The first steps of external quality control

Thirty years ago

Sakari Närvänen, Helsinki University Central Hospital (HYKS), Children’s Hospital. Moodi 3-1993.

In the early 1960s, I set up a serum iron method in the Children’s Hospital laboratory that I felt was good, met the required criteria, was suitable for clinical use as a micromethod, and so on. I felt a certain professional pride in my expertise.

There were only a few of us hospital chemists at the time. We all knew each other. I greatly enjoyed the company of my colleague Matti Nuutinen, a chemist at the Finnish Red Cross Blood Service. We met often. On one occasion, I mentioned that I had developed an excellent iron method that worked very well. But Matti had done the same at the Blood Service. He, too, had an excellent method for iron analysis.

I suggested that we compare our methods by analysing iron concentrations from the same serum samples. And so we did.

When I then went to meet Matti with my results, there was a kind of manic determination in my stride. Apparently, Matti felt the same way. But when we placed the results side by side, we were speechless. They differed rather uncomfortably. Naturally, each of us suspected the other, although such collegial suspicion could not be voiced aloud. How could we, as good friends, resolve the matter without leaving a bad aftertaste?

Then we had an idea. Let’s send the same serum samples to a few laboratories at HYKS clinics and city hospitals and ask them to analyse serum iron and TIBC. We would easily and conveniently get an answer as to whose iron method was in order.

We received the responses. Once again, we spent a moment in silence, as it still did not become clear whose method was better. What did become clear, however, was that the variation in iron methods across Helsinki’s clinical laboratories was outrageously large. The lowest value was 55, and the highest was 135 micrograms per cent. We sent the results to our colleagues, along with our perhaps unnecessary remark: “something should be done.”

Matti and I considered the reasons for the large variation in results. Soon, we sent out new serum samples, iron standards and a haemoglobin solution to determine whether the problem lay more with the methods themselves or with the poor condition of the instruments. We received responses and information about the equipment used, including Klett-Summerson and Beckman B photometers. Our colleagues had clearly become more alert. The variation in the second round was significantly smaller than in the first. Matti’s and my discussions and meetings had produced results in iron analyses.

In the 1960s, I also developed a creatinine method, incorporating Lloyd’s reagent to improve specificity. From textbooks, I had read that this was necessary when analysing children’s blood samples. The method seemed to work well, just like the iron method had earlier. When, for comparison, I contacted my colleague Aulis Hyvärinen and suggested comparing creatinine methods between the Children’s Hospital and the Meilahti laboratory, my bravado was considerably more subdued than it had been with iron. Our suspicion about the variability of results felt very likely. Once again, there were new investigations, new serum samples for analysis, and comparisons of results. It was no great revelation to realise and wonder about the existence of other methods as well, and the likely variability inherent in them. The work threatened to expand in challenging ways.

Interest in our comparisons had also been awakened in other laboratories. We felt the need to expand our circle, which was then joined by Chief Chemist Nils-Erik Saris from Meilahti and Chief Physician Erkki Leskinen from the Kivelä–Hesperia laboratory. The four of us sat together for months, every Thursday, in Aulis’s office, discussing the situation and how it could be improved.

In 1967, our first publication appeared, presenting our results and views [1]. Based on the literature, the reliability of laboratory results in Finland was better than in Canada and Norway, from which comparative data were available. “Laboratories operating with commendable accuracy were most numerous in Finland at 19%, compared with 10% in Norway and ‘very few’ in Canada.”

At the end of 1967, Nisse Saris and I had the rewarding opportunity to attend an international quality control symposium in Geneva, where we did not have to be ashamed of the standard of our country’s laboratory results. I wrote a brief summary of the symposium’s views for Duodecim [2].

While we were in Geneva, the elite of winegrowers were holding their own magnificent celebration at our hotel. In our circle, we strove for uniform quality. In theirs, professionals whose products stood out from the crowd were celebrated and praised.

Interest in the reliability and comparability of laboratory results spread, as did our desire to compare. However, creating systems required both serum and the processing of results. At first, it seemed that there was plenty of leftover serum at the Meilahti hospital laboratory, so we could simply collect it. There was indeed plenty—but what if it contained hepatitis and we spread it around? Could we take that risk? At first we did, but then we began to worry. Mixing, aliquoting and sending the sera was another task altogether, which Aulis handled admirably, sacrificing much of his own time. Even then, Aulis was dedicated to developing control serum systems. His doctoral dissertation [3] dealt with the statistics of results obtained from control sera.

When we considered where to obtain serum for controls, we realised that there was blood other than leftover samples from Meilahti. Surely cow blood and serum would be just as usable as human serum. We called the city slaughterhouse and asked about the possibility of collecting cow blood. We received a friendly and positive invitation.

I remember how, the night before, I prepared rubber boots and a raincoat for the morning. Aulis had the same gear when we set off to the slaughterhouse in his Peugeot, bringing at least a few buckets with us.

Once there, surrounded by a particular smell, we felt it appropriate to explain to the butchers what the blood would be used for. Judging by their expressions, they did not understand us, but things nevertheless began to move forward. We were directed to the right place, where cows were hanging by one leg on chains. Then a professional made a precise cut with his blade, and the bucket was placed underneath. It was easy and straightforward to collect enough blood. We drove carefully and without spilling back to the Children’s Hospital.

Since we were building a control serum system, we naturally did not use any anticoagulant. The aim was to obtain a pure product—true serum. At the Children’s Hospital, the buckets were lifted onto a table, and we confidently announced to the girls that “we went to get enough blood.”

We waited for the moment when the blood would clot sufficiently, and the serum would rise to the surface, just as it does in a centrifuge tube. Nothing happened. The contents of the buckets remained unchanged. No serum appeared. What should we do? We did not have bucket-sized centrifuge tubes. What about filtration? There were no bucket-sized funnels either. Let’s pour the whole mess onto a sufficiently large piece of gauze—it should filter through. We found a square metre of gauze, secured it well at the corners, and poured the bucket contents to filter. The result was a coffee cupful of bright red “serum.” The whole operation had clearly gone wrong.

We could not come up with any way to obtain clear serum. We had failed. Fortunately, the slaughterhouse did not charge us or HYKS, so we did not have to worry about administration or auditors. But what should we do with the fresh blood pudding—bucket-sized lumps of fresh blood? The kitchen would hardly be interested in making blood pancakes. There was only one option: down the toilet.

However, toilet plumbing is designed for slimmer products, so we fetched a long knife and sliced the pudding. “That should fit.” But when the flush began, the pudding clogged the drain. The level of bloody water rose higher and higher in the toilet bowl. There were no tools anywhere, let alone a plunger. Sleeves up quickly, arm in up to the elbow, mashing the pudding in the bowl. After all, it was clean blood, as was the water flushing it away. The belch of the toilet as the bowl finally emptied put an end to our experiment with cow blood. All that remained was “Mansikki,” the nickname by which the control serum was long known.

I still have, from the 1960s, a metre-high and three-metre-long sheet of graph paper on which control serum values from different methods were recorded at the time. Laboratory technicians used secret codes alongside the entries. The system revealed information about each technician’s skills and honesty, and also showed whether, for example, Mondays differed somehow from other weekdays. When a control batch changed, no one knew where to place their point—cheating was impossible. Oh, those days!

In 1968, the situation began to grow beyond our control. I do not know the details of the end of that year, because I left for three years of development aid work in Ethiopia. While there, I learned that “Kliinisten laboratoriotutkimusten Laaduntarkkailu Oy” (Quality Control of Clinical Laboratory Investigations Ltd) had been founded.

References

1. Saris N-E, Nuutinen M ja Närvänen S. Maamme kliinisten laboratorioiden analyysitulosten vaihtelu. Duodecim 83: 1104 – 111, 1967.
2. Närvänen S. Laboratorion kontrollijärjestelmät muissa maissa. Duodecim 84: 743 – 744, 1968.
3. Hyvärinen A. A Model for the Statistical Description of Analytical Errors Occurring in Clinical Chemical Laboratories with Time. Scanj J Clin & Lab Invest vol 45, suppl 175, 1985.