New SDP Data Available Spring 2021

New Data Data Discovery

Get details on new variables related to labor & delivery services, antenatal care, vaccinations, facility shipment schedules, and more!

Matt Gunther (IPUMS PMA Senior Data Analyst)

We’re excited to announce the release of several new Service Delivery Point samples this month over at! As always, you’ll find the new data harmonized with older samples wherever the new surveys repeat questions you’ve seen before. The new samples also contain a big batch of new variables derived from questions that were posed for the very first time in PMA surveys, so we’d like to introduce a few of the highlights here.

The new samples included in this release represent data collected from:

Labor and Delivery

We’ve added a variable group in the Other Health Services topic offering more than 60 new variables related to Labor and Delivery. Many of these are currently available only for the Ethiopia 2019 sample, which piloted the new questions.

For example, you’ll find new variables about delivery personnel, including those showing whether a facility has a skilled birth attendant or a provider able to perform C-section delivery available 24 hours per day. Other variables describe the infrastructure available for labor and delivery, including the number of delivery rooms and beds, labor rooms, maternity waiting rooms, and newborn resuscitation tables. You’ll also find a number of variables describing the environment inside of a facility’s delivery rooms, including whether they are private, heated, and whether they have several specific delivery guidelines and protocols posted in the room.

Several labor and delivery statistics are also provided for the month preceding the interview. These include the total number of facility deliveries, cesarian deliveries, stillbirths (both fresh and macerated), and neonatal deaths (reported separately for those occurring within 24 hours and one week). Other variables report whether certain services were provided any time within 3 months preceding the interview, including:

Lastly, a number of variables indicate whether a particular service is typically provided at a facility. These include,

Antenatal Care

As with Labor and Delivery, a new Antenatal Care group contains a number of variables that are currently available only for the Ethiopia 2019 sample, such as:

Beyond Ethiopia 2019, several of the other new samples included questions related to topics that are normally discussed with patients during an antenatal visit:

In earlier survey rounds, PMA questionnaires included questions on whether these topics were discussed with a new mother after birth. The new samples differentiate between whether these topics were covered before the mother left the facility after delivery (e.g DISPPSPACE) or at a postnatal care visit later on (e.g. DISPNCSPACE).


The Ethiopia 2019 sample also includes some of the first PMA variables related to vaccination. You’ll find indicators for whether a facility typically provides immunization services, whether those services were provided on the day of the interview, and whether a woman visiting the facility for her child’s immunization would be offered family planning services or counseling during the visit.

The availability of the following vaccines are also provided:

Facility Stockouts & Expected Shipments

PMA has included variables related to contraceptive stockouts for many samples dating back to 2014. Four of the new samples dig deeper into the reasons why facilities experience stockouts, and also report the expected delivery time for methods that were out of stock on the day of the interview.

For each of the following methods, the expected delivery time is reported by two variables: a numeric value and a unit (e.g. days, weeks, months) that defines the value.

It’s recommended that users construct their own derived variables for expected delivery times using whichever unit of time they prefer. For example, notice that the values for expected delivery of Depo Provera DEPOVAL are reported either in weeks or months in DEPOUNIT:

dat %>% count(DEPOVAL, DEPOUNIT)
# A tibble: 18 x 3
                      DEPOVAL                    DEPOUNIT     n
                    <int+lbl>                   <int+lbl> <int>
 1  0                          1 [Weeks]                      3
 2  1                          1 [Weeks]                     55
 3  1                          2 [Months]                    29
 4  2                          1 [Weeks]                     16
 5  2                          2 [Months]                     9
 6  3                          1 [Weeks]                      8
 7  3                          2 [Months]                     9
 8  4                          1 [Weeks]                      2
 9  4                          2 [Months]                     1
10  6                          2 [Months]                     1
11  7                          1 [Weeks]                      1
12 12                          2 [Months]                     1
13 14                          1 [Weeks]                      1
14 30                          2 [Months]                     5
15 60                          2 [Months]                     1
16 99 [NIU (not in universe)] 97 [Don't know]                92
17 99 [NIU (not in universe)] 98 [No response or missing]     1
18 99 [NIU (not in universe)] 99 [NIU (not in universe)]   1492

Suppose you wanted to create a derived variable called DEPO_WKS that simply reports the expected delivery time of Depo Provera in weeks. For any value that’s currently reported in months (DEPOUNIT == 2), you might decide to multiply the value in DEPOVAL by 4. Don’t forget to handle non-response values (e.g. 97, 98, 99) separately!

dat %>% 
  mutate(DEPO_WKS = case_when(
      DEPOUNIT == 2 ~ DEPOVAL * 4, 
      DEPOUNIT > 90 ~ NA_real_, 
      T ~ as.double(DEPOVAL) 
    )) %>% 
# A tibble: 15 x 2
   DEPO_WKS     n
      <dbl> <int>
 1        0     3
 2        1    55
 3        2    16
 4        3     8
 5        4    31
 6        7     1
 7        8     9
 8       12     9
 9       14     1
10       16     1
11       24     1
12       48     1
13      120     5
14      240     1
15       NA  1585

For facilities that were currently out of stock of a method that they normally provide, these new samples include variables explaining why the method was out of stock. With Depo Provera, for example, we can now see that a majority of stockouts across samples are caused by shipments that were ordered, but did not arrive:

dat %>% count(OUTWHYDEPO)
# A tibble: 8 x 2
                                         OUTWHYDEPO     n
                                          <int+lbl> <int>
1  1 [Did not place order for shipment]                23
2  2 [Ordered but did not receive shipment]           154
3  3 [Did not order right quantities]                   8
4  4 [Ordered but did not receive right quantities]    17
5  5 [Unexpected increase in consumption]               5
6  9 [Other]                                           27
7 97 [Don't know]                                       1
8 99 [NIU (not in universe)]                         1492

National Health Programs

Lastly, we’ve created a new variable group related to participation in national health programs. While we may see more data in upcoming samples, these variables currently describe facility participation in programs provided by the National Hospital Insurance Fund (NHIF) for Kenya. Specifically, you’ll find an indicator for whether a facility in the Kenya 2019 sample provides family planning methods / services covered by NHIF and, if so, whether it provides each of these:


If you see mistakes or want to suggest changes, please create an issue on the source repository.