Article Text
Abstract
Introduction Exercise ASKARI SERPENT (Ex AS) is a British Army exercise that provides primary healthcare (PHC) to Kenyan civilians in support of local health authorities. It is conducted in partnership with the Kenya Defence Force Medical Services (KDFMS). Accurate epidemiological data is critical in planning the exercise and for any future short-notice contingency operations in similar environments. This paper reports epidemiological data for Ex AS using a novel data collection system.
Methods PHC on Ex AS was delivered by trained and validated combat medical technicians (CMTs) using a set of Read-coded protocols. The CMTs were also directly supported and supervised by medical officers and nurses.
Results A total of 3093 consultations were conducted over a 16-day period. Of these, 2707 (87.5%) consultations fell within the remit of the CMT protocols, with only 386 consultations (12.5%) being conducted exclusively by the medical officers or nurses.
Discussion A Read-coded matrix built on CMT protocols is a simple and useful tool, particularly in civilian populations, for collecting morbidity data with the vast majority of conditions accounted for in the protocols. It is anticipated that such a system can better inform training, manning, medical material and pharmaceutical procurement than current category-based morbidity surveillance systems such as EPINATO (NATO epidemiological data). There is clear advantage to directly linking data capture to treatment algorithms. Accuracy, both in terms of numbers and condition, is likely improved. Data is also captured contemporaneously rather than after indeterminate time. Read coding has the added benefit of being an established electronic standard. In addition, the system would support traditional reporting methods such as EPINATO by providing increased assurance.
- EPIDEMIOLOGY
Statistics from Altmetric.com
Key messages
Medical support delivered by the UK Defence Medical Service (DMS) often includes support to civilian patients, and this is likely to continue with the return to contingency.
Accurate epidemiological data of patient contact is required to inform and support the rapid deployment of an effective medical capability.
Current epidemiological tools used by the military are not designed to record the full spectrum of medical treatment facility (MTF) activity.
A Read-coding matrix built on combat medical technician (CMT) protocols is an easy and useful tool for collecting morbidity data without significant resource constraints.
Such a system can better inform training, manning, medical material and pharmaceutical procurement than current morbidity surveillance systems while supporting and enabling audit.
Introduction
What is now the British Army Training Unit Kenya has been a base for British military exercises since Kenyan independence in 1964. Exercise ASKARI SERPENT (Ex AS) is a long-standing partnered training exercise with the Kenyan Defence Force Medical Services (KDFMS). It is directed by, and in support of, the local Kenyan health authorities.
The scope of medical support delivered by the UK Defence Medical Service (DMS) has, at times, included civilians.1 With the return to contingency, the DMS will likely find that this support continues whether secondary to conflict or as part of deliberate operations. In 2010, the National Security Strategy2 and the Strategic Defence Review3 defined the requirements for the future utility of force, including international terrorism, major accidents or natural hazards as well as state-on-state military crisis. The future character of military deployments will remain unpredictable,4 and so, accurate data on the likely patients will inform and support the rapid deployment of effective medical capability.
The requirement for epidemiological data is often revisited on the horizon of contingency operations. Jefferson and Demicheli wrote in 1998 of the requirement to define predictor variables of the consumption of healthcare assets by type and geographical location for operations other than war.5 Data sources previously used were generally either individual clinical collected data, admission and discharge data and/or medical reports and returns.6 Bricknell, whilst recognising the importance of a robust evidence base and a requirement for standardisation of clinical reporting, concluded that the usage of reports and returns as the primary data source was preferable.6
A variety of ad hoc data collection systems have been previously used, although incomplete collection and minimal interpretation resulted in negligible value to decision-making processes.7 Data collection became standardised with the implementation of J95, a morbidity return based on the International Classification of Diseases (ICD),8 which allowed the aggregation of data on attendance, referral, restriction of duties and time off work for all episodes of ill-health at Role 1.8 J95 aimed to do more than gather simple primary care morbidity data as it attempted to provide the morbidity surveillance used for strategic-level reports identifying needs, trends and consequent resources requirements.9 This was a classification-based system and has further evolved to reflect changes to ICD coding and North Atlantic Treaty Organization (NATO) adoption, into EPINATO (NATO epidemiological data) and now EPINATO-2. It should be noted that EPINATO-2 is not designed to record the full spectrum of medical treatment facility (MTF) activity, but to report those disease and non-battle injury stressors deemed relevant to operational effectiveness.
EPINATO-2 exemplifies a system with temporal and geographical separation between data collection, categorisation and reporting. Even where the temporal separation is reduced (EPINATO-2 can deliver daily reports), without regular audit of returns against the raw data collected in MTFs, the summarised data reported by such systems may be subject to unknown and varying levels of categorisation and reporting errors.
A review of dermatological conditions on operational deployment revealed that EPINATO style classifications were far too broad10 despite dermatological conditions accounting for nearly a quarter of all patients seen.11 Similar concerns were expressed on the accuracy of returns secondary to human interpretation.10 Flaws were recognised in the acquisition of data on diarrhoeal diseases.12 In one report, 5% of patients (n=102) on a 6-month operational tour in a forward operating base were coded into EPINATO category 16, whereas only 0.2% were described in category 20 (Table 1), which is likely to represent mistakes in coding.
Categorised reporting systems may result in a level of information that is too broad to inform future planning as was found on recent UK operations where literature review and expert opinion were required to ‘plug the gaps’.13 If categorised reports were supported by routine audit of underlying MTF data, this data would then also be available and retained to support future contingency planning.
This paper aims to demonstrate a system that could form the conceptual basis by which a contemporaneous point of treatment dataset could be generated, upon which EPINATO-2 (or other system) categorisation could then be standardised and audit of that categorisation could be routinely undertaken. In addition, it will show how this data collection can be delivered at the most forward points of clinical delivery in austere and challenging environments without significant resource implication.
Methods
Force Support Squadron, 5 Medical Regiment, with partnered KDFMS, deployed three troops of combat medical technicians (CMTs) class 1 and 2 to the Kenyan districts of Samburu, Lakipia and Meru who provided primary healthcare (PHC) to Kenyan local nationals over 16 days. The CMTs were clinically supervised by a general duties medical officer (GDMO) and a PHC nursing officer (NO). Overarching clinical support was provided by a general practitioner—the regimental senior medical officer. The capability of each troop was significantly enhanced by the KDFMS partners, which included NOs, senior non-commissioned nurses, pharmacy technicians and Kenyan military medics (akin to CMTs).
Prior to departure, a 2-day Kenya-specific training package was provided to CMTs on a platform similar to that used for Op HERRICK 18,13 followed by a clinical-scenario-based validation package to assess CMTs and GDMOs. CMTs were mandated to use CMT protocols when consulting patients with CMT 2s working to support the more senior CMT 1s. Protocol templates were selected during patient triage by an experienced CMT 1; each CMT protocol is coded (in this case using the Read-coding system), and thus, coding data is generated at the point of delivery of clinical care. The actual code used is operator independent.
The medics were assisted clinically by KDFMS personnel who also provided highly valuable translation skills. Nursing staff (with the exception of nurse prescribers) worked within patient group directions. Consultations involving pregnant women or children aged 16 or younger required mandatory review by a PHC nurse prescriber or GDMO.
The coding data was collected and reported daily along with pharmaceutical usage. The protocol headings thereby formulated a matrix by which clinical data was captured. The protocol headings contain Read codes compatible with widely used electronic patient record systems, including the UK Defence Medical Information Capability Programme (DMICP). Although deployed DMICP assets were unavailable for this exercise, Read coding allows simple integration of this data to any future electronic capture. If the GDMOs or NOs saw patients for whom the presentation (or diagnosis) was outwith the CMT protocols, this was reported as ‘other’. The ‘other’ categories were later tabulated by reviewing all GDMO and NO consultations. The data was assessed to look for commonality and trends.
Results
A total of 3093 consultations were conducted over 16 days, and 2707 (87.5%) fell within the remit of the CMT protocols with the remainder being conducted exclusively by doctors or PHC nurses. A median of 159 consultations occurred per day (mean 159, range 33–273), and 67 of the 83 (81%) CMT protocols were used; a further 44 categories (Table 2) were required when doctors or PHC nurses saw patients exclusive of protocols, many of which could retrospectively be coded into a matrix built around current CMT protocols (Table 3).
The most commonly seen condition was back pain, accounting for 11% of all consultations and 12.6% of CMT consultations, and the 10 commonest conditions made up 65.6% of the protocol matrix (Table 4). Abdominal pain of unknown aetiology remained a common problem and a diagnostic frustration. It may be relevant that indigestion and gastric symptoms were also common, and could represent symptoms secondary to Helicobacter pylori colonisation, which is thought to be highly prevalent in Kenyan adults and children suffering abdominal pain and dyspepsia.14 Of 2707 CMT consultations, 711 (26%) presented with musculoskeletal joint pain, with a further 59 patients presenting to the doctors or NOs with generalised joint pain (25% overall, n=770).
In patients seen exclusively by doctors or nurses, the commonest condition noted was ‘deworming’, making up over one-third of the ‘other category’ (Table 2). While generalised joint pain is uncommon in soldiers and is more likely to reflect a rheumatological condition, in the rural manual population of Kenya, widespread arthralgia was the second most common condition seen outside of the protocol matrix; for the most part, this was thought to be secondary to generalised muscle aches/osteoarthritis. The third most common presentation seen exclusively by doctors were eye symptoms. Although CMT protocols include various eye presentations, confidence in managing eyes was low among CMTs and may represent a training gap, which should be further investigated. However, the vast majority of eye presentations also fell outside of protocols. The patients regularly presented with symptoms of eye irritation, which, in discussion with KDFMS personnel, were attributed to cooking over open fires coupled with the effects of dust. One troop was enhanced for a day in Samburu with a Kenyan ophthalmic nurse, which accounted for the 21 cases (5.4%) of ‘specialist eye clinic attendance ?diagnosis’. Her input into Kenyan eye problems was invaluable to both CMTs, nurses and doctors.
All results were compared by region to assess for regional trends (Figure 1). Lakipia had the widest variety of presentations reflecting a more urbanised culture with some access to healthcare. It was, therefore, not uncommon for patients to already present with a diagnosed condition either for further management or a second opinion. Toothache was more common in Meru, which is likely to represent the presence of UK and KDFMS dentists in that troop. The large amount of undiagnosed abdominal pain seen in Meru was thought to be due to constipation, a side effect of the recreational use of khat (Catha edulis) grown in the region.
Overall, 19% of the CMT protocols were not used. No patients were seen with a chalazion, although this could well be due to confusion/crossover between this and a stye (n=4). No patients were reported to be seen with impetigo, which may reflect the low population densities of rural Kenya, similarly for lice. Although no patients were seen with cutaneous larva migrans specifically, 144 patients were seen by doctors for ‘deworming’, and 44 patients were seen with dermatitis, either of which could have been a manifestation of hookworm infection.
Sunburn and ‘pigmenting lesions’ representing different types of skin cancer were unsurprisingly uncommon in the African population, with no patients recorded by the medics alone. The authors do recall a patient seen with a possible squamous cell carcinoma of the lip coded in the ‘other’ section as cancer. Pitted keratolysis was unsurprisingly not seen with rural Kenyans commonly barefoot. Consequently, with barefoot patients being the usual, it was wrongly anticipated that jiggers (Tunga penetrans) would be a common cause of presentation. This may reflect public health efforts by the Kenyan Health Services in setting up mobile clinics. The lack of scorpion, leech and tick presentation may reflect a local knowledge of self-management, although scabies, perhaps due to the enduring itching, was seen in 15 instances (0.6%).
Discussion
This paper gives detailed epidemiological data from Ex AS generated at the point of delivery through the use of coded protocols. The data would allow direct audit of any categorised reports and be available to support future planning purposes.
It is acknowledged that just over 12% of cases seen were not covered by coded CMT protocols. However, 90% (2707) of cases generated consistent and auditable coding data at the point of clinical delivery. The gold standard for epidemiological data collection remains real-time hierarchical coding independent of the user. The usage of CMT protocols reduces variance by attaching a Read code to individual patient records. True independence of coding is probably unachievable in a deployed environment. Precoding of protocols (or templates for independent practitioners) requires understanding of those likely conditions in the deployed theatre. This paper, in part, contributes to that understanding for Kenya and similar environments.
Historic operations have validated the accuracy of the original EPINATO;6 however, recent operations, despite the ongoing training requirement for EPINATO, have resulted in question marks placed over the accuracy of the data collection.9 ,10 It is acknowledged that the newer EIPNATO-2 reports do not seek to provide the granularity of data reporting demonstrated by this paper. However, it is critical to the validity of those reports that the underlying dataset is auditable and audited. Ad hoc data collection from individual medical facilities, spread widely across operational environments and potentially across multiple coalition nations, with categorisation delivered by multiple individuals of varying clinical qualification must, at least, introduce the possibility of significant errors in data collection.
The coding mechanism described has delivered detailed information that can directly be used to plan pharmaceutical procurement, manning, training and validation for future Ex AS and future contingency operations. The data was made available to those units deploying on subsequent Ex AS.
Conclusions
A Read-coding matrix built on CMT protocols is an easy and useful tool for collecting morbidity data without significant resource constraints. It supports and enables the audit and quality control of any category-based force health protection such as EPINATO-2 systems. It is anticipated that such a system can better inform training, manning, medical material and pharmaceutical procurement than current morbidity surveillance systems. The utility of such a matrix would be improved with linkage to an Information Technology-based information medical capability programme such as DMICP, informed by detailed understanding of likely disease and morbidity in the deployed environment.
References
Footnotes
Contributors ITP performed the analysis and drafted the manuscript. RJW developed the concept, managed the data acquisition and reviewed the draft. PC performed overarching review and critical revision of intellectual content.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.