Early identification of chronic kidney disease: it is time to enhance patient and population-based informatics tools for general practitioners

Abstract

Chronic kidney disease (CKD) is a global public health issue that can lead to several complications such as, kidney failure, cerebro/cardiovascular disease, and death. There is a well-documented “awareness gap” among general practitioners (GPs) to recognize CKD. As shown by estimates stemming from the Health Search Database (HSD) of the Italian College of General Practitioners and Primary Care (SIMG), no substantial changes were observed in terms of the incident rate of CKD over the last 10 years. Namely, 10.3–9.5 per 1000 new cases of CKD were estimated in 2012 and 2021, respectively. Thus, strategies to reduce under-recognized cases are needed. Early identification of CKD might improve patient’s quality of life and clinical outcomes. In this context, patient- and population-based informatic tools may support both opportunistic and systematic screening of patients at greater risk of CKD. As such, the new effective pharmacotherapies for CKD would be proficiently administered. To this aim, these two complimentary tools have been developed and will be further implemented by GPs. The effectiveness of these instruments in identifying the condition at an early stage and reducing the burden of CKD on the national health system needs to be verified according to the new regulations on medical device (MDR: (EU) 2017/745).

Keywords:

• Computer-based tools
• clinical governance
• clinical decision support system
• primary care

Chronic kidney disease (CKD) is a global public health issue leading to several complications such as kidney failure, cerebral/cardiovascular disease, and death. In the last decades, the burden of CKD has been growing, with a global prevalence of 9.1%¹. Furthermore, the decrease in the disability-adjusted life year (DALY) rate for CKD is not associated to a decrease in rates of CKD mortality or prevalence, suggesting a shift to mortality due to CKD occurring at an older age. By 2030, CKD is expected to become the fifth leading cause of death worldwide¹. In Italy, CKD afflicts more than 6 million people, with a prevalence of about 7.5% and 6.5% among men and women, respectively². However, it is important to highlight that only around 10% of patients affected by CKD were aware about their disease^1,3.

In the Health Search Database (HSD; Italian College of General Practitioners and Primary Care), from 2012 to 2021, some increase in the yearly prevalence of CKD was observed. Namely, the cumulative prevalence of CKD ranged from 5.2% to 13% during this period. Regarding the yearly incidence (i.e. new cases of CKD by year), some decrease was observed between 2012 and 2013 with 10.3 and 7 cases per 1000, respectively. Otherwise, values ranging from 6 to 8 per 1000 were obtained from 2014 to 2021. Of note, during the SARS-CoV-2 pandemic, the incidence rate kept values ranged from 8 to 9.5 per 1000. As such, no substantial difference in capturing new cases of CKD was revealed over the last 10 years.

There is a well-documented “awareness gap” among general practitioners (GPs) to recognize CKD in several countries, including Italy³. Indeed, only 36% of the countries confer to CKD the rule of health-care priority⁴. In a study involving 300 GPs, which recruited a cohort of about 500,000 patients, serum creatinine levels were requested for only 17% of the patients. Although, 16% of this patient subgroup was affected by CKD (operationally defined as eGFR <60 mL/min/1.73m²), only 12.5% of them were identified as CKD sufferers by GPs^5,6. Along this line, it was recently reported that 77% of patients with proven Stage G3 CKD were undiagnosed by their GPs in Italy⁷. The fact that CKD is under-recognized by GPs depends on several reasons. Firstly, there is a lack of CKD knowledge and a poor awareness of CKD guidelines and factors which are all potentially related to the awareness gap in primary care⁸. Furthermore, CKD is asymptomatic in the first stage. It also frequently coexists with other major conditions, mainly cardiovascular diseases, which might draw most of the GP’s attention. Nevertheless, the effect of kidney disease is not limited to the renal environment but extends to cardiovascular diseases for which it is an independent risk factor^9,10. Furthermore, as shown in the GBD 2017 Risk Factor Collaborators¹¹, the DALY rate for impaired kidney function is greater than that for drug use, unsafe sanitation, low physical activity, second-hand smoke, and many other dietary risk factors. In this respect, a consensus emerged on “Early Identification and Intervention in CKD”, highlighting the need to implement of effective screening coupled with risk stratification^12,13.

Regarding the support for GPs in the early identification of CKD, although accurate algorithms are available to raise the suspect of CKD, no effective clinical decision support systems (CDSS) have been largely implemented in primary care^14–16. This is not true for other chronic conditions, for whom the use of CDSS provided a significant improve in their identification and/or management^17–20. As for CKD, several country-specific prediction models have been developed and validated²¹. They comprise algorithms gathered through several techniques, such as traditional score development and validation as well as supervised machine learning approaches^22,23. For the latter, Random Forest, J48 algorithm, gradient boost, support vector machine, neural network and others, have been proposed for CKD prediction. However, they likely suffer from high risk of bias, which is mainly due to the absence of external validation studies. In addition, they have a poor interpretability. In this respect, Generalized Additive Models (GAMs) could represent an adequate response for CKD forecasting, given their feasibility in learning complex associations through the combination of nonlinear functions, thereby uncovering nonlinear associations and interactions and improving the predictive accuracy²⁴. Regarding the patient-based tool, by using a dashboard-based software, the GP would be able to visualize the patient’s risk estimate combined with the list of those risk factors leading to this assessment. In Italy, through MilleDSS^i,25 we developed such a system by which a visual reminder appears when the GP inspects patient’s Electronic Medical Records (EHRs) including at least one of the five risk factors for CKD (i.e. arterial hypertension, diabetes, cardiovascular disease, proteinuria ≥ 300 mg, creatinine >1.5 mg/dL and >1.3 mg/dL for males and women, respectively) in those still undiagnosed with the disease. In the example reported in Figure 1 the hand indicates the red (high risk) section, supporting GP in his/her clinical decision making^21,22. Currently, we are developing and testing a similar prediction model based on GAM methodology²⁴, which will improve the predictive accuracy of the current prediction model.

Figure 1. Examples of MilleDSS (A) and GPG (B) tools (in Italian) for opportunistic and systematic screening of CKD. (A) CDSS: Dashboard shows an example of “high risk” of CKD. (B) Governance tool: process indicators with green and red cursors indicating the absence or presence of criticisms; blu cursor shows the prevalence of patients at risk of incurring in CKD (simulated data of an individual general practitioner having in charge 1535 patients).

In any case, the CDSS should be complemented by other clinical governance applications, in which the GP could visualize his/her cohort of patients according to the aforementioned prediction score and other process indicators²⁶. By doing so, a general picture of those being at different risk levels would be complemented by specific indicators to quantify several specific aspects. For instance, among diabetes/hypertension/cardiomyopathy sufferers being tested with creatinine and/or microalbuminuria/proteinuria as well as BMI or lipid profile assessment at least once. Then, these indicators should be effectively ascertained by viewing the entire cohort at risk when the overall population is the denominator. A bar shows the related points of criticism according to “green” and “red” cursors. As in data simulation shown in Figure 1, the Italian General Practice Governance (GPGⁱⁱ) enables GPs to visualize such a picture of their patients. In addition to viewing the sub-groups size on or out of target (i.e. “green” or “red” cursor), clicking on the respective numbers, patient lists can be generated and ordered according to specific criteria. As such, some categories of patients might be prioritized for future visits and/or prevention, screening, or communication campaigns.

Given the crucial role of GPs in the early diagnosis of CKD, there is the need to require accessible and targeted CKD detection in primary care. Nevertheless, there are still uncertainties on the benefit/risk ratio related to screening or case finding per se. This is particularly relevant considering that cost-benefit analyses including new medications, namely sodium-glucose cotransporter-2 inhibitors (SGLT-2i), are not still available. As for screening programs, many of these showed a lower benefit than expected, and screening for CKD is not excluded^27–29. Indeed, given the unclear relationship between intensive screening (i.e. serum creatinine evaluations) and overt kidney disease, a potential harm exerted by treatments on asymptomatic patients cannot be excluded. As such, there is evidence that screening is unlikely cost-effective, whereas case finding appears the best approach in terms of cost and benefit^13,30,31. This is why clinical investigations should be tailored on certain risk categories, for which the concurrent risk factors would increase the development or the progression of CKD. Along this line, the GPs’ workload in screening their patients for CKD would be affordable. Namely, Italian GPs are allowed to enlist up to 2000 patients on the basis of the regional health care system. According to the data simulation report shown in Figure 1, one-third of these patients might be at risk of developing CKD across different risk categories. In this context, the presence of supportive informatic tools might be pivotal in providing information that entails cost-effective screenings^13,30,32 based on opportunistic and systematic approaches using CDSS and governance tools (i.e. MilleDSS and GPG in Italy), respectively.

In Italy, GPs are independent contractors for the national health service. They invest in their own computer-based tools, thereby participating in networks and creating an epidemiological real-time observatory. Therefore, a real-life comparator could be available given that numerous subgroups of them will benefit these “automatisms”. As such, clinical evidence will depend on the effectiveness of these instruments to reduce the burden of CKD according to the new medical device regulation (MDR: (EU) 2017/745). This is particularly relevant when new effective medications have recently become available for CKD therapy^33–36.

In conclusion, given the multiple facets of CKD, robust tools are needed to early and effectively identify most of the patients with CKD and promptly intervene with appropriate care.

Transparency

Declaration of funding

This article was funded by Genomedics Srl which has the ownership and copyright of MilleDSS technology.

Declaration of financial/other relationships

Lapi F. and Marconi E. provided consultancies in protocol preparation for epidemiological studies and data analyses for AstraZeneca, Mundipharma, MSD. Medea G. and Cricelli C. provided clinical consultancies for AstraZeneca, Mundipharma, MSD. Cricelli I is an employee of Genomedics Srl. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

Cricelli I., Piccinocchi C., Medea G., Cricelli C. and Lapi F. conceptualized the study. Cricelli I., Marconi E. and Lapi F. wrote the manuscript. Lapi F and Cricelli C. are responsible for the integrity of the work. All aspects of the study were led by the authors.

Acknowledgements

None.

Notes

i MilleDSS is a registered trademark of Genomedics Srl - Florence, Italy.

ii GPG is a registered trademark of Genomedics Srl - Florence, Italy.