Assessment of the CDSS
In order to test the guideline against real decisions made by the local MDT, we retrospectively entered the data of 45 consecutive potential live kidney donors who presented to the Oxford Transplant Centre between January and March 2016. All available patient history and investigations were entered via the online forms, and the recommended decisions generated for each of the 4 categories were downloaded. Decisions generated by the CDSS were compared to actual clinical decisions made by the MDT using computer software (pfanalyse).
This retrospective analysis of anonymized data with no resulting alteration to patient care is considered as clinical audit by the NHS research and ethics service, and as such specific ethical approval is not required.
Development of the CDSS
The screening component of CDSS comprises 17 core tasks that are completed by all donors (Table S1, SDC,http://links.lww.com/TP/B606). In addition, a further 17 tasks are only activated in response to entry of specific user data (medical history, age/gender related screening, social history and previous surgery/hospitalization). The screening survey takes around 10 to 15 minutes to complete in the absence of a complicated medical history.
The nurse review component comprises 2 tasks; a review of the medical and social history, and the “red-flag” questionnaire from the ELPAT live donor psychological assessment tool to identify specific social, legal, or psychological concerns regarding donation.14
The laboratory and radiological investigations component comprises 3 core tasks to collate results of routine investigations; blood tests and urinalysis, CT angiogram, and GFR measurement. A further optional task collects results of split-function assessment where the size discrepancy between the kidneys is greater than 10%.
Once complete, decisions are provided in the 4 categories defined above: contraindications, further investigation, special considerations and donation side. Example tasks and decision candidates are illustrated in Figure 1.
Comparison of Decision Recommendations With Real Patient Episodes
Of the 45 consecutive patients presenting to the unit during the 3-month study period, 17 went on to donate a kidney. The reasons for not proceeding in the remaining 28 patients were: decision to proceed with an alternative live donor (n = 16), recipient received a cadaveric kidney transplant (n = 2), donor deemed medically unfit (n = 7), donor deemed psychiatrically unfit (n = 1) and donor withdrew from assessment (n = 2).
For the 17 patients that went on to donate, the median time from initial presentation to decision to proceed was 132 days (range, 56-626 days). The median time from presentation to donation was 204 days (range, 79-650 days).
The CDSS identified 15 major contraindications in 13 potential donors. Of those with major contraindications, 7 proceeded to donation. Two patients had a BMI greater than 35 kg/m2 but were deemed by the clinical team to be fit to donate after specific discussion with the patient as to the additional risks posed. Four patients had persistent traces of leucocytes in their urine in the absence of identifiable infection, and no further investigation was undertaken. One patient had a GFR just below the recommended threshold for age. Another had an initial protein:creatinine ratio greater than 50 mg/mmol, which was lower than 50 mg/mmol on 2 repeat samples with a renal biopsy demonstrating thin basement membrane disease.
Of the 7 patients deemed medically unfit to donate, major contraindications were identified in 3 (2 with BMI greater than 35 kg/m2, 1 with a history of breast cancer). A further 3 had contraindications identified on further investigations recommended by the decision tool; one had retroperitoneal fibrosis apparent on CT scan, another had an adrenal nodule requiring investigation, and the final one had significant cardiac disease identified on a myocardial perfusion study. One further patient had a combination of minor contraindications that were deemed too high risk to proceed.
Twenty-four of the 45 patients had minor contraindications to donation. Of the 17 proceeding to donation, nine had minor contraindications. Common minor contraindications included being a current smoker (n = 11), BMI between 30 and 35 kg/m2 (n = 7), persistent microscopic hematuria (n = 6), history of preeclampsia or hypertension (n = 5), and history of psychiatric disease or poor social support (n = 4).
The CDSS recommended additional investigations in 38 of 45 potential donors. Of the 93 additional investigations recommended, 40 (43%) were actually requested and completed. The majority of discrepancies related to the recommendation for oral glucose tolerance testing and the requirement for routine cancer screening tests. National guidelines recommend oral glucose tolerance testing in patients with high fasting blood sugar, BMI greater than 35 kg/m2, of high-risk ethnicity or with a family history of diabetes. Oral glucose tolerance test (OGTT) was only completed in 4 of 23 cases in which it was recommended. Bowel cancer screening is offered to all patients aged 60 to 74 years every 2 years in the United Kingdom. Of those eligible, only 1 of 16 had undertaken recent bowel cancer screening, and this was not requested before donation in the remaining cases. In 2 patients with a history of renal stones, metabolic screen or 24-hour urine collections were not completed as per national guidelines.
Compliance with other investigations was much higher. Patients with hypertension at presentation are recommended to undergo 24-hour blood pressure monitoring, which was undertaken in 9 of 11 patients in which it was recommended. Requirements for up-to-date cervical smears and mammograms in female patients were adhered to in 8 of 10 patients in whom these were required. For patients with persistent microscopic hematuria (n = 6), cystoscopy was requested for 4, and biopsy for 3.
In 2 patients, investigations were undertaken that were not recommended by the CDSS. These were a thrombophilia screen and an echocardiogram, both deemed clinically necessary by the assessing anesthetist.
The decision tool identified special considerations in 19 of 45 patients. Referral to a geneticist was recommended in 5 patients due to a history of potentially inheritable kidney disease. Referral was deemed unnecessary by the nephrologist in all cases due to the nature of the history provided and documented, suggesting appropriate management of these patients.
Referral to a psychiatrist was recommended on the basis of clinical history or nurse concern for 5 patients, all of whom underwent further assessment by a psychologist or psychiatrist. One patient of childbearing age indicated the desire for a future pregnancy, and the need for pregnancy counseling was correctly identified. Five patients were correctly identified as taking either the oral contraceptive pill or hormone replacement therapy that needed to be stopped prior to surgery.
The decision tool recommended a donation side (left or right kidney) if data from a CT angiogram and GFR measurement were provided. Sufficient data were available for a decision for 25 patients; the remaining patients did not proceed far enough through the donation pathway for these investigations to be completed.
The decision tool identified the same donation side as the clinical MDT in 23 of 25 patients. In 2 patients, a significant size discrepancy was identified (>1 cm difference between kidneys). National guidelines recommend assessment of split function with a DMSA scan in these cases, but this was not completed, and the larger kidney was donated in both cases.
This study demonstrates the feasibility of applying a CDSS to a complex, guideline-driven clinical pathway assessing potential live kidney donors. Use of patient- and nurse-entered data in this way has the potential to identify contraindications and necessary additional investigations, improving compliance with guidelines and ultimately reducing donor risk while allowing true informed consent specific to the individual patient.
We have demonstrated that a CDSS has the potential to identify contraindications and investigations defined by guidelines that may be overlooked by the clinical MDT. This is particularly important given the nature of live kidney donation, where the safety of the potential donor is the overriding consideration. Seven of the 17 patients proceeding to donation in the current study had major contraindications as defined by the national guidelines. Although in some of these, adequate consideration was given to the impact of these contraindications and discussion with the patient was documented, there were a number of cases where no patient discussion or MDT consideration was recorded, suggesting that they may not have been identified by the MDT. Although guidelines ultimately advise as to best practice, clinical practice that contradicts best available evidence should be carefully considered, documented, discussed as part of the informed consent process and future guidelines revision. A CDSS, such as the one developed here, will reduce the chance of overlooking specific clinical risks, providing a personalized risk assessment for each individual patient that can be used as an integral part of the informed consent process. Further developed, it may provide an effective feedback loop in advancing guidelines.
Similarly, the CDSS was able to identify a number of additional investigations that were recommended by national guidelines but were not completed. Again, clinical judgment may dictate that in individual cases the risks of further investigation outweigh the need, but the use of such a tool will provide guideline-based recommendations to ensure that these considerations are not overlooked. In particular, it helps to identify where cancer screening is not up-to-date, especially where algorithms are complicated and difficult to remember.
The finding of poor compliance with national guidelines is not unique to our center. A previous study demonstrated a great deal of variability in live kidney donor assessment between UK centers in factors, such as age and BMI thresholds, exclusion on the basis of medical history, and the use of investigations.5 Practice in many centers disagreed with national guidelines, suggesting a role for a CDSS in highlighting evidence-based practice.
The use of a patient- and nurse-completed CDSS may have additional benefits over and above guideline compliance and patient safety. In the current analysis, median time from presentation to decision to proceed with a donor was 132 days, and to donation was 204 days, with a great deal of variation between patients. Although some of this variation was due to recipient factors or use of the Live Donor Kidney Sharing Scheme, in a number of donors prolonged assessment resulted from sequential requesting of investigations or clinical review by other services (eg, psychiatry). Remote entry of data by patients in their home environment, along with earlier identification of necessary investigations, may improve the coordination of investigations and reduce the duration of assessment and number of clinic visits required. This may be important, as the time required for clinic visits has been identified as a potential barrier to the donation process, with improvements in donation rates seen with a 1-stop clinic model.15 Clinical input to the decision support system use may also prevent unsuitable candidates from progressing further down the assessment pathway. All of this has the potential to save patient and nurse time, and ultimately reduce service costs.
There are a number of limitations to the present study. The single-center nature of our analysis means that we are unable to assess the potential impact of this tool in other transplant centers, where compliance with guidelines may differ. Although retrospective analysis of past decisions suggests some room for improvement, we were unable to establish the true impact of a CDSS used in a prospective setting. Previous studies using CDSS for more straightforward clinical decisions have suggested improved compliance with published guidance when the CDSS is used at the point of decision making. A systematic review of studies reporting the use of CDSS in a number of different settings (drug prescribing, reminders, diagnostics and disease management) identified some evidence for improved practitioner compliance, although very few trials documented improved patient outcomes as a result.16 Similar improvements in compliance have been demonstrated in the more complex setting of cancer treatment decisions,8 and in reducing unnecessary investigation in preoperative assessment.17
Successful use of a CDSS relies on clinicians acting upon the clinical decisions provided by the system. This will only happen if the CDSS correctly reflects clinical practice guidelines, and if clinicians believe in the recommendations produced in such guidelines. Good quality guidelines recommend practice based upon the best available evidence via a robust guideline development process. Recent work has demonstrated that the quality of international guidelines in renal transplantation is variable, with guidelines lacking in robust developmental methodology which may have an impact on their uptake in real clinical practice.18 It is therefore important that clinical decisions do not just rely on guideline-based CDSS, but also take into account local practice and clinical experience as part of the live donor MDT process. However, a CDSS, such as the one presented here, can be adapted to local protocols and practice, which may reduce the risk of noncompliance with recommendations when evidence is lacking or of poor quality.
The CDSS developed in this study relies in part on data entry by patients, replacing the use of an existing screening questionnaire. Many transplant centers use a paper-based or online screening questionnaire to identify basic demographic, medical, and social data and to rapidly screen out unsuitable donors. Use of this data may speed up the assessment process but will still require verification by a healthcare professional to ensure accuracy before a final decision can be made. Completion of the patient tasks in the current CDSS takes less than 15 minutes, making it feasible for the majority of patients. Further refinement by populating patient demographics and investigation results directly from the electronic health record will reduce patient and nurse time further and help to prevent transcription errors. The ability to save the form to return to at a later date would also enhance the patient experience. Future incorporation of tools that allow calculation of patient-specific risk of outcomes, such as end-stage renal disease, may further enhance the ability of a CDSS to assist in patient-tailored informed consent.19
In summary, we have demonstrated the feasibility of turning a complex set of clinical practice guidelines for the assessment of potential live donors into a machine-executable data and decision making model. This system is able to accurately identify contraindications, investigations and the most suitable procedure based upon a combination of patient- and nurse-entered data and has the potential to improve evidence-based clinical decision making and patient-specific informed consent. Planned future work will develop this into a patient-facing system and assess the potential impact of this tool in the prospective setting, assessing external validity in other transplant centers.
2. Truog RD. The ethics of organ donation by living donors. N Engl J Med
3. Tong A, Chapman JR, Wong G, et al. Living kidney donor assessment: challenges, uncertainties and controversies among transplant nephrologists and surgeons. Am J Transplant
5. Arunachalam C, Garrues M, Biggins F, et al. Assessment of living kidney donors and adherence to national live donor guidelines in the UK. Nephrol Dial Transplant
6. Thiessen C, Kim YA, Formica R, et al. Written informed consent for living kidney donors: practices and compliance with CMS and OPTN requirements. Am J Transplant
7. Scheitel MR, Kessler ME, Shellum JL, et al. Effect of a novel clinical decision support tool on the efficiency and accuracy of treatment recommendations for cholesterol management. Appl Clin Inform
8. Séroussi B, Bouaud J, Gligorov J, et al. Supporting multidisciplinary staff meetings for guideline-based breast cancer management: a study with OncoDoc2. AMIA Annu Symp Proc
9. Patkar V, Acosta D, Davidson T, et al. Using computerised decision support to improve compliance of cancer multidisciplinary meetings with evidence-based guidance. BMJ Open
10. Miles A, Chronakis I, Fox J, et al. Use of a computerised decision aid (DA) to inform the decision process on adjuvant chemotherapy in patients with stage II colorectal cancer: development and preliminary evaluation. BMJ Open
11. Niazkhani Z, Pirnejad H, Rashidi Khazaee P. The impact of health information technology on organ transplant care: a systematic review. Int J Med Inform
12. Sutton DR, Fox J. The syntax and semantics of the PROforma guideline modeling language. J Am Med Inform Assoc
13. Fox J. Cognitive systems at the point of care: the CREDO program. J Biomed Inform
14. Massey EK, Timmerman L, Ismail SY, et al. The ELPAT living organ donor psychosocial assessment tool (EPAT): from “what” to “how” of psychosocial screening—a pilot study. Transpl Int
15. Graham JM, Courtney AE. The adoption of a one-day donor assessment model in a living kidney donor transplant program: a quality improvement project. Am J Kidney Dis
16. Garg AX, Adhikari NK, McDonald H, et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. JAMA
17. Sim EY, Tan DJA, Abdullah HR. The use of computerized physician order entry with clinical decision support reduces practice variance in ordering preoperative investigations: a retrospective cohort study. Int J Med Inform
18. O'Donoghue KJM, Reed RD, Knight SR, et al. Critical appraisal of international clinical practice guidelines in kidney transplantation using the appraisal of guidelines for research and education (AGREE) II tool: a systematic review. Transplantation
. 2018; doi:10.1097/TP.0000000000002255.
19. Park JS, Ahn HK, Na J, et al. Development of a screening tool to predict chronic kidney disease risk in post-nephrectomy living kidney donors. Transplant Proc
Supplemental Digital Content
Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.