An audit of coded data - Assignment Example

Auditing Report Executive Summary This paper details an audit of coded data, taking off from 40 data points from data coded by a representative coder A and compared against a reference set of data that is correct, and labeled for the purposes of the audit as coder B data. This audit complies with the protocol prescribed in the ACBA, or the Australian Coding Benchmark Audit. The audit undertakes an analysis of the variances between the actual data in the coder B data set from the data in the coder A data set. The findings are that in general, coder A data is flawed, with near unity error rates for all data except for two data points. Moreover, in terms of the four categories of data examined, the error rates ranged from 40 percent to 75 percent, with the AD coding error rates being the highest. The analytical findings indicate that there is a need to overhaul the coding process. Further steps recommended include doing more iterations of the audit, with more coders, with more data points, and with more refined analytical processes that refine the variance analysis to include different forms of errors in the coding process for the different categories of data (Thompson, 2010; Pavilion Health, 2012; World Health Organization, 2002; Government of Western Australia, 2014).. Table of Contents Executive Summary 1 I. Introduction 3 II. Rationale 4 III. Methodology 5 IV. Reporting Categories 6 V. Scoring 7 VI. Results 8 A.General Results 8 B.Coding Mistakes by Data Category 11 C.Error Rate by Data Category 13 D.Error Frequency by Data Category 14 E.Coding Accuracy 15 F.Coding Errors by Defined Error Categories 16 VII. Interpretation of Results 18 VIII. Recommendations 19 IX. Follow-Up 21 X. Results Dissemination 21 XI. Conclusion 22 XII. Appendices 23 Appendix A: Variances Between Coder A Data and Coder B (Reference/Correct) Data 23 XIII. References 32 I. Introduction This report details the results on an audit of coder A data based on a set of reference data containing the correct inputs as present in Coder B data. This is to determine the accuracy of coding data, with the audit being in line with the prescriptions of the ACBA, or the Australian Coding Benchmark Audit. The basic method consists of analyzing coder A data for variances with the reference data as provided in coder B. Coder B data is correct, against which the accuracy of coder A data is vetted against. This is the fundamental method. The use of 40 records for the audit is in line with the ACBA prescription for audit sample size to be able to make a good assessment of the quality of the coding (The University of Queensland, 2014). The ACBA audit process, in theory, is a convenient, practical and effective process for identifying coding error sources, not only from the coder side, but also from the wider system side, and in so doing allow for a more comprehensive view of the sources of and potential mitigations for the correction of coder and process/system mistakes in the coding process. ACBA in other words also shores up the reliability of coded data, by tracing coding mistakes in a comprehensive and systematic fashion. ACBA also makes it easier to understand that the coding process and coding mistakes that follow from certain coding processes can be as much a source of coding error as human error. Stated another way, in the ideal scenario, correcting the process in coding can also translate to fewer mistakes due to human coding error, if not totally eliminating the human factor in coding by suitable interventions, workarounds, and inherent double checks in the entire coding process. This paper then details the results of an audit conducted on the 40 sample data, as encoded by coder A, and as validated against the correct reference data from coder B (World Health Organization, 2002; Government of Western Australia, 2014). II. Rationale There is a need to make sure that clinical data is coded accurately because of the way such data is used in various decision-making and research purposes both at present and moving forward. Case histories, when they are properly coded, can be used as reliable bases for optimizing the use of resources, and for priming the care setting to be ready for the most probable cases that are to appear on a regular basis. Historical data can be a minefield of insights into the nature of healthcare needs of the immediate population and of the usual client base of the health care facility. Such data therefore, needs to be properly coded and guarded against errors, both from the system side, and from the side of the human coders, who may be both susceptible to making errors in coding, depending on the processes and other mitigating circumstances surrounding the coding process. A proper coding audit that is in compliance with the ACBA processes aids tremendously in ensuring that data is accurately coded, for the uses that have been outlined above, and for any other practical uses for historical data that management, administrators, and the caregivers can come up with moving forward. As well, the audit also allows for the stress testing of current coding methods and processes, and the stress testing of the coding work and coders as well, by way of determining how well current processes are able to accurately process data. The results of such an audit can yield valuable insights into process re-engineering efforts, for instance, to improve coding processes and the accuracy of coded data (Thompson, 2010; Pavilion Health, 2012). III. Methodology The ACBA prescribed process for coding data is ideal in that it is able to allow for double checking of data coded, and allows for a process to resolve errors in coding within the process through redundant coding by different sets of coders. The different coders then are able to validate each others’ work through a process that pits the results of independent coding efforts against each other, with an arbiter that basically also acts as a mechanism for resolving any coding errors. The results of a redundant coding effort is that any errors in coding are caught in the system, so to speak, and corrected before the final data is considered to be accurate for safekeeping and later retrieval and use. The method here on the other hand consists of undertaking an audit of existing coder A data from data from coder B, with the latter being the reference for the correct set of data for an analysis on variances and errors in coding. The bulk of the report details the report of the audit and the analysis of the variances between coder B data and coder A data. For both coder A and B, a common coding template was put to use, and the arbitration process for any mistakes then can proceed from the use of coder B data as a reference, and determining where coder A data deviated from the standard reference data from coder B. In this process moreover, the results of the audit are such that data coder B is not made privy to the results of coder A’s efforts until all of the data had been coded in (Thompson, 2010; Pavilion Health, 2012; World Health Organization, 2002; Government of Western Australia, 2014). As prescribed in the ACBA, the present audit made use of 40 records that were encoded by both coder A and coder B, with again coder B data, for the purposes of this report, being the accurate set of data. This simplifies the audit process in terms of being able to identify how well the coding process as it is is able to correctly code the data and to avoid coding errors from the coding process in use by coder A (World Health Organization, 2002; Government of Western Australia, 2014). IV. Reporting Categories to simplify the classification of coding errors for the purposes of this report, coding errors are classified according to mistakes in coding either of the four coding categories that were considered in the coding data, and those are PD, AD, PP, and AP. Coding mistakes then are designated corresponding to the data categories in the comparison, and mistakes are denoted by the suffix x, so that coding mistake categories are denoted as PDx, ADx, PPx, and APx. These are the designated categories of audit for this exercise. The categories are determined by the nature of the data provided by the exercise. Where the entry sequence is wrong, and the entry is missing, the audit categories as presented here are unable to capture them. The generic audit categories will capture any mistakes in coding, either by way of incorrect entry, incorrect sequence, and missing data. This further simplifies the audit, but also limits the power of the audit analysis in that it does not differentiate mistakes in coding according to those categories that have not been broken down. Any mistakes in coding for each of the coding mistake categories are lumped together into either PDx, ADx, PPx, and APx as already enumerated above (World Health Organization, 2002; Government of Western Australia, 2014). V. Scoring For the purposes of the audit, which mainly focused on variances between the coder A data and the correct coder B data, all of the categories of clinical data are given equal weights, and in turn all of the categories of mistakes in coding, all four of the data categories present in the coder A data set, are likewise given equal weights. By this is meant that in the absence of any other data on the coder A data set, any mistake in any of the categories of data in the data set are given equal significance. The analysis makes this equal importance rule apparent. Equal emphasis is given to the different analyses for variances between coder A and coder B data categories, with the result being that in general all mistakes in the coder A data set are captured and given the proper analytical treatment. Moreover, the equal treatment of the four categories of data in the analytical framework used for this report translates to all four of the data categories being given equal weight in the synthesis of the overall variances of the coder A data from the coder B data. In the concrete, PDx, ADx, PPx, and APx data are given equal weight in a composite measure of the total variance of coder A data from the true values, for evaluation purposes. None of the categories stand out as more important than any other as far as analytical treatment is concerned (Thompson, 2010; Pavilion Health, 2012). VI. Results A. General Results In general, out of the population of 40 records for the analysis, only two records did not have any errors. This corresponds to a failure or error rate of 38 records out of 40, or close to unity. One can say that the accurate records are the outliers, with the rule being that virtually all of the records have mistakes and that the coding process in use by coder A has a very high degree of failure and mistakes. The error rate is 95 percent, or to put it another way, the success rate for the coding work is just 5 percent out of the sample of 40 records. Assuming that the sample population is adequate to make generalizations about the entire data set and the entire coding process, one can say that the coding process is a failure, having an unacceptably high failure rate of 95 percent (Thompson, 2010; Pavilion Health, 2012; World Health Organization, 2002; Government of Western Australia, 2014). The table below details a general overview of the rate of errors in the sample data Data Entries Without Errors 2 5 Percent Data Entries With Errors 38 95 Percent Total 40 100 Percent A bar graph would illustrate the wide disparity between the number of records that are entirely accurate on the one hand, and the majority of records that have one or more wrong entries in one or more of the data categories that are relevant in the coding process. Given that the number of wrongly-coded entries overwhelms, one can imagine the result, n a large bar for wrongly coded data and a miniscule bar for correctly coded data. The bar graph is below: Represented by a pie graph, the glaring disparity between correctly coded and those with errors is more obvious: From this macro perspective, one can see that the coding process is flawed, or the coder is incompetent, or both. This is an opportunity for the organization to examine the processes as well as the coder competence, to determine the root causes of the large rates of errors in the coding process currently in use. Having determined the poor results in this audit, the next step is to be able to come up with interventions to radically shore up accuracy of the coding effort (World Health Organization, 2002). B. Coding Mistakes by Data Category As discussed earlier, there are four categories of data corresponding to the data categories available in the coder A data set, and the mistakes are tracked according to the data category. The errors are then denoted by PDx, ADx, PPx, and APx, for each of the four tracked data categories of PD, AD, PP and AP (Thompson, 2010; Pavilion Health, 2012). For PDx, there are 18 out of 40 records that are relevant, or about 45 percent of all records that are found by the audit to be in error (Thompson, 2010; Pavilion Health, 2012). For ADx, there are 30 out of 40 records that are relevant, for an error rate of 75 percent (Thompson, 2010; Pavilion Health, 2012). For PPx, there are 16 out of 40 records that are relevant, for a failure or error rate of 40 percent (Thompson, 2010; Pavilion Health, 2012). For APx, there are 19 out of 40 records that are relevant, for an error rate of close to 50 percent (Thompson, 2010; Pavilion Health, 2012). Tabulated, the data looks like this: With Coding Errors W/out coding Errors Error Rate (%) PD 18 22 45 AD 30 10 75 PP 16 24 40 AP 19 21 47.5 Graphically, the data above looks like this: In the above table and graphs the error rates for each of the data categories is simply the number of entries with wrongly coded data divided by the total number of data entries considered, in this case the population being 40 sets of data(Thompson, 2010; Pavilion Health, 2012).. Looking at the individual data categories another way: C. Error Rate by Data Category As can be gleaned from the above data, for most of the categories the error rate hovers between 40 and 50 percent, while for AD, the error rate is very large, at close to 75 percent. These results complement the overall findings of a very high error rate for all of the records considered. As already discussed above, overall just two out of 40 records have no errors whatsoever (Thompson, 2010; Pavilion Health, 2012). Honing in on rate of coding errors per data category, the large rate of errors for AD relative to the rate of coding error for the other data categories further stand out: From the above it is clear that only AD has a rate of coding error that exceeds 50 percent, while the others hover below that or close to that, with the floor being at 40 percent for error rates for all data categories(Thompson, 2010; Pavilion Health, 2012). D. Error Frequency by Data Category In terms of error frequency for each of the data categories, rather than percent, the data indicates that the frequency of errors occurs within a range of 16 and 30 coding errors for all data categories(Thompson, 2010; Pavilion Health, 2012).: E. Coding Accuracy The flipside chart below meanwhile details frequency of correct coding for all data categories, indicating some success in the coding process that is not evident from looking at the general error rate for all entries, where the error frequency is 38 out of 40 records, or close to unity(Thompson, 2010; Pavilion Health, 2012).: \ The take from the above chart is that there is a large measure of success coding PP data, as indicated by the high rate of correct coding frequency and the high absolute number of correctly coded data entries/records in the sample population. This input can matter as much as the inputs relating to coding errors in determining what aspects of the coding process contribute to the correct coding of the records and in particular of the high coding accuracy relatively speaking for PP data (Thompson, 2010; Pavilion Health, 2012). F. Coding Errors by Defined Error Categories Where applicable and where data is available, this analysis included looking into coding errors by the error categories for incorrect coding, sequencing and selection for each of the data categories PD, AD, PP and AP. The tabulated results are as follows (Thompson, 2010; Pavilion Health, 2012): Error Type Frequency Error Rate (%) Incorrect sequencing of PD CMjDx 1 0 0 Incorrect coding of PD CMjDx 2 18 45 Incorrect selection of PD CMjDx 3 18 45 Incorrect AD CMnDx1 30 75 Graphing the results, we get the following: VII. Interpretation of Results What stands out from the above results is that there are very large gaping holes in the encoding processes, as reflected in the near-total presence of errors in coding in all but two of the records considered in the sample population of 40 records. Breaking down the errors in coding by the four categories considered, one can see moreover that there are large errors in coding in all of the four data categories, suggesting that the process shortcomings and/or the encoder shortcomings or both are pervasive, to the extent of universally affecting coding accuracy and coding quality in general. PD, PP and AP error rates are close to one another, hovering between 40 percent to half of all records in the case of AP, with AD being the outlier with an error rate for coding of 75 percent. Taking a step back, moreover, one sees that in general, the correctly coded entries can be considered as the outliers in this context, with majority of the coded data for coder A being generally erroneous, that an audit is really an exercise in validating that the entire coding process needs to be revamped altogether, and the core causes of the errors can be attributed either wholly to the process or to the coder, or both. One way to be sure that the process is at fault is to try more iterations of the coding process with a number of different coders, to determine whether or not the results of this audit are not due to the coders but due to the processes for coding itself. the ACBA audit should be able to provide insights into the latter in theory, and in the case where the coder B data are culled using the same existing coding process, then one can say that the errors in coding can be attributed to the lack of competency of coder A (World Health Organization, 2002; Government of Western Australia, 2014). In reference to the differing error rates for the categories of data as discussed above, the outlier high rate of error in the coding of AD data, at 75 percent in comparison to between 40 and 50 percent for the other three data categories, needs to be investigated further. If the mistakes are process-generated, for instance if the encoding process causes coders to make mistakes in the coding of AD data in unusually high rates of occurrence, then this audit presents proof of the need to investigate and probably re-engineer the coding process for AD data in particular. This applies to a lesser degree to those categories of data where the rate of error are significant, but not as large as the AD data coding failure rates. Again, there is a need to investigate to what extent the coding errors are due to the processses, and to what extent the mistakes are due to coder competence or lack of competence in this case. At any rate, what is apparent from the audit is that the data coding process is flawed in some fundamental sense, that needs to be further investigated via further iterations of the coding process. Iterations, as discussed above, can take the form of more coders going through the audit process, to further hone in on the shortcomings in processes as well as shortcomings in the coder capabilities and competence (World Health Organization, 2002; Government of Western Australia, 2014). VIII. Recommendations The preceding discussion already broached further steps to identify, validate and confirm the sources of error in the coding process, given the general prognosis that the entire process is flawed in some fundamental way. The rates of error are close to unity in general, and abnormally high for all the categories of data presented. Most of the data in coder A are wrongly coded, with just two out of 40 records being correctly coded. More iterations are needed with more coders to identify shortcomings with the process, or with the coders, or both. The object of the iterations is to confirm the presence of process and coder errors in the coding of the various categories in each entry (World Health Organization, 2002; Government of Western Australia, 2014). Further recommendations are in the area of going beyond the minimum ACBA data set sample size in order to make more accurate insights from the analysis of discrepancies in coding and actual data in new audits. This is because it may be that a larger sample size can yield more accurate insights. This is a variation of the recommendation to add more coders into an investigation that involves more iterations of the audit. More iterations with greater sample sizes should be aid in determining the weaknesses in the coding processes and in the quality of the coders doing the work (World Health Organization, 2002; Government of Western Australia, 2014). A final set of recommendations has to do with widening the scope of the included data in the audit, and refining the analysis to include more error forms, from just general mistakes in coding that have been examined in this analysis. For instance, mistakes can take the form of wrong sequencing and missing data. These error forms are not investigated in this audit. A refined analysis will yield more insights into the nature of the mistakes in coding currently present in the processes and the coders (World Health Organization, 2002; Government of Western Australia, 2014). IX. Follow-Up Follow-up steps in this audit involves doing more iterations of the audit, as described above, to include more coders, refining the analysis to include different forms of mistakes in coding, and to widen the sample size of data entered to be able to get a more detailed picture of the nature of the problems associated with large mistakes in data coding unearthed in this audit (World Health Organization, 2002; Government of Western Australia, 2014). X. Results Dissemination The results of this audit are to be disseminated to the entire coding team and the management team overseeing the coding work, as well as to the final users of the coded data. This is to make sure that all stakeholders are made aware of the issues associated with the coded data, and to be able to make suggestions to improve the quality of the coded data moving forward (World Health Organization, 2002; Government of Western Australia, 2014). XI. Conclusion It is clear from the audit of the coded data that there are fundamental flaws in the coding processes that need to be addressed. The rate of error in the coding work is very large and close to unity, indicating a general failure in the coding process. More work needs to be done in terms of refining the audit, including more iterations with more coders and with more data, to be able to get to the heart of the problem (World Health Organization, 2002; Government of Western Australia, 2014). XII. Appendices Appendix A: Variances Between Coder A Data and Coder B (Reference/Correct) Data B. Case 10452 PD: K01.1 AD: --- PP: 97323-04 (458) AP: 92514-19 (10910) – 92514-19[1910] C. Case 10950 PD: I83.9 AD: --- - Z92.1 PP: 32508-00 (727) AP:92514-99 (10910), 95550-02 (1916) – 32508-00[727], 92514-29[1910] D. Case 10707 PD: M23.81 – M23.51 AD: T94.1, Y89.9, Y29.9 - M22.3, M23.23,M23.25 PP: 49560-03 (1503) AP:49560-02 (1501), 92514-99 (1910), 95550-03 (1916) - 49560-02[1501], 92515-19[1910], 92508-19[1909], 95550-03[1916] E. Case 10612 PD: I88.0 - I88.9 AD: --- - A09.9 PP: 30571-00 (926)- 92514-19[1910] AP: 92514-99 (1910) - 92514-19[1910] F. Case 10901 PD: S41.0 - S41.80 AD: S40.81, V89.2, Y 92.9, U72 - V48.59,Y92.40, U73.9 PP: 30023-00 (1566) - 90665-00[1628] AP: 92514-19 (1910) G. Case 10600 PD: K43.2 - K43.9 AD: – - K66.0, T81.2, S36.40, Y60.0, Y92.22, U73.8, Z72.0 PP: 30403-00 (993) - AP: 92514-99 (1910) - 30378-00[986], 30375-19[901], 92514-29[1910] H. Case 10611 PD: K36.0 - K35.9 AD: --- - K36, K66.0, Z72.0 PP: 30572-00 (926) AP: 92514-99 (1910) - 30393-00[986], 92514-99[1910] I. Case 10553 PD: J44 AD: Z72.0 - F17.1 PP: 92209-02 (570) - 92209-01[570] AP: --- J. Case 10902 PD: S81.88 - T63.0 AD: T89.00, X20.09, Y92.9, U73.9 - X20.08, Y92.9, U73.9 PP: --- - 90665-00[1628] AP:---- - 92514-19[1910] K. Case 10900 PD: T20.2 - T24.2 AD: T24.2, T21.22, T22.22, T31.00, X12, Y92.04, U73.8 - T20.1, T21.12,T31.00, X12, Y92.04, U73.9 PP: 96092-00 (1870) – blank AP blank blank L. Case 10903 PD: T23.0 - T22.21 AD: T22.01, T31.00, X10.2, Y92.04, U73.1 - T31.00, X10.2,Y92.04,U73.2 PP: --- AP:--- M. Case 10850 PD: P59.09 - P59.9 AD: Z38.0 PP: 90677-00 (1611) AP:--- N. Case 10609 PD: K40.20 AD: Z72.0 PP: 30609-03 (990) AP:92514-99 (1910) - 92514-29[1910] O. Case 10904 PD: T43.2 AD: T43.69, T43.3, T51.8, X61, X65, X92.9, U73.9, F39.90, Z91.5, F12.1, - X61,T43.3, X61,T43.69, T51.0, X65, Y92.09, U73.8, F32.90, Z63.8,Z91.5 PP: 95550-11 (1916) - blank AP:---- P. Case 10705 PD: S22.42 - S22.43 AD: W19, Y92.9, U73.8, J98.1, J90 - W19,Y92.9,U73.9, G40.90,Z72.0 PP: --- - 95550-09[1916] AP: ---- Q. Case 10704 PD: G56.0 AD: --- PP: 39331-01 (76) AP: 92519-29 (1909) - 92515-29[1910], 92519-29[1909] R. Case 10551 PD: J22 AD: J45.9, E10.9, Z86.43, Z86.41 - J45.9, Z86.43 PP: --- AP:--- S. Case 10500 PD:I21.0 AD:Z72.0 PP: --- AP:--- T. Case 10607 PD: K29.30 - K29.5 AD: --- - K31.7 PP: 30473-01 (1008) AP:92515-39 (1910) U. Case 10606 PD: K36 AD: --- PP: 30572-00 (926) AP:92514-13 (1910) - 92514-19[1910] V. Case 10613 PD: E66.9 AD: --- K66.0, Z86.43 PP: 30511-02 (889) - 30511-01[889] AP: 92514-39 (1910) - 30393-00[986], 92514-39[1910] W. Case 10608 PD: K55.9 AD:K26.9, D64.8 - K26.4, D62 PP: 13706-02 (1893) AP: 95550-03 (1916), 95550-01 (1916) - 95550-02[1916] X. Case 10604 PD:K63.2 - R19.4 AD:Z72.0 - T81.8,S36.54,Y83.8,Y92.22, U73.8,Z72.0 PP:30375-23 (907) - 30473-00[1005] AP: 30375-04 (915), 92514-40 (1910), 320920-00 (905), 92515-29 (1910) - 32090-00[905], 92515-29[1910], 32000-00[913], 92514-40[1910] Y. Case 10603 PD:Z12.1 AD:Z80.0 PP:32090-00 (905) AP:92515-29 (1910) Z. Case 10655 PD: Q38.8 - N64.8 AD: --- - Z41.1 PP: 45528-00 (1753) - AP: 92514-29 (1910) - 45588-00[1675],92514-29[1910] AA. Case 10605 PD: K57.30 AD: E11.9, Z72.0 - K63.50, Z72.0 PP:32090-00 (905) AP: 92515-29 (1910) - AB. Case 10906 PD: T88.7 - D68.3 AD:Z44.2, Z92.22, I48.0, I35.10 - Y44.2, Y92.22, U73.8 PP: --- - 92062-00[1893] AP:--- AC. Case 10706 PD: S72.03 AD:W19, Y92,09, U73.9, G30.9, F00.9, - W19,Y92.14, U73.9, G30.9,F00.9* R32 PP: --- - 95550-09[1916] AP: --- AD. Case 10602 PD: K21.0 – K20 AD: E11.9, Z72.0 - K29.5, K63.58, D12.8, Z72.0 PP:32093-00 (911) - AP: 30473-01 (1008), 92515-29 (1910) - 30473-01[1008], 92515-39[1910] AE. Case 10453 PD: H65.3 AD: --- PP:41644-00 (312) - 41632-01[309] AP: 92514-19 (1910) AF. Case 10852 PD:Z38.0 - P22.1 AD:--- PP:--- AP:--- AG. Case 10401 PD:H26.9 AD:--- - Z96.1 PP: 42702-04 (197) AP:92509-29 (1909) - 92515-29[1910], 92509-29 [1909] AH. Case 10610 PD: K60.2 AD: K64.0 - I84.2 PP: 90338-00 (930) AP:32135-00 (941), 92514-99 (1910) AI. Case 10554 PD: J22 AD:I48.9, E05.9, I10, R03, R32, R15 PP:95550-00 (1916) - I48, I10, E05.9, F03, Z86.43 AP:95550-05 (1916), 95550-02 (1916), 95550-03 (1916) - 95550-05[1916], 95550-03[1916], 95550-09[1916] AJ. Case 10712 PD: S73.00 - S73.01 AD:Z96.64, W19, Y92.01, U73.8 - W18.9,Y92.01,U73.9,Z96.64 PP: 47051-00 (1487) AP:92514-29 (1910), 95550-03 (1916), 95550-02 (1916), 95550-11 (1916) - 92514-29[1910], 95550-03[1916], 95550-02[1916] AK. Case 10654 PD: E65 AD: Z41.1 - blank PP:30165-00 (1666) - 30177-00[1666] AP:92514-29 (1910) AL. Case 10709 PD: S42.41 AD:W07.9, Y92.9, U73.9, - W07.9,Y92.09, U73.9 PP:47456-00 (1413) AP:92514-10 (1910) AM. Case 10708 PD: Z47.0 AD:--- PP:47927-00 (1554) - 47930-00[1554] AP: 92514- 19 (1910) AN. Case 10851 PD: P28.83 - P22.1 AD:Z38.0 - Z03.71, Z29.2,Z38.0 PP: --- - 96199-02[1920] AP: --- - 96199-07[1920] AO. Case 10703 PD: M17.1 AD:Z72.0 PP: 49534-01 (1518) - 49518-00[1518] AP:92508-29 (1909), 95550-03 (1916), 95550-02 (1916) - 92515-29[1910], 92508-29[1909], 95550-02[1916], 95550-03[1916] XIII. References Government of Western Australia (2014). Clinical Coding in Western Australia. Government of Australia Department of Health. Retrieved from http://www.clinicalcoding.health.wa.gov.au/about/ Pavilion Health (2012). Performance Indicators for Coding Quality (PICQTM) and Quality System Data Analysis. HIMAA. Retrieved from http://www.himaa.org.au/2012/Presentations/Tue%2030%20Oct/Coding%20Stream/PICQ%20Workshop%20-%20J%20Berry/HIMAA%20PICQ%20and%20Quality%20System%20Data%20Analysis.pdf The University of Queensland (2014).Module 5: Coding mortality data. Strengthening civil registration and vital statistics for births, deaths, and causes of deaths. Retrieved from http://www.uq.edu.au/hishub/docs/Resource%20Kit/Module%205%20-%20Coding%20mortality%20data.pdf Thompson, T. (2010).ICD-10-AM Classification for Morbidity Coding. Government of New Zealand. Retrieved from http://www.stats.govt.nz/~/media/Statistics/browse-categories/health/injuries/icd-coding-workshop/morbidity-coding.pdf World Health Organization (2002).Assessment of ICD-10 Coding in the South-East Asian Region. World Health Organization Regional Office for Southeast Asia. Retrieved from http://whqlibdoc.who.int/searo/2002/SEA_HS_220.pdf Read More