Changes in energy content of lunchtime purchases from fast food restaurants after introduction of calorie labelling: cross sectional customer surveys

Dear Editor,

We thank Pande and Soumerai for their comments. However, we believe
that they present an overly simplistic critique of the study's findings
and methodology that is not grounded in the reality of public policy
evaluations. First, and most importantly, we did not conclude, as they
suggest, that the intervention was ineffective. We were similarly
disappointed by reports from some press outlets that suggested otherwise.

The 15% of patrons who reported seeing and using calorie information
purchased fewer calories. In 2009, almost 1.3 million New Yorkers reported
eating fast food at least once a week , resulting in approximately 3
million visits weekly (1). This means that very substantial numbers of New
Yorkers will use this information to purchase fewer calories each week.
Furthermore, there were statistically significant declines in the total
number of calories purchased in several major chains within the overall
restaurant sample.

Since mandatory calorie labeling by restaurant chains has been
adopted nationwide in the U.S. and will go into effect next year, and with
this expansion the incentive to industry to change its practices grows, we
concluded that further rounds of assessment are needed to measure the full
impact. It should be noted that the calorie labeling regulation also
serves the simple purpose of assuring a consumer's right to know. In that
regard, calorie posting was considered useful by 84% of New Yorkers(2).
Nevertheless, that information alone will not be sufficient to reverse the
obesity epidemic is not a surprise. Rather, it is a challenge for future
action.

Pande and Soumerai's initial point suggests that evaluation of
calorie posting in New York City could be meaningfully pursued only with
Cochrane-quality randomized trials. While this may represent a scientific
ideal, the restaurant industry, which aggressively opposed the policy
through the courts, would simply not have cooperated with randomization.
Randomized trials have not been feasible for key public health policies
such as those that required seatbelts, raised tobacco taxes or removed
lead from paint. While using other jurisdictions as controls is a
reasonable alternative suggestion, it was not feasible within the narrow
timeframe of a few months between approval and implementation. New York
City rapidly designed and funded this study to ensure a large and
representative sample of affected fast food chains and their customers. We
agree with Pande and Soumerai that the study has important limitations,
which were clearly discussed in the publication. However, the presence of
these limitations does not obviate our findings and, as with all
publications, serves to inform the reader as to how the results should and
should not be interpreted.

Unfortunately, no other jurisdiction, subsequent to the
implementation of this measure, has been able to carry out even a
similarly scaled evaluation. This reflects the typical difficulties
confronting policy evaluation, where timely studies are often the only
option, and while imperfect, still furnish critically useful information.
We welcome other studies with different designs that complement our work,
particularly if baseline measures of sufficiently representative samples
can be taken before a calorie labeling policy goes into effect elsewhere.

In regard to Pande and Soumerai's concern about contamination of the
baseline data, the information previously provided by Subway at baseline
was restricted to a small sticker on the display case and provided calorie
values for only a few of their products. Thus it was not equivalent to
earlier implementation of comprehensive calorie posting. More
importantly, the evaluation of a public policy, which was ultimately the
goal of the study, is to measure what changes occur in society with its
implementation. If one company voluntarily provided some information at
baseline, that was, in fact, the societal condition prior to the policy
and hence does not invalidate our findings.

It seems obvious, but perhaps merits note, that the food industry is
well positioned to assess the impact of calorie labeling upon its sales
through its cash register data. However, with the exception of the
Starbucks chain, which supported a large study whose results showed a 6%
decline in calories purchased without loss of revenue at their
establishments,(3) no other company has chosen to share its sales data.

We welcome discussion and debate on how to assure the evaluation of
public policy innovations, and believe there is an urgent need for funding
pools designed to support the assessment of innovative public health
policies within typical fast moving timeframes. As nations, we need a
bigger commitment to the evaluation of large scale public health measures
with the potential to change population health status.

Lynn Silver MD, MPH, FAAP

Mary T. Bassett, MD, MPH

(1) New York City Community Health 2009, Unpublished data.

(2) Quinnipiac University Polling Institute. New Yorkers big on sugar
drink tax, calorie counts 2011.
www.quinnipiac.edu/x1302.xml?ReleaseID=1425

(3) Bollinger B, Leslie P, Sorenson A. calorie posting in chain
restaurants. The National Bureau of Economic Research, Working paper no.
15648. 2010. Nber.org/papers/w15648