Risk assessment of failure during transitioning from in-centre to home haemodialysis

Background Introducing a de-novo home haemodialysis (HHD) program often raises safety concerns as errors could potentially lead to serious adverse events. Despite the complexity of performing haemodialysis at home without the supervision of healthcare staff, HHD has a good safety record. We aim to pre-emptively identify and reduce the risks to our new HHD program by risk assessment and using failure mode and effects analysis (FMEA) to identify potential defects in the design and planning of HHD. Methods We performed a general risk assessment of failure during transitioning from in-centre to HHD with a failure mode and effects analysis focused on the highest areas of failure. We collaborated with key team members from a well-established HHD program and one HHD patient. Risk assessment was conducted separately and then through video conference meetings for joint deliberation. We listed all key processes, sub-processes, step and then identified failure mode by scoring based on risk priority numbers. Solutions were then designed to eliminate and mitigate risk. Results Transitioning to HHD was found to have the highest risk of failure with 3 main processes and 34 steps. We identified a total of 59 areas with potential failures. The median and mean risk priority number (RPN) scores from failure mode effect analysis were 5 and 38, with the highest RPN related to vascular access at 256. As many failure modes with high RPN scores were related to vascular access, we focussed on FMEA by identifying the risk mitigation strategies and possible solutions in all 9 areas in access-related medical emergencies in a bundled- approach. We discussed, the risk reduction areas of setting up HHD and how to address incidents that occurred and those not preventable. Conclusions We developed a safety framework for a de-novo HHD program by performing FMEA in high-risk areas. The involvement of two teams with different clinical experience for HHD allowed us to successfully pre-emptively identify risks and develop solutions. Supplementary Information The online version contains supplementary material available at 10.1186/s12882-022-03039-4.


Background
The past decade has seen a growing literature reporting superior outcomes for home haemodialysis (HHD) patients compared to patients dialysing in-centre (ICHD) [1,2]. The benefits of HHD extend beyond flexibility in scheduling, reduced travel time, and improved quality of life when receiving dialysis treatment in a familiar home environment [3,4]. Recently published studies of HHD have even reported comparable survival to deceased donor renal transplant patients [5,6]. Many renal programmes are now advocating HHD, in addition to peritoneal dialysis (PD), as the preferred modality choice for chronic maintenance dialysis patients.
Although HHD is not a new modality and has a good safety record in established programmes [7,8], Singapore is only just initiating a pilot programme. Our programme will offer suitable local end stage kidney failure (ESKF) patients who are already receiving haemodialysis (HD) or have chosen HD as their long-term option to be treated at home. The adoption of home dialysis has been slow, in part because Singapore is geographically small and well connected with one of the world's leading public transportation systems. Older patients have also been reluctant to accept that dialysis treatments can be performed safely at home. Fortunately, with wider exposure and better health literacy over recent years, there are now an increasing number of patients keen to participate in HHD as a long-term dialysis modality [9][10][11].
The fundamentals of HD are essentially the same, whether ICHD or HHD. In Singapore 80% of patients with chronic kidney disease treated by dialysis attend ICHD [12]. For our pilot trial of HHD, we excluded patients using tunnelled catheters for vascular access, given the higher risk of adverse incidents [13]. We decided to make a caregiver a requirement for the initial patients starting HHD to improve patient confidence while starting this new modality option to patients used to ICHD. The prerequisite requirement of a mandatory caregiver is more restrictive in allowing access to HHD. Although previously published reports do not demonstrate that the presence of a care giver translates into reducing the risk of adverse events [13] but a caregiver was chosen to allay the anxieties of patients and stakeholders.
For HHD, the major goals are to prepare the staff both as educators and trainers, the patient and the home ready for dialysis. As we are starting and adopting a new modality that will change the landscape of chronic maintenance dialysis locally in Singapore, we strengthened our HHD risk assessment by evaluating the risk of failure in the processes of HHD using the Failure Mode Effects Analysis (FMEA) tool. FMEA has been used in many healthcare settings, including ICHD, to assess various new critical policies and procedures before implementation and also to identify areas for improvement [14][15][16][17]. Our study will be the first to use FMEA to evaluate the risk of failure in HHD. We collaborated with one of the pioneers of HHD, the Royal Free Hospital in London, to further strengthen and validate our results. We aimed to review potential failures in the HHD process and identify the highest risk areas to help us develop prioritised interventions designed to prevent failures or at least minimise risks. We also defined remedial actions to mitigate the impact and consequences of any failures.

Methods
We conducted a FMEA between July 2020 to February 2021 with teams from the National University Hospital Singapore (NUH) and Royal Free Hospital (RFH) in London. The FMEA was performed independently and adjusted collaboratively. We performed our FMEA in the following manner: Step 1: Selection of team members.
Step 2: Process scope identification and listing of all key processes (using process flowchart).
Step 3: Identification of failure mode.
Step 4: Scoring based on risk priority numbers.
Step 5: Designing solutions to eliminate and mitigate risk.
Ethics approval for the study was not required by The National Healthcare Group Institutional Review Board in Singapore as failure mode effect analysis does not meet the definition of human-subject research. All methods were performed in accordance with the relevant institutional guidelines and regulations.

Step 1: team selection
We identified the key members for the FMEA process and recruited cross-functional members with diverse and in-depth knowledge of HD and HHD from both NUH and RFH. Teams from these two hospitals shared common expertise in HD but were distinct in that one is planning for the initiation of a HHD programme, whereas the other has decades of experience in HHD. Together, we provided unique perspectives of a HHD programme at different phases of maturity.

Step 2: process identification and process flowchart preparation
We reviewed the established processes involved in the HHD journey at RFH and conducted a search on the HHD processes in PubMed. Each team identified and listed all key processes and subprocesses in HHD independently, and this was subsequently collated to map the final processes and subprocesses.
Step 3: failure mode identification We reviewed the HHD experience at the RFH and used "brainstorming" sessions and an extensive review of published literature to identify all the possible failure modes in HHD. We listed all potential failures that can occur at each step in the flowchart and found that the area with the highest risk involves the transition phase of patients from in-centre haemodialysis to home.
A list of potential failure modes during the process of transitioning to home was generated. The list was categorised, reviewed for accuracy and completeness by both the NUH and RFH teams. The possible areas of failure were identified, considering contributing factors and potential consequences.
Step 4: scoring based on risk priority numbers In the areas of greatest concern, we rated each process failure for likelihood (remote to very high), severity (none to catastrophic), and detectability (almost certain to absolutely uncertain) using a consensus approach. The risk priority number (RPN) was then calculated for each potential failure mode. The RPN is the quantitative estimate of the risk associated with each failure mode (Table 1) [18]. FMEA teams assigned an RPN to each failure mode based on three factors: (1) the likelihood of occurrence (L), (2) the degree of severity if it does occur (S), and (3) the likelihood of detecting the occurrence (D). The RPN was calculated using the formula: L x S x D, where high numbers indicate a high priority for intervention and action [19,20]. The scores were determined based on consensus, following discussions between NUH and RFH team members. A failure mode with an RPN of 100 or greater was considered a high priority and was further investigated and documented in the FMEA worksheet (Table 2) [21,22].
Step 5: designing solutions to eliminate, mitigate risk and risk review For the highest area of concern, potential solutions to mitigate risks and interventions were evaluated. As risk mitigation in this area usually involved a bundle approach and so was often overlapping, we discussed all possible solutions for each failure in the area concerned. We had extensive discussions for areas with RPN scores more than 100. A risk review process is planned prospectively as the HHD program starts to refine the FMEA.

Results
The NUH team consisted of three nephrologists with one having formal training in patient safety and healthcare quality, a senior renal nurse manager, one newly trained HHD nurse, a representative from hospital clinical governance experienced in conducting regular FMEA, an administrative executive, and a non-renal physician trained in healthcare quality from another independent (non-affiliated) institution. The team from RFH consisted of a nephrologist, two nurse managers of the HHD and home therapies programs, two senior renal technologists, and two patients each with more than 10 years HHD experience. The teams had 5 meetings of approximately 2 h. FMEA was conducted separately by each team and then through video conferencing platform (ZOOM) meetings.

Flowchart of key processes in HHD
We identified all the key areas, processes and sub-processes in setting up HHD. The four major areas in HHD involve setting up the program, training of staff and patients, transferring patients from in-centre to HHD, and ongoing maintenance therapy (Fig. 1).

Failure modes and risk priority numbers
After reviewing various main areas, transitioning to HHD was found to have the highest risk of failure (Fig. 1). The FMEA on the transitioning to HHD process was independently completed by the two teams, but the final score was determined after joint deliberation. The three main processes in transitioning to HHD area are setting up the HHD system at home, completing tests before starting HHD, and performing HHD.
There was a total of 5 main steps with 13 failure modes in setting up the HHD system at home and 2 main steps with 2 failure modes in completion of the tests before starting HHD. We identified that performing Table 1 Risk priority definition and rating scales [18] Risk priority number Definition Rating scales

Likelihood
The perceived chance of the failure happening within a defined period Rating of 1-10: 'failure is unlikely' to 'very likely or inevitable' Severity How severe the outcome is to the patient should failure occur Rating of 1-10: 'no severity at all' (would not affect individual or system) to 'moderate' (significant effect with no injury) to 'major injury' to 'death' Detectability Is the area of failure readily known, or is it discovered only when an adverse outcome occurs?
Rating of 1-10: 'almost certain the process or steps will detect potential cause(s)' to 'absolute uncertainty that the control will not detect potential cause(s) and subsequent failure mode (s) HHD treatment has a total of 7 sub-processes; including (1) preparation to start HHD treatment, (2) general patient and dialysis equipment evaluation before starting dialysis, (3) managing vascular access during dialysis treatment, (4) dealing with treatment interruption and troubleshooting machine alarms, (5) administering medication during dialysis, (6) others, and (7) ending dialysis, with a total of 27 steps and 44 failure modes (Table 2). Among the sub-processes, managing vascular access during treatment has the highest failure modes (Additional file 1: Appendix 1). Vascular access management had a total 9 steps with 9 failure modes. Failure mode effect analysis revealed scores ranging from 9 to 256 with a mean of 135 and median of 144 respectively. The highest RPNs were failure to monitor vascular access, a defective technique in cannulation access, and troubleshooting vascular access alarm with scores of 252, 256, and 180 respectively. As vascular access management and troubleshooting access alarms are overlapping and have higher RPN scores, we explored the failure modes, effect, consequences, and identified risk mitigation strategies in all 9 areas of access-related medical emergencies. In particular we emphasized 3 main areas with high RPN scores, highlighted in grey in Table 3. Table 3 provides strategies for risk mitigation of failure modes in the 9 areas related to vascular access. We identified 46 risk mitigation strategies. We discussed all areas related to vascular access with focus on the three main steps with the highest RPN (grey shading in Table 3). Some of the potential strategies for risk mitigation were overlapping.  3. Butterfly style taping method preferentially or in selected patient Chevron styles [33,34] 4. Blood lines looped loosely to allow movement of patient and prevent blood lines pulling on needles especially in nocturnal dialysis [32] 5. Setting the lower limit of the venous pressure alarm as close as possible to current venous pressure alarm 6. Wetness detector, especially for nocturnal haemodialysis 7. Regular assessment of patient performing treatment observed during home visit and during physician review at HHD unit (at least once every two months)

Risk mitigation strategies
The most significant concern was a defective technique in access cannulation that usually occurs most frequently during the early transitioning phase to HHD. We suggest risk reduction in this area needs to be initiated at the start of HHD training. In addition to identifying trained staff dedicated to HHD training, we propose an individualised vascular access approach to self-cannulation for each patient [26,[38][39][40]. The individualised approach includes a single dedicated trainer, type of techniques, cannulation type, and weekly review during the first three months of HHD training by a designated HHD nurse with subsequent discussion with the HHD team at weekly meetings. The most suitable cannulation technique will help overcome the fear associated with cannulation and improve confidence in cannulation [41][42][43]. Rope-ladder cannulation remains the preferred cannulation technique for HHD, and our programme encourages this over buttonhole cannulation [44]. Although buttonhole cannulation is possible for HHD patients self-needling and patients with short segment for cannulation, it should be avoided  [45,46]. Patients that are MSSA and MRSA colonised should undergo successful eradication therapy before training with buttonhole cannulation. In the group of patients in the maintenance phase, the use of an audit tool using a checklist is helpful to identify patients that are at high risk for access failure and intervention and retraining [26,[47][48][49]. As cannulation failure will not allow the patient to proceed with dialysis treatment, the patient should be instructed to abandon further attempts when appropriate, rest the access, apply ice for a minimum of 10 min and report to the training centre [50]. When possible, images of vascular access should be transferred and shared via a secured platform to assist in decision-making. If the problem cannot be readily resolved, the HHD training unit should review the patient the next day or perform a home visit. The general vascular access alarms can be divided into venous and arterial pressure alarms. The arterial alarm may indicate line disconnection or access dysfunction, while venous alarm indicates thrombosis or clotting in the dialyzer or circuit, or kinking of lines. One of the most serious and life-threatening complications of HHD is venous needle dislodgement leading to significant blood loss, [32] and although this may not trigger the dialysis machine venous pressure alarm, wearing a wetness detector underneath the fistula alarm will trigger a separate alarm to alert the patients. As such wetness alarms are recommended for those HHD patients opting for nocturnal HHD. Training should highlight the importance of the alarms, the consequences of dismissing alarms, and troubleshooting steps. The use of educational tools such as the teach-back method should be considered to increase understanding and improve patient confidence [51,52]. Competency checks much be performed during training and regularly in the maintenance phase. As some alarms may not often be triggered, we suggest a short and concise patient reminder card in addition to the HHD patient manual on troubleshooting these alarms as visual reminders [53,54]. If the vascular access alarm is not resolved, patients are advised to contact HHD nursing or the technical hotline.
The recent Kidney Disease Outcomes Quality Initiative (KDOQI) Clinical Practice Guidelines for Vascular Access 2019 focused on regular physical examination and checking the vascular access to detect clinical indicators of flow dysfunction by trained and experienced dialysis nurses or physicians over regular access flow monitoring [50]. In the outpatient dialysis centre, vascular access surveillance is often conducted using various devices in addition to physical examination. However, this may be limited to purely clinical examination for HHD patients. Given that the KDOQI recommendations were based on weak evidence and the concern, in particular with reports of HHD patients at higher risk of vascular access thrombosis, possibly due to a higher intensity of dialysis, such as more prolonged and more frequent dialysis sessions, we suggest teaching patients to monitor and report access dysfunction and pseudo-aneurysm formation. This needs to be supplemented by physical examination by nursing and physician review during clinic encounters and six-monthly access flow monitoring [55,56]. When required, the HHD patient will need to undergo retraining with an audit tool to ensure that they are able to detect, record, and report access problem promptly to the HHD nursing team [47]. The HHD program should collaborate closely with the vascular access team to facilitate urgent vascular access evaluation and intervention when needed [57].

Discussion
The concept of the hospital at home (HAH) that substitutes hospital-level services at home for what would otherwise be inpatient hospitalisations has gained popularity over recent years, even locally, in Singapore [58]. Although HAH is distinctly different from HHD, which replaces an inpatient service rather than an outpatient service, the data reporting that HAH is safe and effective is promising to suggest the management of patients with acute medical illness in a home setting is safe [59][60][61]. Nevertheless, the data on safety is still relatively limited, and management in the home setting should be considered as supplementation to a healthcare facility rather than replacement in suitable individuals [62]. Worldwide, there are different approaches to HHD, with some centers restricting access to programs for patients living alone, whereas others provide assistance at home, and even help with access cannulation and decannulation. As we were initiating a HHD program de-novo, we restricted patient selection to those with fistula access and a partner or carer at home.
While HHD shares some similarities with ICHD, it also varies significantly with a more selected group of patients that are relatively healthier and having lesser human interaction involved in their treatment. The similar prescription and processes in HHD allow implementation of a system in a more controlled environment. HHD can, therefore can be undertaken safely, if not safer despite performing a high complex treatment at home. Notably, the risk and the safety strategies for HHD vary from that of ICHD [63]. While the common safety problems at ICHD were falls, medication errors, access-related, dialyser prescription errors, and excess blood loss or prolonged bleeding, we identified vascularaccess as the main safety area for HHD similar to that reported by Holey et al. [64,65]. The risk in other areas was successfully reduced by the selection and detailed planning, and framework focussing on risk reduction.
Our safety framework for the HHD program is focused mainly on risk reduction and pre-emptive evaluation. This includes strict assessment criteria for the suitability, training with frequent competency checks with individualised treatment plans for patients where appropriate, and regular checks using a standardised audit tool [25,66,67]. Patient selection for HHD is a defined and vigorous process and includes a thorough clinical assessment, evaluation of the home environment as well as psychosocial background [68]. Subsequent to this assessment, the patient will need to undergo a period of training, which will vary between patients, initially aiming for for 8-16 weeks with multiple clinical and theoretical competency assessments at completion of each skill and weekly. For the first haemodialysis treatment at home, a dialysis nurse and technologist will be present to observe the patient performing the session correctly and monitor for any technical errors. If there are no problems and the patient is confident after the first session, the first clinic visit will then be scheduled 4 weeks later Thereafter, clinic visits are scheduled at monthly intervals for the initial 6 months. Patients are provided 24-h nursing and technical tele consult at all times [66]. Ad-hoc home visit arrangements during office hour can be made if needed or patients will be asked to return clinic for review by physician or additional test.Home visits are scheduled to alternate between clinic visits. We used FMEA in our study to pre-emptively identify processes that can fail, prioritise failure modes with the higher risk, and set risk reduction strategies.
In terms of HHD infrastructure and equipment, there are often two or more critical stakeholders involved, with one being the contractor responsible for plumbing modifications and electrical works, and the dialysis vendor(s) providing dialysis machines, portable reverse osmosis machine and other components to the dialysis water system. Depending upon the quality of the potable water, additional or larger water softeners, carbon filters may be required depending on of the local water supply. The contractor and installation company for HHD will need to be familiar with and ensure full compliance with all regulatory requirements Adequate assessment of space and placement of equipment with regular maintenance by the vendor will reduce the risk of failure. HHD machines designs and consumables such as blood lines should be chosen with considerations for simplicity, patient-friendly set-up and interface, easy to trouble shoot with features to improve safety [66,69]. In the event of equipment failure, repairs will be initiated at the earliest possibility, and patients will be directed to dialyse at the HHD unit under the physician's direction.
Self-cannulation competency is the most crucial competency in completing HHD training. Failure in cannulation will not allow dialysis to be initiated regardless of competency in other areas. While assisted home dialysis, both for HHD and PD is possible,there is a significant health care cost.. The manpower cost can be prohibitive, and there may also be a shortage of dialysis trained nurses. Even if the trained staff are only required to start and finish the treatment session, the required hours including travel time for the nurse adds a significant cost factor, especially in countries with high wages. Vascular access-related events were found to have the highest RPN in our assessment of the early transition and maintenance phase of HHD. This is consistent with previous reports that vascular access is the most common category leading to severe adverse events in the HHD programme, with calls received frequently deemed severe [8,13,70]. We discussed the main areas with the highest RPN as agreed by the two different teams. As many of the potential risk mitigation solutions overlap, we suggest a bundle approach to addressing vascular-access related failure ( Table 4). As our program matures, we intend to include patients with catheter access in the programme but this will warrant a detailed review of risk and pre-emptive risk reduction strategies.
Previously published data from well-established programmes reported no correlation between the experiences of HHD programme with the occurrence of adverse events but found that these events occurred primarily in patients with some degree of HHD experience with a median vintage of two years, suggesting that there may be some degree of experimentation, serendipity, complacency, burnout, and non-compliance [13,47,71]. Our new programme's framework will include a minimum of two monthly reviews either at the HHD training centre or at home with an audit as the patient is dialysing. We will emphasize specific attention every two years through a thorough review using an audit tool and theoretical competency assessment and sharing incidents  [47] Close collaboration with vascular access team Retraining when necessary in the programme to highlight the importance of adherence to protocols [47]. Similarly, a two-year review of the patient's suitability to continue HHD will be conducted to ensure safety in continuing the treatment with progression or new cognitive and psychological changes. Experienced centres have reported technical issues and human error as a contributory cause to incidents in HHD programmes [13]. While our safety framework for the program is focused mainly on risk reduction and preemptive evaluation, not all failures can be wholly preventable, and measures will need to be put in place for the early detection of failure with an established workflow to resolve acute problems, with a rapid evaluation cycle to prevent future occurrences. Examples of early detection and workflow for acute access-related issues will include a wetness detector, blood sensors, or needle dislocation sensors both at the access site and under the fistula arm, coupled with accessibility to technical and nursing support [72,73]. We will be conducting regular audit of the programme going forward and conducting concise incident analysis or root cause analysis adopting a latent approach with multidisciplinary team involvement, comprising physician, nursing, quality expert, technical team, and possibly patients with each near miss or incident reported [74,75]. The action plan will be designed for prevention and early detection. Changes to workflow will be then subsequently made as necessary, emphasizing communication to all patients regarding any changes in a timely manner.
Failure mode and effects analysis (FMEA) is a proactive risk management tool for identifying the possible failure modes of a system, process, product or service, analysing the causes and effects of the failures, and eliminating or reducing the most significant ones by proposing risk mitigation actions [76]. FMEA is effective in evaluating both new and existing processes and systems. While FMEA has not been validated for HHD, we have used FMEA as an assessment tool before starting a new clinical program focussing on processes with highest risk. Given that there was no prior experience in HHD at our hospital, we sought collaboration from another unit. The involvement of the two HHD teams with different experiences and perspectives is unique and is also the strength of our study. It generates the opportunity to review HHD as a start-up with different challenges and to use the lessons from a mature HHD programme to help craft solutions to these challenges. The collaboration between the two teams provided an excellent platform for sharing and discussion as we focused on the safety aspects of our pilot HHD programme. The minor difference between HHD vascular access management between the two programmes are outlined in Table 5.
Although the processes of HHD are similar in many instances, the ability to generalise the failure modes and potential solutions for risk mitigation in two different settings may limit our study findings. There are inherent differences in our population in terms of logistics, demographics, and even simple differences such as climate and housing compared to that in London. To reduce unknown confounders related to these differences, we aim to conduct a second FMEA after starting our HHD programme and to examine our suggested solutions prospectively.

Conclusions
Although HHD may appear to be a complex treatment performed by patients at home without direct supervision, HHD is a safe therapy as evident by its past record. Enhancing safety and patient experience will encourage and motivate patient to choose HHD [55]. The reported adverse incidents are similar to that of ICHD. However, HHD patients must have the ability to manage any treatment related complications occurring at home, occasionally with the presence of a caregiver and hence, the enhanced focus on safety and the reason for this FMEA approach. Our study reports the use of FMEA with the involvement of two teams, at different ends of the clinical experience scale in HHD. We summarize here the risk assessment of possible areas of failure in starting HHD from different perspectives. The risk reduction strategies are not new, but we have designed a framework that addresses the specific areas in a bundle approach. As vascular access related medical emergencies were the most prominent in our risk assessment, our FMEA was primarily directed to this area of risk and we successfully identified risk reduction strategies. Table 5 Outline of differences between vascular access management in new and an established HHD programme

New HHD programme Established HHD programme
Frequency of review Monthly for the first 3 months then, 2 monthly clinic review and physical examination of vascular access by HHD nurse and physician 3 monthly review of vascular access by vascular surgeon with vascular access scans if necessary 6 monthly clinic review and physical examination of vascular access by HHD nurse Competency assessment 2 monthly review alternating home visit and HHD centre review by both HHD nurse and physician 6 monthly review alternating home visit and HHD centre review by HHD nurse