Patient risk management in laboratory medicine: an international survey to assess the severity of harm associated with erroneous reported results

Introduction

All international quality standards for medical laboratories have updated their recommendations and requirements in recent years, focusing the quality management process on patient risk. ISO/IEC Guide 51:2014 defines risk as the “combination of the probability of occurrence of harm and the severity of that harm” [1]. The general concept of risk can be expressed as the probability of harm=probability of occurrence x the severity of harm [2]. In the clinical laboratory, risk analysis and risk management are focused on the risk of patient harm from erroneous and delayed results. The probability of occurrence is the probability of producing an erroneous or incorrect result. With an erroneous result being defined as a patient sample measurement that contains measurement error in excess of the quality specification for the analyte being measured. Erroneous results can be produced when the test method is “in control” state or “out of control” state of the test method. The probability of producing erroneous results is the sum of the probabilities of producing erroneous results when the test method is “in control” and “out of control” [3]. The severity of harm (SOH) in this context is the consequence to the patient when an erroneous result is produced [4]. As much as the occurrence can be estimated objectively with quantitative data, SOH determination is by nature subjective. Unfortunately, there is actually no guidance to estimate SOH except recommendation for using a qualitative five level scale of harm [4], 5] (Table 1). Some descriptive studies propose to evaluate the SOH caused by lab errors from declarative failure reports. Nevertheless, they do not provide any information on the specific criticality of each examination which can considerably differ from one analysis to another [6], 7]. We propose here, an international survey approach to estimate objectively and specifically the SOH in case of erroneous reported results for a bundle of 20 routine analysis. The aim of this study is to propose some guidance for SOH assignments that can be used by the clinical diagnostic laboratory community to improve compliance with ISO 15189:2022 requirements.

Materials and methods

Results

Severity of harm survey results

Severity of harm level distribution

A total of 267 clinical laboratory professionals categorized for 20 analytes the amount of harm that would be potentially caused if a patient were to be mistreated based on an incorrect laboratory result. We observed a 99.1 % overall response rate for the SOH categorization part of the survey. The most chosen category is serious (33.9 %) followed by critical (30.1 %), catastrophic (16.3 %), minor (16.2 %), and to a much lesser extent the negligible category (2.7 %). The 20 analytes were analyzed individually to establish the SOH category distribution (Table 2). The SOH categorization results are also presented in a Likert plot (Figure 1) which can be used to classify the analytes from the potentially most harmful to the most harmless. We observed that troponin is considered by the responders to be the most harmful analyte; on the opposite blood iron is considered to be the most harmless in case of wrong results.

Table 2:

Distribution of severity of harm category by analyte (in percentage).

Analyte	Severity of harm category, %
	No answer	Negligible	Minor	Serious	Critical	Catastrophic
Albumin	1.12	4.49	42.7	38.2	11.99	1.5
ALT	0.4	2.6	33.7	48.7	12	2.6
CA 19-9	1.5	1.87	14.23	40.82	28.09	13.48
Creatinine	0	0	9	43.4	41.9	5.6
D-dimer	1.5	0.4	6.7	25.8	43.8	21.7
Digoxin	3	0.7	3	19.9	40.4	33
Fibrinogen	2.2	0.4	12.4	42.7	33.7	8.6
Glucose	0	0.4	5.6	31.8	40.4	21.7
Haemoglobin	0.37	0.75	4.12	25.09	38.58	31.09
hCG total	1.5	3.7	11.6	35.6	32.6	15
Iron	1.12	16.48	46.82	27.34	7.87	0.37
Lipase	2.25	3	18.73	45.32	25.84	4.87
Phosphorus	1.5	5.24	37.83	37.45	17.6	0.37
Platelet	0.7	0	3.7	27	42.7	25.8
Potassium	0	0	1	16	35	48
Protein (total)	1.9	7.5	37.1	41.2	12	0.4
Prothrombin time	1.5	0	6.4	28.5	41.6	22.1
Sodium	1.1	1.1	9.4	42.7	33.3	12.4
Troponin	0.04	0	0	10	32	57
TSH	1.5	2.25	20.22	49.81	22.47	3.75

1. The survey respondent’s category designation is presented in percentage in alphabetical order. The specific matrix of each analyte is serum or plasma for albumin, ALT, CA 19.9, creatinine, digoxin, glucose, hCG, total, iron, lipase, phosphorus, potassium, protein (total), sodium, troponin and TSH; whole blood for haemoglobin and platelet and at last plasma for D-dimer, fibrinogen and prothrombin time.

Figure 1:

Likert plot representation of survey results of severity of harm categorization. The likert plot centered bars with splits the serious category. Analytes are ordered by the combined critical and catastrophic category. The “no answer” category was removed and not taking into account.

Severity of harm category attribution

To propose guidance for SOH determination, we decided to assign a unique SOH category for each parameter named the “Prevailing Severity of Harm Category”. This main assignment is designed to help clinical laboratory professionals to associate a ranked level of damage with a clinical analysis mistake. By default, the modal category of the survey is assigned to each parameter. In some cases, it could be irrelevant when the modal category is very close to another one. To remedy this lack of significance we choose to assign the highest rank of SOH when the difference between the modal and the second category is less than 5 %. It concerns 5 parameters and four were reclassified in the upper category (Albumin, Creatinine, Phosphorus and Protein Total). After reclassification two parameters were associated with a catastrophic SOH, eight were considered critical, nine as serious, only one as minor and no one was assigned to the negligible category of SOH. The Prevailing SOH Categories are presented for each tested parameter in Table 3 (modal categories are available in the supplementary materials in Supplementary Table S1).

Table 3:

Prevailing severity of harm category for the survey parameters.

Analyte	Prevailing severity of harm category
Potassium (serum/plasma)	Catastrophic
Troponin (serum/plasma)	Catastrophic
Creatinine (serum/plasma)	Critical
D-dimer (plasma)	Critical
Digoxin (serum/plasma)	Critical
Glucose (serum/plasma)	Critical
Haemoglobin (whole blood)	Critical
hCG total (serum/plasma)	Critical
Platelet (whole blood)	Critical
Prothrombin time (plasma)	Critical
Albumin (serum/plasma)	Serious
ALT (serum/plasma)	Serious
CA 19-9 (serum/plasma)	Serious
Fibrinogen (plasma)	Serious
Lipase (serum/plasma)	Serious
Phosphorus (serum/plasma)	Serious
Protein (total) (serum/plasma)	Serious
Sodium (serum/plasma)	Serious
TSH (serum/plasma)	Serious
Iron (serum/plasma)	Minor

Discussion

Risk management in medical laboratories has become a central part of the requirements for quality and competence in the new ISO 15189:2022 [10] which is in alignment with the ISO 22367:2020 specifically devoted to this topic [5]. In ISO 22367:2020 patient harm is specifically addressed in the standard Section 5.7 and states that “Reasonably foreseeable harms that could result from each hazardous situation shall be identified and classified along with the severity of each harm. This process and the identified harms, shall be documented.” [5]. Although this ISO standard was elaborated for medical laboratories, there are actually few literature or consensus guidelines to document patient harm in a risk analysis process. With this study we propose to specifically estimate with an international survey the SOH from erroneous lab results for a panel of 20 analytes.

As recommended in Annex I.4 of ISO 22367:2020 we use an ordinal variable with concise definitions without associated probability elements [5]. Consequently, our five level SOH scale fit with actual requirements and our results can be used in risk assessment process as defined by Janssens et al. [11]. The key point of this study is to define a SOH category which is independent of any clinical context. It cannot be denied that clinical context has major influence on patients’ harm from incorrect lab results. However, the variety of situations encountered for an analyte makes it impossible to consider in a lab risk management system. In addition, the clinical context is not always known by the laboratory. Consequently, in order to provide a usable laboratory tool, we apply CLSI EP23^2nd recommendations which state that the most harmful scenario had to be assumed for determining SOH when key considerations are unable [4]. As only lab professionals had a complete overview of their activity, we decided to address the survey to laboratory professionals rather than clinicians. This recruitment bias is assumed and makes the proposed SOH classification suitable for medical laboratory use.

A few studies are available to evaluate SOH induced by lab errors in real medical situations. Nevertheless they all indicate that most lab failures do not lead to patient harm which is in contrast to our survey results [6], 7]. Indeed, almost all surveyed parameters (19/20) were associated with at least a serious SOH category which is defined by “injury or impairment requiring professional medical intervention”. This should be explained that for real failure, laboratory errors are detected before they reach the patient. It is supported by the fact that postanalytical events, which are more likely to reach patients, contain the highest ratio of severe harm [7], 12]. It is also supported by O’Kane’s publication who ranked both real and potential patient outcomes for 658 quality failure lab reports. It reveals that despite any significant adverse outcomes, most events (67.9 %) were graded at the maximum rank of damage if the potential worst-case was considered [13]. These findings clarify our results and explain the high rate of serious and critical SOH categories attributed to our 20 evaluated parameters. Our results are also consistent with two recent studies which also used a survey approach with a five level scale to estimate the SOH from laboratory errors [14], 15]. Despite some differences with tested analytes, responder’s demography and numbers, they also found that the majority of asked parameters were associated with at least a serious SOH category (55.4 % [14] and 94.8 % [15]).

The main limitation of the current study is the lack of representativeness of the survey responders. This lack of diversity is particularly important from a geographical point of view. Despite a wide range of represented countries (n=43), four countries are largely over-represented and totalized 46.4 % of the participants (France, Italy, Germany, Spain). We also noted that there were virtually no representatives from Asia (2.3 %), the entire American continent (1.5 %) and no one from Australia. Moreover, the European countries totalized 68.2 % of the responder’s origin. This over-representation of the EU is due to the fact that the survey was launched by the European Bio-Rad^® team and therefore was well promoted during congresses or virtual events across Europe. Moving forward, additional efforts in promotion need to take place trying to improve uniformity of the response rate globally.

Hazardous situations occur to patients when an incorrect result is reported to a clinician or when a critical result is delayed [5]. By asking “How would you rate the severity of patient harm if a clinician would interpret incorrect reported results from these analytes?” our SOH evaluation cannot assess harm caused by delayed results. This limitation should be coped in further survey investigations by adding question(s) which specifically deal with this topic. However, our results can document the consequences of laboratory failure for many steps of the total test process and could help to identify the most critical ones. As requested by the ISO 15189:2022 standard Section 7.1 our proposed SOH assignment can provide guidance for laboratories to monitor and evaluate (for a specific test) the identified risk and effectiveness of mitigation process according to the potential harm to the patient [10].

The prevailing SOH category classification can also be used for internal quality control (IQC) establishment strategy by applying the model presented in CLSI document EP23^2nd [4]. This model proposed to map a SOH category to a corresponding maximum acceptable probability of patient harm from incorrect results. The laboratory can then design a quality control strategy that have a predicted probability of patient harm below this maximum. The use of a such model permits laboratories to satisfy the ISO 15189:2022 subclause 7.3.7.2 which states that IQC shall be performed at a frequency that is based on the stability and robustness of the examination method and the risk of harm to the patient from an erroneous result [10].

Conclusions

To the best of our knowledge, this is the first international attempt to investigate how the laboratory community thinks about patient harm from an erroneous reported result. While severity of harm designations should be specific to the local clinical system, guidance can be provided by collecting the opinion of a wide variety of clinical laboratory professionals. This proposed SOH category evaluation constitutes a valuable tool for laboratories to document their risk management policy and responds to a need prompted by the normative evolution of all international laboratory quality standards. Nevertheless, this approach needs to be extended to a much more exhaustive set of parameters and matrix with more participants to become routinely usable and recognized by the scientific community of clinical diagnostic laboratories. During the last decade this community was deeply involved in elaborating quality measurement tools as Quality Indicators [16] to reduce error rate and enhance patient safety: this study contribute to the global effort and give some guidance to improve risk assessment standardization in laboratory medicine.