Abstract

Each medicine have indications for various disease. It is not rare that one patient is suffering from various diseases, thus a patient takes various medicines. So it is difficult to comprehend the medicine actually prescribed for the treatment of each disease, and its frequency. We construct a database for clinical study, that we call Clinical Database System, which allow us to investigate the relation between disease name, laboratory data, and medicine.

We researched all prescriptions for the patient of hypertension, diabetics mellitus, hyperlipidemia, and angina pectoris in April, 1996. Various medicines besides anti-hypertensive agent are used in the patients with hypertension. The same tendency was observed in other disease. The frequency of each medicine prescribed for the patient with the disease was compared with that without the disease. By selecting the medicine whose frequency was significantly higher in the patient with the disease than that without the disease, we could efficiently get the medicine used for treatment of the disease.

Next, we researched all prescriptions for the patient under leukocytopenia from September, 1996 to March, 1997. In the group of leukocytopenia, various medicines were used. In each patient under leukocytopenia, we list up the medicine which prescribed during the period of leukocytopenia, but did not prescribe before the period. The frequency of the medicine prescribed in the group of leukocytopenia was compared with that in all patient, and the medicines whose frequency were significantly higher in the group of leukocytopenia than that in all patient were listed up. In this list, there were many antibiotics, anti-inflammatory drug, H₂-blockers and other kinds of medicine. However, most of the medicines are thought to be used for treatment of infections and its complications, some medicine may be the cause of leukocytopenia. But we must carefully interpret these data and further examinations are necessary for certification.

Now, we can mechanically search medicines for treatment of the objective disease and the medicines inducing leukocytopenia. We expect that our method is very useful for development of automatically warning system for adverse effect of medicine.

Keywords:

Database; Prescription; Disease name; Laboratory data;

Introduction

There are various medicines using for various disease. We cannot perfectly know the any medicine is used for any purpose. And there are various adverse effect induced by various medicines. We research the way of selecting medicines actually prescribed for the treatment of each disease and monitoring the adverse effects of medicine.

In our Osaka University Hospital, the Hospital Information System(HIS) is running on main frame computer. It is called HUMANE(Human-oriented Universal Medical Assessment System with Network Environment).

Every doctor in Osaka University Hospital inputs his all medical prescription and all order of laboratory examination, radiological study and other physiological examinations into the hospital information system. HIS displays one patient's history of medical prescriptions and laboratory data. Doctors refer the data from HIS and decide next step of treatment.

The database system in mainframe of HIS is well tuned for searching individual patient's record. For example, we can easily access to one patient's medical prescription. But, it takes long time to get the list of patient's name who takes the specific medicine. Furthermore, if we access the database of HIS for clinical study, the load of our query might cause the slow down of the mainframe computer.

Therefore, we constructed the other database beside that of HIS, it is easy to use horizontal search. We call the database system as Clinical Database System.

Structure of database

The Clinical Database System is constructed on Work Station(NEC EWS4800) running the UNIX operating system. The Oracle7 is used for database management system. Daily or monthly, various clinical data is downloaded from host computer.

The database server is connected to private local area network. The client system is small computer, running Windows95 or MacOS. The KeySQL is running on each client computers. The KeySQL is a query program with a graphical user interface. The client program of KeySQL access the KeySQL server running on the work station. The KeySQL server program access the database system. The data file from the Clinical Database System can be formatted for Microsoft Excel, Microsoft Access, and any other application's format.

Now, our clinical database system stores the list of disease name of every patient, all medical prescriptions and a part of the result of laboratory examination . The other data will add to the Clinical Database System in the near future.

Materials and Methods

The frequency of prescribing the medicine to use for a treatment of disease is higher in the group of the disease. The patient who is prescribed the medicine has possibility to suffer the disease which the medicine is used for treatment. We compared the frequency of prescribing the medicine in the two groups; i.e., a group suffering from the disease and another not suffering from the disease.

Part 1. Search the medicine for treatment of the disease.

Object

The object of this part was the outpatient who visited Osaka University Hospital from April 1,1996 to April 30,1996. The number of outpatient was 19,190. We selected patients who had hypertension, hyperlipidemia, diabetics mellitus, and angina pectoris.

Method

First, we selected outpatient who had one of these disease, and their group was named the disease group. The other group, who did not have the disease and a distribution of age and sex are matched with the disease group, was named the non-disease group.

We listed up medicines from prescriptions of two group of the outpatients. From this list, we calculated the frequency of each medicine prescribed. The frequency of each medicine prescribed for the patient in the disease group was compared with that in the non-disease group. If there was a significant difference between the two frequencies of one medicine by chi-square test, we nominated it as the medicine of treatment for the disease.

We calculated the ratio of the number of prescriptions in the disease group and that in the non-disease group. The number of prescriptions in the disease group was named A, and the number of prescriptions in the non-disease group was named B. We calculated the ratio which equals B divided by the quantity A plus B. If the ratio was under 0.3, then we selected the medicine for treatment.

Part 2. Search the medicine that is caused leuko-cytopenia.

Object

The object was the outpatient and inpatient who visited Osaka University Hospital between September 1,1996 and March 31,1997. The number of patients was 99,000. From the laboratory data of blood picture, we selected the patient whose white blood cell counts under 3,000 per cubic millimeters.

We excluded the example from the patient who had cancer and blood disease, because those patient has a obvious reason for leukocytopenia and we wanted to select the example of "accidental" leukocytopenia.

Method

We listed up the medicine used in leukocytopenia patients. In each leukocytopenia patients, two prescriptions were searched. One prescription was during the period that the number of white blood cell was in normal range and another was during the period that the number of white blood cell decreased,. From the two prescriptions, we selected the medicine which used in the period of leukocytepenia but not used in the period when the number of white blood cell was in normal range.

Next, the frequency of prescribing the medicine in the group of the patient with leukocytopenia was compared with that in all patient. If there was a significantly difference by chi-square test, the frequency of the medicine was compared with the frequency of the medicine that dose not reported the adverse effect of leukocytopenia. If there was a significant difference in two frequency by chi-square test, the medicine was nominated for the cause of leukocytopenia..

Results

Part 1.

We calculated the frequency which was the ratio between the number of prescribing the medicine and the number of the patients with a disease. In the patient of hypertension, the most frequently used medicine was teprenone. Most frequently used medicine which is not medicine for the treatment of hypertension was nifedipine, but the frequency was under 10%.

In the group of hypertension, we selected 36 kinds of medicines by our methods. In these medicines, 34 kinds of medicines were permitted to use the treatment of hyper-tension(Table 1).

In the group of diabetics mellitus, we selected 15 kinds of medicines and 11 kinds of medicines were permitted(See Table 2). In the group of hyperlipidemia, we selected 6 kinds of medicines and 6 kinds of medicines were permitted(Table 3). In the group of angina pectoris, we selected 11 kinds of medicines and 9 kinds of medicines were permitted(Table 4).

*ISDN = isosorbide dinitrate

ISMN* = isosorbide mononitrate

Table 5-selected medicines in leukocytopenia patients

medicine	number of prescriptions in all patients	frequency of leukocytopenia
amphotericin B	173		4.6%
furosemide	198		3.5%
polymyxin B	116		4.3%
ciclosporin	107		2.8%
famotidine(powder)	116		2.6%
ubidecarenone	323		1.9%
ciclosporin 2	125		2.4%
digoxin	196		1.7%
Co-trimoxazole	220		1.8%
isoniazid	235		1.7%
norethisterone	120		1.7%
methotrexate	101		2.0%
vancomycin	105		1.9%
sodium alginate	185		1.6%
salazosulfapyridine	206		1.5%

Part 2.

The total number of inpatients and outpatients was 59,707. The number of medicine was 136,436. Famotidine was the most number of prescription medicine in the group of leukocytopenia. The number of patients taking famotidine was 2889. By the comparing the frequency of the prescriptions, 15 medicines were selected. In these medicines, the most specific medicine in leukocytopenia was amphotericin B. The frequency in Table 5 is the ratio between the number of prescriptions in the patients of leukocytopenia and the number of prescriptions in all patients.

Discussion

In our methods, we can easily select the medicine for treatment. In the patient's group of hypertension, ciclosporin and benzbromarone are selected in spite of no permission for treatment of hypertension. One of the adverse effect of a ciclosporin is hypertension, and the hyperuricemia frequently complicated hypertension. In these cases, we might happen to point out the hypertension as adverse effect or complication.

In the patient's group of diabetics mellitus, ISDN, dimethylpolysiloxane, propentofylline, kallidinogenase, delapril, bifonazole, and polyenephosphatidylcholine are selected in spite of no permission for treatment of diabetics mellitus as shown in Table 2. The propentofylline and the kallidinogenase can use for the treatment of the disturbance of brain circulation, and the disturbance of brain circulation is a complication of diabetics mellitus. The polyene-phosphatidylcholine and clino-fibrate can use for the hyperlipidemia, and the diabetics mellitus is frequently complicated by the hyperlipidemia. The polyene-phosphatidylcholine can use for the treatment of the adverse effect of acarbose and voglibose. ISDN and delapril are listed up because circulatory disease is the complication of diabetics mellitus. The reason that bifonazole and polymyxin-B being might be the infectious disease is often complicated with diabetics mellitus.

In the group of hyperlipidemia, insulin and procaterol are selected. Diabetics mellitus is often complicated with hyperlipidemia, hypertension and ischemic heart disease.

In the group of angina pectoris, as shown in Table 4, aspirin and warfarin potasium are selected. The aspirin is frequently used as antithrombotic agent. The warfarin potasium used for anticoagulant therapy. These therapy is important of the prevention of the myocardial infarction and other cardiac event. Disopyramide, procainamide and mexiletine mean the arrythmia is complication of ischemic heart disease.

The medicines that have possibility to induce leuko-cytopenia are various kinds. We selected 15 medicines by our method. Many of them, antibiotics are used for the treatment of infections. H₂-blocker can be used for the prevention of stress ulcer in the severe infections. But ubidecarenone and some other medicine has no reason to use for the treatment of leukocytopenia and complications. They are suggested the cause of leukocytopenia. We must carefully research these medicine.

Conclusion

We made a database for clinical study. Using this system, we surveyed the relation of medicine, disease name, and laboratory data. In some disease, we can mechanically select the medicine for the treatment by the survey of prescriptions and disease name. And we select some medicine as the cause of leukocytopenia by the survey of prescriptions and laboratory data. We expect that our method is very useful for the development of automatically warning system for adverse effect of medicine.

References

[1]Evans RS, Pestotnik SL, Classen DC, Bass SB, Burke JP Prevention of adverse drug events through computer-ized surveillance. Proc Annu Symp Comput Appl Med Care. 1992:437-41