Nutritional screening for improving professional practice for patient outcomes in hospital and primary care settings
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
Main objective:
To examine the effectiveness of nutritional screening in improving quality of care (professional practice) and patient outcomes compared with usual care.
Secondary questions:
The review may also contribute to answering some of the following questions:
-
Are the recipients of nutritional screening willing partners (i.e. is it an acceptable intervention)?
-
Are nutritional screening programmes likely to be cost‐effective?
-
Are there any alternative interventions for nutritional screening that are more (cost) effective or acceptable?
-
What are the characteristics of a (cost) effective nutritional screening programme?
Background
It has not been long since the publication of the Association of Community Health Councils report "hungry in hospital" (ACHCEW 1997). In their report, the authors argued that a lot of patients in hospitals are not receiving the nutritional support they need. Before that the King's Fund estimated that proper nutritional interventions could save a lot of money for the NHS by reducing length of hospital stay and improving recovery (Lennard‐Jones 1992). Similarly, concerns have been raised about the nutritional status of hospitalised patients in several countries (Roubenoff 1987; White 1991; Corish 2001; Beck 2002). Since then efforts have targeted the nutritional status of patients and the quality of nutritional support in hospitals. Several studies have identified that undernutrition, whether caused in hospital or not, is widespread among in‐patients (Stratton 2000; Elia 2003). Many screening tools are being developed in order to help identify undernourished patients (Ferguson 1999; Jones 2002). Routine assessment of weight and height in hospitals, as well as in high risk groups in the community, has been recommended (Lennard‐Jones 1992; Sizer 1996; Elia 2003). Despite these efforts and publicity, recent studies suggest that weight and height of patients are not systematically recorded in hospitals, making it difficult to estimate body mass index (BMI), change in weight and risk of undernutrition (Campbell 2002). It is also known that the development of many of the suggested nutritional screening tools did not follow defined methodological criteria (Arrowsmith 1999; Jones 2002). However, recently an easy to use undernutrition screening tool has been developed (Elia 2003). The use of this screening tool or equivalent tools have been recommended for improving quality of nutritional care in hospitals (NMPDU 2002; Elia 2003).
Given the prevalence of undernutrition and lack of proper nutritional management of patients in hospitals and community, nutritional screening may play a role in resolving the problem (Beck 2002). Nutritional screening could result in early identification of patients who may otherwise have been overlooked (Elia 2003). It may also help in establishing reliable pathways of care for patients with undernutrition. These could include provision of support, advice for junior clinicians, access to dietitians and provision of adequate follow‐up. On the other hand, screening programmes can invoke costs to health systems (personnel time and treatment costs) and patients (e.g. because of false negative and false positive results, and from side effects of potential treatments (Gigerenzer 2002). It is therefore important to assess the effectiveness of nutritional screening programmes. Similar studies have contributed to the establishment of screening programmes in other areas of care (Sackett 2000).
Objectives
Main objective:
To examine the effectiveness of nutritional screening in improving quality of care (professional practice) and patient outcomes compared with usual care.
Secondary questions:
The review may also contribute to answering some of the following questions:
-
Are the recipients of nutritional screening willing partners (i.e. is it an acceptable intervention)?
-
Are nutritional screening programmes likely to be cost‐effective?
-
Are there any alternative interventions for nutritional screening that are more (cost) effective or acceptable?
-
What are the characteristics of a (cost) effective nutritional screening programme?
Methods
Criteria for considering studies for this review
Types of studies
We will include Clustered Randomised Controlled Trials, Randomised Controlled Trials (RCTs), Block Randomised Designs, Controlled Clinical Trials (CCTs), (Randomised) Controlled Cross‐over designs, Controlled Before‐After studies and Interrupted Time Series studies. Clustered randomised studies and Block Randomised Designs, e.g. Latin Square designs (Pocock 1983) are considered to be of high value here. However, given the difficulties of conducting such studies quasi‐experimental designs with appropriate controls will also be included.
Blinding is not considered essential since it is virtually impossible to undertake in screening studies. Concurrent controls are preferred. Historical controls will be considered, but their limitations will be noted specifically as the intervention may be linked to other quality improvement initiatives and therefore historical controls tend to be unreliable. Concurrent control is defined as "a group of similar patients or participants, in similar setting, followed up for the period of study". Historical control is defined as "a group of similar patients in the same setting in the past, for whom reliable data is available".
Other study designs may be discussed but will not be 'included' in the systematic review. These studies may be valuable in answering the review's secondary questions.
Types of participants
We will include studies conducted in different settings including acute hospitals, long term care institutions, community and primary care. Participants will be adult patients aged 16 years or over, in hospital, out‐patient clinic, primary care or long term care and also individuals that are otherwise considered healthy but may be at risk of undernutrition. The definition of the latter group may differ but may include e.g. elderly groups etc.
There is no restriction in terms of gender or the underlying status of health of the participants (given this is the evaluation of the screening programmes). However, these variations will be taken into account in the interpretation of the findings.
Types of interventions
The intervention ('screening programme') is loosely defined as the application of a screening tool on a group of patients, or otherwise healthy people, for whom the level of undernutrition risk is unknown, to establish whether they are at undernutrition risk. Nutritional screening is different from nutritional assessment. The latter is usually more detailed and targets those patients who are already considered to be 'at risk' of undernutrition. Nutritional screening for undernutrition, in contrast, targets patients for whom the risk of undernutrition is unknown and usually involves application of simpler tools. In practice, however, the line between the two is blurred. Therefore, careful attention should be given to what is meant by screening and assessment in different studies.
The intervention could be offered as a stand alone intervention or as part of a wider strategy. For example it may be part of a wider strategy for quality improvement in which nutritional screening is one part. The intervention may also be offered as part of a more comprehensive screening package in which screening is being conducted for a few ailments. In the latter forms, the included studies should report outcomes that can be related to a nutritional screening programme. It is likely that nutritional screening programmes will have been implemented alongside other interventions to improve quality of care. Therefore, it may be difficult to disentangle the effects of screening per se. On the other hand, contamination of the 'control' groups with the intervention is a real possibility especially if the period to devise, implement and evaluate the intervention is prolonged, which is the case for many screening programmes.
Types of outcome measures
Outcomes include process outcomes and patient outcomes:
Primary outcomes
-
Patient outcomes: mortality, morbidity (e.g. wound infection), health related quality of life (measured by validated generic or condition specific instruments).
Secondary outcomes
-
Patient outcomes: e.g. change in BMI or weight.
-
Process outcomes: identification of patients requiring nutritional care, data (e.g. weight and BMI) recording, referral of patients to dietitians or similar, nutritional interventions, dietary advice to patients.
Adverse outcomes
-
Patient or process outcomes: false positive rates, adverse outcomes of nutritional interventions.
Search methods for identification of studies
Electronic searches
The search strategy will have two main components: "nutrition" component and "screening" component. No study design filter will be used, to increase the sensitivity of the search. It is also to acknowledge that apart from searches devised for systematic reviews and RCTs, there are not adequately validated search strategies for quasi‐experimental designs.
Electronic searches will include the following databases: MEDLINE, EMBASE, CINAHL, The Cochrane Library, Current Controlled Trials and National Research Register. We will also search NHS Economic Evaluation Database (NHS EED) and Health Economic Evaluations Database (HEED) for economic literature. We will conduct forward searching for key papers through the Web of Knowledge and Science Citation Index (SCI). The MEDLINE search strategy can be seen in Appendix 1 and will be adapted for all other databases searched.
Searching other resources
Key individuals will be contacted for ongoing and unpublished studies or any studies that may have been missed in our search. System for Information on Grey Literature in Europe (SIGLE) database will also be searched for grey literature. Reference lists of all included studies and relevant reviews and editorials will be searched for appropriate studies.
Data collection and analysis
Titles of all retrieved studies will be checked by one reviewer to exclude those titles that are obviously irrelevant to the review. The abstracts of all the remaining titles will be sought. The abstracts will be independently assessed by two reviewers. The full‐text of the papers will be ordered if both reviewers feel the studies are potentially relevant to the review according to the defined criteria. If only one reviewer considers the abstract as relevant, the view of a third reviewer will be sought. The full‐text of all obtained studies will be independently assessed by three reviewers against the inclusion criteria. Disagreements will be resolved by discussion.
Two reviewers will independently extract the data. Disagreements will be resolved by seeking the view of a third reviewer. Data about the intervention and control groups, setting, patient or participant characteristics, length of follow‐up, outcome measures, and the source of funding will be extracted. Any further data that may help the interpretation of the findings will also be extracted.
The methodological quality of included studies will be assessed by two reviewers (Jadad 1996). Disagreement will be resolved by seeking the view of a third reviewer. Quality assessment will follow approaches recommended by the Cochrane PaPaS and EPOC Review Groups. We will note the limitations of scoring quality of included studies (Juni 1999; Moher 1998). Any potential sources of bias in the included studies will be recorded and the implication of those biases for the reported outcomes will be elaborated upon.
Pre‐planned subgroup analyses are: within different settings (e.g. hospital versus primary care); according to the definitions used for the establishment of malnutrition in the population (e.g. different nutritional screening tools or cut‐off points); and the prior probability of risk in the screened population (e.g. if certain wards are screened versus screening the whole hospital).
The results of each study will be reported as point estimates with corresponding 95% confidence intervals (CIs). Potential sources of clinical heterogeneity will be elaborated on. Homogeneity will be statistically tested, if possible. If data are statistically heterogeneous, reasons for heterogeneity will be explored. Regardless of any evidence of statistical heterogeneity, the influence of specific differences between the studies will be investigated and reported. Quantitative synthesis will be implemented if the studies provide sufficient data and are considered homogeneous. Standardized or weighted mean differences will be calculated for continuous outcomes. Results will be expressed as relative risks (RRs) and risk differences (RDs) for dichotomous outcomes (Sutton 2000; Petitti 2000; Pogue 1998). 95% CIs will be reported. Presence of publication bias will be assessed using appropriate statistical tests (Sterne 2001). Meta‐regression will be attempted if technically possible (large enough number of studies with limited number of co‐variates).
Quantitative synthesis may not be possible, due to heterogeneity or because of scarcity of evidence for individual outcomes. In that case, the data will be synthesised using narrative approaches. Two reviewers will independently synthesise the findings. If there are any discrepancies, a third reviewer will be asked to synthesise the findings using narrative techniques. The results will be presented to the rest of the review group. The differences will be discussed and the views of other reviewers will be sought on the analysis and for the interpretation of the findings.
Sensitivity analysis will be used when appropriate to assess the extent to which the methodological quality of studies (using different cut‐off points), the characteristics of the participants or settings lead to variation in primary outcomes. These analyses will be considered only if meta‐analyses have sufficient numbers of studies.
