✁ ✂
✄ ☎
✁ ✆
✝ ✞
✟ ✄
✠ ✁
✆ ✠
✁ ✡
✞ ☛
✂ ✁
✂ ✄
☛ ✂
✄ ✠
☛ ☞
✌ ✍
☞
✜ ✢
2.2 SURVEY DATES
All sampled clinics were given one randomly assigned date for data collection, which was a working day for all. The following days were excluded:
SXEOLFKROLGD\V ZHHNHQGVLQFOXGLQJULGD\6DWXUGD\DQG6XQGD\
0RQGD\0RQGD\VDUHXVXDOO\EXVLHVWIRUPRVWSULPDU\FDUHFOLQLFV DZHHNDIWHUPDMRUIHVWLYHFHOHEUDWLRQV
ZRUNLQJGD\VLQEHWZHHQSXEOLFKROLGD\V If the clinic was closed on the survey date, the doctor had the option to change the survey date to the next
available working day. The research team however was informed of the new survey date.
2.3 DATA COLLECTION AND FOLLOW-UP
7KH VDPSOHG FOLQLFV ZHUH HDFK VHQW DQ LQYLWDWLRQ OHWWHU WR DWWHQG D EULHÀQJ LQ HDFK VWDWH ULHÀQJV for doctors in the public clinics were held on weekdays whereas for private doctors, it was during the
weekends from June to July 2012. A research pack which contained the survey forms and instructions ZHUHGLVWULEXWHGGXULQJWKHEULHÀQJV
7RHQFRXUDJHIXUWKHUSDUWLFLSDWLRQFOLQLFVWKDWGLGQRWDWWHQGWKHEULHÀQJZHUHODWHUFRQWDFWHGE\SKRQH If the doctor refused to participate in the survey, the team did not pursue further. However, if they agree
to participate the research pack would be sent either E\FRXULHUIROORZHGE\WHOHSKRQHFDOOWRHQVXUHWKDWWKHUHVHDUFKNLWKDGEHHQUHFHLYHGULHÀQJ
would be done over the phone to explain about the survey form. E\SHUVRQDOYLVLWWRWKHFOLQLFVZLWKLQWKHYLFLQLW\RI.ODQJ9DOOH\DQGDVKRUWSULYDWHEULHÀQJ
would be given to the doctornurse in-charge Two weeks and one day before the survey date, a reminder via telephone was made to the clinic about
the project and to answer any questions pertaining to the survey. Instructions would be repeated when necessary. After the survey date, follow-up phone calls were made if the research pack was not returned
after 3 weeks, and subsequently at 5 weeks.
Various approaches were also taken to increase the acceptance and response rate of private clinics. This included the following:
a Approaching the managersenior management of the chain clinicsgroup practices
E 2UJDQLVLQJSULYDWHLQGLYLGXDOEULHÀQJVDORQJVLGH0HGLFDO3UDFWLFHLYLVLRQ·VHQIRUFHPHQW activities
c Getting support and assistance from Malaysian Medical Association MMA at the state level
8SRQFRPSOHWLRQRIGDWDFROOHFWLRQSDUWLFLSDQWVZHUHJLYHQFHUWLÀFDWHVZKLFKWKH\ZRXOGODWHUXVHWR claim for continuing professional development CPD points. An individualised feedback, questionnaire
on primary care prescribers, and a copy of the National Medical Care Statistics 2012 report were also sent to all participants.
✁ ✂
✄ ☎
✁ ✆
✝ ✞
✟ ✄
✠ ✁
✆ ✠
✁ ✡
✞ ☛
✂ ✁
✂ ✄
☛ ✂
✄ ✠
☛ ☞
✌ ✍
☞
✜ ✣
2.4 RESEARCH PACK AND QUESTIONNAIRE
A pre-testing session of the questionnaire was carried out by requesting a convenience sample of doctors from public and private clinics to complete the forms. A total of 32 encounters were recorded, and
FRPPHQWVIURPWKHGRFWRUVZHUHWDNHQLQWRFRQVLGHUDWLRQ7KH106IRUPZDVIXUWKHUPRGLÀHG DQGWKHÀQDOLVHGIRUPLVHQFORVHGLQSSHQGL[DFKUHVHDUFKSDFNFRQWDLQHG
Survey Pads Each pad has
- 40 forms - One set of instructions
- One case study - One example of a completed form
6XUYH\DWH1RWLÀFDWLRQ
Survey date and day
Project Summary Purpose and objective of NMCS 2012
Public Notice
Notice to be displayed in the participating clinic to inform patient that the clinic is currently undertaking a MOH survey
Data Dictionary One-page description of variables in the survey form
ICPC-2-code List ICPC-2-Code list
Also included in the research pack: Call Letter
Letter signed by State Health Directors to inform the participating clinics of the survey
Prepaid Envelope Two sets 1 for National Healthcare and Establishment Survey NHEWS
and 1 for NMCS. The one for NHEWS contains a NHEWS survey form and a small envelope containing the user ID and password for those who prefer
online questionnaire for NHEWS. NHEWS Primary Care is a 4-pages questionnaire focusing on services and
workforce in the primary care setting
Report NHEWS 2010 Primary Care Report
2.5 DATA MANAGEMENT
2.5.1 Data entry
All data entry was done through the use of data entry web application and performed by trained data entry personnel.
Prior to the start of data entry, data entry personnel were given reading materials containing the detailed GHVFULSWLRQ RI VWXG\ H[DPSOHV RI IRUPV FODVVLÀFDWLRQ DQG FRGLQJ V\VWHPV DQG GDWD HQWU\ UXOHV DQG
regulations. This was followed by two sessions of data entry training which lasted for 2 hours each.
Session 1: Demonstration 3UHVHQWDWLRQRQGDWDHQWU\PRGXOH
LYHGHPRQVWUDWLRQRIGDWDHQWU\PRGXOH LYHGHPRQVWUDWLRQRIFRGLQJV\VWHP
LVFXVVLRQRQGDWDHQWU\DQGFRGLQJV\VWHP 3UDFWLFDOVHVVLRQGDWDHQWU\DQGFRGLQJRIHQFRXQWHUVSHUSHUVRQ
✁ ✂
✄ ☎
✁ ✆
✝ ✞
✟ ✄
✠ ✁
✆ ✠
✁ ✡
✞ ☛
✂ ✁
✂ ✄
☛ ✂
✄ ✠
☛ ☞
✌ ✍
☞
✜ ★
Session 2: Question and answer LVFXVVLRQRQGDWDHQWU\DQGFRGLQJLVVXHVDIWHUFOHDQLQJDQGUHYLHZLQJWKHGHPRHQWULHV
8SGDWHRQWKHFRGLQJUXOHV 2.5.2
Quality check on data entry
Double data entry was incorporated as part of the quality check on data. This form of quality check has been recommended and known to correct data entry errors from the original entry.
1
Double data entry was done for approximately 10.0 of the total entries in two batches where batch 1 was completed in November 2012 and batch 2 in February 2013. The selection of clinics for the double entry
was random and data entry personnel were blinded towards the double data entry process.
Type of entry Number of forms
Single data entry 11,220
Double data entry 1,377
RUHDFKEDWFKRIGRXEOHGDWDHQWU\DOOGLVFUHSDQFLHVEHWZHHQWKHÀUVWDQGVHFRQGVHWRIUHFRUGVZHUH YHULÀHG E\ FKHFNLQJ HLWKHU ZLWK WKH RULJLQDO IRUPV RU WKH FRGLQJ GHÀQLWLRQV 7UXH HQWU\ HUURUV ZHUH
LGHQWLÀHGDQGDFRUUHFWWKLUGUHFRUGZDVXSGDWHGLQWRWKHGDWDEDVH7KHSHUFHQWDJHRIGDWDHQWU\HUURU for each available variable was then calculated. The maximum data entry error percentage from each
batch was then compared.
Table 2.2: Data entry error rate for NMCS 2012
Variable Data entry error
Batch 1 Batch 2
Coded variables
ICPC code 5.85
8.96 ATC code
0.23 5.28
Non-coded variables
Visit disposition 7.63
0.87 Gender
5.02 1.84
The two variables for the non-coding section were the variables with the highest data entry error rate for batch 1. There was marked improvement in error rate for these variables from batch 1 to batch 2.
Increase of data entry error rate for the coded variables can be attributed to recruitment of new data entry personnel, resulting in more variation in coding. Most of the coding errors were systematic errors;
data entry personnel had a misconception of the way a certain diseasemedication is coded, thus making a repetitive error of the same nature.
There does not appear to be a general consensus of acceptable data entry error rate worldwide. Previous study has shown that error rates detected by double-entry method for clinical databases were 2.3 to 5.2
demographic data and 10.0 to 26.9 treatment data.
2
Similarly, Fontaine P et al. reported an overall rate of 7.3 for data entry strategies used in clinical trial.
3