The impact of pre-operative education on recovery following coronary artery bypass surgery

Transcription

1 European Heart Journal (2002) 23, doi: /euhj , available online at on The impact of pre-operative education on recovery following coronary artery bypass surgery A randomized controlled clinical trial C. M. Shuldham, S. Fleming and H. Goodman Royal Brompton & Harefield Trust, London, U.K. Aims Research has demonstrated the beneficial impact that pre-operative education exerts on the postoperative recovery of patients having surgery but little work has focused specifically on cardiac surgery. Therefore a randomized controlled trial was designed to elucidate the consequences of pre-operative education, given before admission, on postoperative pain, anxiety, depression and wellbeing in the 6 months following a first episode of coronary artery surgery. Method and Results Three hundred and fifty-six people were randomized into the study, with 188 in the experimental and 168 in the control arms. Patients in the experimental group received the intervention, a day of education by members of the multidisciplinary team, prior to admission for surgery. Experimental and control subjects had the usual care, which involved education on admission and throughout their stay in hospital. Measurement was conducted on entry to the study, before randomization, and at 3 days, 6 weeks, 3 months and 6 months following operation. A variety of tools were used: the SF-36 Health Status questionnaire, the Hospital Anxiety and Depression scale, the General Well-Being questionnaire and a pain measurement tool. Analysis was done using the intention-totreat principle and non-parametric statistics. There were no significant differences between groups in the primary outcomes namely anxiety (P=0 09) and pain (P=0 48), or in depression (P=0 62) and wellbeing ( worn out P=0 11; tense and uptight P=0 29) 6 months after operation. This was also the case 3 days after coronary artery surgery. There was a significant difference in length of hospital stay (P=0 01) with the experimental group having the longer stay. These findings contrast with much of the existing evidence. Conclusion The findings demonstrate that there is no benefit to be gained from this form of pre-operative education and that there is an associated increase in length of hospital stay. Future research could examine an ongoing programme of education and support, and might use alternative methods such as CD-ROM or the Internet. (Eur Heart J, 2002; 23: , doi: /euhj ) 2002 The European Society of Cardiology Key Words: Pre-operative education, coronary artery surgery, anxiety, depression, wellbeing. See page 600, doi: /euhj for the Editorial comment on this article Introduction There is a considerable body of research examining the impact of pre-operative education on recovery following surgery. In the U.K., pre-operative teaching was shown to be effective in reducing stress [1] and pain and anxiety [2] in postoperative general surgical patients. Many of the reported studies are from North America and meta-analyses have shown a beneficial effect from Revision submitted 10 July 2001, and accepted 11 July Correspondence: Dr C. M. Shuldham, Royal Brompton and Harefield Trust, Sydney Street, London SW3 6NP, U.K X/02/ $35.00/0 pre-operative education. For example Hathaway [3] analysed 68 studies and found a 20% improvement in postoperative outcomes in experimental subjects. Devine and Cook s [4] meta-analysis of 102 studies focused on how psycho-educational interventions influence recovery. Analysis showed a positive effect in respect of pain, psychological wellbeing and satisfaction with care in this and a later meta-analysis [5]. Overall, the meta-analyses [3 6] demonstrated that preoperative education has a positive influence on recovery after surgery, despite the fact that individually many of the studies fail to show a significant difference between experimental and control groups. Studies have been criticized, by for example Lindeman [7] and Shuldham [8], 2002 The European Society of Cardiology

2 Pre-operative education and bypass surgery 667 for lack of blinding, unclear randomization protocols, lack of homogeneity in the sample and insufficient size. Detail is rarely given about the usual care received, and experimental interventions are short, lasting on average 42 min [4] and falling in later studies to 30 min [5]. Most researchers collected data from subjects approximately 3 4 days after operation. The longer term effects of teaching are not known, the impact of illness severity, age and sex are also unclear. Although positive effects were found irrespective of who provided the education it is not possible to examine whether the results are generalizable to conditions where regular staff rather than researchers conduct the teaching. Cardiac surgery Several studies explore the issue of education in relation to cardiac patients. Some include all cardiac surgical patients whether they are having coronary artery bypass grafts (CABG) or valve surgery [9,10]. Others recruit only CABG patients [11 16]. Review of these studies [17] suggests that the impact of pre-operative education for people having coronary artery bypass surgery is far from clear and further research is warranted. That said, a recent study [18] not included in the review showed that a multidimensional pre-operative intervention led to a reduction in length of stay and improved quality of life. Research design and method A randomized, single blind, controlled clinical trial was designed. This set out to test the null hypothesis that, compared to a referent group, there would be no reduction in postoperative anxiety, pain, depression and length of stay and no increase in wellbeing in patients who have received pre-operative education, at 6 months following a first episode of coronary artery bypass surgery. Data from a previous study [19] were used for a power calculation. This was set to demonstrate a 25% reduction in outcome measures (pain, anxiety and depression) in the experimental as compared to control subjects, at the 0 05 level of significance with 90% power. It was calculated that 276 patients were required. As mortality was known to be 2 5% this also needed to be accounted for in recruitment. Ethical approval for the proposal was obtained from the local research ethics committee. All letters, consent forms and patient information leaflets were reviewed. Each patient was given an information sheet and a copy of the signed consent form. Entry criteria The population under study was all the people at the study hospital awaiting coronary artery bypass surgery for the first time. There were approximately 700 patients a year. Patients had to be 18 years of age or older, having a first episode of coronary artery bypass graft (CABG) surgery at the study hospital and be willing and able to attend the education clinics. As the education and questionnaires were in English, patients had to be fluent in that language. Study protocol Patients who fulfilled the entry criteria were approached by telephone. Those who could not be reached in this way were sent a letter of invitation. People who expressed a willingness to consider the study were sent a letter, information and a consent form. Following receipt of the consent form, the patient was sent the initial outcome measures. There were four questionnaires. The SF-36 [20] was used for patient characterization prior to randomization to allow identification of important differences between experimental and control groups in analysis. The Hospital Anxiety and Depression (HAD) scale [21] measured anxiety and depression and a Visual Analogue Scale (VAS), body map and categorical rating scale for pain intensity and relief, all focused on pain. This was a new composite questionnaire. A reliability study on 101 pairs of questionnaires using McNemar s test showed no difference in pain intensity (P=1 00) or relief (P=0 73) between assessments. Finally the General Well-Being Questionnaire (GWBQ) [22] was used. Once these had been returned patients were randomized by another member of the department, using computer generated random numbers to reduce bias in allocation of the patients to the two groups. In this way the researchers, members of the educational programme and staff on the ward were blinded to the patient group. People in the experimental group were then invited to the education clinic, run by an independent team of staff. Patient education programme: the intervention Pre-admission patient education in the study hospital was already offered to some patients [23,24]. Audit evaluation showed that patients enjoyed the day, found it beneficial [25] and that their pain was reduced [24]. The hospital education programme for cardiac surgical patients comprized both the pre-hospital component, the independent variable in this study, and the usual care, namely the programme of inpatient education. These involve members of all the health care professions (nurses, physiotherapist, doctor) and although there is a team, over time individuals changed. In order to overcome the difficulties this presented, an agreed brief was used for each session in the intervention and usual education programmes.

3 668 C. M. Shuldham et al. Table 1 Education intervention Pre- and postoperative care* Coronary artery disease v Coronary vein grafts v Internal mammary artery grafts Hospital stay v Process of admission v Pre-operative procedure v Postoperative process v Expected stay Medical care v Revascularization v Potential complications v Medication v Health promotion Rehabilitation v Physiotherapy v Breathing exercises v Walking v Exercise and recreation v Food and eating v Cardiac rehabilitation *An informal approach was used, encouraging discussion and questions and answers. Each patient assigned to the experimental arm attended the hospital once for approximately 4 h in a group of between 10 and 15 people early in the waiting period. Relatives could choose to attend. The education programme provided information on pre-operative events and likely progress. Factors personal to the patient such as recovery and stay in intensive care/ recovery, pain and analgesia were discussed (Table 1). The atmosphere was informal with questions and discussion encouraged. Videos were used and a package of written information given to everyone. An optional visit to the wards and intensive care unit was arranged. The researchers did not join in with the day. The intervention was more comprehensive in the range of topics, duration and variety of staff than any example in the literature. The usual education provided, informally on a oneto-one basis by staff as part of the normal care which control and intervention groups received, started at the time of admission, anything from a few days to the day before surgery. Nurse, doctor, physiotherapist, occupational therapist, pharmacist and dietician all participated. This individual teaching was supplemented by a regular series of sessions on the wards to which patients were invited. Patients not in the research also attended both the pre- and in-hospital programmes and staff were unaware of whether a person was or was not in the study. Study patients were asked not to alert staff to their assigned group. Prior to commencing the main study a convenience sample of ten patients was assigned randomly for approach by letter or telephone call. Five people contacted by each method returned consent forms and questionnaires. Each person was visited on the ward, 3 days after operation. Ten more in-patients were recruited to test further the postoperative data collection process. No material changes were made following this. Measurement Baseline measures were used to characterize the sample on entry to the study (Table 2). The questionnaires were next administered in a face-to-face interview on the third postoperative day. If the patient was still in the intensive care unit then these questionnaires were given 3 days after transfer to the ward. Thereafter questionnaires with accompanying letters were sent by post to the patient 6 weeks, 3 and 6 months after operation. On each occasion the questionnaires were administered in the same order: the SF-36, followed by the HAD, Well-Being and Pain. Nonrespondents were followed-up 2 weeks after the questionnaires had been sent with a letter and further questionnaires. If after 7 more days they still had not replied they were telephoned. When two sets of questionnaires were returned together the one labelled for that sequence was used. Data were entered on an Access database and checked, blind to the patient s group. Analysis, using SPSS Version 9, was only undertaken once all the patients had completed the study. Missing data Each questionnaire was examined for missing responses. The SF-36 Manual [26] recommends that a scale score be calculated if a respondent answered at least half of the items in each of the two major scales. Using this formula, 14 questionnaires were removed at baseline and 19 from the 6 month data. For the HAD scale, questionnaires were deleted where more than one response on a scale (anxiety or depression) was missing. Where only one response was absent, the mean of the rest was entered in the vacant slot. This is recommended practice in the German manual (Herrmann: personal communication) and the majority of questionnaires could be used. The General Well-Being Questionnaire had the most missing data and the same principles were applied. Forty-one questionnaires were removed for the tense and uptight dimension across the five measurement points and 50 for the worn out scale. The pain questionnaire comprized of a series of related questions that did not produce one overall summative score and there was no reasonable way of estimating a value for missing responses. These were treated as missing values. Length of hospital stay was calculated from the day of surgery and included the day of discharge. An index of case-mix severity, the Parsonnet score [27] was used on admission and the APACHE 11 score [28] for patients in ICU. Trial profile Over one thousand people were approached to enter the study and of the 374 who agreed, 188 were randomized

4 Pre-operative education and bypass surgery 669 Table 2 Baseline measurements Physical health* Mental health* Anxiety Depression Worn out Tense and uptight Pain n Mean Standard deviation Median Range Source: SF-36; HAD; GWBQ and VAS. *The SF-36 works in the opposite direction to the other tools, so a higher score reflects better health. into the experimental arm and 168 to the control (Fig. 1). Some patients did not follow the protocol, usually because they were admitted before the education intervention could be given. The per-protocol group was small and so later in the study 2:1 randomization was introduced to increase the numbers entering the experimental group. Rigorous attempts were made to keep in contact with participants, however despite this, 27 patients were lost to follow-up or withdrawn. The reasons include the fact that some patients did not have the planned operation and that some patients died. Characteristics of the total study sample In order to establish whether randomization had ensured equality between the study groups, experimental and control patients were compared using the intentionto-treat principle. There were no statistically significant differences for sex, age, consultant or the category of urgency on the waiting list between these two groups (Table 3). There was a difference in marital status, with proportionately slightly more divorced people in the control as compared to the experimental arm and a few single people. There wasadifference in the length of wait with double the number of experimental patients admitted within 3 months compared to control and more control patients waiting over a year. Analysis showed no significant differences in ethnic or occupational group. Risk, as defined by the Parsonnet score was similar between the groups. Mann Whitney U tests were also used on the results of the SF-36 physical and mental health scores, the anxiety and depression scales of the HAD, Pain and the two dimensions of General Well-Being ( worn out, and tense and uptight ). All showed no significant differences. Thus the only statistically significant and potentially important differences between the two groups were in marital status and length of wait for operation, the control group having the longer wait. Comparison between people who completed the protocol and those who did not Because several patients did not follow the protocol fully, comparison was carried out to compare those who followed the protocol with the people who did not. Differences were established in respect of the consultant responsible for their care (chi-squared=17 15 df=6; P=0 01), ethnic group (chi-squared=27 87 df=1; P<0 001) and the length of wait (chi-squared=28 35 df=3; P<0 001). This probably resulted from the fact that some patients were admitted prior to receiving the intervention and some who waited a long time were transferred to another hospital. Outcomes 6 months after surgery For the purposes of the primary analysis, the point 6 months after operation was selected, as existing research had examined the early but not the longer-term impact of education given to patients before operation. The difference between baseline measurement and that at 6 months for each outcome was calculated. This showed that people had generally improved on all dimensions. Where one group had improved more than the other, the changes always favoured the control (Table 4). Differences in anxiety and pain, on the Visual Analogue Scale, between experimental and control groups for all people in the study were examined as the primary outcome and showed no significant difference between the groups. Analysis also demonstrated no significant difference in depression between experimental and control subjects. The overall picture of no significant differences on the outcomes was confirmed for wellbeing i.e. worn out and tense and uptight, 6 months after operation. Thus the null hypothesis that the educational intervention had no impact on the outcomes measured was supported, in respect of 6 months after operation (Table 4). However, the findings in respect of the dimensions of wellbeing should be viewed with some caution as the GWBQ questionnaires suffered from a number of missing responses. Outcomes 3 days after operation It is important to consider whether a difference existed early on, but had disappeared later in the postoperative course. Analysis of the variables that represented the change from baseline to 3 days after surgery showed no significant differences between experimental and

5 670 C. M. Shuldham et al. Eligible patients approached n = 1028 Agreed to enter n = 374 Refused the trial n = 654 Randomized Yes n = 356 No n = 18 Experimental n = 188 Control n = 168 Followed protocol n = 124 Followed protocol n = 145 Did not follow protocol n = 49 Did not attend clinic n = 33 Operation at other hospital n = 11 Different operation n = 4 Attended clinic at other hospital n = 1 Did not follow protocol n = 11 Different operation n = 5 Operation at another hospital n = 4 Early discharge research project n = 1 Attended clinic by choice n = 1 Withdrawn n = 15 Died before admission n = 3 Died following operation n = 2 Too ill after surgery n = 1 Heart transplant n = 1 Did not have operation n = 3 Did not want to continue n = 1 Operation at another hospital n = 1 Withdrawn n = 12 Died n = 2 Did not have operation n = 4 Operation at another hospital n = 6 (too late for follow up) Completed trial n = 173 Completed trial n = 156 Figure 1 Trial profile. control groups. This included anxiety (Mann Whitney U=11 636; Z= 0 28; P=0 78), pain (Mann Whitney U= ; Z= 0 72; P=0 47), depression (Mann Whitney U=10 756; Z= 1 24; P=0 22) and both aspects of wellbeing: worn out (Mann Whitney U=9717 5; Z= 1 49; P=0 14) and tense and uptight (Mann Whitney U=10 008; Z= 1 27; P=0 21). These analyses demonstrate that not only was there no significant difference between groups late in the postoperative course but that contrary to the findings from other studies, there was also no significant difference early after coronary artery bypass surgery. This was the case even when only the patients who had followed the research protocol were examined. The fact that the finding of no difference was sustained when all the subjects analysed in the experimental group will have received the education intervention, adds weight to the overall result that education given pre-operatively has no impact on postoperative anxiety and pain or on depression and wellbeing. However analysis of baseline data had shown a difference between the experimental and control group in the length of time they had waited for operation. This could act as a possible confounding factor, associated with the outcomes. A Kruskal Wallis test showed no differences between the groups (Table 5) in respect of the outcomes discussed. This suggests that the variation in waiting time for operation that occurred between experimental

6 Pre-operative education and bypass surgery 671 Table 3 Summary comparison of experimental and control groups at baseline: intention-to-treat Patient characteristic Experimental n=173 Control n=156 Chi-squared df P-value Sex Male 155 (90%) 133 (85%) Female: n (%) 18 (10%) 23 (15%) Age: mean (SD) 62 7 (7 46) 62 3 (8 46) Ethnic group: n (%) White 146 (84%) 143 (92%) Other 22 (13%) 10 (6%) Not classified 5 (3%) 3 (2%) Occupation: n (%) 1 and 2 60 (34%) 66 (42%) N and 3M 70 (41%) 61 (39%) 4 and 5 14 (8%) 11 (7%) Unclassified 29 (17%) 18 (12%) Marital status: n (%) Married 130 (75%) 125 (80%) Single 14 (8%) 3 (2%) Widowed 7 (4%) 9 (6%) Divorced 8 (5%) 13 (8%) Not known 14 (8%) 6 (4%) Consultant: n (%) A 30 (17%) 31 (20%) B 12 (7%) 12 (8%) C 17 (10%) 16 (10%) D 6 (4%) 4 (3%) E 35 (20%) 35 (22%) F 61 (35%) 55 (35%) G 12 (7%) 3 (2%) Urgency: n (%) Very urgent 43 (25%) 32 (20%) Urgent 26 (15%) 15 (10%) Not urgent 103 (59%) 107 (69%) Not categorized 1 (1%) 2 (1%) Length of wait: n (%) 0 3 months 20 (12%) 11 (7%) months 58 (34%) 49 (31%) 7 12 months 84 (47%) 75 (48%) >12 months 11 (6%) 21 (14%) Mann Whitney U Z score Parsonnet: median (range) 3 (0 29) 3 (0 18) Anxiety: median (range) 7 (0 20) 7 (0 20) Depression: median (range) 5 (0 15) 5 (0 16) Worn out: median (range) 21 (2 39) 19 (5 41) Tense and uptight: median (range) 14 (0 33) 13 (1 37) Pain: median (range) 40 (0 98) 37 (0 94) Physical health: median (range) ( ) ( ) Mental health: median (range) ( ) ( ) and control subjects did not affect the findings of no difference between the groups. Length of stay in hospital The final question was whether there would be a reduction in length of stay in hospital for patients who had received teaching before admission as compared to counterparts who had not participated in this education. The pattern of the data suggested that it may have been log normally distributed; however, transformation and analysis using the Kolmogorov Smirnov one sample test for normality (P<0 001), demonstrated that it was not normally distributed. Therefore, the non-parametric Mann Whitney U test was used. A significant difference between groups (P=0 01) was found with the control patients having the shorter stay (Table 6). Again because the disparity in the length of wait for operation between experimental and control subjects could have acted as a confounding factor, patients were categorized according to the time waited. Analysis with

7 672 C. M. Shuldham et al. Table 4 Outcomes at 6 months after operation Outcome n= Experimental n= Control Mann Whitney U Z score P-value Anxiety Median change (range) ( 9 to+13) 151 2( 8 to +13) Pain Median change (range) ( 86 to +98) ( 83 to +82) Depression Median change (range) ( 7 to+14) 148 2( 8 to +12) Worn out Median change (range) ( 19 to +20) ( 14 to +30) Tense and uptight Median change (range) ( 13 to +21) ( 13 to +34) Length of stay Mean (SD) (5 04) (4 38) Table 6 Length of hospital stay Intention-to-treat Mean length of stay (in days) Standard deviation Median (in days) Range Experimental group (n=162) Control group (n=152) Table 7 Summary of findings Outcomes Anxiety Pain Depression Worn out Tense and uptight Length of stay Six months P= Three days P= Hospital stay P= 0 01 Mann Whitney U tests and levels of significance. Table 5 Comparisons of experimental and control groups: length of wait Outcome Chi-square df P-value Anxiety Pain Worn out Tense and uptight Depression Kruskal Wallis test. a Kruskal Wallis test revealed no statistically significant differences between the two arms of the trial in each of the subgroups (chi-squared=1 13; df=3; P=0 77). In order to see if a difference in postoperative illness severity might have existed between groups, data on APACHE scores were analysed. These showed no significant differences between groups in relation to their APACHE score 24 h after surgery (Mann Whitney U 2228; Z= 0 21; P=0 84) or on discharge from intensive care (Mann Whitney U 1720; Z= 0 71; P=0 48). This all supports the proposition that education given before admission is actually associated with a longer length of stay following a first episode of coronary artery surgery. From this analysis, the hypothesis of no difference between trial groups in respect of anxiety, pain and the secondary outcomes of depression and wellbeing has been upheld (Table 7). When it came to the impact of education on the time spent in hospital the hypothesis was rejected. The control group had the shorter stay. Discussion When considering the results it is worth bearing in mind the study sample. The majority of patients in this research were white, male and from the third (manual) occupational group upwards. People had to understand and write English. The other important factor is that the extra visit to the hospital acted as a deterrent to many

8 Pre-operative education and bypass surgery 673 patients who did not wish to travel or who could not attend the clinic. Perhaps also they did not feel well enough to take on a commitment to the research, although the median Parsonnet score was the same for people who declined the study and those who joined. Such factors nevertheless limit the extent to which findings can be generalized to the whole population of people having coronary artery surgery and the results should be reviewed with this in mind. Analysis of data from patients when they entered the study showed that randomization had ensured equality in the groups, with the exception of length of wait for operation. The control group had the longer wait; however, statistical tests suggested that this had not acted as a confounding factor. It is important to consider whether some other circumstance influenced the length of stay for one group more than the other. During the research a greater proportion of the control patients as compared to intervention went through the fast-track system of care but about the same percentage from each group went to intensive care. However, most of the small number of people who returned to ICU came from the experimental group. Was one group more sick than the other and did this have an impact on their stay in hospital? Although, from the data collected in the study, small dissimilarities were seen in the progress of patients, the scale of these appears insufficient to have affected the variation between groups. For patients requiring intensive care there was no significant difference in their APACHE scores 24 h after surgery or on discharge from intensive care. Patients from both groups remained in the unit for similar lengths of time. In addition to these known factors, there are other unknown issues that may have exerted an influence. These include severity of disease as rated by findings from coronary angiography, length of time on cardiopulmonary bypass, number of grafts and intraoperative blood loss. The study took place over nearly 4 years and patients had surgery during a period of 3 5 years. Innovations, such as beating heart surgery, were introduced in that time. And yet the differences in length of stay between experimental and control groups was not accounted for by the timing of surgery, either early or late in the study. However, whilst the difference is indeed statistically significant, it is not clear the extent to which a differential stay of one day is important clinically for patients. Critique of the study There are many reasons why a study might fail to find a significant difference between groups. The first is that there really is no significant difference on the outcomes chosen between experimental and control patients. For this to be the case the research sample has to be large enough for important differences, if present, to be detected. The fact that 269 patients followed the protocol completely gives a power estimate of well over 80%, to show a 25% variation between groups at the 0 05 level of significance. Confounding was minimized by randomization and patients in the two groups were similar at baseline, however, some factors, such as the interval between education and surgery, could not be controlled. The patient was treated as the unit of analysis (Altman 1997) throughout the study, which can mean more insignificant, but true results are likely. There was a differential loss of subjects from the comparison groups, however, the finding of no difference was confirmed when only the per-protocol patients were analysed, suggesting that difficulty with the process of the study did not change the result. Discussion of the lack of differences between groups has to consider the measurement tools. All suffered more than expected from missing data, although the loss of patients to follow-up described above together with levels of missing data were below the 15% overall and 10% difference between groups, suggested as important by Devine and Cook [6]. A strength of this study, and a reason as to why the significant and non-significant results may be true where others have reported a different answer, is that, although subjects knew to which group they had been assigned, staff were unaware of, i.e. blind to, the allocation. Data were collected, inputted and checked, blind to the person s assigned group. No subjective estimates of the patients condition postoperatively were made: patients responses were the basis of all data and analyses. Therefore the chance for staff to affect the results were considerably less than in some comparable studies. Education, both control and intervention, was given by hospital staff as part of their normal practice and as the independent variable in this study, the education intervention, was stronger than any other in the literature and used all the strategies recommended. Devine [5] had made the point that whereas positive findings resulted from education, this was usually given by the researcher and it was not possible to examine whether findings could be generalized to situations where regular staff provide the teaching. The present study suggests that they cannot. The study compared one form of education with another, and not with no education. In generating an insignificant result on many of the outcome measures, the research does not prove that education fails to benefit patients, but rather that education before admission does not benefit patients in respect of certain measures. It also suggests that in some cases it may have a negative impact, namely on length of stay. There may also have been differences on outcomes other than the factors that were the focus of this study. Implications for practice Research seeks to provide definite answers to important questions, which can inform clinical practice, enabling evidence based patient care. The results challenge some of the generally accepted beliefs about pre-operative education. The dilemma they generate is over whether to

9 674 C. M. Shuldham et al. continue additional education in the absence of demonstrable benefit, and in the light of the association with a longer length of stay. The education clinic could be a service available for those who chose it, rather than a part of the routine pre-operative programme. For the wider population of people having cardiac surgery at other centres and where such a programme is not available then a cautious approach should be adopted, and new programmes evaluated. Future research directions The waiting period for cardiac surgery is long, frequently over a year, although recent government plans have target improvements. A single day of education without an associated programme of support may be inadequate. Thus an alternative is to develop the intervention into a programme of education and support throughout the waiting period, with a trial to evaluate the impact. Telephone interventions, education using CD-ROM and the Internet all offer possibilities. Programmes could be tailored more accurately to meet individual needs and patients could help develop them. There are many changes in technology that could be exploited to benefit people having all kinds of treatment and which are in their infancy. Now is the time to evaluate them in well-designed trials, before they become established without evidence of effectiveness. Finally the emphasis in the literature of pre-operative education has been on the impact postoperatively, and the present study is no exception. Little is known about the effect on pre-operative measures. Many staff of the Trust participated in this research study and we wish to acknowledge their contribution. In particular Professor Anthony Newman Taylor and Professor Susan McLaren who supervized the lead author during the study. We also want to thank the people having cardiac surgery for participating in the study at a very stressful time in their lives. The research was supported by the North Thames Responsive Funding Group and Royal Brompton and Harefield NHS Trust. References [1] Boore J. Prescription for Recovery. London: RCN, [2] Hayward J. Information a prescription against pain. London: RCN, [3] Hathaway D. Effect of pre-operative instruction on postoperative outcomes: a meta-analysis. Nurs Res 1986; 35: [4] Devine EC, Cook TD. Clinical and cost saving effects of psychoeductional interventions with surgical patients: a meta-analysis. Res Nurs Health 1986; 9: [5] Devine EC. Effects of psychoeducational care for adult surgical patients: a meta-analysis of 191 studies. Patient Educ Couns 1992; 19: [6] Devine EC, Cook TD. A meta-analytic analysis of effects of psychoeducational interventions on length of post-surgical hospital stay. Nurs Res 1983; 32: [7] Lindeman CA. Patient Education. Ann Rev Nurs Res 1988; 6: [8] Shuldham CM. Pre-operative education a review of the research design. Int J Nurs Stud 1999; 36: [9] Grady KL, Buckley DJ, Cisar NS, Fink NM, Ryan SD. Patient perception of cardiovascular surgical patient education. Heart Lung 1988; 17: [10] Recker D. Patient perception of pre-operative cardiac surgical teaching done pre and post-admission. Crit Care Nurse 1994; 14: [11] Miller P, Shada EA. Pre-operative information and recovery of open heart surgery patients. Heart Lung 1978; 7: [12] Christopherson B, Pfeiffer C. Varying the timing of information to alter pre-operative anxiety and post-operative recovery in cardiac surgery patients. Heart Lung 1980; 9: [13] Marshall J, Penckofer S, Llewellyn J. Structured postoperative teaching and knowledge and compliance of patients who have had coronary artery bypass surgery. Heart Lung 1986; 15: [14] Lepczyk N, Raleigh EH, Rowley C. Timing of pre-operative patient teaching. J Adv Nurs 1990; 15: [15] Cupples SA. Effects of timing and reinforcement of preoperative education on knowledge and recovery of patients having coronary artery graft surgery. Heart Lung 1991; 20: [16] Rice VH, Mullin MH, Jarosz P. Pre-admission self-instruction effect on post-admission and post-operative indicators in CABG patients: partial replication and extension. Res Nurs Health 1992; 15: [17] Shuldham CM. Pre-operative education for the patient having coronary artery bypass surgery. Patient Educ Couns 2001; 43: [18] Arthur HM, Daniels C, McKelvie R, Hirsh J, Rush B. Effect of a pre-operative intervention on pre-operative and postoperative outcomes in low-risk patients awaiting elective coronary artery bypass graft surgery A randomized controlled trial. Ann Intern Med 2000; 133: [19] Shuldham CM, Cunningham G, Hiscock M, Luscombe P. Assessment of anxiety in hospital patients. J Adv Nurs 1995; 22: [20] Ware JE, Sherbourne CD. MOS 36 item Short Form Health Survey (SF36) 1. Med Care 1992; 30: [21] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiat Scand 1983; 67: [22] Cox T, Gotts G. The General Well-Being Questionnaire (GWBQ Manual). Nottingham, UK, University of Nottingham [23] Nelson S. Pre-admission clinics for thoracic patients. Nursing Times 1995; 91: [24] Nelson S. Pre-admission education for patients undergoing cardiac surgery. Br J Nurs 1996; 5: [25] Johnson K. Pre-Surgery Preparation Day: 1997 Report. London: Royal Brompton Hospital, [26] Ware JE, Snow KK, Kosinski MD, Gandek MS. SF-36 health survey: manual and interpretation guide. Boston: Medical Outcomes Trust, [27] Parsonnet V, Dean D, Bernstein AD. A method of uniform stratification of risk for evaluating the results of surgery in acquired adult heart disease. Circulation 1989; 79: [28] Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE 11: A severity of disease classification system. Crit Care Med 1985; 13: