Socio-demographic characteristics of study participants
The study enrolled a total of 386 department heads in 83 health facilities with a response rate of 98%. Of the total facilities enrolled, 76(92%) of them were health centers, followed by primary hospital 5(6%) and referral hospital 2(2%).
The mean age of respondents was 28.2 years with a standard deviation of 5.4 years. Of the total study participants included in the study 238(62%) of them were female participants, 294(82.4%) of them were below the age of 30 years, 225(58.3%) of them were diploma graduates, and 115(29.8%) of them were clinical nurses by profession. While considering the year of experience, 239(63.7%) of the study participants had less than five years of experience, 281(74.3%) % of them were earned five thousand and less Ethiopian birr per month, and 233(60.4%) of them were rural dwellers. Of the total departments, 97(25.1%) of departments heads were from Maternal and Child Health (MCH) departments, followed by Out-patient departments 76(19.7%) (Table 1).
Table 1
Socio-demographic characteristics of study participants in health facilities in the Amhara region, Northwest Ethiopia, 2019
Variables
|
Number
|
Percent (%)
|
Sex (N-= 384)
|
|
|
Male
|
146
|
38
|
Female
|
238
|
62
|
Age in year (N = 357)
|
|
|
<=30
|
294
|
82.4
|
>30
|
63
|
17.6
|
Level of Education (N = 386)
|
|
|
Grade ten completed
|
2
|
0.5
|
Diploma
|
225
|
58.3
|
Bachelor
|
155
|
40.2
|
Master
|
4
|
1.0
|
Profession (N-= 386)
|
|
|
BSC Midwifery
|
18
|
4.7
|
BSC nurse
|
51
|
13.2
|
Health Officers
|
56
|
14.5
|
Clinical Nurse Diploma
|
115
|
29.8
|
Health Information Technician
|
42
|
10.9
|
Laboratory diploma
|
17
|
4.4
|
Midwifery diploma
|
37
|
9.6
|
Pharmacy diploma
|
39
|
10.1
|
Others
|
11
|
2.8
|
Experience in year (N = 375)
|
|
|
<=5
|
239
|
63.7
|
6–10
|
102
|
27.2
|
>= 11
|
34
|
9.1
|
Salary in ETB (N = 378)
|
|
|
<= 5,000
|
281
|
74.3
|
5,001–10,000
|
94
|
24.9
|
>10,000
|
3
|
0.8
|
Work Location (N = 386)
|
|
|
Urban
|
153
|
39.6
|
Rural
|
233
|
60.4
|
Departments by type (N = 386)
|
|
|
Anti-Retroviral Therapy (ART)
|
17
|
4.4
|
Emergency Outpatient department
|
18
|
4.8
|
Facility head office
|
24
|
6.2
|
HMIS Unit
|
41
|
10.6
|
Laboratory
|
17
|
4.4
|
Maternal and Child health (MCH)
|
97
|
25.1
|
Out-patient department (OPD)
|
76
|
19.7
|
Pharmacy
|
42
|
10.9
|
TB clinic
|
16
|
4.1
|
Youth friendly service
|
18
|
5.2
|
Others
|
20
|
3.4
|
Performance Monitoring Team Discussions
Of the total 386 department heads observed, 252(65%) and 244(63%) of them mentioned Performance Monitoring Team (PMT) meetings were conducted regularly and maintained official minutes that contains the discussion points and decisions made during their meeting, respectively. Regarding PMT meeting contents, 219(89%) of minutes showed management of RHIS activities, 183(75%) indicated discussions made on RHIS findings, 163(66%) revealed the decisions made, and 124(51%) showed follow up actions taken place regarding the decisions. However, only 63(25%) showed RHIS related issues or problems referred to a district or regional level for further actions. Of the total 386 departments investigated, 200(52%) calculated target against achievement, and only 130(33.7%) provided feedback to lower-level health workers. As a result, the study revealed that only 46.9% of department heads utilized routine health information for evidence-based decisions (Fig. 1).
Among the total study participants, 245(63.6%) received annual or monthly targets on RHIS performance, however; only 189(48.9%) of them mentioned that they received senior management directives concerning the use of information, 146(37.8%) received a district office report, and 150(38.9%) had documents showing the use of information for advocacy purposes. Besides, 173(44.8%) of them were participated in meetings held at the district level.
Display Of Information
Of the total 386 respondents, only 118(30.6%) and 156(40.4%) of them displayed the catchment area map and summary of demographic information such as population by age group, respectively. However, feedback provided by district office or higher organization was found only in 129(33%) of the departments (Table 2).
Table 2
Type of information displayed in health facilities in the Amhara region, Northwest Ethiopia, 2019
Variables
|
Number
|
Percent (%)
|
Catchment area map (N = 386)
|
|
|
Yes
|
118
|
30.6
|
No
|
268
|
69.4
|
Demographic data (N = 386)
|
|
|
Yes
|
156
|
40.4
|
No
|
230
|
59.6
|
Feedback available (N = 386)
|
|
|
Yes
|
129
|
33.4
|
No
|
257
|
66.6
|
Feedback review strategy (N = 129)
|
|
|
Yes
|
111
|
86.1
|
No
|
18
|
13.9
|
Feedback review personal responsibility (N = 129)
|
|
|
Yes
|
115
|
89.2
|
No
|
14
|
10.8
|
Feedback indicate resource mobilization (N = 129)
|
|
|
Yes
|
87
|
67.4
|
No
|
42
|
32.6
|
Feedback indicate data use for advocacy (N = 129)
|
|
|
Yes
|
83
|
64.3
|
No
|
46
|
35.7
|
Rhis Supervision
Among the total 386 department heads assessed, only 158(41%) of them received RHIS supervision from higher-level institutions. Of which, majority 148(93.7%), 146(92.4%), and 140(88.6%) explained that supervisors use a checklist, check the quality of data, and discussed performance based on target and achievement, respectively; however, only 98(62%) of supervisors sent official reports to supervisee after the supervision had taken place. Among 383 department heads, only 129(33.4%) mentioned that internal supervision was conducted regularly, and 66(51.1%) of them were provided written feedback that addressed the weaknesses and strengths of RHIS performance.
Planning
Among a total of 386 department heads assessed, 66(17.1%) of them had no strategic plan of the facility. However, 242(62.7%) of them had received an annual plan for the 2019 fiscal year, of which 119(49.2%) reflected the use of routine data for problem identification. Of the total 386 departments observed, the practice of data use for the target set during annual planning was observed only in 167(43.3%) of departments (Table 3).
Table 3
Characteristics of the planning process in selected health facilities of the Amhara region, Northwest Ethiopia, 2019
Characteristics
|
Frequency
|
Percent (%)
|
Strategic plan available (N-= 386)
|
|
|
Yes
|
66
|
17.1
|
No
|
320
|
82.9
|
Annual plan available (N = 386)
|
|
|
Yes
|
242
|
62.7
|
No
|
144
|
37.3
|
Plan reflect use of data for problem identification (N = 242)
|
|
|
Yes
|
119
|
49.2
|
No
|
123
|
50.8
|
HMIS data use for target setting (N = 386)
|
|
|
Yes
|
167
|
43.3
|
No
|
123
|
50.8
|
Technical Constraints
The study indicated that, among 384 department heads, 137(32.7%) faced challenges while using HMIS data as detailed below (Figure.2) of which, 110(80.3%) of them provided feedback to the HMIS unit to improve the challenges, and 88(80.6%) of them explained the feedback get addressed. Among a total of 384 department heads examined, 256(66.3%) explained they had the skill to make an evidence-based decision. The study also indicated that 288(74.6%) of study participants need training on the use of data for decision making followed by data collection 152(39.4%), and data analysis 151(39.1%).
Organizational Constraints
From a total of 386 departments assessed, 324(83.9%) conducted regular staff meetings in the last three months of which, 240(74%) conducted senior staff meetings, 310(96%) carried out all-staff meetings, and 295(91%) implemented department-level meetings. Only 43.4% of department heads received feedback from seniors to improve their performance. Among 377 respondents, 173(45.9%) agreed that the facilities arranged an incentive or motivation mechanism to improve the use of information for evidence-based decision making that in turn improved the quality of healthcare.
Behavioural Constraints
Among the total 380 study participants, 211(55.5%) and 119(31.3%) of them agreed that not using collected data discouraged them, and collecting information led them to feel bored, respectively. However, 342(89.8%) of respondents mentioned that collecting information was meaningful for them, 346(91.5%) explained collecting information gave them that data were needed, and 308(81.5%) of them mentioned that collection information was appreciated. Fewer 45(11.8%) of them explained that collecting information gave them forced activities.
Factors Affecting The Level Of Information Use
The study revealed that displaying demographic and performance data, and providing feedback were individual-level predictors whereas maintaining PMT minute, issuing directives by senior management, using HMIS data for target setting, receiving HIS supervision, and location were organizational level predictors. (Table 4).
The study revealed that, department heads who displayed demographic data were 12.42 (AOR = 12.42; 95% CI [5.52, 27.98]) times more likely to use the information for decision making than those who did not display demographic data. Besides, the odds of use of information for decision making was 1.68 (AOR = 1.68; 95% CI [1.33, 2.11]) and 2.29 (AOR = 2.29; 95% CI [1.05, 5.00]) times higher among department heads who displayed performance data for monitoring and provided feedback to the HMIS unit regarding data quality than their counter parts those who did not, respectively (Table 4).
The study indicated that, facilities that maintained PMT minute and received senior management directives had 3.53 (AOR = 3.53; 95% CI [1.61, 7.75]) and 3.56 (AOR = 3.56; 95% CI [1.76, 7.19]) times higher odds of the level of information-use for decisions, respectively. Besides, the level of information-use for decision making was 3.43 (AOR = 3.43; 95% CI [1.66, 7.09]) and 2.84 (AOR = 2.84; 95% CI [1.33, 6.07]) times more likely in health facilities which utilized HMIS data for target setting and received RHIS supervision, respectively. However, the level of information-use in rural facilities was 84% (AOR = 0.16; 95% CI [0.07, 0.39]) less likely than urban facilities (Table 4).
Model Appropriateness
The score of the Log-likelihood ratio for the null model in the GLMM was subtracted from the ordinary logistic regression model and resulted in -19.65784 (degree of freedom = 1). As a result, the test statistics (-2*Log likelihood ratio) was 39.316, which indicated that the between-facility variance was non-zero. Besides, the ANOVA model employed for the null model in the GLMM and ordinary logistic regression resulted in a statistically significant chi-square test (χ2 = 39.316, P = < 0.001), which supported to include additional terms (random effect) in the model. Moreover, the ICC score in the null model was 0.35, which implied that 35% of the chance of having a good level of information use was explained by between-cluster differences (facilities) (Table 5).
Model Comparison
The scores for AIC and BIC decreased from the null to the global model in a generalized linear mixed-effect model. Unlike that, the value for PCV increased in cascading models (Table 5)
Model Fit And Performance Diagnosis
The ratio of residual deviance to the degree of freedom for the global model (model with both individual and organizational level variables) was 0.97, which explained that the model was not over-dispersed.
Multicollinearity Test
Though a multicollinearity test conducted for the final model, none of the variables scores showed above 10 for the test statistic.
Table 4
Factors associated with the level of information-use for decision making in the Amhara region, Northwest Ethiopia, 2019
Variables
|
Null model
|
Model with level-1
variables
|
Model with level-2
variables
|
Model with both level-1 and level-2 variables
|
AOR (95% CI)
|
P-value
|
AOR (95% CI)
|
P-value
|
AOR (95% CI)
|
P-value
|
AOR (95% CI)
|
P-value
|
Intercept
|
0.85 (0.59, 1.24)
|
0.406
|
0.004 (0.001, 0.02)
|
0.001 ***
|
0.25 (0.05, 1.15)
|
0.076
|
0.03 (0.005, 0.21)
|
0.001***
|
Display demographic data
|
-
|
-
|
7.46 (3.52, 15.83)
|
0.001 ***
|
-
|
-
|
12.42 (5.52, 27.98)
|
0.001***
|
Received office report
|
-
|
-
|
5.33 (2.58, 10.99)
|
0.001 ***
|
-
|
-
|
-
|
-
|
Supervisor provide feedback timely
|
-
|
-
|
1.22 (1.03, 1.45)
|
0.05*
|
-
|
-
|
-
|
-
|
Supervise staff
|
-
|
-
|
1.64 (1.11, 2.43)
|
0.05*
|
-
|
-
|
-
|
-
|
Staff display data for monitoring
|
-
|
-
|
1.48 (1.18, 1.85)
|
0.000 ***
|
-
|
-
|
1.68 (1.33, 2.11)
|
0.001***
|
Provide feedback to HMIS unit
|
-
|
-
|
2.93 (1.42, 6.07)
|
0.01**
|
-
|
-
|
2.29 (1.05, 5.00)
|
0.05*
|
Feedback from seniors available
|
-
|
-
|
2.79 (1.29, 6.07)
|
0.01**
|
-
|
-
|
-
|
-
|
PMT minute maintained
|
-
|
-
|
-
|
-
|
2.78 (1.29, 5.98)
|
0.01**
|
3.53 (1.61, 7.75)
|
0.01**
|
Received planned target
|
-
|
-
|
-
|
-
|
2.63 (1.17, 5.92)
|
0.05*
|
-
|
-
|
Senior management issued directives
|
-
|
-
|
-
|
-
|
3.06 (1.58, 5.90)
|
0.001***
|
3.56 (1.76, 7.19)
|
0.001***
|
Have annual plan
|
-
|
-
|
-
|
-
|
2.93 (1.24, 6.92)
|
0.05*
|
-
|
-
|
HMIS data use for target setting
|
-
|
-
|
-
|
-
|
2.20 (1.09, 4.45 )
|
0.05*
|
3.43 (1.66, 7.09)
|
0.001***
|
Received HIS supervision
|
-
|
-
|
-
|
-
|
2.34 (1.13, 4.88)
|
0.05*
|
2.84 (1.33, 6.07)
|
0.01**
|
Location
|
-
|
-
|
-
|
-
|
0.27 (0.12, 0.60)
|
0.01**
|
0.16 (0.07, 0.39)
|
0.001***
|
*** P-value < 0.001, ** P-value < 0.01, *P-value < 0.05, ‘.’ P-value < 0.1 |
Table 5
Summary of information for the generalized linear mixed effect model in the Amhara region, Northwest Ethiopia, 2019
Model Components
|
Null model
|
Model with level-1
variables
|
Model with level 2
variables
|
Model with both level-1 and level-2 variables
|
Variance of random effect (τ2)
|
1.778
|
1.574
|
1.102
|
0.7187
|
LR
|
-247.2
|
-173.7
|
-171.3
|
-142.7
|
PCV
|
Reference
|
11.4%
|
37.9%
|
59.6%
|
ICC
|
0.351
|
0.324
|
0.251
|
0.179
|
AIC
|
498.3
|
365.4
|
360.6
|
305.5
|
BIC
|
506.2
|
401.0
|
396.1
|
344.9
|
LR; log-likelihood ratio, PCV: proportional change in the variance, ICC: intra class correlation, AIC: Akaike information criteria, BIC: Bayesian information criteria.
|