Product recall: a Croatian experience (2000–2010)

Abstract

Background

Timely and efficient recall of products known or suspected to be non-conforming is an important measure in the prevention of adverse events and in patients' safety. Product recall in the transfusion service is regulated by professional standards and legal acts, but publications presenting results related to the implementation of these procedures are quite rare.

Materials and methods

Data from the Croatian Institute of Transfusion Medicine (CITM) on the procedures of product recall during an 11-year period (2000–2010) were retrospectively analyzed. Reasons for product recall, their frequency, level of severity and efficiency of the procedures are presented and discussed.

Results

During the study period, there were 245 procedures of product recall, for an average of 22 (18–29) procedures/year, all of low extent (1–25 products). Recall was required for 1/3,571 blood products issued, while the frequency of laboratory test report recalls was 1/5,447 patients. The leading reasons for product recall were suspected bacterial contamination of blood products (30.2%) and suspected or demonstrated non-conformity of laboratory test reports (28.6%). In total, 99 (40.4%) product recalls were categorized as class I, 30 (12.2%) as class II and 116 (47.3%) as class III.

Discussion

According to the available literature data, the product recall procedures were performed quite infrequently by the CITM and were of low extent. There was a remarkable decreasing trend in the rate of product recall due to non-conformities or errors made at the CITM, along with a constant or increasing rate of recalls because of biological variability of blood products.

Introduction

Timely and efficient recall of products known or suspected to be non-conforming is an important measure in the prevention of adverse events and in patients' safety. In transfusion medicine, these procedures are defined by professional standards and legal provisions, which is no surprise considering that product recall is relatively common in this segment of medical science. It is so primarily because of the biological origin of blood products and thus of the specific risks associated with their use. Not all the risks of transfusion treatment can be predicted and some of them cannot even be prevented. It is, therefore, necessary to establish an efficient system of quality management in transfusion services, including systematic surveillance of the overall activities in the whole transfusion chain (haemovigilance). Such a system enables timely detection of any deviation from the specific requirements and implementation of corrective actions, including product recall. At the Croatian Institute of Transfusion Medicine (CITM), product recall is performed in line with legal provisions and written procedures as part of the established quality management system. The results of 11 years of experience in product recall are presented here to enable comparison with other transfusion services and to re-evaluate our practice to date by identifying potential drawbacks and weaknesses for further improvements.

Materials and methods

Data from 11-year reports on product recall of the CITM Quality Assurance Department were used in the study. The procedures of product recall were analysed according to their number, efficiency and reasons for recall. The trends observed were analysed and both favourable experiences and shortcomings of the system during the study period are discussed. The procedure methodology is described in detail by the standard operating procedure (SOP) at the CITM, and is based on national legislation and professional standards. The main determinants of the procedure performance at our institution are shown in a flowchart (Figure 1).

Figure 1.

Flowchart of product recall activities at the CITM.

At the CITM, a product is defined as a tangible material good as well as any process result including provision of various services. This means that besides blood products, laboratory test reports are also considered as products. The recall procedure is undertaken for the products that have been distributed or issued by the CITM, and which are suspected or definitely known to be non-conforming. A decision on product recall can refer to the products manufactured from a single donation or to all products that have been manufactured from a particular donor's blood and are still in date.

The classification of criticality of the product recall procedures is based on the Food and Drug Administration (FDA) requirements and has three grades¹:

• - class 1: justified suspicion that the use of the product might result in a fatal outcome or another very severe consequence;

• - class 2: the product non-conformity might result in some reversible health consequences, but the probability of a fatal outcome or severe consequences is very low;

• - class 3: the product non-conformity is not likely to induce any health damage; however, it cannot be excluded.

The levels of criticality were determined when deciding on product recall, not upon testing completion, thus the level of criticality referred to the potential rather than actual (demonstrated) risk.

Statistical analysis

The test of proportion differences was used for determining the statistical significance of differences between study parameters in different periods; differences yielding P values less than 0.05 were considered statistically significant. Kendal's tau non-parametric test was used to analyse trends throughout the study period. A positive or negative trend was considered statistically significant at P <0.05.

Results

During the 11-year period, 245 product recall procedures were undertaken, for an average of 22 procedures/year. The number of products recalled in each procedure ranged from 1 to 25 (mean 2.6); thus, all the procedures were of minor extent. Out of 643 products recalled, 396 products were returned, yielding a 62% efficiency of the procedures (37–96% per year). Blood products intended for clinical use accounted for the highest proportion of products recalled (n =498; 77.4%), followed by laboratory test reports (n =103; 16%) and blood products issued for fractionation (n =42; 6.5%). Of 245 recall procedures, 41 (16.7%) were initiated because of a suspected transfusion reaction (22 suspected bacterial contamination, 12 suspected transmission of hepatitis, 5 suspected transfusion-related acute lung injury [TRALI], and 2 reactions of unknown cause).

During the study period, 1,928,586 blood products were issued in total, of which 1,585,700 were for clinical use. Thus, recall was required for 1/3,571 of all blood products and 1/3,184 of blood products for clinical use. The number of patients was used as a denominator for monitoring recalls of laboratory test reports. During the 11-year study period, a testing service was provided for 555,600 patients, with a frequency of test report recalls of 1/5,447 patients.

A total of 99 (40.4%) product recalls were classified as class I, 30 (12.2%) as class II, and 116 (47.3%) as class III. Out of 245 recall procedures, 109 (44.5%) were undertaken as a consequence of customer complaints.

The reasons for launching the procedure of product recall are summarised in Table I. As seen in Table I, a suspicion of bacterial contamination of a blood product and demonstrated or suspected non-conformity of laboratory test finding reports were the predominant reasons for product recall.

Table I.

Product recall in 2000–2010: distribution according to reasons.

Reason for recall: demonstrated or potential	n	%
BP bacterial contamination	74	30.2
Test report non-conformity	70	28.6
Donor risk (post-donation information)	21	8.6
Error on BP issuing	20	8.2
Non-conforming BP label	17	6.9
Transfusion-transmitted viral infection	11	4.5
Presence of irregular antibodies	11	4.5
Error in laboratory testing	5	2.0
Inadequate BP quality	5	2.0
Other	11	4.5
Total	245	100.0

Legend BP: blood product.

The pattern of product recall procedures during the study period is shown in Table II, and results of trend analysis in Table III. Blood product recall is expressed as the number of procedures per 10,000 blood products issued, and laboratory test report recall as the number of procedures per 10,000 patients.

Table II.

Frequency of product recall procedures in 2000–2010.

Reason for recall	Year
	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010
BP bacterial contamination	0.51	0.26	0.40	0.52	0.24	0.35	0.48	0.19	0.37	0.60	0.28
Test report nonconformity	0.00	1.58	0.91	1.26	2.10	1.61	1.20	2.00	0.89	1.21	0.89
Donor risk	0.13	0.26	0.17	0.12	0.06	0.17	0.18	0.00	0.00	0.10	0.05
Error on BP issuing	0.19	0.32	0.11	0.35	0.06	0.00	0.00	0.00	0.11	0.05	0.00
Non-conforming BP label	0.06	0.06	0.23	0.12	0.00	0.29	0.00	0.06	0.05	0.10	0.00
Transfusion-transmitted viral infection	0.19	0.00	0.00	0.00	0.00	0.00	0.00	0.12	0.11	0.05	0.14
Presence of irregular antibodies	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.19	0.27	0.10	0.05
Error in laboratory testing*	0.06	0.06	0.00	0.06	0.06	0.06	0.00	0.00	0.00	0.00	0.00
Inadequate BP quality	0.06	0.13	0.00	0.00	0.00	0.06	0.00	0.00	0.00	0.05	0.00
Other	0.00	0.06	0.11	0.12	0.06	0.00	0.12	0.00	0.00	0.05	0.09

Legend

^*Errors in laboratory testing other than test report non-conformity;

BP: blood product.

Table III.

Results of trend analysis of product recall in 2000–2010.

Reason for recall	P	T	95% CI
BP bacterial contamination	0.755	−0.055	−0.520 to 0.480
Test report non-conformity	0.8065	−0.038	−0.602 to 0.590
Donor risk	0.057	−0.426	−0.766 to 0.02
Error on BP issuing	0.024	−0.506	−0.869 to −0.178
Non-conforming BP label	0.322	−0.212	−0.611 to 0.307
Transfusion-transmitted viral infection	0.235	0.298	−0.351 to 0.817
Presence of irregular antibodies	0.023	0.555	0.319 to 0.744
Error in laboratory testing*	0.008	−0.591	−0.799 to −0.155
Inadequate BP quality	0.159	−0.305	−0.688 to 0.377
Other	0.866	−0.0197	− 0.488 to 0.463

Legend

^*Errors in laboratory testing other than test report non-conformity;

BP: blood product.

The results presented in Tables II and III clearly show a decreasing trend in the frequency of product recalls in all categories except for product recalls due to a suspected presence of irregular anti-erythrocyte or anti-leucocyte antibodies in donor blood and due to suspected transfusion-transmitted viral infection. However, a statistically significant decline was only achieved in two categories (error on blood product issuing and error in laboratory testing), and borderline significance in the category of product recall because of post-donation information on possible donor risk.

In order to get a more detailed insight into the pattern of product recalls, their frequency was compared between two periods (2000–2005 vs 2006–2010). These two periods were chosen for comparison not only for their comparable length but also because a number of measures and procedures in blood product preparation and testing were introduced at the CITM in 2005, which had a favourable impact on the product and service quality and safety. Table IV shows a comparison of the frequency of product recalls and recall procedures between these two periods. There was no statistically significant difference in blood product recalls between the two study periods, while statistical significance was only achieved for laboratory test reports when comparing the frequency of reports recalled. This is so because a large number of test reports (n=25) were recalled in a single recall procedure.

Table IV.

Comparison of product recall frequency between two study periods (2000–2005 vs 2006–2010).

Period	Type of product	Products issued (n)	Products recalled			Recall procedures
			n	%	P	N	1/products issued	P
2000–2005	Blood products	998,188	282	0.028	0.970	101	1/9,883	0.154
2006–2010		930,398	258	0.028		74	1/12,573
2000–2005	Test reports	280,219	65	0.023	0.008	36	1/7,784	0.963
2006–2010		275,381	37	0.013		34	1/8,099

In order to assess the trends of blood product recalls according to the reasons for recall, the frequency of blood product units recalled and blood product recall procedures was compared for each particular category in two periods (2000–2005 vs 2006–2010). The results are shown in Table V.

Table V.

Comparison of blood product recalls (units and procedures) between two study periods (2000–2005 vs 2006–2010) according to reasons for recall.

A. Blood product units recalled
Reasons for blood product recall	2000–2005 (n)	2000–2005 (%)	2006–2010 (n)	2006–2010 (%)	P
BP bacterial contamination	139	0.014	105	0.011	0.073
Donor risk (post-donation information)	19	0.002	7	0.0008	0.044
Error on BP issuing	47	0.005	4	0.0004	<0.001
Non-conforming BP label	20	0.002	11	0.0012	0.230
Transfusion-transmitted viral infection	28	0.003	52	0.006	0.003
Presence of irregular antibodies	0	0	68	0.007	<0.001
Error in laboratory testing*	8	0.0008	0	0	0.018
Inadequate BP quality	6	0.0006	1	0.0001	0.147
B. Blood product recall procedures
Reason for blood product recall	2000–2005 (n)	2000–2005 (1/BP issued)	2006–2010 (n)	2006–2010 (1/BP issued)	P
BP bacterial contamination	38	1/26,268	36	1/25,844	0.961
Donor risk (post-donation information)	15	1/66,546	6	1/155,066	0.111
Error on BP issuing	17	1/58,717	3	1/310,133	0.006
Non-conforming BP label	13	1/76,784	4	1/232,600	0.069
Transfusion-transmitted viral infection	3	1/332,729	8	1/116,300	0.193
Presence of irregular antibodies	0	-	11	1/84,582	0.002
Error in laboratory testing*	5	1/199,638	0	-	0.088
Inadequate BP quality	4	1/249,547	1	1/930,398	0.394

Legend

^*errors in laboratory testing other than test report non-conformity;

BP: blood product.

As shown in Table V, there was a declining trend in the frequency of blood product recalls and of recall procedures initiated due to errors in blood product issuing, post-donation information on donor risk, erroneous product labelling and suspicion of product poor quality, and errors in laboratory testing that resulted in issuing non-conforming blood products. However, an increase was recorded in the blood product recalls within the scope of trace-back procedures (investigation of suspected transfusion-transmitted viral infection), and in recall of products suspected for the presence of irregular anti-erythrocyte or anti-leucocyte antibodies. Although differences between the frequencies compared did not always reach statistical significance, these results should be interpreted with caution because even the least positive or negative trends may considerably influence user perception of the product and service quality and safety. Table VI shows a more detailed distribution of the reasons for product recall within particular categories and recall efficiency.

Table VI.

Distribution of reasons for product recall within particular categories and recall efficiency.

Reasons for product recall	Recall procedure		Products
			Required	Returned
	n	%	n	n	%
BP bacterial contamination	74		244	146	59.8
Test report non-conformity
Erroneous result	26	37.1	58	50	86.2
Erroneous first/last name on test report	22	31.4	25	24	96.0
Erroneous birth date on test report	5	7.1	7	6	85.7
Test report sent to erroneous institution	4	5.7	4	4	100.0
Erroneous reg. number on test report	4	5.7	4	4	100.0
Erroneous unit ID on test report	3	4.3	6	6	100.0
Other	6	8.6	8	7	87.5
Total	70	100.0	112	101	90.2
Donor risk (post-donation information)
Hepatitis suspected in donor	9	42.9	10	5	50.0
Risky sexual behaviour	4	19.0	8	4	50.0
Enquiring about findings	3	14.3	3	2	66.7
Risk suspected	3	14.3	3	1	33.3
Surgical procedure	1	4.8	1	0	0.0
Elevated temperature	1	4.8	1	1	100.0
Total	21	100.0	26	13	50.0
Error on BP issuing
Erroneous type of BP issued	4	20.0	7	7	100.0
BP of erroneous blood group issued	5	25.0	7	7	100.0
BP issued to erroneous institution	3	15.0	10	10	100.0
Other	8	40.0	27	27	100.0
Total	20	100.0	51	51	100.0
Non-conforming BP label
Erroneous first/last name on cross-match label	4	23.5	7	7	100.0
Erroneous product name on label	4	23.5	13	10	76.9
Erroneous blood group on BP label	3	17.6	4	2	50.0
BP labelled with 2 different ID numbers	2	11.8	2	1	50.0
Erroneous date on BP label	2	11.8	3	3	100.0
Volume discrepancy between BP label and documentation	2	11.8	2	2	100.0
Total	17	100.0	31	25	80.6
Transfusion-transmitted viral infection	11		80	10	12.5
Presence of irregular antibodies
TRALI suspected	5	45.5	42	17	40.5
IAT positive	3	27.3	5	2	40.0
Passive transmission of anti-erythrocyte antibodies suspected	3	27.3	21	8	38.1
Total	11	100.0	68	27	39.7
Error in laboratory testing	5		8	6	75.0
Inadequate BP quality
Icteric plasma	1	20.0	3	3	100.0
Lipaemic plasma	1	20.0	1	1	100.0
BP unusable - clots	1	20.0	1	1	100.0
BP with excess leucocytes issued	1	20.0	1	1	100.0
Damaged product package	1	20.0	1	1	100.0
Total	5	100.0	7	7	100.0
Other	11		16	10	62.5
Total	245		643	396	61.6

Legend BP: blood product; IAT: indirect antiglobulin test; TRALI: transfusion-related acute lung injury.

Discussion

In spite of maximal efforts invested by transfusion services to achieve high quality in all segments of their activity, some products and services issued are not in accordance with specified requirements and/or customer expectations. This is primarily a consequence of biological variability, but also of errors in performance and subsequently identified deviations in product quality. It is, therefore, necessary to establish an efficient system of timely recall of the products known or suspected to be non-conforming. These procedures should be based on written protocols, with all activities described in detail and responsibility of those involved in the process precisely defined. Performance of these procedures should be in line with legal provisions, regularly reviewed and revised, and approved by the authorised person.

Many legislative bodies, regulatory agencies and professional organisations require, and to a lesser or greater extent define recall procedures of biological products. In the 22^nd edition of its standards, the American Association of Blood Banks requires establishment of procedures to prevent unintentional use of non-conforming products². Product recall is an integral part of the Council of Europe and World Health Organisation - Good Manufacturing Practices recommendations, and it is described in the PIC-S SOP Procedure for handling rapid alerts and recalls arising from quality defects^3–5. The FDA defines recall of biological products in Title 21, Code of Federal Regulations (CFR), Section 7.3, and describes manufacturer's responsibility and surveillance process of recalls by FDA in Title 21, CFR 7.40–7.59¹. The basic principles of product recall are also defined in Chapter 8, Complaints and Product Recall of EU Guidelines to Good Manufacturing Practice, Medicinal Products for Human and Veterinary Use⁶. Despite the relatively good regulation of this segment of quality management by legal provisions, there is a paucity of literature reports on the reasons and frequency of product recall.

The need for product recall may ensue from the processing of user complaints, deviation from the prescribed product/service quality identified after its issuing or report on a transfusion treatment side effect, or upon demand from inspection or regulatory bodies. In the present study, 109 (44.5%) recall procedures resulted from complaint management, whereas others were caused by nonconformities and errors identified at the CITM. It should be noted that at the CITM, the reports of transfusion treatment side effects that are suspected to be caused by inappropriate blood product quality are processed through the system of complaint management. Thus, this category included reports of suspected bacterial contamination of the blood product and investigation of suspected transfusion transmitted viral infection.

The ever more complex procedures of donor selection, and of blood product preparation and testing, which are accompanied by the ever more stringent quality requirements, have resulted in the increasing number of product recalls. In the late 1990s, one per 2,000 products available to hospitals was recalled in the USA, mostly because of post-donation information on a possible donor risk¹. Our study results indicated a lower frequency of product recall (1/3,184 products for clinical use), however, these differences could be attributed to different work organisation, different criteria for deciding on product recall, and different study periods. Considerable differences are also observed in the reasons for undertaking the procedure of product recall. In our study, recalls for suspected bacterial contamination of blood products predominated and the frequency of such recalls remained stable throughout the study period. While in this group, the rate of product recalls undertaken within the procedure of examining transfusion reactions suspected to be caused by product contamination decreased, the rate of product recalls during the procedure of investigating the initially positive results of bacterial testing of blood products increased. According to Ramsey, post-donation information on donor risk was the leading cause of product recalls in the late 1990s¹. In our study, this cause ranked third, following recall of laboratory test reports, which are not monitored by the FDA.

In our institution, there was no significant decline in the overall frequency of product recalls during the 11-year study period; however, considerable differences were recorded within particular categories. A decrease was recorded in the frequency of product recalls due to errors in product issuing, post-donation information on donor risk, erroneous labelling, suspicion of poor quality of blood products, and errors in laboratory testing that resulted in issuing non-conforming blood products. Notably, these mostly were the reasons deriving from errors in the work of CITM employees. The number of these product recalls has been reduced by upgrading the quality of work and implementation of appropriate corrective measures. Such a favourable trend should first be ascribed to the more stringent criteria of process control in the preparation of blood products, implementation of double administrative control of blood product issuing, and improved safety in all phases of laboratory testing. However, the rate of product recalls within the procedure of testing for suspected presence of irregular anti-erythrocyte or anti-leucocyte antibodies in blood products showed a significant increase. Five of 11 product recall procedures in this category referred to testing for suspected TRALI in the patient. This is not surprising considering the fact that this side effect of transfusion treatment has recently attracted great attention, with consequentially better recognition and reporting. In Croatia, testing for the presence of irregular anti-erythrocyte antibodies is compulsory for blood samples from all first-time donors, and in repeat donors when indicated by the medical history. Given the extended scope of process control in immunohaematology testing where donor samples are used and improved sensitivity of analytical methods, donors with irregular anti-erythrocyte antibodies are more frequently detected as accidental findings. A part of product recalls in this group resulted from testing for suspicion of passive antibody transfer in a blood product recipient. Based on these results, the CITM is planning to introduce screening of all donations for anti-erythrocyte antibodies.

Along with following the reasons and frequency of product recalls, it is of utmost importance to monitor the efficiency of these procedures in order to identify and improve the critical points. According to Ramsey¹, the majority of products needing to be recalled are being transfused because of the relatively small blood product reserves at hospitals and short expiry date of particular blood products. In our experience, the efficiency of product recalls greatly depends on the reasons for which the recall is undertaken, and on the stocks of particular blood products. For example, the efficiency of recalls is lowest in trace-back procedures (12.5%) because of the relatively long time elapsed from transfusion to the presentation of symptoms that may be related to a transfusion-induced viral disease. On the other hand, in the group of testing for suspected bacterial contamination, the efficiency of recalls was much higher (59.8%) because sepsis as an acute reaction is promptly reported to the blood product manufacturer. In the present study, the recall procedures related to blood product quality, labelling and issuing, and those undertaken for suspected non-conformity of the laboratory test reports were most efficient. Proper handling of the product returned includes correct labelling and storage in a separate space until the final decision on further procedures. Records must be kept on the disposal of these products. Recalled laboratory test reports must be clearly labelled as non-conforming and attached to the documents on product recall (or complaint/non-conformity records).

It is of the utmost importance to assess the level of product recall criticality according to the potential adverse effects on patients' health and safety. The FDA classifies product recalls in three classes, from the most severe class I through to the least harmful class III. Ramsey and Sherman analysed blood product recalls published by the FDA during the 1990–1997 period⁷. In these 8 years, approximately 1 per 700 products available to hospitals was withdrawn: 75% of these were class III recalls, 24% were class II, and only 12 units were class I. According to our study results, the greatest proportion of product recalls belonged to class III (47.3%). However, class I recalls were recorded with a comparable high frequency (40.4%). These differences could be explained by different transfusion practices although different criteria for assessing the risk or different timing of risk assessment cannot be excluded.

According to available literature data, products are recalled quite infrequently by the CITM and the extend of the recalls is small. Analysis of trends over time revealed a decreasing rate of recalls in some product categories. On the other hand, an increase in recalls was recorded in some other product categories, as a result of implementation of new procedures and methods, and ever more stringent quality requirements and mechanisms of surveillance. Over the years, there has been a notable change in the perception of clinicians and colleagues from hospital transfusion departments related to product recalls. The initial unease and distrust have been overcome and prejudices have been eliminated and replaced by quality collaboration to the benefit of all those involved in the process of transfusion treatment.