Extended application of statistical process control- quantitative risk assessment techniques to monitor surgical site infection rates
- Mostafa Essam Ahmed Eissa , Department of Microbiology and Immunology, Cairo University, Cairo, Egypt
Article Information:
Abstract:
Surgical wound infections are major challenges affecting the health and even the life of patients subjected to invasive operations in hospitals. Healthcare providers should ensure the appropriate quality control of surgical site infection (SSI). World Health Organization (WHO) provides a comprehensive global database record of SSI. Monitoring of the trend of SSI rate and magnitude using statistical process control (SPC) tools would deliver useful information about the previous, current, and expected behavior and pattern of the inspected quality characteristic. The degree of compliance with good practices guidelines and rules (GXP) and the improvement achieved or required might be assessed and quantified using Shewhart or process-behavior trending charts. The following study demonstrates the application of SPC analysis using commercial software packages in the assessment of SSI in selected cases of countries from the WHO dataset.
Keywords:
Article :
Introduction:
Infections related to surgical procedures are a global challenge that may impact a patient’s health and may even threaten life in extreme cases [1]. Control measures of surgical site infection (SSI) are being actively sought to contain it. These measures should minimize the contamination of surgical wounds from both endogenous and exogenous microorganisms [2]. The worldwide efforts to set standards for control of SSI are supported by WHO guidelines, resources, tools, and database records collected from different nations [3-5]. The present study demonstrates the application of statistical process control (SPC) on SSI rates trends of selected nations from WHO records as could be shown in Table 1, which lists the categorization and the abbreviation code for each country [5]. It should be noted that in this SSI work Tajikistan (TJK) is listed alone and as part of the Central Asian Republics Information Network (CARINFONET) database record [5].
SSI Pattern Monitoring Using Statistical Software:
It could be observed that the general trend line of SSI % was toward decreasing rates except for Belgium (BEL) which showed a gradual increase. The average SSI value was low for CARINFONET region and relatively high for BEL despite greater annual variations that could be seen in TJK as could be seen from Figure 1.
Table 1. WHO record for SSI rates of the selected countries under investigation [5]
|
Country |
ISO 2 |
WHO |
WHO |
Members |
Members |
Members |
Commonwealth |
Central Asian |
|
group full |
|
Code/ |
European |
of the |
of the |
of the |
of Independent |
Republics |
|
name† |
|
ISO 3 |
Region |
European |
European |
European |
States |
Information Network |
|
|
|
|
|
Union |
Union |
Union |
|
members |
|
|
|
|
|
|
before |
after May |
|
(CARINFONETΨ) |
|
|
|
|
|
|
May 2004 |
2004 |
|
|
|
|
|
|
|
|
(EU15) |
(EU13) |
|
|
|
Belgium |
BE |
BEL |
yes |
yes |
yes |
no |
no |
no |
|
Portugal |
PT |
PRT |
yes |
yes |
yes |
no |
no |
no |
|
Tajikistan‡ |
TJ |
TJK |
yes |
no |
no |
no |
yes |
yes |
ΨCARINFONET consists of five nations: Kazakhstan, Kyrgyzstan, Tajikistan, Turkmenistan, and Uzbekistan.
‡Independent SSI % record from CARINFONET is available for TJ.
†Data source: https://gateway.euro.who.int/en/datasets/european-health-for-all-database/.
Statistical software (GraphPad Prism for Windows version 6.01) analysis showed that TJK data showed significant skewness from Gaussian distribution at a 99% confidence level and passed only Kolmogorov-Smirnov (KS) normality test while other groups passed all normality tests. Spearman correlation showed weak (BEL with PRT and TJK), moderate (BEL with CARINFONET and PRT with TJK and CARINFONET), and very strong (normally, TJK with CARINFONET) levels of relations between datasets [6].
On the other hand, histograms of Figure 2 illustrate data distribution showing their degree of distortion from the bell-shape of the normal distribution spreading. Moreover, Box plot diagram shows the difference between general trend, spreading, distortion in distribution and outlier point(s) (evident in TJK only marked by a red colored single point) of SSI rates. One-Way ANOVA at α=0.05 in Table 2 shows that BEL SSI rate is significantly higher than other groups which are supported by Box-and-Whisker plot. When TJK record is included with CARINFONET SSI rate, the net result is lower mean value for SSI percent and minimal data spreading without outliers. The details of the report generated using a statistical program (GraphPad Prism for Windows version 6.01) are numerically presented in Table 2.
Table 2. Box-and-Whisker plot and One-Way ANOVA test showing data spreading, skewness, and outlier(s) (marked as a red dot) combined with the assessment of the significance of variation between SSI percent of each group (Data generated using GraphPad Prism for Windows version 6.01/Minitab® version 17.1.0)
|
One-Way ANOVA |
|||||||||
|
Significance level (Alpha) |
0.05 |
||||||||
|
Tukey's |
multiple |
comparisons test |
Mean diff. |
95% CI of diff. |
Significant? |
Summary |
DF |
||
|
PRT VALUE vs. TJK VALUE |
-0.15 |
-0.42 to 0.13 |
No |
ns |
62 |
||||
|
PRT VALUE vs. CARINFONET VALUE |
0.036 |
-0.24 to 0.31 |
No |
ns |
62 |
||||
|
PRT VALUE vs. BEL VALUE |
-0.89 |
-1.2 to -0.62 |
Yes |
**** |
62 |
||||
|
TJK VALUE vs. CARINFONET VALUE |
0.18 |
-0.11 to 0.47 |
No |
ns |
62 |
||||
|
TJK VALUE vs. BEL VALUE |
-0.75 |
-1.0 to -0.46 |
Yes |
**** |
62 |
||||
|
|
-0.93 |
-1.2 to -0.64 |
Yes |
**** |
62 |
||||
|
Test details |
Mean diff. |
SE of diff. |
n1 |
n2 |
q |
||||
|
PRT VALUE vs. TJK VALUE |
-0.15 |
0.10 |
20 |
15 |
2.0 |
||||
|
PRT VALUE vs. CARINFONET VALUE |
0.036 |
0.10 |
20 |
15 |
0.50 |
||||
|
PRT VALUE vs. BEL VALUE |
-0.89 |
0.10 |
20 |
16 |
12 |
||||
|
TJK VALUE vs. CARINFONET VALUE |
0.18 |
0.11 |
15 |
15 |
2.3 |
||||
|
TJK VALUE vs. BEL VALUE |
-0.75 |
0.11 |
15 |
16 |
9.7 |
||||
|
CARINFONET VALUE vs. BEL VALUE |
-0.93 |
0.11 |
15 |
16 |
12 |
||||
ns: Not Significant, CI: Confidence Interval, DF: Degree of Freedom, SE: Standard Error, Raw data source: WHO [5]
Figure 2. Histogram showing SSI rates for four WHO data records along with correlation matrix at a 95 % confidence interval (Data generated using GraphPad Prism for Windows version 6.01).
Laney modified attribute control charts were constructed using SPC software and the methods were detailed and described through Minitab® version 17.1.0 [7,8]. Laney process-behavior charts are more accurate in isolating assignable-cause variations from common- cause variations when data do not show compliance with the presumed distribution required for ordinary trending charts [9]. Attribute control charts present SSI rates as the number of cases per 10,000 patients and the red points represent special-cause out-of-control SSI years. Warning signals above the midline of the mean value (green solid line) are required to be investigated to set further measures to control SSI. In the same line, red dots below U' line provide a good opportunity for improvements and to learn from it where SSI % at minimum values (Figure 3).
Discussion:
Despite the general stability of SSI % for PRT, there was unusual aberrant spiking in two successive years due to possible extraneous factors that require investigation to set better control. A similar situation could be found with TJK (and normally linked to CARINFONET region trend) at one year. However, in case of BEL, the drop from high SSI value to low rates was campaigned by a gradual increase through the following years and the shift in the process means to reach a new higher value. Thus, a continuous update of data is required to elucidate the new drift in the balance and to verify the process stability. However, it would be desirable to bring SSI rates to the downside again to maintain the infection risk at minimum values in hospitals. Laney modified control charts have been applied in various fields in the healthcare industry including pharmaceutical processes [10-12]. The ease with which trending charts could be applied will facilitate on-time process monitoring and control to guard against any possible excursions that may ruin money and efforts in any industry. The current work shows that further improvements and GXP measures are required to ensure better control over infections in hospitals that are stemmed from surgical wound contamination by microbial cells. Years of SSI rate excursion should be reviewed to analyze the root causative factors to set protective measures for patient health and life. Also, the stability of SSI percent from year to year is indicative for the adequacy of the countermeasures against microbial intrusion into the surgical wound. Other factors should not be overlooked such as the significant rise in the number of admitted individuals to hospitals, ecological and demographical changes impact.
Simple and fast risk analysis method could be derived from control charts parameters by considering multiplication of both means (as a severity factor) and upper control limit (as a brink of excursion). The multiplication will amplify small differences which will expose the significance between the comparisons. For instance, SSI risk for BEL (2.11) is the highest followed by TJK (1.04) then CARINFONET (0.33) and PRT (0.30).
Conclusion:
SPC tools are useful for monitoring global records of SSI rates to monitor the past and current states of control and to investigate excursions to set corrective actions and preventive actions (CAPA). Nevertheless, the area of improvements could be spotted from the process-behavior charts to provide better control measures on microbial infections within healthcare facilities in order to stabilize SSI trends at low levels.
References:
- Li B, Webster TJ. Bacteria antibiotic resistance: new challenges and opportunities for implant‐associated orthopedic infections. J Orthop Res 2018;36:22- 32.
- Eissa ME. Minimization of microbial infection risk from surgery. Open Journal of Surgery 2018;1:20-3.
- World Health Organization (WHO). Global guidelines on the prevention of surgical site infection [Internet]. World Health Organization 2018 [cited 5 Jan 2019]. Available from: https://www.who.int/gpsc/ssi-guidelines/en/.
- World Health Organization (WHO). Surgical site infections tools and resources [Internet]. World Health Organization 2018 [cited 5 Jan 2019]. Available from: https://www.who.int/infection-prevention/tools/surgical/en/.
- World Health Organization (WHO). WHO European health information at your fingertips [Internet]. Gateway.euro.who.int 2018 [cited 5 Jan 2019]. Available from: https://gateway.euro.who.int/en/datasets/european-health-for-all-database/.
- Spearman’s correlation [Internet]. Statstutor.ac.uk. 2019 [cited 5 Jan 2019]. Available from: http://www.statstutor.ac.uk/resources/uploaded/spearmans.pdf.
- Methods and formulas for Laney U' Chart. Methods and formulas for Laney U' Chart - Minitab [Internet]. Support.minitab.com 2017 [cited 5 Jan 2019]. Available from: https://support.minitab.com/en-us/minitab/18/help-and-how-to/quality-and-process-improvement/control-charts/how-to/attributes- charts/laney-u-chart/methods-and-formulas/methods-and-formulas/.
- Attribute Control Charts. Minitab assistant white paper [Internet]. Support.minitab.com 2018 [cited 5 Jan 2019]. Available from: https://support.minitab.com/en-us/minitab/18/Assistant_Attribute_Control_ Charts.pdf.
- Interpret the key results for Laney U' Chart. Interpret the key results for Laney U' Chart – Minitab [Internet]. Support.minitab.com 2018 [cited 5 Jan 2019]. Available from: https://support.minitab.com/en-us/minitab/18/help-and-how-to/quality-and-process-improvement/control-charts/how-to/attributes- charts/laney-u-chart/interpret-the-results/key-results/.
- Eissa M. Application of Laney control chart in the assessment of the microbiological quality of oral pharmaceutical filterable products. Bangladesh J Sci Ind Res 2017;52:239-246.
- Eissa M. Adulterated pharmaceutical product detection using statistical process control. Bangladesh Pharmaceutical Journal 2018;21:7-15.
- Eissa M. Application of attribute control chart in the monitoring of the physical properties of solid dosage forms. Journal of Progressive Research in Modern Physics and Chemistry 2018;3:104-113.