8 April 2011

'MOTHER' study was really looking at babies, not mothers.

“Treatment for Pregnant Opioid Abuse Dependent Women”
Karol Kaltenbach, PhD
Clinical Associate Professor, Thomas Jefferson University, Jefferson Medical College

“Live at St Lukes”. Psychiatry Grand Rounds Wednesday 23rd March 2011.

This presentation was a well constructed description of the MOTHER study in a context somewhat removed from the science in my view.

Dr Kaltenbach began by saying that "Families I speak to are all real worried about babies withdrawing from drugs when they are born to mothers on methadone". She said that this was the main purpose of their study: to examine paediatric outcomes from pregnancies randomised to methadone or buprenorphine. We were reminded how little formal research has been done on opiate addicted mothers and their offspring even though tens of thousands of cases have been treated in the case of methadone which is considered the ‘gold standard’. While technically neither drug is approved in pregnancy, methadone treatment is considered a priority option for opioid dependent pregnant women in most health systems (and waiting lists are often dispensed with). Further, and reassuringly, to date several hundred babies have been reported born to mothers on buprenorphine. So the best ‘research’ has actually been field experience with this matter as well as scores of observational reports from Finnegan, Kandell and others. There have been no consistent problems reported from either group apart from the occurrence of neonatal abstinence syndrome (NAS). It is my belief that such concerns over NAS can be addressed by reassurance that existing treatments are satisfactory but may delay baby’s discharge by a few days in some cases.

To her credit Dr Kaltenbach stressed at the outset that from their findings and others’, NAS is no more or less likely with either drug, about half the babies exhibiting symptoms in both groups. We were then given an overview of the trial, initial 1074 screened candidates from seven sites (US, Canada, Austria) , 175 randomised to methadone (89) or buprenorphine (86) on double blind, double dummy bases and with 58 buprenorphine and 73 methadone subjects completing the study. We were briefly shown data supporting no substantive differences between the groups in the usual relevant demographic details including drug use (those dependent on sedatives or alcohol had been excluded at initial assessment). Then we saw numerous tables of primary outcomes and secondary outcomes of the study (all fetal results, including mostly non-significant differences - eg. NAS score, birth weight, head circumference, gestational age, etc). It was emphasised by the speaker that three fetal characteristics were significantly better in the buprenorphine group. These were (1) total amount of morphine used (2) infant’s hospital stay and (3) duration of NAS (I noted that these are all derivatives of the same thing, viz: the severity of NAS as perceived by the treating doctors). My feeling is that it is an overstatement to say that these are three (separate) beneficial findings.

To my mind this trial, for the first time, proved that methadone is the drug of choice for opioid dependency in pregnancy … it also showed that the 58 mothers who finally had a baby on buprenorphine (almost 1000 original applicants didn't) had babies with slightly less protracted and/or severe withdrawal syndromes. And this is no surprise since buprenorphine is the weaker agonist drug and thus mothers with more severe dependencies may be more likely to have dropped out. While the authors provided some comparative evidence against this, on the other hand it would appear to be supported by their finding that ten times as many of the buprenorphine drop-outs stated that they were 'dissatisfied with the medication' (20 versus 2 in the methadone drop-outs).

Dr Kaltenbach did emphasise the health care budget and the issue of hospital stays for treatment of NAS. More important than concerns over NAS in my experience is that most expectant parents want to know: 'will our baby grow up healthy?' NAS can be managed relatively simply and its occurrence is not known to be associated with any negative long term outcomes (Finnegan, personal communication). Most important are the proven negative outcomes known for pregnancies of opioid dependent women who are denied adequate treatment. The costs of such complications to the health system, insurance companies, managed care organisations and society generally are substantial and would be vastly more than even the most comprehensive opioid dependency treatment program.

Despite the modest benefits to the offspring and disastrous consequences to the retention rates, we were told that the researchers believed that their data supported a conclusion that buprenorphine is now a ‘first line option’ in pregnancy. Their institution’s press office went further, stating that that buprenorphine is ‘superior to methadone’ in pregnancy, something which I believe is patently untrue and highly misleading.

If indeed this trial were intended to fairly determine outcomes of the respective drugs then the analysis should have included the patients who ‘failed to comply with the protocols’ and were discharged from treatment. In my clinical experience involuntary 'discharges' are exceedingly rare and in most cases would be involuntary ‘transfers’ so treatment would not be curtailed under any circumstances against the wishes of the patient.

My impression from this presentation was that one of the main goals of this study was to support buprenorphine treatment in the community in America where methadone is often not available. Indeed the very title MOTHER would seem to indicate that fetal outcomes were secondary rather than primary. This is intriguing since buprenorphine is already being used widely in pregnancy with excellent results in those women who can tolerate it. In fact the main conclusion from this study is that both drugs are needed for such women since neither has sufficiently high retention rates. As Dr Kaltenbach implied early in her talk, such research during pregnancy has serious ethical limitations (and my feeling is that without a safety net this trial was in the same category).

A doctor in England tried to do a trial of this nature in non-pregnant patients and over a two year period in their busy treatment clinic she reported that she could not recruit one single patient to randomise (Pinto). It is of concern that in the ‘MOTHER’ study that subjects stood to earn up to $5800 in incentives for compliance and so this may explain why they were able to recruit the women, many of whom had been in stable methadone treatment at the time of joining the study (although this detail was not given in the paper or presentation but in a parallel publication by the group). Thus approximately half such stable patients had their prescribed medication altered during the trial. Such an incentive for a pregnant woman in stable treatment seems to me to be unfair and unbalanced. That such women consented to altered treatment is little justification or consolation for the excess number who dropped out of treatment in my view. Dr Kaltenbach implied that the incentives were used as a ‘normal’ part of high quality treatment, ensuring that there would be no criticism that the women were denied state-of-the-art medical attention. I have never heard of incentives of this degree being used or recommended as “contingency” management (the ‘rewards’ for ‘good behaviour’ are usually modest compared to these sums and sometimes the equivalent of bubble gum or lottery tickets). It appears inescapable, however, that this trial could not have gone ahead without some such strong incentives.

Another detail was that daily attendance, including weekends is an onerous task, especially during pregnancy and is hardly consistent with high quality treatment claimed by the researchers. Yet this is just one more deficiency in a very troublesome trial protocol.

At her talk Dr Kaltenbach seemed uncertain of the Canadian site situation (which ended up providing no subjects at all for the trial, rather like Dr Pinto in England). Initially she said that the patients had the opportunity to get pharmacy pick-ups and daily attendance was no longer acceptable to them. Later in response to my comments she said that there were problems getting the Subutex, Suboxone and placebos to the hospital in Canada.

Dr Kaltenbach stated that the researchers originally wanted to recruit 350 candidates but since the numbers were only about half that they changed their alpha significance level to 0.03 … something I confess I find bizarre and unconvincing. In my view a difference in outcomes is either significant or it isn’t using traditional p values for group comparisons. Even just looking at the raw figures on drop-outs I believe most concerned parties would be able to decide which group they would liked their family member or patient to be in.

Dr Kaltenbach described the excess drop-outs in the buprenorphine group as ‘pretty dramatic’. Yet there is a claim in the paper that this difference is not significantly different! We then had some speculation about the reasons behind the poor buprenorphine results, including inadequate induction doses or inadequate withdrawals before starting (and I wonder if it is ethical to induce withdrawals in pregnant women at all). The reader should note that this trial used pure buprenorphine (Subutex) which is the agent used in nearly all the quality research while the combination drug (Suboxone) has still not been subjected to much if any comparative research to my reading.

While Dr Kaltenbach said that there was no reason to do so, it seemed traditional nowadays to put up a disclosure notice. She said that there were no declarations of interest as she did no consultancies for drug companies.


Comments by Andrew Byrne ..