> For the complete documentation index, see [llms.txt](https://ourfuturehealth.gitbook.io/our-future-health/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://ourfuturehealth.gitbook.io/our-future-health/data-types/clinic-measurements-data/poct-lipid-profile-data/poct-validation-study.md).

# POCT validation study

## Validation of the Mission® point-of-care testing (POCT) lipid profile <a href="#validation-of-of-the-mission-r-cholesterol-point-of-care-testing-poct-device" id="validation-of-of-the-mission-r-cholesterol-point-of-care-testing-poct-device"></a>

*The results from this validation study will be submitted for publication in due course.*

### Introduction   <a href="#study-overview-aims-and-objectives" id="study-overview-aims-and-objectives"></a>

Our Future Health used the Mission® Cholesterol (ACON Laboratories, Inc., San Diego, USA) point-of-care testing (POCT) device to obtain finger-prick blood lipid measurements during baseline clinic visits, including total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and triglycerides (TG). The TC:HDL-C ratio and non-HDL cholesterol were derived manually as TC divided by HDL-C, and TC minus HDL-C, respectively.&#x20;

The Mission® POCT device was selected as one out of four devices that met the National Cholesterol Education Program (NCEP) required standard of evidence for bias and precision (bias ≤3%) in a report by the National Institute for Health and Care Research (NIHR)[<sup>1</sup>](#references). This NIHR report included the findings from a single validation study of the Mission POCT device by Kurstjens et. al.[<sup>2</sup>](#references) In a sample of 59 adults, Kurstjens et. al.[<sup>2</sup>](#references) found the Mission® device to have high correlation (Pearson's r = 0.96) with laboratory measurements for TC. The device showed high sensitivity (96%) but low specificity (67%), with a high negative predictive value (NPV, 96%), and lower positive predicted value (PPV, 65%) for TC. Measurements of TG were systematically overestimated, while HDL-C was underestimated, resulting in compounded inaccuracies in derived measures. Consequently, calculated metrics such as the TC:HDL-C ratio were found to be less reliable.&#x20;

To assess the accuracy and reliability of the Mission device and POCT lipid profile data, Our Future Health conducted a cross-sectional paired-sample validation study involving 260 adult volunteers. Participants attended an Our Future Health appointment at one of two centres in Bristol or Nottingham during June 2023. Ethical approval was obtained from the Research Ethics Committee to collect an additional blood sample from participants for quality assurance purposes during their standard Our Future Health appointment.

#### Aim

To assess the validity of the Mission® POCT device for lipid profile testing against a laboratory-based reference standard (Roche Cobas® 702 Chemistry Analyzer).

#### Objectives

The primary objective was to evaluate the accuracy of the Mission® POCT device compared with laboratory measurements (NHS standard) for six lipid-related analytes: TC, HDL-C, LDL-C (calculated), TG, TC:HDL-C ratio (calculated), and non-HDL cholesterol (calculated).

Secondary objectives were to:

1. Assess the classification performance (sensitivity and specificity) of the Mission® POCT device for detecting elevated total cholesterol (TC > 5 mmol/L and TC > 7.5 mmol/L) and elevated LDL-C (> 3 mmol/L).
2. Compare the validity of LDL-C derived using the device-reported Friedewald equation[<sup>3</sup>](#references) with LDL-C calculated using the updated Sampson–NIH equation[<sup>4</sup>](#references).

### Methods

#### Eligibility criteria

1. age ≥18 years
2. prior enrolment in the Our Future Health programme&#x20;
3. sample collection at either Bristol or Nottingham Our Future Health centres in June 2023

No exclusion criteria were applied. All participants had previously consented to participate in the programme. The study population was not selected based on demographic characteristics such as age, sex, or ethnicity.&#x20;

#### Sample processing

Rapid lipid profile using the Mission® POCT device was performed on capillary fingertip whole blood by trained staff. An additional 3.5 ml venous blood sample was collected from each participant using a BD Vacutainer® SST II Advance tube, with informed consent.&#x20;

Venous blood samples were transported the same day to the UK Biocentre (Milton Keynes, UK) laboratory for processing. Any sample with haemolysis or visible lipaemia were recorded. After end-of-day centrifugation, samples were stored at +4°C overnight and transferred the following morning to the Department of Pathology at Northampton General Hospital NHS Trust for lipid analysis using the Roche Cobas® 702 Chemistry Analyzer. The laboratory is UKAS-accredited (No. 8115) and provides routine NHS lipid testing.&#x20;

All samples were disposed of in accordance with HTA regulations.&#x20;

#### Statistical analyses&#x20;

We evaluated six lipid-related analytes: TC, HDL-C, LDL-C (calculated), TG, the TC:HDL-C ratio, and non-HDL cholesterol.

LDL-C was calculated by the Mission® POCT device using the Friedewald equation[<sup>3</sup>](#references), but this calculation is only valid when the participants are fasting and TG levels are below a certain threshold[<sup>5</sup>](#references). As the participants in this study did not fast, we also assessed LDL-C based on the updated Sampson-NIH equation[<sup>4</sup>](#references). &#x20;

TC:HDL-C ratio and non-HDL cholesterol were derived manually:&#x20;

* TC:HDL-C ratio = `TC/HDL-C`
* non-HDL cholesterol  = `TC - HDL-C`

To assess the accuracy and agreement between Mission® POCT device and laboratory-based measurements, we conducted the following statistical analyses:

1.  Pearson correlation coefficients to evaluate the linear relationship between the two methods.
2.  Passing-Bablok regression, a robust non-parametric method that is resistant to outliers and does not assume a specific distribution of the data, was used to test bias. Specifically, the regression slope and intercept were interpreted as indicators of proportional and constant bias, respectively.&#x20;
3.  Bland-Altman analysis was used to test agreement through the calculation of the mean difference (bias) and limits of agreement (LOA), supplemented by one-sample paired t-tests to evaluate whether the mean difference was statistically significant.&#x20;

To evaluate the classification performance of the Mission® POCT device in detecting high TC (>5.0 mmol/L), very high TC (>7.5 mmol/L), and elevated LDL-C (>3.0 mmol/L), we calculated:

1. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV).
2. Cohen’s kappa coefficient to assess agreement beyond chance.&#x20;

&#x20; These thresholds were chosen for the following reasons:&#x20;

* A total cholesterol level above 5 mmol/L is considered high for most healthy adults. Clinical actions are rarely based on this number alone, but rather on an overall cardiovascular risk assessment (including LDL-C, HDL-C, smoking status, blood pressure, and BMI).
* Based on NICE guidelines and NHS lipid management pathways, total cholesterol levels of >7.5 mmol/L (<30 years) and >9.0 mmol/L (>30 years) are considered severe and raise a strong suspicion of Familial Hypercholesterolaemia, a genetic condition requiring immediate clinical action to reduce the high risk of premature cardiovascular disease.
* An LDL-C level greater than 3 mmol/L is considered raised according to UK guidance and generally merits clinical action, particularly in patients with cardiovascular risk factors.

### Results <a href="#results" id="results"></a>

Complete paired data from Mission® POCT device and laboratory cholesterol results were available for 260 participants (44% male, 6.4% non-white ethnicity, 52% age 50 years or older).&#x20;

#### Accuracy and agreement between Mission® POCT device and laboratory-based measurements

**Total Cholesterol (TC)**&#x20;

For TC, values from Mission® POCT device yielded slightly lower mean and median estimates compared with laboratory measurements (Table 1). &#x20;

The Mission® POCT device method showed a very strong correlation with the laboratory method (Pearson r = 0.899).&#x20;

Passing-Bablok regression indicated slight but statistically significant proportional bias (slope = 1.07, 95% CI: 1.00–1.14) and systematic bias (intercept = -0.34, 95% CI: -0.75 to -0.03). Although mean POCT values were slightly lower than laboratory measurements, proportional bias indicated that POCT may slightly overestimate values at higher concentrations (Figure 1).&#x20;

Bland-Altman analysis showed no evidence of systematic bias (a mean bias of 0.027, 95% CI: -0.03 to 0.09). The paired t-test provided no evidence of differences in mean TC (p-value of 0.357). The LOA ranged from -0.95 to 1.00, with relatively narrow confidence intervals, suggesting consistent variability across the measurement range (Figure 2). The LOA suggests that 95% of POCT measurements of TC could be expected to be between 0.95 mmol/L lower and 1.00 mmol/L higher than the laboratory value (Figure 2).&#x20;

**HDL Cholesterol**&#x20;

For HDL-C, both POCT and laboratory measurements had the same mean estimate (1.56 mmol/L), but the median estimate was slightly higher for POCT (1.59 vs 1.52 mml/L, Table 1).

HDL-C showed strong correlation with laboratory measurements (Pearson’s r = 0.821) and good overall agreement.

Passing-Bablok regression indicated no evidence of systematic or proportional bias (Figure 1), with a slope of 1.05 (95% CI: 0.97 - 1.13) and an intercept of -0.04 (95% CI: -0.16 to 0.09), both not statistically significantly different from 1 and 0, respectively.

Bland-Altman analysis showed a negligible mean bias (-0.003 mmol/L), with LOA ranging from -0.49 to 0.49 mmol/L, indicating narrow variability across the measurement range (Figure 2). This LOA suggests that 95% of POCT measurements of HDL-C would be expected to be between 0.49 mmol/L lower and 0.49 mmol/L higher than the laboratory value (Figure 2).

**LDL Cholesterol**&#x20;

LDL-C measurements provided by the POCT device (Friedewald equation) were missing for 5 of the 260 participants and the mean and median estimates were lower for POCT compared with laboratory measurements (Table 1). Mean and median LDL-C based on the Sampson-NIH equation were also lower for POCT compared with lab (Table 1).

Both LDL-C results showed very strong correlations with laboratory tests (Friedewald: Pearson r = 0.884; Sampson-NIH: Pearson r = 0.911), with Sampson-NIH demonstrating slightly stronger agreement.&#x20;

For LDL-C calculated with the Mission® POCT device using the Friedewald equation, a systematic bias was evident.&#x20;

* The significant intercept of the Passing-Bablok regression (-0.26, 95% CI: -0.44 to -0.12) indicates a constant offset between POCT and laboratory methods, while there was no evidence of proportional bias (1.03, 95% CI: 0.98–1.09; Figure 1).&#x20;
* Bland-Altman analysis revealed a moderate positive bias, with a mean bias of 0.174 mmol/L (95% CI: 0.12 to 0.23) and moderately wide limits of agreement (-0.77 to 1.12 mmol/L), which indicates that it is expected that 95% of POCT results will fall between 0.77 mmol/L lower and 1.12 mmol/L higher than the laboratory value (Figure 2).&#x20;

For LDL-C re-calculated using the Sampson-NIH equation, agreement with the laboratory method was improved.&#x20;

* The Passing-Bablok regression showed a slope closer to unity (1.01, 95% CI: 0.97–1.06) and a similar intercept (-0.22, 95% CI: -0.36 to -0.09), indicating better calibration across the measurement range (Figure 1).&#x20;
* Bland-Altman analysis revealed a moderate positive bias, with a mean bias of 0.174 mmol/L (95% CI: 0.13 to 0.22) and slightly narrower LOA (-0.61 to 0.96 mmol/L), which indicates that 95% of POCT results are expected to fall between 0.61 mmol/L lower and 0.96 mmol/L higher than the laboratory value (Figure 2).&#x20;

Overall, although the mean bias was nearly identical between methods, the Sampson-NIH calculation showed improved agreement with narrower limits of agreement and better calibration compared with Friedewald. This indicates that the Sampson-NIH equation may provide a more reliable estimate of LDL cholesterol in this dataset, particularly given the observed variability in triglyceride measurements. &#x20;

**Triglycerides (TG)**&#x20;

For TG, values from Mission® POCT device yielded considerably higher mean and median estimates compared with laboratory measurements (Table 1). &#x20;

A moderate correlation was observed for TG (Pearson r = 0.582).&#x20;

Although the regression slope (1.10, 95% CI: 1.03–1.23) suggested proportional bias, the intercept (0.04, 95 CI: -0.08 to 0.11) did not show significant constant offset in regression (Figure 1).&#x20;

Bland-Altman analysis revealed a substantial negative bias (-0.431 mmol/L, 95% CI: -0.54 to -0.32) and wide LOA (-2.24 to 1.38 mmol/L), indicating relatively poor agreement; 95% of POCT results for TG were expected to be between 2.24 mmol/L lower and 1.38 mmol/L higher than the laboratory value (Figure 2).  &#x20;

**TC:HDL-C Ratio**&#x20;

For the TC:HDL-C ratio, the mean values were very similar between POCT and lab measurements, but POCT had a slightly lower median (Table 1).&#x20;

The TC:HDL-C ratio showed very strong correlation with laboratory measurements (Pearson r = 0.892).

Passing-Bablok regression indicated both systematic and proportional bias, with a slope of 0.91 (95% CI: 0.86 - 0.97) and an intercept of 0.20 (95% CI: 0.03 to 0.33), suggesting underestimation of POCT-derived ratios at higher values (Figure 1).

Bland-Altman analysis showed minimal mean bias (0.007, 95% CI: -0.06 to 0.08), with relatively wide LOA (-1.11 to 1.13), indicating that 95% of POCT results are expected to be between 0.61 mmol/L lower and 0.96 mmol/L higher than the laboratory value (Figure 2).

**Non-HDL Cholesterol**&#x20;

For non-HDL-C, both the median and mean values were very similar between POCT and lab measurements (Table 1).&#x20;

Non-HDL cholesterol showed a strong correlation (Pearson r = 0.939).&#x20;

Passing-Bablok regression indicated excellent agreement. The slope (1.00, 95% CI: 0.96–1.04) and intercept (-0.04, 95% CI: -0.21 to 0.11) were close to the expected values of 1 and 0, respectively (Figure 1).

The mean bias was 0.03 mmol/L (95% CI: -0.02 to 0.08) in the Bland-Altman analysis, and LOA were -0.71 to 0.78 mmol/L, indicating that 95% of the time, POCT measurements of non-HDL-C are expected to be between 0.71 mmol/L lower and 0.78 mmol/L higher than laboratory measurements (Figure 2).&#x20;

***Table 1: Mean, median and range of POCT and lab measurements for TC, HDL-C, LDL-C, TG, TC:HDL-C ratio and non-HDL-C.** All measurements are in mmol/L. \* POCT data on LDL cholesterol missing for 5 participants.*

<table><thead><tr><th>Variable</th><th width="123" align="center">Total Cholesterol (TC)</th><th width="123" align="center">HDL Cholesterol</th><th width="131" align="center">LDL Cholesterol (Friedewald)</th><th align="center">LDL Cholesterol (Sampson-NIH)</th><th width="135" align="center">Triglycerides (TG)</th><th width="132" align="center">TC:HDL-C ratio</th><th width="116" align="center">non-HDL-C</th></tr></thead><tbody><tr><td>N Data points</td><td align="center">260</td><td align="center">260</td><td align="center">255*</td><td align="center">260</td><td align="center">260</td><td align="center">260</td><td align="center">260</td></tr><tr><td>Mean POCT</td><td align="center">5.14</td><td align="center">1.56</td><td align="center">2.81</td><td align="center">2.91</td><td align="center">1.80</td><td align="center">3.52</td><td align="center">3.58</td></tr><tr><td>Median POCT</td><td align="center">5.04</td><td align="center">1.59</td><td align="center">2.66</td><td align="center">2.80</td><td align="center">1.44</td><td align="center">3.25</td><td align="center">3.54</td></tr><tr><td>Range POCT</td><td align="center">2.61 to 8.56</td><td align="center">0.54 to 2.59</td><td align="center">0.10 to 5.90</td><td align="center">0.53 to 5.76</td><td align="center">0.17 to 7.34</td><td align="center">1.85 to 9.79</td><td align="center">1.37 to 6.70</td></tr><tr><td>Mean Lab</td><td align="center">5.17</td><td align="center">1.56</td><td align="center">2.99</td><td align="center">3.08</td><td align="center">1.37</td><td align="center">3.53</td><td align="center">3.61</td></tr><tr><td>Median Lab</td><td align="center">5.10</td><td align="center">1.52</td><td align="center">2.90</td><td align="center">3.03</td><td align="center">1.20</td><td align="center">3.34</td><td align="center">3.53</td></tr><tr><td>Range Lab</td><td align="center">2.40 to 8.10</td><td align="center">0.68 to 3.06</td><td align="center">0.80 to 5.99</td><td align="center">1.00 to 5.59</td><td align="center">0.33 to 5.26</td><td align="center">1.78 to 7.67</td><td align="center">1.20 to 6.87</td></tr></tbody></table>

<figure><img src="/files/KRJWCQYzEmtkI4T352jc" alt=""><figcaption><p><em><strong>Figure 1. Scatter plots showing the relationship between POCT and laboratory measurements.</strong></em> A) Total Cholesterol: Strong correlation with slight proportional bias. B) HDL Cholesterol: Good agreement, minimal bias. C) LDL Cholesterol (Friedewald): Strong correlation, but systematic overestimation. D) LDL Cholesterol (Sampson-NIH): Strong correlation with improved agreement and reduced variability compared to Friedewald. E) Triglycerides: Moderate correlation, high variability. F) TC:HDL Ratio: Strong correlation, underestimation at higher ratios. G) Non-HDL Cholesterol: Excellent agreement, minimal bias.</p></figcaption></figure>

<figure><img src="/files/AJpRoWcEEWpHnPIJB1vE" alt=""><figcaption><p><em><strong>Figure 2. Bland-Altman plots.</strong></em> A) Total Cholesterol: Mean bias near zero, narrow limits of agreement. B) HDL Cholesterol: Minimal bias, consistent agreement. C) LDL Cholesterol (Friedewald): Positive bias, wider limits. D) LDL Cholesterol (Sampson): Positive bias with narrower limits and improved agreement compared to Friedewald. DE) Triglycerides: Significant negative bias, wide variability. FE) TC:HDL Ratio: Acceptable bias, but broader limits. GF) Non-HDL Cholesterol: Tight limits, excellent agreement.</p></figcaption></figure>

#### Classification performance of Mission® POCT device

For TC >5.0 mmol/L (high TC), Mission® POCT device achieved 88.6% sensitivity and 87.5% specificity (κ = 0.76), indicating good agreement with laboratory classification (Table 2). At a higher threshold of TC >7.5 mmol/L (very high TC), specificity increased (98.4%) but sensitivity decreased to 66.6% (κ = 0.56), indicating reduced ability to detect very high cholesterol values (about 3% of 260 participants had TC >7.5).

For elevated LDL-C (>3.0 mmol/L) calculated with the Mission® POCT device using the Friedewald equation, sensitivity was 79.3% and specificity 94.0%, with κ = 0.74, indicating substantial agreement (Table 2). LDL-C (>3.0 mmol/L) calculated using the Sampson-NIH equation, sensitivity was 80.5% and specificity 94.5%, with κ = 0.75.

***Table 2: Sensitivity, specificity, NPV and PPV for Mission® POCT device vs gold standard (laboratory test)***

<table><thead><tr><th>Variable</th><th width="167" align="center">Hypercholesterolaemia  (TC >5 mmol/L)</th><th width="166" align="center">Hypercholesterolaemia  (TC >7.5 mmol/L)*</th><th width="169" align="center">High LDL-C (Friedewald)**  (>3 mmol/L)</th><th width="176" align="center">High LDL-C (Sampson-NIH)  (>3 mmol/L)</th></tr></thead><tbody><tr><td>N POCT </td><td align="center">133 (51.2%) </td><td align="center">8 (3.1%) </td><td align="center">104 (40.8%) </td><td align="center">114 </td></tr><tr><td>N Lab </td><td align="center">132 (50.8%) </td><td align="center">6 (2.3%) </td><td align="center">121 (47.5%) </td><td align="center">133 </td></tr><tr><td>Sensitivity </td><td align="center">88.6% </td><td align="center">66.6% </td><td align="center">79.3% </td><td align="center">80.5% </td></tr><tr><td>Specificity </td><td align="center">87.5% </td><td align="center">98.4% </td><td align="center">94.0% </td><td align="center">94.5% </td></tr><tr><td><p>Positive predicted  </p><p>value (PPV) </p></td><td align="center">88.0% </td><td align="center">50.0% </td><td align="center">92.3% </td><td align="center">93.9% </td></tr><tr><td><p>Negative predictive  </p><p>value (NPV) </p></td><td align="center">88.2% </td><td align="center">99.2% </td><td align="center">83.4% </td><td align="center">82.2% </td></tr><tr><td>Cohen’s Kappa, k [95% CI]</td><td align="center"><mark style="color:$tint;">0.76 [0.68 to 0.84]</mark> </td><td align="center">0.56 [0.24 to 0.88] </td><td align="center">0.74 [0.66 to 0.82] </td><td align="center">0.75 [0.67 to 0.83] </td></tr></tbody></table>

*\* Less than 10 participants with TC >7.5 mmol/L and no participants with >9 mmol/L*&#x20;

*\*\* Device-derived data on LDL cholesterol missing for 5 participants*

### Conclusions and Considerations for Research using Our Future Health POCT data  <a href="#conclusions-and-considerations-for-research-using-ofh-poct-data" id="conclusions-and-considerations-for-research-using-ofh-poct-data"></a>

The study evaluated the analytical agreement and diagnostic performance of the Mission® cholesterol POCT device against a laboratory reference in 260 participants. The results of this validation study are intended to inform research using Our Future Health POCT lipid profile data.

Overall, device measurements showed strong correlations with laboratory measurements for most analytes, particularly total cholesterol, LDL-C, TC:HDL-C ratio, and non-HDL cholesterol. Total cholesterol, HDL-C, and non-HDL cholesterol showed good agreement with minimal bias, supporting their use in large-scale research settings.&#x20;

* Total cholesterol demonstrated good agreement with minimal bias and relatively narrow limits of agreement.&#x20;
* HDL cholesterol also showed good agreement with negligible bias, indicating reliable measurement.&#x20;
* The derived variable non-HDL cholesterol had the strongest agreement and appeared to be the most robust parameter.&#x20;

LDL cholesterol calculated with the Friedewald equation showed greater systematic bias and variability, which improved when using the modified Sampson-NIH calculation. These findings are consistent with the known limitations of the Friedewald equation, which relies on a fixed triglyceride-to-VLDL relationship and is more susceptible to error in the presence of TG variability.[<sup>6</sup>](#references) In this dataset, where TG measurements showed substantial variability, the Sampson-NIH method appears to provide a more robust and reliable estimate of LDL-C for research applications. Given these results, in Our Future Health data releases we have replaced the originally provided LDL-C values with LDL-C values recalculated using the Sampson–NIH method to improve robustness and reduce estimation error in research analyses.

The level of analytical agreement between POCT and laboratory measurements shown in this study reflect the benefits and challenges of measuring lipids in real world settings. Although trained clinical staff conducted the POCT lipid profile testing in accordance with [Standard Operating Procedures](/our-future-health/data-types/clinic-measurements-data/procedure-for-clinic-measurements.md#what-is-the-aim-of-a-sop) (SOPs) and the Mission® Device [user manual (external link)](https://www.manualslib.com/manual/1636526/Acon-Mission.html?page=2#manual), factors such as operator variability may contribute to measurement error. Where possible, we implemented operational measures throughout the data collection period to improve data quality and minimise bias. These included staff training, updates to SOPs, and enhanced monitoring of temperature control and storage conditions.&#x20;

Overall, this study supports the use of POCT for rapid lipid testing for research data collection. Certain measurements, particularly triglycerides and values derived from them, should be interpreted with caution. However, the Our Future Health POCT data is likely to be suitable for studies involving risk stratification and other population-level analyses of lipid profiles.&#x20;

***

### References

1. National Institute for Health and Care Research, Newcastle In Vitro Diagnostics Cooperative. *Point of care testing for cholesterol measuring: a rapid review and presentation of the scientific evidence.* 18 May 2022. <https://newcastle.mic.nihr.ac.uk/wp-content/uploads/2022/09/NIHRNewcastleMIC_AHSNLipidPOCT_Report_vFinal.pdf>.
2. Kurstjens S, Gemen E, Walk S, Njo T, Krabbe J, Gijzen K, Elisen MG, Kusters R. *Performance of commercially-available cholesterol self-tests.* Ann Clin Biochem. 2021 Jul;58(4):289-296. doi: [10.1177/0004563221992393](https://doi.org/10.1177/0004563221992393). Epub 2021 Feb 17. PMID: 33478240.
3. Friedewald WT, Levy RI, Fredrickson DS. *Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge.* Clin Chem. 1972 Jun;18(6):499-502. doi: [10.1093/clinchem/18.6.499](https://doi.org/10.1093/clinchem/18.6.499). PMID: 4337382.&#x20;
4. Sampson M, Zubiran R, Wolska A, Meeusen JW, Donato LJ, Jaffe AS, Melloni GEM, Giugliano RP, Sabatine MS, Marston NA, Remaley AT. *A Modified Sampson-NIH Equation with Improved Accuracy for Estimating Low Levels of Low-Density Lipoprotein-Cholesterol.* Clin Chem. 2025 Nov 4;71(11):1125-1137. doi: [10.1093/clinchem/hvaf073](https://doi.org/10.1093/clinchem/hvaf073). PMID: 40629956.
5. Martins J, Steyn N, Rossouw HM, Pillay TS. Best practice for LDL-cholesterol: when and how to calculate. J Clin Pathol. 2023 Mar;76(3):145-152. doi: [10.1136/jcp-2022-208480](https://doi.org/10.1136/jcp-2022-208480). Epub 2023 Jan 17. PMID: 36650044.
6. Martins J, Steyn N, Rossouw HM, Pillay TS. Best practice for LDL-cholesterol: when and how to calculate. J Clin Pathol. 2023 Mar;76(3):145-152. doi: [10.1136/jcp-2022-208480](https://doi.org/10.1136/jcp-2022-208480). Epub 2023 Jan 17. PMID: 36650044.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://ourfuturehealth.gitbook.io/our-future-health/data-types/clinic-measurements-data/poct-lipid-profile-data/poct-validation-study.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.