Association of chronic kidney disease (CKD) and failure to monitor renal function with adverse outcomes in people with diabetes: a primary care cohort study

Diabetes mellitus is an increasingly common condition and is associated with increased risk of cardiovascular events and mortality [1, 2]. Furthermore, diabetes is the most common cause of end-stage renal disease (ESRD); in the United States it account for up to 45% of all new cases of ESRD [3]. Chronic kidney disease (CKD) is a risk factor for cardiovascular disease and mortality both in the general population [4, 5] and amongst people with diabetes [6]. The two components of CKD; reduced estimated glomerular filtration rate (eGFR) and proteinuria have been shown to be independently associated with cardiovascular events and mortality in the general population and in high risk populations [7‐9].

The National Institute for Health and Clinical Excellence (NICE) currently recommends all people with diabetes in England and Wales regularly have their renal function tested, including testing for albuminuria [10]. However, people with CKD complicating diabetes are not always identified and are sometimes sub-optimally managed in primary care [11]. Early identification of CKD and intervention with renoprotective measures, particularly the use of angiotensin converting enzyme inhibitors (ACEIs) and angiotensin receptor blockers (ARBs), has been shown to be effective in slowing progression of renal disease and in reducing cardiovascular events [12‐16], and treatment that reduces proteinuria also reduces the risk of progression [17].

We investigated the association of eGFR and proteinuria on adverse vascular and renal outcomes in people with diabetes, with or without hypertension, in a community setting. In addition we investigated the association of these outcomes in people who did not have monitoring of their renal function.

We performed a cohort analysis on all adults with diabetes in the Quality Improvement in Chronic Kidney Disease (QICKD) trial database. The QICKD trial was a three-armed cluster randomised controlled trial to analyse the impact of quality improvement interventions on blood pressure in people with renal disease [18]. Randomisation occurred at the primary care practice level. Practices were allocated to usual practice (no intervention), provision of clinical guidelines and prompts, or audit-based education. A reduction was of 2.41 mmHg (CI 0.59-4.29 mmHg; p = 0.012) was demonstrated with audit based education. However, neither intervention was found to have any impact on cardiovascular outcomes during the follow-up period. The trial database comprises routinely collected general practice (GP) data from 127 primary care practices across England; a nationally representative sample of urban, sub-urban and rural practices in London, Surrey, Sussex, Leicester, Birmingham and Cambridge between January 2006 and December 2010 [19]. There were additional records, of varying durations, for each person prior to these dates. Data recorded between January 2006 and June 2008 were used to determine the baseline characteristics of the people included in the study. A second data collection was undertaken at 30 months to obtain follow-up data.

We included all adults with type 1 and 2 diabetes. These patients were identified using a validated method for identifying correctly coded cases of diabetes from UK primary care records [20]. This method included analysis of diabetes read codes (e.g. read codes C10E and C10F are used for type 1 and type 2 diabetes respectively in UK primary care) in combination with documented investigation results. A person was defined as having diabetes if they had read codes for either type 1 or type 2 diabetes or laboratory values consistent with the World Health Organisation (WHO) diagnostic criteria for diabetes [21, 22] prior to, or during the baseline period; a fasting blood glucose > 7.1 mmol/l, non-fasting blood glucose > 11.1 mmol/l, or HbA1c > 59 mmol/mol (7.5%). People with a recorded diagnosis of gestational diabetes, or elevated blood glucose results during pregnancy were excluded unless they subsequently developed diabetes outside of pregnancy.

England has a registration based primary health care system. With very few exceptions the whole population is registered with a single GP. Patients access non-emergency services through their GP and the GP receives letters about all hospital attendances including emergencies. Patients all have a unique ID called NHS number, which is attached to all their medical records. The health system has been progressively computerised since the 1990s [23]. Since 2004 there have been pay-for-performance indicators for chronic disease management, to which CKD was added in 2006 [23]. Remuneration for these is based on extracts from routine computer data [24]. Routinely collected primary care data in the UK, therefore, provides a highly comprehensive patient record.

A total of 741,913 people were included for analysis, of whom 35.914 (4.8%) had diabetes. We excluded people who died or left the practice before the start of the follow up period (n = 123), or if they were aged less than 18 at the start of the follow up period (n = 142,533). From this remaining adult cohort 35,502 people met the diagnostic criteria for diabetes.

The mean age of this cohort was 63.6 years (standard deviation of 14.3 years). Males made up 54% of the population. Diabetes was coded as type 1 diabetes in 1,202 (3.4%) people and type 2 in 30,767 (86.7%). The remaining 3,533 (9.9%) had other codes consistent with a diagnosis of diabetes or investigations consistent with the diagnostic criteria but where diabetes type was not specified.

Overall 15,813 (44.5%) people were identified with CKD. Hypertension was found to be present in almost everyone with diabetes and CKD; 15,244 (96.4%) (Figure 2). 5,862 people were found to have hypertension without CKD.

A total of 3,144 people (8.9%) in the cohort had an adverse outcome during the 30 month follow up period (Figure 3). All-cause mortality was the most common adverse outcome with 1,495 cases (4.2%), equating to an annual mortality of 17 per 1,000 person years. Non-fatal cardiovascular events occurred in a total of 1,531 people (4.3%). 1,691 (4.8%) people were lost to follow up before suffering an adverse event. These people were included in our final analysis. The demographics of the population are provided in Table 1.

A diagnosis of heart failure, stroke, TIA, ischaemic heart disease, or MI before the start of follow up were all found to be independent predictors of adverse outcome (Table 2). No association with adverse outcome was found with ethnicity, BMI, HbA1c, total cholesterol, and the use of aspirin. These variables were therefore removed from the multilevel logistic regression model.

The odd ratios for adverse outcomes increased with declining renal function, proteinuria, and hypertension (Table 3). Cox regression analysis also demonstrates increasing hazard ratios with declining renal function (Table 4). Event free survival analysis for people with CKD with and without hypertension showed their summative effect on adverse outcomes (Figure 4).

A total of 4,460 (12.6%) people had incomplete CKD screening during the 2.5 year baseline period. This comprised 1,574 (4.4%) people with no serum creatinine recorded and 3,478 (9.8%) untested for proteinuria. People with unmonitored renal function, both with and without hypertension, were found to have significantly higher incidence of adverse vascular and renal outcomes than those with normal renal function. People with unmonitored renal function were found to have lower prescription rates of ACE inhibitors and ARBs (41.4%, 95% CI 40.2-42.6%) than people with no evidence of CKD (54.8%, 95% CI 54.1-55.4%) and people with CKD (71.1%, 95% CI 70.3-71.9%). They were also younger, more likely to drink excessive alcohol and smoke, had lower prescription rates of other medications, were more likely to have other missing data, and had worse mean cholesterol values (Table 5).

CKD and proteinuria are associated with worse health outcomes in people with diabetes, and this effect is additive. People with diabetes, who are not monitored for renal disease, have an increased risk of adverse events and appear to receive suboptimal medical therapy.