What blood tests are justified for routine monitoring of long-term conditions?

Problem

It is unclear what blood tests should be monitored periodically in people with hypertension (HTN), type 2 diabetes mellitus (T2DM) and chronic kidney disease stage 3 (CKD3). As part of our larger Optimal Testing project, we convened an expert panel that identified greatest uncertainty around monitoring sodium and potassium in HTN, T2DM and CKD3; haemoglobin in T2DM; and thyroid function in CKD3. Here we examine whether abnormalities of these analytes are more common in these long-term conditions than in those without.

Approach

We used the clinical practice research datalink (CPRD) to identify people with HTN, T2DM and CKD3 diagnosed between 01/01/2011-30/12/2019 and age- sex- and practice-matched controls. We used Cox regression to analyse time to first abnormal blood result (sodium <135 or >145mEq/L, potassium <3.5 or >5.5 mEq/L or haemoglobin <120 or 130 g/L), to first clinically significant abnormal result (sodium <130 or >150 mEq/L, potassium <3.0 or > 6.0 mEq/L, or haemoglobin <100 g/L), and MED3 coded hypothyroidism.

Findings

The total number of cases and 1:1-matched controls for HTN, T2DM and CKD3 were 83,563, 25,820, and 18,012, respectively. The proportion with at least one blood test was higher in cases than controls: 74%, 78% and 84% versus 39%, 40% and 48% for electrolytes in T2DM, HTN and CKD3, respectively; 71% versus 36% for haemoglobin in T2DM; and 47% versus 33% for TSH in CKD3. Among those with at least one test performed, the frequency of testing was still higher in cases, ranging from 1.6 versus 0.7 tests/year for sodium in T2DM to 0.7 versus 0.5 tests/year for TSH in CKD3. The hazard ratios for measuring an abnormal blood result were all significantly >1 for the LTCs, ranging from 5.4 (95%CI 4.9-6.1) for potassium in T2DM to 2.5 (2.3-2.7) for sodium in CKD3. To account for ascertainment bias due to higher testing in cases we restricted the analysis to those with at least one blood test and further adjusted for the yearly frequency of testing. The magnitude of all HRs fell, ranging from 1.7 (1.5-1.8) for haemoglobin in T2DM to 1.1 (1.0-1.2) for sodium in CKD3. There was a similar pattern for clinically significant abnormalities and coded hypothyroidism, with all HRs > 1 in the unadjusted analysis. But after accounting for testing frequency only potassium abnormalities in CKD (HR1.6, 94%CI 1.2-2.1) and sodium in HTN (1.2, 1.1-1.4) remained significantly > 1.

Consequences

Our results suggest that routinely testing electrolytes, haemoglobin and TSH in selected LTCs leads to increased detection of small abnormalities – producing potential downstream work such as retesting to check resolution. However, the underlying rate of clinically significant test abnormalities is similar between those with LTCs and controls.