendations to each patient with good compliance. Moreover, more studies are needed to study the association between adherence to treatment and OD outcomes.

Conclusions
Compliance with treatment recommendations for OD in older hospitalized patients was low and increased after hospital discharge. One-month self-reported adherence to recommendations was associated with lower rates of mortality and other complications associated with OD. More research is needed to address how to improve compliance with recommendations and to explore if a higher adherence is associated with better outcomes.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Hospital Universitario Ramón y Cajal (protocol code 120/18).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy and permission restricted to the publication of this article only.