JP5355302B2 - Accredited research institution verification system and accredited research institution verification program - Google Patents

Description

  The present invention relates to an accredited research institution verification system and an accredited research institution verification program for verifying an accredited research institution that conducts an accreditation survey on an applicant for certification for long-term care in a long-term care insurance system.

  The long-term care insurance system is operated by each municipality as an insurer (group) under the Ministry of Health, Labor and Welfare, which supervises and supervises. In this long-term care insurance system, when a resident of a municipality wants to use a long-term care service, an application for certification for long-term care is applied to the municipality. When an application is made, the municipality dispatches an authorized investigator from the research institution, and conducts an accreditation survey on the physical condition of the applicant for the need for nursing care. This certification survey is performed by selecting an option set for each item for a plurality of preset certification survey items. In addition, the investigator makes a free statement as a special note about the items that cannot be expressed in the choices of each item. At the same time, the municipality asks the attending physician to create a written opinion.

  The results of the certification survey are input to a computer, and the standard time required for certification for long-term care is calculated by a predetermined certification program using a tree diagram. The degree of care required based on the calculated standard time required for certification for long-term care (not applicable, required Support 1, 2 and each stage from need of nursing care 1 to need of nursing care 5) are output as primary determination results.

  After this, at the nursing care certification committee consisting of health, medical and welfare specialists, comprehensively check the results of the primary judgment, the opinion of the attending physician and special notes, as well as the contents of the reference index presented by the Ministry of Health, Labor and Welfare, and make a secondary judgment. It is determined whether care (support) is required, and if care (support) is required, how much care (support) is required. Then, based on the judgment of the care certification examination committee, the insurer city / town / village performs a need for nursing care (support required) certification (see, for example, Patent Document 1). That is, in the certification of long-term care, a person who is recognized as needing long-term care is recognized as one of the seven categories (degree of care required) of support 1 or 2 and long-term care 1 to 5, and services corresponding to this category. Can receive.

JP 2001-5880 A

  The standard for determining the level of care required is based on how long it takes to take care (standard time for certification for long-term care), and the standard time required for long-term care is the survey agency entrusted to the local government or local government as described above. The survey results (74 basic survey items) of interviews by the surveyors are input to the computer and calculated. As described above, the survey results by the investigators of this survey organization become basic data for determining the degree of care required. Therefore, this survey must be accurate and it is desirable to be able to verify the results of each survey.

  One way to verify the content of a research organization's survey is to determine the standard time required for long-term care, calculated by a research organization, for a model group (for example, a group of ideal research organizations that conduct appropriate accreditation surveys, such as investigators). It is conceivable to check whether it is prominent as compared with the average value of the population of the research institutions, which is a group of research institutions specified in advance, where education instruction is thorough. If the result of the check is prominent, there is a possibility that the investigation method or judgment of the relevant research institution is not appropriate.

  Conventionally, there is no mechanism for clarifying such a prominent value (protrusion value), and there is no mechanism for clarifying the factor. Therefore, it has not been possible to identify an appropriate survey method or survey organization that has made a judgment and to examine its validity.

  An object of the present invention is to provide an accredited research institution verification system and an accredited research institution verification program that can identify an inappropriate research institution by verifying whether an accredited survey result by the research institution is appropriate.

  The accredited research institution verification system according to the present invention retains the accreditation data including the nursing care accreditation standard time, which is the result of the long-term care accreditation survey conducted in the past, in a grouped state for each research institution that conducted the survey. Using the data storage means and the authorization data held in this authorization data storage means, extract the average nursing care certification reference time for each research institution, and obtain the average value and standard deviation of each average nursing care certification reference time A standard measure calculating means for obtaining a standard measure for the average required care-required reference time for each research institution from the average value and standard deviation and the average required care-required reference time for each research institution, and the standard measure calculating means. Protruding value detecting means for detecting a probing survey organization whose standard measure of the average time required for certification for long-term care is out of a preset range And features.

  Furthermore, the accredited research institution verification program according to the present invention operates the accredited research institution verification system on a computer.

  According to the present invention, it is possible to verify the certification survey result by checking whether or not the nursing care certification standard time of a certain research organization is prominent compared with the average value of the population of the research organization. In addition, if an overhang is confirmed, it is possible to check the outliers of more detailed items step by step, clarify the overhaul, and determine whether the investigation method and judgment of the relevant research institution are appropriate. It can be considered and can provide guidance and explanations to the research organization.

It is a conceptual block diagram explaining one Embodiment of the research organization verification system by this invention. It is a functional block diagram explaining the standard measure calculating means in one embodiment of the present invention. It is a functional block diagram explaining the protrusion value detection means in one embodiment of the present invention. It is a functional block diagram explaining the chi-square calculation means in one embodiment of the present invention. It is a functional block diagram explaining the chi-square test means in one embodiment of the present invention. It is a flowchart explaining the operation | movement for excluding the research organization of the peculiar age structure from verification object in one embodiment of this invention. It is a flowchart explaining the operation | movement for detecting the research organization in which the nursing care certification | authentication reference | standard time protruded in one embodiment of this invention. It is a figure showing the age composition of investigation object of each investigation organization in one embodiment of the present invention in a tabular form. It is a figure showing the average nursing care certification | authentication reference | standard time of each investigation organization in one embodiment of this invention with a table | surface form. It is a figure which represents the average nursing care certification | authentication reference | standard time according to classification | category of each investigation organization in one embodiment of this invention with a table | surface form. It is a figure which represents the verification content using the chi-square value according to id of the private research organization in one embodiment of this invention in a table format. It is a figure showing the verification content using the chi-square value according to id of the direct management research organization in one embodiment of the present invention in a tabular form. It is a figure showing the tree diagram for calculation of the nursing care certification | authentication reference | standard time in a meal division.

  Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows a conceptual configuration of a research institution verification system in this embodiment. This survey organization verification system, for example, is the presence or absence of prominence of each average nursing care certification standard time, which is the survey result of multiple survey organizations, such as a directly managed survey organization of the municipality or a private research organization entrusted to the municipality, in a certain municipality unit. And has an authorization data storage means 11, an exclusion means 12, a standard measure calculation means 13, a protruding value detection means 14, a chi-square calculation means 15, and a chi-square test means 16. Each of these means 11, 12, 13, 14, 15, 16 is realized by a computer and its peripheral devices, and executes predetermined processing by a preset program.

  The accreditation data storage means 11 categorizes the accreditation data, including the time required for long-term care certification, which is the result of the long-term care-required certification survey conducted by each research institution for each certification subject requiring long-term care. Is held in a state.

  The excluding means 12 has a function of excluding, from the verification target, the research institutions whose certified data number does not reach the predetermined number from among the authorized data held in the authorized data storage means 11 and grouped by research institutions. In other words, when the number of data is small, such a research institution that is strongly influenced by the value of the data with a small verification result and makes accurate verification difficult is excluded. In addition, this exclusion means 12 is a research organization with a specific age structure in which the average age of a specific nursing care application subject to be surveyed is not within a preset range compared to the average age of the population. Also, this research institution is excluded from the verification. That is, in general, the long-term care-required certification reference time increases as the age of the long-term care-recognized survey subject increases, and decreases as the age decreases. Therefore, if the age structure of the subject requiring certification for long-term care (which is also the applicant for certification for long-term care) is biased to either the older age or the younger age, even if the survey results are appropriate, As a result, it may appear as a prominent value in the survey results of the population, making accurate verification difficult. Therefore, survey institutions with a specific age structure are also excluded from the verification target.

  When there are a plurality of data to be verified, the standard measure calculation means 13 obtains a standard measure value indicating where each piece of data is located. That is, as shown in FIG. 2, for a plurality of data xi to be verified, the average value μ is obtained by the average value calculation unit 131 and the standard deviation σ is obtained by the standard deviation calculation unit 132. Then, the standard measure ti is obtained by the following formula (1) by the standard measure calculation unit.

ti = (xi−μ) / σ (1)
However, i = 1 to n
The protrusion value detecting means 14 determines whether each of the standard measures obtained by the standard measure calculating means 13 is out of the preset range. That is, as shown in FIG. 3, a threshold value S (for example, ± 1.96σ or the like) for specifying a predetermined range is set in the protruding value detection unit 14, and each standard measure ti is within the range of this threshold value. Judging whether to enter. As a result, data that does not fall within the threshold range is determined as protrusion value data. When the threshold value is ± 1.96σ, individual data xi of the population having an average of μ and a standard deviation of σ falls within a range of μ ± 1.96σ with a probability of 95%. Therefore, the remaining 5% becomes the protrusion value data.

  The chi-square calculation means 15 assigns ids to a plurality of occurrence events in the verification target research institution, and the degree of deviation is calculated from the occurrence frequency (measured frequency) for each id and the theoretical frequency by a chi-square value X ^ 2. Calculate as Although the details will be described later, the theoretical frequency is calculated from the occurrence frequency (measured frequency) for each corresponding id of the model research institution set in advance, and is held in the authorized data storage means 11 as shown in FIG. Further, from the actual measurement data of the verification target research organization, the actual measurement frequency extraction unit 151 for each id extracts the actual measurement frequency A [id] for each id in the verification target research organization. Also, using the actually measured frequency for each id in the measured data of the model research institution held in the authorized data storage means 11, the theoretical frequency for each id in the verification target research institution is calculated by the id-specific theoretical frequency calculation unit 152. [Id] is calculated.

  Here, the theoretical frequency B [id] is the total value of the measured frequency (frequency distribution) for each model research institution by id, and the total value of the actually measured frequency A [id] by id of the target research organization to be compared. The expected frequency distribution of the target research institution calculated by matching The theoretical frequency B [id] is obtained by the following equation (2).

Theoretical frequency B [id] by id of the target research organization
Total value of measured frequency A [id] by id of the target research organization ×
(Measured frequency by id of model research institute / id by model research institution
Total value of measured frequencies) (2)
The model research organization is an ideal research organization that conducts appropriate accreditation surveys. For example, it is a research organization that has been specified in advance and has thorough educational guidance for investigators.

  The id-specific chi-square value calculation unit 153 calculates the id-specific chi-square value X ^ 2 [id] by using the measured frequency A [id] for each id and the theoretical frequency B [id] for each id. Calculate in (3).

X ^ 2 [id] = (A [id] -B [id]) ^ 2 / B [id] (3)
The chi-square value calculation unit 154 obtains a chi-square value X ^ 2 by summing the id-specific chi-square values X ^ 2 [id] obtained by the above equation (3).

  The chi-square test means 16 tests the set hypothesis. As hypotheses, the null hypothesis (the research method (result) between the model research institution and the target research institution is not different) and the alternative hypothesis (the model research institution and the target research institution have a deviation) are set. Therefore, it is verified whether there is a difference in the distribution of the survey results between the model survey organization and the target survey organization. That is, the chi-square calculation means 15 calculates how much the obtained data is deviated from the theoretical value as the chi-square value X ^ 2, so if the probability that the numerical value is generated is extremely small, the null hypothesis is obtained. Reject and adopt the alternative hypothesis (the hypothesis you want to prove, the real intention).

  As shown in FIG. 5, the chi-square test means 16 is set with a degree of freedom n and a significance level α. The degree of freedom n is a value obtained by subtracting −1 from the number of possible categories (in this case, the number of ids). That is, if one is decided, it is the number that can be moved freely. The significance level α is a chi-square value X ^ 2 in a chi-square distribution determined by the degree of freedom n (represented by a curve in FIG. 4. If the degree of freedom n changes, the chi-square distribution (the shape of the curve also changes)). The probability of the occurrence of. That is, this is a criterion for determining whether the probability of occurrence of the chi-square value X ^ 2 is significant (the null hypothesis is adopted) or rejected. For example, when the significance level is 1% (α = 0.01), the null hypothesis is broken if the probability of occurrence of the chi-square value X ^ 2 represented by the horizontal axis is less than 1%.

  Next, the operation of the above configuration will be described with reference to FIGS. This research institution verification system compares the values of the standard time required for long-term care certification obtained from a long-term care accreditation survey conducted by the research institution, and there is a research institution that has outstanding data for the population. It is to check whether to do. This check is performed in units of local governments, for example. In this embodiment, as shown in FIG. 8, XX city will be described as an example. In XX city, there are 01 to 20 research organizations, of which survey organizations 01 and 02 are directly managed by XX city, and the other research organizations 03 to 20 are private research organizations.

  Here, accreditation data of insurers (municipalities) nationwide is stored and stored in the accreditation support center under the supervision of the care insurance system. Therefore, the department that verifies the research organization takes out the certification data of each research organization 01 to 20 in the city XX from this certification support center and stores it in the certification data storage means 11 of the research organization verification system shown in FIG. .

  In the research institution verification system, first, authorization data is grouped according to the research institutions 01 to 20 (step 601). The research institution verification system compares the average time required for certification of long-term care required for each grouped research institution and checks whether there is a prominent value compared to other research institutions, but performs an accurate check. Therefore, as a precondition, a research institution that is significantly different from other research institutions in the number of data samples and data characteristics of the research institution is excluded from the verification target by the exclusion means 12.

  That is, if the number of samples of data held by the research institution is small or the survey subject is extremely old, there is a possibility that a prominent value may occur depending on the characteristics of these data. For example, if the survey target includes a large number of elderly people, the certification time required for long-term care inevitably increases. Therefore, research institutions with such data characteristics are excluded from the verification targets.

  For this purpose, the excluding means 12 sets the threshold value of the number of samples to x or more in advance, and determines whether or not the number of samples of authorized data is x or more for each of the research institutions 01 to 20 (step 602). As a result, survey institutions that are less than x are excluded (step 602: No). On the other hand, with regard to the accreditation data of the research institution determined to be x or more (step 602: Yes), it is checked whether the age structure is an exclusion target.

  First, a population ratio of 65 years old and over and a standard measure are calculated (step 603), and a list thereof is created as shown in FIG. 8 (step 604). In FIG. 8, in the top row, the average age of XX city, the population (number of cases) for each age group above a specific age (here 65 years), and the ratio thereof are shown. In addition, the said average age is an average age more than specific age (65 years old). The second row and the subsequent lines in FIG. 8 show data of each research organization 01 to 20 in XX city. For each research institution 01-20, from the left column in the figure, the average age of a specific age (65 years old) or more, the average ratio of the average age of XX city and the average age of the relevant research institution, the standard measure of the relevant research institution, The population (number of cases) by age group over a specific age (65 years old) and the ratio are shown. In addition, the value of the standard deviation with respect to each average age is shown outside the bottom line.

  The standard measures in each of the survey institutions 01 to 20 are obtained by the standard measure calculation means 13 as follows. That is, the standard measure calculation means 13 inputs the average age data of the research institutions 01 to 20 as the plurality of data xi shown in FIG. The average value calculation unit 131 calculates the average value μ of the population, and the standard deviation calculation unit 132 calculates the standard deviation σ. Based on the average age μ and standard deviation σ of certification applicants requiring nursing care at a specific age or higher for all survey institutions (populations) calculated in this way, and the same average age xi of each survey institution, standard measure calculation unit 133 Thus, the standard measure ti is calculated by the above equation (1).

  Next, it is judged by the protruding value detection means 14 whether the standard measure ti in each research organization 01-20 calculated in this way is outside the preset threshold range (step 605). That is, as shown in FIG. 3, a threshold value (here, ± 1.96σ) is set in the protruding value detection means 14 so as to specify a planned range, and the survey organization calculated by the standard measure calculation means 13 is set. The standard measure ti of the average age for every 01 to 20 is input, and among these, the survey organization with the standard measure ti that falls outside the preset range is detected as a survey organization with a specific age structure.

  In the example of FIG. 8, for the threshold value ± 1.96σ, the research organization 6 has a standard measure of 2.08 and deviates upward, and the survey organization 18 has a standard measure of −2.51 and deviates downward. Yes.

  Here, since the threshold value is set to ± 1.96σ, the individual data xi of the population whose average is μ and the standard deviation is σ, that is, each average over the specific age (65 years old) of each research institution 01 to 20 Age falls within the range of μ ± 1.96σ with a probability of 95%. Therefore, in this example, the remaining 5% is the protruding value data.

  The exclusion means 12 is a research institution whose age structure detected by the protruding value detection means 14 is unique, that is, a research institution of an age structure determined to be outside the preset threshold range (step 605: Yes) (FIG. 7). In the example, the research organization 6 and the research organization 18) are excluded from the verification targets.

  In the above description, the presence or absence of a prominent value was checked based on the average age of all survey subjects over 65 years old. For example, the ratio of 75 years old or older corresponding to the late elderly may be used as a criterion. .

  On the other hand, the accreditation data of the survey organization of the age structure determined to be not outside the preset threshold range (step 605: No) is used to check whether the average nursing care accreditation reference time is prominent as the next stage. Transition.

  In order to identify the research institutions with which the average nursing care certification standard time is prominent, use the certification data from which the research institutions 06 and 18 were excluded in the previous stage. The ratio and standard measure (by total / by category) are calculated (step 701), and the comparison table shown in FIG. 9 and FIG. Is created (step 702).

  Here, the long-term care certification standard time is the total of the long-term care certification standard times calculated for each of the eight categories, and in order to show these easily, It is written as time (total), and the certification time required for long-term care required for each category is written as the standard time required for certification for long-term care (by category). As described above, the certification time required for long-term care (by category) is divided into eight categories: “meal”, “excretion”, “movement”, “cleanliness maintenance”, “indirect care”, “BPSD”, “function” The “training” and “medical-related” sections are calculated using a tree diagram, which will be described later. The nursing care certification standard time (total) is a value obtained by summing the above eight nursing care certification standard times (by category).

  FIG. 9 is a comparison table regarding the certification time required for long-term care (total), and FIG. 10 is a comparison table regarding the standard time required for certification of long-term care (by category). These certification dates for certification of long-term care (total) and certification times of certification for long-term care (by category) are calculated for each certification applicant for certification for long-term care. The column shows the average values calculated by calculating the standard time data of certification applicants for long-term care required by the research organization. These are the average long-term care standard reference time (total) and average long-term care certification as mentioned above. It is called the reference time (by category).

  The case where the protruding value of these average nursing care certification reference time (total) and average nursing care certification standard time (by category) is detected will be described below.

  First, regarding the average certification for long-term care required (total), in the comparison table of FIG. 9, the “standard time” column shows the value of the average required certification for long-term care (total) as described above. ing. In this example, a comparison between a directly managed research institution group and a private research institution is taken as an example, so the top two lines show the directly managed research institutions 01 and 02 of XX City, and the third line shows the average of these directly managed groups. Values are shown. In the fifth line and below, private research organizations (research organizations 03 to 20 excluding the survey organizations 06 and 18) are shown. The fourth line shows the standard deviation value for each research institution calculated using the average value of the directly managed group. In the “average ratio” column, the ratio of the average nursing care certification reference time (total) of each research institution to the average value of the directly managed group is shown. Further, in the “standard measure” column, the standard measure value of each research institution calculated by the standard measure calculating means 13 is shown.

  In the standard measure calculation means 13, the value of the average nursing care certification reference time (total) of each research institution extracted from the certification data is input, and the average value of each research agency for each of these average nursing care certification reference times (total). The standard deviation is obtained, and the standard measure for the average nursing care certification reference time (total) of each research institution is obtained from the average value and standard deviation and the average nursing care certification standard time (total) of each research institution. That is, the standard measure calculating means 13 inputs the average nursing care certification reference time (total) of each research organization as the plurality of data xi shown in FIG. Then, an average value μ of each research organization is obtained by the average value calculation unit 131. In this embodiment, as described above, since the directly managed research organization is compared with the private research organization, the average value of the directly managed research organization is used. For this reason, the data xi input to the average value calculation unit 131 is only the average nursing care certification reference time (total) of the two directly managed research institutions 01 and 02 shown in FIG. In addition, the standard deviation calculation unit 132 obtains the standard deviation σ from the average value μ obtained by the average value calculation unit 131 and the average nursing care certification reference time (total) xi of each research organization. From the average value μ and the standard deviation σ thus determined and the average nursing care certification reference time (total) xi of each research institution, the standard measure calculating unit 133 calculates the standard measure ti according to the above equation (1). Ask for each.

  If the research organization is private only and does not need to be compared with the directly managed research organization, use the average nursing care certification reference time (total) of all private organizations as xi, and calculate the mean value μ and standard deviation σ. Further, the standard measure ti may be obtained from the above formula (1) from these values and the average nursing care certification reference time (total) xi of each research institution.

  Next, the protruding value detection means 14 determines whether or not the standard measure ti of the average nursing care certification reference time (total) at each research institution obtained in this way is outside the preset threshold range (step 703). . That is, as shown in FIG. 3, a threshold value (± 1.96σ) for specifying the planned range is set in the protruding value detection means 14, and each survey organization obtained by the standard measure calculation means 13 is set. The standard measure ti of the average nursing care certification reference time (total) is input, and out of these (step 703: Yes), the survey organization data of the standard measure ti is converted into the average nursing care certification reference time ( Detected as total value).

  In the example of FIG. 9, with respect to the threshold value ± 1.96σ, the research institution 7 has a standard measure of 2.80, so it has deviated upwards. It will protrude upwards compared to the research institution. That is, the survey organization 7 is detected as a survey organization with an average required nursing care certification reference time (total).

  Next, in order to verify the details of the probing of the research institution 7 where protrusions were detected, prominence about the average long-term care certification reference time (by category) constituting the average long-term care certification reference time (total) Perform value check. That is, as shown in FIG. 10, the average nursing care certification reference time (by category) is “meal”, “excretion”, “movement”, “cleanliness maintenance”, “indirect care”, “BPSD”, “functional training”. ”And“ Medical-related ”, and“ reference time ”,“ average ratio ”, and“ standard measure ”are required for each of these categories. Therefore, it is determined which of these sections protrudes (step 704).

  In the example of FIG. 10, the standard measure of “meal” is 2.33 and the standard measure of “medical” is 2.06, which is outside the range of the threshold value ± 1.96σ. “Medical related” is extracted as a protruding segment (step 705). That is, in the survey organization 7, the average nursing care certification standard time (total) was prominent because the average nursing care certification standard time (by category) for the two categories “meal” and “medical related” was prominent. It turns out that.

  Here, the nursing care certification standard time (by category) is “meal”, “excretion”, “movement”, “cleanliness maintenance”, “indirect care”, “BPSD”, “functional training”, “medical-related” It is known that each of the eight sections is calculated using a tree diagram.

  FIG. 13 is a tree diagram for calculating the nursing care certification reference time (by category) for the category “meal”. Such tree diagrams are also prepared for the other seven categories. The nursing care certification reference time (by category) is calculated using the tree diagram corresponding to. Hereinafter, for the category “meal”, a method of calculating the nursing care certification reference time (by category) will be briefly described using the tree diagram of “meal” shown in FIG. 13.

  The tree diagram has a plurality of judgment items (id numbers [0] to [18] attached) arranged in a tree shape, and the final end after passing through these judgment items is called a leaf. This leaf is set with a nursing care certification reference time (minutes) and individual ids (id numbers [19] to [38]). For example, the leftmost leaf in the drawing has a nursing care certification reference time of 18.6 minutes and id is [19]. By applying the corresponding items of the nursing care certification survey results to the judgment items of this tree diagram, the time set for the last leaf reached is the certification of nursing care certification in the category “meal” of the subject requiring nursing care certification survey. Calculated as the reference time (by category).

  Note that among the determination items, items having a prefix Q are set with determination criteria corresponding to the number of evaluation levels of the items. In addition, for items that are not prefixed with Q and are enclosed in square brackets, the intermediate evaluation point itself is set as a criterion.

  For example, “Q meal intake” with id [0] has four evaluation levels (not shown), ie, “1. Not canceled, 2. Watching, etc. 3. Partial assistance, 4. Full assistance”. If the survey result is “2. watching etc.”, it is smaller than the criterion “2.5”, so the process moves to the left in the figure, and then the “life function” of id [1] is determined. Since “life function” is an aggregate of a plurality of intermediate evaluation items, the intermediate evaluation score “31.25” is a criterion. If the intermediate evaluation score of the survey result is smaller than “31.25”, the process moves to the left in the figure, and the “cognitive function” of the next id [3] is determined. Since “cognitive function” is also a collection of intermediate evaluation items, the intermediate evaluation score “40.35” is a criterion. If the intermediate evaluation score of the survey result is smaller than “40.35”, the process moves to the left in the figure and reaches the leaf of id [19] in the final stage. Therefore, the nursing care certification reference time (by category) in the “meal” category of this survey target person (nursing care certification applicant) is calculated as “18.6 minutes”.

  The average nursing care certification standard time (by category) of each research institution shown in FIG. 10 is calculated based on the nursing care certification standard time (by category) of each individual requiring nursing care certification calculated as described above. Separated and averaged. In addition, the certification time required for long-term care (by category) is integrated with the id of the terminal leaf of the tree diagram in that category. It is a collection of leaf ids in the corresponding “category” of each individual who needs to be certified for long-term care at the research institution.

  Therefore, by verifying the distribution of the frequency of occurrence of leaf id in the collection of leaf ids in each category that is the basis of the average certification required for long-term care (by category) ) Can be found. This factor is verified by the chi-square calculation means 15 and the chi-square test means 16 in which the occurrence frequency of leaves in the protruding section is different from the occurrence frequency of leaves in the same section of an ideal model research institution. Or verify by

  Here, as mentioned above, the model research institution is an ideal research institution that conducts appropriate accreditation surveys. For example, thorough education and guidance for investigators and appropriate accreditation surveys are conducted. Is a research institution that has been verified in advance, and has been selected in advance from a nationwide scope by a management oversight body.

  In performing the above verification, the frequency of occurrence of each leaf id in the tree diagram, which is the basis of the average nursing care certification reference time (by category), is obtained for each research institution. In other words, the frequency of occurrence of leaves, each having an individual id, which is a plurality of occurrence events in the verification target research organization, is obtained. Then, as shown in FIG. 11, the frequency of occurrence for each id in the verification target category (here, “meal”) of the preset model research organization and the private research organization (for example, the research organization 7) to be verified. The chi-square value X ^ 2 [id] for each id is obtained from the occurrence frequency A for each id in the same section, and the chi-square value X ^ 2 of the corresponding section is obtained by summing the chi-square values for each id. Ask for.

  In other words, the chi-square calculating means 15 extracts the actually measured power A [id] for each id in the verification inspecting institution by the actually measured power extracting section 151 for each id shown in FIG. Further, the theoretical frequency calculation unit 152 by id uses the actually measured frequency by id of the model research institution to calculate the theoretical frequency B [id] by id in the verification target research institution 7 by the above-described equation (2).

  Then, the id-specific chi-square value calculation unit 153 uses the actually measured frequency A [id] for each id and the theoretical frequency B [id] for each id in the verification target research institution to determine the chi-square value X ^ 2 for each id. [Id] is calculated by the aforementioned equation (3). Further, the chi-square value calculation unit 154 adds the chi-square value X ^ 2 [id] for each id to obtain a chi-square value X ^ 2.

  Next, the chi-square test means 16 shown in FIG. 5 sets the chi-square value X ^ 2 for each category (in this example, for each “meal” category) obtained by the chi-square computing means 15 and preset. Based on the significance level α, it is determined whether there is a difference in data for the classification corresponding to the model research institution and the verification target research institution.

  The chi-square test means 16 tests whether the null hypothesis is rejected or adopted with respect to the set null hypothesis “There is no difference between the model research institution and the target research institution”. As shown in FIG. 5, the chi-square test means 16 is set with a degree of freedom n and a significance level α. In the example of FIG. 11, since there are 20 ids, the degree of freedom n is 19. The significance level α is set to 20% (α = 0.2), 10% (α = 0.1), 5% (α = 0.05), 1% (α = 0.01), Test whether the null hypothesis is rejected or adopted at the significance level.

  Here, when the degree of freedom is 19, the threshold for testing is 23.900 when the significance level is 20%, 27.204 when 10%, 30.144 when 5%, and 36.191 when 5%. That is, at a degree of freedom of 19 and a significance level of 1%, the probability of being greater than the chi-square value X ^ 2 = 36.191 is less than 1%. In FIG. 11, the chi-square value of a private research institution is X ^ 2 = 48.825, and the P value, which is the probability that this chi-square value will appear, is 0.02%, which is less than the significance level of 1%. Therefore, the private research institution in Figure 11 rejected the null hypothesis “There is no difference between the model research institution and the target research institution” and the alternative hypothesis “There is a difference between the model research institution and the target research institution” was adopted. The

  Next, with respect to a private research institution in which a deviation detected in this manner exists, it is checked which id deviation is most prominent. As a method for checking the protrusion id, an average value μ and a standard deviation value σ of X ^ 2 [id] are calculated, and a standard measure z of each id is calculated. Then, it is checked whether or not the calculated standard measure z is rejected at a preset significance level. It can be judged that the possibility of being rejected at a low significance level is more prominent. In the example, the significance level is lowered from 36% ⇒10% ⇒5%.

  That is, the standard measure calculating means 13 calculates the chi-square value X ^ 2 [id] for each id for the corresponding classification of the verification target research institution determined that the chi-square test means 16 has a difference in data with the model research institution. The average value μ and the standard deviation value σ are obtained. Then, the standard measure z for each id is calculated from the chi-square value X ^ 2 [id] for each id and the average value μ and the standard deviation σ according to the equation (1). However, X ^ 2 [id] is substituted for xi in Equation 1, and ti is replaced with z [id].

  Next, out of the standard measure z [id] for each id obtained by the standard measure calculating unit 13, the standard measure id that is out of the preset range is detected by the salient value detection unit 14 as an id with a particularly large deviation. To do. For this reason, the significance level of 36%, 10%, and 5% is set as the threshold value S in the protruding value detection means 14. In the private research institution of FIG. 11, id28 is deviating from the significance level of 36%, and is rejected at this significance level of 36%. In addition, id32 deviates from any of the significance levels of 36%, 10%, and 5%, and is rejected at the significance levels of 36%, 10%, and 5%, respectively. That is, it can be seen that these two ids have a larger amount of deviation than the other ids, and these two ids (id28, id32) are the cause of the protrusion.

  The example of FIG. 12 shows a case where the null hypothesis “the model research institution is not different from the target research institution” is adopted. Although the target research institution is directly managed here, it may of course be a private research institution. Also in this case, the actually measured frequency for each id of the model engine is the same as in FIG. The theoretical frequency B of the directly managed research organization to be verified is calculated from the frequency of occurrence for each id in the verification target category (here, “meal”) of this model research organization, and for each id in the same category of the directly managed research organization. The chi-square value X ^ 2 [id] for each id is obtained from the occurrence frequency A and the theoretical frequency, and the chi-square value X ^ 2 for the corresponding section is obtained by summing the chi-square values for each id. In the example of FIG. 12, the chi-square value of the “meal” category is X ^ 2 = 5.563.

  Is there a difference in data in the corresponding categories between the model research institution and the verification target research institution from the chi-square value X ^ of the “meal” category and the preset significance level α? . That is, the chi-square test means 16 tests whether to reject or adopt the set null hypothesis “There is no difference between the model research institution and the target research institution”. The degree of freedom set in the chi-square test means 16 is n = 19, and the significance level α is 20% (α = 0.2), 10% (α = 0.1), 5% (α = 0. 05) 1% (α = 0.01).

  The chi-square value of the directly managed research organization in FIG. 12 is X ^ 2 = 5.563, and the P value, which is the probability that this chi-square value appears, is 99.88%, which is higher than the significance level of 20%. Therefore, the directly managed research organization in FIG. 12 is adopted without rejecting the null hypothesis “the research method (result) between the model research organization and the target research organization is not different”.

  As described above, in the present invention, it is possible to verify the accredited survey result by checking whether or not the standard time required for nursing care certification of a certain research institution is prominent compared with the average value of the population of the research institution. . In addition, if an oversight can be confirmed, the cause can be checked step by step for more detailed items, the oversight factor can be clarified, and the investigation method and judgment of the relevant research institution can be examined. It will be possible to provide guidance and explanations to research institutions.

DESCRIPTION OF SYMBOLS 11 ... Authorized data storage means 12 ... Exclusion means 13 ... Standard measure calculation means 131 ... Average value calculation part 132 ... Standard deviation calculation part 133 ... Standard measure calculation part 14 ... Protrusion value detection means 15 ... Chi-square calculation means 151 ... By id Measured frequency extraction unit 152 ... theoretical frequency calculation unit by id 153 ... chi-square value calculation unit by id 154 ... chi-square value calculation unit 16 ... chi-square test means

Claims (2)

Accreditation data storage means for holding the accreditation data including the nursing care accreditation standard time, which is the result of a survey requiring accreditation for nursing care conducted in the past, in a grouped state for each research organization that conducted the survey,
Using the authorization data stored in this authorization data storage means, for each research institution, the average nursing care certification reference time is extracted, and the average value and standard deviation of each of these average nursing care certification certification reference times are obtained. A standard measure calculating means for obtaining a standard measure for the average nursing care certification reference time for each research organization from the standard deviation and the average nursing care certification standard time for each research organization;
An accredited research institution verification system comprising: probing value detecting means for detecting a prominent research institution in which the standard measure of the average required nursing care reference time obtained by the standard measure calculating means is out of a preset range. The authorization data is read from the authorization data storage means that holds the authorization data including the nursing care certification reference time, which is the result of the nursing care certification survey conducted in the past, in a grouped state for each research organization that conducted the survey. ,
The standard measure calculation means extracts the average nursing care certification reference time for each research institution from this read authorization data, and calculates the average value and standard deviation of each of these average nursing care certification reference times. And the standard deviation and the average measure for certification of long-term care required for each research institution from the standard deviation and the average standard of certification for long-term care required for each survey institution, respectively.
An accredited research institution verification program for causing a computer to realize a function for detecting a probing research institution in which the standard measure of the obtained average standard value of certification required for long-term care is out of a preset range by means of the outlier detection means.

Images