Method
Research Design
The key variables in the study are the method of serving nutritional supplement drinks, which is the independent variable, and its effect on compliance and consumption which are the primary dependent variables (Allen, Methven & Gosney, 2014). Nutritional status is a secondary dependent variable. The research problem was communicated well. The authors provided an introduction on the issue of the risk of or actual undernutrition among older adults and the health consequences. They detailed how oral supplementation is an evidence-based medical nutrition therapy to prevent or treat undernutrition but that poor compliance is a barrier to adequate consumption limiting the effectiveness of this therapy.
Subsequently, the authors discussed how nursing interventions can influence the physical and social environment during meal times. In conjunction with supportive care consisting of good communication, prompts or encouragement, and assistance, these interventions have been associated with higher food intake. However, preliminary research by the authors revealed that the conventional method of administering oral nutritional supplements (ONS) was by the use of a straw but that this practice seemed to be associated with lesser consumption than when a beaker or glass is used (Allen, Methven & Gosney, 2014). The study therefore aimed to establish through a more rigorous research method if serving method is a significant factor affecting nutritional intake. Presenting what is known about the topic, the authors established the gap in research that justified the study.
There was no theoretical framework included in the research report. It is unknown whether the researchers did not incorporate a framework or if space limitations in the publishing journal led to its omission. A theoretical framework pertains to the underlying theory or model that serves as rationale for the study as well as basis for the interpretation of results (Kaplan, 2012). Having a theoretical framework ensures that the study is related to nursing and will expand the nursing body of knowledge. However, a descriptive review of previous studies by Perry et al. (2012) showed that there was insufficient indication that nurses employed an integrated psychosocial model, concept, or theory pertaining to nutritional care. This may partly explain the lack of theoretical framework in the study.
The study employed a parallel group randomized controlled trial (RCT). Blinding was not feasible because it is impossible for both the nurses and patients not to know their serving method allocation (Allen, Methven & Gosney, 2014). An RCT is the most rigorous possible design because it allows for mechanisms to reduce bias lending greater confidence in the conclusions of the relationship between the independent and dependent variables (Grove, Burns & Gray, 2013). The purpose of the study is to establish if there is a relationship between serving method and selected outcomes, namely compliance and nutritional intake. The mechanisms to reduce bias include random allocation to groups and blinding.
Two serving methods are being compared and are represented by the two study groups – conventional serving using a straw or serving using the patient’s preferred method which is typically through a beaker or glass (Allen, Methven & Gosney, 2014). The comparison will generate evidence supporting the conventional or alternative method so that evidence-based recommendations can be made. In regards to ethics, the study sought approval from the Wales Research Ethics Committee and also complied with the Helsinki Declaration which states that the participants’ health and wellbeing must be a paramount consideration over and above the advancement of science (Williams, 2008). The researchers also underwent Good Clinical Practice training offered by the UK National Health Service (2014) that builds knowledge in the regulatory framework of research, individual and organizational roles, and eliciting informed consent as well as skills in reporting safety issues if these arise. Informed consent by proxy was obtained from the participants’ families.
Sampling and Setting
The study was conducted in multiple sites, majority of which were care homes but also included hospitals. However, no information was given on which wards the recruitment took place in. The 11 sites were located within four geographical regions. Researcher availability, geographical proximity, and the presence of organizational support were the criteria for choosing the sites (Allen, Methven & Gosney, 2014). The population was adequately identified and described using inclusion and exclusion criteria. Inclusion criteria included age above 65 years and having mild cognitive impairment and dementia as validated by a screening tool. The exclusion criteria included poorly-controlled diabetes, suspected diagnosis, swallowing difficulties precluding the administration of ONS in their original form, lactose intolerance, allergy to the products, celiac disease, delirium, and fluid restrictions to less than a liter per day (Allen, Methven & Gosney, 2014). Both criteria ensure that medical conditions which can potentially confound the findings, in favor of conventional feeding method or alternative feeding method, are controlled.
Potential participants were approached by care managers during the times that researchers were at the site. Patients who expressed their interest were requested to give proxy informed consent. This sampling method is consistent with convenience sampling because not all potential participants have an equal chance of being chosen but rather, recruitment depended on researcher availability at the site until the desired sample size was reached (Grove, Burns & Gray, 2013). A power analysis was done to determine, based on the literature and statistics, the number needed to establish the significance of findings with 90% power and a confidence level of 95% (Allen, Methven & Gosney, 2014). In addition, additional participants were recruited to achieve balance between both groups in terms of gender, age, number of medications, nutritional status, and mental status.
Data Collection and Measurement
The full and empty bottles were weighed to determine the standard weight of the glass which was used to determine the standard weight of the contents. After each feeding, the residual food was measured and subtracted from the standard weight of the content to ascertain the amount of nutritional supplement intake (Allen, Methven & Gosney, 2014). Nutritional information on the ONS was employed to measure the intake of specific nutrients. Compliance is measured by the participant’s willingness to consume the ONS rather than refuse it and was noted after each feeding. As such, the number of data collection points is appropriate. The measurement tools also gave an idea of how the different variables were operationalized.
In addition, secondary measurements of nutritional status using the Mini-Nutritional Assessment (MNA) that is based on six objective indicators and has high validity when used among geriatric patients were also measured (Allen, Methven & Gosney, 2014), but it was not specified when the tools was administered. If the typical rate of weight gain in adults over time is considered, then data collection points can be adapted to ensure weight gain is measured at a time when it usually occurs. Nevertheless, the choice of tools was appropriate as these provided objective data given that self-reports of intake cannot be given by older adults with dementia.
Validity was conferred by objective measures used such as anthropometric measurements and reliance on the weight of ONS given as compared with visual percentage estimations of intake or chart documentations are employed, the latter shown to have only 78% accuracy (Allen, Methven & Gosney, 2014). However, compliance was measured by observations and the nurses’ judgment of participant refusal or lack thereof. There was no discussion on how nurses objectively judged refusal of the drink or patient resistance to the nurse administering it given the relative difficulty of communicating with patients with dementia (Stokes, 2013).
Procedures
It was not clear how the dosing of the ONS given at specific times at midmorning, midafternoon, and evening was decided on. However, participants in both groups were given the ONS at these times. There was no data showing if there was deviation from the schedule or omission of ONS based on unavoidable circumstances. The energy content of the ONS brands used in the study varied by as much as 41.8 kilocalories, and the protein content varied by as much as 4.7 grams (Allen, Methven & Gosney, 2014). The use of different brands was intentional with the aim of improving the variety of flavors. The three flavors were chosen based on an unpublished study showing the preferences of the study population.
There was also no report on whether a patient assigned to the serving method by straw was switched to serving method by preference when he or she refused the ONS. Nurses may consider doing this in hopes of increasing supplement intake given that doing so is in the best interest of the patient and patient welfare comes first over research interests. In addition, there was no mention of a protocol on feeding by straw and feeding by glass/beaker. As such, intervention fidelity cannot be fully determined. Moreover, there was also no mention of training given to the nurses who participated in the study and if data collection forms were standardized to ensure completeness and accuracy of data.
Results
Data Analysis
The researchers employed the Mann-Whitney test to compare the two groups (Allen, Methven & Gosney, 2014). The test is appropriate because the assumption, based on the hypothesis, is that there is no difference in compliance and consumption despite differences in the serving method used. Another assumption is that the data is not normally distributed since the dependent variable is the percentage of ONS that should have been consumed versus what was actually consumed, and the independent variable which is serving method consists of two groups, i.e. conventional or based on preference. There was also variation in the brands and nutritional content of the ONS used. There was no discussion on the reduction of type I and type II errors.
Control/Validity
There are two techniques of research control used. One mechanism is randomization. Assigning a patient to each treatment arm by chance reduces participant bias or the tendency that patient characteristics influence the results (Groves, Burns & Gray, 2013). Another mechanism is blinding researchers and participants to the brand of ONS, although the process was not described, and blinding outcome evaluators of the patient’s treatment. The purpose is to reduce researcher bias or measurements influenced by the evaluator’s subjective expectations of a brand and serving method which reduce the accuracy of findings. Standardizing the intervention is still another way of reducing bias as it permits the clear attribution of outcomes solely to the independent variable.
Only statistical measurements were used as basis for establishing whether the study provided valid evidence. There was consideration of the fact that several ONS doses were omitted but were not considered as zero intakes, although there was no documentation as to the nurses’ reasons for omission (Allen, Methven & Gosney, 2014). Nevertheless, omissions were removed from the total number of doses that should have been consumed. Otherwise, the difference in consumption between the two groups would have been too small as to be insignificant.
Findings
The researchers reported a P value of 0.027 indicative of the existence of an effect (Sullivan & Feinn, 2012) of the independent variable on the two main outcomes measured and, hence, statistical significance. However, nutritional content as an outcome was slightly significant in regards to nutritional status despite the significantly higher ONS intake in the personal preference serving method. Because of the marginal significance in the outcomes, the authors assumed that other unknown factors, possibly related to metabolic conditions, influenced the relationship between nutritional intake and nutritional status.
Novice users of quantitative studies will have trouble understanding the implications of the findings because of insufficient information presented on the effect size and precision of estimates or confidence intervals. Effect size pertains to the magnitude of the difference noted between the two arms (Sullivan & Feinn, 2012). The study obtained a statistical significance of P = 0.027, but there was no measure of the magnitude which is in fact a small effect size because it is near Cohen’s d of 0.2. The confidence interval is set at 0.05.
Summary Assessment
The limitations include the lack of blinding of the participants and staff to the serving method used. This is fair because nurses who administer the ONS would know which serving method they are using. Another limitation is the variability in the brand, flavor, and nutritional content of the ONS. The different flavors are also necessary to enhance the palatability of the ONS and encourage consumption and compliance. An alternative would have been to use research-grade ONS to reduce variability as is done in studies of alternative medications such as cranberry juice for the prevention of urinary tract infections (Efros et al., 2010). However, this might not have been readily available in the U.K.
Despite its limitations, the study calls attention to the fact that undernutrition is a common issue among institutionalized older adults. Despite the employment of preferred serving methods, consumption is just 64.6% which means that more interventions must be developed, integrated, and tested to address the problem of nutritional intake. The findings support the need to redesign how nurses provide nutritional care in general so that the psychosocial and other aspects of eating are factored into the plan of care. The goal is to reduce the incidence of malnutrition in older adults in order for them to achieve a higher quality of life as they age.
References
Allen, V. J., Methven, L., & Gosney, M. (2014). Impact of serving method on the consumption of nutritional supplement drinks: Randomized trial in older adults with cognitive impairment. Journal of Advanced Nursing, 70(6), 1323-1333. doi:10.1111/jan.12293.
- Efros, M., Bromberg, W., Cossu, L., Nakeleski, E., & Katz, A.E. (2010). Novel concentrated cranberry liquid blend, UTI-STAT with Proantinox, might help prevent recurrent urinary tract infections in women. Urology, 76(4), 841-845. doi: 10.1016/j.urology.2010.01.068.
- Grove, S. K., Burns, N., & Gray, J. R. (2013). The practice of nursing research: Appraisal, synthesis, and generation of evidence (7th Ed.). Philadelphia, PA: Saunders. ISBN: 978-1- 4557-0736-2.
- Kaplan. L. (2012). Reading and critiquing a research article. Retrieved from http://www.americannursetoday.com/reading-and-critiquing-a-research-article/
- Perry, L., Hamilton, S., Williams, J., & Jones, S. (2012). Nursing interventions for improving nutritional status and outcomes of stroke patients: Descriptive reviews of processes and outcomes. Worldviews on Evidence-Based Nursing, 10(1), 17-40. doi: 10.1111/j.1741- 6787.2012.00255.x
- Stokes, G. (2013). Tackling communication challenges in dementia. Nursing Times, 109(8), 14- 15. Retrieved from http://www.nursingtimes.net/nursing-practice/specialisms/older- people/tackling-communication-challenges-in-dementia/5055312.article
- Sullivan, G.M., & Feinn, R. (2012). Using effect size – or why the P value is not enough. Journal of Graduate Medical Education, 4(3), 279-282. doi: 10.4300/JGME-D-12-00156.1
- United Kingdom National Health Service (2014). Introduction to Good Clinical Practice (GOP). Retrieved from https://www.wales.nhs.uk/sites3/page.cfm?orgid=580&pid=52229
Williams, J.R. (2008). The Declaration of Helsinki and public health. Bulletin of the World Health Organization, 86(8), 650-652. doi:10.2471/BLT.08.050955.