Aberystwyth University, Information Management, Libraries and Archives, Llanbadarn Centre, AberystwythUK, SY23 3AL Show Find articles by Christine Urquhart Suffolk NHS Primary Care Trust, Public Health Directorate, Rushbrook HousePaper Mill Lane, Bramford, IpswichUK, IP8 4DE Find articles by Rosemary Currell University of Salford, School of Health & Society, SalfordUK, M6 6PU Find articles by Maria J Grant University of Huddersfield, School of Human & Health Sciences, R1/19 Ramsden BuildingQueensgate, HuddersfieldUK, HD1 3DH Find articles by Nicholas R Hardiker Author information Copyright and License information Disclaimer Aberystwyth University, Information Management, Libraries and Archives, Llanbadarn Centre, AberystwythUK, SY23 3AL Suffolk NHS Primary Care Trust, Public Health Directorate, Rushbrook HousePaper Mill Lane, Bramford, IpswichUK, IP8 4DE University of Salford, School of Health & Society, SalfordUK, M6 6PU University of Huddersfield, School of Human & Health Sciences, R1/19 Ramsden BuildingQueensgate, HuddersfieldUK, HD1 3DH Christine Urquhart, Email: ku.ca.reba@ujc. Corresponding author.Editorial Group: Cochrane Effective Practice and Organisation of Care Group. Copyright © 2018 The Cochrane Collaboration. Published by John Wiley & Sons, Ltd. A nursing record system is the record of care that was planned or given to individual patients and clients by qualified nurses or other caregivers under the direction of a qualified nurse. Nursing record systems may be an effective way of influencing nurse practice. To assess the effects of nursing record systems on nursing practice and patient outcomes. For the original version of this review in 2000, and updates in 2003 and 2008, we searched: the Cochrane Effective Practice and Organisation of Care (EPOC) Group Specialised Register; MEDLINE, EMBASE, CINAHL, BNI, ISI Web of Knowledge, and ASLIB Index of Theses. We also handsearched: Computers, Informatics, Nursing (Computers in Nursing); Information Technology in Nursing; and the Journal of Nursing Administration. For this update, searches can be considered complete until the end of 2007. We checked reference lists of retrieved articles and other related reviews. Randomised controlled trials (RCTs), controlled before and after studies, and interrupted time series comparing one kind of nursing record system with another in hospital, community or primary care settings. The participants were qualified nurses, students or healthcare assistants working under the direction of a qualified nurse, and patients receiving care recorded or planned using nursing record systems. Two review authors (in two pairs) independently assessed trial quality and extracted data. We included nine trials (eight RCTs, one controlled before and after study) involving 1846 people. The studies that evaluated nursing record systems focusing on relatively discrete and focused problems, for example effective pain management in children, empowering pregnant women and parents, reducing loss of notes, reducing time spent on data entry of test results, reducing transcription errors, and reducing the number of pieces of paper in a record, all demonstrated some degree of success in achieving the desired results. Studies of nursing care planning systems and total nurse records demonstrated uncertain or equivocal results. We found some limited evidence of effects on practice attributable to changes in record systems. It is clear from the literature that it is possible to set up the randomised trials or other quasi‐experimental designs needed to produce evidence for practice. Qualitative nursing research to explore the relationship between practice and information use could be used as a precursor to the design and testing of nursing information systems. Keywords: Humans, Nursing, Nursing Records, Outcome and Process Assessment (Health Care), Patient Care Planning, Professional Practice, Randomized Controlled Trials as Topic When patients are in hospital or sick at home and visited by a nurse, it is important that the care they receive is recorded properly. Nurses record a wide variety of information about a patient’s care and progress. For example, nurses would record a patient’s status while in ICU every hour, or when and how pain medication should be given and when it was given, or the progress of a pregnant woman visiting a clinic. These nursing records are a way for nurses to share care information with other nurses, other health care professionals and sometimes with patients. This is Information that can ensure patient care is consistent when staff changes shifts or information that can be used later as a history of previous care. But what is the best way to record and share this information? Is there a system or way of recording care information that is best? It has been suggested that there may be a difference in how nurses practice or how well a patient does with the use of one record system compared to another. A review of the effect of different nursing record systems was conducted. After searching for all relevant studies, 9 studies were found. These studies compared nursing records filled out on paper with nursing records done on computer; nursing records that were held by patients themselves to records kept at a hospital or clinic; and nursing records which used different types of forms. The evidence shows that nursing record systems which aim to fix a specific problem, such as reducing lost notes, decreasing the time required for data entry, or the amount of paper files, may be successful at fixing that problem. But it is uncertain whether changing an entire system of recording nursing care may improve how nurses practice or how well a patient does. What this review did show, is that there needs to be more work with the nursing professions to understand exactly what needs to be recorded and how it will be used, and that it is important to involve the nursing staff in the design and development of the nursing record systems. As health care grows more complex and the boundaries between the professions and different sectors become increasingly blurred, the ability to communicate effectively about patient care is more important than ever. The delivery of good nursing care has always been dependent upon the quality of the information available to the nurse, and nurses have long been recognised as key collectors, generators, and users of patient or client information. Moreover, the role of the nurse in providing 24‐hour care and in coordinating the care given by others means that the exchange and transfer of information is a significant nursing activity. A nursing record system is the record of care that is planned or given to individual patients and clients by qualified nurses or by other care givers (including nursing students) under the direction of a qualified nurse. The format of these records has varied over time, across different specialties, and in different countries. The development of nursing records has been concerned with their use as vehicles for the storage and exchange of information. They have also been used to support different philosophies of nursing practice. Over the last 30 years, nursing in the UK and North America has been developed by nursing theorists as a planned, problem‐solving activity. Although eminent nurse researchers (Henderson 1982) have suggested that there is more to nursing than problem solving, this has provided the focus for discussion about the provision of nursing care and the recording of information about patient care. The potential effects of recording nursing information in rigid frameworks based on this approach are considerable. The benefits to be expected from the consistent and directed recording of patient care information are clear, and include, for example, patient safety and continuity of care. The possible adverse effects on patient care, nursing practice, and the development of nursing knowledge, of recording patient information in structures which may be inadequate or inappropriate for the purpose, could be highly significant. All that can be readily retrieved and known about the nursing care of a patient or client after the event resides in these paper and electronic systems. The recognition that good information is necessary for effective nursing practice has led nursing theorists and practitioners, information scientists, and the commercial world to invest considerable resources in the development of computerised nursing information systems. The benefits to patients of this investment are by no means established. For example, an earlier review of nursing information systems (Audit Commission(UK) 1992) was critical of developments that had done no more than automate existing manual systems; and yet investment in these systems continues. A review of what is known about these systems and their effects, therefore, appeared to be necessary. Since the last update of this systematic review there have been several related reviews published. A review (Urquhart 2005) of qualitative research on nursing record systems addressed some questions raised in the conclusions to the previous update (Currell 2003). This review concluded that the formal record with legal status may not reflect nursing practice and care; the benefits of structured versus free text entry were doubtful and, therefore, time savings hard to realise. Poissant 2005 examined the impact of electronic health records on documentation time of physicians and nurses, and noted that a goal of decreased documentation time may be illusory. In a review of the research on electronic health record systems, Häyrinen 2008 noted that while many studies examined the completeness and accuracy of the documentation, fewer studies examined information use or the impact on communication between different health professionals. This review also noted the lack of evidence on nursing documentation or patient self‐documentation. A systematic review by Müller‐Staub 2006 examined effects on documentation of assessment quality; the frequency, accuracy, and completeness of nursing diagnoses; and on coherence between nursing diagnoses, interventions, and outcomes. The review concluded that standardised nursing diagnoses led to better documentation but that better documentation did not necessarily lead to better patient care outcomes. A narrative review of information systems in nursing (Oroviogoicoechea 2008) stressed the need for a more socio‐technical approach to evaluation; perhaps a realistic evaluation approach would be to consider coherency, context, mechanism, and outcomes with an appropriate mix of methods to answer the research questions. Such multi‐method evaluation is expensive and demands a range of research skills in the evaluation team. Not surprisingly, therefore, in a systematic review of the effects of health information technology on the quality, efficiency, and costs of medical care Chaudhry 2006 found that 25% of the studies came from four academic institutions that had implemented internally developed systems (full details in AHRQ 2006). Delpierre 2004 systematically reviewed the medical literature since 2000 for the impact of computer‐based record systems on the quality of medical practice, and user and patient satisfaction. The review (which identified three studies of nurse satisfaction with computer‐based record systems) concluded, in line with other reviews, that the structure of care and the organisational factors required more consideration in evaluations. A systematic review (York 2007) of the effectiveness of computerised decision support within nursing, combined with empirical research, found limited evidence of effectiveness and concluded that more information was required on the real requirements of nurses for information in decision making. The objective of the review was to establish the impact of nursing record systems on nursing practice and patient outcomes. The literature suggests that there may be a difference in nursing practice or patient outcomes with the use of one nursing record system compared with another, and that nursing record systems may be an effective medium by which to influence the way nurses practice. The comparisons made in the review were of the use of any one kind of nurse record system compared with another. The comparison groups included reflect the changes in nursing philosophy and nursing practice over time, and changes in the organisational features for the provision of health care. The developments in health information technology were also reflected in the different types of nursing record systems that the review covers. The questions for the review were whether there is a measurable difference in nursing practice or patients outcomes between the use of: 1. a structured nursing record system and an unstructured system; 2. a care plan and a nursing record system which does not include a formal care plan; 3. a manual nursing record system and a computerised nursing record system; 4. centrally‐held nursing records and patient‐held nursing records; 5. a nursing record and a multidisciplinary record; 6. a problem‐oriented nursing record and a non‐problem oriented nursing record; 7. daily progress notes and charting by exception (i.e. only recording abnormalities or deviations from the plan instead of recording at regular intervals, even if there is no change or abnormality to report). The review aimed to identify both beneficial and adverse effects of the use of different nursing record systems. The review also set out to establish gaps in knowledge and identify areas for further research, both in nursing and in informatics. Types of studiesThe types of study designs included were those described in the Cochrane Effective Practice and Organisation of Care Group (EPOC) Module (see 'Methods used in Reviews', in 'Group Details' for complete definitions of study designs). These were:
Types of participants1. Qualified nurses (how ever defined in the country of origin of the study) and nurse students or other healthcare practitioners working under the direction of a qualified nurse. The term 'nurse' is used to include all qualified nurses, midwives, and health visitors; and to include all those working in hospital, the community, or primary care settings. 2. Patients receiving care that was recorded or planned using different nursing record systems. Types of interventionsStudies that compared the use of one kind of nursing record system with another in hospital, community, or primary care settings, including: 1. multidisciplinary care records and patient‐held records, where they formed the only or principal record of nursing care for individual patients; 2. systems based on standard care plans, applied to the care of individual patients; 3. systems designed to record specific aspects of direct nursing care and which form an integral part of the nursing record, such as pain control or wound management; 4. records in paper and electronic formats. We excluded the following systems: 1. nurse management systems, such as those designed for rostering or workload measurement; 2. Systems designed for nurse education, unless they are applied to real patient care; 3. ephemeral or informal means of communicating nursing care, such as nurses' personal notebooks, ward diaries, or verbal communication; 4. systems such as risk assessment instruments which are not designed to be an integral part of the principal nursing record but are adjuncts only. Types of outcome measuresWe included studies if they gave objective measures of provider performance or patient outcomes. The original review considered only English‐language publications but the updates have included a search for publications in other languages. The EPOC Group Specialised Register, MEDLINE, CINAHL, the Royal College of Nursing (RCN) Database, and the King's Fund Database were the starting points for the search. The particular difficulties of searching the nursing literature are the lack of abstracts for much of the work, and the changes over time in the keywords used for indexing. The electronic search strategies were gradually refined for each database and the following strategies identified all the studies included in the review. They still identified large amounts of irrelevant material. For the original review, considerable handsearching of printed indexes included handsearches of the RCN Bibliography of Nursing Literature (1856 to 1983) (including the card index for 1983 which had not been included in either the printed bibliography or the database), the RCN Steinberg Collection, the Royal College of Midwives Midwifery Index (1976 to 1991), and the Cumulative Index to Nursing Literature (CINL) (1956 to 1982). These were carried out to complete the search back to 1948. The other major nursing index is the International Nursing Index, which was included in MEDLINE from 1963. A comparison was made of the International Nursing Index and the Cumulative Index to Nursing Literature for the year 1975. The journals not covered by CINL are the non‐English language journals and a number of the American State Nursing Association journals and newsletters. CINL does include more general health service journals not covered in the International Nursing Index, and also scans the major medical journals. For the purposes of this review, therefore, we thought it was unnecessary to search both the International Nursing Index and CINL. For the original review, the authors searched a large number of conference proceedings: Alliance for Engineering in Medicine and Biology; Annual Symposium on Computer Applications in Medical Care; British Computer Society Nursing Specialist Group Conferences; Global Conference on Patient Cards and Computerisation of Health Records; Health Care Computing Conference (UK); IEEE Engineering in Medicine and Biology Society; International Congress on the Use of Computers and Information Science in Nursing; International Society for Systems Science in Health Care; Medical Informatics Europe; The MEDINFO World Conference on Medical Informatics. For the 2003 update, the authors searched the MEDLINE, EMBASE, CINAHL, British Nursing Index, and Aslib Index to Theses databases (1998 to 2002). We handsearched The Journal of Nursing Administration, Computers in Nursing, and Information Technology in Nursing (up to 2002). We did not update the searches of the other databases and grey literature included in the original review (except for Health Care Computing Conference and Med Info) as the original searches produced little relevant material. For the original review, the authors sent a standard letter outlining the project aims and inclusion criteria to 55 experts in the field of nursing informatics around the world, and sent a similar letter to suppliers of nursing computer systems. The authors also contacted relevant nursing internet groups. There was very little response from outside the UK. The authors sent letters to the authors or institutions involved in studies which were either reported as ongoing at the time of publication, or in which there was inadequate information in the published account. These studies were up to 20 years old, so the lack of response was not surprising. A search for the 2008 update failed to find any further publications on these projects and these studies have now been removed from the list of ongoing studies. For the 2008 update, we searched (2003 to 2007) MEDLINE (in April 2008), EMBASE (in July 2008), CINAHL (in April 2008), ISI Web of Knowledge, BNI (in August 2008); and ASLIB Index of Theses (not date restricted) (in February 2008). We handsearched Computers, Informatics, Nursing (Computers in Nursing); Information Technology in Nursing; and the Journal of Nursing Administration (2003 to 2007). We also undertook searches of the references lists of retrieved articles, as well as the Agency for Healthcare Research and Quality 2006 report on electronic patient record systems. We searched the conference programmes of recent and upcoming conferences via search engines, and the Commons Ground conference site, to identify ongoing studies. Search strategies appear in the appendix. Only one of the two new included studies was in MEDLINE: as in the original review, several databases had to be searched to find relevant studies. Two of the authors reviewed all relevant studiesusing the criteria for review set out in the EPOC data collection checklist. We assessed the quality of all eligible trials using the criteria described by the EPOC Group (see 'Editorial information' under 'Group details' for 'Methods used in reviews'). Two reviewing authors (in two pairs) independently assessed the quality of each study and extracted the data, resolving any differences by discussion among all reviewing authors. The EPOC Editor was contacted only for advice on technical points. We have summarised relevant data on the quality and results of studies in the 'Characteristics of included studies' and 'Results' tables. We excluded studies that were so compromised by flaws in their design or execution as to be unable to provide reliable data; the reasons for exclusion are provided in the 'Characteristics of excluded studies' table and in the 'Results' section of the review. We made considerable efforts to contact the authors of studies that came within the scope of the review and appeared to meet the methodological criteria but for which the published account lacked sufficient information or data to be certain. We also tried to contact the authors of studies that were reported in the literature as 'ongoing' but which could reasonably have been expected to have been completed by now. Some of the studies are more than 20 years old and not only was it impossible to trace most of the authors, some of the organisations had disappeared. Of the six authors who did respond, two had discarded the relevant material. We have removed studies that were listed as ongoing in previous versions of the review (prior to the 2008 update) following a final check to identify any further publications on these studies. We have summarised and presented the data in natural units. For dichotomous variables in the RCTs, we have reported the absolute percentage differences between the two groups and the percentage differences relative to the control group. For controlled before and after studies, we have reported both the absolute change between the experimental and control groups after the intervention and the percentage change relative to the control group, and also the absolute change from baseline to post‐intervention in both groups, together with the difference of the change between the two groups. We have followed the convention of reporting outcomes as unfavourable events as far as was possible and sensible. In some cases this would have been counter‐intuitive and, although we had aimed for consistency, this has not always been possible. We have, therefore, added tags to some outcomes in the results table to indicate whether the result favoured the experimental group or the control group. We have reported P values as described by the study authors. Because so few studies were identified as being suitable for inclusion in the review, and because of the heterogeneity between studies, it was not necessary to conduct any pooled statistical analyses. In the 2008 update, and following the assessment of abstracts, the team identified 193 documents of potential relevance to the review; some of these were duplicates as the database searching was done at two different sites. The comparable figures for the 2003 update (publication date limits 2000 to 2002) were MEDLINE (264 retrieved, 18 identified for further analysis); CINAHL (716 retrieved, 21 identified for further analysis); and BNI (15 retrieved, seven identified for further analysis).
*This search was done last and no new unique records were identified. ** No date restrictions were applied. Further analysis resulted in a total of 38 candidate studies. Many of these had to be excluded for reasons of research design. For 17 of the 29 excluded studies, there were insufficient data available about the methods or the results, particularly the nursing contribution to the record (see 'Characteristics of excluded studies' table). Nine studies met all the criteria and have been included in the review. The range of nursing record systems identified in the literature search was wide and included aspects of paper records, computerised records, problem‐oriented records, multidisciplinary records, care planning, patient‐held records, and the siting of records or computer terminals. However, the studies included in the review did not address all of the research questions posed in the review protocol, and we found no appropriate studies addressing questions 1, 2, 5, 6, and 7. There were four studies comparing a manual and a computerised system, which addressed question 3; three studies compared client‐held with centrally‐held records, which addressed question 4; and two studies compared differently structured paper records but did not compare structured with unstructured records, and so did not directly address question 1. Of the nine studies included in the review, eight were RCTs: two of client‐held maternity records, one of a parent‐held child health record, one of a paediatric pain management sheet, one of a medical record with medical and nursing care planning forms, and three of computerised nursing record systems. One study was a controlled before and after study of a computerised nursing care planning system. These records were all designed for different healthcare and nursing purposes, and it was clearly not appropriate to combine the findings of these studies. There are two new studies (Bosman 2003; From 2003) in the 2008 update. One study (Scharf 1997) that was included in previous versions has now been excluded, to comply with revised EPOC quality criteria for controlled before and after trials. As indicated above, the previous studies listed as ongoing have been removed. There is one possible ongoing study but only limited information is available. Studies comparing client‐held records with centrally‐held recordsIn one of the client‐held maternity record studies (Elbourne 1987) it was hypothesised that the women holding their own maternity records would: be more satisfied with their maternity care, feel more confident and in control, find it easier to communicate with staff, suffer less depression, have better outcomes for labour and delivery, smoke less, be more likely to make use of maternity services, be more likely to breastfeed. It was also hypothesised that the availability of notes would be increased and clerical time saved. It was a RCT in which the necessary overall sample size was calculated to be 280 for the patient satisfaction factors in the study. Data were obtained from four self‐administered patient questionnaires and case notes. There were 317 women recruited to the study and, at the time of the final questionnaire, the response rate was 85% of the patients originally recruited. Lovell 1987 also hypothesised that giving mothers their own maternity notes would have a positive effect on: their satisfaction with the care they received, their sense of control and self‐confidence, communications with staff, involvement of the babies' fathers, smoking and alcohol behaviour, breastfeeding, and attendance at clinics. They also aimed to assess any practical problems or effects on clinical outcomes. This was a RCT carried out in one hospital antenatal clinic in a socially disadvantaged area of South London. Data were obtained from self‐administered questionnaires at booking, 32 to 34 weeks' gestation, and postnatally (mostly in hospital); clinical and demographic information from the case notes; observations in the antenatal clinic; and 20 semi‐structured interviews with staff. There were 246 mothers considered eligible, before attending the clinic, of whom nine were subsequently found to be ineligible (for example having miscarried or moved away). It was difficult to work out the exact response rates for the questionnaires as the authors reported that the numbers of women responding to different questions varied, and in some tables they only presented the findings as percentages. Lakhani 1984 aimed to evaluate the efficiency and effectiveness of a parent‐held child health record, as a means of improving communications between healthcare agencies and improving parents' uptake of health services. A RCT was carried out comparing groups of parents given or not given a child health record booklet. The study involved two postnatal wards in one teaching hospital over a period of four months, in 1980, and used a balanced random plan so that each ward was a control ward or a study ward for the same number of weeks during the study. The calculated necessary sample size was 225 in each group assuming an immunisation uptake of 70%, rising by 12% over the study period. No details were provided on adjustments to sample size or analysis with the type of clustering randomisation used. The final sample was 157 babies in the experimental group and 142 in the control group. The methods of evaluation were: interviews with the mothers; questionnaires for the health visitors, clinical medical officers, and general practitioners; and review of the entries in the record booklets. Studies comparing paper records with differing structuresIn the paediatric pain management study (Stevens 1990), the aim was to improve the pain management of children following surgery through the development and testing of a paediatric pain assessment and management flow sheet. This flow sheet was not just a tool for pain assessment but was also used for the whole process of evaluating, treating, and assessing the effectiveness of pain management. The author described the pre‐testing and implementation of the pain sheet that preceded the RCT, which compared the use of the pain sheet with the usual recording method. In the trial, 43 eligible patients were randomly assigned to one of two paediatric surgical units, and 24 nurses were randomly assigned to one of two nursing teams. Data were collected for 48 hours postoperatively (or until the child went home, if sooner) from the nurses' charts and medication records, and from a nurse satisfaction survey. The planning form study report (From 2003) contained two related studies. The first was a before and after study (with no control) and data from that were not included in the review. The second study was a RCT of the effect on patient care of including a multidisciplinary care‐planning form in the medical record. The care planning was a two‐step process. In step one, the receiving nurse identified problems and drew up plans of action. These were recorded in the nursing records and on a green planning form included in the medical record. In step two, the medical specialist identified problems, plans of action, and a length of stay schedule, and this information was also recorded on the planning form. Patients admitted to the internal medical clinic of a university clinic over a period of six weeks were randomly allocated to continue with the planning form in the medical record (n = 154) or to have the planning form removed (n = 150) (control). The clinic had four specialised sub‐wards. The outcome measures were length of stay (LOS), length of stay comparisons (actual versus planned), risk of readmission, and accomplishment of plan of action. The mean LOS for the four different wards was compared (analysis by ward of discharge). Studies comparing manual nursing care planning with computerised nursing care planningOne computerised nursing care planning study (Spranzo 1993) used a randomised block pre‐test, post‐test design. Four nursing wards in a large Veteran Affairs medical center were matched on two characteristics (nursing care needs and head nurse motivation) and then randomly assigned to an experimental or control group. The intervention was the introduction, on the experimental wards, of a computerised nursing care planning system; which was compared with the existing paper nursing care plans. The system was for care planning only and did not include assessment or progress notes, so was only one part of the total flow of nursing information. Data were collected contemporaneously by questionnaires to nurses six weeks before the implementation of the computerised system, and three months post‐implementation; by telephone interviews with patients three days after discharge, two months pre‐implementation, and three months post implementation; and by review of patients' clinical records. The number of patients involved in the study was 41 in the experimental group and 32 in the control group at the pre‐test stage, and 40 in both the experimental group and control group at the post‐test stage. There were 19 nurses in the experimental group and 21 in the control group in the pre‐ and post‐test stages of the project. Data were also collected from the hospital computerised administration systems for nursing hours, overtime, and sickness; and also for patients' length of hospital stay. The unit of allocation was nursing wards, although the units of analysis were nursing wards, nurses, and patients. Multiple regression analyses were carried out to control for selected covariates. In the second study (Daly 2002), in one long‐term care facility, the intervention was the introduction of a computerised care‐planning system incorporating standardised nursing nomenclature, which was compared with an existing paper‐based system. In this study, 30 patients were randomly allocated to either the computerised or written care‐planning system on admission. The four registered nurses in the facility who were computer literate were assigned to the computerised system, and the four nurses who were not computer literate were assigned to the paper‐based system. It appeared that both systems were used for recording care as well as for care planning. Patient demographic data were collected on admission, together with all other study variables. Every month, for six months thereafter, all data were collected for each patient; except the Mini Mental State Examination (MMSE), care plan data, and time for care plan development, which were collected by the investigator every three months. The study lasted for 30 months in total. The third study (Ammenwerth 2001) compared the conventional paper‐based documentation system in a psychiatric department of a university medical centre with a newly introduced computer‐based nursing documentation system. This was a RCT in which 60 patients were allocated to either the paper‐based or the computerised system on admission to the ward. Ten nurses worked continuously on the ward throughout the study and used both systems. The time period covered by the study was 13 weeks. A mixture of qualitative and quantitative data collection methods were used. All the nurses were interviewed at the beginning and after the end of the data collection, and were asked about their acceptance of computers. The computerised system was installed seven weeks in advance of the study. Measurements were started three weeks after the first patients were admitted and ended three weeks after the last patients were admitted. The average length of patient stay was 20 days. Sixty patients in total were recruited to the study, but full results were reported for only 40 patients, as the first 20 were discounted because of possible learning effects immediately after the training on the use of the system. Later publications by this team on their longitudinal evaluation (listed under Ammenwerth 2001) noted that acceptance of a computerised system may depend on attitudes towards documentation, and the ward culture (Ammenwerth et al 2003); a FITT framework (fit between individuals, task, and technology) was proposed to explain the different patterns of adoption found (Ammenwerth et al 2006). An interrupted time series design was used by the same team to examine the quality and quantity of nursing documentation (Mahler 2007) on four wards, but with only three time points. We could not include this in the present review. The fourth study (Bosman 2003) compared the manual registration system used on an intensive care unit (ICU) with the Intensive Care Information System that was fully implemented for routine use seven months prior to the trial. This was described as a crossover randomised trial with patients randomly allocated to one or the other arm; the nurses were assigned (in a blinded manner) to the patient, so that the nurses were alternately exposed to both methods. The trial lasted for six weeks. The authors stated that all nurses were experienced and familiar with both methods of charting. Only patients with uncomplicated cardiothoracic surgery were included in the trial (145 out of 174 admitted were eligible). The observers were two experienced ICU nurses; activities observed included a set of admission tasks, and time‐motion analysis was used to observe the two nurses involved in admissions. For the registration tasks, work sampling was applied to observations made every five minutes; the activity was assigned to one of 148 items grouped into four categories (patient care, documentation, unit‐related activities, personal activities). Observations were made up to 11 hours after admission (mean 182 ± 122 min after admission, inter‐quartile 181 min, no reported differences between groups). The work‐sampling methodology was adapted from an existing published scheme. Only validated and manually entered values from the computerised system were collected for comparison, for the first 24 hours of data after ICU admission. Changes over time in the proportion of time spent on documentation and patient care activities were also recorded. One pilot prototyping study (Choi 2004a) on a hand‐held information system for point of nursing care was identified as a possible ongoing study, but there were no details about the type of evaluation planned. AllocationEight of the nine included studies were RCTs (Ammenwerth 2001; Bosman 2003; Daly 2002; Elbourne 1987; From 2003; Lakhani 1984; Lovell 1987; Stevens 1990). A power calculation appeared to have been done in three RCT studies (Elbourne 1987; Lakhani 1984; Lovell 1987). Lakhani 1984 used a 'balanced random plan' to allocate wards but there were few other details of the clustering technique used; and outcomes were measured by individual patient (one baby per set of twins included). The client‐held record studies (Elbourne 1987; Lakhani 1984; Lovell 1987) described how well the control and experimental groups were matched after the randomisation process, with tables comparing baseline characteristics in two (Elbourne 1987; Lovell 1987). Of the nursing care planning, nursing record, and pain management RCTs, three (Bosman 2003; From 2003; Stevens 1990) tabulated the baseline characteristics of the experimental and control groups. Minimal details were provided in two (Ammenwerth 2001; Daly 2002). In the pain management study (Stevens 1990), both the nurses and patients were randomly allocated. Bosman 2003 was a patient randomised RCT but only nursing care process outcomes were compared, and nurses were not randomised. No details were provided on the patient randomisation process but the authors stated that the nurses were assigned to the patients by a head nurse who was unaware of the patient randomisation results. The patients in the experimental group had higher APACHE II and APACHE III scores than did the controls but were otherwise similar. From 2003 completed care planning prior to randomisation of patients; the groups were broadly similar in demographic characteristics and the patient outcomes measured. Only one RCT (Elbourne 1987) described adequate allocation concealment. The ninth included study (a CBA design) was a study of computerised nursing care planning (Spranzo 1993) and was described as a randomised block pre‐test, post‐test design. Four nursing units were randomly allocated to control or experimental groups but the units of analysis for almost all the outcomes were nurses or patients. There were baseline differences between the wards in that the experimental wards had a significantly higher level of patient intensity and a higher nursing workload than the control wards. In the implementation phase there was also a significant difference between the two experimental wards, and the author suggested that this might have added to the stress of implementing the computerised system. BlindingBlinded assessment of the outcomes was not reported in any of the studies, and might not have been possible for any of them because of the nature of the interventions. The records that were the source of the data for clinical outcomes were the intervention in each case, and even independent observers of nursing activity with a record system (as described in Bosman 2003) could not be ignorant of the allocation. The nurses in Ammenwerth 2001 were themselves responsible for recording time spent on care documentation, whereas in Daly 2002 and Bosman 2003 the timings were done by the investigator. From 2003 noted that one of the two investigators was not independent, practising on one of the wards part of the time. Incomplete outcome dataIn one computerised record study (Daly 2002), 30 patients were randomly allocated but four patients died and six were discharged in less than seven months and were not included in the analysis. Bosman 2003 noted that 31 of the 145 eligible admissions were not observed but data were collected on the registration activities for all 145 patients. From 2003 noted that records were reviewed daily but, as not all categories of action could be recorded, the emphasis was on actions demanding collaboration with other departments. Patients were followed up for 120 days from the date of the second medical examination: to collect data on readmissions, date of discharge, and deaths after discharge. Ammenwerth 2001 excluded the first 20 of the 60 patients allocated from the data analysis as the first three weeks counted as the training period for the nurses. The three client‐held record studies (Elbourne 1987; Lakhani 1984; Lovell 1987) reported the exclusion criteria used after allocation and the reasons for drop‐outs later in the studies. Withdrawals were mainly for medical or social reasons. Elbourne 1987 sent the final questionnaire to 93% of the original randomly allocated participants, and Lovell 1987 noted that most of the withdrawals for medical reasons were picked up in the clinical outcomes. Lakhani 1984 noted that 24% of the 479 mothers who started in the study had left the district after a year; of those interviewed (322 of 343 remaining in the study), data could only be obtained from 299 (62% of the original 479). Selective reportingThe truncation in the presentation of data and results for publication of all these studies (for example reporting results as percentages only when the total numbers were not specified) has caused some difficulty in assessment for this review. In Bosman 2003, some of the activities compared were only possible with one or the other of the systems. Time comparisons were reported for the top five most frequently observed documentation activities in both systems, but comparisons of two activities between the systems were not possible as the activities did not fall in the top five of both. The computerised system automatically calculated fluid balance and dealt with laboratory results and both these activities were in the top five of the paper documentation group. With Ammenwerth 2001, the care plan quality was difficult to assess reliably given the lack of care planning with the paper system. Other potential sources of biasThere were several sources of contamination possible. Elbourne and Lakhani (Elbourne 1987; Lakhani 1984) both reported episodes in which clients from the control group were given the experimental record, and vice versa. They reported the steps taken to deal with this in the analysis. In the computerised nursing care planning study, Spranzo 1993 described steps taken to prevent contamination between the control and experimental wards; but it was not clear in the pain management study (Stevens 1990) whether there might have been contamination between the nurses working on the experimental and the control wards. The effect of explicitly taking the parents' views of the children's pain into account in the experimental group, but not in the control group, was also not clear. The author also noted the variety of ages and diagnoses in the two patient groups, the small sample size, and the lack of independent pain ratings or the consideration of any pain‐relieving interventions other than drugs. In two computerised nursing record system studies (Ammenwerth 2001; Daly 2002), the paper and computerised systems were being used in the same departments; and in the intensive care information system study (Bosman 2003) nurses were 'alternately exposed' to both paper and computerised charting. In Ammenwerth 2001 the nurses were using both systems and knew which patients had which kind of record; and in Daly 2002, although the nurses were assigned to use one system or the other, it must have been possible for them to take part in the care of patients from the other group. In From 2003 it was unclear how nurses may have used the planning information separately copied into their own records but there seemed to be potential for contamination between the groups. Several of the studies had unit of analysis errors. In the child health record study (Lakhani 1984), postnatal wards were the unit of allocation but babies were the unit of analysis. In Bosman 2003 the patients were the unit of allocation but nurses (individuals in the registration phase or pairs during admission) were the unit of analysis. In Spranzo 1993 the unit of analysis error caused difficulties in the analysis and significance of the results. The author also noted the small numbers in the study and that three months from the introduction of the intervention to measurement of outcomes might have been too short a time period. The author noted the possibility of systematic error in data collection and a threat to internal validity through compensatory rivalry between the groups of nurses. However, the author described the selection of potential covariates and the multiple regression analyses undertaken to account for them. This appeared to have been done both logically and thoroughly. Repeated measure ANOVA were computed for each of the study variables, with type of care plan as the independent variable, in one of the other computerised record system studies (Daly 2002). This appeared to have been done systematically and the authors reported on between participants, within participants, and interaction effects. The effects of timing of intervention, training periods, measurement periods, and the attitudes of nurses were hard to disentangle. In two of the computerised record studies (Ammenwerth 2001; Spranzo 1993), the attitudes of the nurses to computer use were assessed; but there was no mention of this in Daly 2002, only that half the nurses were 'computer literate' and half were not. In Daly 2002 the nurses who were assessed as computer literate were assigned to the computerised system, the non‐computer literate nurses were assigned to the paper system, which made valid comparisons more difficult. The study by Bosman 2003 was conducted some months after implementation of the computerised system, and the authors stated that nurses were familiar with both types of system (paper and computerised). Studies comparing client‐held records with centrally‐held recordsThe results of the client‐held record studies (Elbourne 1987; Lakhani 1984; Lovell 1987) showed no consistent positive or negative effect for clients holding their own records. The maternity record studies (Elbourne 1987; Lovell 1987) showed no statistically significant differences in clinical outcomes between the women holding their own maternity notes and those with conventional cooperation cards, except that Lovell 1987 reported significantly more assisted deliveries in the study group. They suggested that this was related to the higher number of women in this group having epidural anaesthesia. Elbourne 1987 reported that women holding their own notes were more likely to say they felt more in control of their pregnancies (author reported rate ratio 1.45; 95% confidence interval (CI) 1.08 to 1.95), and to say they found it easier to talk to the doctors and midwives antenatally (rate ratio 1.73; 95% CI 1.16 to 2.59); although they also reported that women tended to find the notes more difficult to read and understand. Lovell 1987 reported statistically significant differences between the study and control groups for the women who thought they had not been at all well informed during pregnancy (author reported P < 0.01), and for those who had been able to have their chosen companion with them in labour (author reported P < 0.05). Asked about which kind of notes they would prefer in future, Elbourne 1987 reported that women showed a tendency towards the system they were familiar with, although this was greater for women holding their own notes (author reported rate ratio 1.55; 95% CI 1.34 to 1.81); Lovell 1987 reported significantly more women in the study than the control group wishing to carry their notes in a future pregnancy (P < 0.001) and approximately one‐third of the control group also stated a preference for carrying their case notes. No other significant effects were reported, although both studies suggested that there were some administrative benefits because of fewer problems with missing notes; although women did very occasionally forget to take their notes to the clinic they never lost their own notes. In the parent‐held child health record study (Lakhani 1984), after one year's use the authors reported favourable comments on the booklet: from all the health professionals who responded to the questionnaires, and from the mothers who had received a booklet. There was, however, little apparent difference between the two groups in the uptake of immunisation (diphtheria/tetanus/polio first dose (experimental versus control): 89.8% versus 94.4%; second dose: 75.8% versus 85.2%; third dose: 22.3% versus 31.0%. Pertussis first dose (experimental versus control): 64.3% versus 57.8%; second dose: 58.0% versus 59.2%; third dose: 19.1% versus 19.0%) or developmental checks (first assessment (experimental versus control): 96.2% versus 97.9%; second assessment: 78.3% versus 83.8%). Studies comparing paper records with differing structuresThe pain management study (Stevens 1990) indicated a positive effect for the use of the flow sheet. This was based on the mean pain intensity ratings of the two groups reported on a scale of 0 to 10 as: 6.8/10 in the control group, and 4.2/10 in the experimental group (author reported P < 0.01); measured using an adapted form of the Children's Hospital of Eastern Ontario Pain Scale (CHEOPS). The author also reported that the children in the experimental group were assessed more frequently than those in the control group (5.4 versus 3.9 times in 24 hours, P < 0.01), and the experimental group received more analgesia than the control group (total doses 78 versus 34, P < 0.01). The study was extended for four weeks to give all the nurses experience in using the pain management sheet and in a subsequent satisfaction survey: 15 of the 24 nurses preferred the new flow sheet, eight preferred the existing method, and one was undecided. In the integrated record study (From 2003), the use of planning forms with the medical record was associated with a higher accuracy of planned length of stay (LOS) (the unadjusted mean difference of actual LOS minus planned LOS was 2.5 ± 5.6 days compared to 3.8 ± 6.5 days). The authors stated that the difference was significant (P = 0.02) when tested in a general linear model adjusting for ward of discharge: adjusted mean difference of 1.5 days (2.5 days compared to 4.0 days longer than planned) (95% CI 0.2 to 2.8 days). The mean LOS for patients with the planning form (using a general linear model) was 6.1 days compared to 7.6 days, a difference of 1.5 days (P = 0.02, 95% CI 0.2 to 2.7 days) (figures in table VII of paper, 6.2 days compared to 7.48 days, were not adjusted for ward of discharge). The authors stated that the accomplishment of the plan of action and the risk of readmission were unaffected. Studies comparing manual nursing care planning with computerised nursing care planningThe computerised nursing care planning study (Spranzo 1993), after controlling for selected covariates, showed only a negative effect of the computerised system on documented nursing care planning. Care planning increased on the control wards at the time of the post‐test (pre‐test: 28.08 versus 26.61, post‐test: 28.03 versus 33.30) and the author suggested that compensatory rivalry between the nurses might have contributed to this finding. No significant effects on patient outcomes were demonstrated, although there was a nonsignificant inverse relationship between the computerised system and patient satisfaction (pre‐test: 50.83 versus 48.84, post‐test: 49.98 versus 51.55) and readiness for self care (pre‐test: 20.73 versus 20.75, post‐test: 20.23 versus 20.83). There was also an inverse, but not significant, relationship between the computerised system and nurse attitudes to care planning (pre‐test: 67.1 versus 63.2, post‐test: 59.7 versus 63.4). No significant effect was found on nurse sickness, overtime, or job turnover. The baseline imbalance between the two groups may have been responsible for differences in the between‐group and within‐group values for some outcomes in this study. The study comparing a paper‐based record system with a computerised nursing record system (with a standardised nursing nomenclature) in a long‐term care facility (Daly 2002) showed no differences in patient outcomes between the two groups. The study did, however, show significant differences in the recording process itself. The time taken to produce the computerised record was significantly longer than for the paper‐based record (for example 3.78 h versus 2.31 h for the last time period (P = 0.002) for preparation time differences over all seven time periods measured), although the time spent on documentation generally decreased in both groups over the study period. Nursing diagnoses were similar for the two groups, although the authors reported a few more diagnoses made in the computerised group; there were considerable differences in the number of recorded nursing interventions (author reported P = 0.001) and activities (author reported P = 0.007). The authors reported that the director of nursing at the study site indicated that the nurses using the paper‐based system did not record all the care that was actually given whilst the nurses using the computerised system recorded everything. The third study comparing paper‐based and computerised nursing records (Ammenwerth 2001) did not include any patient outcomes. The authors reported that planning and documentation of tasks took significantly more time with the computerised system (4.8 min compared to 2.0 min per day, P = 0.004); similarly, report writing took significantly longer (6.6 min compared to 4.7 min per day, P = 0.019). There was no significant difference between the groups for care planning (16.4 min compared to 43.3 min per day, P = 0.131). The authors suggested that this was because only a small number of care plans (11 experimental and six control) were prepared. In this study, the documentation was reviewed by external nurse experts who reported that although the average number of aims and tasks were higher in the computerised care plans, only 80% of the planned tasks were carried out in the experimental group whilst 100% were carried out in the control group. The expert nurses gave a similar rating for the overall quality of the care plans in both groups. The legibility of the computerised plans was considerably better than in the paper‐based plans. In this study the physicians were asked their opinion of the two systems and three of the five doctors said that they read the computerised nursing record more often than in the previous system. The later interrupted‐time series study (three time points) (Mahler 2007) of four wards found that documentation quantity increased after the implementation of the computerised system, and the quality of documentation increased on three of the four wards. The fourth study, Bosman 2003 compared the proportion of time devoted to different categories of nursing activities as well as the time spent on various documentation activities (recalculated to indicate total time spent based on an eight‐hour shift). Time‐motion analysis for the admission phase indicated that the mean duration of the admission was slightly longer with the computerised system (18.1 ± 4.1 min compared to 16.8 ± 3.1 min, and the difference (1.3 min) was statistically significant (95% CI 0.04 to 2.72). For the registration phase, the nurses using the computerised system spent significantly less time on documentation activities (14.4% of observations compared to 20.5% (difference ‐6.1%; 95% CI ‐8.1 to ‐4.0)) and significantly more time on patient care (61% of observations compared to 54.9% (difference 6.1%; 95% CI 3.5 to 8.8). The differences for unit‐related and personal activities were not significant (difference of proportion 95% CI ‐2.1 to 0.0; ‐1.2 to 3.2, respectively). The difference in proportion of time spent on documentation compared to patient care varied with time after admission, but at the majority of time points the time spent on documentation and patient care varied significantly. The reduction was attributed to the time spent on hourly checks, entering of laboratory results (only done on the manual system), and calculation of fluid balance (only done on the manual system). The average number of vital sign annotations was significantly higher, on 17 out of the 24 hours, for the computerised intensive care information system (ICIS) compared to the paper system. The proportion of observations of time spent on reviewing or writing medical and nurses' notes in fact increased with the ICIS (9.4% compared to 5.0%), as did the shift change documentation (10.1% compared to 6.7%). The logging and validation of hourly checks remained the same as a proportion of documentation activities (29.4% in both groups), although the absolute number of observed logging and validation activities decreased. The estimates of time spent, based on an eight‐hour shift, indicated that the logging and validation of hourly checks took less time with the ICIS (20.3 min compared to 28.8 min) and two activities took longer with the ICIS: the shift change (7.0 min versus 6.6 min) and the reviewing and writing of medical and nurses' notes (6.5 min versus 5.0 min). This review of the nursing records literature was initiated in 1998 in response to the move to develop effective computerised systems for recording nursing care. It was realised that merely automating existing paper record systems was not the answer, and that the means to develop systems that accurately captured nursing practice lay in first acquiring a deep understanding of the complexities of the art and science of nursing itself. The review of the last 60 years of the literature on nursing records suggested that nurses and other healthcare professionals believe that there should be a link between nurse record keeping and the quality of care that patients receive. Although this review suggests that the body of rigorous research evidence on the topic is still very sparse, there is no doubt that nursing records is a topic that has been widely and extensively discussed in the professional literature. However, none of the studies produced results that could be interpreted as evidence of a change in practice that resulted from a change of record system. The studies indicate that nurses concur with professional bodies (for example ANA 2008; NMC 2008) that there is a link between their clinical practice and the nursing record that is beyond its obvious use as a notekeeping system. The client‐held record studies suggest altered and improved relationships between clients and health professionals; the pain management study sought to demonstrate a direct effect on patient care. The use and effectiveness of nursing care planning as a nursing activity is brought into question by the computerised care planning studies. In the controlled before and after study (Spranzo 1993), the author noted that the measured quality of nursing care remained constant despite the difference in documented care planning, and the study in a long‐term care facility (Daly 2002) showed no differences in patient outcomes between the computerised and paper‐based systems. The expert nurses who reviewed the documentation in the third study (Ammenwerth 2001) noted that there were no differences in the overall quality of the nursing documentation (although the extended study found that documentation quality improved on three of the four wards). Spranzo 1993 suggests that simply automating an existing paper system might not be expected to have an effect on the quality of care, Daly 2002 thought that the use of a standardised nomenclature within the computerised system would have an effect on patient care; but the results of these studies show that whilst the documentation changed the patient care given remained the same. This may not be surprising as in two of these studies the different systems were being used by a small number of nurses on the same ward. Preparation for change may need to be planned more carefully and with more awareness of the needs of nurses. A pre‐test, post‐test study (Thoroddsen 2007) of the implementation of nursing languages in nursing documentation details substantial collaborative work involved in the implementation, as well as research to develop a valid measurement instrument. What is being documented, how, and why deserves attention. The objective of the review was to establish the impact of nursing record systems on nursing practice and patient outcomes. The identified studies provide no evidence of any measurable difference, in nursing practice or patient outcomes, between the use of one kind of nursing record system or another. There may be social or administrative reasons for giving clients custody of their own records but there is no clear evidence of any clinical or psychological benefit. The results of the pain management study (Stevens 1990) suggest a beneficial effect might be obtained from a highly structured recording system. The study might be repeated, in a larger trial over a longer period of time, and might be extended to the care of adults as well as children, to see whether this is a real effect and whether it could be sustained. The use of a care planning form, with both the nurses' care plan and the medical care plan, that is integrated with a medical record may have an effect on patient care outcomes. Yet we have no details about the use the nurses may have made of a nursing care planning form retained within a separate nursing record, and details about the estimation of accomplishment of plans of action are very limited. There is a need for more research on how a multidisciplinary record might be used by the professionals who contribute to the record. It is unclear whether LOS is a good process indicator to use in relation to patient outcomes (From 2003). The computerised nursing care planning study (Spranzo 1993) raises a number of important issues. It is well thought out and carefully and thoroughly executed. The author discussed how the equivocal results of this study resembled the conflicting findings of the many other similar (but less well conducted) investigations in this area. The computerised care planning system used here was not linked to a patient assessment system or to progress reporting and did not provide any prompts or reminders. The author noted that the nurses were not involved in the development of the system, so that its use in practice had not been well thought out. She stated that it might thus have been possible to "have moved the Computerised Nurse Care Planning beyond a documentation tool and into the realm of actual use to improve provider strategies for care". Two other computerised record studies did include the full range of nursing documentation but they too failed to show changes in patient care or patient outcomes, and both studies indicated that more nurse time was required for the computerised systems. However, the numbers of patients and nurses in these studies were very small and larger studies, more rigorously controlled, might provide more robust results. Another computerised record study (Bosman 2003) showed that the proportion of time allocated to patient care increased with the use of computerised system (the proportion of time spent on unit‐related or personal activities remained very similar). The decrease in time spent on documentation is not unexpected, as some of the tasks required for the paper record system were wholly or partly automated with the computerised system (for example logging of hourly checks, laboratory results, and calculating fluid balance). The calculations indicate that the times required for reviewing and writing medical and nurses' notes and shift change documentation were slightly greater with the computerised system, although these activities do not account for as much time as some of the other documentation activities. The methodological shortcomings of the studies both included and excluded from this review have been discussed in detail above. It is clear that in the field of nursing record systems, as the methodology of nursing, health services and health informatics research develop, almost all these problems could be overcome. Several of the included and excluded studies suggest that an interrupted time series design would be appropriate and feasible, but the difficulty would be to ensure sufficient planning to collect data from an adequate number of time points before systems are changed. However, Spranzo 1993 provides the clue to the more fundamental problem which appears to have bedevilled the research in this area to date. This is the problem of regarding the nursing record simply as a documentation tool that has to be completed to standards, for example of the Joint Commission on Accreditation of Healthcare Organisations (JCAHO), or as paperwork that takes time to complete which might otherwise be spent at the bedside. One of the criteria for including studies in this review is that there should be objective measures of professional behaviour or patient outcomes, and the studies described here have all targeted their intervention towards some ultimate improvement in patient care. It would seem that it is only as the different purposes of the nursing record, and the ways in which it underpins care, are explored and understood that the right questions can be framed and the appropriate research studies designed, which can then lead the profession towards nursing record systems that are proven to be effective and are experienced as a necessary part of nursing practice. Qualitative nursing research could provide better understanding of nursing practice and of the kinds of record systems that would fully support effective nursing practice, but a search for qualitative data in this area (Urquhart 2005) revealed a surprisingly small number of studies. Imaginative qualitative research could be used as the precursor to well‐designed trials of nursing information systems in which nurses, health services, system suppliers, and patients could have confidence. The studies in this review (Bosman 2003; Elbourne 1987; Lakhani 1984; Lovell 1987; Stevens 1990) that set out to solve a relatively discrete and focused problem, for example effective pain management in children; empowering pregnant women and parents; reducing the loss of notes, time spent on data entry of test results, transcription errors, the number of pieces of paper in a record, all demonstrate some degree of success in achieving the desired results. The studies which, despite sophisticated study designs, still leave uncertain and equivocal results are the studies of nursing care planning systems and of the total nurse record. The recent, wider literature shows that, despite the developments in technology and investment in electronic records in health systems over the last 10 years, nursing records remain problematical. Record keeping is still seen as an extra burden, an administrative concern (NHS 2006) that is not an integral part of the care of patients, as professional regulators stress it should be (NMC 2008). Even the literature on the provision of decision support systems has failed to show conclusive benefits for nurses in their practice (York 2007). Many authors have identified the lack of engagement of clinical nurses in the development of nursing record systems as a cause of failure, and this may well be a contributory factor. We suggest, however, that the problem lies far deeper, in the nursing professions' understanding of nursing itself and how it is taught and practised. Nursing is variously seen as a process, a problem‐solving activity, or a caring art or science, which all offer different ways of articulating what is done with patients. But so far no demonstrably effective record systems have been developed to support any of these approaches. It is suggested that the design of nurse record systems has to be returned to the profession as an integral part of the development of the theory of nursing. Further research is required in this area. This review has shown that this is a topic of concern to healthcare professionals and to administrators. However, the focus of this concern and the research questions asked could now be better informed by the growing body of knowledge and expertise in health informatics. There is an opportunity to gain greater understanding of nursing practice through an exploration of the information it uses rather than simply addressing administrative issues, important as these may be. There is still a need for research into the relationship between clinical practice and its understanding and written expression by nurses. It is clear from the literature that the full range of the functions of a clinical nursing record are not well understood, and that exploratory research using qualitative techniques is required before more focused trials are undertaken. Effective nursing record systems will probably be shown to be as diverse as nursing practice itself. The challenge for nursing is to understand what characteristics of nursing practice are best served by which characteristics of record systems (Urquhart 2005). For example, which kind of practice needs a structured record, which kind needs to allow mostly free text? What aspects of nursing care need to be recorded and shared electronically, and which aspects of care are best served by a paper or voice format? Indeed, what aspects of nursing care need to be recorded at all? But the key question is ‐ how will an effective nursing record be recognised? This is a fundamental question which cannot be answered until the purposes of record keeping have been fully articulated. The review shows that this has yet to take place. These conceptual issues must be tackled, but the need to design and carry through research projects (both quantitative and qualitative) based on sound scientific methods must also be addressed. It is disturbing that so many studies of poor methodological quality should have been found in this topic area as they represent a considerable amount of effort and expense for little benefit or reward. There are many reasons why so much work in this area of nursing is so unsatisfactory. Some of these reasons are related to the lack of basic health informatics research, described above; and some are made explicit in the publications, particularly the late involvement of nursing staff in the procurement and implementation of computerised systems, and the difficulties of carrying out research projects across a number of units within healthcare institutions that are never static. This is particularly well illustrated by the studies by van Gennip 1995 and Burkle 1995. However, the study by Spranzo 1993 shows that it is possible to carry through a well thought out and rigorous investigation; and that many of the studies could easily have been improved. It is clear that RCTs and other experimental and quasi‐experimental designs would be able to address questions in this area. The most valid timing of an RCT needs to be discussed. Bosman 2003 conducted the RCT over six months after implementation of the computerised system, other RCTs have been done close to the time of implementation. Practice changes may require time to develop and this means that longitudinal evaluation may be helpful. For example, this could be done by preparing ahead of time for a change; and taking measurements at appropriate intervals for a time series analysis; or including control units in both the before and after stages of a study and continuing testing beyond the implementation phase of projects, into the maturity of the record system. There are, however, indications that finding a valid control site is difficult (Ammenwerth 2001‐ study by Ammenwerth 2003). The development of effective nursing record systems, together with all health and medical record systems, is certainly a difficult area for research but one in which rigorous research is both necessary and possible. The review authors wish to acknowledge the support of the Wales Office of Research and Development in Health and Social Care, who funded the original review, and the 2007 to 2008 Department of Health (England) Incentive Scheme for Cochrane reviews and updates for funding for the update. We would like to acknowledge the work of Paul Wainwright, who was a co‐reviewer for the original review. We are also grateful to the EPOC Editorial Team for their support and encouragement. The review and updates could not have been completed without considerable help from the staff of several libraries, in the UK and USA, but particular thanks go to the staff of the libraries at Aberystwyth University, the Royal College of Nursing, University of Salford, and Suffolk Primary Care Trust. And finally, we are very grateful to all those in the nursing research and nursing informatics community who gave us help, support, information about their studies, and encouragement. Appendix ‐ MEDLINE OVID Search Interface 1. Evaluation Studies/ 2. Nursing Evaluation Research/ 3. Nursing Administration Research/ 4. "Costs and Cost Analysis"/ 5. Cost‐Benefit Analysis/ 6. exp "Outcome and Process Assessment (Health Care)"/ 7. exp "Task Performance and Analysis"/ 8. Clinical Nursing Research/ 9. Workload/ 10. exp Nursing Staff/ 11. Medical Records/ 12. "Forms and Records Control"/ 13. Medical Records, Problem‐Oriented/ 14. Nursing Records/ 15. Medical Records Systems, Computerized/ 16. Documentation/ 17. Point‐of‐Care Systems/ 18. Patient Care Planning/ 19. Hospital Information Systems/ 20. randomized controlled trial.pt. 21. controlled clinical trial.pt. 22. Intervention Studies/ 23. experiment$.tw. 24. time adj series).tw. 25. pre test or pretest or (post test or posttest)).tw. 26. Random Allocation/ 27. impact.tw. 28. intervention?.tw. 29. chang$.tw. 30. evaluat$.tw. 31. effect?.tw. 32. comparative study.pt. 33. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 34. 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 35. 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 36. 33 and 34 and 35 37. limit 36 to yr="2003 ‐ 2007" o / indicates subject heading. o exp indicates retrieving results using the selected term and all of its more specific terms. o .pt. indicates publication type. o $ indicates unlimited truncation retrieves all possible suffix variations of the root word. o .tw. indicates searching for the exact word or phrase in the titles and/or abstracts of articles. o ? indicates a ‘wild card’ can be used within or at the end of a search term to substitute for one or no characters. Appendix ‐ EMBASE NLH Search 2.0 1. MEDICAL RECORD/ 2. MEDICAL DOCUMENTATION/ 3. HOSPITAL INFORMATION SYSTEM/ 4. PATIENT CARE PLANNING/ 5. 1 OR 2 OR 3 OR 4 6. EVALUATION/ 7. NURSING EVALUATION RESEARCH/ 8. COST/ 9. COST BENEFIT ANALYSIS/ 10. OUTCOME ASSESSMENT/ 11. TASK PERFORMANCE/ 12. CLINICAL NURSING RESEARCH/ 13. WORKLOAD/ 14. NURSING STAFF/ 15. NURSING ADMINISTRATION RESEARCH/ 16. 6 OR 7 OR 8 OR 9 OR 10 OR 11 OR 12 OR 13 OR 14 OR 15 17. RANDOMIZED CONTROLLED TRIAL/ 18. random*.af 19. experiment*.af 20. (time ADJ series).af 21. ((pre ADJ test) OR pretest OR (post ADJ test) OR posttest).af 22. impact.af 23. intervention*.af 24. chang*.af 25. control*.af 26. evaluat*.af 27. COMPARATIVE STUDY/ 28. CONTROLLED CLINICAL TRIAL/ 29. 17 OR 18 OR 19 OR 20 OR 21 OR 22 OR 23 OR 24 OR 25 OR 26 OR 27 OR 28 30. 5 AND 16 AND 29 31. 5 AND 16 AND 29 [Limit to: Publication Year 2003‐2007] o / indicates subject heading. o * indicates unlimited truncation retrieves all possible suffix variations of the root word. o .af indicates a search undertaken in any field. o ADJ indicates that the words separated by ADJ have to be next to each other. Appendix ‐ CINAHL Ovid Search Interface 1. exp Nursing Protocols/ 2. Nursing Records/ 3. Problem Oriented Records/ 4. Computerized Patient Record/ 5. Charting/ 6. DOCUMENTATION/ 7. Pain Measurement/ 8. Clinical Information Systems/ 9. exp Patient Record Systems/ 10. Hospital Information Systems/ 11. Clinical Assessment Tools/ 12. Nursing Outcomes/ 13. exp "Process Assessment (Health Care)"/ 14. "Record Review"/ 15. exp Evaluation Research/ 16. "Outcomes (Health Care)"/ 17. exp "Task Performance and Analysis"/ 18. Program Evaluation/ 19. Change Management/ 20. "Cost Benefit Analysis"/ 21. "COSTS AND COST ANALYSIS"/ 22. Quality Assurance/ 23. PRODUCTIVITY/ 24. Job Performance/ 25. clinical trial.pt. 26. exp Pretest‐Posttest Design/ 27. exp Prospective Studies/ 28. Random Assignment/ 29. Comparative Studies/ 30. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 or 10 31. 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 32. 25 or 26 or 27 or 28 or 29 33. 30 and 31 and 32 34. limit 33 to yr="2003 ‐ 2007" o exp indicates retrieving results using the selected term and all of its more specific terms. o / indicates subject heading. o .pt. indicates publication type. Appendix British Nursing Index OVID Search Interface 1. health economics/ 2. "outcomes and prognoses"/ 3. "standards and guidelines"/ 4. patients records/ 5. evaluat*.ti. 6. evaluat*.ab. 7. management information systems/ 8. "care plans and planning"/ 9. randomised controlled trial*.ti. 10. randomised controlled trial*.ab. 11. randomized controlled trial*.ti. 12. randomized controlled trial*.ab. 13. controlled clinical trial*.ti. 14. controlled clinical trial*.ab. 15. pre test.ti. 16. pretest.ti. 17. pre test.ab. 18. pretest.ab. 19. post test.ti. 20. posttest.ti. 21. post test.ab. 22. posttest.ab. 23. time series.ti. 24. time series.ab. 25. impact.ti. 26. impact.ab. 27. intervention*.ti. 28. intervention*.ab. 29. chang*.ti. 30. chang*.ab. 31. effect*.ti. 32. effect*.ab. 33. 1 or 2 or 3 or 4 or 5 or 6 34. 7 or 8 35. 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 36. 33 and 34 and 35 37. limit 36 to yr="2001 ‐ 2002" 38. limit 36 to yr="2003 ‐ 2007" o / indicates subject heading. Withdrawn from publication for reasons stated in the review Experimental nursing record system compared with conventional record
Last assessed as up‐to‐date: 1 November 2008.
Review first published: Issue 2, 2000
For the 2008 update, Maria Grant prepared the major database search strategies, ran the searches, and coordinated the sifting and selection processes together with obtaining full text of studies for review. Rosemary Currell, Maria Grant, Nicholas Hardiker and Christine Urquhart sifted the search outputs. Rosemary Currell, Nicholas Hardiker and Christine Urquhart undertook the handsearches. Christine Urquhart edited the review, with contributions from all the team.
The review authors are aware of no conflicts of interest relevant to this review. This review is out of date; although it was correct as of the date of publication. For further information, please contact the Managing Editor at EPOC References to studies included in this review
References to studies excluded from this review
References to studies awaiting assessment
References to ongoing studies
Additional references
|