Mealtime management in Australian residential aged care: Comparison of documented, reported and observed care.

Abstract Purpose: Mealtime management in Residential aged care facilities (RACFs) should be holistic and comply with the principles of person-centred care (PCC) to ensure residents’ medical, nutritional and psychosocial mealtime needs are met. However, this is not always achieved and multiple issues with mealtime management in RACFs exist. The aim of the current study was to compare documented, reported and observed mealtime management to explore factors influencing optimal mealtime care. Method: Data were triangulated from: (a) review of 14 resident files; (b) observation of 41 mealtimes; (c) questionnaires with 14 residents; and (d) questionnaires with 29 staff. Result: Results revealed multiple discrepancies between data sources leading to the delivery of sub-optimal mealtime care. Poor documentation impacted staff knowledge of required mealtime practices resulting in occasions of inconsistent and inappropriate care. Observational and interview data highlighted discrepancies between residents’ mealtime preferences and actual practice. In many instances observed care was not holistic nor consistent with PCC. Conclusion: Given the significant medical, nutritional and psychosocial risks associated with poor mealtime management, systematic changes in policy, staff training and multidisciplinary care are needed.

Acknowledging the complexity of these challenges and to understand why inconsistencies and inadequacies in mealtime management persist, mealtime management must be examined in context using data from a variety of sources. To date only one study has attempted to use triangulation of data sources to explore issues associated with mealtime care in RACFs. With a specifi c focus on examining assistant in nursing clinical knowledge about dysphagia, Pelletier (2004) compared data from: staff analysis of a simulated mealtime scenario; individual semistructured interviews; and observation of mealtime care. Results revealed discrepancies between assistant in nursing knowledge and knowledge translation to daily care and a consistent pattern across data sources strengthening the fi ndings of the study and illustrating the benefi t of data triangulation. The aim, therefore, of the current study was to explore factors infl uencing mealtime management in RACFs by comparing information from documented, reported and observed care. The objective was to increase understanding of these factors to inform practice change and improve mealtime care.

Method
Permission for the study was granted by the Behavioural and Social Sciences Ethics Committee of The University of Queensland, Australia, and the participating RACF providers. This study was based on post-positivist, reality-oriented inquiry (Campbell & Ruso, 1999), using triangulation of data to describe as close to truth as possible, current mealtime management in participating RACFs. Triangulation was used to increase the accuracy and credibility of the fi ndings, exploring current mealtime management considerate of context, resident need and the interplay between documented, reported and observed care (Patton, 2002;Thurmond, 2001).
Two regional RACFs providing high care services were recruited for this study. One was a 56 bed forprofi t organization; the other a 61 bed not-for-profi t organization. The two facilities were chosen to represent typical aged care services within the local area in which the research was conducted. Each had bed capacity within the average range for Australian facilities (Australian Institute of Health and Welfare, 2012) and each facility was governed by long standing aged care organizations that managed multiple (27 and 35, respectively) RACFs statewide. Within each RACF, both resident and staff participant groups were recruited. Recruitment was a two-stage process with residents recruited fi rst and staff recruited during data collection.

Residents
Due to known challenges in recruiting representative resident samples in RACFs (Worrall & Hickson, 2003;Zermansky, Alldred, Petty, & Raynor, 2007), maximum variation sampling was used to select a diverse cohort (Patton, 2002) ensuring: (1) at least 50% of residents had moderate-to-high mealtime needs determined by a rating of " C " or " D " on the Nutrition sub-scale of the Aged Care Funding Instrument (ACFI) (Australian Government, Department of Health and Ageing, 2008); and (2) at least 50% of residents experienced moderate-tosevere communication diffi culties confi rmed by RACF staff and/or documented in the summary page of residents ' fi les. To receive a rating of " C " or " D " on the Nutrition sub-scale of the ACFI, residents must require either: (a) one-to-one physical assistance from another person to eat the majority of their meal (e.g. placing or guiding food to the resident ' s mouth); or (b) supervision to eat (e.g. standing by to provide physical or verbal assistance) and one-to-one physical assistance to prepare to eat (e.g. cutting up or vitamizing food). Potential residents were then further categorized based on classifi cation across fi ve ACFI sub-categories: (1) Mobility, (2) Verbal Behaviour, (3) Physical Behaviour, (4) Cognitive Skills and (5) Complex Healthcare. Each of these sub-categories is rated on a 4-point scale (A ϭ least need, D ϭ highest need). Final sampling included variability across these categories (Table I). A total of 14 residents were recruited from the 19 who provided consent for their fi le and ACFI data to be reviewed for potential inclusion. The fi nal sample included fi ve men and nine women, aged between 60 -99 years ( M ϭ 84 years; SD ϭ 10.9 years). Length of residency in the RACF ranged from 6 months to 27 years ( M ϭ 5.1 years; SD ϭ 8.3 years).
The purpose and procedure of the study was explained to interested residents verbally and in writing, using large font and picture symbols to aid comprehension. Written consent was obtained from all residents or their legal guardian prior to resident fi les being accessed. All residents were required to be English speaking; however, residents were not excluded based on their current communication or cognition abilities. Residents with an active palliative care plan were excluded.

Staff
Recruitment of staff was conducted as a two-step sequential process. Initially, all RACF staff with direct resident contact were provided with written information about the study and informed they may be observed during their daily interaction with residents. Staff who did not wish to be observed were asked to inform the Care Manager. No staff member declined observation. Following this, the staff member observed as the primary carer for the resident during each mealtime observation was approached to participate further. Inclusion criteria specifi ed that all staff had worked with the resident they were observed with for a minimum of 3 months and had suffi cient English skills to respond to a questionnaire. A total of 39 staff provided care to the residents during the observations; 29 met recruitment criteria and consented to participate further. Staff included 27 women and two men, aged 19 -62 years ( M ϭ 36.2 years; SD ϭ 12.6 years), from the following professional groups: assistant in nursing ( n ϭ 19), registered nurse ( n ϭ 7), recreation activity offi cer ( n ϭ 2) and enrolled-endorsed nurse ( n ϭ 1). These staff had worked with each matched resident participant for between 6 months and 14 years ( M ϭ 4 years; SD ϭ 4.4 years), for between 3 -5 days per week.

Procedure
Data collection began with a review of resident fi les, followed by mealtime observations, then the resident and staff questionnaires. Data collection for the chart review and the mealtime observational assessments was completed by a speech-language pathologist (SLP) with 9 years experience working in aged care. To minimize observer bias the researcher was independent to the setting and unfamiliar to both staff and the residents.

Resident fi le review
Once residents were selected, a comprehensive fi le review was conducted to identify (a) the nature of any information relevant to mealtime management and (b) the location of this information in each resident ' s fi le. Resident fi les included the resident ' s care plan, ACFI summary and progress notes (online and paper based).

Mealtime observations
Each resident was observed during three mealtime sessions including one breakfast, one lunch and one dinner and on different days of the week to allow for variation between meal types and staffi ng. A total of 41 observations were completed with each mealtime observed in it ' s entirely from meal set-up to completion. One observation session for one resident could not be completed due to resident illness. All observations were conducted by the primary researcher, who took a passive role in the observations, maintaining a seated position adjacent to the resident and at a distance of between 2 -4 metres from the resident. This distance ensured accurate observation of the mealtime, but minimized disruption to the resident and staff.
During the observations the researcher completed a purpose built form to record the mealtime data in detail (see the Supplementary Appendix to be found online at http://informahealthcare.com/doi/ abs/10.3109/17549507.2014.987816). Content of the form was developed from a broad-based literature review of factors infl uencing mealtime function (Amella, 2004;Evans, Crogan, & Schultz, 2003;Miller & Patterson, 2014;Steele et al., 1997). The form categorized the mealtime into eight main sections: mealtime environment; location and seating; Table I. Resident maximum variation sampling characteristics.

Sampling categories
Resident ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Nutrition * C D C C D C C C C C C D C D Communication# 1 4 1 1 4 4 4 4 1 2 3 3 4 4 Mobility * * A ϭ lowest care need, D ϭ highest care need. # 1 ϭ nil diffi culty, 2 ϭ mild diffi culty, 3 ϭ moderate diffi culty, 4 ϭ severe diffi culty. meal presentation; feeding assistance; resident -staff interaction; diet; use of specifi c mealtime management strategies; and researcher recommendations. The form was designed to be used by any health professional involved in mealtime care to provide a single rater perspective, identifying potential issues in mealtime management for further multidisciplinary evaluation.

Resident questionnaire
Following the observations, residents were asked four questions: (1) Where do you like to eat?
(2) How do you like your meals to be set-up?
(3) What foods and drinks do you like? (4) What foods and drinks don ' t you like?
The intent of these questions was to understand residents ' mealtime preferences, enabling comparison with staff knowledge of these preferences. Of the 14 residents, fi ve were unable to answer these questions due to severe developmental delay ( n ϭ 1), poor medical state with decreased level of alertness ( n ϭ 1) and late stage dementia ( n ϭ 3). The resident questionnaire was completed after the observations to maintain the researchers ' independence and avoid bias during the observations.

Staff questionnaire
Staff completed a questionnaire to explore their knowledge of the residents ' mealtime needs and mealtime preferences. Staff were initially asked fi ve questions: (1) Does the resident have any preferences about meal set-up or procedure? (2) Are there any foods or fl uids that the resident prefers or particularly likes? (3) Are there any foods or fl uids that the resident refuses or particularly dislikes? (4) Does the resident have a specifi c mealtime management plan or mealtime management recommendations? (5) Does the resident eat their meals in their bedroom?
Staff then were asked whether the resident required any of four specifi c types of mealtime assistance to complete their meal: (a) visual supervision, (b) verbal prompting, (c) physical assistance or (d) special positioning. Responses were recorded as " yes " or " no " . Staff responses of " sometimes " were recorded as " yes " .

Data analysis
The content of data obtained from the fi le reviews and residents ' and staff questionnaires were analysed qualitatively to identify key categories of information from each of these three data sources describing residents ' mealtime needs (Patton, 2002). Identifi ed categories from each of these three data sources were compared to identify similarities and differences between resident and staff responses and documented description of residents ' mealtime needs. Yes/no responses from the staff questionnaires, documentation of specifi c mealtime management recommendations and items from the mealtime observation form were further analysed using descriptive statistics calculating frequency counts and percentiles across the 14 fi les and 41 completed questionnaires and observations. Data obtained from the four data sources were cross-compared to complete the process of triangulation. Frequency counts of specifi c mealtime management recommendations in resident fi les were compared to staff yes/no responses and observational data to provide a direct comparison of documented, reported and observed use of visual supervision, verbal prompting, physical assistance and special positioning.

Results
Results obtained from each of the four data sources are presented below, followed by cross comparison to triangulate the analysis.

Resident fi le reviews
Information relating to mealtimes was found in several locations in residents ' fi les, including in their care plan, progress notes and ACFI summary. Content analysis revealed fi ve key categories of information, but no consistency in the depth of information provided across different residents. The fi ve categories of information were: formal assessment of mealtime diffi culties by SLPs and dieticians and resulting recommendations, including recommendations for meal texture and fl uid consistency (present in 5/14 fi les); resident likes/dislikes and preferences for meal set-up and procedure (in 9/14 fi les); recommended meal texture and fl uid consistency without reference to SLP or dietician assessment (in 9/14 fi les); strategies to facilitate the resident ' s mealtime experience (in 2/14 fi les); and informal report of mealtime diffi culties and strategies to assist the resident by facility staff, medical offi cers, family members and/or the resident (found in 9/14 fi les). Further analysis of these nine fi les identifi ed that fi ve of these nine residents required visual supervision, fi ve required verbal prompting, three required full physical assistance and four required special positioning, during mealtimes.

Mealtime observations
Both physical and verbal interaction between residents and staff during mealtimes was observed to be minimal. During a third (13/41) of the mealtime observations the resident ate alone. For most of these mealtimes the resident ate alone in their bedroom, with no staff interaction other than to deliver and collect the meal. Across all the mealtime observations there was minimal to no natural ongoing resident -staff communication, with 63.4% of resident -staff interaction coded as " no ongoing interaction " (26/41), 31.7% coded as " minimal appropriate interaction " (13/41) and 4.9% coded as " minimal inappropriate interaction " (2/41). All observations coded as either " minimal appropriate interaction " or " minimal inappropriate interaction " involved residents with moderate or severe communication diffi culties. Therefore, for residents with nil or mild communication diffi culties, no ongoing resident -staff interaction was noted during any observation.
Both participating facilities had the capacity to seat all residents in dedicated lounge/dining areas for meals. Observation, however, revealed 70% of residents ate in their bedroom during more than one observation (29/41), with most residents seated in recliner chairs for their meals. Almost all (40/41) meals were presented to residents with all courses on a single tray. Few environmental barriers to the mealtime due to odours, lighting, noise or physical obstructions were noted during the observations. Despite the majority of residents eating in their bedrooms, level of privacy was rated as minimal or fair during 88% (36/41) of the observations. These ratings were assigned as a result of: (1) the presence of other residents and their family members in shared bedrooms, (2) staff in the room carrying out care tasks unrelated to the meal and (3) observed lack of action to increase privacy for residents positioned in full view of the passing public, such as by drawing a curtain or closing the resident ' s door.
Each residents ' regular carer provided mealtime assistance during the majority of observations (39/41), with assistants in nursing delivering and collecting resident meals and providing mealtime assistance during 36/41 observations, family members during three observations and during the remaining observations a registered nurse and a kitchen staff member delivered one meal each.
All residents received thin fl uids during all observations. A normal diet was given during 34 observations, a soft diet during one observation and a puree diet during six observations. Modifi ed cutlery was provided during three observations and modifi ed crockery during 21 observations. Visual supervision was observed during 13 observations (31.8%) and recommended as needed by the researcher during 20 observations (48.8%). Verbal prompting was observed during six observations (14.6%) and recommended by the researcher during 10 observations (24.4%). Physical assistance was provided throughout the entire meal during 14 observations (34.2%) and recommended by the researcher during 15 observations (36.6%). Residents were specifi cally positioned for their meal during six observations (14.6%), with special positioning recommended by the researcher during 18 observations (43.4%). Four residents in this study were legally required to be fed by an RN based on their ACFI classifi cation. Mealtime assistance for these residents was observed to be provided by assistants in nursing and family members. During two observations no assistance to these residents other than meal-set up was observed. The assistance given to these residents was noted as insuffi cient in meeting the resident ' s mealtime needs during 7/11 observations, with the resident observed to struggle to complete his/her meal (63.6%).
Overall, mealtime management was observed to be consistent throughout the duration of the meal during 80% of observations and suffi cient to minimize the resident ' s risk of mealtime-related medical and nutritional complications during 61% of observations. Limited explicit management of residents ' psychosocial mealtime needs was observed during any observation.

Resident questionnaire
Content analysis of data from the resident questionnaire indicated most residents had a preference for both general food groups (e.g. sweet vs savoury) and specifi c food and fl uid items (e.g. apple juice vs orange juice). Five residents also expressed preferences regarding meal set-up and procedure, such as requesting meal items to be placed in a set position on the table and given in a set order. All nine residents expressed specifi c opinions about where they would like to eat and their preferred level of privacy during meals; fi ve of the nine residents preferred a high level of privacy during mealtimes.

Staff questionnaire
Over half (56%) of staff indicated that to their knowledge the resident participant did not have any mealtime likes/dislikes or preferences. Staff reported that 68% of residents preferred a high level of privacy during mealtimes, with 60% of residents eating in their bedroom at least " sometimes " . Just over half (63%) of staff were not aware that the resident had documented mealtime management strategies. Of the four residents who had previous SLP involvement 33% of matched staff were not aware of these recommendations. Staff reported that 57.5% of residents required visual supervision, 57.5% required verbal prompting, 50.0% required physical assistance and 52.5% required special positioning during their meal.

Triangulation of data across data sources
Similar to the residents, staff reported residents preferred a high level of privacy during mealtimes; residents classifi ed with the highest mealtime needs (rating of D) and the one other resident with documented SLP recommendations (i.e. residents 2, 5, 8, 12 and 14). Of these fi ve residents, four were also classifi ed as having severe communication diffi culty and one with moderate communication diffi culty. For those residents where greater consistency in mealtime management was noted there were no distinguishing demographics other than a pattern of higher cognitive skills.

Discussion
This study aimed to compare documented, reported and observed mealtime management in RACFs to explore ongoing issues and interactions in holistic mealtime management for residents with varying mealtime needs. Triangulation of data sources enabled a multi-faceted analysis of current mealtime management identifying inconsistency in mealtime management across staff and residents. Discrepancies were found between what was documented in residents ' fi les, what was observed and what staff reported residents needed. Past literature indicates mealtimes are highly valued by RACF staff (Crogan et al., 2001;Ullrich et al., 2014) and residents (Chan & Pang, 2007;Crogan, Evans, Severtsen, & Shultz, 2004;Palacios-Cena, Losa-Iglesias, Cachon-Perez, Gomez-Perez, Gomez-Calero, & Fernandez-de-las-Pena, 2013); however, data from this study revealed mealtime management was limited in meeting residents ' holistic mealtime needs.
Many care staff in this study were not aware that residents had documented mealtime management recommendations and most were unable to describe in any detail the mealtime preferences of residents in their care. These preferences were also poorly documented in residents ' fi les. Lack of detailed information in resident fi les in RACFs has been however, resident preference for privacy was not documented in any resident fi le. While all residents who completed the resident questionnaire expressed specifi c mealtime preferences, over half of the staff were not aware of these. Resident mealtime preferences were documented in only 9/14 resident fi les, with minimal information provided and limited to resident preference for meal location and one or two specifi c food likes/dislikes. During 51% of observations residents received a diet inconsistent with their documentation. In nearly all cases a regular diet was given instead of a soft or pureed diet, representing a discrepancy with the resident ' s documented diet. Modifi ed cutlery was supplied to the one resident documented as requiring it during only one of the three observation sessions with this resident. This resident was observed to experience diffi culty managing normal cutlery. Modifi ed crockery was supplied to residents as documented during most observations (83%), with the resident observed to use this crockery without diffi culty 76% of the time.
Comparison of recommended and observed use of: (1) visual supervision, (2) verbal prompting, (3) physical assistance and (4) special positioning revealed marked inconsistency across data sources. Staff reported need for the use of all four types of assistance was higher than that documented, researcher recommended or observed. Visual supervision, verbal prompting and special positioning were also observed less frequently than documented as needed or researcher recommended.
Staff were observed to follow the resident ' s documented mealtime recommendations during 51% of observations, 18% for residents with additional SLP recommendations. Overall consistency in mealtime management across data sources for individual residents is illustrated in Table II. Inconsistency in mealtime management can be seen for all residents; however, greatest inconsistency is seen for the four Table II. Triangulation of data sources (fi le review, mealtime observations, resident and staff questionnaires) for individual residents.

Management categories
Resident ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 x Total no. Categories Consistent ( n /11) 8 1 8 9 0 5 8 1 5 7 6 2 6 2 % Categories Consistent 73 9 73 82 0 45 73 9 45 64 55 18 55 18 x ϭ inconsistency between any two or more of the four data sources.  consistency across all four data sources. noted previously to be particularly problematic (Blackford, Strickland, & Morris, 2007;Pye, Worrall, & Hickson, 2000). Ensuring staff are aware of what is documented for residents is also a challenge, with investigators noting that recommendations are often documented without verbal handover to care staff (Pye et al., 2000). Overall, in the current study observed care did not refl ect either documented care or residents ' reported preferences, resulting in mealtime management strategies that were often poorly matched to residents ' needs. This study highlighted a need for change at both management and service levels to facilitate accurate documentation of individual resident ' s mealtime needs and better processes to ensure staff understand and address these needs.
Only four of the 14 (29%) residents in this study had previous SLP involvement and any formal diagnosis of mealtime diffi culties. Care provided to these residents was found to be less consistent than care provided to residents with lower level mealtime needs, raising concern for the wellbeing of residents at most risk of mealtime complications. However, it is also important to note that, from a SLPs perspective, mealtime management was limited in meeting the holistic mealtime needs for all residents. Hence, although some residents were identifi ed through documentation to have specifi c mealtime needs, many others may have also benefi tted from greater support. As very few residents had undergone any formal mealtime assessment the true extent of support needs for most residents in this study may be under-identifi ed. To confi rm this, multidisciplinary assessment of each resident ' s mealtime needs and observation of current mealtime care would be needed. Such assessment would facilitate evidencebased identifi cation of residents ' mealtime diffi culties and needs and inform the development of appropriate care plans to support individual residents during mealtimes.
However, more comprehensive resident assessment will only be benefi cial if the resulting recommendations are adequately communicated to and followed by all involved staff. Adherence to SLP recommendations in the current study was found to be poor, with staff observed to follow documented SLP recommendations during only two of 11 observations. This fi nding likely refl ects the identifi ed limitations in staff knowledge of residents ' mealtime needs and suggests the need for increased or more systematic communication between specialist health professionals and care staff. The need for greater staff awareness and understanding of mealtime issues has been highlighted numerous times in past literature, with a consistent call for increased training for RACF staff in mealtime care (Aselage & Amella, 2010;Crogan et al., 2001;Pelletier, 2004;Reimer & Keller, 2009).
By Australian law, residents of RACFs rated as D on the ACFI sub-scale of Nutrition are required to be fed their entire meal by a registered nurse (Australian Government, Department of Health and Ageing, 2008). However, this was not observed to occur for the four residents in this study with this rating. This fi nding most likely refl ects resource limitations, particularly limitations relating to time pressure and staffi ng (Bennett et al., 2014;Crogan et al., 2001;Kayser-Jones & Schell, 1997;Simmons & Schnelle, 2006). Of further concern was poor adherence to documented food texture. Texture modification is used as a therapeutic treatment for dysphagia, with the prescription of specifi c foods or fl uids dependent on the nature and severity of the diffi culty. Non-adherence can place individuals at increased aspiration risk (The Dieticians Association of Australia and The Speech Pathology Association of Australia, 2007). These fi ndings suggest residents ' with mealtime diffi culties in this study were at-risk for medical and nutritional complications and possible unplanned hospitalizations. Such consequences can affect facility funding and accreditation.
Most residents ate in their bedroom, limiting opportunity for social dining. This fi nding is consistent with resident preference for increased level of privacy during mealtimes, but inconsistent with the concepts of group and social dining (Amella, 2004;Barnes, Wasielewska, Raiswell, & Drummond, 2013). This fi nding suggests further research is needed to explore how residents defi ne the concept of social dining in RACFs. This suggestion is supported by the fi ndings of Adams, Anderson, Archuleta, and Kudin (2013) in comparing resident mealtime preferences in the family home prior to admission and the RACF post-admission. The authors found that, regardless of where residents ate at home, most residents preferred a quiet dining experience in the RACF, with the value of social interaction during mealtimes in the RACF limited compared to eating with one ' s family and loved ones at home. The authors recommended further exploration of resident mealtime and dining style preferences and further involvement of residents in mealtime preparation and planning. Philpin et al. (2014) again support this need, discussing the complexity of the construct of meatime experience in RACFs, including the interplay between physical and sociocultural elements of the mealtime and the importance of shared understanding about mealtimes between residents and staff.
Achieving shared understanding requires effective communication. In this study, minimal resident -staff interaction, either verbal or physical, was observed, with the least resident -staff communication noted with residents with minimal communication difficulty. While it is acknowledged that these residents often required less mealtime assistance and as such may not have been a priority for staff during mealtimes, they were the residents most capable of carrying an ongoing conversation. Hence, while limited resident -staff interaction during mealtimes may not have a signifi cant impact on meal intake for these residents, it may have a signifi cant impact on their psychosocial mealtime needs. Barnes et al. (2013) found similar results, noting that residents who were independent in meal intake were generally left to themselves. Pelletier (2004), however, found contrary results, with staff noted to initiate more ongoing and varied communication with residents with better communication skills. Neither of these studies examined the nature of resident -staff communication as a major factor in their research. Exploration of resident -staff communication in this study, although adding valuable data, was also limited. Investigation of resident experience of resident -staff interaction and the mealtime care they receive, as well as exploration of resident -resident interaction during mealtimes was not included. Classifi cation of resident mealtime diffi culties in this study was based on the ACFI sub-category of Nutrition. Although this measure is used in RACFs to determine residents ' mealtime needs and, therefore, directly mediates assistance given to the resident during mealtimes, the ACFI is not a diagnostic tool (Australian Government, Department of Health and Ageing, 2008). Assessment on the ACFI is limited in scope and cannot determine the nature or severity of the resident ' s specifi c mealtime diffi culties, nor does it provide recommendations for individualized mealtime management. Further analysis of mealtime management for residents with differing mealtime needs would require thorough multidisciplinary assessment. Classifi cation of resident communication diffi culty in this study was largely subjective and, therefore, also problematic. Again, comprehensive multidisciplinary assessment of residents communication needs would be ideal; however, the suitability of use of many commonly used communication assessment tools with the RACF population and in the RACF setting has long been questioned (Hopper, Cleary, Oddson, Donnelly, & Elgar, 2007;Pye et al., 2000;Worrall & Hickson, 2003). Although measures were taken to reduce observer bias, including independence of the researcher to the setting and participants and the use of a rigid procedure and data collection protocol, future research would be strengthened by the inclusion of multiple raters from different professions.

Conclusion
By comparing data across multiple sources this study revealed inconsistency in mealtime management in two RACFs, limitations in addressing residents ' holistic mealtime needs and lack of compliance with the principles of PCC. The study design and fi ndings reiterate the complexity of achieving optimal mealtime management for residents in RACFs. Barriers and challenges identifi ed in the two facilities in this study were not singular, simple problems, rather they were complex and arising from breakdowns and interactions at multiple levels from documentation to implementation. Facility management, specialists, researchers and policy-makers must acknowledge this complexity and work together to fi nd sustainable solutions to further support residents and staff during mealtimes. Priority must also be given to comprehensive investigation of residents ' psychosocial mealtime needs, incorporating the perspective of the resident and exploration of resident -resident mealtime interaction.

Declaration of interest:
The authors report no confl icts of interest. The authors alone are responsible for the content and writing of the paper.