Clinicians’ management of patients potentially exposed to rabies in high-risk areas in Bhutan: A cross-sectional study

Background Rabies is endemic in southern Bhutan, associated with 1–2 human deaths annually and accounting for about 6% of annual national expenditure on essential medicines. A WHO-adapted National Rabies Management Guidelines (NRMG) is available to aid clinicians in PEP prescription. An understanding of clinical practice in the evaluation of rabies risk in endemic areas could contribute to improve clinicians’ PEP decision-making. Methods A cross-sectional survey of clinicians was conducted in 13 health centers in high-rabies-risk areas of Bhutan during February–March 2016. Data were collected from 273 patients examined by 50 clinicians. Results The majority (69%) of exposure was through dog bites. Half the patients were children under 18 years of age. Consultations were conducted by health assistants or clinical officers (55%), or by medical doctors (45%), with a median age of clinicians of 31 years. Rabies vaccines were prescribed in 91% of exposure cases. The overall agreement between clinician’s rabies risk assessment and the NRMG for the corresponding exposure was low (kappa =0.203, p<0.001). Clinicians were more likely to underestimate the risk of exposure than overestimate it. Male health assistants were the most likely to make an accurate risk assessment and female health assistants were the least likely. Clinicians from district or regional hospitals were more likely to conduct accurate risk assessments compared to clinicians in Basic Health Units (Odds Ratios of 7.8 and 17.6, respectively). Conclusions This study highlighted significant discrepancies between clinical practice and guideline recommendations for rabies risk evaluation. Regular training about rabies risk assessment and PEP prescription should target all categories of clinicians. An update of the NRMG with more specific criterions for the prescription of RIG might contribute to increase the compliance, along with a regular review of decision-making criteria to monitor adherence to the NRMG. Author summary Human rabies remains an important public health threat in Bhutan, especially in southern regions where canine rabies is endemic. The steady increase in number of patients reporting to hospitals following dog bites means escalating costs of post-exposure prophylaxis for the country. We investigated attitudes and practices of clinicians who manage patients with potential rabies exposure, in the endemic area. The risk of rabies exposure in the study area is mostly associated with dog bites, involving children half the time. Rabies vaccines were prescribed in 9 out of 10 exposure cases, while immuno-globulins were rarely prescribed. The study confirmed the perceived lack of compliance of clinicians with guideline recommendations for assessing rabies risk. This results in under-estimating the rabies risk in potentially risky exposures in high-rabies-risk areas. Our work underscore the importance of targeted training of female health assistants, doctors, and clinicians in basic health units to improve the management of rabies exposure. In particular there is need to update the national guidelines regarding indications and use of rabies immune-globulins.


59
In Asia, rabies remains a major public health threat, with an estimated 39,000 deaths annually, mostly increased from 1000 to over 7000 [7]. As per the Ministry of Health record, an estimated annual cost of 77 PEP in Bhutan is about Nu. 8.5 million (USD 131,000) -approximately 6% of the essential medicines 78 budget (S1)

79
To assess the rabies risk in potentially exposed patients, the Ministry of Health in Bhutan recommends while ARV associated with rabies immunoglobulin (RIG) are recommended for Category 3 exposures.

85
However, a shortage of RIG supply in Bhutan resulted in RIG treatment being reserved for patients with 86 exposure to suspected or confirmed rabid animals only. The current wording in the NRMG around RIG 87 prescription lacks clarity and does not provide clear direction for clinicians.

88
Despite the availability of the NRMG, there is public concern that both sporadic human deaths due to 89 rabies and rising PEP expenditure could be results of inappropriate rabies PEP prescription by clinicians.

90
However, there is currently no published evidence supporting this assumption. Therefore, this study 91 investigated rabies risk assessment and PEP prescription practices in potentially exposed patients by 92 clinicians in the high-rabies-risk areas of southern Bhutan. The aim was to provide a better 93 understanding of clinician's knowledge and practices with respect to managing patients potentially 94 exposed to rabies and identify measures to improve these where necessary to strengthen the national 95 effort to reduce the burden of rabies.

97
We conducted a cross-sectional study to assess clinicians' rabies risk assessment process and PEP 98 prescription decisions during the management of patients potentially exposed to rabies.

Study sites and participants 101
All 13 health centers with doctors in the medical staff, i.e. hospitals and grade I Basic Health Units (BHU-102 I), located in the high rabies-risk belt of southern Bhutan were included in the study (Fig 1). All clinicians 103 involved in the management of patients potentially exposed to rabies infection from animals were 104 included. The term 'clinician' in this study refers to staff who treat patients, including doctors and 105 paramedical staff (clinical officers and health assistants).

Data analysis
126

Demographics of clinicians and patients
127 Summary statistics were calculated to describe the population of clinicians performing clinical 128 assessments of patients potentially exposed to rabies, and the demographics of the patient population 129 included in the study. Given individual clinicians managed a varying number of cases potentially exposed 130 to rabies, summary statistics for clinician demographics were weighted according to the number of such 131 cases managed by each clinician. All personal identifiable information of both patient and clinician were 132 removed to anonymize and protect their privacy before the analysis of data was conducted 133 Types of exposure

134
The different types of potential exposure and the animal species involved were determined from each 135 patient's account of the exposure event. Exact binomial tests were used to test for equi-probability of 136 males versus females for various types of exposure.

138
Our questionnaire comprised a set of 23 epidemiological questions to evaluate the level of rabies risk in 139 a patient exposed to a potentially infected animal. The questionnaire was prepared using the NRMG and 140 a rabies expert panel and covered date, type of exposure, animal species involved, vaccination status 141 (dog and cat), potential rabies status of the animal and past PEP history of the patient.

142
We identified sets of relevant epidemiological questions for three types of exposures, namely: direct 143 exposure to an owned animal or a stray animal, and indirect exposure to any animal through contact

147
Firstly, we described the proportion of clinicians asking the relevant epidemiological questions 148 pertaining to cases with each type of exposure. Secondly, we independently classified each case into 149 one of three rabies risk categories (none, moderate, severe) by comparing the epidemiological 150 information provided by the patient, either during or following the consultation, with the current NRMG

151
[11] . The criteria that we used for rabies risk classification are presented in Table 1.
152 Table 1. Criteria for rabies risk assessment and recommended PEP prescription in case of 153 exposure to "suspect or rabid animals", extracted from the WHO adapted National Rabies 154 Management Guidelines (2014) in Bhutan

Exposure type Risk category Recommended PEP
Licks on intact skin, touching, feeding of animals.
Petting, bathing or coming in contact with ustensils used on a suspected rabid animal. 168 health center type: Basic Health Unit, district hospital, regional referral hospital;

170
actual level of risk of the exposure event according to the NRMG (none, moderate, severe).

171
Variables significant at P<0.3 were used to fit a multivariate model. Observations were clustered by 172 clinician, hospital and district, hence these 3 variables were used as nested random effects in the model.

173
We used a stepwise backward model selection process using the lowest Akaike Information Criterion 174 (AIC). Potential interactions between fixed effect variables in the final model were evaluated and 175 selected using the lowest AIC.

176
Assessment of PEP prescription by the clinician 177 We described the clinicians' practices in relation to the prescription of PEP, in particular ARV and RIG, of age were infrequently exposed (Fig 2). Of the patients presenting, 97% were Bhutanese nationals; 202 55% were male, and 45% were female (P>0.05). Half of the patients were preschoolers or students, and 203 16% were farmers (Fig 3). The majority of patients (208/267, 78%) presented to the health center or 204 hospital on the day of exposure or the following day. The majority of patients were potentially exposed to rabies as a result of dog bites (189/273, 69%).

212
Among dog bites, 67 (35%) were inflicted by free-roaming (also referred to as 'stray') dogs and 123 213 (65%) by pet dogs. There was no significant difference between the proportions of male or female 214 patients for each of the exposure categories presented in Table 3. 215

Rabies risk assessments 219
In nearly all consultations, the clinicians investigated the type of rabies exposure by asking relevant 220 questions during the consultation. The exact type of exposure, the date, the wound site and the species 221 involved were obtained in over 95% of clinical assessments. Details of the epidemiological information 222 that clinicians collected to assess rabies risk are detailed in (Fig 4). The risk profile of the 272 patients for whom information was available to independently classify rabies risk according to the NRMG was 57% 224 severe risk, 43% moderate risk and only 1 (0.3%) was no risk.

249
Male health assistants were three times more likely to make an accurate risk assessment than male 250 doctors, whereas female doctors were twice as likely to be accurate than female health assistants. Male 251 health assistants were 12 times more likely to make an accurate risk assessment than female health 252 assistants. Clinicians from district or regional hospitals were much more likely to conduct accurate risk 253 assessments compared to clinicians in Basic Health Units (Odds Ratio of 7.8 and 17.6, respectively),

254
independent of the clinician's designation in the different healthcare facilities. The random effects 255 (clinicians nested in hospitals nested in districts) suggested that after taking into account the variation in 256 assessment accuracy associated with clinicians, hospitals and the fixed effects, there was no residual 257 variation between districts.

Clinicians' PEP prescription practices 269
The number of patients for whom clinicians prescribed ARV by rabies risk category as assessed by the 270 clinician and as independently assessed according to the NRMG is shown in Table 6. Neither ARV nor RIG 271 was prescribed for 1 patient assessed by the clinician to be severe risk and ARV was not prescribed for a 272 second patient assessed to be moderate risk. Conversely, clinicians prescribed ARV for 10 (38%) of 26 273 patients whom they assessed as having no rabies risk and 75 (95%) of 79 patients for whom they did not 274 assess rabies risk. Considering the independent risk categorization of patients according to NRGM, 7

275
(5%) of 154 severe risk and 16 (14%) of 117 moderate risk patients were not prescribed ARV. PEP and 276 other treatments prescribed by the clinicians in this study are described in (Fig 5). 277

287
This study was the first attempt to describe and evaluate clinical practices in the management of 288 patients potentially exposed to rabies through contact with animals, in high-rabies-risk areas of Bhutan.

289
We described and analyzed consultations for 273 patients who were potentially exposed to rabies, for children under two years old, which were rarely presented. However, the age distribution of the 297 underlying study population was not taken into account, neither was the possible differential bias of 298 reporting to health centers following potential rabies exposure in adults versus children. Age-specific 299 risks thus cannot be inferred from these data.

300
We attempted to minimize information biases from the clinicians through prior communication.

301
Observational biases from the interviewer were mitigated by training on information recording.

302
However, clinicians could have been influenced towards more rigorous risk assessment during the study, 303 so the results likely over-estimate the performance of clinicians.

342
Specific rabies risk assessment and PEP training should target all clinicians involved in managing cases 343 potentially exposed to rabies, including doctors, since the latter is equally in the frontline and tend to 344 not perform as well as trained Health Assistants (particularly males), similar to Indian study conducted in 345 eight cities [14]. Gender parity in the training of health professionals should also be pursued to ensure 346 the engagement of female clinicians, as well as targeted training of staff in basic health units, as this is 347 likely to make an important difference in improving the accuracy of rabies assessments.

348
According to the NRMG, a rabies risk assessment is recommended in cases exposed to "suspected or 349 confirmed rabid animals". However, there is no clear definition for a "suspected" rabies case in animals.

350
In fact, since rabies is considered endemic in southern Bhutan, all potential vector animals involved in an 351 exposure event should be considered as suspected rabies and followed by a risk assessment in a health 352 facility. Our result indicated, by contrast, that only 71% of clinicians actually performed and documented 353 a rabies risk assessment. The findings of this study suggest that risk assessment by clinicians and PEP 354 decisions was mostly based on the type of exposure (i.e. bite, bite with puncture wound, licks, nibbles,

355
indirect exposure) which are clearly outlined in the national guidelines (Table 1). However, they tend to 356 misclassify the risk based on the patient's answer. It is a concern that 13% of patients were mis-classified 357 as having no risk, while they had moderate or severe risk.

358
Irrespective of the risk assessment, the vast majority of patients (91.6%) still received ARV treatment 359 even when the risk was not assessed. Clinicians in high-risk areas of Bhutan thus proved relatively 360 conservative in their attitude towards PEP prescription. However, for the 23 patients in the study who 361 did not receive ARV, 16 were misclassified as having no rabies risk category, including three patients that 362 were in fact in the severe risk category. Conversely, the only patient with no risk of rabies was still 363 prescribed ARV, and another in the category of moderate risk was unduly prescribed RIG. This is similar 364 to findings of a nationwide study conducted between 2005 and 2008, which reported frequent PEP 365 administration in category I exposures [10]. The discrepancy between clinician's practice and the 366 national guidelines in our study lies in underestimating the rabies risk in the first place, rather than a 367 lack of compliance with recommendations in subsequent PEP prescription. As a result, seven patients

368
(2.6%) in the highest risk category had not received the appropriate treatment (neither ARV nor RIG). In 369 this study, eight patients were exposed to "confirmed rabies" cases, all of which were cows (the 370 exposure consisting of handling the carcass and drinking raw milk). Another 76 patients were exposed to 371 animals classified as "suspected of rabies" by the clinician, including 62 dog bites. However, RIG was 372 prescribed to only 3 patients (1%) which is in concurrence with the results of the earlier study that RIG 373 was not regularly administered to dog bite victims in Bhutan [10]. Two of the patients receiving RIG

374
were bitten by dogs suspected of rabies, the third after drinking milk and handling the carcass of a 375 confirmed rabid cow. The risk for this patient was mis-classified by the clinician from moderate risk of rabies exposure [27]. By contrast, our study in high risk area showed that due to a tendency to