EMPLOYMENT OF DATA MINING TO HELP IN INFECTION CONTROL FOR IN PATIENTS

EMPLOYMENT OF DATA MINING TO HELP IN INFECTION CONTROL FOR IN PATIENTS

Abstract

Hospital related infections are affecting patients in current times. Apart from increasing the hospital stay, they also increase the healthcare costs. Data mining can be used to reduce the occurrence of these infections. Data collected in hospitals when analyzed can reveal patterns, which are significant in the prevention of the hospital related infections. One of the hospital related infection is pneumonia specifically the ventilator related pneumonia. The use of a ventilator machine in the medical intensive care units and high-risk nursery leads to the ventilator-associated pneumonia (VAP). VAT leads to a high mortality more than community-based pneumonia. A study in prevalence of ventilator related pneumonia in Huntington Hospital revealed that 57 % of all the patients who uses the ventilator acquire ventilator related pneumonia. With this pattern, it is easy to determine the measures to take in the prevention of the infection.

INTRODUCTION

Hospital acquired infections are the most complicated infections affecting many patients today. A more common scientific name for these infections is Nosocomial infections. A study of the Harvard medical practice found that surgical wound infection was the second largest category with adverse effects. Nosocomial present the largest risk to patients admitted to hospitals. In the year 1957 and 1958, when there was an epidemic of Nosocomial in England, the problem became the focus of public health. Since then public health, practitioners with an emphasis on surveillance have done the study of these infections (Burke, 2003 p651). Nosocomial is not the only adverse effect in health care but also the most studied. In current times, more than 5% of patients admitted to hospitals acquire one or more infections, and the risk is steadily increasing. Nosocomial infections affect more than two million patients every year in the United States resulting in the death of more than 90,000 patients. These infections also result in more than five billion U.S dollars as the cost of patient care. This makes infection control a significant component of patient safety and health concerns.

The most prevalent infections include the Urinary tract, bloodstream, surgical site, and pneumonia infections. These include more than 80% of all the Nosocomial infections. A quarter of the infections affect patients in intensive care units who are more susceptible to such infections. In addition, microorganisms that are resistant to many antibiotics cause 70% of the infections. This poses a danger to public health given that patients depend on antibiotics to stay safe from microorganisms. When a microorganism is resistant to more than one antibiotic, then it can affect many people or cause an outbreak. These hospital-acquired infections are increasing regarding frequency, associated costs, and mortality rate. As outlined by Burke (2003 p651), urinary tract infections are the most common followed by surgical site infections. Pneumonia and blood stream infections are less common but are associated with high mortality rates and associated costs. This only shows that among the four most prevalent infections, all of them have varied adverse effects. The occurrence of bloodstream infections has tripled since the year 1975, but the surgical infections have slightly decreased.

With the rate at which the adverse effects of the Nosocomial infections is increasing, there is need to manage the issue. Data mining is among the techniques that can help control this challenge. Currently, most researchers and hospitals extract patient’s data manually from the files, which is tedious and time-consuming (Wisniewski et al., 2003 p454). In the case of research, this limits the sample size due to the labor involved. With the current developments in technology, it is easy to have a data warehouse, which is more efficient and accurate. With a data warehouse, one can easily extract the data required and conduct an analysis. This can improve research and enable the prevention of Nosocomial infections. With data mining, it is possible to combine and analyze data from many databases to reveal patterns that can predict outbreaks and hospital infections. This can help control the occurrence of these diseases and save lives as well as reduce the costs associated with patient care. Lobdell et al. (2012 p 65) argue that by the year 2002, the occurrence of hospital-acquired infections was estimated to be 1.7 million patients resulting in a mortality rate of more than 6% in the United States. This exceeds the mortality rate associated with colon and breasts cancer combined. These are worrying statistics given that these infections such as hospital related pneumonia are controllable. , which is more chronic. Hospital-related infections have been very prevalent all over the world. It is the high time public health concentrated in controlling these infections by use of data mining to predict their occurrence.

The concept of data mining lies at the interface of database technology, statistics, visualization, and machine learning. A database is a collection of information organized in such a way that its contents are efficiently and easily accessible. Manipulation of data in a database to generate all kinds of patterns and results is easy and fast. Since hospitals have already integrated the traditional statistical methods such as probability, data is available for use though it may not be sufficient. All hospitals have integrated the traditional methods of statistics and probability, and other hospitals have separately implemented data mining techniques for better results. With a database, one can access the required data and conduct an analysis. In addition, with data mining, it is possible to conduct an analysis of the whole population analyzed as opposed to taking a sample. Data from several databases is brought together to draw conclusions about a certain infection and develop litigations (Lobdell et al., 2012 p 67). However, without data mining, this is hard since the availability and accessibility of data will be inefficient. This is the most efficient way through which health care quality is improved

In hospitals, there are many microorganisms and other infectious bacteria, which are harmful. There are cases of microorganisms mutating and becoming immune to antibiotics and other medical agents. With such instances, hospital-related infections have become more common. Patients go to the hospital for health care, and it is very disturbing for one to contract other infections at the hospital. In addition, the fact that hospitals are the providers of health care means that they are responsible for preventing the hospital-related infections.

Research questions

This paper seeks to unravel how the data mining technique can help in solving the challenge of hospital-related infections. This is a challenge that is associated with adverse effects and huge costs. Data mining generates infinite possibilities of hidden data patterns, which can help to generate a countless diagnosis. The specific question under study is;

  • Can data mining techniques solve the hospital related infections (HRI) challenge?

The sub-questions include;

  • Can data mining generate patterns for diagnosis and treatment of Pneumonia as an HRI?
  • How effective is data mining in the management of Pneumonia as an HRI?

Patients with life-threatening illnesses commonly contract pneumonia while in the intensive care unit. This is more so due to the use of the ventilator. Ventilator-associated pneumonia increases the mortality rate and health care costs. The fact that critically ill patients in the intensive care unit can contract other diseases such as pneumonia raises a critical question. 86 % of the nosocomial pneumonia is associated with the use of ventilators. In the Unites States alone, more than 300, 000 people die from ventilator-related pneumonia. Estimates reveal that more than 27 % of the patients mechanically ventilated contract pneumonia. It is the second most killer infection in the ICU and the most common in those ventilated. The high rates of ventilator-associated pneumonia call for preventive measures and control. The CDC is currently concentrating on the prevention and control of Nosocomial infections among them ventilator-associated pneumonia. The CDC collects data from different hospitals and determines the reduction rates in the occurrence of hospital-related infections. Hospitals use this data to make conclusions about individual reduction rates.

LITERATURE REVIEW

Identifying the risk factors associated with hospital-related infections is significant. Burke (2003 p 651) argues that identifying the risk factors can reduce the risk of infection. It is possible to use strategies in the prevention of HRI. These strategies are divided into systems based, process based, and education based. Some of the most suggested interventions including education and trainings have been very vague and difficult in terms of implementation. One of the major obstacles to the control of HRI is behavioral change. Health care workers commonly infect patients with contaminated hands. Despite spending a lot of money in trainings and education, health care workers still fail to adhere to the set standards of hygiene. Failure to comply is blamed on understaffing and impractical guidelines. The widely recognized method of infection control method of hand washing is not enough to control the HRIs. Some bacteria and microorganism will still survive under water and others are airborne. In the developing countries, the lack of resources makes it hard for health care providers to maintain hygiene. This increases the occurrence of the hospital related infections and raising the mortality rates in these countries.

The process based and the education-based strategies of prevention are widely applied in health care facilities but still the challenge of HRIs still exists. The system-based strategies such as data mining have not been widely accepted. Active surveillance is significant in identifying the risk factors that health care providers can change. There has been wide evaluation of widely in the development of health care processes to control infections. However, these infections remain persistent. Without surveillance, it is hard to determine the success and failures of the efforts put in prevention and control of these infections. Surveillance ensures that data capture and storage in databases occurs where it is accessible for further analysis. With surveillance, it is possible to build patterns and statistical models to predict the occurrence of an infection. The process-based indicators are not very efficient in controlling the infections in that the adherence to set procedures and processes does not always occur.

In his study, Burke (2003 p 653) presents two examples of cases where process indicators have failed illustrating the need for more surveillance. Quality improvement projects have identified errors in the administration of antibiotics before surgery. This has increased the occurrence of surgical-site infections. Errors in the administration of antibiotics are associated with an increase in surgical site infections by a factor of two. In surgeries, there is the set procedure followed and if the health care provider deviates from the process, infections are bound to occur. In a New Yolk case study, errors had occurred in 27 of the 54 people used as a sample in the study. This would mean that errors occurred in half of the patients operated. Studies have shown that patients who undergo improper surgical processes do not contact any infections. This masks the problem in individual hospitals. Therefore, if proper monitoring of such an error is not properly done, then, it would be hard to identify the risk factors that contribute to infections in the hospitals.

Burke (2003 p 654) again argues that at least 5 % of the hospital related infections occur in outbreaks that can be detected by careful surveillance of information. Recognition of most of the outbreaks occurs in the laboratories. It is significant o detect and control outbreaks to prevent catastrophes. Outbreaks are the major causes of high mortality rates. When Ebola broke out in Africa, many people died translating to a high mortality rate. With enough and efficient infection surveillance, it was easy to discover the outbreak and control it. The overdependence on lab results to reveal the outbreak led to many deaths. Scientists have recently applied data mining techniques to detect outbreaks and prevent them. Molecular technique shows that transfer of genes in different species causing antibiotic resistance causes unrelated infections.

Data Mining

Data mining is the process of finding unknown patterns and trends by analyzing data in databases and using the results to make predictions. It includes a variety of techniques and methods. This method is considerably a new methodology developed in 1994. Its objective is to identify potential useful and valid patterns and correlation sin data. This happens through combining data sets to form patterns that are subtle for humans to detect. As argued by Obenshain (2004 p 690), database vendors are inputting the mining algorithms into the database to enable an efficient data management process. The value of doing this is to ensure that the data can be manipulated easily and efficiently without much effort or system knowledge. Such an automated surveillance system can be advantageous in that it reduces time and increases efficiency. Data mining holds a great potential for the health care industry. Many industries have adopted data mining to predict and model customer experiences and make necessary changes. However, the health care industry lags behind in the adoption of technology. Data mining in the healthcare, data mining has been an academic exercise and very few people have successfully succeeded in practical implementation. With the potential in practice, it is the high time health practitioners adopt the data mining controlling and preventing hospital-related infections.

In health care, the use of data mining is becoming increasingly popular. Heath care insurers to minimize the medical insurance abuse and fraud are using data mining techniques more. In addition, the health care transactions generate data so voluminous for the traditional methods to analyze (Koh and Tan, 2011 p 65). However, with data mining, decision making through the discovery of patterns and trends in large volumes of data is easy and efficient. This kind of analysis is increasingly becoming necessary with the high health care costs and increased mortality rates related to HRIs. The insights from data mining can influence revenue, cost, and operating efficiency while at the same time raising the quality of health care. As Koh and Tan, (2011 p 65) puts it; data mining generates information that can be helpful to the stakeholders in the health care industry. From improving the quality of health care through reducing HRIs, it also reduced health care insurance frauds and abuse. This gives a more insightful idea since healthcare facilities and healthcare insurance industries can collaborate in implementing data mining techniques. The health insurance companies have adopted data mining as opposed to health care facilities, which still insist on the traditional methods of laboratory analysis to predict outbreaks.

Data Mining in Health Care Management

To aid in the management of health, data mining applications and methods should be developed to better track and identify chronic diseases and high-risk patients. For the hospital related infections, the most common include the Urinary tract, bloodstream, surgical site, and pneumonia infections. However, these infections do not affect virtually all patients admitted to hospitals. An infection such as surgical site infection occurs mostly when the right surgery procedure is not adhered to. However, of all the patients who undergo surgery and the right procedure are not followed, not all of them contact the surgical site infection. This solely means that there are those who are more prone to such an infection more than others are. The risk is high in some patients than others (Koh and Tan, 2011 p 66). In this kind of a case, data mining comes in to identify the most common nosocomial infections, identify the characteristics of the high-risk patients, and develop appropriate interventions. This will help in the management of the infections thus lowering the mortality rate as well as the cost of health care.  Already some of the countries and states have adopted the data mining technique in the management of health care and have witnessed its success.

A good example is the Arkansas Data Network, which reviews resource utilization and readmission and after comparing the data with the current literature in health care develops a treatment option. This is using evidence to provide better health care by use of data mining. The Group Health Cooperative based in Seattle arranges patients population based on characteristics such as demographic and medical condition. This helps determine the group requiring much resources and introduce programs to educate the population on management and control of the medical condition. The group has been involved in several data mining efforts to make health care more affordable and of high quality. The Seton Medical Center uses data mining to reduce patients stay in the hospital, maintain high standards in service delivery, avoid complications and provide information to the health care providers (Koh and Tan, 2011 p 66).  To reduce expenditure and improve outcomes, Blue Cross has implemented several data mining initiatives. The organization uses claim records, physician interviews, emergency department data, and pharmaceutical records to identify any unknown asthmatic and develop interventions.

The health care process generates a large amount of data, which if manipulated through technology can generate trends and patterns. The computer-based record systems have made it easy to record and store data in virtual databases accessible and easy. A health care facility with a computer will record a vast amount of data (Prather et al., 1997 p 101). With a large amount of data, techniques are needed to search through the data and generate patterns and relationships. These techniques include neural networks, statistical methods, and machine learning methods. The use of these techniques to generate relationships and patterns from a large amount of data is part of data mining. Data mining encompass diverse techniques and methods of data analysis. The technique used depends on the objective and aims of the researcher and the data available. The crucial part of the whole process is to extract the required data, clean it, and then convert the qualitative data to quantitative data for statistical analysis. Obenshain (2004 p 691) recommends two strategies of data mining. These include unsupervised and supervised learning. In supervised learning, the output data sets are placed into the system to train it and get the desired outputs. However, in unsupervised learning, the data appears in groups.

There are two types of pneumonia including community acquired (CAP) and health care-associated pneumonia (HCAP).  A study by Micek et al. (2007) revealed that patients with HCAP were more likely to require machine ventilation as compared to those with CAP. The study involved a sample of patients suffering from CAP and d HCAP revealed that patients with HCAP are more likely to die as compared to patients suffering from CAP. This is a clear indication that the hospital related pneumonia more so the ventilator-associated pneumonia has a higher mortality rate than the normal pneumonia. The ventilator machines save lives of patients in intensive care but when a patient in intensive care acquires a pneumonia infection due to the ventilator machines, the chances of survival becomes minimal. This increases the mortality rate because of hospital-related infections; data mining can help determine those at high risk of acquiring ventilator-related pneumonia and develop an instigation plan.  Pneumonia is among the infections that increase mortality rate not only in the developing countries but more so in the developed countries. Despite the advances made in improvement of therapies such as the antimicrobial therapy and the use of preventive measures, hospital related pneumonia remains a significant cause of high mortality and morbidity rates not only in the United States but also all over the world. The diagnosis of VAP is not efficient as compared to the diagnosis of Community-associated pneumonia. This makes it the most prevalent kind of pneumonia. It is the highest nosocomial infection killer among the intensive care unit patients. This is more because patients in the ICU are suffering from other diseases and occurrence of VAT just makes their condition worse.

Ventilator Associated Pneumonia

            Ventilator-associated pneumonia (VAP) complicates the treatment course of many critically ill patients. The mortality rate of VAP ranges from 24% to 50%. The rate can rise to as high as 76% in some settings. Since antimicrobial treatment improves the outcome of the patients, the identification of the affected patients is significant to improving the status.  The lack of standardized criteria for the diagnosis of VAP limits the epidemiology of VAP. VAP is normally present when a patient develops chest radiograph or progressive infiltrate (Bouza et al., 2001 p 1035). As opposed to community-acquired pneumonia, the limited criteria of epidemiology limit the diagnosis. Most of the diagnosis criteria are based on lab testing which can take time. Though the attribute of death has decreased over the years, the mortality rate related to VAP is still high.

The kind of organism causing VAP largely depends on the duration of mechanical ventilation. Early VAP is caused by pathogens, which are sensitive to antibiotics while drug resistant and bacteria bring about the late onset. This means that once a patient contracts VAP from the ventilator, it can later become drug resistant and thus untreatable. This is the major reason why the mortality rate for VAP is high (Mellor, A., 2000 p10). The diagnosis of VAP remains a mystery since most of the clinical diagnosis methods still miss some of the patients. Different health organization and bodies recommend different diagnosis methods, but none is extremely efficient. The American Thoracic Society (ATS) for example recommends the use of respiratory tract microbiology and culture samples. Sample analysis is done in the laboratory, and the diagnosis determined.

Frequency of Ventilator-associated Pneumonia

In the year 1992, a study done to determine the prevalence of VAP concentrated in 1417 ICUs and evaluated 10038 patients. The study revealed that 21% of the patients acquired ICU related infections. The prevalence of pneumonia was 10% (Mellor, A., 2000 p8). The study concluded that there are seven major risk factors for ICU-acquired infections including mechanical ventilation. Another study done in European ICUs concluded that mechanical ventilation increased the risk of in ICU patients. Many studies conducted in the prevalence of VAP have shown that VAP frequencies vary from 8% to 28% (Chastre and Fagon, 2002 p 868). This high rate calls for control and prevention. Chastre and Fagon (2002 p893) concluded that antimicrobial therapy and adequate supportive measures remain the major treatments of VAP. However, the increased mortality rates related to VAP calls for a change in the techniques and control measures. The treatments methods are not effective enough to control the infection. The major control method, in this case, would be to prevent the occurrence of VAP through looking at the risk factors and controlling them. If the rate of these who contract VAP reduces from 28% to 10%, then mortality rate related to VAP would reduce dramatically.

Risk factors for VAP

Epidemiology studies identify the risk factors in the development of VAP. Weinstein et al. (2004) argue that VAP is the most lethal infection among patients in the ICU. Most estimates of the rate of occurrence of VAP conclude the rate to be 10-40%. However Weinstein et al. (2004) argues that the rate can be higher since most of the studies concentrate on patients undergoing ventilation for less than 48 hrs. If the high rates are integrated to the many people undergoing mechanical ventilation, then VAP would be the most common killer infection in the world. VAP increases the mortality rate and prolongs the hospital stay by more than seven days. The prolonged hospital stay increases health care costs and reduces the patient-doctor ratio. Patients in the ICU require continuous treatment and these calls for a high patient-doctor ratio. However, when many patients are in the ICU, this ratio decreased rapidly affecting the quality of health care accorded to the patients.

One of the major risk factors for the development of VAP is intubation. This is the placement of a plastic tube on the trachea to facilitate the administration of certain drugs. This combined with mechanical ventilation leads to VAP. Weinstein et al. (2004) recommend that unnecessary intubation be avoided. However, in some health cases, intubation is unavoidable. An alternative could be the use of Noninvasive positive-pressure ventilation (NIPPV) which has been determined to be beneficial as opposed to intubation. However, it also poses a danger to the development of VAP, and unnecessary use is unrecompensed. Another risk factor is the duration in the ventilator machine (Morehead and Pinto, 2000 p1930). The risk of VAP is however not constant depending on the time of ventilation. The conclusion of a cohort estimated the risk to be 3%, 2%, and 1% in the first week, second week and third week. As a result, reducing the ventilation period can reduce VAP development. The argument is that the body is not used to mechanical ventilation in the first and second week and thus increasing the development of VAP. However, by the third week, the body has accustomed to the ventilator machine. Reducing the ventilation period in the first two weeks will reduce the occurrence and development of VAP.

Another risk factor is the resistance to drugs. It is common for a human body to resist antibiotics and other drugs. When a human body resist drugs, then the treatment becomes hard. In cases of VAP, it sometimes takes long to diagnose. This gives the body time to adjust and resist the drugs administered. In addition, VAP occurs in patients under treatment for other diseases. The different drugs administered to the patient can react to complicate the infection. Besides, with continued use of the ventilator machine, the drugs may not work given the increase in the causation factor. Other factors such as enteral nutrition and combined interventions also complicate the matter.  There is no specific intervention program for VAP and health care providers are left to try the different interventions available. Depending on the availability of resources, different intervention programs can be used which increased the chances of failure in intervention. A prevention program based on education has recently shown to reduce the occurrence of VAP. Although there are many interventions to reduce VAP, they are rarely used.

The intervention measure largely depends on factors such as availability of resources. There cannot be a universal intervention program, and the health facilities should develop their prevention programs. However, this can only be possible through data mining. He risks factors and then develops intervention programs. Through data mining, hospitals can identify the risk factors and then work to control and prevent the occurrence of VAP (Weinstein et al., 2004). Through data mining, a hospital can predict the occurrence of VAP and then adjust accordingly to save lives and resources. Once data is analyzed and a prediction made, the hospital can take preventive measures. With the high mortality rate related to VAP, prevention is better than treatment. The prevention of VAP can only be dome through data mining. Data mining generates information that can be helpful to the stakeholders in the health care industry. From improving the quality of health care through reducing HRIs, it also reduced health care insurance costs.

This paper will show how using data mining a hospital can prevent the occurrence of VAP and install preventive measures. It is clear that the only way to reduce mortality rate associated with VAP is to prevent its occurrence rather than treatment. In addition, the overdependence on the tests from laboratories, which can take time, is not effective in the preservation and diagnosis of VAP. Data mining predicts the occurrence of VAP and helps in the prevention of VAP before it occurs. The use of data mining in the healthcare industry is gaining popularity in recent times and it is time all facilities took it up in not only the prevention of VAP but also other hospital-related infections.

METHODOLOGY

Data mining technique

The predication data mining technique will be applied. The technique models a relationship between the dependent variable and the independent variable. In this case, the pneumonia is dependent on the use of the ventilator machine. The idea is to build a relationship between the use of the ventilator machine and the contracting of pneumonia. With the relationship identified, it is then easy to predict future occurrences of VAT and take necessary preventive measures. The predication technique applies when a relationship between the independent and dependent variables predicts future independent variables. In this study, the pneumonia is the independent variable and the ventilator machine is the dependent variable.

Study Area

Data collection will occur in Huntington Hospital, which is equipped with an intensive care unit (ICU) and High-risk nursery (HRN). Located in Huntington town in New Yolk, it ranks among the best hospitals. The hospital is equipped with an MICU and HRN. In both of these hospital units, a ventilator device moves air in and out of the lungs. The machine helps patients who are insufficiently breathing and thus found in the ICU and HRN units. With a ventilator machines, patients can contract pneumonia infection, which is the major focus of this study. Pneumonia is among the most prevalent hospital-related infections. The objective is to determine if one can use data mining to control pneumonia as a hospital-related infection.

Data collection

Primary data

Data collection will occur for one month. The process will occur with the help of a structured form for data recording. The target population includes patients in the Medical Intensive care unit (MICU) and children over seven pounds in weight and above in the high-risk nursery (HRN). Data to be captured include the ventilator days and the patient days.

Date Pneumonia infections Ventilator days Patient days
       
       
       
       
       
  Total Total Total

 

The ventilator days are the total number of days of exposure to the ventilator by all the patients, that is, the total number of patients exposed to the ventilator for the whole month. The patient days on the other hand are the total number of days in the MICU and HRN, that is, the total number of patients in the MICU and HRN in the course of the month. The Pneumonia infections column will record the total number of patients who contracts pneumonia because of using the ventilator. With the totals above, the calculation of ventilator associated infection rate and the device utilization rate (DU).

Secondary data

To enhance analysis and interpretation, recommended rates will be sources from Center for Disease Control (CDC) report for comparison with the results above, after comparison, it will be easy to interpret the data and make a conclusion based on the interpretation. The CDC regularly publishes device associated data rates, which a facility can use to make decisions after analyzing individual data. The CDC uses data reported by hospitals participating in the National Nosocomial Infections Surveillance (NNIS). The CDC basis its results on the combined data and one can compare the results of the combined data with individual hospital data to make conclusion about a Nosocomial infection in the hospital.

Data Analysis

After collecting data using the methodology defined above, and summing up the data, the following are the totals. (The full table is located at the end of this paper)

Pneumonia infections Ventilator days Patient days
60 105 160

From the table above, there has been 60 ventilator related pneumonia related infections in Huntington Hospital for the month of September 2016. 105 patients have utilized the ventilator for the month and given that only 60 acquired pneumonia, 45 did not contact the pneumonia. For the whole of the month, there were 150 patients in the MICU and HRN. A significant number of the patients in MICU and HRN did not use the ventilator in the course of the month.

= 0.57

=57 %

The ventilator associated infection rate of 57 % shows that of all the patients who utilized the ventilator for the month, 57% acquired the pneumonia infection.

= 0.65

                                        =65%

The device utilization rate shows that 65 % of all the patients in the MICU and HRN utilized the ventilator in the course of the month. In a recent, report by CDC, it was estimated that the number of pneumonia infections in acute care hospitals would be 157, 500. This is a high number given that it is only one infection. CDC recommends that prevention of HAIs is possible only if the stakeholders including doctors and nurses are aware of the infection. The only way to be aware of the infection is to follow the statistics and make predictions. The CDC releases report regularly on the national statistics on hospital-related infections. Hospitals are encouraged to compare their data with the national data to determine if the hospital rate is a low or high outlier. This happens through determining whether the infection rates and the device utilization ratios are above the median and CDC recommends percentile. If the rate is below the 10th percentile and then it is a low outlier.  However, if it is above the 90th percentile, then it is a high outlier. It is significant to examine the device utilization rate together with the infection rate since ventilation is a risk factor for pneumonia. If the infection rate is above the 90th percentile and the device utilization rate between 75 and 90th percentile, then preventive measures need to be installed. Such measures may be a reduction of ventilation time or use of other drugs. As suggested by Morehead and Pinto, 2000 p1930), the risk of VAP is not constant depending on the time of ventilation. The conclusion of a cohort estimated the risk to be 3%, 2%, and 1% in the first week, second week and third week. Therefore, if the Huntington Hospital followed these findings, they can reduce the time spent in the ventilator in the first and second week and manage VAT.

Discussion and Conclusion

From the data above, a higher number of patients in the MICU and HRN who utilize the ventilator acquire pneumonia. The rate of acquisition is 57 %. More than half of the patients using the ventilator acquire pneumonia. After comparing the data with the CDC recommended rates, the device utilization rate is between the 75th and 90th percentile. In addition, the infection rate is above the 90th percentile. This makes the rate a high outlier, which indicates there is a problem. The statistics indicate that most of the patients using the ventilator machine end up acquiring the pneumonia infection. Given that the patients need the machine, it is not advisable to stop them from using the machine. However, it is possible to limit the ventilation time. Ventilation is a significant risk in pneumonia and limiting the duration can lower the number of patients acquiring pneumonia. Weinstein et al. (2004) argue that most estimates of the rate of occurrence of VAP to be 10-40%.

In this study, the rate of occurrence is 57%, which is higher than the rate found by Weinstein et al. most of the studies conclude that mostly the rate lies between 8 to 40%. The rate at Huntington Hospital is 57% indicating that there is a higher likelihood of a patient using the ventilator machine will contract VAP. This means that the hospital needs to rethink the use of the ventilator machine. The hospital needs to rethink reducing the period spent in the ventilator machine especially in the first and second day. Previous studies have identified VAP as a major cause of the increased mortality and hospital costs. From the results above, if the VAP rate is 57%, this means that most of the patients using the ventilator machine will contract pneumonia. the continued collection and analysis of data in the two ICUs will reveal a pattern which can then be used to generate a course of action. Contracting pneumonia and combined with the initial illness can lead to death or prolong the hospital stay. Prolonging the hospital stay increases the health care costs not only to the hospital but also to the individuals. Economically, reducing the occurrence of VAT will increase the consumable income on the patients who spend a lot of money for treatment of infections acquired in the hospital. The mortality rate because of the hospital-acquired infection is high and can be reduced through statistical methods such as data mining.

Most of the hospital-related infections are preventable through predictions (CDC, 2004). Data mining enables the prediction based on past occurrences. With predictions, it is possible to concentrate on prevention rather than waiting for the occurrence of VAT and then treatment. Data mining can be used to predict the occurrence rate of other hospital-related infections such as surgical site infection and take necessary measures. Data mining can be very effective in predicting the occurrence of VAT and then work to prevent its occurrence. In this same way, all other hospital-related infections are preventable (Chastre and Fagon, 2002 p890). For decades, data has been used in determining the best course of action. From business to health, data mining is used to control the occurrence of diseases and reduce costs of health care. Virtually all areas of health care can adopt the concept of data mining to reduce the occurrence of all health condition. The hospitals, however, need to record all available clinical data for analysis.

Recommendation for further research

I recommend further research to determine if data mining can reduce the occurrence of other diseases such as cancer. Some of the radiotherapy machines used in hospitals increase the occurrence of cancer. In the treatment of cancer, however, such, machines are used which might increase the severity of one’s condition. Data mining is a diverse concept, which is applicable in many areas of health care.

Data Collection Results

 

Day Pneumonia Infections Ventilator days Patient days
1 1 2 4
2 2 3 7
3 4 7 9
4 2 6 8
5 4 5 7
6 5 8 10
7 1 3 8
8 1 2 4
9 2 6 8
10 2 2 4
11 1 2 3
12 2 2 2
13 0 1 1
14 4 5 7
15 2 2 4
16 2 5 7
17 4 6 9
18 1 5 9
19 2 6 8
20 3 4 7
21 4 4 5
22 2 2 3
23 1 1 2
24 2 3 4
25 1 2 3
26 1 1 2
27 1 4 5
28 1 2 4
29 0 1 1
30 2 3 5
Totals 60 105 160

 

 References

Bouza, E., Brun-Buisson, C., Chastre, J., Ewig, S., Fagon, J.Y., Marquette, C.H., Muñoz, P., Niederman, M.S., Papazian, L., Rello, J. and Rouby, J.J., 2001. Ventilator-associated pneumonia. European Respiratory Journal, 17(5), pp.1034-1045.

Burke, J.P., 2003. Infection control-a problem for patient safety. New England Journal of Medicine, 348(7), pp.651-656.

Chastre, J. and Fagon, J.Y., 2002. Ventilator-associated pneumonia. American journal of respiratory and critical care medicine, 165(7), pp.867-903.

Koh, H.C. and Tan, G., 2011. Data mining applications in healthcare. Journal of healthcare information management, 19(2), p.65.

Lobdell, K.W., Stamou, S. and Sanchez, J.A., 2012. Hospital-acquired infections. Surgical Clinics of North America, 92(1), pp.65-77.

Mellor, A., 2000. Risk factors for ventilator associated pneumonia. Critical Care, 3(1), p.1.

Micek, S.T., Kollef, K.E., Reichley, R.M., Roubinian, N. and Kollef, M.H., 2007. Health care-associated pneumonia and community-acquired pneumonia: a single-center experience. Antimicrobial agents and chemotherapy, 51(10), pp.3568-3573.

Morehead, R.S. and Pinto, S.J., 2000. Ventilator-associated pneumonia. Archives of Internal medicine, 160(13), pp.1926-1936.

CDC, 2004. National Nosocomial Infections Surveillance (NNIS) System Report, data summary from January 1992 through June 2004, issued October 2004. American journal of infection control, 32(8), p.470.

Obenshain, M.K., 2004. Application of data mining techniques to healthcare data. Infection Control & Hospital Epidemiology, 25(08), pp.690-695.

Prather, J.C., Lobach, D.F., Goodwin, L.K., Hales, J.W., Hage, M.L. and Hammond, W.E., 1997. Medical data mining: knowledge discovery in a clinical data warehouse. In Proceedings of the AMIA annual fall symposium (p. 101). American Medical Informatics Association.

Wisniewski, M.F., Kieszkowski, P., Zagorski, B.M., Trick, W.E., Sommers, M. and Weinstein, R.A., 2003. Development of a clinical data warehouse for hospital infection control. Journal of the American Medical Informatics Association, 10(5), pp.454-462.

Weinstein, R.A., Bonten, M.J., Kollef, M.H. and Hall, J.B., 2004. Risk factors for ventilator-associated pneumonia: from epidemiology to patient management. Clinical Infectious Diseases, 38(8), pp.1141-1149.

Do you need an Original High Quality Academic Custom Essay?