Family Practice © Oxford University Press 1992
Vol. 9, No. 4 Printed in Great Britain
General Practitioner Records on Computer—Handle with Care AEW GILLILAND, KA MILLS AND K STEELE
INTRODUCTION An increasing number of general practitioners utilize computerized medical records systems to store clinical data on their patients. Computerized data-bases should prove invaluable if not essential to general practitioners if they are to meet the conditions of their new contract. However, the data are of little use if incomplete. Although a number of studies have addressed the accuracy of computerized age/sex registers1'2 few have considered the completeness of the computerized clinical data.3"5 Practices vary as to how they collect the clinical information needed to keep their computerized records up to date. One method involves using an encounter form to record what happens during consultations. Any additions, amendments or deletions to be made to the computerized record are noted on this form which is then passed on to a computer clerk for keying in. Now that many general practitioners have desk-top computers in their surgeries it may be argued that the manual recording form is obsolete. However, not all doctors are willing to use the computer during or in
between consultations to record patient information on a regular basis. A limited encounter form may still be a necessary means of gathering data in general practice. Such an encounter form had been used by the two study practices for some years. As their computer system was not multi-user the option of direct entry by the doctors was not available. In addition to providing the study practices with information for their day to day service work, academic staff in the local department of general practice have, in the past, been able to access the patient data-base to look at several clinical areas, in particular prescribing.6'7 This had highlighted areas where information seemed incomplete and it was decided to look at this in more detail to see if the level of recording could be improved. The initial hypothesis was that recording of selected aspects of clinical information in the forementioned practices would be incomplete and could be improved by appropriate intervention. The study aims were 2-fold. 1. To determine by comparison with a manual record how accurately the doctors were recording certain laboratory investigations and the electrocardiograms (ECGs) they request.
Department of General Practice, Queen's University of Belfast, Dunluce Health Centre, 1 Dunluce Avenue, Belfast BT9 7HR. Correspondence to AEW Gilliland.
441
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
Gilliland AEW, Mills KA and Steele K. General practitioner records on computer—handle with care. Family Practice 1992; 9: 441-450. This 1 year prospective study involved nine general practitioners in an urban health centre who routinely record all patient contacts on computer. The study determines by comparison with a manual record how accurately doctors record laboratory investigations on computer and compares the effectiveness of three interventions in improving the completeness of computerized recording of presenting symptoms, problems/diagnoses and laboratory investigations. Recording was analysed for 1 month prior to and for two 1 month periods following each intervention. A control group was used. A total of 7983 patient contacts were analysed. Intervention led to an improvement in the recording of presenting symptoms and problems/diagnoses. Recording of investigations on the computer showed no improvement, remaining at one-third of the total in the treatment room book for both study and control doctors. The effectiveness of the different forms of intervention depended on both the aspect of the consultation considered and the familiarity of individual doctors with the method of data collection. Aspects considered less important required greater intervention to bring about a marked improvement, as did doctors relatively new to the practice. It may not be possible to get all aspects of the consultation recorded with the same degree of accuracy. This has implications for the accuracy of retrospective studies dependent on existing computerized data.
442
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
2. To determine the effect of three methods of intervention on improving the completeness of recording of presenting symptoms, problems/ diagnoses and investigations.
Informal Individual Meeting with no Feedback of Results The first meetings with the general practitioners were held on November 30 1989. They took the form of a quick informal request by the authors to individual doctors taking about 10 minutes after surgery. The purpose of the study was explained and each doctor was asked to try and improve their recording in the three areas of interest. No feedback was given at this time about their current level of recording. Each general practitioner was unaware of which of his colleagues were in the study group and was asked not to discuss the study with any of them. Informal Individual Meeting with Feedback The second meetings were held on March 22 1990. Each study doctor was seen individually for approximately 15 minutes. He or she was given their own results for November, December and February and shown how these compared with their colleagues (unidentified) in both the study and control groups. They were reminded about the purpose of the study, given further encouragement to improve and encouraged to discuss any problems they had with data recording. Formal Group Meeting with Feedback The third intervention at the end of July involved all the study general practitioners as a group. They were given feedback about both their individual and the overall study group recording levels as compared with the control group. The doctors were given the opportunity to discuss their individual results and encouraged to suggest reasons for the observed changes in recording. Their encounter recordings were subsequently analysed for August and October to assess the impact of the third group meeting. RESULTS The numbers of encounters recorded by the doctors each month are shown in Figure 1. The total for the study doctors (/» = 5) was surgery, 3715; home visits, 603 and for the control doctors (n = 4) surgery, 2957; home visits, 618. The study doctors saw a significantly higher proportion of their patients in the surgery in December \p < 0.01), June (P < 0.05) and October (P < 0.01). Presenting Symptoms The percentages of encounters for which one or more presenting symptoms were recorded are shown in
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
METHODOLOGY The study took place over a 1 year period commencing in November 1989 and involved nine general practitioners based in an urban health centre. Each doctor was assigned randomly to either a study or control group. The study group consisted of three full-time and two part-time principals and a trainee. The control group contained two full-time and one part-time principal and a trainee. The control group were unaware of the study and were left to record their consultations in the usual way. The researchers did not take part in the study. The study considered the doctor's recording of three distinct aspects of the consultation: presenting symptoms, problems/diagnoses and the following investigations—full blood picture (FBP), erythrocyte sedimentation rate (ESR), biochemical profile (SMAQ and ECGs. Throughout this paper the term encounter refers to an individual consultation. The encounter form used routinely by the study and control doctors contains a field which lists 11 common presenting symptoms— cough, headache, tiredness, chest pain, abdominal pain, dizziness, back ache, indigestion, rash, joint pain and disturbance of bowel function. These are based on Morrell's findings.8 If one or more of these applied to a particular consultation the doctor was asked to circle the relevant symptom(s). (This section was introduced 3 years prior to the study for research purposes.) The general practitioners were asked to record a problem/ diagnosis for every consultation. In addition, any laboratory investigation or ECG requested via the treatment room was to be recorded on the encounter form. The doctors recorded the above information routinely for all surgery and home visits. The nurses only took blood samples, swabs, etc. at the request of a doctor. Details of all investigations requested by the doctor were entered into a day book by the treatment room nurses. The two methods of data collection—the investigation field of the computerized encounter form and the treatment room day book— should therefore record the same events. We were able to compare the two sets of figures for each investigation for both the study and control doctors. This was only done for investigations requested during surgery consultations as very few investigations are requested following home visits. The study doctors were unaware of the study prior to the December 1 1989 when they were interviewed and encouraged to improve their recording in the three areas of interest. This enabled us to use the practice computer to determine what percentage of each doctor's consultations (both in the surgery and at home) had presenting symptoms, a problem/diagnosis
or a request for a laboratory investigation or ECG recorded throughout November 1989. Three forms of intervention were used in turn. Consultation data were analysed for the month immediately following each intervention and again for the third month afterwards. The three forms of intervention used were as follows.
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS
443
Study Doctors a M Surgery
l\\\N Homo
Numb«r of Encounters 700
Dae 88 Fab 80
Apr SO Jun SO
Aug 80 Oet BO
Month
Control Doctors Number of Encounters 700
NOV BS
Dae BS Fab SO
Apr 80 Jun 80
Aug so Oet so
Month FIGURE 1 Surgery and home encounters during study months.
Figure 2. No data are presented for March and July as a month was needed between the study collection periods in which to analyse and prepare the data for feedback to the doctors. In November, prior to the start of the study, there was no statistically signficant difference in the recording of presenting symptoms (in the surgery or at home) between the study and control doctors (Tables 1 and 2). In all subsequent months (excluding home visits in February) the recording of presenting symptoms by the study doctors was significantly higher. There was a marked increase in recording by the doctors in the study group in the month immediately after each of the three meetings with a subsequent decline by the third month (Figure 2). For surgery con-
sultations the largest increase in recording of presenting symptoms occurred after the first informal meeting. The figure rose from 7 to 24% (i.e. a 17% increase). For home visits the largest increase (24%) occurred after the second informal meeting with the individual general practitioners. However, it should be noted that if the doctors are considered individually three of them show their largest increase after the third formal meeting. Although the number of doctors involved was small we found a marked variation in how individual doctors responded to encouragement to improve their recording. For surgery recording (Table 1) this ranged from no increase (Dr D) to a 66% increase after only one informal meeting (Dr Q . Three doctors (A, B and Q
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
Nov 88
444
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
Study Doctors • Surgery
ESS] Home
% of Encounter*
Dec 89Feb 90
•
Apr 90Jun 80
•
Aug 90Oct 90
•
Aug 900ct 90
Month •• Intervention
Control Doctors % of Encounters
Nov 89
•
Dec SSFeb 90
•
Apr 90Jun 90
Month - Intervention
FIGURE 2 Percentage of encounters where presenting symptom(s) were recorded.
showed their largest percentage increase in recording after the first informal meeting. All the control doctors maintained a constant level of symptom recording averaging 3% of total encounters throughout the study. It is interesting to note that the greatest increase in recording for home visits (Table 2) did not occur until after the third more formal meeting for Drs A, B and C. Dr C's recording appears to fluctuate greatly due to the small number of home visits undertaken by this genera] practitioner. Drs D and E showed a very similar pattern of response to that found for surgery visits.
Problems/Diagnoses The results for problems/diagnoses (Figure 3) differed to those for presenting symptoms. The first brief intervention produced a 3-fold increase in recording (from 14 to 50% for surgery consultations and from 14 to 38% for home visits). This increase was maintained and improved upon after subsequent intervention. The group intervention did not improve recording (in terms of percentage increase) to a greater extent than the second individual intervention. The recording of problems/diagnoses by both the study and control doctors is shown in Table 3 (surgery) and Table 4 (home visits). Individual figures for con-
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
NovSO
445
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS
TABLE 1 Recording of presenting symptoms in the surgery. Number (%) of each doctor's consultations where one or more presenting symptoms were recorded Month Dec
Nov Individual study doctors A B C D E
16 (15) 17 (15) 0 0 1 (1)
"
21 20 33 0 6
(31) (32) (66) (7)
April
Feb
24 38 31 1 1
(22) (25) (34) (1) (1)
51 18 57 0 42
(31) (22) (45) (26)
June
28 27 8 _ 51
(20) (15) (14) (23)
Aug
63 37 18 _ 43
(31) (25) (21)
Oct
(34)
34 (20) 42 (30) 53 (25) 0 -
34 (7) All control doctors 18 (4) P value
0.2
80 (24)
95 (17)
168 (30)
114 (19)
161 (29)
129 (22)
13 (6)
15 (3)
26(5)
12 (2)
5 (1)
4 (1)
0.001
0.001
0.001
0.001
0.001
0.001
* Intervention. -, No consultations. P value taken to be significant if less than 0.05. TABLE 2 Recording of presenting symptoms on home visits. Number (ft) of each doctor's consultations where one or more presenting symptoms were recorded Month Nov Individual study doctors A B C D E
8 (53) 2 (8)
Dec
(60) (28) (100)
Feb
(28) (24) (5)
Aug
Oct
(35) (27) (50)
4 (27) 5 (24) 0
9 (41) 15 (45) 3 (50)
4 (15) 9 (45) 6 (35) 0
(43)
9 (33)
5 (50)
0 0
9 7 3 0 3
10 (10)
19 (17)
12 (12)
29 (36)
18 (28)
32 (45)
19 (23)
3 (3)
7 (6)
5 (6)
8 (10)
1 (1)
5 (8)
6 (6)
0.2
0 .02
0.2
0.001
0.001
0.001
0.01
(10)
5 5 1 0 1
June
April
(5)
11 4 8 0 6
All study doctors All control doctors P value
• Intervention. -, No consultations P value taken to be significant if less than 0.05.
trol doctors have been omitted. Although they varied in their individual recording rates each remained very constant in terms of their own level of recording throughout the study. Surgery Consultations (Table 3) In November, prior to the start of the study, the control doctors showed a significantly higher rate of
recording of problems/diagnoses. The situation was reversed after the first intervention and the study doctors then maintained a significantly higher rate of recording throughout the study. As with presenting symptoms the doctors varied greatly in their individual responses to intervention. Dr C showed a very high existing level of recording for problems/diagnoses prior to the study and this
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
All study doctors
446
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
Study Doctors • i
ESS Home
% of Encounters
Nov 89
•
Dae 80 Fab SO
—i •
i i Apr 90 Jun 90
1— • Aug 90 Oct 90
Month ' Intervention
Control Doctors 100
% of Encounters
9080706060403020100 Nov «»
•
Dae SB Fab 90
•
Apr 90 Jun 90
•
Aug 90 Oct 90
Month 1
Intervention FIGURE 3 Percentage of encounters where problem/diagnosis
was maintained. The other doctors showed the same 'response time' found for presenting symptoms i.e. Drs A and B responded more rapidly than Dr E and Dr D showed no marked alteration in recording. Home Visits (Table 4) The improvement in recording of problems/diagnoses for home visits took longer to achieve. It was only from February onwards that the study doctors showed a significantly higher rate of recording. In addition, from February onwards, the percentage of home visits where a problem/diagnosis was recorded by Drs A, B and E was consistently higher than their corresponding figures for surgery consultations. Dr C's figures are
was recorded.
erratic due to the small number of home visits done by this doctor. Overall Dr D did not show any marked change in recording. This doctor was away from the practice from May to September and was not present for the formal group meeting. We are therefore unable to explain why his/her recording of problems/diagnoses should have increased to 50% in October. Investigations For the purposes of the study we looked at the doctors' recording of ECGs and three laboratory investigations—FBP, ESR and admission profile (SMAQ. The study doctors showed no trend toward increased
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
100 00 80706060 40302010-
Surgery
447
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS TABLE 3 Recording of problems/diagnoses—surgery.
Nov Individual study doctors A B C D E
4 6 29 7 29
*
Dec
57 31 44 8 24
(4) (5) (94) (7) (20)
(84) (49) (88) (12) (28)
Number (%) of each doctor's consultations where a problem/diagnosis was recorded
Feb
80 114 78 17 20
•
Month April
140 66 106 7 65
(73) (75) (86) (14) (21)
(85) (82) (84) (21) (40)
June
99 126 47 60
•
Aug
(72) (71) (81)
162 (80) 123 (84) 73 (87)
(28)
39 (31)
Ocl
126 117 182 8 -
(73) (84) (86) (12)
All study doctors 164 (49)
309 (54)
384 (68)
332 (56)
397 (71)
433 (73)
108 (26)
63 (31)
126 (27)
155 (28)
136 (26)
101 (30)
141 (32)
0.001
0.001
0.001
0.001
0.001
All control doctors 0.001
0.001
P value * Intervention. -, No consultations.
TABLE 4 Recording of problems/diagnoses—home
visits. Number (%) of each doctor's consultations where a problem/ diagnosis was recorded Month
Nov Individual study doctors A B C D E
3 1 4 6
(20) (4) (14) (23)
•
Dec
11 12 3 4 12
(73) (48) (100) (11) (40)
Feb
16 19 5 3 5
(89) (91) (26) (14) (26)
•
April
28 12 10 0 6
(90) (80) (63) (43)
June
14 14 2 15
(93) (67) (100) (56)
*
Aug
21 31 5 7
(95) (94) (83) (70)
Oct
21 15 14 9 -
(78) (75) (82) (50)
All study doctors 14 (15)
42 (38)
48 (49)
56 (70)
45 (69)
64 (90)
59 (72)
25 (28)
40 (35)
22 (24)
21 (25)
20 (24)
24 (40)
29 (31)
0.05
0.7
0.001
0.001
0.001
0.001
0.001
All control doctors P value • Intervention. -, No consultations.
recording of these investigations over the study period (Figure 4). The recording of each investigation (considered month by month) was not significantly different between the two groups of doctors with the exception of FBPs in April (P < 0.05) and June (P < 0.01) and ECGs in June (P < 0.05). In each case the control doctors showed the higher level of recording. Comparison with Treatment Room Figures The recording of investigations via the computer was significantly lower than in the treatment room book
for both study and control doctors throughout the study (Table 5). Detailed figures are included to show that the lower level of recording via the computerized system applied to all the investigations considered and was not attributable to just one of the investigations being under-recorded. When all the investigations were considered together for each month it was found that for the study doctors 32% of the investigations recorded in the manual record were also recorded on the computer (monthly variation 22-39). The comparable figure for the control doctors was 35% (monthly variation 22-44).
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
75 (15)
448
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
^ H Study
^ ^ Control
% of Encounters
•
Dec 89Feb 90
•
Apr 90Jun 90
*
Aug 900ct 90
Month • Intervention FIGURE 4 Percentage of surgery encounters where investigation(s) of interest were recorded.
DISCUSSION This study was carried out over the same time period that the new contract was being introduced. We chose to look at aspects of clinical care which general practitioners are, as yet, not required to record under this new contract. However, this may change as the requirements for practice audit increase. Under-recording of such 'non-essential' aspects of the consultation was perhaps not unexpected, being seen as a low priority compared to ensuring that immunization and screening data were up to date. Even so, we were able to demonstrate a significantly increased level of recording of both presenting symptoms and problems/diagnoses by the study doctors. The control doctors showed no change in their recording of presenting symptoms, problems/diagnoses or laboratory investigations throughout the study. One might have predicted that the contract would affect this type of recording in some way. The recording of problems/diagnoses altered differently to that of presenting symptoms. A much greater improvement in recording occurred after the first informal intervention. The improvement was maintained over the months subsequent to the intervention and improved further after the second intervention. A 'problem/diagnosis' is applicable to the majority of consultations whilst specified presenting symptoms and the need for certain laboratory investigations apply less frequently. The less 'routine' an aspect of the consultation is, the harder it may be to ensure that it is recorded completely. In addition,
recording of a problem/diagnosis for every consultation may be seen as more clinically relevant by doctors than the recording of presenting symptoms. This would support the currently held view that one of the best uses of computers at present is for recording diagnoses,9 particularly where the diagnosis results in a prescription being issued.10 Various reasons were suggested to account for the less than 100% recording of problems/diagnoses. The study doctors were asked to record a problem/ diagnosis (i.e. a reason for attending) for all consultations including reviews of chronic conditions. The comment was made during the feedback sessions that many patients seen both at home and in the surgery have chronic conditions which have been documented previously. When such patients present for review or with an exacerbation the doctor may be less inclined to repeatedly record the same diagnosis. The general practitioners felt that their recording of presenting symptoms might be less complete for home visits because encounter forms and/or patients' charts are not always to hand during a home visit and have to be filled in after the event, perhaps in the car. There may only be time to fill in essential information or the doctor may be unable to remember all the details. A number of the doctors commented that it was very useful to see their results in black and white and they felt that this had had more effect than a purely verbal request to improve and just being told that they were not recording well. Presenting the results of individual doctors to the whole group has to be handled
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
Nov 89
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS
carefully and the use of individual meetings is preferable initially to allow the doctors to grasp what is required of them. Familiarity with the method of data collection influenced the degree of intervention necessary to imTABLE 5 Recording of investigations requested by study doctors. Number of consultations (%) where investigation was recorded Computer
12 8 3 3
December FBP ESR Adm. profile ECG Total consultations
2 5 1 1
February FBP ESR Adm. profile ECG Total consultations
20 12 1 1
April FBP ESR Adm. profile ECG Total consultations
10 (2) 12 (2) 2 (0) 0
June FBP ESR Adm. profile ECG Total consultations
13 (2) 12 (2) 0 2 (0)
August FBP ESR Adm. profile ECG Total consultations
15 13 3 3
October FBP ESR Adm. profile ECG Total consultations
13 11 2 3
(2) (2) (1) (1)
PC)
40 28 8 11
(8) (6) (2) (2)
0.001 0.001 _ 0.008
20 19 2 6
(6) (6) (1) (2)
0.001 0.001 _ -
48 28 8 15
(8) (5) (1) (3)
0.001 0.001 0.016 0.001
43 36 5 6
(8) (6) (1) (1)
0.001 0.001 0.032
36 22 1 4
(6) (4) (1) (1)
0.001 0.01 0.016 _
43(8) 33 (6) 11 (2) 13 (2)
0.001 0.001 0.008 0.01
40 31 12 12
0.001 0.001 0.01 0.004
498 (1) (2) (0) (0) 332 (4) (2) (0) (0) 576
569
590 (3) (2) (0) (0) 559 (2) (2) (0) (1)
(7) (5) (2) (2)
591
• P value based on McNemar Test (Siegel, 1956) or binomial test if expected frequency < 5 . -, Binomial test not valid.
prove recording. The three study doctors who showed a marked improvement in recording of presenting symptoms after the first informal meeting had been involved in the development and implementation of the encounter form. They had been using it for 4 years and were more familiar with it than Dr D (part-time) and Dr E (a trainee). Dr E's response improved noticeably after the second intervention in April where he realized that he had been filling in the encounter form incorrectly. This had resulted in the presenting symptom not being stored on the computer although it appeared in the printed clinical notes. The slower improvement in recording of problems/ diagnoses by the trainee in the study gToup again proved to be partly due to lack of familiarity with the encounter form. If a diagnosis is only put down tentatively e.g. ?asthma it will not be coded and stored on the computer. A trainee who is less experienced is less willing to commit himself to a diagnosis and might make more use of the 'query diagnosis'. Once this was discussed with the trainee at the second informal meeting his recording of problems improved. This highlights the need for new or temporary doctors in a practice to be given adequate instruction as to the details required on any data collection forms used routinely by the practice. Practices may not give enough attention to this area which is potentially a major source of missed, incomplete or incorrectly coded information being captured in the records. The recording of investigations showed little change after any of the forms of encouragement. A possible explanation for this was that the investigations considered were being recorded completely to start with. However, this was not supported by comparison with the manual record. As all requests for laboratory investigations resulting from surgery visits are processed by the treatment room nurses there should be no case where an investigation is recorded on the computer and not in the treatment room book. The latter was found to provide the more accurate measure of investigation usage. A number of reasons may account for the underrecording on the computer. If no encounter forms are available when the patient is seen the details will not be captured for transfer to the computer. This is more likely if patients are seen outside normal surgery hours. With hindsight, the space provided on the encounter form for recording investigations is too small and easily overlooked. The doctor has already written on a request form for the treatment room nurse which investigations are to be carried out. He has thus 'written down the details once' and may not see the need to repeat this process by writing them again on the encounter form. In addition, the results are filed in the patients' notes so a permanent record of the investigation will exist. The danger here is that the doctor is assuming that the investigation will be performed and the result will come back. If this doesn't happen it is very difficult for the doctor or his col-
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
November FBP ESR Adm. profile ECG Total consultations
Manual
449
450
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
leagues to determine which investigations a particular patient has had. If doctors are not convinced of the need to record something it will not be done. This is compounded if the details to be recorded have to be written down more than once. It may be that the doctor is not the best person to be recording what laboratory investigations are requested. A better way to achieve accurate computer recording might be for the nurses to key the details directly into the computer instead of both doctor and nurse filling in a paper record.
ACKNOWLEDGEMENTS We are grateful to the Eastern Health and Social Services Board, Northern Ireland for their financial
REFERENCES Voss SN, Thomas HF. How wdl do family practitioner committee and general practice records agree? Experience in a semi-rural practice. Br J Gen Pract 1991;41:293-294. 2 Difford F, Hook PM, Sledge M. Maintaining the accuracy of a computer practice register: household index. Br MedJ 1985; 290: 519-521. 3 Johnson N, Mant D, Jones L, Randall T. Use of computerised general practice data for population surveillance: comparative study of influenza data. Br MedJ 1991; 302: 763-765. 4 Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. Br Med J 1991; 302: 766-768. 5 Mant D, Tulloch A. Completeness of chronic disease registration in general practice. Br Med J 1987; 294: 223-224. 6 Irwin WG, Mills KA, Steele K. Effect on prescribing of the limited list in a computerised group practice. BrMedJ 1986; 293: 857-859. 7 Steele K, Mills KA, Gilliland AEW, Irwin WG, Taggart A. Repeat prescribing of non-steroidal antiinflammatory drugs excluding aspirin: how careful are we? Br Med J 1987; 295: 962-964. 8 Morrell DC. Symptom interpretation in general practice. J R Coll Gen Pract 1972; 22: 297-309. 9 Pringle M, Hobbs R. Large computer databases in general practice. Br Med J 1991; 302: 741-742. 10 Lee PR. Medication and prescribing. In Fry J. (ed.) Primary care. London: Heinemann Medical, 1980; 230-251. 1
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
CONCLUSIONS Our initial hypothesis that data recording would be incomplete was confirmed. We were able to demonstrate an improvement in the recording of presenting symptoms and problems but not in the case of laboratory investigations. It may not be possible to get all aspects of the consultation recorded with the same degree of accuracy. The effectiveness of the different forms of intervention depended on both the aspect of the consultation considered and the familiarity of the individual doctors with the method of data collection. Aspects seen as possibly less important or applying to fewer consultations required greater intervention to bring about a marked improvement, as did doctors relatively new to the practice. Despite intervention, the recording of investigations on the computer was on average one-third of the total in the treatment room book for both study and control doctors.
support and to Dr Desmond Merrett for statistical advice.
Vol. 9, No. 4 Printed in Great Britain
General Practitioner Records on Computer—Handle with Care AEW GILLILAND, KA MILLS AND K STEELE
INTRODUCTION An increasing number of general practitioners utilize computerized medical records systems to store clinical data on their patients. Computerized data-bases should prove invaluable if not essential to general practitioners if they are to meet the conditions of their new contract. However, the data are of little use if incomplete. Although a number of studies have addressed the accuracy of computerized age/sex registers1'2 few have considered the completeness of the computerized clinical data.3"5 Practices vary as to how they collect the clinical information needed to keep their computerized records up to date. One method involves using an encounter form to record what happens during consultations. Any additions, amendments or deletions to be made to the computerized record are noted on this form which is then passed on to a computer clerk for keying in. Now that many general practitioners have desk-top computers in their surgeries it may be argued that the manual recording form is obsolete. However, not all doctors are willing to use the computer during or in
between consultations to record patient information on a regular basis. A limited encounter form may still be a necessary means of gathering data in general practice. Such an encounter form had been used by the two study practices for some years. As their computer system was not multi-user the option of direct entry by the doctors was not available. In addition to providing the study practices with information for their day to day service work, academic staff in the local department of general practice have, in the past, been able to access the patient data-base to look at several clinical areas, in particular prescribing.6'7 This had highlighted areas where information seemed incomplete and it was decided to look at this in more detail to see if the level of recording could be improved. The initial hypothesis was that recording of selected aspects of clinical information in the forementioned practices would be incomplete and could be improved by appropriate intervention. The study aims were 2-fold. 1. To determine by comparison with a manual record how accurately the doctors were recording certain laboratory investigations and the electrocardiograms (ECGs) they request.
Department of General Practice, Queen's University of Belfast, Dunluce Health Centre, 1 Dunluce Avenue, Belfast BT9 7HR. Correspondence to AEW Gilliland.
441
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
Gilliland AEW, Mills KA and Steele K. General practitioner records on computer—handle with care. Family Practice 1992; 9: 441-450. This 1 year prospective study involved nine general practitioners in an urban health centre who routinely record all patient contacts on computer. The study determines by comparison with a manual record how accurately doctors record laboratory investigations on computer and compares the effectiveness of three interventions in improving the completeness of computerized recording of presenting symptoms, problems/diagnoses and laboratory investigations. Recording was analysed for 1 month prior to and for two 1 month periods following each intervention. A control group was used. A total of 7983 patient contacts were analysed. Intervention led to an improvement in the recording of presenting symptoms and problems/diagnoses. Recording of investigations on the computer showed no improvement, remaining at one-third of the total in the treatment room book for both study and control doctors. The effectiveness of the different forms of intervention depended on both the aspect of the consultation considered and the familiarity of individual doctors with the method of data collection. Aspects considered less important required greater intervention to bring about a marked improvement, as did doctors relatively new to the practice. It may not be possible to get all aspects of the consultation recorded with the same degree of accuracy. This has implications for the accuracy of retrospective studies dependent on existing computerized data.
442
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
2. To determine the effect of three methods of intervention on improving the completeness of recording of presenting symptoms, problems/ diagnoses and investigations.
Informal Individual Meeting with no Feedback of Results The first meetings with the general practitioners were held on November 30 1989. They took the form of a quick informal request by the authors to individual doctors taking about 10 minutes after surgery. The purpose of the study was explained and each doctor was asked to try and improve their recording in the three areas of interest. No feedback was given at this time about their current level of recording. Each general practitioner was unaware of which of his colleagues were in the study group and was asked not to discuss the study with any of them. Informal Individual Meeting with Feedback The second meetings were held on March 22 1990. Each study doctor was seen individually for approximately 15 minutes. He or she was given their own results for November, December and February and shown how these compared with their colleagues (unidentified) in both the study and control groups. They were reminded about the purpose of the study, given further encouragement to improve and encouraged to discuss any problems they had with data recording. Formal Group Meeting with Feedback The third intervention at the end of July involved all the study general practitioners as a group. They were given feedback about both their individual and the overall study group recording levels as compared with the control group. The doctors were given the opportunity to discuss their individual results and encouraged to suggest reasons for the observed changes in recording. Their encounter recordings were subsequently analysed for August and October to assess the impact of the third group meeting. RESULTS The numbers of encounters recorded by the doctors each month are shown in Figure 1. The total for the study doctors (/» = 5) was surgery, 3715; home visits, 603 and for the control doctors (n = 4) surgery, 2957; home visits, 618. The study doctors saw a significantly higher proportion of their patients in the surgery in December \p < 0.01), June (P < 0.05) and October (P < 0.01). Presenting Symptoms The percentages of encounters for which one or more presenting symptoms were recorded are shown in
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
METHODOLOGY The study took place over a 1 year period commencing in November 1989 and involved nine general practitioners based in an urban health centre. Each doctor was assigned randomly to either a study or control group. The study group consisted of three full-time and two part-time principals and a trainee. The control group contained two full-time and one part-time principal and a trainee. The control group were unaware of the study and were left to record their consultations in the usual way. The researchers did not take part in the study. The study considered the doctor's recording of three distinct aspects of the consultation: presenting symptoms, problems/diagnoses and the following investigations—full blood picture (FBP), erythrocyte sedimentation rate (ESR), biochemical profile (SMAQ and ECGs. Throughout this paper the term encounter refers to an individual consultation. The encounter form used routinely by the study and control doctors contains a field which lists 11 common presenting symptoms— cough, headache, tiredness, chest pain, abdominal pain, dizziness, back ache, indigestion, rash, joint pain and disturbance of bowel function. These are based on Morrell's findings.8 If one or more of these applied to a particular consultation the doctor was asked to circle the relevant symptom(s). (This section was introduced 3 years prior to the study for research purposes.) The general practitioners were asked to record a problem/ diagnosis for every consultation. In addition, any laboratory investigation or ECG requested via the treatment room was to be recorded on the encounter form. The doctors recorded the above information routinely for all surgery and home visits. The nurses only took blood samples, swabs, etc. at the request of a doctor. Details of all investigations requested by the doctor were entered into a day book by the treatment room nurses. The two methods of data collection—the investigation field of the computerized encounter form and the treatment room day book— should therefore record the same events. We were able to compare the two sets of figures for each investigation for both the study and control doctors. This was only done for investigations requested during surgery consultations as very few investigations are requested following home visits. The study doctors were unaware of the study prior to the December 1 1989 when they were interviewed and encouraged to improve their recording in the three areas of interest. This enabled us to use the practice computer to determine what percentage of each doctor's consultations (both in the surgery and at home) had presenting symptoms, a problem/diagnosis
or a request for a laboratory investigation or ECG recorded throughout November 1989. Three forms of intervention were used in turn. Consultation data were analysed for the month immediately following each intervention and again for the third month afterwards. The three forms of intervention used were as follows.
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS
443
Study Doctors a M Surgery
l\\\N Homo
Numb«r of Encounters 700
Dae 88 Fab 80
Apr SO Jun SO
Aug 80 Oet BO
Month
Control Doctors Number of Encounters 700
NOV BS
Dae BS Fab SO
Apr 80 Jun 80
Aug so Oet so
Month FIGURE 1 Surgery and home encounters during study months.
Figure 2. No data are presented for March and July as a month was needed between the study collection periods in which to analyse and prepare the data for feedback to the doctors. In November, prior to the start of the study, there was no statistically signficant difference in the recording of presenting symptoms (in the surgery or at home) between the study and control doctors (Tables 1 and 2). In all subsequent months (excluding home visits in February) the recording of presenting symptoms by the study doctors was significantly higher. There was a marked increase in recording by the doctors in the study group in the month immediately after each of the three meetings with a subsequent decline by the third month (Figure 2). For surgery con-
sultations the largest increase in recording of presenting symptoms occurred after the first informal meeting. The figure rose from 7 to 24% (i.e. a 17% increase). For home visits the largest increase (24%) occurred after the second informal meeting with the individual general practitioners. However, it should be noted that if the doctors are considered individually three of them show their largest increase after the third formal meeting. Although the number of doctors involved was small we found a marked variation in how individual doctors responded to encouragement to improve their recording. For surgery recording (Table 1) this ranged from no increase (Dr D) to a 66% increase after only one informal meeting (Dr Q . Three doctors (A, B and Q
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
Nov 88
444
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
Study Doctors • Surgery
ESS] Home
% of Encounter*
Dec 89Feb 90
•
Apr 90Jun 80
•
Aug 90Oct 90
•
Aug 900ct 90
Month •• Intervention
Control Doctors % of Encounters
Nov 89
•
Dec SSFeb 90
•
Apr 90Jun 90
Month - Intervention
FIGURE 2 Percentage of encounters where presenting symptom(s) were recorded.
showed their largest percentage increase in recording after the first informal meeting. All the control doctors maintained a constant level of symptom recording averaging 3% of total encounters throughout the study. It is interesting to note that the greatest increase in recording for home visits (Table 2) did not occur until after the third more formal meeting for Drs A, B and C. Dr C's recording appears to fluctuate greatly due to the small number of home visits undertaken by this genera] practitioner. Drs D and E showed a very similar pattern of response to that found for surgery visits.
Problems/Diagnoses The results for problems/diagnoses (Figure 3) differed to those for presenting symptoms. The first brief intervention produced a 3-fold increase in recording (from 14 to 50% for surgery consultations and from 14 to 38% for home visits). This increase was maintained and improved upon after subsequent intervention. The group intervention did not improve recording (in terms of percentage increase) to a greater extent than the second individual intervention. The recording of problems/diagnoses by both the study and control doctors is shown in Table 3 (surgery) and Table 4 (home visits). Individual figures for con-
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
NovSO
445
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS
TABLE 1 Recording of presenting symptoms in the surgery. Number (%) of each doctor's consultations where one or more presenting symptoms were recorded Month Dec
Nov Individual study doctors A B C D E
16 (15) 17 (15) 0 0 1 (1)
"
21 20 33 0 6
(31) (32) (66) (7)
April
Feb
24 38 31 1 1
(22) (25) (34) (1) (1)
51 18 57 0 42
(31) (22) (45) (26)
June
28 27 8 _ 51
(20) (15) (14) (23)
Aug
63 37 18 _ 43
(31) (25) (21)
Oct
(34)
34 (20) 42 (30) 53 (25) 0 -
34 (7) All control doctors 18 (4) P value
0.2
80 (24)
95 (17)
168 (30)
114 (19)
161 (29)
129 (22)
13 (6)
15 (3)
26(5)
12 (2)
5 (1)
4 (1)
0.001
0.001
0.001
0.001
0.001
0.001
* Intervention. -, No consultations. P value taken to be significant if less than 0.05. TABLE 2 Recording of presenting symptoms on home visits. Number (ft) of each doctor's consultations where one or more presenting symptoms were recorded Month Nov Individual study doctors A B C D E
8 (53) 2 (8)
Dec
(60) (28) (100)
Feb
(28) (24) (5)
Aug
Oct
(35) (27) (50)
4 (27) 5 (24) 0
9 (41) 15 (45) 3 (50)
4 (15) 9 (45) 6 (35) 0
(43)
9 (33)
5 (50)
0 0
9 7 3 0 3
10 (10)
19 (17)
12 (12)
29 (36)
18 (28)
32 (45)
19 (23)
3 (3)
7 (6)
5 (6)
8 (10)
1 (1)
5 (8)
6 (6)
0.2
0 .02
0.2
0.001
0.001
0.001
0.01
(10)
5 5 1 0 1
June
April
(5)
11 4 8 0 6
All study doctors All control doctors P value
• Intervention. -, No consultations P value taken to be significant if less than 0.05.
trol doctors have been omitted. Although they varied in their individual recording rates each remained very constant in terms of their own level of recording throughout the study. Surgery Consultations (Table 3) In November, prior to the start of the study, the control doctors showed a significantly higher rate of
recording of problems/diagnoses. The situation was reversed after the first intervention and the study doctors then maintained a significantly higher rate of recording throughout the study. As with presenting symptoms the doctors varied greatly in their individual responses to intervention. Dr C showed a very high existing level of recording for problems/diagnoses prior to the study and this
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
All study doctors
446
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
Study Doctors • i
ESS Home
% of Encounters
Nov 89
•
Dae 80 Fab SO
—i •
i i Apr 90 Jun 90
1— • Aug 90 Oct 90
Month ' Intervention
Control Doctors 100
% of Encounters
9080706060403020100 Nov «»
•
Dae SB Fab 90
•
Apr 90 Jun 90
•
Aug 90 Oct 90
Month 1
Intervention FIGURE 3 Percentage of encounters where problem/diagnosis
was maintained. The other doctors showed the same 'response time' found for presenting symptoms i.e. Drs A and B responded more rapidly than Dr E and Dr D showed no marked alteration in recording. Home Visits (Table 4) The improvement in recording of problems/diagnoses for home visits took longer to achieve. It was only from February onwards that the study doctors showed a significantly higher rate of recording. In addition, from February onwards, the percentage of home visits where a problem/diagnosis was recorded by Drs A, B and E was consistently higher than their corresponding figures for surgery consultations. Dr C's figures are
was recorded.
erratic due to the small number of home visits done by this doctor. Overall Dr D did not show any marked change in recording. This doctor was away from the practice from May to September and was not present for the formal group meeting. We are therefore unable to explain why his/her recording of problems/diagnoses should have increased to 50% in October. Investigations For the purposes of the study we looked at the doctors' recording of ECGs and three laboratory investigations—FBP, ESR and admission profile (SMAQ. The study doctors showed no trend toward increased
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
100 00 80706060 40302010-
Surgery
447
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS TABLE 3 Recording of problems/diagnoses—surgery.
Nov Individual study doctors A B C D E
4 6 29 7 29
*
Dec
57 31 44 8 24
(4) (5) (94) (7) (20)
(84) (49) (88) (12) (28)
Number (%) of each doctor's consultations where a problem/diagnosis was recorded
Feb
80 114 78 17 20
•
Month April
140 66 106 7 65
(73) (75) (86) (14) (21)
(85) (82) (84) (21) (40)
June
99 126 47 60
•
Aug
(72) (71) (81)
162 (80) 123 (84) 73 (87)
(28)
39 (31)
Ocl
126 117 182 8 -
(73) (84) (86) (12)
All study doctors 164 (49)
309 (54)
384 (68)
332 (56)
397 (71)
433 (73)
108 (26)
63 (31)
126 (27)
155 (28)
136 (26)
101 (30)
141 (32)
0.001
0.001
0.001
0.001
0.001
All control doctors 0.001
0.001
P value * Intervention. -, No consultations.
TABLE 4 Recording of problems/diagnoses—home
visits. Number (%) of each doctor's consultations where a problem/ diagnosis was recorded Month
Nov Individual study doctors A B C D E
3 1 4 6
(20) (4) (14) (23)
•
Dec
11 12 3 4 12
(73) (48) (100) (11) (40)
Feb
16 19 5 3 5
(89) (91) (26) (14) (26)
•
April
28 12 10 0 6
(90) (80) (63) (43)
June
14 14 2 15
(93) (67) (100) (56)
*
Aug
21 31 5 7
(95) (94) (83) (70)
Oct
21 15 14 9 -
(78) (75) (82) (50)
All study doctors 14 (15)
42 (38)
48 (49)
56 (70)
45 (69)
64 (90)
59 (72)
25 (28)
40 (35)
22 (24)
21 (25)
20 (24)
24 (40)
29 (31)
0.05
0.7
0.001
0.001
0.001
0.001
0.001
All control doctors P value • Intervention. -, No consultations.
recording of these investigations over the study period (Figure 4). The recording of each investigation (considered month by month) was not significantly different between the two groups of doctors with the exception of FBPs in April (P < 0.05) and June (P < 0.01) and ECGs in June (P < 0.05). In each case the control doctors showed the higher level of recording. Comparison with Treatment Room Figures The recording of investigations via the computer was significantly lower than in the treatment room book
for both study and control doctors throughout the study (Table 5). Detailed figures are included to show that the lower level of recording via the computerized system applied to all the investigations considered and was not attributable to just one of the investigations being under-recorded. When all the investigations were considered together for each month it was found that for the study doctors 32% of the investigations recorded in the manual record were also recorded on the computer (monthly variation 22-39). The comparable figure for the control doctors was 35% (monthly variation 22-44).
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
75 (15)
448
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
^ H Study
^ ^ Control
% of Encounters
•
Dec 89Feb 90
•
Apr 90Jun 90
*
Aug 900ct 90
Month • Intervention FIGURE 4 Percentage of surgery encounters where investigation(s) of interest were recorded.
DISCUSSION This study was carried out over the same time period that the new contract was being introduced. We chose to look at aspects of clinical care which general practitioners are, as yet, not required to record under this new contract. However, this may change as the requirements for practice audit increase. Under-recording of such 'non-essential' aspects of the consultation was perhaps not unexpected, being seen as a low priority compared to ensuring that immunization and screening data were up to date. Even so, we were able to demonstrate a significantly increased level of recording of both presenting symptoms and problems/diagnoses by the study doctors. The control doctors showed no change in their recording of presenting symptoms, problems/diagnoses or laboratory investigations throughout the study. One might have predicted that the contract would affect this type of recording in some way. The recording of problems/diagnoses altered differently to that of presenting symptoms. A much greater improvement in recording occurred after the first informal intervention. The improvement was maintained over the months subsequent to the intervention and improved further after the second intervention. A 'problem/diagnosis' is applicable to the majority of consultations whilst specified presenting symptoms and the need for certain laboratory investigations apply less frequently. The less 'routine' an aspect of the consultation is, the harder it may be to ensure that it is recorded completely. In addition,
recording of a problem/diagnosis for every consultation may be seen as more clinically relevant by doctors than the recording of presenting symptoms. This would support the currently held view that one of the best uses of computers at present is for recording diagnoses,9 particularly where the diagnosis results in a prescription being issued.10 Various reasons were suggested to account for the less than 100% recording of problems/diagnoses. The study doctors were asked to record a problem/ diagnosis (i.e. a reason for attending) for all consultations including reviews of chronic conditions. The comment was made during the feedback sessions that many patients seen both at home and in the surgery have chronic conditions which have been documented previously. When such patients present for review or with an exacerbation the doctor may be less inclined to repeatedly record the same diagnosis. The general practitioners felt that their recording of presenting symptoms might be less complete for home visits because encounter forms and/or patients' charts are not always to hand during a home visit and have to be filled in after the event, perhaps in the car. There may only be time to fill in essential information or the doctor may be unable to remember all the details. A number of the doctors commented that it was very useful to see their results in black and white and they felt that this had had more effect than a purely verbal request to improve and just being told that they were not recording well. Presenting the results of individual doctors to the whole group has to be handled
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
Nov 89
ACCURACY AND COMPLETENESS OF COMPUTERIZED RECORDS
carefully and the use of individual meetings is preferable initially to allow the doctors to grasp what is required of them. Familiarity with the method of data collection influenced the degree of intervention necessary to imTABLE 5 Recording of investigations requested by study doctors. Number of consultations (%) where investigation was recorded Computer
12 8 3 3
December FBP ESR Adm. profile ECG Total consultations
2 5 1 1
February FBP ESR Adm. profile ECG Total consultations
20 12 1 1
April FBP ESR Adm. profile ECG Total consultations
10 (2) 12 (2) 2 (0) 0
June FBP ESR Adm. profile ECG Total consultations
13 (2) 12 (2) 0 2 (0)
August FBP ESR Adm. profile ECG Total consultations
15 13 3 3
October FBP ESR Adm. profile ECG Total consultations
13 11 2 3
(2) (2) (1) (1)
PC)
40 28 8 11
(8) (6) (2) (2)
0.001 0.001 _ 0.008
20 19 2 6
(6) (6) (1) (2)
0.001 0.001 _ -
48 28 8 15
(8) (5) (1) (3)
0.001 0.001 0.016 0.001
43 36 5 6
(8) (6) (1) (1)
0.001 0.001 0.032
36 22 1 4
(6) (4) (1) (1)
0.001 0.01 0.016 _
43(8) 33 (6) 11 (2) 13 (2)
0.001 0.001 0.008 0.01
40 31 12 12
0.001 0.001 0.01 0.004
498 (1) (2) (0) (0) 332 (4) (2) (0) (0) 576
569
590 (3) (2) (0) (0) 559 (2) (2) (0) (1)
(7) (5) (2) (2)
591
• P value based on McNemar Test (Siegel, 1956) or binomial test if expected frequency < 5 . -, Binomial test not valid.
prove recording. The three study doctors who showed a marked improvement in recording of presenting symptoms after the first informal meeting had been involved in the development and implementation of the encounter form. They had been using it for 4 years and were more familiar with it than Dr D (part-time) and Dr E (a trainee). Dr E's response improved noticeably after the second intervention in April where he realized that he had been filling in the encounter form incorrectly. This had resulted in the presenting symptom not being stored on the computer although it appeared in the printed clinical notes. The slower improvement in recording of problems/ diagnoses by the trainee in the study gToup again proved to be partly due to lack of familiarity with the encounter form. If a diagnosis is only put down tentatively e.g. ?asthma it will not be coded and stored on the computer. A trainee who is less experienced is less willing to commit himself to a diagnosis and might make more use of the 'query diagnosis'. Once this was discussed with the trainee at the second informal meeting his recording of problems improved. This highlights the need for new or temporary doctors in a practice to be given adequate instruction as to the details required on any data collection forms used routinely by the practice. Practices may not give enough attention to this area which is potentially a major source of missed, incomplete or incorrectly coded information being captured in the records. The recording of investigations showed little change after any of the forms of encouragement. A possible explanation for this was that the investigations considered were being recorded completely to start with. However, this was not supported by comparison with the manual record. As all requests for laboratory investigations resulting from surgery visits are processed by the treatment room nurses there should be no case where an investigation is recorded on the computer and not in the treatment room book. The latter was found to provide the more accurate measure of investigation usage. A number of reasons may account for the underrecording on the computer. If no encounter forms are available when the patient is seen the details will not be captured for transfer to the computer. This is more likely if patients are seen outside normal surgery hours. With hindsight, the space provided on the encounter form for recording investigations is too small and easily overlooked. The doctor has already written on a request form for the treatment room nurse which investigations are to be carried out. He has thus 'written down the details once' and may not see the need to repeat this process by writing them again on the encounter form. In addition, the results are filed in the patients' notes so a permanent record of the investigation will exist. The danger here is that the doctor is assuming that the investigation will be performed and the result will come back. If this doesn't happen it is very difficult for the doctor or his col-
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
November FBP ESR Adm. profile ECG Total consultations
Manual
449
450
FAMILY PRACTICE—AN INTERNATIONAL JOURNAL
leagues to determine which investigations a particular patient has had. If doctors are not convinced of the need to record something it will not be done. This is compounded if the details to be recorded have to be written down more than once. It may be that the doctor is not the best person to be recording what laboratory investigations are requested. A better way to achieve accurate computer recording might be for the nurses to key the details directly into the computer instead of both doctor and nurse filling in a paper record.
ACKNOWLEDGEMENTS We are grateful to the Eastern Health and Social Services Board, Northern Ireland for their financial
REFERENCES Voss SN, Thomas HF. How wdl do family practitioner committee and general practice records agree? Experience in a semi-rural practice. Br J Gen Pract 1991;41:293-294. 2 Difford F, Hook PM, Sledge M. Maintaining the accuracy of a computer practice register: household index. Br MedJ 1985; 290: 519-521. 3 Johnson N, Mant D, Jones L, Randall T. Use of computerised general practice data for population surveillance: comparative study of influenza data. Br MedJ 1991; 302: 763-765. 4 Jick H, Jick SS, Derby LE. Validation of information recorded on general practitioner based computerised data resource in the United Kingdom. Br Med J 1991; 302: 766-768. 5 Mant D, Tulloch A. Completeness of chronic disease registration in general practice. Br Med J 1987; 294: 223-224. 6 Irwin WG, Mills KA, Steele K. Effect on prescribing of the limited list in a computerised group practice. BrMedJ 1986; 293: 857-859. 7 Steele K, Mills KA, Gilliland AEW, Irwin WG, Taggart A. Repeat prescribing of non-steroidal antiinflammatory drugs excluding aspirin: how careful are we? Br Med J 1987; 295: 962-964. 8 Morrell DC. Symptom interpretation in general practice. J R Coll Gen Pract 1972; 22: 297-309. 9 Pringle M, Hobbs R. Large computer databases in general practice. Br Med J 1991; 302: 741-742. 10 Lee PR. Medication and prescribing. In Fry J. (ed.) Primary care. London: Heinemann Medical, 1980; 230-251. 1
Downloaded from http://fampra.oxfordjournals.org/ at Freie Universitaet Berlin on July 1, 2015
CONCLUSIONS Our initial hypothesis that data recording would be incomplete was confirmed. We were able to demonstrate an improvement in the recording of presenting symptoms and problems but not in the case of laboratory investigations. It may not be possible to get all aspects of the consultation recorded with the same degree of accuracy. The effectiveness of the different forms of intervention depended on both the aspect of the consultation considered and the familiarity of the individual doctors with the method of data collection. Aspects seen as possibly less important or applying to fewer consultations required greater intervention to bring about a marked improvement, as did doctors relatively new to the practice. Despite intervention, the recording of investigations on the computer was on average one-third of the total in the treatment room book for both study and control doctors.
support and to Dr Desmond Merrett for statistical advice.
