A pragmatic bottom-up approach to harmonize the units of clinical chemistry tests among Belgian clinical laboratories, focusing on immunoassays

Introduction

The primary goal of medical laboratories is to provide analytical results of high quality to the prescribers in order to assist medical decision making and optimal health care [1].

Unfortunately, even today, results from different laboratories are not always comparable [2]. Such comparability is a basic requisite for patient safety and has become even more compelling over the last few years due to increasing patient mobility, consolidation and networking between primary care and hospitals and the electronic exchange of patient records [3].

The importance and crucial need of harmonizing the units used for the reporting of analytical results is strongly supported by a large number of scientific publications [4]. Indeed, the use of different units has been reported as leading to misinterpretations and, thereby, to wrong treatments, adverse patient outcomes and possible increased healthcare costs [1, 5, 6, 7, 8]. Therefore, harmonization has been considered for many years as a major challenge at the international level [4, 8, 9, 10]. The harmonization includes a standardization of the assays themselves and a harmonization of the units used for reporting the analytical results and the use of common reference values [6, 11]. It improves the quality of the communication between healthcare professionals and the understanding, trueness and comparability of results [7].

The harmonization of units and, more specifically, the use of SI units are considered as a key aspect of improvement of the laboratory practice at the international level. For that reason, several decades ago, the International Union of Pure and Applied Chemistry (IUPAC), the International Federation of Clinical Chemistry (IFCC), and the World Health Organization (WHO) published several recommendations in order to encourage the medical laboratories to use the International System of Units (SI units) [7, 10, 12, 13, 14].

Similarly, some scientific journals require that authors use SI units when they are publishing their work [15, 16]. For instance, the Author center of the New England Journal of Medicine has published recommendations in order to encourage authors to “express all measurements in conventional units, with Système International (SI) units given in parentheses throughout the text. Figures and tables should use conventional units, with conversion factors given in legends or footnotes. In accordance with the Uniform Requirements, however, manuscripts containing only SI units will not be returned for that reason” [15, 17]. The use of SI units becomes thus more and more recommended.

However, despite the efforts of various international organizations, it seems that the use of SI units remains heterogeneous at the national and international levels [5, 7, 12]. For example, in Europe, the use of SI units differs significantly between countries [4]. As illustrated in Figure 1, some countries predominantly (>80%) use SI units whereas in other countries the penetration of SI units is rather moderate (10%–80%) to low (<10%). Most of the northern European countries (except Ireland) adopted SI units for expressing laboratory results, while the southern countries (Spain, Italy, Albania, Greece, Turkey, Cyprus) are still using traditional units [4]. Similarly, in the US a low use of SI units has been reported as well [16, 18].

Figure 1:

Percentage use of SI units in Europe (from Ref. [4]).

Red color, <10% use of SI units; orange color, 10%–80% use of SI units; green color, >80% use of SI units; white color, no data available.

In Belgium, the situation was rather diffuse [3]. For many parameters, SI and conventional units were used side by side, which often led to confusion when results from different laboratories were compared. As illustrated in Table 1, some parameters show marked differences in values depending on the unit that is used. For example, a free T4 result of 5 pmol/L (reference interval for adults: 9.0–26 pmol/L; [19]) indicates hypothyroidism while the same numeric value in ng/dL (reference interval for adults: 0.7–2.0 ng/dL; [19]) is indicative for hyperthyroidism.

It seems that the lack of use of SI units is related to longstanding habits and practical difficulties of implementation [10, 18, 20]. The main reasons given for refusing the use of SI units is avoiding confusion and misinterpretations of results by the physicians who were accustomed to the previously used units [9, 10]. However, as the shift to the use of SI units could be successfully performed in some European countries and that international recommendations also favor their use, it became evident for Belgian experts that an in-depth thinking and adaptation of the use of SI units by the Belgian laboratories should be performed.

As it appeared that this harmonization process can be challenging, a step-by-step awareness campaign of harmonization of measuring units was started in 2007 by the Scientific Institute of Public Health (WIV-ISP) (Brussels, Belgium) and the Belgian Commission for Clinical Biology.

In this campaign, there was no intention to impose SI units. Instead, a more pragmatic bottom-up approach involving awareness of the laboratories to the importance of harmonizing units and data collection over time in order to assess the compliance with the recommended units was chosen. The proposed harmonization process was a pragmatic mix of SI and conventional units, based on the units that were actually used by Belgian laboratories. The participation took place on a voluntary basis which stimulated an active engagement from the laboratories. It was decided not to change all parameters at once and to use a two-step harmonization campaign with a transition period. The goal was to leave sufficient time between the two phases of conversion in order to allow for familiarization with the new reporting units.

In addition to the support of the laboratories, other stakeholders were also involved in the project (e.g. the Belgian Endocrine Society) and a lot of attention was given to communication in order to inform all clinicians and not only clinical chemists. Widely accessible media, such as general medical magazines or widespread professional periodicals were used for this purpose and the topic was also introduced at symposia and discussed with practitioners on several local medical quality team meetings.

The timeline of this campaign is shown in Figure 2. The harmonization program encompassed about 140 different parameters in the field of chemistry and immunochemistry. To evaluate the impact of the harmonization campaign, we conducted a survey about the use of units in all Belgian laboratories. For this purpose, all clinical chemistry laboratories were asked to send their results of the national proficiency testing that was conducted in 2015 in a hardcopy routine analysis report. These reporting units were compared to previously collected data in 2010. This article only focuses on 23 immunoassay parameters included in the project.

Figure 2:

Time line of harmonization program for units in Belgium.

Materials and methods

Results

The participation rate in both studies was very similar. In 2010, 97.9% (185/189 laboratories) of laboratories completed the survey. In 2015, 97.1% (167/172 laboratories) sent the patient report to the WIV-ISP.

The units used in 2010 and 2015, i.e. before and after the harmonization program, are compared in Figures 3–5.

Figure 3:

Boxplots representing the percentage of use of preferred units per type of parameter and per year.

Figure 4:

Percentage of use of the preferred units for type 1 (A) and type 2 (B) parameters in 2010 and 2015.

(A) Percentage of use of the preferred units for type 1 parameters in 2010 and 2015. CA15.3, carcinoma antigen 15.3; CA 19.9, carcinoma antigen 19.9; CA 125, carcinoma antigen 125; CEA, carcinoembryonic antigen; FSH, follicle-stimulating hormone; GH, growth hormone; hCG, human chorionic gonadotropin; LH, luteinizing hormone; PTH, parathyroid hormone; PSA, prostate-specific antigen; TSH, thyroid-stimulating hormone. (B) Percentage of use of the preferred units for type 2 parameters in 2010 and 2015. AFP, alpha-fetoprotein; FT3, free triiodothyronine; FT4, free thyroxine; Vit B12, vitamin B12.

Figure 5:

Evolution in the use of units for AFP (A) and prolactin (B) between 2010 and 2015.

(A) Percentage of use of every unit used by Belgian medical laboratories in 2010 and 2015 for alpha-fetoprotein (AFP, preferred unit μg/L). (B) Percentage of use of every unit used by Belgian medical laboratories in 2010 and 2015 for prolactin (preferred unit μg/L).

Because the harmonization program occurred in different phases, we present the results by sorting the parameters into “type 1 parameters” and “type 2 parameters”.

Supplementary Table SM1 shows the data collected in 2010 and 2015 (percentage of use of each reported unit for all the 23 parameters of the EQA and the number of laboratories) for the type 1 and type 2 parameters separately.

As represented in Figure 3, the percentages of laboratories following the new recommendations sharply increased from 2010 to 2015 for all the parameters under investigation. For the type 1 parameters, a large heterogeneity in reported results was observed before the harmonization program (Figure 3, type 1 2010). A larger homogeneity and the high rate of use of the preferred unit can be observed for the 2015 survey (Figure 3, type 1 2015). The box and whisker plots in Figure 3 show that for the type 2 parameters (Figure 3 type 2 2010 and 2015), unit harmonization was lower in 2015 compared to the type 1 parameters. In 2010, the average use of the preferred unit was 25% for the type 1 parameters and 11% for the type 2 parameters. In 2015, the average of use of the preferred units was 91% for type 1 parameters and 84% for type 2 parameters.

The results are illustrated for each biological parameter in Figure 4. For type 1 parameters, Figure 4A shows that in 2010 the use of the preferred unit varied between 5% for cancer antigen (CA) 15.3 and CA 19.9 and 45% for human chorionic gonadotropin (hCG). As the use of the preferred units was extremely low in 2010 for tumor markers, the most important increase of the use of the preferred unit between 2010 and 2015 has been observed for these parameters (e.g. 89% of use of the preferred unit in 2015 for CA.15.3). In 2015, the percentage of use of the preferred unit was generally high with at least 85% for growth hormone (GH) and a maximal percentage observed for 96% for follicle-stimulating hormone (FSH) and luteinizing hormone (LH).

Figure 4B shows the percentage of use of preferred units for type 2 parameters. As for the type 1 parameters, the change to the preferred unit was well implemented with an increase in the percentage of use of the preferred units between 2010 and 2015 with a minimal increase for FT3 from 23% to 85% and a maximal increase in the use of the preferred unit for estradiol from 9% in 2010 to 93% in 2015.

In 2010, the parameter for which the smallest percentage use of the preferred unit was insulin (3% of the laboratories used pmol/L to express their results). In 2015, this percentage increased to 70%.

Among the type 2 parameters, Figure 5 shows two parameters with the largest number of users of each unit in 2010. For alpha-fetoprotein (AFP), there was a large heterogeneity in the units used in 2010. In 2015, the majority of the laboratories (77% – 87/113 laboratories) used μg/L. Similarly, four different units were used in 2010 for prolactin (11% of laboratories used μg/mL) and the percentage of use of the preferred unit (μg/L) increased in 2015 to 84%.

In order to know whether a change in the numerical value of the result influenced the laboratory to adopt the preferred unit, we took into account the use of units at a laboratory level and asked ourselves whether the change of units would mean a change in numerical value. As such, three categories were created. A first category included the cases in which the unit used in 2010 was already the preferred unit. This category was not considered in Tables 2 and 3. The second category included the cases for which a modification of the unit induced no change in the numerical value of the result for the laboratory itself and the third category included the case for which a modification of the unit induced a change in the numerical value of the result. With this division of the results we can also calculate the proportion of laboratories which changed none, some or all their units (Table 2). Note that this division is not parameter-based, as was the division in type 1 and type 2 parameters as described before.

Most of the laboratories changed their units to the preferred units. Some laboratories only partially changed the units to the preferred ones (Table 2).

There is a significant difference between the proportion of change when there is a change in the numerical value or not (p-value <0.001). Most laboratories changed their units when it induced no change in the numerical result (Table 3).

Discussion

Harmonization of units and reference intervals units has been on the agenda for more than 25 years now, and little progress has been made recently. We feel that increased mobility of patients, and the emergence of electronic health records that aggregate patient data from multiple providers might provoke a new impetus to harmonization.

Although there are several initiatives and recommendations [7, 10, 12, 13, 14, 15, 17] for harmonization of assays and reporting units, there are very few articles that report on the implementation of harmonization. Christie and Marallo implemented SI units in two hospitals in the US in 1992, with little resistance in the beginning, but they had to backtrack because medical staff voted to discontinue the use of SI units, because of problems with patient referrals and the transfer of information to physicians either unacquainted with SI units or unwilling to learn the system [22]. Our study shows that with the harmonization program set up by the WIV-ISP and the commission of clinical biology, SI units were successfully implemented for 19 parameters and a harmonization of conventional units was obtained for four parameters measured by immunoassays (Table SM1). The high compliance rate is regarded as a success.

Some factors related to the design of the awareness campaign can explain the high rate of participation of the Belgian laboratories of clinical chemistry. Among them, we strongly believe that the bottom-up strategy that was used created more good-will towards change than a top-down decision by a government agency. The fact that national proficiency testing results can now only be reported in the preferred units was an extra impetus and we think that if a large majority of the laboratories support the harmonization program, the rest of the laboratories tend to follow.

The endorsement of clinicians was also essential to overcome skepticism about the usefulness and relevance of the changes of units that might exist among prescribers. Finally, the use of a two-step process may have helped to avoid the emergence of uncertainty among the clinicians concerning the reporting units used by their laboratories and enabled clinicians to get everyone acquainted with the new reporting units. To facilitate this transition, especially for the parameters in the second phase with numeric changes, some laboratories opted for a double reporting of results in both the old and new units during a certain period.

Notwithstanding the high compliance rate that was obtained, some laboratories choose not to follow the recommendations for harmonization, or only partially made the switch. In a survey in 2015, the main reasons to remain outside the switch were fear of the reactions from the clinicians and ICT problems. Of the laboratories that did change their units, 16% received remarks from their users. Most of them were confused and had problems with the interpretation of results, due to the loss of their frame of reference. When changing units, one should not underestimate the time needed to get accustomed to the new numeric values and reference interval. To our knowledge, no laboratory changed back to the former units.

An important note to this project of harmonization is that it was tailored to the Belgian situation, where no dominance in either SI or conventional units pre-existed. When performing a similar exercise within Europe, or worldwide, it will be much more difficult to find a consensus between the SI oriented countries and the regions that still adhere to conventional units.

We do realize that this is a halfway solution, because we did not go totally for SI units, and many more efforts will be needed to go to a full harmonization in Europe and in the world. However, our approach is consistent with the recommendation of Plebani [3] that “while international consensus on this issue would be welcome, on a local basis a pragmatic solution should be sought, at least by different laboratories in the same geographical area, in order to harmonize the units of measure, and to obviate the reporting of confusing results”.

Moreover, harmonization of units is only a start on the way to full standardization of results. An International Consortium for Harmonization of Clinical Laboratory Results has been set up [6]. According to the website of the International Consortium for Harmonization of Clinical Laboratory Results (http://www.harmonization.net/measurands/; [23]), 31 parameters have been harmonized, 19 parameters are in the active process of being harmonized, eight are incompletely harmonized, and for 11, harmonization is needed.

Conclusions

About a decade ago, there was a large variation in reporting units used among the Belgian clinical laboratories. SI and conventional units were often used side by side, complicating the communication of laboratory results between healthcare providers. The Belgian Commission for Clinical Biology and the Scientific Institute of Public Health therefore took the initiative to promote the harmonization of reporting units in Belgium. A bottom-up approach was followed whereby units were chosen that were actually used by the Belgian laboratories. In case of doubt, relevant professional clinical organizations were consulted. This led to a proposal which was a pragmatic mixture of SI and conventional units. Laboratories were invited to participate in this harmonization on a voluntary basis and adjustments were grouped, depending on the impact on the numeric values, with a clear timeline for the proposed changes. An extensive information campaign was set up for the clinicians to explain the necessity for harmonization.

This strategy was successful, with a large majority of the laboratories switching to the proposed units. It is also important to emphasize that the figures from the 2015 survey are not endpoints, but only a snapshot of a trend towards harmonization which is still ongoing. Several laboratories that, for various reasons (mergers, IT restrictions, …), did not switch at the proposed time points have harmonized their units since then. This suggest that the effects of the harmonization program are still ongoing.