A Prospective, Randomized Controlled Trial Comparing 3 Dressing Types Following Sternotomy
Dressings are an important aspect of post-sternotomy care. A prospective, randomized, controlled trial was conducted between July 2010 and August 2012 to compare wound and patient outcomes among 3 marketed postoperative dressings.
Using convenience sampling methods, 315 patients requiring a sternotomy incision followed by hospitalization were randomly assigned to postoperative care with a dry sterile (n = 106), metallic silver-containing (n = 105), or ionic silver-containing (n = 104) dressing. Application time and ease of dressing application were assessed. The dry dressings remained in place for 24 to 48 hours.
On postoperative day 5 or at time of discharge, the silver dressings were removed and wound outcomes (degree of wound approximation, skin integrity, wound exudate, presence/absence of necrotic tissue), presence/absence of surgical site infection, patient comfort, and dressing factors (ease of removal, dressing integrity) were evaluated. All participants received the same preoperative skin preparation and intravenous antibiotics. Data were analyzed using Fisher’s exact chi-squares and 1-way ANOVA. No statistically significant differences were found among the dressings in terms of wound healing or infection rate, but statistically significant differences were noted in patient comfort and certain dressing factors. When comparing all 3 dressings, the dry sterile dressing took less time to apply (P = 0.000) and was easiest to apply (P = 0.000). Use of the metallic silver dressing resulted in the least patient-reported pain upon removal (P = 0.015), and incision assessment was easiest with the ionic silver dressing (P = 0.000). When comparing the 2 silver dressings, the metallic silver dressing was easiest to remove (P = 0.003) but had less integrity at the time of removal (P = 0.007). None of the patients in the ionic silver and 1 patient (1%) in the metallic silver dressing group developed a deep infection. Additional research is warranted to determine best practice in postoperative, clean surgical incision management, including sternotomy wounds.
Potential Conflicts of Interest: Dressings for this study were provided at no charge by the respective manufacturers.
A Clinical Trial to Investigate the Effect of Silver Nylon Dressings on Mediastinitis Rates in Postoperative Cardiac Sternotomy
Silver Deposition and Tissue Staining Associated with Wound Dressings Containing Silver
In 2010 in the United States, 397,000 coronary artery bypass surgeries were performed.1 Most cardiac surgery requires a large chest incision with opening of the sternum (sternotomy). Sternal wound infections following cardiac surgery can be catastrophic, resulting in prolonged hospital stays, greater costs, and increased mortality and morbidity.2,3 Postoperative dressings can play an important role in wound healing and infection prevention.4
General guidelines for prevention of surgical site infections following cardiac surgery have been established. According to the Association for Professionals in Infection Control and Epidemiology,5 postoperative dressings should be impermeable to outside contamination sources and provide a protective barrier for the incision site. Dressings should be applied in the operating room directly after skin closure and left intact for 24–48 hours to facilitate wound healing except in the case of excessive drainage. Dressings impregnated with silver were developed to help improve healing and prevent infection; in these products, silver is embedded in the dressing in an ionic or metallic nanocrystalline form and provides an antibacterial effect not associated with antibiotic drug resistance.6
Researchers have examined silver’s antibacterial properties to determine efficacy and effectiveness. In in vitro research, Hooper et al7 studied 2 silver-containing antimicrobial dressings using 9 bacterial strains. Antimicrobial effect was equivalent for 4 strains, with variations in activity and duration against the other strains. The authors concluded that in vitro the silver dressings have comparable, broad-spectrum antimicrobial efficacy; clinical studies are needed to compare the efficacy and effectiveness of silver-containing dressings.
Researchers have evaluated silver dressings in various patient populations. In their nonconcurrent cohort study with real-time clinical data collection, Childress et al8 examined lower extremity revascularization wound complications in 216 patients who underwent 248 interventions (n = 118 control; n = 130 treatment). Conventional, nonsilver-containing dressings were used in the first 15 months of the study, followed by use of metallic silver (MSD, Acticoat® Smith & Nephew, St. Petersburg, FL) and occlusive plastic. The rate of wound complications (defined as failure to heal/wound dehiscence, superficial and deep soft tissue infections) decreased 64% after introduction of the MSD (control 14% [17/115], treatment 5% [7/130]; P = 0.016).
Storm-Versloot et al9 conducted a systematic review of 26 randomized controlled studies (2,066 patients with a variety of wounds, including burns). They noted most of the studies were small and of poor quality and evidence that silver-containing dressings prevent wound infection or promote wound healing was insufficient. Moore and Young’s10 systematic review of studies on the effectiveness of silver dressings identified 5 clinical trials (3 randomized controlled) conducted between 2000 and 2009 (N = 410) that included various wound types but no clean surgical wounds. Of those trials, although silver dressings positively impacted wound healing in all 5 studies, only 2 compared ionic silver dressings (Contreet foam, hydrofiber AG, Coloplast Corp, Minneapolis, MN) to nonsilver dressings for effectiveness and safety. In contrast to the previous systematic review, Moore and Young10 concluded the silver dressings used in these studies promote wound healing and appear to be safe and effective treatments for both acute and chronic wounds. The authors also noted studies in this area are limited and recommended additional research to confirm findings regarding the effectiveness of silver dressings. This review of studies did not conclusively identify a specific silver dressing or method of delivery as superior in clean, infected, or chronic wound care.
Huckfeldt et al11 conducted a comparison study of 1,600 cardiac surgery patients retrospectively (n = 1,235; standard gauze dressing) and prospectively (n = 365; silver nylon dressing). They found no cases of mediastinitis (deep chest infection) in the silver group although 13 in the gauze/control group (1%) experienced this infection. In a more recent study of 199 patients undergoing a variety of clean surgical procedures, Schwartz et al12 compared metallic silver to nonsilver dressings and found no significant difference in infection rates.
Based on the currently available literature, there are no consistent, evidence-based recommendations for postoperative silver dressings with clean surgical incisions.
In the post-sternotomy population with its historically low deep sternal wound infection rate of 0.15% to 3.66%,3 there has been limited urgency to conduct a randomized controlled trial to study the impact of various dressings. However, even though the sternal wound infection rate is low, it is associated with significant increases in morbidity and mortality and can be costly.2 The purpose of this prospective, randomized controlled trial was to compare wound healing, patient comfort, surgical site infection rates, and dressing factors among 3 dressing types in patients with clean sternotomy incisions. The study’s hypothesis was cardiac surgery sternotomy patients who received either of the silver-impregnated dressings would have better outcomes (better wound healing, less discomfort, lower incidence of infection) than persons who received the dry dressing.
The study was conducted between July 2010 and August 2012. Approval to conduct the study was received from the Carilion Clinic Institutional Review Board. The principal investigator (PI) formed a study team of nurses who screened patients for eligibility, informed them about the study, assisted with obtaining informed consent, applied the dressings, and completed assessments according to protocol. Manufacturers’ representatives provided extensive education on purpose, use, placement, and management of the silver dressings. To maximize adherence to the protocol, researchers conducted regular information sessions with the study team.
Participants/setting. A convenience sample of 351 inpatients awaiting cardiac surgery or outpatients seen in the presurgical testing area before admission for surgery were recruited for this study. Patients were eligible if they were: 1) adults at least 21 years of age, 2) undergoing cardiac surgery requiring sternotomy incisions, 3) hospitalized at the study setting, 4) English-speaking, 5) able to understand and give consent, 6) approved by their cardiothoracic surgeon to participate, 7) not sensitive to silver, and 8) not sensitive to alginates.
The study setting was a Magnet-designated, 703-bed Level I trauma center located in the southeastern United States. The center’s cardiac surgery program began in 1982; approximately 500 patients have surgeries performed there annually. All cardiac surgeries are performed in a dedicated operating area and patients are transferred to the Cardiac Surgery Intensive Care Unit (CSICU) immediately following surgery and to the Cardiac Surgery Progressive Care Unit (CSPCU) thereafter.
Study dressings. All study dressings were approved for marketing and received US Food and Drug Administration clearance for human use. The dry sterile dressing (DSD, Primapore®, Smith & Nephew, Andover, MA) is a 1-piece, peel-and-stick, nontransparent dressing. This dressing had been used in the facility for almost 30 years (since the initiation of the cardiac surgery program). This dressing was left in place for either 24 or 48 hours per standard facility protocol; it is only water-resistant and thus does not permit the patient to shower or bathe with the dressing in place. The dressing is indicated for postoperative wounds and other minor lesion sites.13
The metallic silver dressing (MSD, Acticoat Post-Op®, Smith & Nephew, St. Petersburg, FL) is a 1-piece, peel-and-stick, nontransparent, absorbent postoperative dressing. It consists of a nanocrystalline, silver-coated polyurethane layer, a white polyurethane foam layer, and an adhesive-coated waterproof polyurethane film layer. The MSD was left in place over the incision for 5 days. This dressing is indicated for a maximum 7-day wear time for postoperative surgical wounds with light to moderate drainage.14 It can cause transient discoloration of the skin and should not be used in patients with known silver allergies. Because the dressing is waterproof, the patient may shower or bathe with the dressing in place.
The ionic silver dressing (ISD, Dermanet® Ag+, DeRoyal, Powell, TN) is a semi-transparent dressing embedded with silver, alginate, and maltodextrin. The dressing was cut to fit the incision and then covered with a transparent dressing (Transeal®, DeRoyal, Powell, TN). The dressing was left in place for 5 days postoperatively. The ISD is indicated for a maximum wear time of 5 days for infected and noninfected wounds, including ulcers, burns, skin lesions, and skin graft donor sites.15 This ionic silver dressing should not be used on patients with known sensitivity to alginates (a seaweed-based component) or on third-degree burns. This dressing is waterproof, allowing patients to shower or bathe with it in place.
Procedure. Before surgery, all participants received sterile skin preparation using 2% chlorhexidine gluconate and 70% isopropyl alcohol (ChloraPrep with Tint®, CareFusion, El Paso, TX) per the standard protocol at the authors’ institution. All participants received intravenous antibiotics within the appropriate timeframe before surgery according to guidelines published by the Joint Commission in its Surgical Care Improvement Program (SCIP-Inf-1b).16 The surgical team closed chest incisions according to individual surgeon protocols.
Interventions. All dressings were applied to participants in the operating room following incision closure. Before placement of the silver dressings, a liquid barrier product (Skin Prep®, ConvaTec, Skillman, NJ) was applied to the area around the participant’s incision and permitted to dry to enhance adherence. The dry sterile dressing was the standard care for post-sternotomy patients at the study site at the initiation of the study. In the initial 8 months of the study, the facility’s incision care protocol dictated the dressing remain in place for 24 hours postoperatively. The facility changed the dressing policy during the study, and the research protocol was modified to leave the dry sterile dressing in place for 48 hours postoperatively. The silver dressings remained in place for 5 days postoperatively or until the patient was discharged from the hospital, whichever came first, and study nurses assessed and reported outcomes at this time.
Outcomes measured. The outcome measures in this study are described in Table 1. The primary outcome measure was wound healing, defined as the degree of wound approximation, skin integrity, exudate, and presence/absence of necrotic tissue. These parameters were assessed by study team nurses at postoperative day 5 (POD 5) or day of discharge. Wound approximation was assessed as total, partial (<2 cm of superficial separation), moderate (>2 cm of superficial separation), or dehisced (complete separation of layers). Skin integrity was evaluated as normal (pink, no redness), inflamed (heat, redness, swelling), or macerated within a 2.5 cm border of the incision. Exudate quality was assessed as purulent, bloody, serosanguinous, or serous. Presence or absence of necrotic tissue was noted.17
The secondary outcome measures in this study were patient comfort, specific dressing factors, and presence/absence of sternal wound infection. Patient comfort with the dressing in place and with dressing change was measured at POD 5 or day of discharge. Comfort was measured using a verbal rating 0–10 scale, with 0 signifying no pain and 10 signifying maximum pain.18 Dressing factors of ease of application, removal, and incision assessment were measured on a 5-point Likert-type scale (1 = very easy, 2 = moderately easy, 3 = neither easy nor difficult, 4 = moderately difficult, and 5 = very difficult). A standard wall clock in the operating room was used to measure the time to apply dressings. Dressing integrity was assessed as suture line exposed, poorly sealed (imperfect covering of the suture line with the potential for organism entry), or well-sealed.17 Sternal wound infection presence and type (superficial or deep) was evaluated using Centers for Disease Control/National Healthcare Safety Network surveillance definitions.19
Study nurses completed and documented assessments of participants on a standardized, investigator-developed, paper-and-pencil data collection tool. On POD 5 or on the patient’s day of discharge, a study nurse assessed the incision and the intactness of the silver dressing. Study participants who required premature removal of their sternotomy dressings (of any type) for any reason (additional chest surgery or catheter placement, need for closer incision assessment, patient request, patient confusion) were removed from the study. No dressing was removed for infection-related concerns. The patient’s electronic medical record was accessed to determine presence, absence, and type of sternal surgical site infection at POD 30. After thorough review, the PI entered all data into an Excel spreadsheet. All data were maintained under the control of the PI in a secure office. Patient confidentiality was maintained through use of a patient list stored separately from the data collection tools.
Sample size and randomization. A sample size of 339 patients with 113 patients in each group was determined adequate to achieve 80% power at a significance level (alpha) of 0.05 to detect a difference of 10% where 85% of DSD dressing patients achieved satisfactory wound approximation compared to 95% of silver dressing patients. Due to low infection rates at the study site, it was not possible to power the study to include infection as a primary outcome.
The PI used a statistician-generated, random numbers table to assign participants to each of the 3 dressing groups. Following randomization, the PI took the appropriate dressing to the operating room and communicated the dressing assignment directly to the nursing staff. Participants were not told of their group assignment until they awakened after surgery (see Figure 1). Due to the nature of the dressings, no aspect of this study was blinded.
Statistical analysis. SPSS software (Version 14.0 for Windows, SPSS, Chicago, IL) was used for data input and analysis. Fisher’s exact chi-squares and 1-way ANOVA were used to test the hypothesis that a larger proportion of patients with silver dressings would experience satisfactory wound approximation compared to patients with standard dressings. If ANOVA test results were statistically significant for any variable, the researchers conducted further analyses on the 2 silver dressings using Tukey’s test to look at each of the pairs individually. Statistical significance level was set at P <0.05.
The final sample consisted of 315 participants: DSD = 106, MSD = 105, and ISD = 104 (see Figure 1). Thirty-six participants were withdrawn from the study because they did not wear their assigned dressing until the appropriate removal time due to additional surgeries or inadvertent removal. Two participants requested their metallic silver dressing be removed early (postoperative day 3) due to itching.
No significant differences were found among patients in the category of wound healing. On POD 5, for each of the 3 groups, total wound approximation occurred for at least 96% (DSD 78, 97.5%; MSD 97, 96%; ISD 98, 97%; P = 0.847) (see Table 2). Normal skin integrity assessed on POD 5 was achieved in at least 97% of each of the groups (DSD 78, 97.5%; MSD 98, 97%; ISD 99, 98%; P = 0.903). Purulent exudate was observed in no more than 1.3% of patients (DSD 1, 1.3%; MSD 1, 1%; ISD 0, 0%; P = 0.554). No necrotic tissue was present in any group. When the 2 silver dressings were compared, no statistically significant differences were noted between them in any of the wound healing subcategories.
In addition, no statistically significant differences were found among the dressings in the category of comfort with dressing in place. On a 0–10 pain scale, average participant ratings for comfort with the dressings in place were DD 0.98; MSD 0.67; ISD 0.75; P = 0.265. A statistically significant difference was noted between the silver dressings in the category of patient comfort with dressing removal (P = 0.025); use of the MSD resulted in the least patient-reported pain upon removal (DSD 2.37; MSD 1.47; ISD 2.38).
Statistically significant differences were noted in the category of dressing factors. The DSD (87, 100%) and the MSD (96, 98%) were rated highest in ease of application compared with the ISD (51, 51.5%) (P = 0.000). The ISD required greater effort because it was a 2-piece dressing that needed cutting while the others were 1-piece, peel-and-stick bandages. Incision assessment was easier with the ISD because it was the only transparent dressing used in the study (DSD 5, 9%; MSD 3, 3%; ISD 27, 27%; P = 0.000) When the silver dressings were compared for ease of removal, the MSD (71, 70%) was rated “very easy” to remove more often than the ISD (50, 50%; P = 0.003). In terms of dressing integrity, the MSD (23, 23%) was more frequently assessed as “poorly sealed” or “suture line exposed” at time of removal than the ISD (9, 9%; P = 0.007). With regard to application time, the DSD had the shortest average application (1.10 minutes), followed by the MSD (1.76 minutes) and the ISD (2.65 minutes) (P = 0.000).
No statistically significant differences were found among the patients with regard to surgical site infection (superficial, deep). Three patients (3%) with DSD had infections: 1 (1%) had superficial and 2 (2%) had deep infections. Among silver dressing patients, 1 (1%) who received MSD had a surgical site infection (superficial 0, 0%, deep 1,1%); 2 patients (2%) who received ISD had infections (superficial) (see Table 2).
Although not statistically significant, no deep infections were seen in patients who received ISD. In addition, no participants had allergic reactions to the silver dressings.
In this study, investigators compared the outcomes of using 3 different dressings (2 silver and 1 nonsilver) in patients with clean surgical wounds. This study of post-sternotomy cardiac surgery patients did not identify a best dressing based on wound healing or infection prevention, a finding consistent with Schwartz et al.12
No statistically significant differences were found among the dressings in the primary outcome of wound healing (approximation, skin integrity, exudate, and necrotic tissue). No statistically significant difference was noted among the dressings regarding patient-reported comfort with the dressing in place, but patient-reported comfort upon dressing removal was highest for the MSD. Few researchers have reviewed silver dressing factors such as ease of application, assessment, and removal, dressing integrity, and time to apply. In this study, statistically significant differences between the dressings were seen in all of these categories.
Although no statistically significant differences were found among the dressings in the incidence and type of surgical site infection, patients receiving the ISD experienced fewer deep infections, a finding that is clinically significant to providers. However, given the low rate of infection at the study site, it was not feasible to conduct an adequately powered study to determine statistically significant differences in infection rates.
The current study generated important evidence upon which to make a clinical decision regarding post-sternotomy care. Although no statistically significant differences were found among the dressings in the primary outcome of wound healing or in the secondary outcome of wound infection, statistically significant differences among the dressings with regard to patient comfort and certain dressing factors were noteworthy.
Based on findings, the study team recommended a change in practice to the ISD for post-sternotomy incisions. This dressing allows practitioners to assess the incision site: it adheres well, is easy to remove, and does not stain the skin. During the course of the study, manufacturers developed a 1-piece, peel-and-stick version of the dressing that increased ease of application for future use. For patients with silver allergy, the dry dressing will remain standard care.
This study had several limitations. First, no patient demographic data were collected that may have allowed for further differentiation among the categories and help explain observed infection rates. Second, the investigators recognize the potential for bias because the companies provided the dressings free of charge and dressing selection was not blinded to the investigators or study team. In addition, the 0–10 pain scale may have not been sensitive enough to capture subtle differences in comfort levels. Lastly, even though study team members were trained before implementation, individual variation in incision assessments may have occurred. Due to the nature of the dressings and the recommendations for use, the dry dressing was left in place for only 24 to 48 hours, while the silver dressings were left in place for 5 days or until discharge, whichever came first. Another limitation is that the study did not include an intention-to-treat analysis. Although this statistical strategy is recommended20 to maintain the integrity of randomization that might be affected by patient withdrawal from the study, the study team elected to complete analysis on patients who completed the study protocol and focus on their specific outcomes.
A prospective, randomized, controlled trial was conducted to compare wound healing, comfort, and dressing factors in post-sternotomy patients randomized to 1 of 3 different dressings. The incidence of surgical site infection also was assessed. No statistically significant differences among dressings for the assessed wound healing outcomes were observed, and the rate of infection was also similar. Patient comfort and other dressing factors were significantly different — notably, pain upon removal, ease of application, and ease of removal were rated significantly better in the MSD than in the ISD group and application time was shorter. However, ease of assessment and dressing integrity on POD 5 were significantly better in the ISD group. None of the patients in the ISD group developed a deep infection. Although patients reported the MSD was more comfortable on removal, it more often was poorly sealed or left the suture line exposed than the other dressing.
Further research is warranted to determine best practice in dressings for clean surgical incisions. Additional large, prospective, randomized controlled studies are needed to evaluate dressing efficacy and infections rates for the cardiac surgery patient population. Postoperative dressings can play an important role in wound healing and prevention of infection. Although this study did not determine best practice, it adds information about sternotomy incision care outcomes that facilitate the clinical decision making process.
This study was possible because of the efforts of a research team of 12 frontline nurses representing Pre-Surgical Testing, Cardiac Surgery Operation Room, Intensive Care Unit, and Progressive Care Unit. Team members screened patients for participation, consented patients, coordinated placement and removal of dressings, and performed incision assessments: Sheri Altizer, Beth Bond, Sherry Boone, Debra Huff, Lisa Jones, Lisa Kuppler-Lee, Kristina Massey, Debra Pullen, Sheri Smith, Jane Thomas, Sarah White, and Lisa Zettler. Special thanks go to consultants from wound care, infection prevention, and statistical analysis: Doris Jones, Kim Hall, Sandra Sebestin, Katherine Shaver, Robert Herbertson, and Ellen Rachel Lockhart.
1. American Heart Association. Heart disease and stroke statistics – 2014 Update. Circulation. 2014;129(1):328–e292.
2. Hollenbeak CS, Murph DM, Koenig S, Woodward RS, Dunagan WC, Fraser VJ. The clinical and economic impact of deep chest surgical site infections following coronary artery bypass graft surgery. Chest. 2000;118(2):397–402.
3. Toumpoulis, IK, Anagnostopoulos CE, Derose, JJ, Swistel DG. The impact of deep sternal wound infection on long-term survival after coronary artery bypass grafting. Chest. 2005;127(2):464–471.
4. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guideline for prevention of surgical site infection. Infect Control Hosp Epidemiol. 1999;20(4):247–278.
5. Kohut K. Guide for the Prevention of Mediastinitis Surgical Site Infections Following Cardiac Surgery. Washington, DC: Association for Professionals in Infection Control and Epidemiology (APIC);2008.
6. Lansdown AB. Silver I: Its antibacterial properties and mechanism of action. J Wound Care. 2002;11(4):125–130.
7. Hooper SJ, Williams D, Thomas DW, Hill KE, Percival SL. An in vitro comparison of two silver-containing antimicrobial wound dressings. Ostomy Wound Manage. 2012;58(1):16–22.
8. Childress BB, Berceli, S, Nelson P, Lee WA, Ozak CK. Impact of an absorbent silver-eluting dressing system on lower extremity revascularization wound complications. Ann Vascular Surg. 2007;21(5):598–601.
9. Storm-Versloot MN, Vos CG, Ubbink DT, Vermeulen H. Topical silver for preventing wound infection (review). Cochrane Database System Rev. 2010;3:CD006478.
10. Moore C, Young J. Effectiveness of silver in wound care treatment. Phys Ther Rev. 2011;16(3):201–209.
11. Huckfeldt R, Redmond C, Mikkelson D, Finley PJ, Lowe C, Robertson J. A clinical trial to investigate the effect of silver nylon dressings on mediastinitis rates in postoperative cardiac sternotomy. Ostomy Wound Manage. 2008;54(10):36–41.
12. Schwartz J, Gross S, Facchin F, Manizate F, Gendics C, Braitman E, Lantis J. A prospective two-armed trial assessing the efficacy and performance of a silver dressing used postoperatively on high-risk, clean surgical wounds. Ostomy Wound Manage. 2014;60(4):30–40.
13. Primapore® Product Information. Available at: www.smith-nephew.com/professional/products/advanced-wound-management/pri.... Accessed February 9, 2015.
14. Acticoat® Post-Op. 2008. Smith & Nephew quick reference guide, St. Petersburg, FL.
15. Dermanet® Ag+. 2012. DeRoyal Wound Care Catalog, Powell, TN.
16. Specifications manual for national hospital inpatient quality measures. NQF-endorsed voluntary consensus standards for hospital care. Available at: www.jointcommission.org/specifications_manual_for_national_hospital_inpa.... Accessed December 17, 2014.
17. Wynne R, Botti M, Stedman H, Holsworth L, Harinos M, Flavell O, et al. Effect of three wound dressings on infection, healing comfort, and cost in patients with sternotomy wounds: a randomized trial. Chest. 2004;125(1):43–49.
18. McCaffery M, Pasero C. Pain: Clinical Manual, 2nd ed. St. Louis, MO: Mosby;1999.
19. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control. 2008;36(9):309–332.
20. Polit D, Gillespie BM. Intention-to-treat in randomized controlled trials: recommendations for a total trial strategy. Res Nurs Health. 2010;33(4):355–368.