Overall, 300 men and women enrolled in the study. We targeted 422 potential participants for interview, 21 were ineligible, 2 refused, and we were unable to locate 99 for a response rate among those eligible to participate of 75%. Temporary outmigration for school or employment accounted for most (62%) of those we were unable to locate. A majority of the participants were women (63%), were currently married (73%), and did not work outside the home in the last 30 days (80%) (Table 1). Participants ranged in age from 18 to 96 (median age: 40) and lived in households with, on average, 5 people. One-third of participants were BIGPIC Family microfinance group members. Members were more likely to be women, older, and currently married. There were no significant differences by microfinance membership for educational attainment, formal employment, household assets, and household size.
Table 1. Characteristics of the study population of 300 residents of two communities in rural western Kenya, 2018-2019
|
|
Microfinance group membership*
|
|
|
Total
N=300
|
Yes
N=100
|
No
N=200
|
P**
|
Socio-demographic characteristics
|
|
N (%)
|
N (%)
|
N (%)
|
|
Gender
|
|
|
|
<0.0001
|
Male
|
112 (37.3)
|
19 (19.0)
|
93 (46.5)
|
|
Female
|
188 (62.7)
|
81 (81.0)
|
107 (46.5)
|
|
Age
|
|
|
|
0.0006
|
<20
|
24 (8.0)
|
1 (1.0)
|
23 (11.5)
|
|
20-29
|
51 (17.0)
|
11 (11.0)
|
40 (20.0)
|
|
30-39
|
68 (22.7)
|
32 (32.0)
|
36 (18.0)
|
|
40-49
|
76 (25.3)
|
30 (30.0)
|
46 (23.0)
|
|
50+
|
81 (27.0)
|
26 (26.0)
|
55 (27.5)
|
|
Marital status
|
|
|
|
<0.0001
|
Never married
|
45 (15.1)
|
1 (1.0)
|
44 (22.1)
|
|
Currently married
|
219 (73.2)
|
84 (84.0)
|
135 (67.8)
|
|
Divorced/separated
|
35 (11.7)
|
15 (15.0)
|
20 (10.1)
|
|
Missing
|
1
|
0
|
1
|
|
Education
|
|
|
|
0.2
|
None/ Some primary
|
92 (30.7)
|
31 (31.0)
|
61 (30.5)
|
|
Primary
|
116 (38.7)
|
45 (45.0)
|
71 (35.5)
|
|
Secondary
|
69 (23.0)
|
20 (20.0)
|
49 (24.5)
|
|
Post-secondary
|
23 (7.7)
|
4 (4.0)
|
19 (9.5)
|
|
Work outside home (last 30 days)
|
|
|
|
0.4
|
Yes
|
59 (19.7)
|
17 (17.0)
|
42 (21.1)
|
|
No
|
240 (80.3)
|
83 (83.0)
|
157 (78.9)
|
|
Missing
|
1
|
0
|
1
|
|
Household asset quartile***
|
|
|
|
0.7
|
Q1
|
70 (25.6)
|
21 (23.6)
|
49 (26.5)
|
|
Q2
|
70 (25.6)
|
27 (30.3)
|
43 (23.2)
|
|
Q3
|
61 (22.3)
|
19 (21.4)
|
32 (22.7)
|
|
Q4
|
73 (26.6)
|
22 (24.7)
|
51 (27.6)
|
|
Missing
|
26
|
11
|
15
|
|
Current NHIF coverage
|
|
|
|
0.2
|
Yes
|
33 (11.0)
|
14 (14.1)
|
19 (9.5)
|
|
No
|
266 (89.0)
|
85 (85.9)
|
181 (90.5)
|
|
Missing
|
1
|
1
|
0
|
|
|
Mean (SD)
|
Mean (SD)
|
Mean (SD)
|
|
Household size
|
5.0 (2.27)
|
5.2 (2.37)
|
4.8 (2.21)
|
0.2
|
Health screening characteristics
|
HIV screening
|
|
|
|
<0.0001
|
Yes
|
252 (84.0)
|
96 (96.0)
|
156 (78.0)
|
|
No
|
48 (16.0)
|
4 (4.0)
|
44 (22.0)
|
|
Diabetes screening
|
|
|
|
<0.0001
|
Yes
|
124 (41.3)
|
77 (77.0)
|
47 (23.5)
|
|
No
|
176 (58.7)
|
23 (23.0)
|
153 (76.5)
|
|
Hypertension screening
|
|
|
|
<0.0001
|
Yes
|
191 (63.7)
|
96 (96.0)
|
95 (47.5)
|
|
No
|
109 (36.3)
|
4 (4.0)
|
105 (52.5)
|
|
Tuberculosis screening
|
|
|
|
0.0001
|
Yes
|
31 (10.3)
|
20 (20.0)
|
11 (5.5)
|
|
No
|
269 (89.7)
|
80 (80.0)
|
189 (94.5)
|
|
Cervical cancer screening
|
|
|
|
0.005
|
Yes
|
36 (19.2)
|
23 (28.4)
|
13 (12.2)
|
|
No
|
152 (80.9)
|
58 (71.6)
|
94 (87.9)
|
|
Disease management outcomes, among those who report HIV, diabetes, or hypertension diagnoses (n=32)
|
Medical visit in last 6 months
|
|
|
|
0.1
|
Yes
|
23 (71.9)
|
10 (90.9)
|
13 (61.9)
|
|
No
|
9 (28.1)
|
1 (9.1)
|
8 (38.1)
|
|
Currently taking medication
|
|
|
|
0.1
|
Yes
|
17 (53.1)
|
8 (72.7)
|
9 (42.9)
|
|
No
|
15 (46.9)
|
3 (27.3)
|
12 (57.1)
|
|
*Member of BIGPIC Family microfinance group for at least 6 months prior to interview
**p-value reported for chi-square test for categorical variables and t-test for continuous variables. P-values calculated among observations with non-missing values.
***Measured by adding up the self-reported value (at time of purchase) of 20 key items in participant’s household
Overall, 11% of all participants had active NHIF health insurance coverage. Health screening rates varied by condition: HIV (84%), diabetes (41%), hypertension (64%), tuberculosis (10%), and cervical cancer (19% - calculated among women only). Among those who reported a diagnosis with HIV, diabetes, or hypertension (n=32), nearly three-quarters (72%) reported a medical visit in the last six months. Just over half reported being currently on medication for their condition (53%).
Health insurance coverage among microfinance group members was 14% compared to 10% in non-members; however, we did not observe a statistically significant association [aPR (95% CI): 1.26 (0.64, 2.27)] (Table 2). There were strong associations between microfinance group membership and health screening. Microfinance members were over three times as likely to report diabetes screening [aPR (95% CI): 3.46 (2.60, 4.60)], about twice as likely to report hypertension screening [aPR (95% CI): 1.96 (1.56, 2.46)], over three times as likely to report tuberculosis screening [aPR (95% CI): 3.31 (1.56, 7.03)]. Among women, microfinance group members were over twice as likely to report cervical cancer screening [aPR (95% CI): 2.43 (1.21, 4.86)]. For each of these outcomes the unadjusted results were similar in magnitude to the results adjusted for age, gender, and marital status. HIV screening was also higher among microfinance members (96%) compared to non-members (78%); however, we did not observe statistically significant associations between microfinance membership and HIV screening [aPR (95% CI): 1.11 (0.64, 2.47)].
Table 2. Relationship between microfinance group membership, health screening, and disease management, among 300 residents of two communities in rural western Kenya, 2018-2019
|
Unadjusted
|
Adjusted*
|
|
PR
(95% CI)
|
p
|
aPR
(95% CI)
|
p
|
Current NHIF coverage
|
1.49
|
(0.78, 2.84)
|
0.2
|
1.26
|
(0.64, 2.47)
|
0.5
|
Health screening outcomes
|
HIV screening
|
1.23
|
(1.13, 1.34)
|
<0.0001
|
1.11
|
(0.96, 1.29)
|
0.2
|
Diabetes screening
|
3.28
|
(2.50, 4.30)
|
<0.0001
|
3.46
|
(2.60, 4.60)
|
<0.0001
|
Hypertension screening
|
2.02
|
(1.74, 2.35)
|
<0.0001
|
1.96
|
(1.56, 2.46)
|
<0.0001
|
Tuberculosis screening
|
3.64
|
(1.81, 7.29)
|
0.0003
|
3.31
|
(1.56, 7.03)
|
0.002
|
Cervical cancer screening
|
2.42
|
(1.27, 4.64)
|
0.008
|
2.43
|
(1.21, 4.86)
|
0.01
|
Disease management outcomes, among those diagnosed with HIV, diabetes, or hypertension (n=32)
|
Medical visit in last 6 months
|
1.47
|
(1.00, 2.16)
|
0.05
|
1.20
|
(0.68, 2.10)
|
0.5
|
Currently taking medication
|
1.70
|
(0.92, 3.13)
|
0.1
|
1.30
|
(0.55, 3.08)
|
0.5
|
*Adjusted for age (categorized at above/below age 40 years), marital status (categorized as currently married vs not), and gender. As cervical cancer screening rates were only calculated among female participants, the adjusted results for this outcome were not adjusted for gender.
Among those who reported diagnoses with HIV, diabetes, or hypertension (n=32), the associations between microfinance membership and disease management outcomes were not statistically significant (Table 2). The prevalence ratios for the relationship between microfinance group membership and both disease management outcomes were above the null, but small in magnitude [Medical visits within the last 6 months: aPR (95% CI): 1.20 (0.68, 2.10); Reporting current medication for their health condition: aPR (95% CI): 1.30 (0.55, 3.08)]. Due to the small sample size of this sub-group, our ability to precisely measure these associations was limited.
The relationship between BIGPIC microfinance membership and the tuberculosis screening outcome was stronger in men compared to women (Table 3, Wald p-value=0.05). We observed no statistical difference between men and women for the relationships between microfinance and the other health screening and management outcomes we assessed, though small sample sizes in the gender-stratified cells limited our ability to estimate our results with precision. Similarly, the relationship between BIGPIC microfinance membership and the HIV screening outcome was stronger in households with lower assets compared to higher assets (Wald p-value=0.05). No statistical differences between household asset status levels were observed for the relationship between microfinance and other health screening and management outcomes.
Table 3. Relationship between microfinance group membership, health screening, and disease management, stratified by gender and socioeconomic status among 300 residents of two communities in rural western Kenya, 2018-2019
|
Women (n=188)*
|
Men (n=112)*
|
Low wealth**
|
High wealth**
|
|
PR (95% CI)
|
p
|
PR (95% CI)
|
p
|
PR (95% CI)
|
p
|
PR (95% CI)
|
p
|
Current NHIF coverage
|
1.61
|
(0.73, 3.53)
|
0.2
|
1.09
|
(0.26, 4.64)
|
0.9
|
1.60
|
(0.51, 4.97)
|
0.4
|
1.16
|
(0.42, 3.18)
|
0.8
|
Health screening outcomes
|
HIV screening
|
1.21
|
(1.08, 1.35)
|
0.007
|
1.26
|
(1.10, 1.44)
|
<0.0001
|
1.32
|
(1.16, 1.51)
|
<0.0001
|
1.11
|
(0.97, 1.25)
|
0.1
|
Diabetes screening
|
3.12
|
(2.16, 4.50)
|
<0.0001
|
4.00
|
(2.74, 5.86)
|
<0.0001
|
3.22
|
(2.17, 4.79)
|
<0.0001
|
3.06
|
(2.06, 4.54)
|
<0.0001
|
Hypertension screening
|
1.67
|
(1.40, 1.98)
|
<0.0001
|
2.67
|
(2.02, 3.52)
|
<0.0001
|
2.10
|
(1.68, 2.62)
|
<0.0001
|
1.88
|
(1.52, 2.33)
|
<0.0001
|
Tuberculosis screening
|
2.20
|
(1.02, 4.78)
|
0.02
|
12.23
|
(2.56, 58.46)
|
0.002
|
3.19
|
(1.24, 8.26)
|
0.02
|
3.63
|
(1.26, 10.43)
|
0.02
|
Cervical cancer screening
|
-
|
-
|
-
|
-
|
-
|
-
|
2.38
|
(0.93, 6.10)
|
0.07
|
2.35
|
(1.04, 5.30)
|
0.04
|
Disease management outcomes, among those diagnosed with HIV, diabetes, or hypertension (n=32)
|
Medical visit in last 6 months
|
1.67
|
(0.98, 2.82)
|
0.06
|
1.20
|
(0.84, 1.72)
|
0.3
|
1.37
|
(0.82, 2.31)
|
0.2
|
1.40
|
(0.57, 3.42)
|
0.6
|
Currently taking medication
|
1.94
|
(0.95, 3.96)
|
0.07
|
1.00
|
(0.20, 4.95)
|
1.0
|
1.65
|
(0.77, 3.53)
|
0.2
|
1.17
|
(0.23, 5.95)
|
0.9
|
*Wald p-values for interaction terms between gender and microfinance group membership were <0.05 for the tuberculosis screening outcome only
**Wald p-values for interaction terms between gender and household wealth asset (dichotomized at median) were <0.05 for the HIV screening outcome only
Our sensitivity analysis operationalizing microfinance exposure as length of membership instead of as member vs. non-member showed point estimates in the same direction as observed in the primary analyses, though they were not statistically significant (Table 4). Among current microfinance members (n=96), those with longer memberships tended to have more health insurance coverage, higher rates of health screening, and better disease management outcomes. The point estimates for each of these outcomes were above the null (with the exception of the estimate for HIV screening and medical visit within the last 6 months), though these estimates were calculated imprecisely with wide confidence intervals often spanning the null.
Table 4. Association between duration of microfinance group membership* and key health screening and disease management outcomes, among current microfinance group members (n=96)
|
PR
(95% CI)
|
p
|
Current NHIF coverage
|
1.38
|
(0.40, 4.75)
|
0.6
|
Health screening outcomes
|
HIV screening
|
0.94
|
(0.88, 1.00)
|
0.05
|
Diabetes screening
|
1.42
|
(1.01, 1.98)
|
0.04
|
Hypertension screening
|
1.10
|
(0.95, 1.28)
|
0.2
|
Tuberculosis screening
|
1.09
|
(0.42, 2.82)
|
0.9
|
Cervical cancer screening
|
1.28
|
(0.57, 2.90)
|
0.5
|
Disease management outcomes, among those diagnosed with HIV, diabetes, or hypertension
|
Medical visit in last 6 months
|
0.83
|
(0.58, 1.19)
|
0.3
|
Currently taking medication
|
2.50
|
(0.85, 7.31)
|
0.1
|
|
|
|
|
|
*Microfinance group membership duration cutpoint at above/below 12 months