The effect of automated hemolysis index measurement on sample and test rejection rates

Introduction

Laboratory test results affect clinical decision-making, patient safety and outcomes [1]. Therefore, data given by laboratories have to be accurate, reliable, cost-effective and time-saving. Preanalytical errors account for up to 60% of total testing process and among preanalytical errors hemolyzed samples are the most frequent ones, approximately 40–70% of the unsuitable samples [2], [3], [4]. Hemolysis is pathological break down of red blood cells (RBC’s) and the release of hemoglobin and other components of RBC’s into serum or plasma, which can occur either in vitro or in vivo [5]. In vitro hemolysis may be result of inappropriate blood collection, specimen handling, processing, storage and transportation [6] and referred as a preanalytical quality marker [7].

Hemolysis can interfere many laboratory tests and alter test results via biological, chemical or spectrophotometric interference [8], [9]. Therefore, it is essential to observe and identify hemolyzed samples for accurate results which are traditionally noticed visually by laboratory staff after centrifugation [10]. But many studies revealed that visual determination of hemolysis is unreliable, subjective and highly variable according to laboratory staff judgment. Automated measurement of serum indices are one of the technological interventions introduced to laboratory use by manufacturers. Serum indices also called as hemolysis, icterus and lipemia (HIL) indices on clinical chemistry analyzers measures free hemoglobin, bilirubin and lipemia semi-quantitative or quantitatively. The calculation of HIL index is easy, cost effective, fast and reliable especially identifying hemolysis [11], [12], [13], [14]. Thus, automated integration of HIL parameters are recommended to laboratories by Clinical and Laboratory Standards Institute in CLSI C56A guideline [15].

In this study we intended to investigate the impact of hemolysis index implementation on sample rejection rates and suppressed test results in current clinical chemistry analyzers.

Materials and methods

This retrospective study was performed in Antalya Public Health Care Laboratory which serves to family practitioners in Antalya in 194 distinct centers. As a routine procedure samples are collected by family health staff into 5 mL serum gel separator tubes with clot activator (BD Vacutainer; Becton Dickinson and Co., Franklin Lakes, NJ, USA). Family Health Centers (FHC) in Antalya province can be categorized by means of location as central and district FHCs. Samples are collected by family health staff between 8:30 and 10:30 in all FHCs, centrifuged in the FHC and collected by couriers once a day to the Health Care Laboratory. Delivery time of samples from central FHCs to laboratory after sample collection ranges from 15 min (as the nearest FHC is in the same building with laboratory) to 120 min (furthest FHC is 65 km to laboratory) with a mean of 90 min and 35 km. From district FHCs delivery time ranges ranges from 60 min (the nearest FHC is 30 km away from the laboratory) to 240 min (furthest FHC is 270 km away from the laboratory) with a mean of 180 min and 120 km. So, preanalytical variabilities are highly difficult to control in public health care laboratories when compared with a hospital laboratory.

Being aware problems of visual determination of serum indices, in June 2014 automated serum index was adapted to clinical chemistry analyzers Beckman Coulter AU2700 (Beckman Coulter, Brea, CA, USA). Analyzers use HIL reagent to measure interfering substances with spectrophotometric measurement which are carried out at following wavelengths; lipemic judgment at 660 and 800 nm, icterus judgment at 480, 570, 600 and 800 nm and hemolytic judgment at 410, 480, 600 and 800 nm. By using formulas analyzer calculates interfering substance semi-quantitatively into one of these six categories; normal, one positive, two positive, three positive, four positive and five positive. Table 1 summarizes the warning flags of analyzer and their approximate concentration given by manufacturer information sheet.

Warning flags derived from analyzer were integrated to Laboratory Information System (LIS). Results of serum indices could be seen in a separate column among patient’s test results both by laboratory technicians and clinical biochemistry specialists.

Twenty-eight routine test are performed on clinical chemistry analyzers; alanine aminotransferaz (ALT), albumin, alkaline phosphatase, amylase, Antistreptolysin O, aspartat aminotransferase (AST), total bilirubin, direct bilirubin, blood urea nitrogen, calcium, cholesterol, chloride, C-reactive protein, creatine kinase, creatinine, gamma glutamyltransferase, glucose (Glu), HDL cholesterol, iron, iron binding capacity, lactate dehydrogenase, LDL cholesterol, magnesium, phosphate, potassium, rheumatoid factor, total protein, triglyceride, sodium, and uric acid.

According to the information sheet given by the manufacturer, tests influenced by hemolysis at different hemolysis levels are given in Table 2. Test results affected by hemolysis are rejected in agreement with the information sheet by clinical biochemistry specialists.

In this study we evaluated data from LIS in terms of sample rejection rates and rejected test results 01 June–31 August 2013 which is before adapting hemolysis index and after 01 June–31 August 2014. Statistical analyses were performed using OpenEpi version 3.01 (www.openepi.com). In order to compare proportional differences between two periods chi-square test was used and p<0.05 considered as statistically significant.

Results

After implementation of HIL indices detection of hemolyzed sample increased by five-fold compared with visual detection. We found rejected sample rate increased (p<0.001) but rejected test rate decreased (p<0.001) (Table 3).

Within a total of 1480 rejected sample the most frequent group was found to be Hem 1 with 75.74% and rejected test rate 72.33% (Table 4).

Most of the patients were female (65.61%) and when analyzed by means of age the most rejected samples belonged to 19–65 years population (Table 5).

The most rejected test in Hem 1 and Hem 2 group was AST, Hem 3 group AST and ALT, Hem 4 group Glu (Table 6).

Discussion

In this study our aim was to evaluate the effect of automated HIL usage on rejected sample rate and rejected test rate. Our results revealed that rejected sample rate was increased but rejected test rate was decreased.

Sample quality is one of the major quality indicators in laboratory diagnostics and for precise measurements factors affecting results, patient safety and clinical decision-making have to be identified and managed. Hemolysis, one of the most rejection motives, is recommended to be determined by automated analyzers because visual determination is considered unreliable and subjective [10], [16]. Automated chemistry analyzers can detect hemolysis with increased productivity, high throughput and decreased error rates. Also, it improves detection of mildly hemolyzed samples which have free hemoglobin levels of 30–60 mg/dL or cannot be identified visually or underestimated by laboratory staff [17]. So, incorrect results especially rising from mild hemolysis can damage patient safety.

In our study after introduction of automated serum indices to analyzers, rejected samples due to hemolysis raised significantly when compared the same period 1 year ago. Vermeer et al. reported that after introduction of serum indices for hemolysis detection in spite of visual determination hemolyzed sample (>0.5 g/L) were found to be six-fold increased [18]. Also in another study, Tamimi et al. [19] reported that after implementation of serum indices sample rejection rates were raised to 0.21% from 0.13% and the most significant sample rejection reason was hemolysis. They concluded that the increase in sample rejection rates assisted by automated analytical systems improved staff’s awareness. These findings were showed consistency with our data but none of these studies categorized hemolysis degree or rejected test profile.

In a study performed by Adiga and Yogish [20] they found that majority of the hemolyzed samples determined by automated analyzer had small to intermediate degree of hemolysis and reported that hemolysis index estimation defeats limitations of visual assessment and provide more unbiased and exact results. In our study we found 75.74% of hemolyzed samples were in Hem +1 group which reflects 50–99 md/dL hemoglobin presence was underestimated or could not be determined by laboratory staff with naked eye.

In our study rejected test rate was significantly lower after introduction of serum indices to laboratory than visual inspection period. This may be a consequence of test based rejection according to manufacturers’ suggestions in automation period. But in visual determination period all tests requested in the same sample were rejected regardless whether or not influenced test existed. Among rejected tests according to manufacturers’ suggestions, the most rejected test in Hem 1 and Hem 2 group was AST, Hem 3 group AST and ALT, Hem 4 group Glu. In the literature we could not find any data about sample rejection rates or test distribution after introducing HIL indices to analyzers. Further comparison studies are needed to see the effect of automated HIL indices usage on test rejection rates and rejected parameters.

In a study published by Söderberg et al. [21] which consisted primary health care centers, nursing homes and emergency departments analyzed hemolysis prevalence and age and gender distribution. They revealed that patients over median age had higher frequency of hemolysis which may be a consequence of difficulties at venipuncture in aged population. They also found in primary heath care centers women had higher frequency of hemolyzed sample. In our study samples from patients between 19 and 65 years old women had higher prevalence of hemolysis. Age and sex distribution analysis of total accepted samples to laboratory showed consistency with our finding, as the majority of patients were women between 19 and 65 years old.

Finally, this study has some limitations. First of all, in this retrospective study we were not able to find data about scale of hemolysis in visual evaluation period and rejected parameters prevalence cause of recording failures. Secondly, studies suggest laboratories should verify hemolysis effect on parameters provided by manufacturer [22]. In our study we relied on manufacturers’ documents cause of resource limitations.

In conclusion, to provide accurate test results and patient safety, interfering substances should be detected on chemistry analyzers by HIL indices which is fast and highly reliable.

The authors declare that there is no conflict of interests regarding the publication of this article.