Background: Control of the quality of pharmaceutical and healthcare products in the market is mandatory to ensure the safety and efficacy of the delivered product to the final consumers. The United States Food and Drug Administration (FDA) is providing a continuous and comprehensive updated list for various healthcare issues including drug recalls. Methods: This study provides a multidimensional analysis using statistical process control (SPC) tools to evaluate the risk associated over a 3-year period (2016–2018). Results: The study showed a simple implementation of the combination of SPC tools, which demonstrated that the major contributors to recalls are microbiological quality issues, problems with product compositions, and packaging defects. Months that contributed by more than 60% of the total recalls were from May to August, November, and December. Conclusion: The general trend of drug recall rates is increasing yearly, which should be a warning signal for the regulatory agencies to take preventive measures to control and prevent excessive cases of recalls.

Introduction

Safety and quality of healthcare products generally and pharmaceutical preparations specifically are of great concern to control by regulatory agencies such as the United States Food and Drug Administration (FDA). Every year, there are many recalls of drugs and healthcare products from the market due to various reasons of defects, which may include violation of Good Manufacturing Practice (GMP), lack of adequate manufacturing control, inaccuracy of the measured results, software updates, stability failure, defective device parts, and many others.[1]

However, it is important to deploy effective measures to correct and prevent the reoccurrence of these problems. Statistical process control (SPC) tools are valuable in bringing focus on the dimensions of the challenge and determining the most probable root causes. Moreover, the availability of commercial statistical software packages has facilitated the analysis and processing of data in a timely manner. With this regard, control charts have been used as useful means in the monitoring of the inspection characteristics in the healthcare industry, including pharmaceutical field.[2,3]

This study aimed to apply SPC to analyze FDA recall data during 3 years of observations to determine crucial investigation points to work on to minimize recall rates and minimize financial loss due to nonconforming medicinal dosage forms and devices losses with subsequent withdrawal from the drug market.

Materials and Methods

Recall data records were collected from the beginning of the year 2016 till the end of 2018. The data source was the FDA drug recall website.[1] Recall records were arranged chronologically then segregated according to broad classification:

  1. Composition issues: These recalls may be due to various reasons of inappropriate components of products as recorded by the FDA such as undeclared/misdeclared ingredients, impurities (N-nitrosodiethylamine was very commonly observed) and sub/overamount of a chemical entity.

  2. Lack of sterility assurance: The probability of finding nonsterile unit among injectable products after being subjected to a validated terminal sterilization process is called the Sterility Assurance Level (SAL). The failure to demonstrate this assurance compromises product safety and quality.

  3. Microbial contamination: This defect may include the following bugs: Pseudomonas aeruginosa, Salmonella spp., Scopulariopsis brevicaulis (fungal pathogen), Burkholderia cepacia (and other Burkholderia spp. which was found to be very common), and other microorganisms such as unidentified bacteria and moulds (which may form particles in liquid products).

    It should be noted that both lack of sterility assurance and microbial contamination can be grouped under microbiological quality issues.

  4. Packaging issues: Different problems that are related to packaging system defects were observed: mislabeling, misinformation, mix up, incorrectly marked or damaged primary packaging material, out-of-order packaging, un-updated leaflet instructions, dosage forms mixing, inadequate blister pack seal, leakages, and missing batch number along with expiry date information.

  5. Particles: The presence of particulate matters either exogenous or endogenous in the product that may be viable or nonviable was detected during the recall observation period.

  6. Unapproved drugs: These are marketed drugs that did not have prior FDA approval to prove efficacy and safety.

  7. Miscellaneous: These are rare or isolated causes of recalls that could not be grouped under any of the previous categories and not frequent enough to create own class.

Overall record and separated groups were subjected to statistical analysis using statistical software, namely Minitab, version 17.1.0, and GraphPad Prism, version 6.01, and were used according to their electronic manual.[4,5] hese programs are used in SPC, which depends on the statistical techniques to visualize variation pattern of data for a monitored process to predict the improvement required and if it was achieved or not.[6]

Results

Figure 1 (upper graph) shows box-and-whisker plot for the number of recalls per each month for each year (as three separate groups), which demonstrate a progressive increase in the number of product withdrawal cases each year. Despite the low correlation between recall rates of each year at 95% confidence interval, yet Figure 1 (lower graph) illustrates an almost similar recurring pattern of peaking in recall rates specifically evident between May and August of each year. Pareto diagram in Figure 2 (upper graph) demonstrates the major contributory causes of overall recalls from the 3 years—by more than 70% —which are composition issues, microbial contaminations, packaging defects, and particulate matters, respectively. However, if both lack of sterility assurance and microbial contamination were grouped together, the main causes will have the following sequence: microbiological quality defects, composition issues, and packaging issues.

Figure 1:

Box-and-whisker diagram of the monthly recall trend on yearly basis (upper figure). Time series plot of the monthly recall rates for years 2016 (red), 2017 (green), and 2018 (blue) showing general trend line (lower figure).

Figure 1:

Box-and-whisker diagram of the monthly recall trend on yearly basis (upper figure). Time series plot of the monthly recall rates for years 2016 (red), 2017 (green), and 2018 (blue) showing general trend line (lower figure).

Figure 2:

Pareto chart showing main contributors causes for recalls (upper figure). Rare event chart for the monitoring of United States Food and Drug Administration drug recalls from 2016 to 2018 (lower figure).

Figure 2:

Pareto chart showing main contributors causes for recalls (upper figure). Rare event chart for the monitoring of United States Food and Drug Administration drug recalls from 2016 to 2018 (lower figure).

Rare events trending using G control charts for overall recalls could be seen in Figure 2 (lower graph) but grouped recalls could be viewed in Figures 35 showing the possibility of the occurrence of each root cause for product withdrawal from the market. Generally, the frequency of overall recalls is increasing with time (evident by decreasing interval days between recalls). Despite the constant rate of lack of the sterility assurance recalls and even increasing intervals between products recalls for unapproved drugs, particulate matter, and Burkholderia spp. contamination, the general trend line of recall rates tends to increase due to greater influence from the overall impact of microbiological quality issues, packing, and composition defects, which tend to increase in the rate of recalls. Pareto chart was used in assessing periods of higher recall rates as could be seen in Figure 6 and it was found that more than 62% of the withdrawal rates occurred in November, December, and between May and August months. Another approach for charting of recalls was based on the product withdrawal frequency per month for each year arranged chronologically and trended using attribute type of control charts. However, before applying the conventional process–behavior chart, a diagnostic test was conducted to determine its validity for the existing data without the emergence of false alarms. It was found that dataset was complied with Poisson probability fit plot.[7] Accordingly, control chart was constructed showing excursions in recall rates in 2018 at two peak points in May and August, where the number of recalls reached 12 recalls per each of these two months as could be seen in Figure 6 (lower graph). Interestingly, previous years showed similar twin peaks but lesser in magnitude.

Figure 3:

G charts for monitoring unapproved drug, particles, and packaging issues recalls.

Figure 3:

G charts for monitoring unapproved drug, particles, and packaging issues recalls.

Figure 4:

G charts for monitoring composition-related, lack of sterility assurance, and microbial contamination issues recalls.

Figure 4:

G charts for monitoring composition-related, lack of sterility assurance, and microbial contamination issues recalls.

Figure 5:

G charts for monitoring total recalls related to microbiological issues and Burkholderia spp. contaminated products–related recalls.

Figure 5:

G charts for monitoring total recalls related to microbiological issues and Burkholderia spp. contaminated products–related recalls.

Figure 6:

Test for validity of ordinary attribute chart for monthly recall data and process-behavior chart for recall rates from 2016 to 2018 (middle and lower figures). Pareto chart for monthly recalls rate from 2016 to 2018 (upper figure).

Figure 6:

Test for validity of ordinary attribute chart for monthly recall data and process-behavior chart for recall rates from 2016 to 2018 (middle and lower figures). Pareto chart for monthly recalls rate from 2016 to 2018 (upper figure).

Discussion

This study showed that the general trend of recall rates is ascending gradually from 2016 till 2018, which is evident by box-and-whisker diagram. SPC analysis shows that there are two high peaks in the summer period, which increase in magnitude progressively each year due to the increase in the number of recalls in 2 months during this time. Other researchers have claimed similar outcome and concluded from this ascending rate of recalls that standard FDA guiding rules and GMP are still underestimated by most firms and were handled with carelessness.[8] This is not strange in view of observation of drug recall surge in previous years.[9] This should send a clear warning signal to avoid future excursions and to set mutual collaborative corrective and preventive measures between healthcare firms and regulatory bodies. Although there are rigorous control and record for traceability in developed countries for products affecting the final consumer health, this is seldom practiced in many developing nations.[10,11] A literature review showed that the withdrawal rate of drugs due to harmful effects has not improved consistently over the last 60 years, especially in the African continent where the rate of drug recalls per country was very low (1.17%) if compared with the western countries (6.18% for Europe and 5.83% for North America).[12] A thorough and close monitoring and control should be ensured during times of year where higher rates of recalls are expected based on the previous recent trends. Nevertheless, more than half of the recall problems could be solved by setting corrective and preventive actions (CAPA) for microbiological quality and composition issues. The common reasons for drug recall in this study are in agreement to a great extent to a long-term analysis conducted by other investigators on FDA recalls.[13] Rare event charts are useful to determine incident rates and their probability of occurrence. Drug recall events are considered of rare occurrence.[14] Thus, G control chart is suitable for this type of analysis for monitoring of drug recall rates. It showed usefulness along with attribute charts in monitoring and data visualization.[15]

SPC tools provide an effective solution to monitor and control product recalls from drug market by affording corrective and preventive visions for the problem of defective products through effective two-way communication between regulatory agencies and drug manufacturing firms. Accordingly, this will save human life and avoid financial losses. This work is limited to the analysis of product recall type, rate, time, and the causative factors to the recall. Also, the reason for spiking in the number of monthly recalls during specific times especially during summer requires further study. However, further investigation is required because the study did not address the manufacturers of the recalled products and the root cause of their failure to deliver conforming products to the standard measures. Nevertheless, the current research provided simple, effective yet fast mean of using SPC tools to provide insight into data and predict the trend behavior in addition to guiding the regulatory agencies in the actions required to control the recall rate. Moreover, data can be continuously updated and SPC can detect any potential drift or change in the monitored process and measure the degree of improvements (if any) if actions have been already taken to minimize the number of recalls.

Conclusion

Thorough and close monitoring of healthcare and medicinal products in the US market is maintained as it is crucial to citizens’ health and life quality. Nevertheless, the potential value of gathered data will be revealed upon using SPC tools. They are essential in spotting defects, time of occurrence, and the most likely causative factors. Moreover, they guide the prediction of the possible trend in the future. However, it is not known if the previously observed pattern of recalls would occur again in 2019. Regulatory agencies in developing nations should execute similar comprehensive programs that provide accurate, timely, and rigorous monitoring and control of recalls for healthcare and medicinal products in the market. By doing this, SPC tools could be used to provide insight into the state of the safety and quality of the products due to the availability of the required data on time to deliver the suitable CAPA.

References

References
1.
Drug Safety and Availability
.
Drug Recalls
.
FDA.gov. 2019. Available from: https://www.fda.gov/Safety/Recalls/EnforcementReports/default.htm. [Last accessed on 2019 Jan 17].
2.
Koetsier
A,
van der Veer
SN,
Jager
KJ,
Peek
N,
de Keizer
NF.
Control charts in healthcare quality improvement. A systematic review on adherence to methodological criteria
.
Methods Inf Med
2012
;
51
:
189
98
.
3.
Shah
S,
Shridhar
P,
Gohil
D.
Control chart: A statistical process control tool in pharmacy. Asian Journal of Pharmaceutics (AJP): Free full text articles from Asian
J Pharm
2014
;
4
:
184
92
.
4.
GraphPad Prism Version 6.0.
GraphPad Prism User Guide
.
San Diego, CA
:
GraphPad-Prism Software
;
2013
.
5.
Minitab Inc
.
MINITAB release 17: Statistical software for windows
.
Minitab Inc
.
State College, PA
;
2014
.
6.
Specific Methods of Quality Control
.
Statistical Process Control (SPC)
.
2019
.
Available from: http://web2.mendelu.cz/af_291_projekty2/vseo/print.php?page=7384&typ=html. [Last accessed on 2019 Feb 12].
7.
Minitab® 18 Support
.
Overview for U Chart Diagnostic. Minitab. Support.minitab.com
.
2017
.
8.
Nagaich
U,
Sadhna
D.
Drug recall: An incubus for pharmaceutical companies and most serious drug recall of history
.
Int J Pharm Investig
2015
;
5
:
13
9
.
9.
Gaffney
A.
Number of drug recalls surges at FDA. Regulatory Affairs Professionals Society
.
Led by Mid-Level Concerns. Raps.org
.
2014
.
10.
Devex Editor
.
Poor-quality medicine: A global pandemic
.
Devex
.
2017
.
11.
Johnston
A,
Holt
DW.
Substandard drugs: A potential crisis for public health
.
Br J Clin Pharmacol
2014
;
78
:
218
43
.
12.
Hall
K,
Stewart
T,
Chang
J,
Freeman
MK.
Characteristics of FDA drug recalls: A 30-month analysis
.
Am J Health Syst Pharm
2016
;
73
:
235
40
.
13.
Onakpoya
IJ,
Heneghan
CJ,
Aronson
JK.
Post-marketing withdrawal of 462 medicinal products because of adverse drug reactions: A systematic review of the world literature
.
BMC Med
2016
;
14
:
10
.
14.
Yom-Tov
E.
Predicting drug recalls from internet search engine queries
.
IEEE J Transl Eng Health Med
2017
;
5
:
4400106
.
15.
Eissa
ME.
Evaluation of microbiological purified water trend using two types of control chart
.
Eur Pharma Rev
2018
;
23
:
36
8
.

Financial support and sponsorship

The authors disclosed no funding related to this arwticle.

Conflicts of interest

The authors disclosed no conflicts of interest related to this article.

Author notes

For reprints contact:reprints@medknow.com