Can this package handle several response/treatment series with several predictors/control series?

[Florent Bedecarrats]

Dear colleagues,

Congratulations for this amazing peace of work! If I read well, the package documentation and the paper in AOAS only mention the possibility to test for the impact on one response time series (ie. one treated unit), using several predictor time series (untreated/control units). However, I’d like to know if it is possible to use it to assess the impact of one same treatment on several treated units, provided they began their treatment at different points in time.

If it helps understanding the question, I’ll provide some more details:

I want to evaluate a social policy in a sub-Saharan country. It is an obstetric care financing system offered by the Ministry of Health with the support of an International development organization I work for. In this country, all patients have to pay for every cares their receiving in health centers. The scheme consists in a voluntary insurance offered to pregnant women. Against a fixed premium (~20 USD), women are entitled, at no additional charge, to an obstetric package consisting of 4 pre-natal consultations, prophylactic treatments (blood and urine tests, ultrasound…), care of normal delivery or any complication (cesarean, ambulance, resuscitation unit…) and post-natal care. Tested since 2002 in the capital city, this scheme is progressively being extended to other regions since 2007. Currently 144 affiliated health care centers (out of the 500 in the country) are offering this insurance, and more than 90% of pregnant women attending these facilities chose to use it.

We have collected the monthly activity statistics of all the 500 health care centers in the country between January2009 and December 2013. During this period, about 60 health care centers have started to offer the obstetric insurance scheme, affiliating at different dates. I would like to compare the total number of birth assisted every month in the affiliated and non-affiliated health care centers to determine if the introduction of the insurance scheme induced an increase in the number of institutional deliveries (target 5A of the Millennium Development Goals).

Sorry for this long development, I hope it helps understanding the purpose of my question.

Best regards,

Florent