It is notable that we found that those who used marijuana more frequently prior to 2018 reported greater increases in use from 2018 onward. On one hand this is encouraging in that it suggests that lighter and non-users of marijuana were not necessarily encouraged to use as a result of legalization. On the other hand, it appears that those who were already more regular users may have tended to increase consumption, potentially increasing vulnerability to the risks associated with marijuana use. In contrast to previous studies , we found participants who endorsed greater frequency of marijuana use had greater frequency of use of tobacco products. Following legalization this was particularly true for e-cigarettes. The specific mechanism for this association is uncertain, but there are multiple possibilities. First, it may be that relaxing restrictions on a specific substance reduces substance-specific concerns about harm , which then generalizes to other drugs. Alternatively, the association can be explained by use of products that deliver both drugs at the same time , or newer vaporizing devices that may do so separately. It is plausible that innovations in nicotine vaping devices encourages marijuana vaping, promoting diversified marijuana product use and synergistically increasing use of both products. This is consistent with the strengthening association between marijuana and e-cigarette use frequencies post legalization. The possibility that lessening marijuana barriers increases tobacco use is concerning given evidence that co-use is associated with psychosocial distress , health problems , nicotine dependence , and tobacco cessation failure . The present study has several limitations. It is a secondary analysis of a naturalistic study of young adult tobacco users, which limited the specificity of marijuana-related measures and may have yielded a sample with disproportionately frequent marijuana use. There is a strong need for additional studies that include outcomes beyond simply quantity,growers equipment frequency or prevalence of use . The design may limit generalizability to other young adult samples.

Another limitation is reliance on self-reported substance use data, though evidence suggests self-report tends to be accurate in observational studies, given the lack of strong demand characteristics . Additionally, self-reported data include only some days during 2015–2019 and may not be representative of use during the entirety of this period. Finally, while the study captured self-reported use of marijuana and nicotine/tobacco products before and after legalized sales of recreational marijuana began in California, we did not directly evaluate access to marijuana retail outlets or other methods of product acquisition. In examining marijuana use before and after legalization of recreational sales in California, we found that frequency of use did not change significantly overall, including following legalization. We also found that increases in marijuana frequency tended to coincide with increased tobacco use, and a specific post-legalization association with e-cigarette use. Finally, we found that the most frequent users of marijuana after legalization were those who had used most often prior to 2018. Findings suggest loosening of marijuana restrictions could lead to negative health consequences for young adults. Strengths of the study include the sample size, and the repeated evaluation of a cohort of young adults before and after legalization. Further research is needed to confirm these findings, to understand how risks associated with changes in marijuana policy can be attenuated, and to identify surveillance targets. The continuously evolving marijuana and tobacco landscape also indicates the importance of ongoing evaluation of co-use. Correlation coefficients between environmental and biomarker measurements are widely used in environmental health assessments and epidemiology to explain the exposure associations between environmental media and human body burdens. As a result considerable attention and effort have been given to interpretation of these coefficients. However, there is limited information available on how the variance in environmental measurements, the relative contribution of exposure sources, and the elimination half-life affect the reliability of the resulting correlation coefficients.

To address this information gap, we conducted a simulation study for various exposure scenarios of home-based exposure to explore the impacts of pathway-specific scales of exposure variability on the resulting correlation coefficients between environmental and biomarker measurements. Bio-monitoring data, including those from blood, urine, hair, etc., have been used extensively to identify and quantify human exposures to environmental and occupational contaminants. However, because the measured levels in biologic samples result from multiple sources, exposure routes, and environmental media, the levels mostly fail to reveal how the exposures are linked to the source or route of exposure . Thus, comparison of biologic samples with measurements from a single environmental medium results in weak correlations and lacks statistically significance. In addition, cross-sectional biological sample sets that track a single marker have large population variability and do not capture longitudinal variability, especially for compounds with relatively short biologic half-lives, which can be on the order of days such as pesticides and phthalates. Therefore, in the case where the day-to-day variability of biological sample measurements is large, the use of biomarker samples with a low number of biological measurements in epidemiologic studies as a dependent variable can result in a misclassification of exposure as well as questions of reliability. For chemicals frequently found at higher levels in indoor residential environments than in outdoor environments, it is common to assume that major contributions to cumulative intake are home-based exposure and/or food ingestion. This simplification can be further justified because people generally spend more than 70 percent of their time indoors. Compounds with significant indoor sources and long half-lives in the human body– on the order of years for chemicals such as polybrominateddiphenyl ethers –have been found to have positive associations between indoor dust or air concentrations and serum concentrations in U.S. populations. On the other hand, extant research has not reported significant associations between indoor samples and biomarkers for chemicals primarily associated with food-based exposures, for example, bisphenol-A and perfluorinated compounds.

For chemicals with both home and food-based exposure pathways and short body half-lives , as is the case for many pesticides, a significant association between indoor samples and biomarkers is found less frequently or relatively weak compared to PBDEs. To better interpret these types of findings, we provide here a simulation study for various exposure scenarios to explore the role of the chemical properties and exposure conditions that are likely to give rise to a significant contribution from indoor exposures. We then assess for these situations the magnitude and variance of the associated correlation coefficients between biomarker and indoor levels. The objectives of this study are to generate simulated correlation coefficients between environmental measurements and biomarkers with different contributions of home-based exposure to total exposure and different day-to-day and population variability of intake from both residential environments and food, to interpret the contribution of home-based exposure to human body burden for two hypothetical compounds whose half-lives are on the order of days and years, and to determine how the pattern of variability in exposure attributes impacts the resulting correlation coefficients linking biomarker levels to exposure media concentrations.Because some indoor contaminants are considered potential threats to human health, many studies have applied significant resources to examine the relationship between exposure to indoor pollutants and adverse health effects. However, these studies are potentially limited by the use of a single or a few environmental and biological samples. The significant implications of this situation are reflected in our results. Multi-day, multi-person sample analyses are costly and labor-intensive. In addition, the resulting R2 values from these studies are not interpreted or poorly interpreted in terms of variability and contribution of exposure sources and the biological half-life of a compound. In this regard, the simulation study in this paper provides an important step towards interpreting the relative contribution of home-based exposure to human body burden for two compounds whose biological half-lives are significantly different . Although these two compounds do not cover the full range of chemical substances, bracketing half lives allows us to quantify the significance of source, measurement,plant benches and exposure pattern variability for disaggregating body burden. In particular, it shows that exposure variability and different contributions of exposure sources are more interconnected than commonly considered in many experimental studies. The work also brings to attention the need to understand the impact of a chemical half-life on the relationship between environmental exposures and bio-monitoring data. The sensitivity of day-to-day variability of wipe concentrations and food exposures on the resulting R2 values also points to the importance of understanding variability and contribution of exposure sources. Finally, future work includes computing the relative number of samples needed for various levels of confidence to disaggregate body burden for various types of compounds , environments, and exposure pathways. Despite the lack of experimental data, the simulated results provide key insights on the role of the variability and contribution of exposure sources and biological half-lives in quantifying a relationship between indoor exposure and human body burden. This approach will be useful for designing future exposure and epidemiologic studies that includes indoor environmental samples and bio-monitoring data.In 1996, California became the first state to legalize medical marijuana. Known officially as the Compassionate Use Act, Proposition 215 allowed patients and caregivers to cultivate and possess marijuana for medical use. The campaign in favor of Proposition 215 focused on the benefits for seriously ill patients. Claiming that the Proposition “sends our children the false message that marijuana is safe and healthy,” the campaign against the Proposition focused on anti-drug education .

Neither side addressed potential public health consequences. If Proposition 215 led to an increase in marijuana use, for example, might it also lead to higher rates of all injury deaths , including deaths from assault , deaths from motor vehicle crashes , and—the subject of the present study—deaths from suicide ? Such consequences assume that medical and recreational users are similar. With one exception, the evidence supports this assumption. Since most California medical users were introduced to marijuana as recreational users, for example, it is reasonable to assume that the user-types have similar socioeconomic backgrounds . Compared with recreational users, however, California’s medical users were more likely to report early health problems or disabilities that would warrant medical use . Although Proposition 215 was drafted so loosely that it effectively legalized all uses of marijuana , marijuana use by California juveniles, who were not eligible for medical marijuana certificates, did not increase following Proposition 215 . Nevertheless, at the national level, during a 15-year period when a majority of states loosened their control of medical marijuana, the U.S. suicide rate rose by 24 percent , prompting many to question how legalization and suicide might be linked. The systematic evidence connecting this trend to the availability of medical marijuana is ambiguous, however. Rylander, Valdez, and Nussbaum , for example, find no correlation between a state’s suicide rate and the number of medical marijuana cardholders in the state. Similarly, comparing suicide before and after a state enacts a medical marijuana law, Grucza et al. find no change in a state’s suicide rate. In contrast, Anderson, Rees, and Sabia report a 10.8 percent reduction in suicides averaged across all medical marijuana states. Attributing a suicide trend to the availability of medical marijuana raises questions about the potential mechanisms at play. What theoretical mechanisms could lead us to expect a relationship between the availability of medical marijuana and suicide? Could these mechanisms be more salient for certain types of suicides than others? If the expected relationship is observed, what methodological rules could be used to support a causal interpretation of the relationship? We address these questions in order. responds to changes in opportunity. Holding opportunity constant, risk responds to changes in motivation. Chew and McCleary use motivation/ opportunity mechanisms to explain life course changes in suicide. Kubrin and Wadsworth use motivation/opportunity mechanisms to explain the effects of socioeconomic factors and firearms availability on race-specific suicide. Wadsworth, Kubrin, and Herting use motivation/opportunity mechanisms to explain suicide trends for young Black males. Consistent with this literature, we argue that if medical marijuana affects suicide risk, it must do so through one or both pathways. Mental health theories operate through a motivation pathway. The psychiatric consensus is that suicide is related to depression, anxiety, and other treatable disorders . If marijuana alleviates the acute stress associated with these disorders, then we expect suicide risk to decrease following legalization of medical marijuana.

However, this trend was not significant in the larger subject population presented here, and was further decreased by the additional noise correction steps used in the present study, including removal of motion related and global signal confounds. While we did not find resting-state functional connectivity to be related to vascular differences, it did exhibit a dependence on the magnitude of spontaneous BOLD fluctuations. Previous work has shown that resting-state BOLD connectivity and amplitude also co-vary across behavioral and disease states . If spontaneous BOLD fluctuations are interpreted as faithful measures of intrinsic neural activity, these results may be interpreted as evidence for true variations in both the amplitude and coherence of the underlying neural sources. On the other hand, the results might also reflect variations in the BOLD signal’s sensitivity to spontaneous neural activity. As the BOLD signal becomes more sensitive to underlying neural activity, there can be a relative increase in the signal-to-noise ratio of the resting-state measures, i.e. the magnitude of BOLD signal fluctuations of neuronal origin divided by the magnitude of fluctuations of non-neuronal origin. Since the estimated correlation between two signals tends to increase with SNR, the relation between BOLD functional connectivity strength and fluctuation magnitude may partially reflect the change in SNR with BOLD signal sensitivity. Future experiments with simultaneous electroencephalography and fMRI would be helpful in elucidating the relationship between resting BOLD amplitude and the underlying coherence of intrinsic neural activity. In this study, subjects performed a 5-minute motor task before undergoing resting-state scans. Previous work has shown that spontaneous BOLD fluctuation amplitude and connectivity can be altered by a preceding motor task . However,cannabis drying racks these effects followed a strenuous task causing muscle fatigue, or extended periods of tapping inducing neural plasticity. We expect that these types of effects would be minimal for the relatively short and non-strenuous finger-tapping task used in this study.

Furthermore, as scan order was kept constant for all subjects, it is unlikely that task-related modulation of resting BOLD signals accounted for the inter-subject differences observed here. To conclude, the results presented here indicate that measures of resting-state BOLD fluctuation amplitude and connectivity are relatively insensitive to inter-subject differences in baseline CBF. This lack of dependence on baseline CBF, when compared to the strong inverse dependence exhibited by the task BOLD response, appears to be caused in part by a weakened inverse relationship between relative changes in CBF and baseline CBF during rest. In addition, flow-metabolism coupling is tighter during rest than during a robust motor task, further weakening the relationship between spontaneous BOLD fluctuations and CBF. As resting-state BOLD fluctuations are often used to examine functional connectivity changes in patients, where both disease and medication can alter the vasculature, our findings are very important. However, while we have demonstrated that measures of spontaneous BOLD fluctuations are not affected by vascular differences between subjects, further work is necessary to identify other non-neural confounds before these metrics can be used as reliable estimates of underlying spontaneous neural activity.In the first part of this work, we assessed the effect of a 200 mg dose of caffeine on resting-state BOLD connectivity in the motor cortex across a sample of 9 healthy subjects. As caffeine reduces baseline CBF through adenosine antagonism, and previous work has suggested that vasoconstriction increases the sensitivity of the BOLD signal to neural activity, we expected to find that BOLD fluctuations and functional connectivity were increased. Surprisingly, we found the opposite: that caffeine significantly reduced spontaneous BOLD fluctuations and connectivity. These results suggest that the primary mechanism of caffeine’s action on functional connectivity is probably not through vascular changes, but possibly through caffeine’s direct effect on neural activity. Preliminary work by our group supports this conclusion by directly demonstrating caffeine induced reductions in neural connectivity using MEG . However, the physiological mechanisms behind the caffeine-induced decrease in BOLD signal power remain unknown.

It is possible that a caffeine-induced decrease in the coupling between CBF and oxygen metabolism, which has been found during task , could be responsible for the smaller BOLD fluctuations, but this remains to be investigated. Even if vascular confounds are not responsible for the results presented here, caffeine should still be carefully considered in the design and interpretation of resting-state BOLD fMRI studies. In the second part of this work, we employed a non-stationary analysis approach to gain further insight into the mechanisms of caffeine’s effect on functional connectivity. Specifically, we used a sliding window correlation analysis to assess whether caffeine consistently weakens the correlation over time or if transient periods of strong correlation still exist, albeit less frequently. We found that BOLD correlation was significantly more variable over time following a caffeine dose, and that extended periods of strong correlation still existed between periods of lower correlation. Furthermore, the temporal variability of BOLD signal correlation was driven by phase differences between the BOLD signals in the left and right motor cortices. While a consistent decrease in correlation could be caused by an overall change in the vascular system induced by caffeine, it is unlikely that a shift in the state of the vascular system would give rise to the increase in the non-stationarity of the correlations that we found. Instead, the caffeine-induced increase in the temporal variability of functional connectivity tends to support the existence of greater temporal variability in the coherence of the underlying neural fluctuations. In the third part of this work, we investigated the BOLD signal dependence on inter-subject differences in baseline CBF using a sample of 17 healthy subjects. We acquired simultaneous BOLD and CBF measures during a motor task and resting state. Consistent with prior studies, we found a strong dependence of the task-evoked BOLD response on inter-subject variations in baseline CBF, but found a much weaker and not significant dependence of the resting-state BOLD response on baseline CBF. In addition, inter-hemispheric resting-state BOLD connectivity between motor cortex regions did not show a significant dependence on baseline CBF.

The strong inverse dependence of the task BOLD amplitude on baseline CBF appears to be caused by the direct dependence of %∆BOLD on %∆CBF, which is modeled in the Davis equation . This is because %∆CBF is inversely related to baseline CBF, as CBF0 is the denominator in calculating %∆CBF. We find that both of these relationships are weaker during rest than task. The reduced dependence of percent changes in CBF on baseline CBF during rest is caused by significantly smaller absolute CBF changes, which are independent of inter-subject differences in baseline CBF during both task and rest. The weakened relationship between relative changes in BOLD and CBF appears to be caused by tighter flow-metabolism coupling during rest than during a robust motor task. These two factors work together to produce an insignificant relationship between spontaneous BOLD fluctuations and baseline CBF. These findings suggest that differences in both the amplitude and correlation of spontaneous BOLD fluctuations between subjects are probably more reflective of neural activity differences than vascular differences.As the use of fMRI to estimate functional connectivity in the brain is a new and rapidly growing field, work that identifies potential limitations of the technique is very important. In this dissertation, differences in baseline blood flow have been investigated as confounds to interpreting BOLD connectivity changes as neural connectivity changes.While these findings are very encouraging for the field, more work remains to be done before BOLD connectivity measures can be relied on as robust measures of neural connectivity. In particular, future work is necessary to determine whether a caffeine-induced decrease in the ratio of flow-metabolism coupling is responsible for the reduced resting-state BOLD signal amplitude found in this study. If proven to be the case,vertical grow system it would suggest that flow-metabolism coupling changes can influence functional connectivity measures even if changes in CBF alone do not.Empirically determining the ratio of blood flow to oxygen metabolism changes during rest can be challenging because of the inherently low signal to noise ratio in ASL. While previous work has shown that caffeine reduces flow-metabolism coupling in response to a task , it remains to be seen whether this is the case during resting state. An important future study would therefore use the Davis model introduced in Chapter 4 to determine how caffeine affects flow-metabolism coupling during resting-state, and whether this change could be responsible for the diminished spontaneous BOLD fluctuations . This future experiment would also address some of the limitations in the methods applied in Chapter 4 to determine flow-metabolism coupling during rest. For example, in typical Davis model experiments, a model of the task paradigm is used to obtain estimates of BOLD and CBF amplitude changes that stem from neural activation, however this is not possible in resting-state studies. In Chapter 4 of this work we estimated BOLD and CBF amplitude changes using root mean square values, but these values are not necessarily reflections of neural activity. For example, physiological noise or motion may contribute significantly to the variance of the resting BOLD and CBF fluctuations and would confound Davis model estimates using RMS values. A new technique using independent component analysis has shown promise in identifying and removing signal components of non-neural origin from spontaneous BOLD signals . In this method, multi-echo data is acquired and ICA components are examined for a linear dependence on echo time , which indicates that they are truly representative of changes in deoxyhemoglobin content and thus likely to reflect neural activity rather than noise.

It is still unclear how this method can be used on CBF data, which has low sensitivity to deoxyhemoblobin content, but possibly it could be applied to the raw multi-echo ASL data before performing a sliding window difference on the cleaned first echo data set. Another approach that could improve Davis model estimates during resting-state is to acquire simultaneous EEG/fMRI data, which could provide a reference spontaneous neural signal for estimating BOLD and CBF amplitude changes that directly result from neural activity changes . Furthermore, the EEG measurements could provide information on how the amplitude of electrical power fluctuations are altered by caffeine. These findings would shed light on whether the mag-nitude of neural activity is truly reduced by caffeine and then reflected in the decreased power of spontaneous BOLD fluctuations.For each subject and run, the sliding window correlation coefficients between the measured motor cortex BOLD time courses were plotted in a histogram. The sliding window correlation coefficients between the simulated data-derived SNR signal + noise pairs were plotted in a histogram below. For several subjects, the data produced visibly multi-modal histograms in both the pre-dose and post-dose scan sessions, which are not predicted by noise. Representative examples are shown for the pre-dose scan section in Figure B.3 and post-dose scan section in Figure B.4. These findings suggest that true variations in correlation exist between the neural “signal” components of the BOLD signal.In addition to the multi-modal shape of the measured correlation histograms, it can be seen that the variance in measured r-values is visually larger than the variance of the simulated r-values, particularly for the pre-dose subject. To quantify the likelihood that noise is responsible for the measured variability in BOLD correlation present in our data, we simulated correlation variability values. We created a histogram of variability values, which were calculated as the standard deviation of the sliding window correlation time course between each pair of simulated signal + noise time courses. This resulted in 10,000 variability values. Then the actual measured variability was compared to the simulated data to determine a percentile and associated p-value . An example of this procedure is shown in Figure B.5.This was done for each subject and run. Plots of simulated data percentile versus measured variability values are shown for each subject and run during the pre-dose and post-dose scan sessions . Note there are two runs per subject and session. Data points above the dashed line correspond to p-values less than 0.05. These simulations suggest that it is highly unlikely that noise is primarily responsible for the temporal variability in BOLD correlation found in the present study. While noise may still contribute to correlation variability, it is probable that underlying variability in the coherence of neural activity is responsible as well.Since California’s Compassionate Use Act of 1996, cannabis has been legally available — under state but not federal law — to those with medical permission. Until 2018, however, no statewide regulations governed the production, manufacturing and sale of cannabis. Prior to development and enforcement of statewide regulations, there were no testing requirements for chemicals used during cannabis cultivation and processing, including pesticides, fertilizers or solvents .

Because both the global and RVT signals were obtained in arbitrary units, we first normalized the global and RVT signals to their mean values to derive percent changes of these signals. We then computed the energy below 0.08 Hz and the standard deviation of the percent change signals. To assess changes in the finger tapping BOLD response to caffeine, average BOLD block responses were extracted from the combined motor cortex ROI. Each subject’s BOLD block response was interpolated to a time resolution of 0.25s and the following timing parameters were computed: 1) time to reach 50% of the peak response , 2) time at which the response falls to 50% of the peak response , and 3) the full-width half-maximum . In addition, the peak BOLD response amplitudes were calculated for each subject.Figure 2.1 provides a qualitative summary of the results. Resting-state functional connectivity maps obtained by using the average signal from the left ROI as a reference are shown for a representative slice from each subject. The top two rows display each subject’s average pre-dose and post-dose maps for the caffeine session. The extent of significantly correlated voxels in the post-dose connectivity maps has been visibly diminished in most subjects, as compared to the pre-dose maps. In contrast, for most subjects the control session maps in the bottom two rows do not demonstrate an obvious difference in connectivity between pre-dose and post-dose conditions. Metrics of connectivity strength are graphed in Figure 2.2. The scatter plots in the top row show the mean z scores for each subject from the caffeine session and control session, with the solid black lines representing equality between the pre-dose and post-dose sections. The caffeine session plot shows a significant = 3.3, p = 0.01 caffeine-induced decrease in mean z score across subjects. In contrast,rolling benches the mean z scores in the control session are clustered about the centerline, without a significant = -0.51, p = 0.62 change between pre-dose and post-dose conditions.

These results are consistent with a repeated measures two-way ANOVA, which showed that the interaction between period and session is significant = 9.8, p = 0.01. In addition, the baseline mean z scores in panels and are not significantly different = 0.36, p = 0.73. The mean z scores presented here were determined using the average time course from the left motor ROI as a reference. If the average time course from the right ROI is used as a reference instead, the caffeine-induced decline in mean z score is still significant = 2.8, p = 0.02.We have shown that caffeine reduces resting-state BOLD connectivity in the motor cortex. This reduction was apparent in the connectivity maps from a representative subject and in the quantitative metrics of connectivity obtained for all subjects. In addition, we found that baseline CBF and the magnitudes of the spontaneous BOLD fluctuations were decreased by caffeine. Physiological confounds, such as changes in respiration and arterial CO2 levels, can alter BOLD signal fluctuations. Regression-based removal of global and respiration volume per time signals has been shown to reduce the influence of these physiological confounds . When we included the regression of the RVT or global signal in our data analysis we found that the post-dose connectivity metrics were still significantly decreased. In fact, RVT signal regression did not significantly alter the functional connectivity metrics . Furthermore, we found that the variance or energy in the global and RVT signals were not significantly altered by caffeine . These findings suggest that the caffeine-induced reduction in BOLD connectivity is not primarily due to respiration changes. In agreement with the literature, we found that the caffeine-induced reduction in CBF was associated with an acceleration of the BOLD response to a finger tapping task . In addition, vasoconstriction due to either hypocapnia or nitric oxide synthase blockade has been found to increase low-frequency fluctuations in CBF . We have previously presented a bio-mechanical model that explains how vasoconstriction can increase the dynamic compliance of the arterioles and thus increase the responsiveness of the vasculature to neural stimulus and fluctuations .

However, in this study we found that both the spectral amplitude of low-frequency BOLD fluctuations and the coherence between resting BOLD fluctuations were diminished by caffeine, suggesting that an increase in bio-mechanical responsiveness was not a dominant factor. As binding of adenosine to A2 receptors is associated with vasodilation, caffeine-related antagonism may reduce the ability of adenosine to contribute to functional increases in cerebral blood flow. In a recent study, we found that a 200 mg oral dose of caffeine led to a significant decrease in the absolute functional CBF change in response to a visual stimulus but resulted in a significant increase in the percent CBF change . These results indicated that the drop in the absolute functional CBF change was primarily related to a drop in baseline CBF as opposed to reflecting an impairment of neurovascular coupling. Also consistent with our prior work and the results of this study, Liau et al. did not find a significant change in the BOLD response. Taken together, these results indicate that caffeine’s effect on adenosine-related vasodilation does not significantly reduce the task-related functional BOLD response. If task-related and resting-state BOLD activity share a common neurovascular coupling pathway, then the task-related BOLD results suggest that an impairment of adenosine-related vasodilation was probably not the dominant factor in the reduced functional connectivity observed in this study. Further studies elucidating the similarities and differences in neurovascular coupling for task-related and resting-state BOLD signals would be useful. In addition to its vasoconstrictive effects, caffeine directly influences neural activity. Caffeine stimulates the central nervous system by antagonizing adenosine A1 receptors throughout the brain. This blocks the inhibitory actions of adenosine, which include hyperpolarization of membrane potentials and the inhibition of neurotransmitter release . Although caffeine acts as a neurostimulant, previous work has shown that a 200 mg dose of caffeine reduces the power of resting electroencephalography activity in the alpha, beta, and theta bands . In addition, the coherence of anterior cortex neural fluctuations in the alpha and theta bands is decreased by caffeine when compared to periods of caffeine abstinence .

Simultaneous EEG/fMRI recordings have shown that resting-state BOLD fluctuations are significantly correlated with EEG power fluctuations in the alpha band , the beta band , and the theta band . These prior findings suggest that the reduction in resting-state BOLD fluctuations and connectivity found in this study may primarily reflect changes in neural power fluctuations. Although the physiological effects of caffeine are often beneficial,rolling grow table such as enhanced mood, attention, wakefulness, and motor speed , a 200 mg dose has been shown to impair several types of memory tasks, including motor learning of a finger tapping task . In light of our findings, this observed decrease in motor learning might reflect a caffeine-induced decrease in resting state neural connectivity. Further experiments with simultaneous EEG/fMRI would be useful to determine if caffeine-induced changes in neural power fluctuations are directly related to the observed reduction of BOLD connectivity. In addition to caffeine, a number of pharmacological agents have been found to alter resting-state BOLD connectivity. Both hypercapnia and cocaine have been shown to reduce the magnitude and coherence of resting-state BOLD fluctuations, while anesthesia appears to have varying effects depending on the specific agent and brain region. Cognitive disorders such as Alzheimer’s disease, schizophrenia, multiple sclerosis, and epilepsy have also been shown to modulate BOLD connectivity . While changes in resting-state BOLD connectivity are typically interpreted as changes in coherent neural activity across spatially distinct brain regions, changes to the neurovascular system may also alter connectivity. For example, as mentioned in the Introduction, hypercapnia appears to decrease BOLD connectivity by weakening the neurovascular coupling between spontaneous neural activity and resting-state BOLD fluctuations. Since many pharmacological agents and diseases are likely to affect both the neural and vascular systems, a greater understanding of the neural and vascular mechanisms that give rise to resting-state BOLD connectivity will be critical for the correct interpretation of changes in connectivity. Similar to a prior study examining the effect of caffeine on baseline oxygen metabolism , we used a control session to examine potential changes in baseline CBF, resting-state functional connectivity, and low-frequency BOLD fluctuations that might have been caused by differences in the subject’s state between the pre-dose and post-dose scan sections. We did not find significant differences between the pre-dose and post-dose results obtained during the control session, indicating that the caffeine-induced decrease in BOLD connectivity was not due to factors, such as subject fatigue, associated with participating in two scan sections. As our protocol did not involve the administration of a placebo dose, it is possible that psychological effects associated with taking a dose could have affected the functional connectivity measures.

Future studies would be useful to assess the effect of a placebo dose on resting-state BOLD connectivity. There was a range of caffeine usage in the current sample of subjects. Prior work has demonstrated variability in the task-related BOLD response due to differences in dietary caffeine consumption. Inter-subject differences in caffeine consumption may also influence the effect of caffeine on resting-state BOLD connectivity. In addition, subjects in this study were asked to abstain from caffeine for at least 12 hours prior to being scanned. Caffeine withdrawal has been shown to alter EEG power in the alpha and theta bands . It is possible that caffeine’s effect on resting-state BOLD connectivity will differ based on the subject’s state of withdrawal during the pre-dose scan section. Further investigation of the effects of dietary consumption and withdrawal on caffeine-induced changes in BOLD connectivity will be helpful. The work presented here shows that caffeine reduces resting-state BOLD connectivity in the motor cortex, most likely by reducing the amplitude and coherence of neural power fluctuations. As the distribution of adenosine receptors varies across the brain , it is possible that the effect of caffeine on functional connectivity will vary with the local receptor concentration. While future work is necessary to determine whether caffeine alters connectivity in other functional networks, the findings of this study indicate that caffeine usage should be carefully considered in the design and interpretation of studies involving resting-state BOLD connectivity.Resting-state functional MRI can be used to assess functional connectivity within the brain through the measurement of correlations between spontaneous blood oxygenation level-dependent fluctuations in different regions. Synchronous BOLD fluctuations have been consistently found at rest within functional networks such as the motor cortex, visual cortex, and default mode network . A growing number of studies have shown that functional connectivity is altered for cognitive disorders such as multiple sclerosis, epilepsy, Parkinson’s, and Alzheimer’s disease , suggesting that resting-state studies can aid in disease diagnosis and improved understanding of disease mechanisms. In addition, inter-subject differences in functional connectivity have been shown to correlate with performance on working memory tasks and intelligence . To date, functional connectivity studies have typically employed stationary metrics obtained with seed-based correlations or independent component analysis computed over an entire resting scan. However, recent work has shown that the correlation strength between different brain regions may vary in time. For example, a study using magnetoencephalography found transient formations of widespread correlationsin resting-state power fluctuations within the DMN and task positive network . This non-stationary phenomenon was particularly apparent when considering nodes in different hemispheres, which exhibited very low stationary correlation. Another study using fMRI found that the phase angle between spontaneous BOLD fluctuations in the DMN and TPN varied considerably over time, with frequent periods of significant anti-correlation . These studies indicate that coordination of spontaneous neural activity is a dynamic process, and suggest that time varying approaches can provide critical insights into functional connectivity. Despite the increasing appearance of resting-state functional connectivity studies in the literature, it remains difficult to interpret the physiological mechanisms behind changes in BOLD signal correlations. The BOLD signal provides an indirect measure of neural activity, and is a complex function of changes in cerebral blood flow , cerebral blood volume, and oxygen metabolism . Factors that alter any part of the pathway between neural activity and the BOLD response can change functional connectivity measurements, making it difficult to decipher the origin of this effect. For example, caffeine is a widely used stimulant that has a complex effect on the coupling between neural activity and blood flow .

Comparing states before and after MML enactment to states without such laws, they find that MML enactment leads to a significant reduction of about 16% in the fatality rate for individuals aged 20-40 from motor vehicle accidents. The authors also find that legalization of medical marijuana leads to a significant 13.2% decrease in fatalities from alcohol-involved accidents. Pacula et al. recognized that heterogeneity in MML policy may result in heterogeneous effects, and they repeated the analysis of Anderson et al. distinguishing between MMLs with and without legal dispensary allowances. Their results confirm that MML enactment is negatively associated with alcohol-involved traffic fatality rates, but they show that this negative relationship is almost entirely offset in states that allow dispensaries. These results are supported by Choi , who uses individual-level data from the National Survey of Drug Use and Health from 2004-2012 and finds that allowing marijuana dispensaries is associated with a 9.1 and 23.9 percent increase in reporting driving under the influence of alcohol and drugs respectively. Why should there be differences in the effects of MMLs on traffic fatalities depending on the allowance of dispensaries? For one, Downey et al. finds that high tetrahydrocannabinol levels significantly increase driving impairment,hydroponics flood tray and MML states that allow dispensaries have been shown to have greater diffusion of high-potency cannabis . Comparing Colorado to non-MML states, Salomonsen-Sautel, Min, et al. find no significant increase in the propor-tion of drivers involved in a fatal motor vehicle crash who tested positive for marijuana until the period when medical marijuana commercialization expanded .The evidence from Table 3.1 can help explain the different findings of past research. There is wide variation in medical marijuana penetration depending on the supply regulations established by state MML policies and the level of federal enforcement.

Using the timing of state MML enactment to identify the effects of medical marijuana legalization on traffic fatalities misses important changes in the availability of marijuana that occur long after initial MML passage. These omitted effects may be of particular importance when assessing the impact on youths, who are more likely to self-report willingness to drive after consuming cannabis or after consuming both marijuana and alcohol . Irrespective of its effects on traffic accidents, marijuana’s role as a substitute for alcohol has important health implications. The question of whether alcohol and marijuana are substitutes or complements has received substantial attention in the literature, but findings have varied. Given the breadth of work, I focus here on studies using marijuana liberalization policies to identify the relationship between marijuana and alcohol use. Empirical evidence on the effect of marijuana decriminalization policies is mixed for a review of the literature, and more recent work specifically examining the effects of MMLs on alcohol consumption has found similarly varied outcomes. Using data from the 2004-2012 National Survey of Drug Use and Health , Wen et al. report a positive effect of MML enactment on frequency of binge drinking for adults over 20 years of age but no effect on alcohol use by individuals under age 21. In contrast, Anderson et al. examine data from the 1993-2010 Behavioral Risk Factor Surveillance System and find that MML enactment has a significant negative impact on past-month drinking and binge-drinking among young adults. Differences in these findings may in part be driven by the fact that the studies cover different years, and hence different state laws are used in the identification of the treatment effects. Pacula, MacCoun, et al. caution against using a binary indicator for marijuana decriminalization laws, as there is substantial variation in how these laws are implemented, enforced, and hence understood by citizens.

As discussed by Pacula et al. and demonstrated in section 3.2, state MML regulations vary greatly and can thus be expected to generate heterogeneous effects. Additionally, the categorical MML measure used in past analyses does not capture the later evolution of medical marijuana markets shown in Figure 3.2. The inclusion of state-specific trends in the empirical specification will thus confound preexisting trends with the dynamic effects of the policy . By using medical marijuana registration rates instead of a binary MML indicator as the policy variable of interest, this paper overcomes these limitations. While there are only a few economic studies of the relationship between marijuana and opioid use, clinical studies by Cichewicz and Welch and Ramesh et al. suggest that smoked cannabis and cannabinoids have opioid sparing properties and may prevent the development of tolerance to opiates. Additionally, studies of opioid dependent patients indicate that moderate use of cannabis or synthetic cannabinoids leads to significantly improved outcomes for medication compliance, opioid withdrawal symptoms, and retention in treatment . The potential for medical marijuana to reduce opioid abuse is supported by Bachhuber et al. , who find that MML enactment is associated with a significant 20% decrease in age-adjusted prescription opioid-related mortality, with these effects strengthening several years post-enactment. Powell et al. find no effect of MMLs on opioid abuse or mortality, but they find that legalizing dispensaries reduces opioid abuse and mortality by about 15%. While these results suggest that increased medical marijuana availability offers the benefit of significantly reducing opioid use, past work has not examined to whom these benefits are accruing. While all age groups have seen significant growth in opioid-related deaths, the rise in mortality rates has been most pronounced for adults aged 45-64 .

Indeed, recent work by Case and Deaton shows that drug poisoning deaths have significantly contributed to the reversal in mortality improvement experienced by US white non-Hispanics aged 45-54 between 1999 and 2013. By estimating the effects of increased medical marijuana availability on opioid poisoning mortality separately by age group, this paper contributes toward further understanding whether medical marijuana can serve to improve the deteriorating mortality outcomes for older individuals. To assess whether increased cannabis use results in more automobile accidents, data was compiled from the Fatal Accident Reporting System for 1990-2013. FARS, collected by the National Highway Traffic Safety Administration ,hydro flood table contains detailed information on the circumstances of the accident, in addition to information on the characteristics of occupants and non-occupants. In order to most precisely identify which age groups experience increased risk of causing fatal accidents,traffic fatalities are analyzed separately by age of the driver involved in single-vehicle accidents only. Summary statistics are given in Appendix H-. It should be noted that the traffic fatality variables used in this analysis differ slightly from that used in Anderson et al. . Their analysis separates traffic fatalities by age of the deceased, while my empirical model analyzes fatalities by age of the driver involved. While these two variables should be correlated, focusing on age of the driver involved provides a better indicator of which individuals are changing their alcohol and cannabis consumption in response to increased medical marijuana availability.To investigate substitution between marijuana and other addictive substances, substance related poisoning mortality data from 1990-2013 was downloaded from the Center for Disease Control’s Wide-ranging Online Data for Epidemiologic Research interface. Alcohol poisonings are defined as deaths with ICD-10 code X45, X65, Y15, or F10.0.5 Opioid analgesic poisonings are defined as those under ICD-10 codes X40-X44, X60-X64, or Y10-Y14, where a prescription opioid was also coded. Heroin-related poisonings are defined as those under ICD-10 codes X40-X44, X60-X64, or Y10-Y14, where heroin was also coded. Summary statistics are given in Appendix H-. Deaths are coded based on multiple-cause reporting instead of the underlying cause, since this can provide a more complete representation of all conditions that contributed to the death . A death is counted if the specified condition is listed on the death certificate as a contributing factor, but the condition need not be the specified as the underlying cause of death. Thus, these counts do not necessarily represent unique deaths on all fatal traffic accidents, but these effects are only statistically significant for daytime accidents. For older adult drivers, increased registration rates do not predict any significant change in fatal traffic accidents. However, while not statistically significant, the effects on drivers aged 45-64 are all negative, with the largest effects for nighttime accidents. While Table 3.2 indicates that greater marijuana availability leads to increased traffic fatalities involving young drivers, it is unclear whether these effects are driven by cannabis use alone or the joint use of cannabis with other substances. To disentangle the role of alcohol and marijuana in generating motor vehicle fatalities, Table 3.3 presents estimates of the effects of registration rate growth on traffic fatalities seperately by substance involvement.Panel A reports estimates of the effects of legal market growth on fatalities in which the driver’s blood alcohol content was tested and found to be equal to zero.

For all age groups, the estimates are insignificant. In contrast, for accidents in which the driver had a positive BAC value, Panel B shows that increased medical marijuana availability is associated with a significant 11.6% increase in traffic fatalities involving a driver aged 15-20 and an insignificant 5.3% decrease for drivers aged 45-64. While the results from Panel C are consistent with increased prevalence of cannabis use for drivers of all ages, Panel D suggests that it is the joint use of alcohol and marijuana that generates negative externalities in the form of increased traffic fatalities caused by drivers aged 15-24. The results of Table 3.3 are consistent with the experimental evidence from driving simulator studies, but caution should be taken in interpreting these results. Drivers are not regularly tested for cannabinoids, and the decision to test may well be endogenous with expansion in the medical marijuana market. Also, because THC is lipid-soluble and excreted slowly over time into urine, a positive test for cannabinoids does not necessarily mean the individual has used marijuana recently — let alone that cannabis-impairment caused the accident . Still, the results suggest that differences between youths and older adults in the decision to use marijuana and alcohol jointly may generate different health consequences caused by greater marijuana availability. To further examine substitution behavior, Table 3.4 presents estimates of the effects of registration rates on poisoning mortality involving alcohol , prescription opioid analgesics , and heroin . Panel A shows that higher registration rates predict a large and significant decline in alcohol-related poisoning mortality for adults aged 45-64. With increased medical marijuana access, older adults appear to substitute away from the heavy use of opioids. Registration rates have a significant negative effect on opioid-analgesic poisoning mortality for adults aged 45-64 of 11-15%. These results are smaller but in line with the findings of Powell et al. . In contrast, there is suggestive evidence of complementarity between alcohol and marijuana for youths aged 15-24, consistent with the evidence from Tables 3.2 and 3.3. As stated earlier, the Poisson specification was preferred over the more intuitive loglinear specification and the commonly-used negative binomial regression model. Since in many years, there are relatively few single-vehicle traffic fatalities, a loglinear specification will introduce considerable noise in the analysis, and will result in biased estimates under heteroskedasticity . While the negative binomial regression estimator can account for the over-dispersion apparent in the data, violation of the model’s assumptions about the underlying data-generating process will produce biased coefficients . Still, these alternative models can provide specification checks for the primary analyses. Tables 3.5 and 3.6 thus present coefficients on the registration rate variable from the log-linear and negative binomial specifications for completeness. In line with the estimates from Tables 3.2 and 3.4, these alternative specifications confirm that growth in the legal medical marijuana leads to a significant increase in weekend and nighttime traffic fatalities caused by drivers aged 15-20, and significant declines in alcohol and opioid analgesic poisoning deaths for adults aged 45-64. As the final set of sensitivity analyses, Tables 3.7 and 3.8 estimate the effects of medical marijuana market growth including only states that had enacted an MML as of 2015. The effects are thus estimated from differences in market size within the set of states that presently provide legal protections for medical marijuana. Since all of these states eventually passed laws, they may be considered more similar. For all outcomes, the results are largely unchanged. One difference of note is that the negative effects on traffic fatalities involving a driver aged 45-64 are larger and significant when the sample is restricted to MML states.

And for states that legalized state-licensed dispensaries in their initial MML, there were often substantial implementation lags between law enactment, the licensing of dispensaries, and the opening of dispensaries . Even if the date of first dispensary operation is correctly identified, production-related realities may lead to further lags before the full effects on access and price are realized. These lags are likely not random, but will be correlated with unobservable local attributes as well as enforcement efforts at the federal level. This paper shows that there is substantial heterogeneity both across states and over time in the extent to which users and suppliers have actively participated in state medical marijuana programs. Changes in perceived federal enforcement had far greater effects on medical marijuana take-up than MML enactment alone, and these effects were concentrated in those states that imposed relatively lax restrictions on legal producers. This indicates that medical marijuana participation is largely driven by the expected benefits associated with access to legal supply. A few other findings deserve mention. First, states with MMLs allowing chronic pain as a qualifying condition on average have significantly higher registration rates. This is unsurprising since these states have a larger pool of eligible applicants. However, the fact that the federal memos did not have a differential effect on registration rate trends in these states with laxer qualifying standards suggests the federal policies did not differentially affect physician willingness to recommend a “marginal” patient. Second, higher registration fees significantly reduce medical marijuana participation. This is consistent with anecdotal evidence that registration fees represent a barrier to take-up for many patients ,hydroponic flood table and future work should assess whether there is age or demographic heterogeneity in the elasticity of participation with respect to application costs.

From this paper’s analysis, changes in medical marijuana registration rates seem to follow a pattern largely consistent with economic models of rationality. Overall, both across- and within-state variation in medical marijuana patient registration rates is primarily driven by differences in the costs of obtaining marijuana. Given evidence that supply spillovers from legal medical marijuana markets to illegal markets largely occur through diversion from registered patients to unregistered consumers , changes in registration rates may more precisely reflect the spillover effects of MML policy compared to binary indicators for various dimensions of MML regulations. The conclusions of this paper can be used to inform the methodologies employed by future work studying the effects of MMLs on substance use and other related outcomes. Using a binary measure of MML enactment to identify the effects of marijuana liberalization relies on a coarse measure of the impact of MML policy on the marginal user. By instead evaluating the effect of MMLs through changes in registration rates or the policy aspects that induced such changes, future work can more accurately assess the complex and dynamic effects of liberalization policies on marijuana consumption and its associated health consequences in the general population. Since 1996, growing evidence of the potential medical benefits of cannabis and increasing social acceptance of the drug have led twenty-three states and Washington D.C. to enact medical marijuana laws , which legalize the use and cultivation of marijuana for medical purposes at the state level. Several of these states have also recently expanded legalization to allow the commercial sale of marijuana to adults for recreational purposes. Prior research has sought to estimate the effects of these policies on marijuana consumption by comparing MML states before and after law passage to states without such laws, but findings have varied.However, this approach assumes that the law’s passage had an equal and immediate effect on demand-side or supply-side channels that would increase consumption.

This paper instead uses improved data and a novel instrumental variables approach to directly study the effects of growth in the size of legal markets for medical marijuana on recreational use. Newly collected data on per capita medical marijuana patient registration rates show that changes in market size are driven by policies changing supply costs and not MML enactment alone, and that growth in the legal market has the unintended consequence of significantly increasing recreational marijuana use among both adults and adolescents. Evidence suggests that changes in the supply of medical marijuana, driven by policies changing the costs faced by legal producers, generate these spillovers to adolescent markets. A simple model of supplier behavior argues that if growth in the legal market is driven by lower production costs, then the price and availability of marijuana in the illegal market will track changes in the legal market. Specifically, the model implies that changes in federal enforcement will have larger effects on marijuana availability in states where legal producers are subject to relatively lax production limits. To con- firm the model’s predictions, I exploit two policies that changed federal enforcement against medical marijuana suppliers as cost shifters. Empirical evidence from medical marijuana patient registration rates confirms that supply costs drive changes in the legal market. The Ogden Memo caused an additional 2% of the adult population to register as medical marijuana patients in states with loose supply regulations, compared to an additional 0.2% in states with strict regulations. Similarly, the Cole Memo significantly reduced legal market growth in loosely regulated MML states but had no effect in states that strictly regulated supply. To estimate the causal effect of changes in medical marijuana supply on marijuana consumption, I instrument for market size with the interaction of initial state regulatory laxness and changes in federal enforcement.

Results show that reaching the median state’s legal market size would significantly increase the prevalence of marijuana use in the past month from 7.2% to 7.7% for adolescents aged 12-17 , from 17.3% to 18.9% for 18-25 year-olds, and from 4.4% to 5.2% for adults over age 25. The significant effects found for youths are in contrast to the null findings of prior research.2 This paper contributes a novel approach for estimating heterogeneous effects in markets with limited legal access. However, this approach has some potential limitations. First, the identification strategy relies on the assumption that federal enforcement changes did not impact demand differently in states with different supply regulations; a number of robustness checks address this potential threat to identification. Second, since the measures of marijuana use are self-reported, the estimates may be subject to reporting bias. Evidence from arrest rates is used to support that the results are not driven by changes in reporting behavior. Third, the approach does not account for potential cross-border spillovers of marijuana supply. This would tend to bias the estimates downward, but evidence using Montana as a case-study indicates that this bias is small. The paper proceeds as follows. Section 2.2 explains state variation in medical marijuana policies, outlines a simple model of supplier behavior, and shows that medical marijuana registration rates are a valid measure of legal marijuana market size. Section 2.3 outlines the methodological approach, section 2.4 describes the data, and section 2.5 presents the empirical results. Robustness checks are included in section 2.6, and section 2.7 concludes. With the Compassionate Use Act in 1996, California became the first state to pass a medical marijuana law , removing criminal penalties for the use, possession, and cultivation of medical marijuana. The state law directly contradicts the federal ban on marijuana use and distribution established in 1937,which remains in place today due to concerns cited by the Office of National Drug Control Policy about legalization’s effects on marijuana use and abuse . The potential increase in youth consumption is of particular concern as some research suggests use of marijuana during early adolescence predicts increased risk of dependence,ebb and flood table lower educational attainment, and cognitive impairmentDespite federal prohibition of marijuana, as of 2015, twenty-two additional states and Washington D.C. have enacted laws providing some protections for the use of medical marijuana; five of these states have also legalized the sale of marijuana to adults aged 21 and older for non-medical purposes. Table 2.1 shows a summary of state laws and various dimensions on which they differ.5 These state regulations vary greatly in how medical marijuana can be supplied. Certain states allowed “caregivers” to supply marijuana to an unlimited number of patients. This permitted producers to operate with virtually no quantity limits and little state oversight. Other states had more restrictive supply limits, allowing only home-cultivation by the patient himself or by the patient’s designated caregiver, who was limited to supplying only one patient. Finally, some states legalized state-licensed dispensaries, which could serve many patients but were subject to substantial monitoring and limited production quotas. Different production allowances can be expected to have heterogeneous effects on price, availability, quality, and product variety in the legal market. While laxer production limits should lead to lower access costs for registered medical marijuana patients, they may also have the unintended effect of increasing spillovers to youths. Adolescents largely do not have legal access to medical marijuana,6 but evidence from cigarette markets shows that youth access laws are limited in effectively reducing teenage consumption due to the presence of social markets .

As costs facing legal users in the formal market significantly affect availability to underage users through these secondary markets , changes in the size of the state medical marijuana industry may better predict changes in adolescent cannabis use than the passage of the law alone. Previous work has mostly ignored the wide variation in MMLs and their implementation. Given the heterogeneity in state supply restrictions, differences in the findings of prior work may be partly due to which MML states are used for identification. Pacula et al. recognized the importance of accounting for differences in the specific dimensions of MML policy, but there remains debate over how to categorize these regulatory differences and interpret their effects. For instance, Pacula et al. find evidence that only MMLs that legalized dispensaries saw significant increases in recreational marijuana use. However, their finding that dispensary laws have significant effects even without the existence of operational dispensaries raises questions as to how to interpret these results . And, as shown in Table 2.1, there is heterogeneity even within states that allowed dispensaries in the strictness of production restrictions. Another potential explanation for the varied findings of past work is the incomplete consideration of the role of federal policy in determining the size and structure of medical marijuana markets. While federal law has remained unchanged throughout years of state experimentation with marijuana liberalization, federal enforcement in these states has varied widely. Before 2009, the federal government made direct threats toward MML states, stating that even users and suppliers in compliance with state policy would remain subject to federal prosecution . However, between 2009 and 2012, two federal memos dramatically altered perceived federal enforcement in medical marijuana states. The Ogden Memo, announced on October 19, 2009, formalized guidelines for federal prosecutors in MML states. The memorandum maintained the government’s commitment to prosecuting significant traffickers of marijuana, but emphasized that “prosecution of individuals with cancer or other serious illnesses who use marijuana as part of a recommended treatment regimen consistent with applicable state law, or those caregivers in clear and unambiguous compliance with existing state law who provide such individuals with marijuana, is unlikely to be an efficient use of limited federal resources” . In sum, the Ogden Memo de-prioritized the federal government’s involvement in prosecuting medical marijuana users and suppliers in states with MMLs. On June 29, 2011, the US government reversed this stance by issuing the Cole Memo as a response to the government’s perceived “increase in the scope of commercial cultivation, sale, and distribution and use of marijuana for purported medical purposes” . The Cole Memo stated that individuals involved in the business of medical marijuana sales and distribution would be subject to federal enforcement action. In the months leading up to and following the memo, the Drug Enforcement Administration stepped up raids on medical marijuana producers . If changes in the risk of federal prosecution shift production costs, then both state variation in supply restrictions and time-variation in federal policy will determine the size of the legal market. Moreover, in states where legal and illegal markets for marijuana co-exist, policy changes that shift production costs in the legal market may affect price and availability in the illicit market. To better understand the effects of costs associated with state restrictions and federal enforcement, I outline a simple model of supplier behavior.

This approach is also particularly relevant when studying the provisions of a single umbrella policy. For example, for provisions of recreational cannabis legalization, exposure categories based on the overall approach to legalization in 1 state versus another may be of greater interest than the effects of individual provisions. Similarly, Erickson et al. categorized states into 4 groups on the basis of stringency of the overall alcohol policy environment and found that these categories were associated with levels of past-month alcohol consumption. Several options are available to define clusters, including manual selection, hierarchical cluster analysis, latent class analysis, and principal components analysis . Heatmaps like those presented here can help inform the selection of appropriate clusters by offering an intuitive visual reference for the likelihood that sets of policies were adopted together. Evaluating situations when each clustering approach might be preferable is a future research direction.With the rapid growth of planetary scale web services, the past few years have seen the consolidation of data center facilities at a scale never seen before. Companies like Google, Amazon and Microsoft are building huge data centers comprising several thousands of servers. Economies of scale and advances in virtualization have favored the consolidation of data center facilities resulting in the emergence of mega data centers containing hundreds of thousands of servers with billions of dollars of investment. The emergence of mega data centers came with an accompanying trend in the basic building block of data centers. The cookie cutter approach to building data centers in the early half of this decade involved using racks with 20 to 40 servers and a top of rack switch as the basic building block. But the consolidation of data center facilities and increase in size of data centers has also led to a shift in the basic building block from a rack to a modular shipping container with anywhere from 250 to 1000 servers. These self-contained shipping containers also known as pods not only include servers, but are also geared with networking, power and cooling equipment.

At the scale of a pod,hydroponic trays it is possible to build non-blocking switch fabrics to interconnect all these servers. Interconnecting multiple modular data centers or pods to construct a mega data center requires careful design of the core interconnection network. Providing the required bisection bandwidth between pods is often expensive. The traditional technique of interconnecting pods involves using a few core packet switches such as the Cisco Catalyst 6509 [Cis] and connecting all the pod switches to the core switches. As the data centers grow in size, the problem of providing sufficient inter pod bandwidth has become more and more challenging. A key driving factor that has led to consolidation of resources is the economies of scale afforded by this consolidation and the increased flexibility of placing computation and services across a large cluster. But the flexibility in placing services or virtual machines has to also be supported by sufficient bandwidth between nodes that require it. For example a large scale service such as a search engine might run on thousands of servers spread across multiple pods requiring significant inter-pod bandwidth. But this set of nodes might change over time, or the number of nodes used to serve the users might change due to increased popularity of the service. In general, it is expected that the communication patterns and bandwidth demands between different sets of nodes changes over time. With the traditional data center network architectures, the only way to provision a network with sufficient bandwidth between any set of nodes that changes over time is to build a complete non-blocking network for the entire data center with electrical packet switches. One might claim that provisioning a fully non-blocking network between all the nodes is an overkill since only some nodes would actually require significant bandwidth resources. However, this fully provisioned topology is required to support full bisection bandwidth even between a pair of pods unless the set of pods that require high bisection bandwidth never changes over time.

For example, consider a traditional network with pods containing 1000 servers each connected by 10 Gb/s links and the pods are all connected through a core layer with an over subscription ratio of 2. Now even if only 2 pods had a full bisection bandwidth requirement and the other pods require only a small amount of bandwidth, this network cannot support the full bisection bandwidth required. This makes it essential to construct fully provisioned networks even to support localized bursts that require high bandwidth as long as these sets of nodes that require high bandwidth is not fixed. Two promising technologies to provision bandwidth more flexibly in the data center include optical circuit switching and wavelength division multiplexing . Optical circuit switches are oblivious to bandwidth and a single optical port can carry several parallel channels of 10 Gb/s using WDM. If a lot of bandwidth is required from a particular source to a particular destination, optical circuits offer a cost effective way of provisioning this bandwidth. A key limitation is switching time – it can take tens of milliseconds to switch from one destination to another. So if the bandwidth demand fluctuates very rapidly, then optical switches are not very useful. Optical circuit switching has been used in the telecom industry for a long time for provisioning long haul links where the capacity is typically provisioned or changed once in a few hours if not days. In the data center, using optical circuit switching at the level of individual hosts is infeasible since hosts speak to several other hosts at short timescales of a few milliseconds or seconds. But modular data centers provide a good opportunity to leverage optical switching since the bandwidth demand when aggregated at the level of pods is relatively more stable. It is still likely that there will be some bursty communication from each pod which would be best served by electrical packets switches which can switch bursty traffic even at a nanosecond scale. Optics already form a relatively large fraction of the data center network cost.

The inter-pod network typically uses 10 GigE or faster links that span long distances of few tens of metres and require the use of optic fibres since 10 GigE over copper is only feasible over short distances of around 10 m. The use of optical fibers requires expensive SFP+ tranceivers each of which costs $200 or more. We propose Helios [FPR+10], a hybrid electrical/optical data center network architecture that unifies the benefits of electrical and optical switching to provide nearly the same performance of a traditional electrically switched network but at a much lower cost, power consumption and complexity . Helios uses packet switches and circuit switches to interconnect pods and dynamically forwards traffic over them based on the nature of traffic and provisions circuits between pods that currently require bandwidth. We have built a fully functional prototype of Helios using commercially available networking equipment and by implementing software to perform various tasks required for Helios. The main contribution of this research is the proposed design for combining optical circuit switching and electrical packet switching in a data center for a more efficient network design. We also present a technique for estimating the natural interpod bandwidth demand by ignoring any bottlenecks caused by current network conditions. We identify key challenges in building a large scale deployment like Helios and describe several research opportunities that can advance our ability to build more efficient large scale data center network designs. Our prototype illustrates the feasibility of the design and indicates the opportunity to get large benefits in cost and power by using a hybrid network architecture for interconnecting pods. Helios uses a combination of electrical packet switches and optical circuit switches at the core layer to interconnect pods. Figure 2.1 illustrates a Helios network. If there are N1 electrical core packet switches, then N1 up link ports from each pod switch connect to these N1 core packet switches. The remaining N2 up links ports from each pod switch are connected to core circuit switches. The relative fraction of packet switches and circuit switches in the core layer depends on the extent of stability in communication patterns. If the traffic pattern is very stable, most of the core switches in the network can be optical circuit switches. Intuitively,seedling starter trays this allows the circuit switching overhead to be amortized over a long period of high utilization of the circuit that is just setup. The servers in each pod are connected to the pod switch by copper links which are feasible over short distances. Due to the relatively large distances involved, the links between the pod switches and the core layer are optical links. Each up link port in the pod switches contains an optical transceiver. The links from the pod switches to the core packet switches require an optical transceiver at the core packet switch end as well. The up links which connect to the optical core circuit switches do not require transceivers in the core layer and terminate at the switch directly as it operates entirely in the optical domain. The up links that are connected to optical circuit switches can also make use of wave division multiplexing .

Suppose we use a WDM factor of w, then w up links would be combined through a passive multiplexer/demultip lexer module into a single super link that is connected to a single core circuit switch port. The different up link ports that use a single super link use transceivers of different wavelengths to allow the multiplexer/demultiplexer module to work correctly. Eventually when a circuit is setup through this core circuit switch, this particular WDM super link would be connected to another pod thereby establishing w links from the source pod to the destination pod. Essentially higher values of w allow more data to pass through a single fiber or optical circuit switch thereby reducing the number of optical circuit switches required in the topology.The software for Helios consists of 3 primary components – Pod Switch Manager, Circuit Switch Manager and the Topology Manager. All these components act in a coordinated fashion to provision bandwidth resources where they are required, when they are required. The interactions between these components is illustrated in Figure 2.2. Besides these 3 components, the core packet switches are just traditional switches that can act in a plug-and-play fashion with simple software similar to learning switches. They do not require any dynamic or specialized configuration for Helios. They can be preconfigured with the MAC addresses or IP prefixes for different pods, but this is not required.TLC was performed using Merck 60 F254 aluminum-backed plates. Flash column chromatography was performed using Silicycle silica gel . Melting points were determined using an automated Buchi B-545 melting point apparatus, which provides a specific melting point, not a range, and are corrected. 1H NMR spectra were obtained on a Bruker Avance spectrometer. 13C NMR spectra were obtained on Bruker Avance NEO and Bruker Avance spectrometers. Chemical shifts are referenced to the residual solvent signal . Infrared spectra were recorded on a Bruker Alpha spectrometer. High-resolution mass spectra were obtained using an Agilent 6545 LC/SFC Hybrid Q-TOF spectrometer. Optical rotations were taken on a Rudolph AutoPol IV polarimeter. Circular dichroism experiments were performed on a Jasco J-815 CD spectrometer. Using the procedure of Zhang,60% NaH was portionwise added to a stirred solution of 3-formylindole in tetrahydrofuran cooled in an ice bath and then the reaction was slowly warmed to room temperature. After stirring at room temperature for 30 min, phenylsulfonyl chloride was added drop wise. The reaction was stirred for 24 h at room temperature. The resulting heterogenous mixture was concentrated under reduced pressure into a crude solid. The solid was dissolved in a mixture of water and methylene chloride.The combined organic layer was dried with sodium sulfate and concentrated in vacuo. The resulting solid was dissolved in minimum amount of hot methylene chloride/hexanes mixture and allowed to cool slowly to room temperature to afford off-white crystals . The spectroscopic data of the product agreed with the reported literature.Using the procedure of Zhang,60% NaH was portion wise added to a stirred solution of 3-acetylindole in tetrahydrofuran cooled in an ice bath and then the reaction was slowly warmed to room temperature. After stirring at room temperature for 30 min, phenylsulfonyl chloride was added drop wise. The reaction was stirred for 72 h at room temperature.

Output of this phase interpolator is driven to the next repeater as depicted in Fig. 4.1. The interpolating between the two signals results in the summation of the FIR shown in Fig. 4.1. The phase interpolator weighting adjustment represents the FIR filter coefficient, and the transfer function is given by H = α + z −1 . The added phase interpolator has very little power and area cost and allows programmable α. With a programmable interpolator, the filter function can be adjusted. An α of zero passes the MDLL divided clock to the output, and an α of one forwards the reference clock like a simple buffer. Tuning the phase interpolator setting, α, changes the -3dB bandwidth. Fig. 3.15 shows the cascade of two clock repeaters with FIR filtering for the output clock. The phase noise side-band spectrum of two cascaded clock repeaters with 40% FIR interpolation coefficient is shown in Fig. 4.2. Jitter is reduced with each stage of the filtering. Since each stage can have the filtering coefficient independently adjusted, Fig. 4.3 illustrates the impact of varying α for each of 4 cascaded clock repeaters. The best combination for least jitter is found to be an α of 40% for the first 3 repeaters and an α of 0.6 in the fourth repeater. Proper choice can reduces jitter by up to 50% in a repeater stage. In our system where more repeaters are used, additional stages of FIR filtering do not reduce jitter significantly due the sharp roll-off of filtering beyond a fourth-order filter. MDLL input mux is implemented as a configurable phase interpolator, shown in Fig. 4.4, to change the relative injection strengths of the reference edge and the VCO feedback edge. It can be configured from 0 to 100% injection strength with 20% steps. Where 0%setting denotes the mux is configured to operate as a normal delay cell in the VCO, thus the CMU is configured as a PLL. A 100% setting permits full injection of reference edge, thus CMU is configured as an MDLL. Intermediate settings interpolate between reference and VCO edges,how to dry cannabis and the CMU operates as a semi-PLL/MDLL. At the output of the CMU we are adding a second mux/phase interpolator structure that takes the incoming reference clockand a divided version of the CMU as inputs.

Output of this phase interpolator goes directly to the clock driver, as depicted in Fig. 4.1. With such configurability, jitter accumulation across the different repeaters can be kept low. depending on its location on the cable. Early on in the link, when forward clock is still clean and didn’t suffer jitter accumulation, the CMU is configured in a MDLL mode to reset VCO jitter accumulation. We also set the PI to forward the incoming clock to the next stage, as depicted in Fig. 4.5. Similarly, later on the link, when clock has undergone significant jitter accumulation, we tune the CMU in a PLL mode to filter incoming clock jitter. We also set the PI to forward a divided version of the filtered PLL clock, Fig. 4.5. Intermediate settings can be used in the middle of the cable link. Multiplying DLLs have gained much interest in recent publications because of their inherent ability to reset jitter accumulation inside the VCOs compared to MPLLs. This is attributed to the fact that reference clock edges are injected into the VCO each reference cycle and thus remove the jitter accumulation memory of the VCO. This can be interpreted as designing an MPLL that has a bandwidth equivalent to the reference frequency bandwidth, compared to traditional MPLLs where bandwidth can’t be greater than one tenth of the reference frequency. One challenging aspect in the design of MDLLs is the alignment of the injected reference edge to the VCO feedback signal. As shown in Fig. 4.6, the select logic is responsible for generating an aperture that allows the reference edge inside the loop and blocks the VCO feedback signal. As shown in Fig. 4.6, any delay mismatch between the reference edge and the VCO feedback edge, or a mismatch in the charge pump would cause the pulse following the injected edge different than the remaining VCO pulses, an effect that would manifest itself as period jitter or reference spur in the frequency domain, which limits the minimum jitter attained by the MDLL. Solutions are provided to this problem in literature. In a slave oscillator is injected with the MDLL master oscillator.

The slave oscillator acts as a LPF for the period jitter. In an auxiliary calibration loop is used to measure the duty cycle error of the output. This error is then used to unbalance the charge pump current to absorb this mismatch. In digital duty cycle measurements are correlated between consecutive samples. The error is then used to steer the control voltage of the VCO. This alleviates the need of PFD and charge pump altogether. While in a phase detector that is based on chopping and correlated double sampling is used to minimize mismatches.In all of the published techniques, a select logic block generates the SEL pulse that opens the aperture for reference injection. These prior works assumed that the SEL pulse could be generated quickly enough to select the next reference edge. Moreover the position of the SEL pulse with respect to the Ref edge was overlooked as a possible factor to affect the pattern jitter. As can be seen from Fig. 4.6 the SEL pulse is asserted when the DIV signal goes high and then OUT1 signal, one of the VCO phases, goes high. The delay for generating the SEL signal from the time the VCO signal OUT1 goes high is thus equal to the delay of the CMOS level restoration buffer after OUT1, not shown in Fig. 4.6, plus 3 or 4 gate delays inside the select logic. This has to be shorter than half a clock cycle of the VCO frequency. This delay is also function of process, voltage and temperature variations. A simulation that shows the impact of the SEL pulse phase shift is illustrated in Fig. 4.7. When the SEL pulses arrives early with respect to the VCO and reference edges, a hold violation inside the select multiplexer causes significant period jitter in the VCO output. Similarly, a SEL pulse arriving late would cause a VCO setup violation inside the multiplexer and also a period jitter increase. This constraint made the design of MDLLs for multi-gigahertz applications quite a challenging task. In fact, none of the published MDLLs, as far as we know, exceeded 2GHz operation. Figure 4.8 shows the schematic of the MDLL with the proposed modification. A 360o phase rotator uses the quadrature phases of the VCO to vary delay of the SEL pulse by 1 complete clock cycle. This can compensate for any amount of latency in the select logic generation and precisely positions the SEL pulse with respect to the ref edge to minimize the period jitter. The VCO comprises 4 delay stages.

The MUX stage is configured as a delay cell and used as the third stage of the VCO to maintain the quadrature nature between the two halves of the VCO. By doing this we can have constant phase steps for the SEL pulse tuning. The output of the phase rotators is then connected to the divider and the select logic to generate the required SEL signal. A calibration loop based on a search algorithm reads the duty cycle error of the VCO output and uses this reading as a measure of period jitter. This represents the optimal aperture position. Design of the phase rotator is shown in the inset of Fig. 4.8. The interpolator comprises a bank of capacitors, connected from one terminal together to form the interpolation node,and from the other terminal each capacitor is connected to a pair of digitally controlled pass gates. Each pass gate is connected to one of the clock inputs CLKI and CLKQ. Interpolation takes place at the common terminal of the capacitors bank as a weighted sum of input clock voltages. Because summation takes place with passive components,cannabis drying rack this implementation provides better linearity performance than the conventional current-source based phase interpolator. Conventional phase interpolators suffer from nonlinearity due to finite output impedance of current sources and clock feedthrough from input to output. A 4fF/unit capacitor is used to achieve the 4-bit matching requirement for the phase interpolator. Charge pump is a critical component of this design. Any mismatches in the charge pump will be interpreted as phase mismatches between reference and divided edge. Unlike PLLs, phase mismatches in MDLL are manifested as period jitter. The charge pump used in this design minimizes static and dynamic mismatches to less than 0.5%, which was satisfactory for our application. Fig. 4.9 shows the charge pump design. Static mismatch is corrected by Transistors M1-M4, which form a singled ended replica of the charge pump. Amplifier 1 forces the output node of the replica to be equal to the control voltage by changing the down current. This guarantees that the DC up and down currents of the charge pump are equal . It also guarantees that the down current is tracking the up current with different control voltages. This means that the charge pump can be used with wider control voltage ranges and therefore lower VCO gain. Using a voltage follower amplifier connected between VCTRL and VCTRL1 is used traditionally to minimize dynamic mismatch. Keeping VCTRL1 equals to VCTRL reduces charge sharing caused by switching currents between these 2 nodes. In our design, dynamic mismatch is corrected by another replica charge pump M5-M10. This replica guarantees that current is always flowing through the charge pump branch formed by M13-M14. This branch now is identical to the single ended replica and node VCTRL1 remains equal to VCTRL. Current consumed by this replica is tens of microamperes. To minimize area, the MDLL loop filter capacitor is implemented using core thin-oxide devices rather than thick oxide devices. This reduces loop filter area by 63%.

Due to the high gate leakage of these devices, an additional compensation technique similar to is used. Amplifier 2, M17 and a replica of the loop filter capacitor are connected in a negative feedback configuration as shown in Fig. 4.9. The loop sets M17 current to be equal to the replica capacitor leakage. Consequently, the leakage current in the loop filter is provided by M18 instead of leaking the charge on the loop filter capacitor.The noise analysis in chapter 3 and section 4.1 ensures the clock propagates across the entire cable length with sufficiently low noise. For data transmission, the distance between data repeaters is essentially a point-to-point link. The primary constraint is to minimize the power/meter for the targeted data rate. Longer repeat distances reduce power/meter by amortizing the repeater circuit’s power. However, with more channel attenuation, more power is needed for equalization. For this study, CAT7 cable is used which has approximately 2.2-dB loss per meter at the data Nyquist frequency of 6GHz. This power trade-off with distance is analyzed and illustrated in Fig. 4.10. Total energy per bit for repeated data transmission across 100 meters is shown versus section repeat distance. Short distances require no equalization but pays the power penalty of terminating the cable for a given minimum receiver sensitivity. Modest equalization power is possible if data repeat distance is kept below 20dB channel loss, which is equivalent to 9m of CAT7 cable. Channel loss beyond this requires added filtering which increases the power per repeater. This analysis uses realistic circuit simulations and includes varying the driver power for higher signal swing to compensate for cable loss. A shallow optimum exists between 4 and 8 meters. An 8-m data repeating distance is chosen to minimize the cost of inserting a large number of repeaters. Figure 4.11 shows the transceiver block diagram of each repeater. The transmitter uses a half-rate architecture and comprises a 16-1 data serializer. The delay for the pre-emphasis FIR is performed in the low frequency digital section to save power. The CML driver has 1 precursor and 2 post-cursor pre-emphasis taps to achieve 9 dB of equalization. The pattern generator and the serializer are used to generate data at each repeater for signal quality and BER characterization.

We will conduct malaria vector population surveillance on a monthly basis continuously till at least 8 months after the last round of larvicide application . We will monitor both indoor- and outdoor-biting mosquito abundance using CO2-baited Centers for Disease Control light traps equipped with collection bottle rotators . The collection bottle rotator, which has eight separate plastic collection bottles, will be programmed to collect active mosquitoes at 2-h intervals between 16:00–08:00. We will place two traps within each sampling compound: one inside the living room, the other outside the house 5 m away. We will conduct a total of 64 trap-nights of vector sampling per cluster per month. This will provide an estimation precision of 0.2 mosquitoes using the previously determined standard deviation. Species of collected mosquitoes will be identified and blood-feeding status will be recorded. We will test for P. falciparum sporozoite infection and blood meal source using an enzyme-linked immunosorbent assay on all specimens. For each house where the vector population was sampled, we will record the number of sleeping persons at each house on the same day as the vector survey. We will calculate sporozoite rate and EIR for each cluster. EIRs will be calculated as, and standardized to a monthly basis. The trapping method will allow for comparison of indoor- and outdoor-biting mosquito abundance and determination of nightly biting activity patterns. We will calculate indoor and outdoor transmission intensities separately assuming that all mosquitoes collected from a compound had their blood meal from the same household. We will calculate EIR for the four study periods as describe above: preintervention period: baseline vector surveillance started at least 6 months prior to the application of long-lasting microbial larvicides till intervention, intervention period, the 8-month washout period,grow trays and post intervention period: vector surveillance continued till 8 months after the last round of larvicide application.

To determine whether new malaria vector species are present in the study sites, we will sequence the ribosomal second internal transcribed spacer and mitochondrial CO1 gene in anopheline specimens that are not amplified by the recombinant deoxyribonucleic acid polymerase chain reaction method, and we will conduct phylogenetic analysis to determine whether the new species found by Stevenson et al. are also present in the study sites.We will conduct the intervention using a two-step approach. First, we will conduct a small-scale four-cluster trial to optimize the time, duration, and quantity of LLML application. Second, we will conduct a cluster randomized trial to test the effectiveness and cost effectiveness of LLML. The design has two parallel arms, i.e., control and intervention, and allows for baseline survey without intervention and crossover .We will select four clusters, two in each county, for an entomological evaluation of the optimal larvicide application scheme . We will randomly select two clusters, one in each county, treated with larvicides and the other two sites will serve as controls . We will treat temporary habitats with FourStar controlled release granule formulation, which maintains effectiveness through wet and dry periods for up to 1 month. We will treat semipermanent habitats with FourStar 90-day briquettes and permanent habitats with FourStar 180-day briquettes. Application dosage will follow the recommendation of the manufacturer, Central Life Sciences: 10 lbs per acre of water surface for the granule formulation, and one briquette per 100 ft2 of water surface for the briquette formulations, regardless of water depth. We will re-treat the habitats every 4 to 5 months. On a weekly basis in the treatment and control sites, we will use aerial samplers to determine habitat pupal productivity, and use standard dippers to determine larval abundance. This will allow for determination of habitat productivity with a tolerable error of 0.5 mosquitoes, based on the standard deviation identified in previous studies. We will monitor indoor and outdoor vector abundance using 64 trap nights per cluster per month. This sample size will allow detection of a difference in average vector abundance of 0.12 mosquitoes with 80 % statistical power and 0.05 type-I error.

We will use ELISA methods to determine Anopheles mosquitoes’ sporozoite infection and blood feeding host preference. We will analyze the data immediately after the small scale trial using analysis of variance with repeated measures and appropriate transformation to determine the effects of habitat larviciding on mosquito abundance and transmission intensity. The percentage reduction in malaria transmission intensity will be calculated.We will assign fourteen clusters each in the two counties to intervention or no intervention by a block randomization method on the basis of clinical malaria incidence, vector density, and human population size per site. Year 1 will focus on preparing the study sites and working with clinics and hospitals to help them improve their routine malaria surveillance . In year 2, we will conduct preliminary surveys on all 28 sites to determine clinical malaria incidence, vector density, geographic information system coordinates of larval habitats, and human population size. Human population size for each cluster, stratified into three age groups will be ascertained from our existing data. We will obtain age-group level aggregated morbidity data from local hospitals and clinics where the sampled residents seek treatment. We will determine vector abundance using CO2-baited CDC light traps for 16 trap-nights per cluster per month in each of the indoor and outdoor environments. Using these data, each cluster will be allocated to either treatment or control through randomization using the following procedures. First, each of the four parameters listed above will be standardized with the highest cluster as 1. Second, we will assign the highest weight for clinical malaria cases , the lowest weight for human population size , and intermediate weights for expected vector density and larval habitats , following the method of Corbel et al.. For each cluster a rank score will be computed as the sum of weighted clinical malaria incidence, vector density, habitat abundance, and human population size. Finally, the 14 clusters within each county will be sequentially numbered according to their rank scores and sorted into seven blocks of two clusters having successive rank scores.

We expect the two clusters within each block to have similar risk characteristics for clinical malaria, vector abundance, larval habitats, and human population size. In each block, the ranks of the two clusters are put into two sealed envelopes, one cluster will be randomly allocated to treatment and another to control, using computer-generated random numbers .After the larvicide application optimization and study cluster randomization, we will treat each treatment cluster with LLML at the time interval of 4 or 5 months . The first treatment will be conducted in February-March about 1 month before the beginning of the long rainy season which usually starts in April. After three treatments, we will perform no treatments for the next 8 months. This will provide useful data on the dynamics of action of the LLML and the waning efficacy of LLML over time. These data will be important in analyzing cost-effectiveness to help optimize the timing of re-treatments. After 8 months, a total washout of the LLMLs will be assumed to have taken place. Next, we will perform a crossover and switch of the control and treatment clusters. Former control clusters will receive three rounds of LLML treatment at appropriate time intervals, and the former treatment clusters will receive no LLMLs. This strategy will minimize ascertainment biases that might be attributed to care-seeking behaviors of the population or to malaria detection and reporting by malaria treatment clinics. We will test LLMLs manufactured by Central Life Sciences. The larvicide application regime is as follows: temporary, semipermanent, and permanent habitats will be treated with FourStar controlled release granule formulation, 90-day briquettes, and 180-day briquettes, respectively. Application dosage will follow the recommendation of the manufacturer: 10 lbs per acre of water surface for the granule formulation,dry racks for weed and 100 ft2 water surface per briquette. We will conduct monthly vector surveys throughout the study period to determine indoor- and outdoor-biting vector abundance, using the same sample size of 64 trap-nights per cluster per month, and sporozoite infection and mosquito blood meal analysis will be conducted on all collected specimens. To confirm larviciding efficacy, we will examine larval abundance, age structure, and pupal productivity on a monthly basis in 100 randomly selected larval habitats each from treatment and control sites using our GIS maps and data on sites where LLML was applied.Sample size was calculated based on 2010 and 2011 active case surveillance results from Iguhu and Emutete areas. Then the number of clusters required and the number of individuals required for each cluster were calculated following the methods developed by Hayes and Bennett based on cluster-randomized trials assuming equal population for each cluster. The observed malaria incidence rate was 52.7 cases per 1000 people year in 2011. We calculated the numbers of clusters and individuals required for epidemiological assessment of the long-lasting larvicide treatments to detect a 50 % protective efficacy conferred by the treatment compared with the reference group , with a power of 80 %, significance level of 5 % and the coefficient of variation of true proportions between clusters within each treatment was assumed to be 0.15.

The estimated number of clusters for the intervention will be five and the required number of individuals for each matched-pair will be 1196; assuming a design effect of 0.25 and 20 % of subjects lost to follow-up. The estimated number of clusters for the intervention will be seven and the required number of individuals for each of the matched-pairs will be fewer than 2000. The 28 clusters proposed in the randomized cluster study will detect 50 % malaria incidence reduction with 99.9 % power and 30 % incidence reduction with 85.3 % power. This is based on the current malaria incidence rate in the study sites and a two-tailed alpha with a human population size of 2000 per cluster . If the malaria incidence is 50 % lower than the current value, the design will still detect 50 % incidence reduction with 99.7 % power and 40 % reduction with 95.2 % power .We will monitor primary and secondary endpoint outcomes throughout the 5-year study period ; data analysis will be conducted in year 5. The difference in clinical malaria incidence between treatment and control groups will be compared using Poisson multivariate regression models with intervention, age, and calendar time as covariates, using a generalized estimating equations approach. GEE is necessary since incidence will be modeled monthly as a temporally-correlated repeated measure using grouped data. Intervention will be a time-varying covariate since the treatment crosses over after three intervention rounds. Since there is no intervention in the 8 months during the washout period, interval censoring will be performed to exclude the second 4 months of data during this period. The odds ratio and the 95 % confidence interval for clinical malaria rates between treatment and control groups will be calculated. Difference in vector density and EIR will be analyzed using a negative binomial regression model and the GEE approach. In all these analyses, clusters will be indicated as intervention and control, calendar time will be categorized into: pre intervention, intervention, post intervention , washout , crossover intervention, post intervention , and nonintervention, and months since intervention will also be included as an independent variable. These variables will allow for comparison between intervention and control clusters based on baseline observations, e.g., relative reduction in vector density, and allow for evaluation of cumulative effect,e.g., the second round of treatment may produce added effect following first-round treatment. Finally, for the economic evaluation, we will calculate incremental cost-effectiveness ratios based on the primary endpoint and on long-term health outcomes including malaria deaths averted. Using the “ingredients approach”, costs will be classified according to: initial setup investment , running costs , and costs of program management and quality control . Cost data will be estimated from health facility and Ministry of Health records, LLML manufacturers and financial accounts of the research project. One-way and multi-way sensitivity analysis will be undertaken to examine the implications of potential changes in variables such as larvicide price and larviciding application frequency. ICERs will be reported from both provider and societal perspectives for different transmission intensity scenarios.Larval control and environmental management have played very important roles in malaria elimination in the United States and Europe, where today larval control using biological larvicides is the primary vector control method.