I've told you about the idea of analysis. It's a common word that you see in newspapers, scientific journals, and lots of other places. It can mean slightly different things to different people. But in public health, you should think of analysis as breaking something down so that it can be more easily understood. I want to tell you about analysis in health intelligence, where to start, and what to think about. I've shown you the health intelligence cycle before. So you know that getting hold of the data is only the first step. As a junior member of staff, it's normal to be faced with quite ambiguous tasks from senior colleagues. Your director may say have a look at this and tell me what you think or can you figure out what to do about x. Ambiguity is horrible and it's particularly difficult when you come from a scientific background and you are used to having a very specific task. I tell my postgraduate students and trainees every year that the value of public health is not in just answering questions, it's knowing what to ask in the first place. Let me explain to you what I mean. If I were to ask you for a cross tabulation of the association between ethnicity and glucose control in type two diabetes mellitus for the population of London and here is the data, you'll probably know where to start. But if I were to come to you and ask what should I do about inequalities and cardiovascular outcome in urban areas of England, you'd quite rightly find it more difficult to know what to do next. You see as a public health specialist, you'll be expected to figure out the question and then the answer. You can only begin to piece together the answer by getting the question right. Clients and partners come to us because they don't always know the right question to ask. That's our value as public health experts, and that's why you have to begin. I told you before that in high income settings there's commonly no shortage of data. Perhaps there's a shortage of high-quality data or data precisely related to the question that you're answering but there's normally plenty of data points from which you can make some level of informed estimate using your professional judgment. Let's return to the question, what should I do about inequalities in cardiovascular outcome in urban areas of England. Let's bring that down. In fact, let's bring that down before undertaking any quantitative analysis. Don't even open a spreadsheet or your stats package. I want to introduce you to the diamond. This may seem a bit strange at first but bear with me. This is the diamond. This is where you are and this is where you want to get to. I want you to first think about the task broadly, opening the diamond. Think about what cardiovascular outcomes are, what inequalities there might be, and what urban really means in England. You're expanding your thinking, going wider on the diamond. Now, if you're starting a new career in public health practice, remember that it's fairly unusual to be doing something that no one in the world has done before. There's likely to be someone somewhere who's done something similar. So have a look. Search engines are a wonderful thing. The first half of your diamond is bringing together everything. You bring your experience, your knowledge, evidence from research, expert opinion from colleagues. If you have time and resource, setting up a brainstorming meeting for particularly difficult projects can be really useful at this stage. But the challenge, and I see this very frequently, is knowing when to stop. Finding evidence is a process that is potentially infinite. Set yourself a length of time, minutes, hours, or days, depending on the project significance and the resource available. Once you've hit that time point, that's your inflection point. You are here on the diamond. You have to stop going wide and you have to narrow down. Your job is to develop hypotheses that you can test or lines of inquiry that you can explore. I can't emphasize enough how important it is to write this down. You'll be surprised how many times you need to refine your hypothesis or questions. What you're doing here is breaking down the problem. The end of this diamond is having answerable questions. Let me show you. Let's take the following bucket of connections that we've sourced in the first half of the diamond. Cardiovascular morbidity and mortality is associated with time of diagnosis, quality of primary care services, level of patient activation, the patient's lifestyle, presence of co-morbidities. Now we suspect that outcomes are distributed unevenly across different communities of interest, socioeconomic, geographical, ethnic, or by age. Our job now is to establish on the basis of the above what are the questions that we can potentially ask. This is how that might look. How are the variables connected? Which are the most important inequalities? What are the factors that can be moderated and how? What are the next steps? You are at the end of the first diamond. You've taken an ambiguous task and you've broken it down into answerable questions. But you'll notice that I said this is the end of the first diamond. There are two diamonds. More on that next.
