Brucellosis testing patterns at health facilities in Arusha region, northern Tanzania.

BACKGROUND: Brucellosis is listed as one of six priority zoonoses in Tanzania's One Health strategic plan which highlights gaps in data needed for the surveillance and estimation of human brucellosis burdens. This study collected data on current testing practices and test results for human brucellosis in Arusha region, northern Tanzania.
METHODS: Retrospective data were extracted from records at 24 health facilities in Arusha region for the period January 2012 to May 2018. Data were captured on: the test reagents used for brucellosis, procurement and testing protocols, the monthly number of patients tested for brucellosis and the monthly number testing positive. Generalised linear mixed models were used to evaluate relationships between health facility characteristics and the probability that brucellosis testing was conducted in a given month, and the proportion of individuals testing positive.
RESULTS: Four febrile Brucella agglutination tests were used widely. The probability of testing for brucellosis in a given month was significantly associated with an interaction between year of testing and facility ownership. Test probability increased over time with more pronounced increases in privately owned as compared to government facilities. The proportion of individuals testing positive for brucellosis was significantly associated with facility type and district, with individuals tested in hospitals in Meru, Monduli and Ngorongoro districts more likely to test positive.
CONCLUSIONS: Febrile Brucella agglutination tests, known for their poor performance, were the mainstay of brucellosis testing at health facilities in northern Tanzania. The study indicates that historical data on human brucellosis in Arusha and other regions are likely to provide an inaccurate measure of true disease burden due to poor performance of the tests used and variation in testing practices. Measures to address these identified shortcomings could greatly improve quality of testing and surveillance data on brucellosis and ultimately inform prevention and control of this priority disease.

Introduction

Tanzania’s One Health strategic plan (2015–2020) highlights brucellosis as one of six high priority zoonotic diseases [1, 2]. The strategic plan identifies knowledge gaps on the incidence, testing and surveillance of brucellosis among other priority zoonoses targeted for control and elimination [1, 2]. Previous hospital and community-based febrile studies have estimated that up to 7.7% of the population in the pastoralist communities of northern Tanzania are exposed to Brucella [3], while 6.1% and 3.5% of acute patients from rural pastoralist and peri-urban, agro-pastoral settings respectively have been identified as confirmed brucellosis cases [4, 5]. The use of standardised protocols for testing, use of internationally recognised diagnostic tests and use of a standard case definition to generate surveillance data on human brucellosis are not currently practiced in Tanzanian health facilities [1].

In Tanzania, the guidelines for surveillance and reporting of prioritized diseases, recommend the use of the electronic, Integrated Disease Surveillance and Reporting system (IDSR) [6]. The system, which is used by health facilities to report disease case data to national surveillance systems, is designed to identify cases of priority, notifiable diseases based on the 10th version of the WHO International Statistical Classification of Diseases and Related Health Problems (ICD-10) [6-8]. The IDSR guidelines were first incorporated into the Tanzanian health system in 2001 and included 13 priority diseases. In 2011, the national IDSR guidelines were revised to include surveillance of 34 priority diseases and conditions in its second phase [9]. Although brucellosis is not a notifiable disease, it is a priority zoonosis in Tanzania, that should be reported at health facilities [6]. However, despite these goals, the IDSR system is not fully integrated into health facility information management systems, especially in rural, primary health facilities and brucellosis is currently not routinely included as a priority disease within the IDSR system [6]. The second phase of the IDSR platform is currently used in most district and referral hospitals, but in the primary health facilities paper-based facility logs are still widely used [10, 11]. The variation in testing protocols at different health facilities for the diagnosis of brucellosis also complicates utilization of existing data. Current practices often lead to misdiagnosis of brucellosis as one of a range of other common causes of febrile illness [3, 5].

The Tanzania national guidelines for brucellosis testing are based on international guidelines [6, 12, 13]. The current Tanzanian guidelines for human testing specify the screening of suspected cases with the Rose Bengal Test (RBT), with additional Rivanol precipitation testing and confirmation using either culture and isolation of bacteria or a positive molecular test result [6]. Serology tests cannot be used to confirm a brucellosis diagnosis using a single acute-stage sample nor can they differentiate the species of Brucella in positive samples [14]. A number of research studies conducted in regions where higher prevalence of brucellosis is suspected have successfully applied the CDC and WHO recommended diagnostic protocols [5, 15–18]. To date, however, none of these recommended diagnostic protocols have been applied at scale for the diagnosis of human brucellosis in Tanzanian health facilities [10, 19, 20]. Instead, studies conducted in Tanzania and the East Africa region have identified a range of rapid, febrile Brucella agglutination tests (FBAT) that are used in health facilities to test for brucellosis [18-21]. None of these rapid tests are recommended for brucellosis diagnosis in any national or international guidelines and several studies to date have indicated poor performance of these rapid tests [18, 21, 22].

Tanzania has over 5,000 existing primary health care facilities (518 hospitals and health centers and 4,554 dispensaries) [23, 24]. The health service organization follows a pyramid structure starting at community level with primary health facilities (dispensaries and health centers), then leading to hospitals serving at the district level and finally up to referral hospitals, providing services to whole administrative regions or zones [25]. Basic health service provision is generally focused at primary health facilities in the ward, village and sub-village levels, with oversight and management coordinated at district level facilities. A hospital or health center at the district level can be designated to manage the health services within the district, and is then referred to as a designated district hospital or health center, respectively. The health information system used in primary health facilities is part of the Mfumo wa Taarifa za Uendeshaji Huduma za Afya (MTUHA translated as “health services information management system”) framework. The MTUHA, which is a paper-based record management system, was first introduced into health facility systems in 1993 [11]. A review of the MTUHA framework and incorporation into IDSR systems was done in 2011, leading to a new national training and implementation program in 2012 [11]. Records of patient management at each facility are generated within a health information system designed to comply with the IDSR framework. Departmental logbooks used within each facility record individual level details of patient testing and results, including brucellosis testing, where performed [11, 26]. Patient records are aggregated daily for reporting into the district health information software (DHIS) through a set of facility level log books [11]. These facility level summaries do not record brucellosis as a specific etiology of illness and instead aggregate brucellosis diagnoses among a range of diseases as “other causes of fever”. The DHIS further aggregates data into regional and referral health systems that are compiled for the Ministry of Health Community Development, Gender, Elderly and Children surveillance systems [23]. Data on brucellosis diagnoses specifically are therefore not routinely compiled or reported from the aggregated facility level summary records upwards through the current reporting system.

Previous studies conducted in Arusha region have demonstrated high prevalence of human brucellosis using the WHO and CDC recommended guidelines for diagnosis and case definitions [4, 27–29]. Higher prevalence has also been documented in previous studies within rural, pastoralist communities [17, 30, 31], which make up a large proportion of the population in Arusha region and northern Tanzania overall as compared to urban and peri-urban communities [32, 33].

This study included a survey of health facilities in Arusha region to evaluate the current practices in place for generating data on the diagnosis of human brucellosis. The study aimed to quantify the patterns in brucellosis test performance across facilities, months of the year and the proportion of patients tested who were brucellosis positive at included health facilities. The study also aimed to identify the test reagents used for diagnosis of brucellosis, their procurement and the protocols used for their performance at health facilities within Arusha region. The findings of this study provide a quantitative characterization of current brucellosis testing practices in northern Tanzania and can be used to inform future improvements in surveillance systems for this priority disease. The findings of this study can inform policy for improved diagnosis and reporting of brucellosis, and also more general guidance on surveillance of a wide range of emerging and priority zoonoses in Tanzania as with many lower and middle income countries (LMICs) in sSA [34, 35]. Improved surveillance and reporting of priority diseases has the potential to increase health care services, public health in general and the capacity for health authorities to quickly and effectively respond to disease outbreaks [36].

Methods

Ethics and approval

The study was approved by the Kilimanjaro Christian Medical College Research Ethics and Review Committee (CRERC, Cert. 829) and the National Institute for Medical Research (NIMR/HQ/R.8c/Vol. II/1140). The study also obtained written approval from national, regional and district level authorities as well as permission to access institution level data from the respective health facility authorities, prior to collection of data. The identity of the institutions included in the study were masked for confidentiality. No individual patient level data were accessed for this study.

Study area

Data for this study were compiled over the period from September to November 2018 in the seven administrative districts of Arusha region; Arusha Urban, Arusha District Council (DC), Meru, Longido, Monduli, Karatu and Ngorongoro districts (Fig 1). Arusha Region, located in the northern zone of Tanzania, has a population mainly comprised of pastoralist and agro-pastoralist livestock keepers in the rural areas with a mixed agro-business urban population [32, 37].

Fig 1

Map of Tanzania showing the seven administrative districts in Arusha Region (yellow shading) where the study was conducted.

Yellow shaded polygons in the insert show Arusha districts within an outline map of Tanzania (no shading). Map created using R software version 3.6.3 and the tmap R package [38]. Shape files for administrative boundaries from the 2012 census were sourced from the Tanzania National Bureau of Statistics (https://www.nbs.go.tz/nbs/takwimu/references/Licence-Agreement-NBS.pdf) [39].

Data collection process

To create a list of all registered health facilities certified to perform brucellosis testing, the regional health management team (RHMT) responsible for oversight in Arusha region were consulted. District health facilities designated to manage health services within each district were identified at the Arusha Regional Medical Office (RMO) by accessing the DHIS. Through collaboration with the RHMT personnel and district laboratory technologists, facilities within each district with laboratories that had been certified to conduct brucellosis testing were identified and listed. District Medical Officers (DMOs) of each district were contacted in advance of all data collection visits. Contact details for the health facilities registered and approved to conduct brucellosis testing were then obtained from the respective DMO and district laboratory technologist and all listed facilities were approached to introduce the study. An appointment was made with a management representative for each facility, by calling ahead. At least three call attempts were made to each approached facility in order to set up a visit. All facilities where phone contact was unsuccessful after three attempts were excluded from the study. At each facility visited, a brief introduction of the study objectives was presented to the respective representative authority as well as the key heads of department facilitating the data collection, lab technicians in charge and in most cases, the health secretaries of the facility to enable sharing of data from records. A structured questionnaire was applied to capture details of current brucellosis testing practices, reagent procurement systems in use and protocols used for testing brucellosis at each facility. The respondent in each case was the person responsible for collection of data on brucellosis for the relevant department including the laboratory, in- and out-patient and records manager. For each month in the period from January 2012 to May 2018, the following data were extracted from facility records:

number of patients tested for brucellosis

number of patients recorded as test positive for brucellosis

Data handling and analyses

Data were entered in Microsoft Office Excel, and all analyses performed using R [40]. For initial summaries of testing records, all months with non-missing data on the number of individuals tested for brucellosis and number positive in that month were included. Generalized linear mixed models were used to evaluate variable associations with two outcome metrics. First, the performance of any brucellosis testing (or not) for each month and the proportion of all brucellosis tested patients who were test positive in each month (including only months when testing was performed). For both models, data from one private testing laboratory were excluded. The data from this facility indicated that brucellosis testing was performed on the majority of individuals, indicating a test service provision role as compared to the testing based on clinical patient evaluations apparently performed at the other included facilities. Variables evaluated in these models included the year of data collection, month of the year, district, type of health facility and health facility ownership. For model analyses the facility type variable was simplified to two levels to differentiate health centers (designated district health centers, district health centers, health centers, private health centers and other health facilities) and hospitals (designated district hospitals, private hospitals and regional referral hospitals). Facility ownership was also simplified to two levels to differentiate government and privately owned (including privately owned and those run by faith-based organizations) facilities. Interaction terms evaluated included a year and ownership interaction for the model of brucellosis test performance and a year and month interaction for the model of proportion testing brucellosis positive. A random effect for the health facility ID was included in both models to account for repeat observations at each facility. For the model with proportion of all brucellosis tested patients who were test positive in each month as outcome, an observation level random effect (unique for each combination of facility ID and month of data collection) was also included to account for overdispersion. For each outcome modeled, initial maximal models including interaction terms were fitted and model simplification performed using likelihood ratio tests (LRT), with a significant p value of ≤ 0.05. Residual diagnostics for final models were performed using the ‘DHARMa’ package in R [41].

Results

Facility identification and characteristics

A total of 86 health facilities registered and approved to conduct brucellosis testing were identified and approached within Arusha region. Of these 24 (27.9%) were visited for data collection for this study. Of the remaining facilities, 51 (59.3%) could not be reached after three attempts and 10 (11.6%) were contacted but declined participation in the study stating grounds that they did not perform brucellosis testing. A single facility (1.2%) withdrew after initial enrollment in the study (Fig 2). The characteristics of the health facilities approached for the study and of those that contributed data are shown in Table 1. Additional details for all facilities contributing data are also given in S1 Table. Lower levels of participation in the study are observed for facilities in Arusha DC and Arusha Urban districts as compared to other districts. Participation was also less likely for health centers as compared to hospitals.

Fig 2

Flow diagram showing the process for identification and inclusion of health facilities in the study.

Table 1

Summary of the district location and facility type of all health facilities (n = 86) approached for participation in this study and the subset (n = 24) that contributed data.

Variable		N total facilities approached	N (%) facilities contributing data
District	Arusha DC	18	4 (22.2%)
	Arusha Urban	41	7 (17.1%)
	Karatu	6	3 (50.0%)
	Longido	4	1 (25.0%)
	Meru	7	4 (57.1%)
	Monduli	5	2 (40.0%)
	Ngorongoro	5	3 (60.0%)
Facility type	Hospital	12	8 (66.6%)
	Health centre	74	16 (21.6%)
TOTAL		86	24 (27.9%)

Monthly brucellosis testing data

Over the 77 months of the data compilation period (from January 2012 to May 2018 inclusive), 1362 monthly entries with complete data on brucellosis testing and results were extracted for 23 facilities (one private testing laboratory excluded) from the MTUHA departmental logbooks, with 409 months missing data. Only two of 23 facilities (8.7%) had contributed complete data on monthly testing for all 77 months of the study period. The 23 included facilities contributed a mean of 59 months data with a range of 16 to 77 months. This translates as overall data completeness of 76.9% for all included facilities and ranging 20.8–100% for individual facilities. Of the 1362 monthly facility entries, 776 (57.0%) entries were for months and facilities where testing of one or more persons for brucellosis was recorded. The data completeness for each year of the survey period is shown in S1 Fig. The compiled data include records of 85,013 brucellosis tests, of which 14,481 (17.0%) were test positive.

The final generalized linear mixed model with the outcome of any brucellosis testing (or not) for each month included a significant interaction between facility ownership and year (Table 2, Fig 3). The coefficients for the year variable main effect indicate increased probability of testing over time with a clear increase from 2012 to 2015 and then more stable probability in following years. The influence of this year effect on test probability was variable between private and government owned facilities however, with more pronounced increases (relative to 2012) in test probability for private compared to government owned facilities and variable effects over different years. The predicted probability of brucellosis testing by year and facility ownership is illustrated in Fig 3. The performance of testing or not in each month of the survey period is also shown in Fig 4. A random effect variable for the facility ID was included in the model and the intra-cluster correlation was 0.88. The marginal R2 for this model was 33.7% and the conditional R2 was 88.4% (Table 2).

Table 2

Multivariable model of variables associated with brucellosis test performance by month at health facilities in Arusha region (n = 1300 monthly observations from 23 facilities).

Variable	Levels	Odds Ratio	95% CI	p -value	LRT χ2	p—value
Intercept		0.02	0.00–0.034	<0.001
Year	2012 (ref)	-	-	-
	2013	8.91	1.29–61.64	0.027
	2014	8.32	1.22–56.59	0.030
	2015	19.96	2.94–135.63	0.002
	2016	29.92	4.40–203.65	0.001
	2017	17.47	2.58–118.43	0.003
	2018 January to May	24.85	3.35–184.40	0.002
Facility ownership	Government (ref)	-	-	-
	Private	9.45	0.16–550.47	0.279
Year * Facility own	2013 * Private	1.03	0.08–13.58	0.979	46.17	<0.001
	2014 * Private	63.52	3.39–1191.52	0.006
	2015 * Private	56.64	2.94–1085.85	0.007
	2016 * Private	12.66	0.69–232.22	0.087
	2017 * Private	147.24	7.31–2966.13	0.001
	2018 * Private	19.74	0.91–426.19	0.057

CI- confidence Interval of estimate; LRT–Likelihood ratio test; χ2 –Chi square; Random Effects: σ2 = 3.29; Facility ID = 15.59

Fig 3

Predicted probabilities of brucellosis testing per month by year and facility ownership type.

The red dots indicate point estimates with confidence intervals for government owned facilities. Blue dots indicate point estimates with confidence intervals for privately owned health facilities.

Fig 4

Proportion of individuals tested per month receiving a brucellosis positive laboratory test result over time.

The plots show data from all facilities and months included in the model analyses. Red points just below y = 0 indicate months and facilities in which no brucellosis testing was performed. Panels show the data subset by facility ownership (government or private) and facility classification (health center or hospital).

The mean proportion of individuals tested for brucellosis who had a positive result recorded ranged from 0 to 100 with a mean of 21.0% (95% CI = 0–36.2). The final generalized linear mixed model with the outcome of proportion of individuals testing positive for brucellosis for each month (including only records for months where some brucellosis testing was performed) included two variables that were significantly associated with the proportion positive (Table 3). First, facility type, with higher proportions testing positive in hospitals compared to health centers. Second, the district with a significantly higher proportion of individuals testing positive in Meru and marginally higher proportions positive in Monduli and Ngorongoro as compared to the reference level of Arusha DC. Random effect variables for the facility ID and an observation level random effect (for facility month combination) were included in the model and the intra-cluster correlations were 0.23 and 0.59 respectively. The marginal R2 for this model was 9.4% and the conditional R2 was 14.4% (Table 2).

Table 3

Multivariable model of variables associated with proportion of individuals testing positive for brucellosis at health facilities in Arusha region (n = 776 monthly observations from 20 facilities).

Variable	Levels	Odds Ratios	95% CI	p-value	LRT χ2	p-value
	(Intercept)	0.11	0.05 – 0.28	<0.001
District	Arusha DC (ref)	-	-	-	14.49	0.025
	Arusha Urban	1.04	0.42 – 2.60	0.930
	Karatu	0.83	0.30 – 2.26	0.713
	Longido	2.08	0.23 – 18.78	0.516
	Meru	3.59	1.14 – 11.29	0.029
	Monduli	3.49	0.93 – 13.04	0.063
	Ngorongoro	2.62	0.98–7.00	0.056
Facility type	Health center (ref)	-	-	-	4.56	0.033
	Hospital	1.89	1.07 – 3.33	0.027

CI- confidence Interval of estimate; LRT–Likelihood ratio test, χ2 –Chi square; Random Effects: σ2 = 3.84; Facility-month = 0.55; Facility ID = 0.20.

Testing reagents and procedures used in the diagnosis of brucellosis

A total of 21 (87.5%) of the 24 facilities included in this study reported conducting brucellosis testing at the time of the study visit (September to November 2018). Of these facilities, 18 (85.7%) had brucellosis testing reagents in the laboratories at the time of the study. All 18 participating facilities used a type of rapid Febrile Brucella Antigen Test (FBAT) at the time of the study (Table 4). All 24 study facilities reported previous use of at least one of the named FBAT kits: Arkray, Eurocell, Fortress or Genuine Biosystems. Procurement of test reagents was done entirely through private suppliers in Arusha Urban district. All of the surveyed facilities reported purchasing brucellosis tests from one of two private distributors. All the facilities included in the study used test results to define or inform case definitions for brucellosis. A total of five of 18 facilities (20.8%) also responded that diagnosis of brucellosis was regularly based on clinician judgment (case presentation and history taking), particularly when there were shortages of testing reagent. One (4.2%) of the 24 health facilities reported previous use of the Rose Bengal Test, although it was not in use at the time of the study (Table 4). Five (27.8%) of the 18 facilities with test reagents on premises reported using controls in performance of their routine test runs, or as a regular quality assurance measure for reagents used. The facilities using controls for testing used previously identified “positive sera” in two of five (40%), or kit provided control sera in three of five facilities (60%) as test controls. In the 18 health facility department logs where brucellosis testing was conducted brucellosis was recorded as “Positive” or “Negative” for 15 (62.5%) facilities and as Brucella abortus, melitensis for three (12.5%) facilities.

Table 4

Data on facility level use of brucellosis tests for past reporting (n = 24 facilities) and test kits present at time of survey (n = 18 facilities).

	Reported historical use of tests (n = 24 facilities)		Test present at time of survey (n = 18)
Test/Kit	No. of facilities	Percentage (95% CI)	No. of facilities	Percentage (95% CI)
Eurocell	9	37.5 (18.8–59.4)	17	94.4 (72.7–99.6)
Fortress	8	33.3 (15.6–55.3)	10	55.6 (30.8–78.5)
Genuine Biosystems	8	33.3 (15.6–55.3)	3	16.7 (35.8–41.4)
Arkray	5	20.8 (7.1–42.2)	2	11.1 (1.4–34.7)
Any of the four Febrile Brucella Agglutination Tests	24	100.0 (36.6–100.0)	18	100 (81.4–100)
Rose Bengal Test	1	4.2 (<1.0–21.1)	0	-

Discussion

In this study we observed that the performance of brucellosis testing or not in each month at health facilities was significantly associated with an interaction between the year of data collection and facility ownership. All participating facilities used the FBAT kits for testing, procured from private distributors, and only one facility reported ever using the Rose Bengal Test. The proportion of individuals testing positive for brucellosis for each month was significantly associated with the district and classification of the facility where testing was done.

This study found that 80.6% of the health facilities included in this study had reagents or a kit for brucellosis testing at the time of the study. Four FBAT kits were being used in health facilities (Table 3), with the Eurocell (80.6%) the most commonly used at the time of the study and previously reported to have been used at 47.6% participating facilities during the entire data extraction period. None of the health facilities included in the study used the Rose Bengal Test at the time of the survey, although one (4.2%) reported its previous use for diagnosis of brucellosis during the full study period (Table 3). Three of the FBAT test kits commonly used in health facilities across Arusha region have documented poor performance in testing for brucellosis using patient samples from this region [18, 21]. It was also observed that most facilities (72.2%) did not ensure quality of testing by use of controls, and 40% of those that did reported using example patient sera as positive and negative controls rather than standards with validated results. Records in 12.5% of facilities reported brucellosis test results as B. abortus or B. melitensis, misrepresenting the ability to distinguish the species of Brucella using serology tests. This fact and the observed widespread use of FBAT kits raises concerns about quality of current practice for brucellosis testing. The FBAT tests have been shown to have low specificity and poor diagnostic accuracy [18, 21] in testing for brucellosis. This not only points to high rates of patient misdiagnosis, but also contributes to a likely inaccurate and misleading picture of the epidemiology and overall clinical burden of brucellosis [13, 21, 42]. All the FBAT kits identified in this study were purchased through private suppliers and distributors of reagents. Commercial procurement of reagents through decentralized sources leads to variation in price and availability of reagents for brucellosis testing also observed in this study (S1 Table) [22, 43, 44].

In 2011 the Tanzanian IDSR platform that involves the use of the paper-based log books was reviewed and found to be underutilized as an effective data collection tool [11, 23, 26]. As a result, the new guidelines implemented in 2012, involved training of key health facility staff and improved supervision of data reporting. The completeness of data extracted from facility log-books on brucellosis testing was higher in all years after 2012 (Table 1, S1 Fig). The regression analysis showed that the performance of brucellosis testing (or not) in each month was significantly associated with the year in which testing was done and the ownership of the health facility, with an interaction between these two variables. The probability of testing increased over time and in private as compared to government owned health facilities, but with a less pronounced and more variable increase in probability of testing over time in government as compared to privately owned facilities (Table 5, Fig 3). Both the observed increase in data completeness and proportion of months in which brucellosis testing was performed later in the data collection period could be associated with better reporting of testing at facilities likely indicating a beneficial effect of the update IDSR guidelines and training provided from 2012 onwards. This finding indicates the benefits of the efforts made to provide training, increase awareness and ensure systematic aggregation of data from health facilities to improve data quantity compiled for health facilities [11].

Table 5

Characteristics of monthly level data from 24 health facilities over all months of observation (n = 1362 months of observation).

Variable	Category	Monthly observations (N = 1362) n (%)
Year	2012	65 (4.8)
	2013	171 (12.6)
	2014	219 (16.1)
	2015	255 (18.7)
	2016	263 (19.3)
	2017	275 (20.2)
	2018 January to May	114 (8.4)
District	Arusha DC	190 (14.0)
	Arusha Urban	423 (31.1)
	Karatu	209 (15.3)
	Longido	62 (4.6)
	Meru	228 (16.7)
	Monduli	55 (4.0)
	Ngorongoro	195 (14.3)
Facility type	Hospital	452 (33.2)
	Health center	910 (66.8)
Facility ownership	Government	640 (47.0)
	Private	722 (53.0)

The observation that testing for brucellosis was more likely in private compared to government owned health facilities for several years of the survey period could be because private health facilities are typically better resourced than government facilities and thus equipped to introduce and sustain more brucellosis testing. It could also reflect the fact that services provided in private health facilities are more patient driven, indicating demand-driven brucellosis testing as has been observed in other studies [10, 45, 46]. It is important to recognize that the clinical rationale for the performance of brucellosis testing (or not) is likely to have been variable across the temporal and spatial intervals captured in this survey and that this underlying but un-observed variation in the clinical ‘need’ for brucellosis testing in each month and location is likely to also contribute to the variation observed in these data.

The study found a mean proportion of 21% patients that had a brucellosis test were classified as positive. A previous study among high-risk abattoir workers in Tanga Region using the Rose Bengal Test reported a prevalence of up to 19.5% [47]. This mean 21% reported in Arusha Region over the seven-year study period is slightly higher than the focused abattoir workers’ study, probably due to the higher test positivity reported with most FBAT kits in Arusha health facilities compared to the Rose Bengal Test used in the Tanga study [18, 22]. The higher prevalence observed in this dataset as compared to that recorded with the RBT in the targeted, high-risk group prevalence reported among slaughterhouse workers does not seem epidemiologically plausible.

The proportion of patients testing positive for brucellosis was significantly associated with the classification of the facility with a higher probability of testing positive for brucellosis in hospitals as compared to health centers (Table 2). This association could be due to the referral system of patients that lead to complicated cases (including probable brucellosis presentations) being seen in hospitals as compared to health centers and thus potentially representing a higher proportion of the tested patient population [25]. Brucellosis typically presents with non-specific symptoms and cases are often misdiagnosed as other causes of illness. Cases are therefore more likely to be defined as brucellosis after numerous visits to health facilities, including upward referral to hospitals. Patients tested in Meru district were significantly more likely to get a positive result compared to Arusha DC. Testing in Monduli and Ngorongoro districts was associated with marginally significant increased test positivity as compared to the reference level of Arusha DC. The populations in these three districts are comprised of large proportions of pastoralist communities [33], who experience increased exposure to a range of zoonotic pathogens, including brucellosis. The likely higher true prevalence of brucellosis in Monduli and Ngorongoro districts is indicated by previous reports [10, 29, 31]. However, the observed association with the district could also be explained by a greater experience of clinicians in these areas with brucellosis and as a result, the application of more stringent criteria for brucellosis testing, only testing patients with high clinical suspicion for brucellosis. Improved awareness and knowledge of brucellosis presentations among healthcare workers, including clinical history taking and evaluation of known risk factors in patients [15, 17, 45, 48–50] have been shown to increase the accuracy of diagnosing brucellosis in cases and non-cases, even with tests shown to have low specificity and diagnostic accuracy [12, 42]. In this study, although all facilities reported relying on test results to inform patient diagnoses, 20.8% of facilities reported frequently relying on clinical diagnoses for brucellosis, especially in scenarios when laboratory testing was hampered, mostly due to lack of test reagents (S1 Table). Overall, our findings highlight areas for improvement in diagnosis and data generation on brucellosis through the use of improved diagnostic tests and the use of consistent case definitions (e.g. including both test results and clinical diagnoses for brucellosis).

This study had a number of limitations; first, the low proportion of facilities that took part in the study, with lower proportions of health centers and facilities from Arusha Urban and Arusha DC districts participating. Approximately 70% of the 86 facilities that were identified to have brucellosis testing capacity either declined to take part in the study or were not reachable. This limits the inference that can be made from study findings. We are therefore not able to investigate the influence of test kit on proportion testing positive.

The results of this study help to identify key areas for improvement to ensure future adherence with Tanzanian national brucellosis surveillance and control guidelines in health facilities. Updates to current practice such us uniform use of the RBT for testing, using standardized protocols and reliable, centralized supply are critical to achieve improved diagnosis of brucellosis in Tanzania [18, 51, 52]. The national medical stores department would be an ideal organization to regulate and sustain the supply of approved testing reagents, including the Rose Bengal Test for brucellosis [44, 53, 54]. The implementation of the IDSR in health facilities would also provide a platform for the standardized reporting of brucellosis cases, potentially leading to the improved amount and quality of data to inform on disease burden. These strategies would eventually improve the aggregation of brucellosis data.

Conclusions

This study characterizes brucellosis testing practices at health facilities in Arusha region over a retrospective period from January 2012 to May 2018. The performance of brucellosis testing was more likely to be observed after 2012 and at private health facilities compared to government owned, while test positivity was higher in hospital class facilities and particularly agro-pastoral community districts. Centralized sourcing of test reagents could potentially address variations in access to and prices of recommended reagents in effect reducing variations in probability of testing health facilities. The tests reported in use were all FBATs with known poor performance. Among the health facilities that had test reagents on site and reported testing for brucellosis, different reagents and protocols for testing of brucellosis were observed. The systematic use of the recommended Rose Bengal Test for brucellosis, specific recording and its reporting from primary health centers would provide more accurate disease burden data enforcing public health strategies for improved control and management of brucellosis and other priority zoonoses in health facilities. The lessons learned from this study could apply to surveillance and control programs for various diseases in Tanzania and equally apply to national and healthcare systems in many similar LMICs.

Graph showing data completeness for the 23 facilities contributing data over the seven years of the data collection period.

Data completeness is calculated assuming that all facilities operated for the full study period and accounts for the fact that only five months of data were included for 2018.

(DOCX)

Click here for additional data file.

Characteristics of facilities included in the study from January 2011 to May 2018 inclusive (n = 24 facilities).

CI- Confidence interval.

(DOCX)

Click here for additional data file.

5 Nov 2021

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Aleksandra Barac

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

3. We note that Figure 1 in your submission contain [map/satellite] images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

a. You may seek permission from the original copyright holder of Figure(s) [#] to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text: “I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only. The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/ The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/ Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/ Landsat: http://landsat.visibleearth.nasa.gov/ USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/# Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: 1- In the Methods section, can be improved by starting with Ethics and approval. Also, you mentioned (This study was conducted from September to November 2018), but in the abstract you mentioned (Retrospective data were abstracted from 24 health facility records in the Arusha region from January 2012 to May 2018), please check?

2- The authors need to sort the abbreviations in one table, especial most of them aren't international or familiar to readers.

3- The authors need to provide more explanation for the statistical part of their study, I would suggest re-analyzing the data to integrate these pieces into the manuscript presentation and rewrite the result section by sorting the table with main data: How many health facilities in the Arusha region during all the period of the study? How many patients went to that? How many do the test?

Reviewer #2: This study was to reveal the relationship between health facility characteristics and the probability that brucellosis testing conducted in a given month, and the proportion of tested individuals who had positive results. While the biggest limitation was the low proportion of facilities participating in the study, and the areas where study conducted were only small part of the country, which resulted in the data unrepresentative. However, this study could inform public health authorities to improve survaliance and management of brucellosis and other priority zoonoses in health facilities. So it has some meaning. I suggest it to be modified into a short paper.

Reviewer #3: Manuscript Number: PONE-D-21-21073

Title: Brucellosis testing patterns at health facilities in Arusha region, northern Tanzania

I appreciate the efforts put by the authors to assess the variables associated with probability of testing for brucellosis and proportion of individuals classified as test positive for brucellosis. The authors have tried to gather data through structured survey from various test facilities and factors influencing test positivity and proportion of test done and to assess the areas of focus to achieve improved quality of testing and surveillance data on brucellosis. However, I feel more clarity is needed for some observations such as lack of year wise data for facility level observations. Moreover, correlations between year and monthly positivity data in regression model need to be based on more complex data analysis taking into account of variable interaction and various confounding factors. The model should take in account of these parameters for more reliable models. Apart from this major limitation, some of the major points need to be addressed are listed below:

Overall, grammar mistakes need to be addressed throughout the manuscript. For e.g.,

Line 102: ‘different’ need to be replaced by ‘differentiate’.

Line 105: The sentence is grammatically incorrect, need to be reframed.

Line 114: Grammatical error has to be replaced by ‘have’

Line 132: typographical error for ‘where’ instead of ‘were’

Major comment:

The model should incorporate confounding factors and interaction between variables. The reliability of model needs to be measured by suitable tests.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: Yes: Pankaj Dhaka

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

20 Dec 2021

Point-by-point Responses to Reviewer Comments (bolded text)

Reviewer #1:

R1- In the Methods section, can be improved by starting with Ethics and approval.

We thank the reviewer for this suggestion and in response, the ethics content has been moved to the start of the methods section; see pg. 8, line 148-154.

R1 - Also, you mentioned (This study was conducted from September to November 2018), but in the abstract you mentioned (Retrospective data were abstracted from 24 health facility records in the Arusha region from January 2012 to May 2018), please check?

We thank the reviewer for this inquiry on clarity. The two time periods referred to are the period over which data were gathered for this study (September to November 2018) and the period over which the data were generated (Jan 2012 to May 2018). We have revised the manuscript in to clarify this and ensure consistency of language. The updates made are as follows to reflect this clarification:

In the abstract (Lines 30-31) - “Retrospective data were extracted from records at 24 health facilities in Arusha region for the period January 2012 to May 2018…”

And methods sections (Line 157-158) – “Data for this study were compiled over the period from September to November 2018 in the seven administrative …”

And (Lines 193 – 194) – “For each month in the period from January 2012 to May 2018, the following data were extracted from facility records...”

R1- The authors need to sort the abbreviations in one table, especial most of them aren't international or familiar to readers

A - We thank the reviewer for this observation and in response have rechecked that all abbreviations are fully explained as per journal guidance; i.e. defining abbreviations at every first mention in the manuscript, particularly for repeat and non-standard abbreviations. To ensure that we follow journal format guidance (and after checking with the editorial team) we have opted not to include a table of abbreviations in the manuscript.

R1- The authors need to provide more explanation for the statistical part of their study, I would suggest re-analyzing the data to integrate these pieces into the manuscript presentation and rewrite the result section by sorting the table with main data:

R1 a -How many health facilities in the Arusha region during all the period of the study?

A - We have reviewed and updated the analyses reported in several ways. These include the revision of content in the first results section on facility identification and characteristics (pg 11,). The updates in this section include the addition of the details requested on the total number of tests and positive results included in the dataset. Lines 230-236 now read…

“A total of 86 health facilities registered and approved to conduct brucellosis testing were identified and approached within Arusha region. Of these 24 (27.9%) were visited for data collection for this study. Of the remaining facilities, 51 (59.3%) could not be reached after three attempts and 10 (11.6%) were contacted but declined participation in the study stating grounds that they did not perform brucellosis testing. A single facility (1.2%) withdrew after initial enrollment in the study (Figure 2). The characteristics of the health facilities approached for the study and of those that contributed data are shown in Table 1..”

R1 b -How many patients went to that? How many do the test?

A – We do not have complete data on the total number of individuals seeking care at all facilities in each month so have not presented these data. The data compiled are on the number of individuals tested and positive and we have provided further details on the totals for these populations. This point has also been addressed in lines 264-266 which now read –

“The data completeness for each year of the survey period is shown in supplementary file Figure S1. The compiled data include records of 85,013 brucellosis tests, of which 14,481 (17.0%) were test positive.”

A - We have also updated the methods used for modelling analyses and reporting including evaluation of interaction terms as part of the model building process and addition to the main text of clarification about model diagnostics processes. These are explained in the methods section (lines 216-226), which now reads –

“Interaction terms evaluated included a year and ownership interaction for the model of brucellosis test performance and a year and month interaction for the model of proportion testing brucellosis positive. A random effect for the health facility ID was included in both models to account for repeat observations at each facility. For the model with proportion of all brucellosis tested patients who were test positive in each month as outcome, an observation level random effect (unique for each combination of facility ID and month of data collection) was also included to account for over-dispersion. For each outcome modeled, initial maximal models including interaction terms were fitted and model simplification performed using likelihood ratio tests (LRT), with a significant p value of ≤ 0.05. Residual diagnostics for final models were performed using the ‘DHARMa’ package in R”

A – Given the inclusion of interactions in the models and update to the results as a consequence, figure 3 has also been included in the results to diagrammatically show the interaction reported for the predicted probabilities of brucellosis testing per month by year and facility ownership type.

The reviewers comment “rewrite the result section by sorting the table with main data” was not entirely clearly to us. In response to other comments though we have added Table 1 which now includes a summary of the district location and facility type of all health facilities (n=86) approached for participation in this study and the subset (n=24) that contributed data and update Table 2 which now includes the characteristics of monthly level data from 24 health facilities over all months of observation (n = 1362 months of observation). We hope that these updates address the reviewers’ points in full.

Reviewer #2:

R2 - This study was to reveal the relationship between health facility characteristics and the probability that brucellosis testing conducted in a given month, and the proportion of tested individuals who had positive results. While the biggest limitation was the low proportion of facilities participating in the study, and the areas where study conducted were only small part of the country, which resulted in the data unrepresentative.

A – We thank the reviewer for this observation. In response we have clarified this limitation at the end of the discussion section (lines 435- 442) and have made the limitations more clear.

“This study had a number of limitations; first, the low proportion of facilities that took part in the study, with lower proportions of health centers and facilities from Arusha Urban and Arusha DC districts participating. Approximately 70% of the 86 facilities that were identified to have brucellosis testing capacity either declined to take part in the study or were not reachable. This limits the inference that can be made from study findings. Also, we were unable to extract the data on the specific tests used to generate all the monthly data for the entire period of the study. As such, we could only estimate the three-month period of study conduct as the duration over which tests found in stock had been used.”

A - A statement has also been included in the results we have added a new summary table (Table 1) clearly demonstrating these characteristics of the facilities involved (and not included) in the study. Lines 235 – 240 now read:

“The characteristics of the health facilities approached for the study and of those that contributed data are shown in Table 1. Additional details for all facilities contributing data are also given in SI Table 1. Lower levels of participation in the study are observed for facilities in Arusha DC and Arusha Urban districts as compared to other districts. Participation was also less likely for health centers as compared to hospitals.”

R2 - However, this study could inform public health authorities to improve surveillance and management of brucellosis and other priority zoonoses in health facilities. So it has some meaning. I suggest it to be modified into a short paper.

A - We thank the reviewer for this suggestion and have made some edits to remove redundancy without compromising the content of the manuscript. However we have also needed to add some content to respond to other reviewer comments. Overall, we strongly feel, that the current format of the manuscript is the most appropriate to communicate the context and findings of this study.

Reviewer #3:

Manuscript Number: PONE-D-21-21073

Title: Brucellosis testing patterns at health facilities in Arusha region, northern Tanzania

I appreciate the efforts put by the authors to assess the variables associated with probability of testing for brucellosis and proportion of individuals classified as test positive for brucellosis. The authors have tried to gather data through structured survey from various test facilities and factors influencing test positivity and proportion of test done and to assess the areas of focus to achieve improved quality of testing and surveillance data on brucellosis.

R3 - However, I feel more clarity is needed for some observations such as lack of year wise data for facility level observations.

A – We thank the reviewer for the observations and suggestions made. In response to this query, we have revised the presentation of this content so that the facility level information is given in Table 1 and the observation level data in Table 2 to improve clarity

R3 - Moreover, correlations between year and monthly positivity data in regression model need to be based on more complex data analysis taking into account of variable interaction and various confounding factors. The model should take in account of these parameters for more reliable models.

We have made several updates to the modelling process and their description in the manuscript to address these comments. First, we have included evaluation of interaction between the year and facility type as suggested (Lines 216-222 in the methods section);

“Interaction terms evaluated included a year and ownership interaction for the model of brucellosis test performance and a year and month interaction for the model of proportion testing brucellosis positive. A random effect for the health facility ID was included in both models to account for repeat observations at each facility. For the model with proportion of all brucellosis tested patients who were test positive in each month as outcome, an observation level random effect (unique for each combination of facility ID and month of data collection) was also included to account for over-dispersion”

We analysed the variables inclusive in the extracted data that we thought appropriate given the nature of data extracted from these log books and found no confounding variables.

Apart from this major limitation, some of the major points need to be addressed are listed below:

R3 - Overall, grammar mistakes need to be addressed throughout the manuscript. For e.g.,

Line 102: ‘different’ need to be replaced by ‘differentiate’.

A – We thank the reviewer for pointing out this grammatical error which has now been revised to read as (lines 87 – 89) – “Serology tests cannot be used to confirm a brucellosis diagnosis using a single acute-stage sample nor can they differentiate the species of Brucella in positive samples…”

Line 105: The sentence is grammatically incorrect, need to be reframed.

A – We thank the reviewer for pointing out this sentence structure. The statement has now been broken into two shorter sentences that now read (lines 92 – 96);

“A number of research studies conducted in regions where higher prevalence of brucellosis is suspected have successfully applied the CDC and WHO recommended diagnostic protocols [5,15-18]. To date however, none of these protocols have been applied at scale for the diagnosis of human brucellosis in Tanzanian health facilities [10,19,20]”

Line 114: Grammatical error has to be replaced by ‘have’

We have revised the statement in question to now read (Lines 104-106)

“…None of these rapid tests are recommended for brucellosis diagnosis in any national or international guidelines and several studies to date have indicated poor performance…”

Line 132: typographical error for ‘where’ instead of ‘were’

A - This is correct as is … “logbooks only captured data about brucellosis testing where testing was done”

Major comment:

The model should incorporate confounding factors and interaction between variables.

A - We thank the reviewer for drawing our attention to this detail in the analysis. We have not included additional variables (potential confounders) as the variable set evaluated includes the key variables that we felt legitimate to consider as either likely explanatory variables or confounders and are not entirely clear on the variables referred to or implied by the reviewer for this point. We have though re-run all models to include evaluation of interactions and revised the methods and results to reflect this. Lines 216 – 222 in the methods section now read;

“Interaction terms evaluated included a year and ownership interaction for the model of brucellosis test performance and a year and month interaction for the model of proportion testing brucellosis positive. A random effect for the health facility ID was included in both models to account for repeat observations at each facility. For the model with proportion of all brucellosis tested patients who were test positive in each month as outcome, an observation level random effect (unique for each combination of facility ID and month of data collection) was also included to account for over-dispersion.”

And lines 267 – 276 in the result sections also represent the outcome of this new analysis;

“The final generalized linear mixed model with the outcome of any brucellosis testing (or not) for each month included a significant interaction between facility ownership and year (Table 3, SI Figure S1). The coefficients for the year variable main effect indicate increased probability of testing over time with a clear increase from 2012 to 2015 and then more stable probability in following years. The influence of this year effect on test probability was variable between private and government owned facilities however, with more pronounced increases (relative to 2012) in test probability for private compared to government owned facilities and variable effects over different years. The predicted probability of brucellosis testing by year and facility ownership is illustrated in Figure 3. The performance of testing or not in each month of the survey period is also shown in Figure 4.”

The reliability of model needs to be measured by suitable tests.

A – We thank the reviewer for this observation and have accordingly revised the language at the end of the methods section to reflect the model diagnostics performed. Lines 225 – 226 now read;

“Residual diagnostics for final models were performed using the ‘DHARMa’ package in R ”

________________________________________

Submitted filename: PLOS reviewers response letter_20Dec21.docx

Click here for additional data file.

7 Mar 2022

Brucellosis testing patterns at health facilities in Arusha region, northern Tanzania

PONE-D-21-21073R1

Dear Dr. Lukambagire,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Aleksandra Barac

Academic Editor

PLOS ONE

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

Reviewer #3: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: The authors have fully recognized the limitations of this study and made appropriate explanations and modifications. Due to the insufficiency of medical resources and importance of brucellosis in this region, the significance of this manuscript is obvious, then I suggest this paper be accepted.

Reviewer #3: Thank you for incorporating the suggested changes. The manuscript seems to be much improved in present form.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

Reviewer #3: Yes: Pankaj Dhaka

14 Mar 2022

PONE-D-21-21073R1

Brucellosis testing patterns at health facilities in Arusha region, northern Tanzania

Dear Dr. Lukambagire:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Aleksandra Barac

Academic Editor

PLOS ONE