Determinants and effects of prenatal care services: evidence from a mobile phone RCT Diether W. Beuermann (BID) Rafael Anta (BID) Walter Curioso (Ministry of Health – Peru) Patricia Garcia (Universidad Peruana Cayetano Heredia) Alessandro Maffioli (BID) Jose Perez Lu (Universidad Peruana Cayetano Heredia) Maria Fernanda Rodrigo (BID)
Motivation Mobile phone services most rapidly adopted technology in developing countries (ITU, 2010; Jensen, 2010) Mobile phones penetration has improved the efficiency of agricultural markets and several welfare related outcomes (Jensen, 2007; Muto and Yamano, 2009; Aker, 2010; Beuermann et al., 2012; Beuermann, 2014) However, this technology has not yet been used extensively for preventive health care in developing countries (Blaya et al., 2010)
Motivation Given high mobile penetration and low marginal costs of SMS messages: provision of preventive health information through this mechanism could be, potentially, a cost-effective intervention Theoretically, reminders can mitigate “attentional failure”: motivating welfare enhancing changes in intertemporal allocations by providing associations between future opportunities and current choices (Reis, 2006; Karlan et al., 2010) Therefore, we empirically test these hypotheses through the implementation of a RCT where pre-natal (via SMS) appointment reminders were sent to pregnant women in Ventanilla, Peru.
Previous evidence (SMS in health services) Applications of SMS technologies in health services has been rigorously tested in: – Reminders for outpatient appointments (Koshy et al., 2008) – Promotion of weight-loss behavior (Joo and Kim, 2007; Patrick et al., 2009) – Smoking cessation (Rodgers et al., 2005) – Adherence to diabetes treatment (Franklin et al., 2006) – Adherence to HIV antiretroviral treatment (Lester et al., 2010; Pop-Eleches et al., 2011) – Health workers’ adherence to malaria treatment (Zurovac et al., 2011) – Preventive practices against dengue (Dammert et al., 2014)
However, no application for pre-natal services has been rigorously tested.
Contribution First RCT that evaluates the effectiveness of SMS based appointment reminders on attendance to pre-natal appointments, behavioral changes during pregnancy and newborns’ health Using the intervention as an instrument for number of pre-natal controls attended, we provide the first experimental evidence on the effects of controls attended on birth weight, gestation and maternal weight gain. [Evans (2004) provided non-experimental evidence for US suggesting weak effects of pre-natal controls on birth outcomes] Baseline data collected during the first pre-natal control allowed us to test for heterogeneous treatment effects by education and distance to the health center
The Intervention Conducted in the district of Ventanilla, Peru Within the 16 health centers in Ventanilla, all paper based medical records for pregnant women were replaced by customized electronic medical records (2010 – 2011) From February 2012 until January 2013, women attending their first pre-natal control within the first 20 weeks of pregnancy were randomized to either receive or not the SMS reminders. Final sample of 1,162 women (576 Treated; 586 Control) Baseline information was collected during the first prenatal control Women were followed until birth and also interviewed face to face 2-3 weeks after birth
The Intervention Appointment reminders were sent every Monday and the day before the appointment In addition, every Wednesday and Saturday, other educational messages recommending healthy food, reminding to take vitamins, and hygienic practices during pregnancy were sent Randomization was conducted at the individual level (but stratified at the health center) Successful random assignment: out of 34 baseline characteristics, only 3 significant at the 10% level or lower Attrition rate of 6%: orthogonal to treatment status and baseline characteristics
Baseline balance Adjusted Difference (3)
Treated (1)
Control (2)
25.75
25.58
0.17 (0.32)
1,162
0.33
0.38
1,162
Complete secondary
0.48
0.44
Incomplete tertiary
0.09
0.09
Complete tertiary
0.09
0.09
-0.05 (0.03) 0.03 (0.03) 0.01 (0.01) 0.00 (0.02)
Civil Status Single
0.17
0.17
1,162
Married
0.11
0.14
Not married but living with partner
0.72
0.69
Pre pregnancy weight (kilograms)
58.32
58.40
Height (meters)
1.53
1.53
-0.00 (0.02) -0.03 (0.03) 0.03 (0.03) -0.07 (0.53) -0.00 (0.00)
Women Socio-Demographic Characteristics Age Education Incomplete secondary or lower
N (4)
1,162 1,162 1,162
1,162 1,162 1,159 1,162
Baseline balance Women physical signals (1st control) Weeks of pregnancy Number of prior births
Treated (1)
Control (2)
12.01
11.73
1.02
0.93
Adjusted Difference (3) 0.32 (0.22) 0.08
N (4) 1,154 1,162
(0.05) Temperature (Celsius)
36.55
36.55
Respiratory frequency
18.98
18.92
Pulse
76.29
75.87
929.65
895.12
Tap water
0.76
0.76
Electricity
0.99
0.97
Internet
0.03
0.03
Refrigerator
0.59
0.63
DVD
0.80
0.77
Computer
0.06
0.06
Household Characteristics Distance to health center (meters)
0.01 (0.02) 0.06 (0.14) 0.51** (0.20) 34.71 (47.76) -0.01 (0.02) 0.01 (0.01) 0.00 (0.01) -0.04 (0.03) 0.03 (0.02) 0.00 (0.01)
1,156 1,156 1,156
1,162 1,161 1,161 1,160 1,160 1,161 1,160
Empirical strategy I  We first assess reduced form effects of treatment on attendance to pre-natal care, birth outcomes and behavioral changes during pregnancy đ?‘Œđ?‘–đ?‘— =âˆ?đ?‘— +đ?›˝đ?‘‡đ?‘–đ?‘— + đ?›žâ€˛đ?‘‹đ?‘–đ?‘— + đ?œ€đ?‘–đ?‘—
 For woman i, attending health center j. Include strata (health center) FEs and individual controls (age, education and distance to health center) to increase precision. Estimated standard errors clustered at the health center level in all regressions  Since behavioral changes were measured through multiple outcomes, we follow Kling et al. (2007) to build categorical summary indexes expressed in standard deviations wrt. the control group.
Overall results
Attend >=6 pre-natal controls
Control Group Mean (1) 0.82
Overall Intervention Effect (2) 0.05** (0.02)
Attend >=9 pre-natal controls
0.33
0.07***
Control Group Mean Adherence to Vitamins
(1)
(2)
0.00
0.05 (0.04)
Exercise
0.00
(0.02) Total pre-natal controls
7.48
0.35***
3.10
0.31**
Recommended Food
0.00
Not Recommended Food
0.00
Observations
586
(0.14) Birthweight (grams)
3,316.29
31.11
-0.00 (0.05)
(0.12) Total pre-natal controls on time
Overall Intervention Effect
0.03 (0.02) -0.04 (0.03) 1,162
(28.07) Weeks of gestation
38.89
0.00 (0.09)
Weight gain during pregnancy (kilograms) Observations
10.30 586
0.12 (0.28) 1,162
 Significant effects on pre-natal care. Non-significant effects on birth outcomes or behavioral changes (although the point estimate is consistent with positive effects)
Heterogeneous effects? Theory predicts that individuals would invest in pre-natal care if the expected benefits outweigh its costs In our setting, pre-natal controls and vitamins are provided for free. So the costs involved are reduced to transportation and opportunity costs of time Benefits of pre-natal care are not observed directly or in the short-term potentially difficult to interiorize when awareness or previous knowledge is absent Dupas (2011) shows that the provision of information influences behavior if it is credible and individuals are capable of processing the new information Therefore, our intervention should be more effective among individuals facing: (1) lower costs of attending pre-natal care; and (2) with relatively higher awareness of the potential benefits of pre-natal care
Heterogeneous effects? Effects by Education
Attend >=6 pre-natal controls Attend >=9 pre-natal controls Total pre-natal controls Total pre-natal controls on time Birthweight (grams) Weeks of gestation Weight gain during pregnancy (kilograms) Observations
Distance to Health Center
Below Secondary
Secondary or Higher
Below 500 meters
Above 500 meters
0.02 (0.04)
0.06*** (0.02)
0.09*** (0.02)
0.03 (0.03)
0.04
0.07**
0.07
0.08**
(0.05)
(0.03)
(0.05)
(0.03)
0.24
0.40***
0.53***
0.32*
(0.21)
(0.11)
(0.16)
(0.18)
0.14
0.40**
0.42*
0.26
(0.30)
(0.14)
(0.20)
(0.21)
-20.15
56.74
87.72*
8.24
(54.02)
(35.13)
(44.07)
(46.17)
-0.11
0.07
0.16
-0.08
(0.16)
(0.10)
(0.19)
(0.13)
-0.37 (0.44) 414
0.41 (0.34) 748
-0.16 (0.54) 405
0.47 (0.37) 757
Heterogeneous effects? Effects by Education and Distance to Health Center Below Secondary
Secondary or Higher
Below 500 meters
Above 500 meters
Below 500 meters
Above 500 meters
Attend >=6 pre-natal controls
0.08** (0.04)
-0.01 (0.06)
0.09** (0.04)
0.06** (0.02)
Attend >=9 pre-natal controls
0.03
0.04
0.07
0.09**
(0.06)
(0.07)
(0.06)
(0.04)
0.40
0.19
0.54**
0.39**
(0.24)
(0.32)
(0.20)
(0.16)
0.08
0.06
0.58*
0.35
(0.29)
(0.41)
(0.28)
(0.21)
-42.42
15.72
188.97***
13.74
(112.66)
(76.44)
(57.82)
(54.69)
0.06
-0.26
0.24
0.04
(0.30)
(0.27)
(0.23)
(0.15)
-1.29* (0.62) 151
0.29 (0.62) 263
0.53 (0.74) 254
0.62 (0.42) 494
Total pre-natal controls Total pre-natal controls on time Birthweight (grams) Weeks of gestation Weight gain during pregnancy (kilograms) Observations
Heterogeneous effects? Effects by Education
Adherence to Vitamins Exercise Recommended Food Not Recommended Food Observations
Distance to Health Center
Below Secondary
Secondary or Higher
Below 500 meters
Above 500 meters
-0.09
0.11**
0.15***
0.00
(0.09)
(0.04)
(0.04)
(0.07)
0.12
-0.06
0.01
-0.02
(0.08)
(0.06)
(0.07)
(0.08)
0.03
0.01
0.01
0.03
(0.02) -0.10 (0.07) 414
(0.02) -0.00 (0.03) 748
(0.05) -0.09* (0.05) 405
(0.03) -0.01 (0.03) 757
Heterogeneous effects? Effects by Education and Distance to Health Center Below Secondary
Adherence to Vitamins Exercise Recommended Food Not Recommended Food Observations
Secondary or Higher
Below 500 meters
Above 500 meters
Below 500 meters
Above 500 meters
-0.14*
-0.05
0.33***
0.02
(0.07)
(0.15)
(0.08)
(0.06)
0.07
0.13
-0.01
-0.08
(0.11)
(0.13)
(0.07)
(0.08)
0.01
0.04
-0.01
0.03
(0.06) -0.18** (0.06) 151
(0.03) -0.03 (0.11) 263
(0.05) -0.03 (0.06) 254
(0.03) 0.01 (0.03) 494
Empirical strategy II  We now look at the effects of pre-natal controls on birth outcomes: đ??śđ?‘–đ?‘— =âˆ?đ?‘— +đ?›˝1 đ?‘‡đ?‘–đ?‘— + đ?›žâ€˛đ?‘‹đ?‘–đ?‘— + đ?œ€đ?‘–đ?‘— đ?‘Œđ?‘–đ?‘— =âˆ?đ?‘— +đ?›˝2 Cđ?‘–đ?‘— + đ?›žâ€˛đ?‘‹đ?‘–đ?‘— + đ?œ€đ?‘–đ?‘—  For woman i, attending health center j. Include strata (health center) FEs and individual controls (age, education and distance to health center) as predetermined variables. Instrument number of pre-natal controls (C) with the treatment indicator (T).
Pre-natal controls and birth outcomes Effects by Education
Distance to Health Center
Overall Effect
Below Secondary
Secondary or Higher
Below 500 meters
Above 500 meters
88.39
-85.12
142.3*
166.0*
25.38
(68.45)
(241.3)
(81.64)
(88.75)
(127.5)
8.784
1.323
12.29
10.92
3.165
0.00966
0.268
0.00319
0.00481
0.0955
Observations
1,162
414
748
405
757
Weeks of gestation
0.00244
-0.504
0.166
0.300
-0.252
(0.244)
(0.914)
(0.235)
(0.352)
(0.489)
F-statistic on excluded instrument
8.632
1.226
12.29
10.63
3.165
P-value
0.0102
0.286
0.00319
0.00527
0.0955
Observations
1,161
413
748
404
757
0.321 (0.749) 10.29 0.00587 1,154
-1.269 (1.257) 1.913 0.187 409
1.010 (0.893) 14.34 0.00179 745
-0.271 (0.861) 16.94 0.000917 401
1.389 (1.423) 3.641 0.0757 753
Birthweight (grams) F-statistic on excluded instrument P-value
Weight gain during pregnancy (kilograms) F-statistic on excluded instrument P-value Observations
Pre-natal controls and birth outcomes Effects by Education and Distance to Health Center Below Secondary
Secondary or Higher
Overall Effect
Below 500 meters
Above 500 meters
Below 500 meters
Above 500 meters
88.39
-106.2
83.47
350.5**
34.80
(68.45)
(257.4)
(336.5)
(154.6)
(124.9)
8.784
2.832
0.350
7.077
5.766
0.00966
0.113
0.563
0.0178
0.0298
Observations
1,162
151
263
254
494
Weeks of gestation
0.00244
0.172
-1.375
0.436
0.0979
(0.244)
(0.730)
(3.084)
(0.425)
(0.340)
F-statistic on excluded instrument
8.632
2.484
0.350
7.077
5.766
P-value
0.0102
0.136
0.563
0.0178
0.0298
Observations
1,161
150
263
254
494
0.321 (0.749) 10.29 0.00587 1,154
-2.162 (1.317) 9.671 0.00717 147
1.482 (4.350) 0.362 0.557 262
0.977 (1.288) 7.077 0.0178 254
1.515 (1.274) 6.795 0.0198 491
Birthweight (grams) F-statistic on excluded instrument P-value
Weight gain during pregnancy (kilograms) F-statistic on excluded instrument P-value Observations
Key messages and policy implications SMS messages were generally effective at increasing pre-natal care attendance However, and in line with theory, these effects were concentrated in groups with higher education (proxy for awareness of future benefits of preventive care) and with easier access to health centers (proxy for costs of preventive care) Further, birth outcomes and behavioral changes were only affected for the high education-easier access groups Policy implication: potential benefits of inexpensive interventions within the health care sector exploiting mobile technologies can only be achieved if the basis of human capital and care access is already in place
Next steps and unanswered questions Our estimates may constitute lower bound effects if treatment spillovers exist we are currently exploiting GIS and social network information within our sample to estimate possible spillover effects We evaluated the SMS component of Wawared. However, the implementation of the electronic medical records per se could have motivated higher pre-natal care attendance as duration and waiting periods for prenatal appointments were reduced from 22 to 10 minutes on average (before and after). Therefore, our estimates would be only the additional effect of SMS messages over the effect of electronic medical records Future research should be directed towards testing these effects separately (maybe within a scale-up of Wawared?)
THANKS!