Study Setting: This is an experimental study conducted in a tertiary care hospital in South India. The source of the sample was mainly from outpatient departments and intensive care units.

Study duration and type: The study was conducted over 18 months with data collected prospectively.

Sampling method: All consecutive samples obtained at the lab during normal working hours was collected.

Sample size: For a confidence level of 95%, power of 90%, alpha error of 5% to detect a difference of 10% between the pre and post hemolytic sample results as significant, we needed 211 samples.

Inclusion criteria: blood samples obtained from wards, ICUs and OPDs of our hospital were included in the study

Exclusion criteria: Samples obtained from the Neonatal intensive care unit were excluded

Procedure: Blood samples were collected in citrated vacutainers containing 3.2% sodium citrate for routine coagulation tests (PT and APTT). All the citrated non-hemolysed samples received by the laboratory within 2 hours from the time of phlebotomy were included in the study. Each sample was divided into two parts. Part 1 was centrifuged, and platelet-poor plasma was extracted. Prothrombin time (PT) and Activated partial thromboplastin time (APTT) was estimated by mechanical mode on Destiny plus. Part 2 of the sample was used for experimental hemolysis. The samples were hemolysed by repeated aspiration of entire blood into a syringe with a narrow bore needle (23 G) and pushing back to the same container five times.

Platelet poor plasma was extracted by centrifuging the sample. The samples were again rerun on the same instrument for PT and APTT. Laboratory reference range and target value for PT and APTT are set in our laboratory with each lot of the reagent by running the tests on an equal number of voluntary healthy males and females. The mean +/- 2 standard deviations is our laboratory reference range for that lot of the reagent. During this study, these values were 12.5 to 17 seconds for PT and 27-37 seconds for APTT. Optical mode on Destiny plus was used to obtain results via optical principles.

7.8% and 18%. One hundred and seventy-four (70.2%) samples had normal prothrombin values, and 137 (57.5%) had normal APTT values. Coagulation tests on optical principle in the post haemolysis sample were largely unsuccessful, with only 24 (9.6%) samples yielding results. Hence the comparison of post haemolytic samples was made on values obtained by the mechanical method alone. The results [Median, (IQR)] of mechanical and optical principles of coagulation tests in the pre-hemolysed samples yielded, Prothrombin time (in seconds) of 15.1(13.9–18.7) and 15.2(14–18.6), INR of 1.08(0.97–1.4) and 1.09(0.98–1.39), APTT (in seconds) of 32.1(29–37.1) and 32.3(28.9–35.9) respectively. The difference was not significant. The mean difference between the pre and post hemolytic samples for prothrombin time was 1.2 seconds, with lower pre hemolytic values. In the group who had normal PT (<17 seconds), the mean difference between pre and post hemolytic samples was 0.9 seconds (95% limit of agreement -3.1 to +1.4 seconds). For the samples which had abnormal PT (> 17 seconds), the mean difference was 1.1 (95% limit of agreement -7.2 to +5) seconds. The mean difference between PT values for both normal and abnormal groups was not statistically significant. INR followed a similar trend. The mean difference between the pre and post hemolytic samples for APTT was 5 seconds, with lower pre hemolytic values. In the group who had normal APTT (<37 seconds), the mean difference between Pre and post hemolytic samples was 4.9 seconds (95% limit of agreement -1 to +10.9 seconds). For the samples which had abnormal APTT (> 37 seconds), the mean difference was 1.1 (95% limit of agreement -14.8 to +4.7) seconds. The mean difference between normal and abnormal APTT groups was statistically significant (Table 1).

Figure 1: Bland Altman plots for agreement of pre and post hemolysis parameters

In our pilot study, we found the variation in the results of both PT and APTT when performed repeatedly from the same sample was minimal and did not exceed 10%. Hence, we estimated the proportions of samples in this study with more than 10% variation post hemolysis. Twenty-four percent of the Normal PT samples,4% of samples with abnormal PT, had a variation of more than 10% from baseline value post hemolysis. Seventy percent of samples with normal APTT and 3% with abnormal APTT had more than 10% variation post hemolysis 85.7% of PT values before hemolysis remained normal, and 14.3% reported abnormal results after hemolysis.

95% of abnormal PT values before hemolysis remained abnormal, and only 5% reported normal PT values after hemolysis. The results with APTT were not so encouraging. Only 64.1% of samples that had normal APTT before hemolysis remained normal. Conversely, 95% of samples that had abnormal APTT before hemolysis remained abnormal post hemolysis as well.

Our study shows that the coagulation results of hemolysed samples vary significantly compared to non-hemolytic samples and hence remain inferior to proceed with coagulation tests. The optical principle of coagulation testing is not suitable for hemolysed samples. Results obtained by mechanical principle should be used with utmost caution in case of emergency. The mean values of PT, pre and post hemolysis have shown minimal difference, which is statistically insignificant. In a study like the present one, Arora et al. [9]. have compared the summarised values of pre and post hemolytic samples, which have yielded similar results as ours.

However, in this scenario, we believe a simple comparison of summarised data results in gross generalization. The sample size in the above study is too small to make these results usable. Nevertheless, they have shown a direction to proceed with hemolytic samples. The positive predictive values for PT and APTT are relatively high, making it reasonable to interpret normal values. The clinician may use a normal value on the hemolysed sample to make appropriate management decisions with a risk of being wrong is less than 5%.

Our study shows that a prolonged APTT due to basic pathology is less likely to produce normal results if hemolysed. In summary, a normal coagulation test results in a hemolysed sample more likely to indicate normal coagulation. Very much prolonged coagulation results in the hemolysed sample, less likely to show normal coagulation.

A result in the upper limit of normal and not very prolonged is equivocal and hence not interpretable limitation of the study. The in-vitro haemolytic procedure we have adopted have resulted in significant hemolysis, which might be way higher than seen in a real-life scenario.

Hemolytic samples will interfere to a variable extent in coagulation studies. The optical principle is unsuitable for processing these samples. Prothrombin time is less affected when compared to activated partial thromboplastin time. The influence of haemolysis on prothrombin time is minimal and still makes it clinically usable.

However, in cases of deranged coagulation, the prothrombin time should be cautiously interpreted. Activated partial thromboplastin time is strongly influenced by haemolysis. In cases of prolonged coagulation, APTT values of hemolysed samples could be still representative of actual coagulation status.

In precarious situations where the patient is extremely sick with prolonged coagulation and repeat sample is difficult to obtain, this information could be helpful.

Dr VK constructed the research question, did the study design and data analysis. Dr RA collected the data, did the literature search, and prepared the initial manuscript. Both authors contributed to finetuning the manuscript and approved it in its present form for submission

In the case of hemolysed samples, the mechanical method of coagulation analysis is suitable for processing the sample. Prothrombin time obtained by this method is clinically usable in precarious situations. Activated partial thromboplastin time is not to be reported in this sample.