Discussion of the results
The aim of this review was to describe the impacts of different data structuring methods used in nursing records or care plans. The analytic framework  was used to generate a synthesis of the results found in previous studies. To strengthen the quality of this systematic review, the Preferred Reporting Items for Systematic reviews and Meta-analyses (PRISMA) guidelines have been used . Search strategy and databases were defined using PICO elements, which also helped to formulate the exact research question [43, 44]. The assessment of the studies included was partly based on predetermined exclusion and inclusion criteria.
The nursing process model recognised by the WHO has been widely used for documentation over the decades and still serves as a basic structure to record patient care in various settings. The model has been useful from the planning, delivering, monitoring and assessing perspectives in the paper-based and the later electronic formats. Over the years, the model has involved between four to six phases: assessment, diagnosis, goal setting, planning, intervention and outcome assessment [37, 110, 111]. In this review, the nursing process model was used in 39 out of 61 studies. The process comprised three to seven phases. There was also some concern that the process was not adequately used in documentation despite the many decades it has been available to implement [73, 74, 77-79, 81, 85, 86].
The development of nursing language to be used in documentation has evolved through research since the 1980s [3, 4]. This review also provides evidence that the analysed studies also focused on SNL developments. However, despite advances in terminology developments, the adoption of SNL still remains sporadic also on the international front [2, 39]. Nursing classifications have been developed to describe the nursing process, to document nursing care and to facilitate aggregation of data for comparisons at the local, regional, national and international levels [2, 4, 112]. This review revealed that the development of SNL is seldom local or even national; most often, SNL involved international aspects.
In many countries, cross-mapping has led to the building up of a reference terminology or SNL unification . In English-speaking countries, SNOMED CT  has been used for cross-mapping purposes. From nursing classifications, at least NANDA-I, CCC (formerly HHCC) and ICNP have been cross-mapped with SNOMED [114, 115]. ICNP has also been regarded as a reference terminology, and some nursing classifications have been cross-mapped with it, for example CCC, NANDA-I, Omaha System and NIC [116, 117]. Translation and cultural validations are required for the worldwide use of terminologies. It is an extra endeavour for nurses to be able to use SNL in non-English-speaking countries as most of these classifications originate from the USA [2, 95].
The results highlight that SNL supports the delivery of daily care in various ways. Nursing interventions are more accurately described and outcomes of care defined [71-73, 76, 79, 80, 84, 87, 89, 91, 93-95, 100-105, 107]. This study supports, albeit slightly, recently discussed technology aspects such as usability. The results indicate that nurses accept computerised tools and appreciate the availability of electronic data [72, 88, 90, 93]. However, there still exist negative attitudes towards electronic documentation and the need for support and education [73, 81, 84, 89]. These findings were mostly classified as unexpected impacts of nursing data structures, and they were discovered in connection with education. Nurses were also confused when they had to use parallel systems, paper and electronic records [98, 99]. As Kelley et al.  concluded, understanding the communication patterns on paper before converting to electronic documentation would be ideal in order to address potential obstacles for efficient information exchange following implementation of electronic nursing documentation.
Historically, there has been a long-standing discussion in nursing practice whether a standard format, for example use of a checklist, would be useful for documentation instead of free text notes. Obviously, a checklist would make the collection of information easier, but it does not promote a system that stimulates thought, creativity and response to individual patient and staff needs. In this review, nursing data structures had positive impacts on comprehensive nursing process documentation [75, 81-83, 85, 86, 99]. There have also been critical comments concerning the use of classifications, emphasising that strictly defined hierarchical classifications often serve organisational and administrative needs more than patients and their needs. Nevertheless, nursing classification systems are used directly by nurses during the course of care to record diagnoses, interventions and outcomes [4, 112, 118]. Beyond the many benefits, the use of a nursing classification provides for patient care, the most important today is data reuse [5, 119]. This was also the positive impact of some studies (e.g. [72, 75, 88]).
The use of resources, for example time in electronic documentation, has been of interest in previous studies (e.g. [120, 121]). In this review, handwritten care plans were not as comprehensive as computerised care plans, but they required less time to prepare . Also at times opposite opinions were presented [89, 97], or no evidence of time efficiency [90, 93] was shown. When nurses, after education, understood how rigorous the documentation system was, they started to value both their own and multiprofessional documentation . The use of SNL had positive impacts also on internurse communication , continuity of care [76, 92], legal demands  and increase in nurses' knowledge . These impacts have also been found in previous studies [11, 12]. Further, auditing the documentation model applied for practice had a positive impact on the use of SNL. An auditing tool has been developed, especially to assess the VIPS model in documentation [72, 80, 81].
The findings also revealed secondary impacts of the use of SNL based on the analytic framework. These impacts focused on research activities, supportive leadership and continuous education [75, 88]. Education was regarded as the key component in the successful use of SNL in various studies [73, 77, 78, 81, 84].
Limitations of the study
This review followed a protocol including 12 stages in order to strengthen the validity and reliability of the study. However, some critical decisions need to be discussed. Frequently, and also in this review, search terms pose a problem as the terms used in indexing literature vary between databases. Further, the search terms used for information retrieval in the databases were problematic because nursing documentation as the umbrella concept was difficult to operationalise using keywords. Thus, some bias in search methodology may exist, and some important and relevant articles may have been missed. The bias in the original search is proven by the relatively large number of new papers when screening the reference lists of review papers. To confirm the review process and the validity of the findings, the reviewers read the articles several times. Each study was read and assessed by two reviewers individually, and in case there was some disagreement, the team was consulted. The team also focused on describing the search process accurately so that this review can be updated.
Previous reviews have also criticised the quality of the studies (e.g. Urquhart et al. ). In this review, the study designs varied widely, and there were many descriptive studies with a single measurement. However, more rigorous methods such as randomised trials and pretest and post-test measures were also used.
The time frame of the papers analysed (1989–2010) also caused some confusion in the descriptions of the structures used in the studies. The review focused on studies where both paper-based documentation and electronic documentation were involved. There was also some uncertainty concerning the use of the nursing process model. Although it has been used for decades, it was not clearly stated how many phases the study comprises. Surprisingly, 23 studies had to be excluded as they did not assess nursing data structures at all.
The framework used for both data extraction and analysis proved to be flexible. However, it was not always clear whether the study focused on healthcare inputs, process or outcome factors. Thus, the analysis was very demanding, and while the reviewers worked independently, there was much discussion between the reviewer pairs. The team also discussed the findings thoroughly when analysing and summarising the results and categorising them as positive and unexpected impacts. The decision was made based on the studies assessed and how the original aims and objectives were described in those studies.
The settings and contexts where the studies were conducted varied widely, and based on the analysis, a comprehensive sample of healthcare settings was represented in the studies. The studies also described the status of SNL use in documentation internationally. In all, 16 countries were included in the analysis. However, some countries had only one study included. In most of the interventions, SNL was in use in nursing practice; however, a great number of studies reported about testing classifications in the clinical environment. Thus, what has happened after the piloting remains unclear.