International Journal of Orthopaedic and Trauma Nursing (2015) 19, 92–101
International Journal of Orthopaedic and Trauma Nursing www.elsevier.com/locate/ijotn
Bowel management post major joint arthroplasty: results from a randomised controlled trial Gail M. Ross-Adjie RN, PhD (Nurse Researcher) a,b,*, Leanne Monterosso RN, BNurs (1 Hons), PhD (Chair of Nursing) a,b,c, Max Bulsara BSc (Hons), MSc, PhD (with Distinction) (Chair in Biostatistics) b,d a
Centre for Nursing and Midwifery Research, St John of God Murdoch Hospital, 100 Murdoch Drive, Murdoch 6150, Australia b The University of Notre Dame Australia, Fremantle, Australia c Edith Cowan University, Perth, Australia d Institute for Health and Rehabilitation Research, Fremantle, Australia
KEYWORDS Arthroplasty; Hip replacement; Knee replacement; Analgesia; Narcotic; Opioid; Constipation; Polyethylene glycol
Abstract Aim: To evaluate the effect of a new post-operative bowel protocol in total hip and total knee replacement patients. Background: Up to 65% of total hip and total knee replacement patients experience some degree of constipation post-operatively. A lack of robust evidence to guide bowel management and reduce constipation in this cohort was the impetus for this study. Design: A multisite cluster randomised trial in private secondary and tertiary hospitals. Methods: In total 331 patients were recruited across seven Australian hospitals over 13 months. Control participants (n = 171) received routine bowel management whilst intervention participants (n = 160) received bowel management as per the trial protocol. Results: Intervention patients took 6 days less than controls to return to normal bowel function, and were more than seven times more likely to return to normal bowel function by day 5 post operatively. Age, gender and length of pre-operative fasting had no effect on these outcomes. Conclusion: These results support the use of the Murdoch Bowel Protocol® for hip and knee replacement patients and may be relevant for other patient groups who experience opioid induced bowel dysfunction. © 2014 Elsevier Ltd. All rights reserved.
* Corresponding author. E-mail address: [email protected] (G.M. Ross-Adjie). http://dx.doi.org/10.1016/j.ijotn.2014.04.002 1878-1241/© 2014 Elsevier Ltd. All rights reserved.
Bowel management post major joint arthroplasty
93
Editor comments This is a well-designed cluster RCT comparing approaches to maintenance of normal bowel function and avoidance of constipation following total hip and knee replacement surgery. There are a number of factors that contribute to patients having difficulty to return bowel function including: reduced mobility, opiate and codeine based analgesia, change of environment and diet. The study will help nurses to understand the importance to patients of having normal bowel function and avoidance of constipation and to consider guidelines and practice within their own clinical areas. RJ
Introduction Total hip and total knee replacements remain one of the most commonly performed major orthopaedic procedures with the number having increased rapidly over the past decade in most Organisation for Economic Cooperation and Development (OECD, 2011) countries. On average, the rate of hip replacement increased by over 25% between 2000 and 2009 with the growth rate even higher for knee replacements, nearly doubling over the past decade (OECD, 2011). These patients are at very high risk for developing post-operative constipation for multiple reasons including: a change in diet, reduced fluid intake, pre-operative fasting, the advanced age of many, reduced mobility, the administration of an anaesthetic and the administration of opioid based analgesia both intravenously and orally (Ho et al., 2008; Schmelzer, 1990; Stumm et al., 2001). Despite the scope of this problem little evidence exists to guide bowel management in this cohort.
Background and literature review In 2008 a baseline clinical audit of 30 orthopaedic patients was conducted at the researcher’s 363bed Australian tertiary hospital (one of 17 hospitals within Australia’s third largest private hospital group). The audit was conducted after several major joint replacement patients required extended inpatient stays for management of severe constipation, the return of increasing numbers of post-operative patients for management of faecal impaction and the report of high rates of dissatisfaction with bowel management post discharge. The audit was based on the Practical Application of Clinical Evidence System (PACES) from the Joanna Briggs Institute (JBI), the world’s largest provider of evidence based guidelines for nurses and allied health professionals and based at the University of Adelaide in South Australia. Baseline results using the audit tool ‘Constipa-
tion associated with analgesia’ revealed deficits across all four audit criteria. Discussions with senior nurses from the other surgical divisions within the hospital group confirmed that constipation in the post-operative orthopaedic cohort was a widespread problem. Whilst there is a significant body of evidence reporting the scope of constipation in orthopaedic patients, no robust evidence exists to guide bowel management in this cohort and surprisingly few articles have been published. Of those that have been, most are case studies or discussion papers with most research generally of poor quality, with small sample sizes or demonstrating questionable academic rigor. Whilst often considered a mild selflimiting problem, constipation may lead to significant morbidity and occasionally mortality (Davies et al., 2008; Groth, 1988; Ho et al., 2008; Kaçmaz and Kas¸ikçi, 2007; Linari et al., 2011; Madsen et al., 2010; Stumm et al., 2001). One of the difficulties when comparing the incidence of constipation reported in the literature is the range of definitions used. While some studies used Rome I, II or III criteria (Drossman, 2006; Peppas et al., 2008) others relied on self-reporting which involves a significant degree of subjectivity. The Rome criteria were developed in Rome after a 1988 meeting where gastroenterologists sought to provide consensus guidelines for the diagnosis of functional gastrointestinal disorders. Rome I guidelines were published in 1989 with the updated Rome II guidelines published in 1999 and the current Rome III guidelines published in 2006 (Thompson, 2006). Other studies relied on more general measures of constipation such as laxative use, frequency of bowel actions per week or whether the patient had experienced a degree of incomplete evacuation making comparisons difficult. Despite the lack of a consistent definition of constipation, many studies cite an increased incidence in women (Belsey et al., 2010; Ho et al., 2008; McCrea et al., 2009; Selby and Corte, 2010) with this incidence increasing with age. In 2005 Ramkumar and Rao published a systematic review of the efficacy and
94
G.M. Ross-Adjie et al.
safety of traditional medical therapies for chronic constipation. The authors discussed the paucity of evidence to support the use of many commonly used laxatives such as senna and bisacodyl with macrogol 3350 (also known as polyethylene glycol (PEG) examples include Movicol® and LaxaCon®) the only laxative to be supported by level 1 evidence and a grade A recommendation. Macrogol 3350 (macrogol) was consistently found to be more effective than lactulose and despite being more expensive was more cost effective due to its efficacy (Ramkumar and Rao, 2005). Belsey et al. (2010) undertook a systematic review to assess the efficacy of macrogol with other classes of laxatives. The authors reviewed 20 studies and found that macrogol use resulted in a highly significant increase in defaecations per week when compared with lactulose (p = 0.021) and ispaghula (p ≤ 0.001). These findings are interesting in light of data published from the Laxative Usage in Constipation in the United Kingdom (LUCK) study (Schafe et al., 2011). This study investigated the prescribing trends for laxatives in the UK during 2005–2009 and found that although senna and lactulose were the most commonly prescribed laxatives in the UK in 2005, by 2009 macrogol was the most commonly prescribed agent and senna use had declined significantly. Despite an acknowledgement that ortho-
paedic patients are at very high risk for developing post-operative constipation only one article which reviewed the effectiveness of macrogol in this group was found. The small, single site study (N = 31) conducted in Adelaide, Australia (Madsen et al., 2010) sought to compare the effectiveness of macrogol with a standard bowel treatment. The control group (n = 16) received the standard bowel management protocol which consisted of Coloxyl and Senna®, sorbitol and a Microlax® enema. The intervention group (n = 15) received one to two sachets per day of macrogol from day 1 post operatively. While only 28 patients completed the study, a statistically significant difference in time to first bowel motion (p = 0.001) was found in those patients who took macrogol. Given the lack of clinical guidelines a multidisciplinary team was convened at the researcher’s hospital to develop a new bowel management tool that was practical, based on best available evidence and easy to implement. The Murdoch Bowel Protocol© (see Fig. 1) comprised two key elements. The first was the Bristol Stool Chart (BSC) a simple, visual tool for consistent stool recording which is widely used and has been validated through repeated use over many studies (Heaton and Lewis, 1997). The BSC classifies stool as one of seven types with stool types
Murdoch Bowel Protocol®
Fig. 1
Murdoch Bowel Protocol®.
Bowel management post major joint arthroplasty 1 and 2 indicating a hard, constipated stool; types 3 and 4 considered normal stools; types 5 and 6 indicating loose stool and type 7 a completely liquid stool. Second, the titrated administration of macrogol, an inert, iso-osmotic laxative available over-the-counter in many countries. The systematic review by Ramkumar and Rao (2005), guidelines from an internationally recognised authority (World Gastroenterology Organisation, 2007) and Australian evidence based guidelines (Joanna Briggs Institute, 2008) confirmed the use of polyethylene glycol with electrolytes as the agent of choice for functional constipation. Advice from senior orthopaedic nurses was to commence the protocol on day 2 post-operatively due to the high proportion of patients who report nausea on day 1 post operatively. The dose of macrogol was increased, decreased or ceased depending on the post-operative day and stool type as self-reported by the patient using the BSC with the aim to return the patient to BSC type 3 or 4 (a reflection of normal gut transit time). One change requested by nursing staff was the administration of a Microlax® enema if the patient recorded BSC type 1 or 2 by postoperative days 4 or 5. Microlax® enemas contain a 5 ml volume of sodium citrate, sodium lauryl sulfoacetate and sorbitol and are classified as a stool softener. Whilst the clinical wisdom of expert nurses is acknowledged there was no robust evidence to either support or deny this addition to the protocol, largely a reflection of the poor quality of literature published on the subject. Following considered discussion, the opinion of the working party was to include administration of a Microlax® enema in the protocol with the proviso that it could be withheld by the nurse or refused by the patient if necessary. Patients were also encouraged to select high fibre menu options, ensure adequate fluid intake and increase mobilisation as able while nurses were encouraged to reduce opioids and administer nonconstipating analgesia where appropriate. Should BSC type 1 or 2 (indicative of constipation) be recorded on day six post-operatively, the Protocol recommends referral to a clinical dietician or continence nurse specialist for further assessment and advice. Similarly if the patient reports BSC type 5, 6 or 7 (indicating loose stool or diarrhoea) on day 8 onwards an assessment should be undertaken to exclude constipation with overflow prior to discharge. Whilst it is acknowledged that a dietician or continence specialist nurse may not be available in all clinical settings, the advice of a clinician trained in bowel assessment is recommended in these situations. This cluster randomised controlled trial (RCT) was guided by Neuman’s Systems Model (NSM), devel-
95 oped by Dr Betty Neuman and first published in 1972 (Neuman and Young, 1972). The NSM emphasises a holistic approach to patient care where the nurse and patient work in partnership to achieve optimal health retention, restoration and maintenance (Fitzpatrick and Whall, 2005). Within the NSM, the patient is considered to be affected by intrapersonal stressors (e.g. reduced gut transit time, response to anaesthetic and opioids); extrapersonal stressors (e.g. hospitalisation and reduced mobility) and interpersonal stressors (e.g. privacy concerns, embarrassment about bowel management; and variations in bowel management between different nursing and medical staff) with primary, secondary and tertiary prevention strategies a key components.
The study Aim The primary aim of the trial was to determine the effect of the Murdoch Bowel Protocol© which is based on the titrated administration of polyethylene glycol with electrolytes, on return to normal bowel function in patients who underwent total hip and total knee replacement. In addition the study also sought to determine whether the variables age; gender; length of pre-operative fasting; time to first mobilisation; length of stay; opioid analgesia; anaesthetic type and operation type had any influence on the time taken for these patients to return to normal bowel function.
Design A cluster RCT was conducted following the Consolidated Standards of Reporting Trials (CONSORT) guidelines for the four trial stages (enrolment, intervention allocation, follow-up, and analyses).
Setting, sample and participants A sample size calculation was undertaken to ensure the trial was adequately powered. Data from a pilot study (N = 12) indicated that 97 intervention and 97 control patients would be required to reject the null hypothesis that the experimental and control survival curves were equal with a probability (power) of 0.80. The type 1 error probability associated with this test of the null hypothesis was set at 0.05. For a cluster RCT a variance inflation factor of 0.5 was used. Based on the small pilot study conducted at the researcher’s hospital, the intracluster correlation coefficient (ICC) was estimated to be 0.01 and
96 a cluster size of at least 50 patients, hence 147 control and 147 intervention patients were required. As a contingency to account for an estimated 10% drop-out rate, a minimum of 160 patients were needed to be recruited to each arm of the trial. Proportional sampling provided the minimum proportion of patients to be recruited from each participating hospital based on the average number of major joint replacement operations conducted monthly. Seven surgical hospitals from within the researcher’s health care group volunteered to participate in the trial as their patients had experienced postoperative bowel dysfunction. These hospitals comprised a 70 bed facility in metropolitan Melbourne, Victoria; a 74 bed hospital in regional Victoria; a 117 bed hospital in regional Victoria; a 126 bed hospital in regional Western Australia; a 184 bed hospital in regional Victoria; a 194 bed hospital in regional Victoria and a 548 bed metropolitan hospital in Perth, Western Australia. Due to the risk of confusion and contamination if patients were randomised into control and intervention groups within each hospital, a cluster RCT design was chosen where hospitals rather than patients were randomised (Campbell et al., 2000). Simple randomisation allocated two hospitals to the intervention group and five hospitals to the control group although sample numbers in each group were similar. Where a study involves multiple sites, cluster RCTs are a legitimate form of RCT. This design offered unique advantages for this trial including control for contamination and enhancing compliance which resulted in robust outcomes (Mazor et al., 2007). The total sample size was 331 patients with 160 patients recruited from two intervention hospitals (49.4% THR [n = 79] and 50.6% TKR [n = 81]) and 171 patients recruited from five control hospitals (47.4% THR [n = 81] and 52.6% [n = 90]). Recruitment took place over a 13 month period from May 2011 to May 2012. Patients were eligible for recruitment if they were: aged over 18 years; able to read and understand English; admitted for elective hip or knee replacement surgery; experienced normal bowel function prior to admission and were able to give written informed consent. Patients were excluded if they: were unable to read and understand English; were confused and/or disorientated; had a history of ulcerative colitis, Crohn’s disease, intestinal obstruction or perforation, toxic megacolon or were pregnant or breastfeeding.
Data collection Prior to trial commencement a trial liaison nurse was nominated to collect data, provide guidance to
G.M. Ross-Adjie et al. hospital staff and act as a liaison between each hospital and the researcher. The trial liaison nurses, preadmission clinic nurses and ward nursing staff were provided with comprehensive education from the researcher which encompassed: a background to the trial including aims, patient recruitment procedures including inclusion and exclusion criteria as well as the process for obtaining informed consent from eligible patients, and how to ensure accurate data collection, patient privacy and the secure storage of data. To ensure that consistent advice was given to all ward nurses who may not have been able to attend the education sessions provided, a Caregiver Information Sheet was provided to all nursing staff involved in the trial across all participating hospitals. In addition, the trial liaison nurse acted as the primary resource at each participating site. Data collection took place at three time periods: pre-admission clinic; during the inpatient stay and post discharge. Pre-admission clinic Patients were recruited at the pre-admission clinic (PAC) of each participating hospital after being identified from the operation list for that week as possibly suitable for trial inclusion. Once identified, patients were informed of the trial purpose and procedures, following which written consent was obtained and baseline demographic data and a baseline bowel assessment were recorded (baseline Bristol Stool Chart number, usual stool frequency per week and current use of laxatives). Inpatient stay Patients were asked to refer to a copy of the BSC and self-report a stool number(s) for the preceding 24 hours with this BSC number recorded by a nurse at 1000 hours each day. Patients at control hospitals then received bowel care as per that hospital or doctor’s usual regime while intervention patients received bowel care as per the Murdoch Bowel Protocol©. ‘Usual regime’ differed markedly between clinicians and hospitals but generally comprised one of a range of commonly used aperients e.g. Coloxyl®, Coloxyl with Senna®, Lactulose® or Dulcolax®. Nurses were required to sign for any aperients administered on both the inpatient medication chart and the Stool Recording and Medication Administration Chart. Patients had the option to refuse any aperients and nurses could also withhold aperients if deemed necessary (e.g. vomiting or nil by mouth) with this information recorded in the appropriate section of the medication chart. As the literature suggested that age and gender (McCrea et al., 2009); length of pre-operative fasting (Annells and Koch, 2003;
Bowel management post major joint arthroplasty Miaskowski, 2009); reduced mobility (Miaskowski, 2009; Nazarko, 1996); administration of opioid analgesia (Healy, 2009; Miaskowski, 2009); anaesthetic type (Hu et al., 2009) and operation type (Pang et al., 2000) may influence the time taken to return to normal bowel function, data on these variables were also collected to enable statistical analysis. Post discharge All patients were followed up by telephone approximately 1 week after discharge by the trial liaison nurse at each hospital. This call was made to elicit and record information about analgesia and laxative usage and constipation since discharge. For those patients who had not returned to normal bowel function at the first follow up phone call, a second phone call was made 2 weeks after discharge seeking the same information. Follow up did not extend beyond this time.
Data analysis Data were analysed using the Statistical Package for the Social Sciences (SPSS®) version 19. Analyses were two-tailed with the significance level set at 0.05. An independent samples t test was used to test for differences between normally distributed continuous variables. The chi-square was used to test for differences between categorical variables. Where any cells had an expected count of less than five, the Fisher’s Exact Probability Test was used. The generalised linear model was used to assess the effect of age, gender, group (control or intervention), length of pre-operative fasting, anaesthetic type, operation type and length of stay on the continuous variable days to normal bowel function. As this was a cluster RCT, the generalised linear mixed model was used to test for cluster effects between control and intervention hospitals. As data were collected by multiple nurses at seven different hospitals, the issue of interrater
Table 1
97 variability was considered. Described as the degree to which two or more independent observers agree on what they are coding or scoring (Polit and Beck, 2012) interrater variability is an important part of ensuring the rigor of any trial. Traditional test– retest reliability was not considered necessary for the recording of patient self-reported information or transcribing of data from fluid balance or observation charts by nurses in this trial as the measurement and recording of such patient data is a core part of basic nursing practice.
Results Trial participants were recruited from the orthopaedic wards of seven private Australian regional and metropolitan hospitals. Patient ages ranged between 43 and 88 years with a median age of 67 (SD 9.2). Males comprised 46.5% of the sample and women 53.5%. Missing data were minimal and estimated not to exceed 1–1.5% of the total data volume. This small number did not warrant data imputation, required when larger amounts of data are missing. While day first mobilised; length of stay and days to normal bowel function were found to be highly statistically significant (p ≤ 0.001) the difference between groups for day first mobilised was not considered clinically significant, with differences in length of stay reflecting different hospital norms and surgeon preferences (see Table 1). Of the variables thought to influence days to normal bowel function, the grouping variable (control or intervention) was the most significant although anaesthetic type, operation type and length of stay were also statistically significant (see Table 2). Patients from both control and intervention groups had the right to refuse any aperient with nurses asked to record the reason on the Stool Recording and Medication Administration Chart. In total 15.6% (n = 25) of intervention patients refused aperients compared with 5.3% (n = 9) of control patients. The
Overall group effect for post-operative variables.
Post-operative variables
Control n = 171 M(SD) [Mdn]
Intervention n = 160 M(SD) [Mdn]
p
Day first mobilised
1.175 (0.7) [1.0] 4.96 (2.1) [5.0] 10.64 (7.0) [9.0]
1.437 (0.7) [1.0] 7.07 (2.0) [7.0] 5.06 (2.0) [5.0]
International Journal of Orthopaedic and Trauma Nursing www.elsevier.com/locate/ijotn
Bowel management post major joint arthroplasty: results from a randomised controlled trial Gail M. Ross-Adjie RN, PhD (Nurse Researcher) a,b,*, Leanne Monterosso RN, BNurs (1 Hons), PhD (Chair of Nursing) a,b,c, Max Bulsara BSc (Hons), MSc, PhD (with Distinction) (Chair in Biostatistics) b,d a
Centre for Nursing and Midwifery Research, St John of God Murdoch Hospital, 100 Murdoch Drive, Murdoch 6150, Australia b The University of Notre Dame Australia, Fremantle, Australia c Edith Cowan University, Perth, Australia d Institute for Health and Rehabilitation Research, Fremantle, Australia
KEYWORDS Arthroplasty; Hip replacement; Knee replacement; Analgesia; Narcotic; Opioid; Constipation; Polyethylene glycol
Abstract Aim: To evaluate the effect of a new post-operative bowel protocol in total hip and total knee replacement patients. Background: Up to 65% of total hip and total knee replacement patients experience some degree of constipation post-operatively. A lack of robust evidence to guide bowel management and reduce constipation in this cohort was the impetus for this study. Design: A multisite cluster randomised trial in private secondary and tertiary hospitals. Methods: In total 331 patients were recruited across seven Australian hospitals over 13 months. Control participants (n = 171) received routine bowel management whilst intervention participants (n = 160) received bowel management as per the trial protocol. Results: Intervention patients took 6 days less than controls to return to normal bowel function, and were more than seven times more likely to return to normal bowel function by day 5 post operatively. Age, gender and length of pre-operative fasting had no effect on these outcomes. Conclusion: These results support the use of the Murdoch Bowel Protocol® for hip and knee replacement patients and may be relevant for other patient groups who experience opioid induced bowel dysfunction. © 2014 Elsevier Ltd. All rights reserved.
* Corresponding author. E-mail address: [email protected] (G.M. Ross-Adjie). http://dx.doi.org/10.1016/j.ijotn.2014.04.002 1878-1241/© 2014 Elsevier Ltd. All rights reserved.
Bowel management post major joint arthroplasty
93
Editor comments This is a well-designed cluster RCT comparing approaches to maintenance of normal bowel function and avoidance of constipation following total hip and knee replacement surgery. There are a number of factors that contribute to patients having difficulty to return bowel function including: reduced mobility, opiate and codeine based analgesia, change of environment and diet. The study will help nurses to understand the importance to patients of having normal bowel function and avoidance of constipation and to consider guidelines and practice within their own clinical areas. RJ
Introduction Total hip and total knee replacements remain one of the most commonly performed major orthopaedic procedures with the number having increased rapidly over the past decade in most Organisation for Economic Cooperation and Development (OECD, 2011) countries. On average, the rate of hip replacement increased by over 25% between 2000 and 2009 with the growth rate even higher for knee replacements, nearly doubling over the past decade (OECD, 2011). These patients are at very high risk for developing post-operative constipation for multiple reasons including: a change in diet, reduced fluid intake, pre-operative fasting, the advanced age of many, reduced mobility, the administration of an anaesthetic and the administration of opioid based analgesia both intravenously and orally (Ho et al., 2008; Schmelzer, 1990; Stumm et al., 2001). Despite the scope of this problem little evidence exists to guide bowel management in this cohort.
Background and literature review In 2008 a baseline clinical audit of 30 orthopaedic patients was conducted at the researcher’s 363bed Australian tertiary hospital (one of 17 hospitals within Australia’s third largest private hospital group). The audit was conducted after several major joint replacement patients required extended inpatient stays for management of severe constipation, the return of increasing numbers of post-operative patients for management of faecal impaction and the report of high rates of dissatisfaction with bowel management post discharge. The audit was based on the Practical Application of Clinical Evidence System (PACES) from the Joanna Briggs Institute (JBI), the world’s largest provider of evidence based guidelines for nurses and allied health professionals and based at the University of Adelaide in South Australia. Baseline results using the audit tool ‘Constipa-
tion associated with analgesia’ revealed deficits across all four audit criteria. Discussions with senior nurses from the other surgical divisions within the hospital group confirmed that constipation in the post-operative orthopaedic cohort was a widespread problem. Whilst there is a significant body of evidence reporting the scope of constipation in orthopaedic patients, no robust evidence exists to guide bowel management in this cohort and surprisingly few articles have been published. Of those that have been, most are case studies or discussion papers with most research generally of poor quality, with small sample sizes or demonstrating questionable academic rigor. Whilst often considered a mild selflimiting problem, constipation may lead to significant morbidity and occasionally mortality (Davies et al., 2008; Groth, 1988; Ho et al., 2008; Kaçmaz and Kas¸ikçi, 2007; Linari et al., 2011; Madsen et al., 2010; Stumm et al., 2001). One of the difficulties when comparing the incidence of constipation reported in the literature is the range of definitions used. While some studies used Rome I, II or III criteria (Drossman, 2006; Peppas et al., 2008) others relied on self-reporting which involves a significant degree of subjectivity. The Rome criteria were developed in Rome after a 1988 meeting where gastroenterologists sought to provide consensus guidelines for the diagnosis of functional gastrointestinal disorders. Rome I guidelines were published in 1989 with the updated Rome II guidelines published in 1999 and the current Rome III guidelines published in 2006 (Thompson, 2006). Other studies relied on more general measures of constipation such as laxative use, frequency of bowel actions per week or whether the patient had experienced a degree of incomplete evacuation making comparisons difficult. Despite the lack of a consistent definition of constipation, many studies cite an increased incidence in women (Belsey et al., 2010; Ho et al., 2008; McCrea et al., 2009; Selby and Corte, 2010) with this incidence increasing with age. In 2005 Ramkumar and Rao published a systematic review of the efficacy and
94
G.M. Ross-Adjie et al.
safety of traditional medical therapies for chronic constipation. The authors discussed the paucity of evidence to support the use of many commonly used laxatives such as senna and bisacodyl with macrogol 3350 (also known as polyethylene glycol (PEG) examples include Movicol® and LaxaCon®) the only laxative to be supported by level 1 evidence and a grade A recommendation. Macrogol 3350 (macrogol) was consistently found to be more effective than lactulose and despite being more expensive was more cost effective due to its efficacy (Ramkumar and Rao, 2005). Belsey et al. (2010) undertook a systematic review to assess the efficacy of macrogol with other classes of laxatives. The authors reviewed 20 studies and found that macrogol use resulted in a highly significant increase in defaecations per week when compared with lactulose (p = 0.021) and ispaghula (p ≤ 0.001). These findings are interesting in light of data published from the Laxative Usage in Constipation in the United Kingdom (LUCK) study (Schafe et al., 2011). This study investigated the prescribing trends for laxatives in the UK during 2005–2009 and found that although senna and lactulose were the most commonly prescribed laxatives in the UK in 2005, by 2009 macrogol was the most commonly prescribed agent and senna use had declined significantly. Despite an acknowledgement that ortho-
paedic patients are at very high risk for developing post-operative constipation only one article which reviewed the effectiveness of macrogol in this group was found. The small, single site study (N = 31) conducted in Adelaide, Australia (Madsen et al., 2010) sought to compare the effectiveness of macrogol with a standard bowel treatment. The control group (n = 16) received the standard bowel management protocol which consisted of Coloxyl and Senna®, sorbitol and a Microlax® enema. The intervention group (n = 15) received one to two sachets per day of macrogol from day 1 post operatively. While only 28 patients completed the study, a statistically significant difference in time to first bowel motion (p = 0.001) was found in those patients who took macrogol. Given the lack of clinical guidelines a multidisciplinary team was convened at the researcher’s hospital to develop a new bowel management tool that was practical, based on best available evidence and easy to implement. The Murdoch Bowel Protocol© (see Fig. 1) comprised two key elements. The first was the Bristol Stool Chart (BSC) a simple, visual tool for consistent stool recording which is widely used and has been validated through repeated use over many studies (Heaton and Lewis, 1997). The BSC classifies stool as one of seven types with stool types
Murdoch Bowel Protocol®
Fig. 1
Murdoch Bowel Protocol®.
Bowel management post major joint arthroplasty 1 and 2 indicating a hard, constipated stool; types 3 and 4 considered normal stools; types 5 and 6 indicating loose stool and type 7 a completely liquid stool. Second, the titrated administration of macrogol, an inert, iso-osmotic laxative available over-the-counter in many countries. The systematic review by Ramkumar and Rao (2005), guidelines from an internationally recognised authority (World Gastroenterology Organisation, 2007) and Australian evidence based guidelines (Joanna Briggs Institute, 2008) confirmed the use of polyethylene glycol with electrolytes as the agent of choice for functional constipation. Advice from senior orthopaedic nurses was to commence the protocol on day 2 post-operatively due to the high proportion of patients who report nausea on day 1 post operatively. The dose of macrogol was increased, decreased or ceased depending on the post-operative day and stool type as self-reported by the patient using the BSC with the aim to return the patient to BSC type 3 or 4 (a reflection of normal gut transit time). One change requested by nursing staff was the administration of a Microlax® enema if the patient recorded BSC type 1 or 2 by postoperative days 4 or 5. Microlax® enemas contain a 5 ml volume of sodium citrate, sodium lauryl sulfoacetate and sorbitol and are classified as a stool softener. Whilst the clinical wisdom of expert nurses is acknowledged there was no robust evidence to either support or deny this addition to the protocol, largely a reflection of the poor quality of literature published on the subject. Following considered discussion, the opinion of the working party was to include administration of a Microlax® enema in the protocol with the proviso that it could be withheld by the nurse or refused by the patient if necessary. Patients were also encouraged to select high fibre menu options, ensure adequate fluid intake and increase mobilisation as able while nurses were encouraged to reduce opioids and administer nonconstipating analgesia where appropriate. Should BSC type 1 or 2 (indicative of constipation) be recorded on day six post-operatively, the Protocol recommends referral to a clinical dietician or continence nurse specialist for further assessment and advice. Similarly if the patient reports BSC type 5, 6 or 7 (indicating loose stool or diarrhoea) on day 8 onwards an assessment should be undertaken to exclude constipation with overflow prior to discharge. Whilst it is acknowledged that a dietician or continence specialist nurse may not be available in all clinical settings, the advice of a clinician trained in bowel assessment is recommended in these situations. This cluster randomised controlled trial (RCT) was guided by Neuman’s Systems Model (NSM), devel-
95 oped by Dr Betty Neuman and first published in 1972 (Neuman and Young, 1972). The NSM emphasises a holistic approach to patient care where the nurse and patient work in partnership to achieve optimal health retention, restoration and maintenance (Fitzpatrick and Whall, 2005). Within the NSM, the patient is considered to be affected by intrapersonal stressors (e.g. reduced gut transit time, response to anaesthetic and opioids); extrapersonal stressors (e.g. hospitalisation and reduced mobility) and interpersonal stressors (e.g. privacy concerns, embarrassment about bowel management; and variations in bowel management between different nursing and medical staff) with primary, secondary and tertiary prevention strategies a key components.
The study Aim The primary aim of the trial was to determine the effect of the Murdoch Bowel Protocol© which is based on the titrated administration of polyethylene glycol with electrolytes, on return to normal bowel function in patients who underwent total hip and total knee replacement. In addition the study also sought to determine whether the variables age; gender; length of pre-operative fasting; time to first mobilisation; length of stay; opioid analgesia; anaesthetic type and operation type had any influence on the time taken for these patients to return to normal bowel function.
Design A cluster RCT was conducted following the Consolidated Standards of Reporting Trials (CONSORT) guidelines for the four trial stages (enrolment, intervention allocation, follow-up, and analyses).
Setting, sample and participants A sample size calculation was undertaken to ensure the trial was adequately powered. Data from a pilot study (N = 12) indicated that 97 intervention and 97 control patients would be required to reject the null hypothesis that the experimental and control survival curves were equal with a probability (power) of 0.80. The type 1 error probability associated with this test of the null hypothesis was set at 0.05. For a cluster RCT a variance inflation factor of 0.5 was used. Based on the small pilot study conducted at the researcher’s hospital, the intracluster correlation coefficient (ICC) was estimated to be 0.01 and
96 a cluster size of at least 50 patients, hence 147 control and 147 intervention patients were required. As a contingency to account for an estimated 10% drop-out rate, a minimum of 160 patients were needed to be recruited to each arm of the trial. Proportional sampling provided the minimum proportion of patients to be recruited from each participating hospital based on the average number of major joint replacement operations conducted monthly. Seven surgical hospitals from within the researcher’s health care group volunteered to participate in the trial as their patients had experienced postoperative bowel dysfunction. These hospitals comprised a 70 bed facility in metropolitan Melbourne, Victoria; a 74 bed hospital in regional Victoria; a 117 bed hospital in regional Victoria; a 126 bed hospital in regional Western Australia; a 184 bed hospital in regional Victoria; a 194 bed hospital in regional Victoria and a 548 bed metropolitan hospital in Perth, Western Australia. Due to the risk of confusion and contamination if patients were randomised into control and intervention groups within each hospital, a cluster RCT design was chosen where hospitals rather than patients were randomised (Campbell et al., 2000). Simple randomisation allocated two hospitals to the intervention group and five hospitals to the control group although sample numbers in each group were similar. Where a study involves multiple sites, cluster RCTs are a legitimate form of RCT. This design offered unique advantages for this trial including control for contamination and enhancing compliance which resulted in robust outcomes (Mazor et al., 2007). The total sample size was 331 patients with 160 patients recruited from two intervention hospitals (49.4% THR [n = 79] and 50.6% TKR [n = 81]) and 171 patients recruited from five control hospitals (47.4% THR [n = 81] and 52.6% [n = 90]). Recruitment took place over a 13 month period from May 2011 to May 2012. Patients were eligible for recruitment if they were: aged over 18 years; able to read and understand English; admitted for elective hip or knee replacement surgery; experienced normal bowel function prior to admission and were able to give written informed consent. Patients were excluded if they: were unable to read and understand English; were confused and/or disorientated; had a history of ulcerative colitis, Crohn’s disease, intestinal obstruction or perforation, toxic megacolon or were pregnant or breastfeeding.
Data collection Prior to trial commencement a trial liaison nurse was nominated to collect data, provide guidance to
G.M. Ross-Adjie et al. hospital staff and act as a liaison between each hospital and the researcher. The trial liaison nurses, preadmission clinic nurses and ward nursing staff were provided with comprehensive education from the researcher which encompassed: a background to the trial including aims, patient recruitment procedures including inclusion and exclusion criteria as well as the process for obtaining informed consent from eligible patients, and how to ensure accurate data collection, patient privacy and the secure storage of data. To ensure that consistent advice was given to all ward nurses who may not have been able to attend the education sessions provided, a Caregiver Information Sheet was provided to all nursing staff involved in the trial across all participating hospitals. In addition, the trial liaison nurse acted as the primary resource at each participating site. Data collection took place at three time periods: pre-admission clinic; during the inpatient stay and post discharge. Pre-admission clinic Patients were recruited at the pre-admission clinic (PAC) of each participating hospital after being identified from the operation list for that week as possibly suitable for trial inclusion. Once identified, patients were informed of the trial purpose and procedures, following which written consent was obtained and baseline demographic data and a baseline bowel assessment were recorded (baseline Bristol Stool Chart number, usual stool frequency per week and current use of laxatives). Inpatient stay Patients were asked to refer to a copy of the BSC and self-report a stool number(s) for the preceding 24 hours with this BSC number recorded by a nurse at 1000 hours each day. Patients at control hospitals then received bowel care as per that hospital or doctor’s usual regime while intervention patients received bowel care as per the Murdoch Bowel Protocol©. ‘Usual regime’ differed markedly between clinicians and hospitals but generally comprised one of a range of commonly used aperients e.g. Coloxyl®, Coloxyl with Senna®, Lactulose® or Dulcolax®. Nurses were required to sign for any aperients administered on both the inpatient medication chart and the Stool Recording and Medication Administration Chart. Patients had the option to refuse any aperients and nurses could also withhold aperients if deemed necessary (e.g. vomiting or nil by mouth) with this information recorded in the appropriate section of the medication chart. As the literature suggested that age and gender (McCrea et al., 2009); length of pre-operative fasting (Annells and Koch, 2003;
Bowel management post major joint arthroplasty Miaskowski, 2009); reduced mobility (Miaskowski, 2009; Nazarko, 1996); administration of opioid analgesia (Healy, 2009; Miaskowski, 2009); anaesthetic type (Hu et al., 2009) and operation type (Pang et al., 2000) may influence the time taken to return to normal bowel function, data on these variables were also collected to enable statistical analysis. Post discharge All patients were followed up by telephone approximately 1 week after discharge by the trial liaison nurse at each hospital. This call was made to elicit and record information about analgesia and laxative usage and constipation since discharge. For those patients who had not returned to normal bowel function at the first follow up phone call, a second phone call was made 2 weeks after discharge seeking the same information. Follow up did not extend beyond this time.
Data analysis Data were analysed using the Statistical Package for the Social Sciences (SPSS®) version 19. Analyses were two-tailed with the significance level set at 0.05. An independent samples t test was used to test for differences between normally distributed continuous variables. The chi-square was used to test for differences between categorical variables. Where any cells had an expected count of less than five, the Fisher’s Exact Probability Test was used. The generalised linear model was used to assess the effect of age, gender, group (control or intervention), length of pre-operative fasting, anaesthetic type, operation type and length of stay on the continuous variable days to normal bowel function. As this was a cluster RCT, the generalised linear mixed model was used to test for cluster effects between control and intervention hospitals. As data were collected by multiple nurses at seven different hospitals, the issue of interrater
Table 1
97 variability was considered. Described as the degree to which two or more independent observers agree on what they are coding or scoring (Polit and Beck, 2012) interrater variability is an important part of ensuring the rigor of any trial. Traditional test– retest reliability was not considered necessary for the recording of patient self-reported information or transcribing of data from fluid balance or observation charts by nurses in this trial as the measurement and recording of such patient data is a core part of basic nursing practice.
Results Trial participants were recruited from the orthopaedic wards of seven private Australian regional and metropolitan hospitals. Patient ages ranged between 43 and 88 years with a median age of 67 (SD 9.2). Males comprised 46.5% of the sample and women 53.5%. Missing data were minimal and estimated not to exceed 1–1.5% of the total data volume. This small number did not warrant data imputation, required when larger amounts of data are missing. While day first mobilised; length of stay and days to normal bowel function were found to be highly statistically significant (p ≤ 0.001) the difference between groups for day first mobilised was not considered clinically significant, with differences in length of stay reflecting different hospital norms and surgeon preferences (see Table 1). Of the variables thought to influence days to normal bowel function, the grouping variable (control or intervention) was the most significant although anaesthetic type, operation type and length of stay were also statistically significant (see Table 2). Patients from both control and intervention groups had the right to refuse any aperient with nurses asked to record the reason on the Stool Recording and Medication Administration Chart. In total 15.6% (n = 25) of intervention patients refused aperients compared with 5.3% (n = 9) of control patients. The
Overall group effect for post-operative variables.
Post-operative variables
Control n = 171 M(SD) [Mdn]
Intervention n = 160 M(SD) [Mdn]
p
Day first mobilised
1.175 (0.7) [1.0] 4.96 (2.1) [5.0] 10.64 (7.0) [9.0]
1.437 (0.7) [1.0] 7.07 (2.0) [7.0] 5.06 (2.0) [5.0]
