February 12th, 2011
Editor,
BMC Public Health
Re:
Resubmission of our manuscript previously entitled as “Road traffic injury
surveillance in a low income country: translating numbers into action” with
a revised title of “A successful model of Road traffic injury surveillance in a
developing country: Process and Lessons Learnt”
Thank you for reviewing the above mentioned manuscript and detailed comments. Based on the
comments, we have made major and extensive revision to our manuscript. These changes were
made in the light of the following comments by Dr. Dandona:
“Setting up road traffic injury (RTI) surveillance in developing country setting is not easy. The
effort made by the authors in setting one up in Pakistan is indeed commendable. Sharing of this
experience with colleagues in other developing countries is important and necessary, however,
this manuscript falls short in detailing practical steps for the others to either follow or adapt.”
This paper has now been revised to describe the detailed practical steps of setting up an injury
surveillance system. Some of the previous comments related to the presentation of surveillance
data are not relevant in light of the revised focus on the describing the surveillance system only.
In the following paragraphs we have attempted to address the reviewers’ comments in detail.
Reviewer: Rakhi Dandona
Minor essential revision
1. Replace “low-income” country with “developing” country.
Replaced
2. Page 4 (statement with references 5 and 6) – What kind of data are being indicated
here?
Rephrased
3. Introduction - Some data on the current road traffic injury mortality or morbidity in
Karachi could be added to give context to the readers.
With the change in the paper we have added reference 11, 12, 13 to the setting.
4. It is mentioned that Karachi is managed so far without an approved mass transport
policy. This statement is not clear. Does it mean that there is a policy which is awaiting
approval or there is no policy at all? Also, is transport covered under other policies such
as urban infrastructure?
The statement was removed from the manuscript.
5. Only road length and numbers of hospitals which deal with trauma are mentioned. There
is no mention of the “actual need” based on which RTIRP was put in place. It will be
useful to know how functional or dysfunctional is the current surveillance system in
Karachi because it is likely that police and emergency departments will be generating
some data which may be inadequate or of poor quality.
Need statement is now added to the last paragraph of page 5 and first paragraph of page 6
6. How was the needs assessment carried out?
No formal need assessment was carried out.
7. Page 6, strategic planning – More information is needed to provide context and practical
information for the readers who may wish to attempt similar project in their country.
A lot of practical information on how to set up a system in another country is now give.
8. How the participants for the above were identified?
A whole section on “assembling the team” was added on page 6.
9. How long did it take for the first meeting to materialize?
We unfortunately do have the information on the delay in the first meeting as the
meetings were initially very informal among the group members.
10. What steps were taken to seek funding from the corporate sector? Were other sectors also
explored? What are the possible do’s and don’ts that would be useful for others to keep
in mind while attempting the same?
A description on the funding was added to page 9. We however, have not included any
do’s and don’ts of funding
11. Page 6, data collection: How many staff were recruited for data collection round the
clock in the 5 hospitals?
A total of 18 staff were recruited to work in different shifts. An organogram is added as
figure 1 for the reference.
12. What is meant by on-site data collection?
The data that is collected at the site of crash, which includes road circumstances, type of
crash, number of people involved in the crash, vehicle condition post crash. This data
plus pertinent details from police report and ambulance (if any) report is collected by our
trained staff.
13. Was the usual patient flow studied in these hospitals for RTI patients to ensure that
majority of them were captured in the surveillance? For example – patients post crash
could report/be brought to the emergency department directly; could be taken directly to
the mortuary if brought dead; or could report to the out-patient department if seen
elsewhere for the injury before. It is important to know what were the data collection
points and how these were identified to ensure maximum data capture. If not all possible
data collection points were utilized, justification for the same should be provided.
Karachi has three government designated trauma centers. Most patients go to these
hospitals though there is a possibility that some patients get care at some of the private
hospitals. We therefore included the two major private hospitals. The surveillance does
do the newspaper surveillance as well as regularly reviews the mortuary records to
complete their count. However, it is still possible that some of the patients still go to
smaller government and private hospitals. The description is added to the last paragraph
on page 7.
14. Since data adequacy and quality are the two major concerns in developing country
setting, how were these ensured in RTIRP?
For every data collection site (hospital) there is one supervisor. For data adequacy and
quality checks, a data collection supervisor is responsible for cross checking with the data
sources; validate the number, events and other details from various sources.
15. Since the staff for data collection was available round-the-clock, in which situations was
data sought from hospital or police records?
Follow up information as well as hospital course was retrieved from hospital records.
Also, when patient was unable to give info on spot or in emergency rooms, or when next
of kin was also unavailable, in those instances hospital or police records were sought
primarily.
16. Appendix 1 –
a. Vehicle involvement – Does this variable cover both the patient who reported to
the hospital (victim) and the other party? What is recorded when “there is no
other vehicle” other than that of the patient (for example – scooter hit a tree)?
Yes, both vehicles are mentioned. When it is a single vehicle crash, the vehicle is
noted. For e.g. if a scooter hit a tree, the vehicle would be scooter and the injured
would be identified as scooter rider. .
b. Patient was – Why details of patient vehicles are not recorded (such as
motorcycle, car etc)?
These are actually recorded in the description of the crash.
c. Helmet – It is quite common in India for drivers of two-wheeled vehicles to put
the helmet on but not strap it, which is as good as not wearing one. If such
phenomenon occurs in Pakistan, it will be useful to differentiate it to plan more
effective helmet promotion campaigns.
We agree that it will be very useful but we are not collecting the information in
the current surveillance format. The focus is to see if two wheelers are putting
helmet at all or not.
d. Type of collision – If a crash was as a result of skidding of vehicle (for example –
motorbike skidded due to sand/oil/water on road), how is it captured in this form
as there is no collision involved?
When there is no collision between two vehicles, our surveillance system noted
the description of the event such as “skidding due to oil spill, or lost control of
bike over an acute turn, or slipped due to rough segment of road” etc. These
descriptors were used to define the type of collision.
e. Was alcohol consumption at the time of crash is not documented?
Information related to alcohol use was not collected. Any use of alcohol is a
crime in Pakistan and often information is not shared on alcohol use. The
description of alcohol use questions was added to page 7.
f. Who completed the clinical part of this form?
The clinical part of the form is completed by the data collectors, who have been
trained in medical data abstraction, ICD-9 CM coding and Injury scoring systems.
They have been provided with respective dictionaries and references to fill this
part of the form. The description is added to page 6.
7. Page 7, geographical and root cause analysis
a. How was high frequency of crashes defined for a site?
The frequency of crashes for a given road was defined as crashes per kilometer of road. 50
Roads with highest crashes per km were defined as those with highest frequency of crashes.
b. How many crash sites were identified, and in how many was it possible to carry out the road
safety audit?
50 such sites were identified. All of them had road safety audits.
c. How necessary is the GIS when the resources are limited? In your experience, would only
proper documentation of the site have sufficed for action?
GIS is very helpful especially in instances when the arterials are spanning across a very long
length. It is not very costly and expertise was offered to us by our engineering colleagues. Also
same purpose can be achieved by Google earth, which is free of cost. Exact latitude helps in
identifying sites which would be otherwise missed if someone documents “ MA Jinnah Road, or
Garden” etc
8. Page 7, dissemination of information – Please cite specific government and non-government
organizations (sectors and not names) to which these data were disseminated. It is neither
possible nor necessary to inform all government and non-government organizations of these
data.
So far City District Government of Karachi, Traffic Police, Highway police, City Police, Citizenpolice liaison committee, major ambulance services, all involved hospital administrations,
Karachi transport department, Ministry of Health, NED University of engineering (partners),
Corporate partners, and print media receive the quarterly report of RTIRP. With limited
resources we have been able to disseminate the information to these key players in road safety.
Results
9. How was RTI defined? Ws any injury resulting from a road crash considered as RTI
irrespective of severity of injury?
Yes
10. It should be specified that the RTI incidence is based on the injuries reporting to these
hospitals, and not population-based incidence. For surveillance, this is reasonable but should be
clearly mentioned.
Yes, this is not population based data, and the incidence is based on hospital based numbers
11.Were the data post establishment of RTIRP very different from those that are available in
routine from hospital or police records (despite inadequacy or quality)? What extra or better
information is RTIRP able to provide?
Yes, because prior no recordkeeping was kept at major government hospitals. New and extra info
is added in the form of geographical and root cause analysis, injury severity and case to case
outcome.
12. It is quite likely that there would be data gaps in the data collection form. Can some idea be
given about what kind of gaps the authors had to deal with?
We still have gaps in clinical data which may be in the form of validation of injuries and their
severity. In a low income setting, not all the injuries (especially minor) could be substantiated by
investigations, or autopsy. We follow the standard rules of AIS while classifying these injuries.
13. Page 8, paragraphs 2 and 3 - Though in the previous paragraph on this page it is mentioned
that the aim of this manuscript is to present impact of setting up RTIRP, almost no data are
presented on it. These paragraphs are generic statements of interventions or strategies put in
place but do not suggest or highlight that these were as a result of RTIRP. More information is
needed. In addition, it is important for the readers to know how where these changes
materialized. How much time did it take for a suggestion to be implemented in real time, and
how the various officials were persuaded to do so? Such an understanding is imperative for
those who plan to attempt this elsewhere.
We have removed these statements from the manuscript.
14. Setting up of the central ambulance system and of the RTI prevention centre are indeed very
important and desirable outcomes of this project. Therefore, it is important that this experience
is shared in a manner that can be of more use to the others.
Yes we agree but we cannot do justice to the manuscript describing the surveillance as well as
describing an ambulance service establishment.
15. Cost of surveillance – More specific data are needed on these costs. Even though the time by
many was volunteered and the space provided by an institution, some costing of it needs to be
taken into account to get complete picture. Cost can be presented as a break-up for the various
components – planning, data collection, central monitoring cell, data dissemination etc.
USD 45000/ year is the direct cost of people’s time both paid and volunteered. Space and
utilities were provided in kind by the host institution. .
Discussion
16. This section is quite generic including the challenges encountered. Please add some
specificity based on the comments above.
Again this section has been completely changed and now gives clear description of challenges.
17. It is not clear whether RTIRP is a standalone project or there is reasonable ownership of it
by the government or other relevant stakeholders?
RTIRP is now taken over by the Federal Ministry of Health, and is now designated as “Road
Traffic Injury Research and Prevention Center”. Other stakeholders continue to support the
project as before.
18. In terms of the impact of RTIRP, has it resulted in improvement of RTI documentation in
hospitals where the project was implemented?
We are not monitoring the hospital documentation.
19. Again, was it possible to improve data capture in police and hospital records to achieve
similar outcome rather than setting up RTIRP in a setting with poor resources? Also highlight
specifically the issues related to sustainability of this project.
We do not know if improving police and hospital records can replace the RTIRP data. Financial
sustainability is of major importance, as well as training and retaining full time staff.
Reviewer: Shanthi Ameratunga
1- Major Compulsory Revisions:
The main difficulty the paper has - in its current form – relates to its inability to assure a
successful process without including the evidence for this. As the authors note, the paper ‘aims
to present the impact of setting up the surveillance system’ with formal data to be presented
elsewhere. However, it is difficult to have one without the other when the authors make
assertions regarding the success, validity and reliability of the RTIRP without presenting
evidence to back this up. For many points made in the Results and Discussion sections (e.g.,
regarding validity, reliability, ability to direct road safety initiatives), it is important to have
evidence. Without this, readers are expected to take these conclusions at face value.
We agree with the reviewers comments. In line with Dr. Dandona’s reviews, we have decided to
keep this paper focused on the process of establishing the surveillance system instead of mixing
it with the presentation of the data.
2- It is noted that “27% of trauma victims are dealt [sic] by one particular government
hospital…”. It is not clear if the five hospitals / institutions provide data that can be
considered population-based. That is, is it only these hospitals (and none other) provide
trauma services for all RTI cases in a defined region (presumably Karachi)? If not, the
assertion made is inappropriate.
We agree with your observation that the data is not population based. We have reasons to
believe that a surveillance system based in these five hospitals captures majority of road traffic
injuries. This belief is based on the following facts: a- for many decades, there was a legal
requirement to bring all motor vehicle crash victims to the three government hospitals in the
surveillance system. Despite changes in the law about 6 years ago, the same practice continues
to a large extent. b- trauma cases which do not go to these hospitals are likely to be seen at the
two major private hospital which are also part of the surveillance. However, we have removed
the discussion on the generalizablity of the numbers as the focus of the paper has changed to
describing the surveillance project itself.
3- Even if these hospitals were the only hospitals admitting major trauma for a defined region,
the surveillance system was established in an ED setting, with the data collection form
indicating that minor injuries were included. As minor injuries can present to other settings,
it is not clear how these 5 sites can provide a regional perspective with information that is
generalisable to the spectrum of RTI. This requires clarification.
Please see response to comment 2.
4- The case definition for injury, criteria for deciding when a surveillance form would be
completed, and how levels of injury severity are determined /established, is not clear. This
information and related coding information for variables such as ‘Clinical information, data
on ‘Head and neck…), etc, would need to be uploaded (potentially a data coder’s guide) at
least as electronic files on to a journal website if readers are to interpret the approach
implied.
Any injury resulting from a road crash considered as RTI irrespective of severity of injury. The
clinical part of the form is completed by the data collectors, who have been trained in medical
data abstraction, ICD-9 CM coding and Injury scoring systems including Injury severity score,
TRISS, physiological scoring (all are already available online). They have been provided with
respective dictionaries and references to fill this part of the form. (Also corrected in the
manuscript)
5- The authors note that ‘important outcomes’ included a ‘risk factor analysis of crash’. It is
not clear how this was achieved with the data collection form and process described in the
current manuscript.
The risk factor analysis of crash included vehicle involvement, road user involved, timing and
location of crash, safety information, contributory factors, local condition of the road and injury
severity details. Aim was investigate high risk crash sites. Detailed analysis of top ten arterials
was also carried out based on road wise fatalities and serious injuries per kilometer. A team
involving law enforcement agencies, civil engineers, local officials of the city government was
responsible for the detaiedl road analysis.
6- The discussion notes that data required ‘personal validation’. As noted earlier, the process
of validating data and assuring its reliability requires elaboration with quantitative evidence
of how these aspects were quality assured. This would help justify conclusions such as
RTIRP providing data that was ‘most reliable’ (Discussion / first paragraph).
For data adequacy and quality checks, a data collection supervisor is responsible for cross
checking with the data sources; validate the number, events and other details from various
sources. The discrepancy in details provided by different sources (e.g. event details by police,
ambulance, ED record) required validation by interviewing the victim, next of kin or ED staff
etc. In this sense RTIRP data could be called “reasonably reliable” (correction done).
Minor Essential Revisions
7- The limitations section of the Discussion requires greater care. It is currently presented as a
justification for the current approach rather than an elaboration of the limitations
themselves.
Revised and added on page 11 and 12
8- The authors refer to 'private industry sponsorship' and a 'hosting institution' in the article.
These should be specified.
Revised and added on page 6, para 2
9- It is noted that the initial planning 'helped in defining goals' of the RTIRP. These goals
should be specified in the article (currently not explicitly noted).
The focus has been changed to the setting up of surveillance system rather then RTIRP. The
sentence referred above has been removed.
I hope these responses are adequately resolve the scientific issues raised by the reviewers. Please
do not hesitate to contact me if you need further information.
Best wishes,
Junaid A. Razzak MD PhD FACEP
Director WHO Collaborating Center on Emergency Medicine and Trauma
Department of Emergency Medicine
Aga Khan University
Stadium Road
Karchi